WmlToHtmlConverter.cs 195 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227222822292230223122322233223422352236223722382239224022412242224322442245224622472248224922502251225222532254225522562257225822592260226122622263226422652266226722682269227022712272227322742275227622772278227922802281228222832284228522862287228822892290229122922293229422952296229722982299230023012302230323042305230623072308230923102311231223132314231523162317231823192320232123222323232423252326232723282329233023312332233323342335233623372338233923402341234223432344234523462347234823492350235123522353235423552356235723582359236023612362236323642365236623672368236923702371237223732374237523762377237823792380238123822383238423852386238723882389239023912392239323942395239623972398239924002401240224032404240524062407240824092410241124122413241424152416241724182419242024212422242324242425242624272428242924302431243224332434243524362437243824392440244124422443244424452446244724482449245024512452245324542455245624572458245924602461246224632464246524662467246824692470247124722473247424752476247724782479248024812482248324842485248624872488248924902491249224932494249524962497249824992500250125022503250425052506250725082509251025112512251325142515251625172518251925202521252225232524252525262527252825292530253125322533253425352536253725382539254025412542254325442545254625472548254925502551255225532554255525562557255825592560256125622563256425652566256725682569257025712572257325742575257625772578257925802581258225832584258525862587258825892590259125922593259425952596259725982599260026012602260326042605260626072608260926102611261226132614261526162617261826192620262126222623262426252626262726282629263026312632263326342635263626372638263926402641264226432644264526462647264826492650265126522653265426552656265726582659266026612662266326642665266626672668266926702671267226732674267526762677267826792680268126822683268426852686268726882689269026912692269326942695269626972698269927002701270227032704270527062707270827092710271127122713271427152716271727182719272027212722272327242725272627272728272927302731273227332734273527362737273827392740274127422743274427452746274727482749275027512752275327542755275627572758275927602761276227632764276527662767276827692770277127722773277427752776277727782779278027812782278327842785278627872788278927902791279227932794279527962797279827992800280128022803280428052806280728082809281028112812281328142815281628172818281928202821282228232824282528262827282828292830283128322833283428352836283728382839284028412842284328442845284628472848284928502851285228532854285528562857285828592860286128622863286428652866286728682869287028712872287328742875287628772878287928802881288228832884288528862887288828892890289128922893289428952896289728982899290029012902290329042905290629072908290929102911291229132914291529162917291829192920292129222923292429252926292729282929293029312932293329342935293629372938293929402941294229432944294529462947294829492950295129522953295429552956295729582959296029612962296329642965296629672968296929702971297229732974297529762977297829792980298129822983298429852986298729882989299029912992299329942995299629972998299930003001300230033004300530063007300830093010301130123013301430153016301730183019302030213022302330243025302630273028302930303031303230333034303530363037303830393040304130423043304430453046304730483049305030513052305330543055305630573058305930603061306230633064306530663067306830693070307130723073307430753076307730783079308030813082308330843085308630873088308930903091309230933094309530963097309830993100310131023103310431053106310731083109311031113112311331143115311631173118311931203121312231233124312531263127312831293130313131323133313431353136313731383139314031413142314331443145314631473148314931503151315231533154315531563157315831593160316131623163316431653166316731683169317031713172317331743175317631773178317931803181318231833184318531863187318831893190319131923193319431953196319731983199320032013202320332043205320632073208320932103211321232133214321532163217321832193220322132223223322432253226322732283229323032313232323332343235323632373238323932403241324232433244324532463247324832493250325132523253325432553256325732583259326032613262326332643265326632673268326932703271327232733274327532763277327832793280328132823283328432853286328732883289329032913292329332943295329632973298329933003301330233033304330533063307330833093310331133123313331433153316331733183319332033213322332333243325332633273328332933303331333233333334333533363337333833393340334133423343334433453346334733483349335033513352335333543355335633573358335933603361336233633364336533663367336833693370337133723373337433753376337733783379338033813382338333843385338633873388338933903391339233933394339533963397339833993400340134023403340434053406340734083409341034113412341334143415341634173418341934203421342234233424342534263427342834293430343134323433343434353436343734383439344034413442344334443445344634473448344934503451345234533454345534563457345834593460346134623463346434653466346734683469347034713472347334743475347634773478347934803481348234833484348534863487348834893490349134923493349434953496349734983499350035013502350335043505350635073508350935103511351235133514351535163517351835193520352135223523352435253526352735283529353035313532353335343535353635373538353935403541354235433544354535463547354835493550355135523553355435553556355735583559356035613562356335643565356635673568356935703571357235733574357535763577357835793580358135823583358435853586358735883589359035913592359335943595359635973598359936003601360236033604360536063607360836093610361136123613361436153616361736183619362036213622362336243625362636273628362936303631363236333634363536363637363836393640364136423643364436453646364736483649365036513652365336543655365636573658365936603661366236633664366536663667366836693670367136723673367436753676367736783679368036813682368336843685368636873688368936903691369236933694369536963697369836993700370137023703370437053706370737083709371037113712371337143715371637173718371937203721372237233724372537263727372837293730373137323733373437353736373737383739374037413742374337443745374637473748374937503751375237533754375537563757375837593760376137623763376437653766376737683769377037713772377337743775377637773778377937803781378237833784378537863787378837893790379137923793379437953796379737983799380038013802380338043805380638073808380938103811381238133814381538163817381838193820382138223823382438253826382738283829383038313832383338343835383638373838383938403841384238433844384538463847384838493850385138523853385438553856385738583859386038613862386338643865386638673868386938703871387238733874387538763877387838793880388138823883388438853886388738883889389038913892389338943895389638973898389939003901390239033904390539063907390839093910391139123913391439153916391739183919392039213922392339243925392639273928392939303931393239333934393539363937393839393940394139423943394439453946394739483949395039513952395339543955395639573958395939603961396239633964396539663967396839693970397139723973397439753976397739783979398039813982398339843985398639873988398939903991399239933994399539963997399839994000400140024003400440054006400740084009401040114012401340144015401640174018401940204021402240234024402540264027402840294030403140324033403440354036403740384039404040414042404340444045404640474048404940504051405240534054405540564057405840594060406140624063406440654066406740684069407040714072407340744075407640774078407940804081408240834084408540864087408840894090409140924093409440954096409740984099410041014102410341044105410641074108410941104111411241134114411541164117411841194120412141224123412441254126412741284129413041314132413341344135413641374138413941404141414241434144414541464147414841494150415141524153415441554156415741584159416041614162416341644165416641674168416941704171417241734174417541764177417841794180418141824183418441854186418741884189419041914192419341944195419641974198419942004201420242034204420542064207420842094210421142124213421442154216421742184219422042214222422342244225422642274228422942304231423242334234423542364237423842394240424142424243424442454246424742484249425042514252425342544255425642574258425942604261426242634264426542664267426842694270427142724273427442754276427742784279428042814282428342844285428642874288428942904291429242934294429542964297429842994300430143024303430443054306430743084309431043114312431343144315431643174318431943204321432243234324432543264327432843294330433143324333433443354336433743384339434043414342434343444345434643474348434943504351435243534354435543564357435843594360436143624363436443654366436743684369437043714372437343744375437643774378437943804381438243834384438543864387438843894390439143924393439443954396439743984399440044014402440344044405440644074408
  1. // Copyright (c) Microsoft. All rights reserved.
  2. // Licensed under the MIT license. See LICENSE file in the project root for full license information.
  3. using System;
  4. using System.Collections.Generic;
  5. using System.Diagnostics.CodeAnalysis;
  6. using System.Drawing;
  7. using System.Drawing.Imaging;
  8. using System.Globalization;
  9. using System.IO;
  10. using System.Linq;
  11. using System.Text;
  12. using System.Xml;
  13. using System.Xml.Linq;
  14. using System.Xml.Xsl;
  15. using DocumentFormat.OpenXml;
  16. using DocumentFormat.OpenXml.Drawing;
  17. using DocumentFormat.OpenXml.Drawing.Charts;
  18. using DocumentFormat.OpenXml.Office2010.Word.DrawingShape;
  19. using DocumentFormat.OpenXml.Packaging;
  20. using HTEXLib.COMM;
  21. using HTEXLib.COMM.Helpers;
  22. using HTEXLib.DOCX.OpenXmlTool;
  23. using HTEXLib.Helpers;
  24. using HTEXLib.Helpers.ShapeHelpers;
  25. using HTEXLib.Models.Inner;
  26. // 200e lrm - LTR
  27. // 200f rlm - RTL
  28. // todo need to set the HTTP "Content-Language" header, for instance:
  29. // Content-Language: en-US
  30. // Content-Language: fr-FR
  31. namespace OpenXmlPowerTools
  32. {
  33. public partial class WmlDocument
  34. {
  35. [SuppressMessage("ReSharper", "UnusedMember.Global")]
  36. public XElement ConvertToHtml(WmlToHtmlConverterSettings htmlConverterSettings)
  37. {
  38. return WmlToHtmlConverter.ConvertToHtml(this, htmlConverterSettings);
  39. }
  40. [SuppressMessage("ReSharper", "UnusedMember.Global")]
  41. public XElement ConvertToHtml(HtmlConverterSettings htmlConverterSettings)
  42. {
  43. WmlToHtmlConverterSettings settings = new WmlToHtmlConverterSettings(htmlConverterSettings);
  44. return WmlToHtmlConverter.ConvertToHtml(this, settings);
  45. }
  46. }
  47. [SuppressMessage("ReSharper", "FieldCanBeMadeReadOnly.Global")]
  48. public class WmlToHtmlConverterSettings
  49. {
  50. public string PageTitle;
  51. public bool TableInnerCss;
  52. public string CssClassPrefix;
  53. public bool FabricateCssClasses;
  54. public string GeneralCss;
  55. public string AdditionalCss;
  56. public bool RestrictToSupportedLanguages;
  57. public bool RestrictToSupportedNumberingFormats;
  58. public Dictionary<string, Func<string, int, string, string>> ListItemImplementations;
  59. public Func<ImageInfo, XElement> ImageHandler;
  60. public WmlToHtmlConverterSettings()
  61. {
  62. PageTitle = "";
  63. CssClassPrefix = "pt-";
  64. TableInnerCss = true;
  65. FabricateCssClasses = true;
  66. GeneralCss = "span {/* white-space: pre-wrap;*/ }";
  67. AdditionalCss = "";
  68. RestrictToSupportedLanguages = false;
  69. RestrictToSupportedNumberingFormats = false;
  70. ListItemImplementations = ListItemRetrieverSettings.DefaultListItemTextImplementations;
  71. }
  72. public WmlToHtmlConverterSettings(HtmlConverterSettings htmlConverterSettings)
  73. {
  74. TableInnerCss = htmlConverterSettings.TableInnerCss;
  75. PageTitle = htmlConverterSettings.PageTitle;
  76. CssClassPrefix = htmlConverterSettings.CssClassPrefix;
  77. FabricateCssClasses = htmlConverterSettings.FabricateCssClasses;
  78. GeneralCss = htmlConverterSettings.GeneralCss;
  79. AdditionalCss = htmlConverterSettings.AdditionalCss;
  80. RestrictToSupportedLanguages = htmlConverterSettings.RestrictToSupportedLanguages;
  81. RestrictToSupportedNumberingFormats = htmlConverterSettings.RestrictToSupportedNumberingFormats;
  82. ListItemImplementations = htmlConverterSettings.ListItemImplementations;
  83. ImageHandler = htmlConverterSettings.ImageHandler;
  84. }
  85. }
  86. [SuppressMessage("ReSharper", "FieldCanBeMadeReadOnly.Global")]
  87. public class HtmlConverterSettings
  88. {
  89. public bool TableInnerCss;
  90. public string PageTitle;
  91. public string CssClassPrefix;
  92. public bool FabricateCssClasses;
  93. public string GeneralCss;
  94. public string AdditionalCss;
  95. public bool RestrictToSupportedLanguages;
  96. public bool RestrictToSupportedNumberingFormats;
  97. public Dictionary<string, Func<string, int, string, string>> ListItemImplementations;
  98. public Func<ImageInfo, XElement> ImageHandler;
  99. public HtmlConverterSettings()
  100. {
  101. TableInnerCss = true;
  102. PageTitle = "";
  103. CssClassPrefix = "pt-";
  104. FabricateCssClasses = true;
  105. GeneralCss = "span { /*white-space: pre-wrap;*/ }";
  106. AdditionalCss = "";
  107. RestrictToSupportedLanguages = false;
  108. RestrictToSupportedNumberingFormats = false;
  109. ListItemImplementations = ListItemRetrieverSettings.DefaultListItemTextImplementations;
  110. }
  111. }
  112. public static class HtmlConverter
  113. {
  114. public static XElement ConvertToHtml(WmlDocument wmlDoc, HtmlConverterSettings htmlConverterSettings)
  115. {
  116. WmlToHtmlConverterSettings settings = new WmlToHtmlConverterSettings(htmlConverterSettings);
  117. return WmlToHtmlConverter.ConvertToHtml(wmlDoc, settings);
  118. }
  119. public static XElement ConvertToHtml(WordprocessingDocument wDoc, HtmlConverterSettings htmlConverterSettings)
  120. {
  121. WmlToHtmlConverterSettings settings = new WmlToHtmlConverterSettings(htmlConverterSettings);
  122. return WmlToHtmlConverter.ConvertToHtml(wDoc, settings);
  123. }
  124. }
  125. [SuppressMessage("ReSharper", "NotAccessedField.Global")]
  126. [SuppressMessage("ReSharper", "UnusedMember.Global")]
  127. public class ImageInfo
  128. {
  129. public string base64;
  130. public Bitmap Bitmap;
  131. public XAttribute ImgStyleAttribute;
  132. public string ContentType;
  133. public XElement DrawingElement;
  134. public string AltText;
  135. public string Mathxml;
  136. public const int EmusPerInch = 914400;
  137. public const int EmusPerCm = 360000;
  138. }
  139. public static class WmlToHtmlConverter
  140. {
  141. public static XElement ConvertToHtml(WmlDocument doc, WmlToHtmlConverterSettings htmlConverterSettings)
  142. {
  143. using (OpenXmlMemoryStreamDocument streamDoc = new OpenXmlMemoryStreamDocument(doc))
  144. {
  145. using (WordprocessingDocument document = streamDoc.GetWordprocessingDocument())
  146. {
  147. return ConvertToHtml(document, htmlConverterSettings);
  148. }
  149. }
  150. }
  151. public static XElement ConvertToHtml(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings htmlConverterSettings)
  152. {
  153. RevisionAccepter.AcceptRevisions(wordDoc);
  154. SimplifyMarkupSettings simplifyMarkupSettings = new SimplifyMarkupSettings
  155. {
  156. RemoveComments = true,
  157. RemoveContentControls = true,
  158. RemoveEndAndFootNotes = true,
  159. RemoveFieldCodes = false,
  160. RemoveLastRenderedPageBreak = true,
  161. RemovePermissions = true,
  162. RemoveProof = true,
  163. RemoveRsidInfo = true,
  164. RemoveSmartTags = true,
  165. RemoveSoftHyphens = true,
  166. RemoveGoBackBookmark = true,
  167. ReplaceTabsWithSpaces = false,
  168. };
  169. MarkupSimplifier.SimplifyMarkup(wordDoc, simplifyMarkupSettings);
  170. FormattingAssemblerSettings formattingAssemblerSettings = new FormattingAssemblerSettings
  171. {
  172. RemoveStyleNamesFromParagraphAndRunProperties = false,
  173. ClearStyles = false,
  174. RestrictToSupportedLanguages = htmlConverterSettings.RestrictToSupportedLanguages,
  175. RestrictToSupportedNumberingFormats = htmlConverterSettings.RestrictToSupportedNumberingFormats,
  176. CreateHtmlConverterAnnotationAttributes = true,
  177. OrderElementsPerStandard = false,
  178. ListItemRetrieverSettings =
  179. htmlConverterSettings.ListItemImplementations == null ?
  180. new ListItemRetrieverSettings()
  181. {
  182. ListItemTextImplementations = ListItemRetrieverSettings.DefaultListItemTextImplementations,
  183. } :
  184. new ListItemRetrieverSettings()
  185. {
  186. ListItemTextImplementations = htmlConverterSettings.ListItemImplementations,
  187. },
  188. };
  189. FormattingAssembler.AssembleFormatting(wordDoc, formattingAssemblerSettings);
  190. InsertAppropriateNonbreakingSpaces(wordDoc);
  191. CalculateSpanWidthForTabs(wordDoc);
  192. ReverseTableBordersForRtlTables(wordDoc);
  193. AdjustTableBorders(wordDoc);
  194. XElement rootElement = wordDoc.MainDocumentPart.GetXDocument().Root;
  195. FieldRetriever.AnnotateWithFieldInfo(wordDoc.MainDocumentPart);
  196. AnnotateForSections(wordDoc);
  197. XElement xhtml = (XElement)ConvertToHtmlTransform(wordDoc, htmlConverterSettings,
  198. rootElement, false, 0m);
  199. ReifyStylesAndClasses(htmlConverterSettings, xhtml);
  200. // Note: the xhtml returned by ConvertToHtmlTransform contains objects of type
  201. // XEntity. PtOpenXmlUtil.cs define the XEntity class. See
  202. // http://blogs.msdn.com/ericwhite/archive/2010/01/21/writing-entity-references-using-linq-to-xml.aspx
  203. // for detailed explanation.
  204. //
  205. // If you further transform the XML tree returned by ConvertToHtmlTransform, you
  206. // must do it correctly, or entities will not be serialized properly.
  207. return xhtml;
  208. }
  209. private static void ReverseTableBordersForRtlTables(WordprocessingDocument wordDoc)
  210. {
  211. XDocument xd = wordDoc.MainDocumentPart.GetXDocument();
  212. foreach (var tbl in xd.Descendants(W.tbl))
  213. {
  214. var bidiVisual = tbl.Elements(W.tblPr).Elements(W.bidiVisual).FirstOrDefault();
  215. if (bidiVisual == null)
  216. continue;
  217. var tblBorders = tbl.Elements(W.tblPr).Elements(W.tblBorders).FirstOrDefault();
  218. if (tblBorders != null)
  219. {
  220. var left = tblBorders.Element(W.left);
  221. if (left != null)
  222. left = new XElement(W.right, left.Attributes());
  223. var right = tblBorders.Element(W.right);
  224. if (right != null)
  225. right = new XElement(W.left, right.Attributes());
  226. var newTblBorders = new XElement(W.tblBorders,
  227. tblBorders.Element(W.top),
  228. left,
  229. tblBorders.Element(W.bottom),
  230. right);
  231. tblBorders.ReplaceWith(newTblBorders);
  232. }
  233. foreach (var tc in tbl.Elements(W.tr).Elements(W.tc))
  234. {
  235. var tcBorders = tc.Elements(W.tcPr).Elements(W.tcBorders).FirstOrDefault();
  236. if (tcBorders != null)
  237. {
  238. var left = tcBorders.Element(W.left);
  239. if (left != null)
  240. left = new XElement(W.right, left.Attributes());
  241. var right = tcBorders.Element(W.right);
  242. if (right != null)
  243. right = new XElement(W.left, right.Attributes());
  244. var newTcBorders = new XElement(W.tcBorders,
  245. tcBorders.Element(W.top),
  246. left,
  247. tcBorders.Element(W.bottom),
  248. right);
  249. tcBorders.ReplaceWith(newTcBorders);
  250. }
  251. }
  252. }
  253. }
  254. private static void ReifyStylesAndClasses(WmlToHtmlConverterSettings htmlConverterSettings, XElement xhtml)
  255. {
  256. if (htmlConverterSettings.FabricateCssClasses)
  257. {
  258. var usedCssClassNames = new HashSet<string>();
  259. var elementsThatNeedClasses = xhtml
  260. .DescendantsAndSelf()
  261. .Select(d => new
  262. {
  263. Element = d,
  264. Styles = d.Annotation<Dictionary<string, string>>(),
  265. })
  266. .Where(z => z.Styles != null);
  267. var augmented = elementsThatNeedClasses
  268. .Select(p => new
  269. {
  270. p.Element,
  271. p.Styles,
  272. StylesString = p.Element.Name.LocalName + "|" + p.Styles.OrderBy(k => k.Key).Select(s => string.Format("{0}: {1};", s.Key, s.Value)).StringConcatenate(),
  273. })
  274. .GroupBy(p => p.StylesString)
  275. .ToList();
  276. int classCounter = 1000000;
  277. var sb = new StringBuilder();
  278. sb.Append(Environment.NewLine);
  279. foreach (var grp in augmented)
  280. {
  281. string classNameToUse;
  282. var firstOne = grp.First();
  283. var styles = firstOne.Styles;
  284. if (styles.ContainsKey("PtStyleName"))
  285. {
  286. classNameToUse = htmlConverterSettings.CssClassPrefix + styles["PtStyleName"];
  287. if (usedCssClassNames.Contains(classNameToUse))
  288. {
  289. classNameToUse = htmlConverterSettings.CssClassPrefix +
  290. styles["PtStyleName"] + "-" +
  291. classCounter.ToString().Substring(1);
  292. classCounter++;
  293. }
  294. }
  295. else
  296. {
  297. classNameToUse = htmlConverterSettings.CssClassPrefix +
  298. classCounter.ToString().Substring(1);
  299. classCounter++;
  300. }
  301. usedCssClassNames.Add(classNameToUse);
  302. sb.Append(firstOne.Element.Name.LocalName + "." + classNameToUse + " {" + Environment.NewLine);
  303. foreach (var st in firstOne.Styles.Where(s => s.Key != "PtStyleName"))
  304. {
  305. var s = " " + st.Key + ": " + st.Value + ";" + Environment.NewLine;
  306. sb.Append(s);
  307. }
  308. sb.Append("}" + Environment.NewLine);
  309. var classAtt = new XAttribute("class", classNameToUse);
  310. foreach (var gc in grp)
  311. gc.Element.Add(classAtt);
  312. }
  313. var styleValue = htmlConverterSettings.GeneralCss + sb + htmlConverterSettings.AdditionalCss;
  314. SetStyleElementValue(xhtml, styleValue);
  315. }
  316. else
  317. {
  318. // Previously, the h:style element was not added at this point. However,
  319. // at least the General CSS will contain important settings.
  320. SetStyleElementValue(xhtml, htmlConverterSettings.GeneralCss + htmlConverterSettings.AdditionalCss);
  321. foreach (var d in xhtml.DescendantsAndSelf())
  322. {
  323. var style = d.Annotation<Dictionary<string, string>>();
  324. if (style == null)
  325. continue;
  326. var styleValue =
  327. style
  328. .Where(p => p.Key != "PtStyleName")
  329. .OrderBy(p => p.Key)
  330. .Select(e => string.Format("{0}: {1};", e.Key, e.Value))
  331. .StringConcatenate();
  332. XAttribute st = new XAttribute("style", styleValue);
  333. if (d.Attribute("style") != null)
  334. d.Attribute("style").Value += styleValue;
  335. else
  336. d.Add(st);
  337. }
  338. }
  339. if (htmlConverterSettings.TableInnerCss)
  340. {
  341. SetStyleElementValue(xhtml, htmlConverterSettings.GeneralCss + htmlConverterSettings.AdditionalCss);
  342. foreach (var d in xhtml.DescendantsAndSelf())
  343. {
  344. if (d.Name.LocalName.ToString().Equals("table") || d.Name.LocalName.ToString().Equals("tr") || d.Name.LocalName.ToString().Equals("td"))
  345. {
  346. var style = d.Annotation<Dictionary<string, string>>();
  347. if (style == null)
  348. continue;
  349. var styleValue =
  350. style
  351. .Where(p => p.Key != "PtStyleName")
  352. .OrderBy(p => p.Key)
  353. .Select(e => string.Format("{0}: {1};", e.Key, e.Value))
  354. .StringConcatenate();
  355. XAttribute st = new XAttribute("style", styleValue);
  356. if (d.Attribute("style") != null)
  357. d.Attribute("style").Value += styleValue;
  358. else
  359. d.Add(st);
  360. }
  361. }
  362. }
  363. }
  364. private static void SetStyleElementValue(XElement xhtml, string styleValue)
  365. {
  366. var styleElement = xhtml
  367. .Descendants(Xhtml.style)
  368. .FirstOrDefault();
  369. if (styleElement != null)
  370. styleElement.Value = styleValue;
  371. else
  372. {
  373. styleElement = new XElement(Xhtml.style, styleValue);
  374. var head = xhtml.Element(Xhtml.head);
  375. if (head != null)
  376. head.Add(styleElement);
  377. }
  378. }
  379. private static object ConvertToHtmlTransform(WordprocessingDocument wordDoc,
  380. WmlToHtmlConverterSettings settings, XNode node,
  381. bool suppressTrailingWhiteSpace,
  382. decimal currentMarginLeft)
  383. {
  384. var element = node as XElement;
  385. if (element == null) return null;
  386. // Transform the w:document element to the XHTML h:html element.
  387. // The h:head element is laid out based on the W3C's recommended layout, i.e.,
  388. // the charset (using the HTML5-compliant form), the title (which is always
  389. // there but possibly empty), and other meta tags.
  390. if (element.Name == W.document)
  391. {
  392. return new XElement(Xhtml.html,
  393. new XElement(Xhtml.head,
  394. new XElement(Xhtml.meta, new XAttribute("charset", "UTF-8")),
  395. settings.PageTitle != null
  396. ? new XElement(Xhtml.title, new XText(settings.PageTitle))
  397. : new XElement(Xhtml.title, new XText(string.Empty)),
  398. new XElement(Xhtml.meta,
  399. new XAttribute("name", "Generator"),
  400. new XAttribute("content", "PowerTools for Open XML"))),
  401. element.Elements()
  402. .Select(e => ConvertToHtmlTransform(wordDoc, settings, e, false, currentMarginLeft)));
  403. }
  404. // Transform the w:body element to the XHTML h:body element.
  405. if (element.Name == W.body)
  406. {
  407. return new XElement(Xhtml.body, CreateSectionDivs(wordDoc, settings, element));
  408. }
  409. // Transform the w:p element to the XHTML h:h1-h6 or h:p element (if the previous paragraph does not
  410. // have a style separator).
  411. if (element.Name == W.p)
  412. {
  413. return ProcessParagraph(wordDoc, settings, element, suppressTrailingWhiteSpace, currentMarginLeft);
  414. }
  415. // Transform hyperlinks to the XHTML h:a element.
  416. if (element.Name == W.hyperlink && element.Attribute(R.id) != null)
  417. {
  418. try
  419. {
  420. var a = new XElement(Xhtml.a,
  421. new XAttribute("href",
  422. wordDoc.MainDocumentPart
  423. .HyperlinkRelationships
  424. .First(x => x.Id == (string)element.Attribute(R.id))
  425. .Uri
  426. ),
  427. new XAttribute("target", "_blank"
  428. ),
  429. element.Elements(W.r).Select(run => ConvertRun(wordDoc, settings, run))
  430. );
  431. if (!a.Nodes().Any())
  432. a.Add(new XText(""));
  433. return a;
  434. }
  435. catch (UriFormatException)
  436. {
  437. return element.Elements().Select(e => ConvertToHtmlTransform(wordDoc, settings, e, false, currentMarginLeft));
  438. }
  439. }
  440. // Transform hyperlinks to bookmarks to the XHTML h:a element.
  441. if (element.Name == W.hyperlink && element.Attribute(W.anchor) != null)
  442. {
  443. return ProcessHyperlinkToBookmark(wordDoc, settings, element);
  444. }
  445. // Transform contents of runs.
  446. if (element.Name == W.r)
  447. {
  448. return ConvertRun(wordDoc, settings, element);
  449. }
  450. // Transform w:bookmarkStart into anchor
  451. if (element.Name == W.bookmarkStart)
  452. {
  453. return ProcessBookmarkStart(element);
  454. }
  455. // Transform every w:t element to a text node.
  456. if (element.Name == W.t)
  457. {
  458. // We don't need to convert characters to entities in a UTF-8 document.
  459. // Further, we don't need &nbsp; entities for significant whitespace
  460. // because we are wrapping the text nodes in <span> elements within
  461. // which all whitespace is significant.
  462. return new XText(element.Value);
  463. }
  464. // Transform symbols to spans
  465. if (element.Name == W.sym)
  466. {
  467. var cs = (string)element.Attribute(W._char);
  468. var c = Convert.ToInt32(cs, 16);
  469. return new XElement(Xhtml.span, new XEntity(string.Format("#{0}", c)));
  470. }
  471. // Transform tabs that have the pt:TabWidth attribute set
  472. if (element.Name == W.tab)
  473. {
  474. return ProcessTab(element);
  475. }
  476. // Transform w:br to h:br.
  477. if (element.Name == W.br || element.Name == W.cr)
  478. {
  479. return ProcessBreak(element);
  480. }
  481. // Transform w:noBreakHyphen to '-'
  482. if (element.Name == W.noBreakHyphen)
  483. {
  484. return new XText("-");
  485. }
  486. // Transform w:tbl to h:tbl.
  487. if (element.Name == W.tbl)
  488. {
  489. return ProcessTable(wordDoc, settings, element, currentMarginLeft);
  490. }
  491. // Transform w:tr to h:tr.
  492. if (element.Name == W.tr)
  493. {
  494. try
  495. {
  496. return ProcessTableRow(wordDoc, settings, element, currentMarginLeft);
  497. }
  498. catch (Exception ex)
  499. {
  500. throw new TransException(element, $"{element.Name}{ex.Message}", ex);
  501. }
  502. }
  503. // Transform w:tc to h:td.
  504. if (element.Name == W.tc)
  505. {
  506. try
  507. {
  508. return ProcessTableCell(wordDoc, settings, element);
  509. }
  510. catch (Exception ex)
  511. {
  512. throw new TransException(element, $"{element.Name}{ex.Message}", ex);
  513. }
  514. }
  515. // Transform images
  516. if (element.Name == W.drawing || element.Name == W.pict || element.Name == W._object)
  517. {
  518. try
  519. {
  520. XElement xElement = ProcessImage(wordDoc, element, settings.ImageHandler);
  521. return xElement;
  522. }
  523. catch (Exception ex)
  524. {
  525. throw new TransException(element, $"{element.Name}{ex.Message}", ex);
  526. }
  527. }
  528. // Transform content controls.
  529. if (element.Name == W.sdt)
  530. {
  531. try
  532. {
  533. return ProcessContentControl(wordDoc, settings, element, currentMarginLeft);
  534. }
  535. catch (Exception ex)
  536. {
  537. throw new TransException(element, $"{element.Name}{ex.Message}", ex);
  538. }
  539. }
  540. // Transform smart tags and simple fields.
  541. if (element.Name == W.smartTag || element.Name == W.fldSimple)
  542. {
  543. try
  544. {
  545. return CreateBorderDivs(wordDoc, settings, element.Elements());
  546. }
  547. catch (Exception ex)
  548. {
  549. throw new TransException(element, $"{element.Name}{ex.Message}", ex);
  550. }
  551. }
  552. //处理 数学公式
  553. if (element.Name == M.oMath)
  554. {
  555. try
  556. {
  557. return ProcessOMath(element);
  558. }
  559. catch (Exception ex)
  560. {
  561. throw new TransException(element, $"{element.Name}{ex.Message}", ex);
  562. }
  563. }
  564. if (element.Name == M.oMathPara)
  565. {
  566. try
  567. {
  568. return element.Elements().Select(e => ConvertToHtmlTransform(wordDoc, settings, e, false, currentMarginLeft));
  569. }
  570. catch (Exception ex)
  571. {
  572. throw new TransException(element, $"{element.Name}{ex.Message}", ex);
  573. }
  574. }
  575. //处理形状
  576. if (element.Name == MC.AlternateContent)
  577. {
  578. try
  579. {
  580. return ProcessAlternateContent(wordDoc, element, settings.ImageHandler);
  581. }
  582. catch (Exception ex)
  583. {
  584. throw new TransException(element, $"{element.Name}{ex.Message}", ex);
  585. }
  586. }
  587. //if (element.Name == MC.Choice)
  588. //{
  589. // return ProcessOMath(element);
  590. //}
  591. //if (element.Name == W.drawing &&element.Parent.Name==MC.Choice)
  592. //{
  593. // return ProcessOMath(element);
  594. //}
  595. // Ignore element.
  596. return null;
  597. }
  598. public static XElement ProcessAlternateContent(WordprocessingDocument wordDoc, XElement element, Func<ImageInfo, XElement> imageHandler)
  599. {
  600. var drawing = element.Element(MC.Choice).Element(W.drawing).Element(WP.inline);
  601. if (drawing == null)
  602. {
  603. drawing = element.Element(MC.Choice).Element(W.drawing).Element(WP.anchor);
  604. }
  605. var graphicData = drawing.Element(A.graphic).Element(A.graphicData).Elements();
  606. foreach (var elm in graphicData)
  607. {
  608. if (elm.Name == WPS.wsp)
  609. {
  610. (string html, HTEXLib.Shape shape) = ProcessWps(wordDoc, elm, imageHandler);
  611. XElement img;
  612. string mathxml = Globals.UnsupportedImage;
  613. var imageInfo = new ImageInfo()
  614. {
  615. ContentType = "image/x-emf",
  616. DrawingElement = element,
  617. AltText = "UnsupportedImageError",
  618. Mathxml = mathxml
  619. };
  620. double? widthInPoints = shape.style.position.cx;
  621. double? heightInPoints = shape.style.position.cy;
  622. img = imageHandler(imageInfo);
  623. if (widthInPoints != null && heightInPoints != null)
  624. {
  625. //imageInfo.ImgStyleAttribute = new XAttribute("style",
  626. // string.Format(NumberFormatInfo.InvariantInfo,
  627. // "width: {0}pt; height: {1}pt ; vertical-align: bottom;", widthInPoints, heightInPoints));
  628. img.SetAttributeValue("style", string.Format(NumberFormatInfo.InvariantInfo,
  629. "width: {0}pt; height: {1}pt ; vertical-align: bottom;", widthInPoints, heightInPoints));
  630. }
  631. else
  632. {
  633. //imageInfo.ImgStyleAttribute = new XAttribute("style",
  634. // string.Format(NumberFormatInfo.InvariantInfo,
  635. // "vertical-align: middle;"));
  636. img.SetAttributeValue("style", string.Format(NumberFormatInfo.InvariantInfo,
  637. "vertical-align: middle;"));
  638. }
  639. XElement elmt = XElement.Parse(html);
  640. return img;
  641. }
  642. else if (elm.Name == WPG.wgp)
  643. {
  644. return ProcessWgp(wordDoc, elm, imageHandler);
  645. }
  646. }
  647. return null;
  648. }
  649. public static XElement ProcessWgp(WordprocessingDocument wordDoc, XElement element, Func<ImageInfo, XElement> imageHandler)
  650. {
  651. var grpSpPr = element.Element(WPG.grpSpPr);
  652. var ext = grpSpPr.GetWordNodeByPath("a:xfrm/a:ext");
  653. var cx = ext.Attribute("cx").Value;
  654. var cy = ext.Attribute("cy").Value;
  655. //var elms= element.Descendants(WPS.wsp);
  656. //StringBuilder stringBuilder = new StringBuilder();
  657. //foreach (var elm in elms) {
  658. // var emlt= ProcessWps(wordDoc, elm, imageHandler);
  659. // stringBuilder.Append(emlt);
  660. //}
  661. //string svg = stringBuilder.ToString();
  662. //svg = $"<div>{svg}</div>";
  663. //XElement elmt = XElement.Parse(svg);
  664. XElement img;
  665. string mathxml = Globals.UnsupportedImage;
  666. var imageInfo = new ImageInfo()
  667. {
  668. ContentType = "image/x-emf",
  669. DrawingElement = element,
  670. AltText = "UnsupportedImageError",
  671. Mathxml = mathxml
  672. };
  673. double? widthInPoints = int.Parse(cx) * 96.0 / 914400;
  674. double? heightInPoints = int.Parse(cy) * 96.0 / 914400;
  675. img = imageHandler(imageInfo);
  676. if (widthInPoints != null && heightInPoints != null)
  677. {
  678. //imageInfo.ImgStyleAttribute = new XAttribute("style",
  679. // string.Format(NumberFormatInfo.InvariantInfo,
  680. // "width: {0}pt; height: {1}pt ; vertical-align: bottom;", widthInPoints, heightInPoints));
  681. img.SetAttributeValue("style", string.Format(NumberFormatInfo.InvariantInfo,
  682. "width: {0}pt; height: {1}pt ; vertical-align: bottom;", widthInPoints, heightInPoints));
  683. }
  684. else
  685. {
  686. //imageInfo.ImgStyleAttribute = new XAttribute("style",
  687. // string.Format(NumberFormatInfo.InvariantInfo,
  688. // "vertical-align: middle;"));
  689. img.SetAttributeValue("style", string.Format(NumberFormatInfo.InvariantInfo,
  690. "vertical-align: middle;"));
  691. }
  692. return img;
  693. }
  694. public static XElement ProcessChart(WordprocessingDocument wordDoc, XElement containerElement, Func<ImageInfo, XElement> imageHandler,
  695. int? extentCx, int? extentCy, string altText, XElement element, string hyperlinkUri)
  696. {
  697. var ird = element.Attribute(R.id);
  698. var chartird = wordDoc.MainDocumentPart.Parts.FirstOrDefault(pp => pp.RelationshipId == ird.Value);
  699. ChartPart chartPart = wordDoc.MainDocumentPart.ChartParts.Where(x => x.Uri.ToString() == chartird.OpenXmlPart.Uri.ToString()).FirstOrDefault<ChartPart>();
  700. if (chartPart != null)
  701. {
  702. ChartColorStylePart ChartColorStylePart = null;
  703. ChartStylePart ChartStylePart = null;
  704. foreach (var idp in chartPart.Parts)
  705. {
  706. if (idp.OpenXmlPart is ChartColorStylePart ChartColorStyleParts)
  707. {
  708. ChartColorStylePart = ChartColorStyleParts;
  709. }
  710. if (idp.OpenXmlPart is ChartStylePart ChartStyleParts)
  711. {
  712. ChartStylePart = ChartStyleParts;
  713. }
  714. }
  715. if (ChartStylePart != null)
  716. {
  717. var ChartStyleChildren = ChartStylePart.ChartStyle.ChildElements;
  718. foreach (var child in ChartStyleChildren)
  719. {
  720. if (child is DocumentFormat.OpenXml.Office2013.Drawing.ChartStyle.StyleEntry StyleEntry)
  721. {
  722. // DoStyleEntry(StyleEntry);
  723. }
  724. ///OfficeArtExtensionList cs: extLst
  725. // MarkerLayoutProperties cs: dataPointMarkerLayout
  726. }
  727. }
  728. HTEXLib.Chart charts = DrawChart(wordDoc, chartPart.ChartSpace, element, imageHandler, null);
  729. // charts.links = slide.hyperlinks;
  730. // return new List<Item> { charts };
  731. XElement img = null;
  732. string mathxml = Globals.UnsupportedImage;
  733. var imageInfo = new ImageInfo()
  734. {
  735. ContentType = "image/x-emf",
  736. DrawingElement = element,
  737. AltText = "UnsupportedImageError",
  738. Mathxml = mathxml
  739. };
  740. img = imageHandler(imageInfo);
  741. double? widthInPoints = extentCx * 96.0 / 914400;
  742. double? heightInPoints = extentCy * 96.0 / 914400;
  743. if (widthInPoints != null && heightInPoints != null)
  744. {
  745. //imageInfo.ImgStyleAttribute = new XAttribute("style",
  746. // string.Format(NumberFormatInfo.InvariantInfo,
  747. // "width: {0}pt; height: {1}pt ; vertical-align: bottom;", widthInPoints, heightInPoints));
  748. img.SetAttributeValue("style", string.Format(NumberFormatInfo.InvariantInfo,
  749. "width: {0}pt; height: {1}pt ; vertical-align: bottom;", widthInPoints, heightInPoints));
  750. }
  751. else
  752. {
  753. //imageInfo.ImgStyleAttribute = new XAttribute("style",
  754. // string.Format(NumberFormatInfo.InvariantInfo,
  755. // "vertical-align: middle;"));
  756. img.SetAttributeValue("style", string.Format(NumberFormatInfo.InvariantInfo,
  757. "vertical-align: middle;"));
  758. }
  759. return img;
  760. }
  761. return null;
  762. }
  763. public static HTEXLib.Chart DrawChart(WordprocessingDocument wordDoc, DocumentFormat.OpenXml.Drawing.Charts.ChartSpace chartSpace, XElement element, Func<ImageInfo, XElement> ImageHandler, HTEXLib.Position positionc)
  764. {
  765. var Chart = chartSpace.GetFirstChild<DocumentFormat.OpenXml.Drawing.Charts.Chart>();
  766. var ShapeProperties = chartSpace.GetFirstChild<DocumentFormat.OpenXml.Drawing.Charts.ShapeProperties>();
  767. HTEXLib.Models.HTEX.ShapeStyle shapeStyleChart = WORDHelper.DoShapeProperties(wordDoc, ShapeProperties, element, ImageHandler, positionc);
  768. var TextProperties = chartSpace.GetFirstChild<TextProperties>();
  769. var style = chartSpace.GetFirstChild<DocumentFormat.OpenXml.Drawing.Charts.Style>();
  770. var ColorMapOverride = chartSpace.GetFirstChild<DocumentFormat.OpenXml.Drawing.Charts.ColorMapOverride>();
  771. var style2010 = chartSpace.GetFirstChild<DocumentFormat.OpenXml.Office2010.Drawing.Charts.Style>();
  772. var UserShapesReference = chartSpace.GetFirstChild<UserShapesReference>();
  773. var charts = DoPlotArea(Chart.PlotArea, wordDoc, element, ImageHandler, positionc);
  774. HTEXLib.Models.HTEX.ShapeStyle shapeStyle = null;
  775. if (Chart.Title != null)
  776. {
  777. var ChartShapeProperties = Chart.Title.ChartShapeProperties;
  778. shapeStyle = WORDHelper.DoShapeProperties(wordDoc, ChartShapeProperties, element, ImageHandler, positionc);
  779. }
  780. // var (PPTParagraphs, tBody) = DoChartTitle(Chart.Title);
  781. //TextBody textBody = DrawText(PPTParagraphs, tBody);
  782. //Shape shape= new Shape { type = "Sp",uid ="Sp-" +ShaHashHelper.GetSHA1(slide.slideIndex + "-" + (Chart.Title.GetType().Name + Chart.Title.OuterXml).Trim().ToLower()), shapeType = "rect",
  783. // textBody = textBody , sid = sid + "titel" };
  784. HTEXLib.Shape shape = new HTEXLib.Shape
  785. {
  786. type = "Sp",
  787. uid = Guid.NewGuid().ToString(),
  788. shapeType = "rect",
  789. // textBody = textBody,
  790. // sid = sid + "titel"
  791. };
  792. shape.style.fill = shapeStyle != null ? shapeStyle.fill : null;
  793. shape.style.border = shapeStyle != null ? shapeStyle.border : null;
  794. //var chart= new Chart {sid=sid,charts =charts,title=shape,type=type,uid = type+"-"+ ShaHashHelper.GetSHA1(slide.slideIndex + "-" + (Chart.GetType().Name + Chart.OuterXml).Trim().ToLower()) };
  795. var chart = new HTEXLib.Chart
  796. {
  797. // sid = sid,
  798. charts = charts,
  799. title = shape,
  800. // type = type,
  801. uid = Guid.NewGuid().ToString(),
  802. };
  803. HTEXLib.Position position = new HTEXLib.Position { /*cx = width, cy = height, x = left, y = top, rot = rot */};
  804. chart.style.position = position;
  805. chart.style.fill = shapeStyleChart != null ? shapeStyleChart.fill : null;
  806. chart.style.border = shapeStyleChart != null ? shapeStyleChart.border : null;
  807. chart.style.effect = shapeStyleChart != null ? shapeStyleChart.effect : null;
  808. chart.svg = new HTEXLib.Svg
  809. {
  810. type = "path",
  811. d = "M" + 0 + " " + 0 + ",L" + position.cx + " " + 0 + ",L" + position.cx + " " + position.cy + ",L" + 0 + " " + position.cy + " z",
  812. close = true
  813. //new Rect {
  814. // x=position.x,
  815. // y = position.y,
  816. // width=position.cx,
  817. // height=position.cy,
  818. // type="rect"
  819. // }
  820. };
  821. if (chart.style != null)
  822. {
  823. if (chart.style.fill.type == -1 || chart.style.fill.type == 0)
  824. {
  825. chart.style.fill = null;
  826. }
  827. if (chart.style.border != null && chart.style.border.type == "none")
  828. {
  829. chart.style.border = null;
  830. }
  831. }
  832. return chart;
  833. }
  834. public static List<HTEXLib.CommonChart> DoPlotArea(PlotArea plotArea, WordprocessingDocument wordDoc, XElement element, Func<ImageInfo, XElement> ImageHandler, HTEXLib.Position positionc)
  835. {
  836. List<HTEXLib.CommonChart> charts = new List<HTEXLib.CommonChart>();
  837. foreach (var child in plotArea.ChildElements)
  838. {
  839. string key = child.LocalName;
  840. IEnumerable<XElement> serNodes = null;
  841. switch (key)
  842. {
  843. //break块中不可以随意更换,此条件用于归类不同从Chart
  844. case "pieChart":
  845. case "ofPieChart":
  846. case "pie3DChart":
  847. case "doughnutChart":
  848. HTEXLib.PieChart pieChart = new HTEXLib.PieChart { chartType = "pie" };
  849. if (key.Equals("pie3DChart"))
  850. {
  851. pieChart.is3D = true;
  852. }
  853. pieChart.pieType = key;
  854. if (key.Equals("ofPieChart"))
  855. {
  856. var ofPieType = HTEXLib.ShapeHelper.GetPPTXNodeByPath(child, "c:ofPieType");
  857. if (ofPieType != null)
  858. {
  859. //ofPieChart-pie ofPieChart-bar 子母饼图
  860. pieChart.pieType += "-" + ofPieType.Attribute("val").Value;
  861. }
  862. }
  863. serNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(child, "c:ser");
  864. pieChart.datas = ExtractChartData(serNodes, wordDoc, element, ImageHandler, positionc);
  865. charts.Add(pieChart);
  866. break;
  867. case "lineChart":
  868. case "line3DChart":
  869. HTEXLib.LineChart lineChart = new HTEXLib.LineChart { chartType = "line" };
  870. if (key.Equals("line3DChart"))
  871. {
  872. lineChart.is3D = true;
  873. }
  874. lineChart.lineType = key;
  875. var LineGrouping = HTEXLib.ShapeHelper.GetPPTXNodeByPath(child, "c:grouping");
  876. if (LineGrouping != null)
  877. {
  878. //standard stacked percentStacked
  879. lineChart.lineType += "-" + LineGrouping.Attribute("val").Value;
  880. }
  881. serNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(child, "c:ser");
  882. lineChart.datas = ExtractChartData(serNodes, wordDoc, element, ImageHandler, positionc);
  883. charts.Add(lineChart);
  884. break;
  885. case "barChart":
  886. case "bar3DChart":
  887. var barDir = HTEXLib.ShapeHelper.GetPPTXNodeByPath(child, "c:barDir");
  888. if (barDir != null)
  889. {
  890. if (barDir.Attribute("val").Value.Equals("bar"))
  891. {
  892. HTEXLib.BarChart barChart = new HTEXLib.BarChart { chartType = "bar" };
  893. charts.Add(barChart);
  894. if (key.Equals("bar3DChart"))
  895. {
  896. barChart.is3D = true;
  897. }
  898. barChart.barType = key;
  899. var BarGrouping = HTEXLib.ShapeHelper.GetPPTXNodeByPath(child, "c:grouping");
  900. if (BarGrouping != null)
  901. {
  902. //standard stacked percentStacked
  903. barChart.barType += "-" + BarGrouping.Attribute("val").Value;
  904. }
  905. serNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(child, "c:ser");
  906. barChart.datas = ExtractChartData(serNodes, wordDoc, element, ImageHandler, positionc);
  907. charts.Add(barChart);
  908. }
  909. else if (barDir.Attribute("val").Value.Equals("col"))
  910. {
  911. HTEXLib.ColChart colChart = new HTEXLib.ColChart { chartType = "col" };
  912. if (key.Equals("bar3DChart"))
  913. {
  914. colChart.is3D = true;
  915. }
  916. colChart.colType = key.Replace("bar", "col");
  917. var ColGrouping = HTEXLib.ShapeHelper.GetPPTXNodeByPath(child, "c:grouping");
  918. if (ColGrouping != null)
  919. {
  920. //standard stacked percentStacked
  921. colChart.colType += "-" + ColGrouping.Attribute("val").Value;
  922. }
  923. serNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(child, "c:ser");
  924. colChart.datas = ExtractChartData(serNodes, wordDoc, element, ImageHandler, positionc);
  925. charts.Add(colChart);
  926. }
  927. }
  928. break;
  929. case "areaChart":
  930. case "area3DChart":
  931. HTEXLib.AreaChart areaChart = new HTEXLib.AreaChart { chartType = "area" };
  932. if (key.Equals("area3DChart"))
  933. {
  934. areaChart.is3D = true;
  935. }
  936. areaChart.areaType = key;
  937. var AreaGrouping = HTEXLib.ShapeHelper.GetPPTXNodeByPath(child, "c:grouping");
  938. if (AreaGrouping != null)
  939. {
  940. //standard stacked percentStacked
  941. areaChart.areaType += "-" + AreaGrouping.Attribute("val").Value;
  942. }
  943. serNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(child, "c:ser");
  944. areaChart.datas = ExtractChartData(serNodes, wordDoc, element, ImageHandler, positionc);
  945. charts.Add(areaChart);
  946. break;
  947. case "scatterChart":
  948. case "bubbleChart":
  949. HTEXLib.ScatterChart scatterChart = new HTEXLib.ScatterChart { chartType = "scatter" };
  950. scatterChart.scatterType = key;
  951. if (key.Equals("scatterChart"))
  952. {
  953. var ScatterStyle = HTEXLib.ShapeHelper.GetPPTXNodeByPath(child, "c:scatterStyle");
  954. if (ScatterStyle != null)
  955. {
  956. scatterChart.scatterType += "-" + ScatterStyle.Attribute("val").Value.Replace("Marker", "");
  957. }
  958. }
  959. serNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(child, "c:ser");
  960. scatterChart.datas = ExtractChartData(serNodes, wordDoc, element, ImageHandler, positionc);
  961. charts.Add(scatterChart);
  962. break;
  963. case "radarChart":
  964. HTEXLib.RadarChart radarChart = new HTEXLib.RadarChart { chartType = "radar" };
  965. radarChart.radarType = key;
  966. var RadarStyle = HTEXLib.ShapeHelper.GetPPTXNodeByPath(child, "c:radarStyle");
  967. if (RadarStyle != null)
  968. {
  969. radarChart.radarType += "-" + RadarStyle.Attribute("val").Value;
  970. }
  971. serNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(child, "c:ser");
  972. radarChart.datas = ExtractChartData(serNodes, wordDoc, element, ImageHandler, positionc);
  973. charts.Add(radarChart);
  974. break;
  975. case "plotAreaRegion":
  976. HTEXLib.PlotAreaChart plotAreaChart = new HTEXLib.PlotAreaChart { chartType = "plotArea" };
  977. plotAreaChart.plotAreaType = key;
  978. var PlotSeries = HTEXLib.ShapeHelper.GetPPTXNodeByPath(child, "cx:series");
  979. if (PlotSeries != null)
  980. {
  981. plotAreaChart.plotAreaType += "-" + PlotSeries.Attribute("val").Value;
  982. }
  983. serNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(child, "c:ser");
  984. plotAreaChart.datas = ExtractChartData(serNodes, wordDoc, element, ImageHandler, positionc);
  985. charts.Add(plotAreaChart);
  986. break;
  987. case "stockChart":
  988. HTEXLib.StockChart stockChart = new HTEXLib.StockChart { chartType = "stock" };
  989. stockChart.stockType = key;
  990. serNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(child, "c:ser");
  991. stockChart.datas = ExtractChartData(serNodes, wordDoc, element, ImageHandler, positionc);
  992. charts.Add(stockChart);
  993. break;
  994. case "surfaceChart":
  995. case "surface3DChart":
  996. HTEXLib.SurfaceChart surfaceChart = new HTEXLib.SurfaceChart { chartType = "surface" };
  997. if (key.Equals("surface3DChart"))
  998. {
  999. surfaceChart.is3D = true;
  1000. }
  1001. surfaceChart.surfaceType = key;
  1002. var Wireframe = HTEXLib.ShapeHelper.GetPPTXNodeByPath(child, "c:wireframe");
  1003. if (Wireframe != null)
  1004. {
  1005. surfaceChart.surfaceType += "-" + Wireframe.Attribute("val").Value;
  1006. }
  1007. serNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(child, "c:ser");
  1008. surfaceChart.datas = ExtractChartData(serNodes, wordDoc, element, ImageHandler, positionc);
  1009. charts.Add(surfaceChart);
  1010. break;
  1011. }
  1012. }
  1013. return charts;
  1014. }
  1015. public static List<Dictionary<string, object>> ExtractChartData(IEnumerable<XElement> nodes, WordprocessingDocument wordDoc, XElement element, Func<ImageInfo, XElement> ImageHandler, HTEXLib.Position positionc)
  1016. {
  1017. if (nodes != null)
  1018. {
  1019. List<Dictionary<string, object>> listDict = new List<Dictionary<string, object>>();
  1020. foreach (XElement node in nodes)
  1021. {
  1022. if (HTEXLib.ShapeHelper.GetPPTXNodeByPath(node, "c:xVal") != null)
  1023. {
  1024. Dictionary<string, object> dict = new Dictionary<string, object>();
  1025. var xCvNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(node, "c:xVal/c:numRef/c:numCache/c:pt/c:v");
  1026. if (xCvNodes != null)
  1027. {
  1028. List<string> list = new List<string>();
  1029. foreach (XElement cvNode in xCvNodes)
  1030. {
  1031. list.Add(cvNode.Value);
  1032. }
  1033. dict.Add("xAxis", list);
  1034. }
  1035. var yCvNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(node, "c:yVal/c:numRef/c:numCache/c:pt/c:v");
  1036. if (yCvNodes != null)
  1037. {
  1038. List<string> list = new List<string>();
  1039. foreach (XElement cvNode in yCvNodes)
  1040. {
  1041. list.Add(cvNode.Value);
  1042. }
  1043. dict.Add("yAxis", list);
  1044. }
  1045. dict.Add("colName", "-");
  1046. listDict.Add(dict);
  1047. }
  1048. else
  1049. {
  1050. Dictionary<string, object> dict = new Dictionary<string, object>();
  1051. var spPr = HTEXLib.ShapeHelper.GetPPTXNodeByPath(node, "c:spPr");
  1052. if (spPr != null)
  1053. {
  1054. var ChartShapeProperties = new ChartShapeProperties(spPr.ToString());
  1055. HTEXLib.Models.HTEX.ShapeStyle shapeStyle = WORDHelper.DoShapeProperties(wordDoc, ChartShapeProperties, element, ImageHandler, positionc);
  1056. dict.Add("colStyle", shapeStyle);
  1057. }
  1058. else
  1059. {
  1060. dict.Add("colStyle", null);
  1061. }
  1062. var colNameNode = HTEXLib.ShapeHelper.GetPPTXNodeByPath(node, "c:tx/c:strRef/c:strCache/c:pt/c:v");
  1063. if (colNameNode != null)
  1064. {
  1065. dict.Add("colName", colNameNode.Value);
  1066. }
  1067. //name
  1068. var catNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(node, "c:cat/c:strRef/c:strCache/c:pt/c:v");
  1069. if (catNodes == null)
  1070. {
  1071. catNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(node, "c:cat/c:numRef/c:numCache/c:pt/c:v");
  1072. }
  1073. if (catNodes != null)
  1074. {
  1075. List<string> list = new List<string>();
  1076. foreach (XElement cvNode in catNodes)
  1077. {
  1078. list.Add(cvNode.Value);
  1079. }
  1080. dict.Add("xAxis", list);
  1081. }
  1082. //value
  1083. var valNodes = HTEXLib.ShapeHelper.GetPPTXNodeListByPath(node, "c:val/c:numRef/c:numCache/c:pt/c:v");
  1084. if (valNodes != null)
  1085. {
  1086. List<string> list = new List<string>();
  1087. foreach (XElement cvNode in valNodes)
  1088. {
  1089. list.Add(cvNode.Value);
  1090. }
  1091. dict.Add("yAxis", list);
  1092. }
  1093. listDict.Add(dict);
  1094. }
  1095. }
  1096. return listDict;
  1097. }
  1098. return null;
  1099. }
  1100. public static XElement ProcessDgm(WordprocessingDocument wordDoc, XElement containerElement, Func<ImageInfo, XElement> imageHandler,
  1101. int? extentCx, int? extentCy, string altText, XElement element, string hyperlinkUri)
  1102. {
  1103. var rdm = element.Attribute(R.dm);
  1104. var dgm = wordDoc.MainDocumentPart.Parts.FirstOrDefault(pp => pp.RelationshipId == rdm.Value);
  1105. XElement img = null;
  1106. string mathxml = Globals.UnsupportedImage;
  1107. var imageInfo = new ImageInfo()
  1108. {
  1109. ContentType = "image/x-emf",
  1110. DrawingElement = element,
  1111. AltText = "UnsupportedImageError",
  1112. Mathxml = mathxml
  1113. };
  1114. img = imageHandler(imageInfo);
  1115. double? widthInPoints = extentCx * 96.0 / 914400;
  1116. double? heightInPoints = extentCy * 96.0 / 914400;
  1117. if (widthInPoints != null && heightInPoints != null)
  1118. {
  1119. //imageInfo.ImgStyleAttribute = new XAttribute("style",
  1120. // string.Format(NumberFormatInfo.InvariantInfo,
  1121. // "width: {0}pt; height: {1}pt ; vertical-align: bottom;", widthInPoints, heightInPoints));
  1122. img.SetAttributeValue("style", string.Format(NumberFormatInfo.InvariantInfo,
  1123. "width: {0}pt; height: {1}pt ; vertical-align: bottom;", widthInPoints, heightInPoints));
  1124. }
  1125. else
  1126. {
  1127. //imageInfo.ImgStyleAttribute = new XAttribute("style",
  1128. // string.Format(NumberFormatInfo.InvariantInfo,
  1129. // "vertical-align: middle;"));
  1130. img.SetAttributeValue("style", string.Format(NumberFormatInfo.InvariantInfo,
  1131. "vertical-align: middle;"));
  1132. }
  1133. return img;
  1134. }
  1135. public static (string html, HTEXLib.Shape shape) ProcessWps(WordprocessingDocument wordDoc, XElement element, Func<ImageInfo, XElement> imageHandler)
  1136. {
  1137. // var wps = element.Element(MC.Choice).Element(W.drawing).Element(WP.anchor).Element(A.graphic).Element(A.graphicData).Element(WPS.wsp);
  1138. WordprocessingShape wsp = new WordprocessingShape(element.ToString());
  1139. var ext = element.GetWordNodeByPath("wps:spPr/a:xfrm/a:ext");
  1140. var cx = ext.Attribute("cx").Value;
  1141. var cy = ext.Attribute("cy").Value;
  1142. HTEXLib.Shape shape = new HTEXLib.Shape();
  1143. HTEXLib.Position position = new HTEXLib.Position { x = 0, y = 0, cx = double.Parse(cx) * 96.0 / 914400, cy = double.Parse(cy) * 96.0 / 914400 };
  1144. shape.style.position = position;
  1145. var ShapeStyle = WORDHelper.DoShapeProperties(wordDoc, wsp.GetFirstChild<DocumentFormat.OpenXml.Office2010.Word.DrawingShape.ShapeProperties>(), element, imageHandler, position);
  1146. SlideColor slideColor = WORDHelper.DoShapeStyle(wsp.GetFirstChild<DocumentFormat.OpenXml.Office2010.Word.DrawingShape.ShapeStyle>(), wordDoc);
  1147. if (ShapeStyle.border != null)
  1148. {
  1149. shape.style.border = ShapeStyle.border;
  1150. }
  1151. //从ShapeProperties 获取 p:spPr
  1152. //再从 ShapeStyle 获取 p:spPr
  1153. if (ShapeStyle.border == null || ShapeStyle.border.color == null || ShapeStyle.border.color.type == -1)
  1154. {
  1155. if (slideColor != null)
  1156. {
  1157. shape.style.border.color.solidFill = slideColor.LineColor;
  1158. shape.style.border.color.type = 2;
  1159. shape.style.border.outline = ShapeStyle.border != null && ShapeStyle.border.outline != null ? ShapeStyle.border.outline : null;
  1160. }
  1161. }
  1162. else
  1163. {
  1164. shape.style.border = ShapeStyle.border;
  1165. shape.style.border.outline = ShapeStyle.border != null && ShapeStyle.border.outline != null ? ShapeStyle.border.outline : ShapeStyle.border.outline != null ? ShapeStyle.border.outline : null;
  1166. }
  1167. if (ShapeStyle.fill == null || ShapeStyle.fill.type == -1)
  1168. {
  1169. if (slideColor != null)
  1170. {
  1171. shape.style.fill.solidFill = slideColor.FillColor;
  1172. shape.style.fill.type = 2;
  1173. }
  1174. }
  1175. else
  1176. {
  1177. shape.style.fill = ShapeStyle.fill;
  1178. }
  1179. var wpars = element.GetWordNodeListByPath("wps:txbx/w:txbxContent/w:p");
  1180. List<string> pars = new List<string>();
  1181. if (wpars != null)
  1182. {
  1183. foreach (var wpar in wpars)
  1184. {
  1185. if (wpar.Value != null)
  1186. {
  1187. pars.Add(wpar.Value);
  1188. }
  1189. }
  1190. }
  1191. var prstGeom = element.GetWordNodeByPath("wps:spPr/a:prstGeom");
  1192. if (prstGeom != null)
  1193. {
  1194. var shapeType = prstGeom.Attribute("prst").Value;
  1195. var svg = SvgHelper.GenShapeSvg(new PresetGeometry(prstGeom.ToString()), shapeType, position, null);
  1196. shape.svg = svg;
  1197. }
  1198. else
  1199. {
  1200. var custGeom = element.GetWordNodeByPath("wps:spPr/a:custGeom");
  1201. if (custGeom != null)
  1202. {
  1203. DocumentFormat.OpenXml.Drawing.CustomGeometry shapeTypeCustom = new CustomGeometry(custGeom.ToString());
  1204. shape = PPTXHelper.DoCustomGeometry(shapeTypeCustom, shape);
  1205. }
  1206. }
  1207. return (DrawingShape(shape, pars), shape);
  1208. }
  1209. public static string DrawingShape(HTEXLib.Shape shape, List<string> pars)
  1210. {
  1211. StringBuilder builder = new StringBuilder();
  1212. if (shape.svg != null)
  1213. {
  1214. builder.Append($"<svg xmlns='http://www.w3.org/2000/svg' style='width:{shape.style.position.cx}px;height:{shape.style.position.cy}px;rotate({shape.style.position.rot}deg); overflow: visible;'>");
  1215. string bdc = shape.style.border != null && shape.style.border.color != null && shape.style.border.color.solidFill != null ? shape.style.border.color.solidFill : "none";
  1216. string bds = shape.style.border != null && shape.style.border.stroke != null ? shape.style.border.stroke : "0";
  1217. double? bdw = shape.style.border != null && shape.style.border.outline != null && shape.style.border.outline.width != null ? shape.style.border.outline.width : 0;
  1218. string bgc = shape.style.fill != null && shape.style.fill != null && shape.style.fill.solidFill != null ? shape.style.fill.solidFill : "none";
  1219. if (shape.style.border != null && (shape.style.border.tailEnd != null || shape.style.border.headEnd != null))
  1220. {
  1221. builder.Append("<defs xmlns='http://www.w3.org/2000/svg'>");
  1222. builder.Append($"<marker id='marker{shape.uid}' viewBox='0 0 20 20' refX='1' refY='10' markerWidth='10' markerHeight='20' stroke='{bdc}' fill='{bdc}' orient='auto-start-reverse' markerUnits='strokeWidth'><path d='M 0 0 L 20 10 L 0 20 z' /></marker>");
  1223. builder.Append("</defs>");
  1224. }
  1225. string stm = "";
  1226. string edm = "";
  1227. if (shape.style.border != null && shape.style.border.headEnd != null)
  1228. {
  1229. stm = $" marker-start='url(#marker{ shape.uid})'";
  1230. }
  1231. if (shape.style.border != null && shape.style.border.tailEnd != null)
  1232. {
  1233. edm = $" marker-end='url(#marker{ shape.uid})'";
  1234. }
  1235. builder.Append($"<path xmlns='http://www.w3.org/2000/svg' d='{shape.svg.d}' fill='{bgc}' stroke='{bdc}' stroke-width='{bdw}' stroke-dasharray='{bds}' {stm} {edm} />");
  1236. if (pars.IsNotEmpty())
  1237. {
  1238. builder.Append($"<text x='{shape.style.position.cx / 2}' y='{shape.style.position.cy}' fill='#000000'>{string.Join("", pars)}</text>");
  1239. }
  1240. builder.Append("</svg>");
  1241. }
  1242. return builder.ToString();
  1243. }
  1244. public static XElement ProcessOMath(XElement element)
  1245. {
  1246. //XslCompiledTransform xslTransform = new XslCompiledTransform();
  1247. // xslTransform.Load(BaseConfigModel.ContentRootPath + "/Config/Core/OMML2MML.XSL");
  1248. XmlReader OMML2MML = XmlReader.Create(new StringReader(Globals.OMML2MML));
  1249. XslCompiledTransform xslTransform = new XslCompiledTransform();
  1250. xslTransform.Load(OMML2MML);
  1251. string mathXml = element.ToString();
  1252. string officeML = string.Empty;
  1253. using (TextReader tr = new StringReader(mathXml))
  1254. {
  1255. using (XmlReader reader = XmlReader.Create(tr))
  1256. {
  1257. using (MemoryStream ms = new MemoryStream())
  1258. {
  1259. XmlWriterSettings settings = xslTransform.OutputSettings.Clone();
  1260. settings.ConformanceLevel = ConformanceLevel.Fragment;
  1261. settings.OmitXmlDeclaration = true;
  1262. XmlWriter xw = XmlWriter.Create(ms, settings);
  1263. xslTransform.Transform(reader, xw);
  1264. ms.Seek(0, SeekOrigin.Begin);
  1265. using (StreamReader sr = new StreamReader(ms, Encoding.UTF8))
  1266. {
  1267. officeML = sr.ReadToEnd();
  1268. // Console.Out.WriteLine(officeML);
  1269. }
  1270. }
  1271. }
  1272. }
  1273. officeML = officeML.Replace("mml:", "");
  1274. XElement officeElement = XElement.Load(new StringReader(officeML));
  1275. return officeElement;
  1276. }
  1277. private static object ProcessHyperlinkToBookmark(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings, XElement element)
  1278. {
  1279. var style = new Dictionary<string, string>();
  1280. var a = new XElement(Xhtml.a,
  1281. new XAttribute("href", "#" + (string)element.Attribute(W.anchor)),
  1282. element.Elements(W.r).Select(run => ConvertRun(wordDoc, settings, run)));
  1283. if (!a.Nodes().Any())
  1284. a.Add(new XText(""));
  1285. style.Add("text-decoration", "none");
  1286. a.AddAnnotation(style);
  1287. return a;
  1288. }
  1289. private static object ProcessBookmarkStart(XElement element)
  1290. {
  1291. var name = (string)element.Attribute(W.name);
  1292. if (name == null) return null;
  1293. var style = new Dictionary<string, string>();
  1294. var a = new XElement(Xhtml.a,
  1295. new XAttribute("id", name),
  1296. new XText(""));
  1297. if (!a.Nodes().Any())
  1298. a.Add(new XText(""));
  1299. style.Add("text-decoration", "none");
  1300. a.AddAnnotation(style);
  1301. return a;
  1302. }
  1303. private static object ProcessTab(XElement element)
  1304. {
  1305. var tabWidthAtt = element.Attribute(PtOpenXml.TabWidth);
  1306. if (tabWidthAtt == null) return null;
  1307. var leader = (string)element.Attribute(PtOpenXml.Leader);
  1308. var tabWidth = (decimal)tabWidthAtt;
  1309. var style = new Dictionary<string, string>();
  1310. XElement span;
  1311. if (leader != null)
  1312. {
  1313. var leaderChar = ".";
  1314. if (leader == "hyphen")
  1315. leaderChar = "-";
  1316. else if (leader == "dot")
  1317. leaderChar = ".";
  1318. else if (leader == "underscore")
  1319. leaderChar = "_";
  1320. var runContainingTabToReplace = element.Ancestors(W.r).First();
  1321. var fontNameAtt = runContainingTabToReplace.Attribute(PtOpenXml.pt + "FontName") ??
  1322. runContainingTabToReplace.Ancestors(W.p).First().Attribute(PtOpenXml.pt + "FontName");
  1323. var dummyRun = new XElement(W.r,
  1324. fontNameAtt,
  1325. runContainingTabToReplace.Elements(W.rPr),
  1326. new XElement(W.t, leaderChar));
  1327. var widthOfLeaderChar = CalcWidthOfRunInTwips(dummyRun);
  1328. bool forceArial = false;
  1329. if (widthOfLeaderChar == 0)
  1330. {
  1331. dummyRun = new XElement(W.r,
  1332. new XAttribute(PtOpenXml.FontName, "Arial"),
  1333. runContainingTabToReplace.Elements(W.rPr),
  1334. new XElement(W.t, leaderChar));
  1335. widthOfLeaderChar = CalcWidthOfRunInTwips(dummyRun);
  1336. forceArial = true;
  1337. }
  1338. if (widthOfLeaderChar != 0)
  1339. {
  1340. var numberOfLeaderChars = (int)(Math.Floor((tabWidth * 1440) / widthOfLeaderChar));
  1341. if (numberOfLeaderChars < 0)
  1342. numberOfLeaderChars = 0;
  1343. span = new XElement(Xhtml.span,
  1344. new XAttribute(XNamespace.Xml + "space", "preserve"),
  1345. " " + "".PadRight(numberOfLeaderChars, leaderChar[0]) + " ");
  1346. style.Add("margin", "0 0 0 0");
  1347. style.Add("padding", "0 0 0 0");
  1348. style.Add("width", string.Format(NumberFormatInfo.InvariantInfo, "{0:0.00}in", tabWidth));
  1349. style.Add("text-align", "center");
  1350. if (forceArial)
  1351. style.Add("font-family", "Arial");
  1352. }
  1353. else
  1354. {
  1355. span = new XElement(Xhtml.span, new XAttribute(XNamespace.Xml + "space", "preserve"), " ");
  1356. style.Add("margin", "0 0 0 0");
  1357. style.Add("padding", "0 0 0 0");
  1358. style.Add("width", string.Format(NumberFormatInfo.InvariantInfo, "{0:0.00}in", tabWidth));
  1359. style.Add("text-align", "center");
  1360. if (leader == "underscore")
  1361. {
  1362. style.Add("text-decoration", "underline");
  1363. }
  1364. }
  1365. }
  1366. else
  1367. {
  1368. #if false
  1369. var bidi = element
  1370. .Ancestors(W.p)
  1371. .Take(1)
  1372. .Elements(W.pPr)
  1373. .Elements(W.bidi)
  1374. .Where(b => b.Attribute(W.val) == null || b.Attribute(W.val).ToBoolean() == true)
  1375. .FirstOrDefault();
  1376. var isBidi = bidi != null;
  1377. if (isBidi)
  1378. span = new XElement(Xhtml.span, new XEntity("#x200f")); // RLM
  1379. else
  1380. span = new XElement(Xhtml.span, new XEntity("#x200e")); // LRM
  1381. #else
  1382. span = new XElement(Xhtml.span, new XEntity("#x00a0"));
  1383. #endif
  1384. style.Add("margin", string.Format(NumberFormatInfo.InvariantInfo, "0 0 0 {0:0.00}in", tabWidth));
  1385. style.Add("padding", "0 0 0 0");
  1386. }
  1387. span.AddAnnotation(style);
  1388. return span;
  1389. }
  1390. private static object ProcessBreak(XElement element)
  1391. {
  1392. XElement span = null;
  1393. var tabWidth = (decimal?)element.Attribute(PtOpenXml.TabWidth);
  1394. if (tabWidth != null)
  1395. {
  1396. span = new XElement(Xhtml.span);
  1397. span.AddAnnotation(new Dictionary<string, string>
  1398. {
  1399. { "margin", string.Format(NumberFormatInfo.InvariantInfo, "0 0 0 {0:0.00}in", tabWidth) },
  1400. { "padding", "0 0 0 0" }
  1401. });
  1402. }
  1403. var paragraph = element.Ancestors(W.p).FirstOrDefault();
  1404. var isBidi = paragraph != null &&
  1405. paragraph.Elements(W.pPr).Elements(W.bidi).Any(b => b.Attribute(W.val) == null ||
  1406. b.Attribute(W.val).ToBoolean() == true);
  1407. var zeroWidthChar = isBidi ? new XEntity("#x200f") : new XEntity("#x200e");
  1408. return new object[]
  1409. {
  1410. new XElement(Xhtml.br),
  1411. zeroWidthChar,
  1412. span,
  1413. };
  1414. }
  1415. private static object ProcessContentControl(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings,
  1416. XElement element, decimal currentMarginLeft)
  1417. {
  1418. var relevantAncestors = element.Ancestors().TakeWhile(a => a.Name != W.txbxContent);
  1419. var isRunLevelContentControl = relevantAncestors.Any(a => a.Name == W.p);
  1420. if (isRunLevelContentControl)
  1421. {
  1422. return element.Elements(W.sdtContent).Elements()
  1423. .Select(e => ConvertToHtmlTransform(wordDoc, settings, e, false, currentMarginLeft))
  1424. .ToList();
  1425. }
  1426. return CreateBorderDivs(wordDoc, settings, element.Elements(W.sdtContent).Elements());
  1427. }
  1428. // Transform the w:p element, including the following sibling w:p element(s)
  1429. // in case the w:p element has a style separator. The sibling(s) will be
  1430. // transformed to h:span elements rather than h:p elements and added to
  1431. // the element (e.g., h:h2) created from the w:p element having the (first)
  1432. // style separator (i.e., a w:specVanish element).
  1433. private static object ProcessParagraph(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings,
  1434. XElement element, bool suppressTrailingWhiteSpace, decimal currentMarginLeft)
  1435. {
  1436. // Ignore this paragraph if the previous paragraph has a style separator.
  1437. // We have already transformed this one together with the previous one.
  1438. var previousParagraph = element.ElementsBeforeSelf(W.p).LastOrDefault();
  1439. if (HasStyleSeparator(previousParagraph)) return null;
  1440. var elementName = GetParagraphElementName(element, wordDoc);
  1441. var isBidi = IsBidi(element);
  1442. var paragraph = (XElement)ConvertParagraph(wordDoc, settings, element, elementName,
  1443. suppressTrailingWhiteSpace, currentMarginLeft, isBidi);
  1444. // The paragraph conversion might have created empty spans.
  1445. // These can and should be removed because empty spans are
  1446. // invalid in HTML5.
  1447. paragraph.Elements(Xhtml.span).Where(e => e.IsEmpty).Remove();
  1448. foreach (var span in paragraph.Elements(Xhtml.span).ToList())
  1449. {
  1450. var v = span.Value;
  1451. if (v.Length > 0 && (char.IsWhiteSpace(v[0]) || char.IsWhiteSpace(v[v.Length - 1])) && span.Attribute(XNamespace.Xml + "space") == null)
  1452. span.Add(new XAttribute(XNamespace.Xml + "space", "preserve"));
  1453. }
  1454. while (HasStyleSeparator(element))
  1455. {
  1456. element = element.ElementsAfterSelf(W.p).FirstOrDefault();
  1457. if (element == null) break;
  1458. elementName = Xhtml.span;
  1459. isBidi = IsBidi(element);
  1460. var span = (XElement)ConvertParagraph(wordDoc, settings, element, elementName,
  1461. suppressTrailingWhiteSpace, currentMarginLeft, isBidi);
  1462. var v = span.Value;
  1463. if (v.Length > 0 && (char.IsWhiteSpace(v[0]) || char.IsWhiteSpace(v[v.Length - 1])) && span.Attribute(XNamespace.Xml + "space") == null)
  1464. span.Add(new XAttribute(XNamespace.Xml + "space", "preserve"));
  1465. paragraph.Add(span);
  1466. }
  1467. return paragraph;
  1468. }
  1469. private static object ProcessTable(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings, XElement element, decimal currentMarginLeft)
  1470. {
  1471. var style = new Dictionary<string, string>();
  1472. style.AddIfMissing("border-collapse", "collapse");
  1473. style.AddIfMissing("border", "none");
  1474. var bidiVisual = element.Elements(W.tblPr).Elements(W.bidiVisual).FirstOrDefault();
  1475. var tblW = element.Elements(W.tblPr).Elements(W.tblW).FirstOrDefault();
  1476. if (tblW != null)
  1477. {
  1478. var type = (string)tblW.Attribute(W.type);
  1479. if (type != null && type == "pct")
  1480. {
  1481. var w = (int)tblW.Attribute(W._w);
  1482. style.AddIfMissing("width", (w / 50) + "%");
  1483. }
  1484. }
  1485. var tblInd = element.Elements(W.tblPr).Elements(W.tblInd).FirstOrDefault();
  1486. if (tblInd != null)
  1487. {
  1488. var tblIndType = (string)tblInd.Attribute(W.type);
  1489. if (tblIndType != null)
  1490. {
  1491. if (tblIndType == "dxa")
  1492. {
  1493. var width = (decimal?)tblInd.Attribute(W._w);
  1494. if (width != null)
  1495. {
  1496. style.AddIfMissing("margin-left",
  1497. width > 0m
  1498. ? string.Format(NumberFormatInfo.InvariantInfo, "{0}pt", width / 20m)
  1499. : "0");
  1500. }
  1501. }
  1502. }
  1503. }
  1504. var tableDirection = bidiVisual != null ? new XAttribute("dir", "rtl") : new XAttribute("dir", "ltr");
  1505. style.AddIfMissing("margin-bottom", ".001pt");
  1506. var table = new XElement(Xhtml.table,
  1507. // TODO: Revisit and make sure the omission is covered by appropriate CSS.
  1508. // new XAttribute("border", "1"),
  1509. // new XAttribute("cellspacing", 0),
  1510. // new XAttribute("cellpadding", 0),
  1511. tableDirection,
  1512. element.Elements().Select(e => ConvertToHtmlTransform(wordDoc, settings, e, false, currentMarginLeft)));
  1513. table.AddAnnotation(style);
  1514. var jc = (string)element.Elements(W.tblPr).Elements(W.jc).Attributes(W.val).FirstOrDefault() ?? "left";
  1515. XAttribute dir = null;
  1516. XAttribute jcToUse = null;
  1517. if (bidiVisual != null)
  1518. {
  1519. dir = new XAttribute("dir", "rtl");
  1520. if (jc == "left")
  1521. jcToUse = new XAttribute("align", "right");
  1522. else if (jc == "right")
  1523. jcToUse = new XAttribute("align", "left");
  1524. else if (jc == "center")
  1525. jcToUse = new XAttribute("align", "center");
  1526. }
  1527. else
  1528. {
  1529. jcToUse = new XAttribute("align", jc);
  1530. }
  1531. var tableDiv = new XElement(Xhtml.div,
  1532. dir,
  1533. jcToUse,
  1534. table);
  1535. return tableDiv;
  1536. }
  1537. [SuppressMessage("ReSharper", "PossibleNullReferenceException")]
  1538. private static object ProcessTableCell(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings, XElement element)
  1539. {
  1540. var style = new Dictionary<string, string>();
  1541. XAttribute colSpan = null;
  1542. XAttribute rowSpan = null;
  1543. var tcPr = element.Element(W.tcPr);
  1544. if (tcPr != null)
  1545. {
  1546. if ((string)tcPr.Elements(W.vMerge).Attributes(W.val).FirstOrDefault() == "restart")
  1547. {
  1548. var currentRow = element.Parent.ElementsBeforeSelf(W.tr).Count();
  1549. var currentCell = element.ElementsBeforeSelf(W.tc).Count();
  1550. var tbl = element.Parent.Parent;
  1551. int rowSpanCount = 1;
  1552. currentRow += 1;
  1553. while (true)
  1554. {
  1555. var row = tbl.Elements(W.tr).Skip(currentRow).FirstOrDefault();
  1556. if (row == null)
  1557. break;
  1558. var cell2 = row.Elements(W.tc).Skip(currentCell).FirstOrDefault();
  1559. if (cell2 == null)
  1560. break;
  1561. if (cell2.Elements(W.tcPr).Elements(W.vMerge).FirstOrDefault() == null)
  1562. break;
  1563. if ((string)cell2.Elements(W.tcPr).Elements(W.vMerge).Attributes(W.val).FirstOrDefault() == "restart")
  1564. break;
  1565. currentRow += 1;
  1566. rowSpanCount += 1;
  1567. }
  1568. rowSpan = new XAttribute("rowspan", rowSpanCount);
  1569. }
  1570. if (tcPr.Element(W.vMerge) != null &&
  1571. (string)tcPr.Elements(W.vMerge).Attributes(W.val).FirstOrDefault() != "restart")
  1572. return null;
  1573. if (tcPr.Element(W.vAlign) != null)
  1574. {
  1575. var vAlignVal = (string)tcPr.Elements(W.vAlign).Attributes(W.val).FirstOrDefault();
  1576. if (vAlignVal == "00")
  1577. style.AddIfMissing("vertical-align", "top");
  1578. else if (vAlignVal == "center")
  1579. style.AddIfMissing("vertical-align", "middle");
  1580. else if (vAlignVal == "bottom")
  1581. style.AddIfMissing("vertical-align", "bottom");
  1582. else
  1583. style.AddIfMissing("vertical-align", "middle");
  1584. }
  1585. style.AddIfMissing("vertical-align", "top");
  1586. if ((string)tcPr.Elements(W.tcW).Attributes(W.type).FirstOrDefault() == "dxa")
  1587. {
  1588. decimal width = (int)tcPr.Elements(W.tcW).Attributes(W._w).FirstOrDefault();
  1589. style.AddIfMissing("width", string.Format(NumberFormatInfo.InvariantInfo, "{0}pt", width / 20m));
  1590. }
  1591. if ((string)tcPr.Elements(W.tcW).Attributes(W.type).FirstOrDefault() == "pct")
  1592. {
  1593. decimal width = (int)tcPr.Elements(W.tcW).Attributes(W._w).FirstOrDefault();
  1594. style.AddIfMissing("width", string.Format(NumberFormatInfo.InvariantInfo, "{0:0.0}%", width / 50m));
  1595. }
  1596. var tcBorders = tcPr.Element(W.tcBorders);
  1597. GenerateBorderStyle(tcBorders, W.top, style, BorderType.Cell);
  1598. GenerateBorderStyle(tcBorders, W.right, style, BorderType.Cell);
  1599. GenerateBorderStyle(tcBorders, W.bottom, style, BorderType.Cell);
  1600. GenerateBorderStyle(tcBorders, W.left, style, BorderType.Cell);
  1601. CreateStyleFromShd(style, tcPr.Element(W.shd));
  1602. var gridSpan = tcPr.Elements(W.gridSpan).Attributes(W.val).Select(a => (int?)a).FirstOrDefault();
  1603. if (gridSpan != null)
  1604. colSpan = new XAttribute("colspan", (int)gridSpan);
  1605. }
  1606. style.AddIfMissing("padding-top", "0");
  1607. style.AddIfMissing("padding-bottom", "0");
  1608. var cell = new XElement(Xhtml.td,
  1609. rowSpan,
  1610. colSpan,
  1611. CreateBorderDivs(wordDoc, settings, element.Elements()));
  1612. cell.AddAnnotation(style);
  1613. return cell;
  1614. }
  1615. private static object ProcessTableRow(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings, XElement element,
  1616. decimal currentMarginLeft)
  1617. {
  1618. var style = new Dictionary<string, string>();
  1619. int? trHeight = (int?)element.Elements(W.trPr).Elements(W.trHeight).Attributes(W.val).FirstOrDefault();
  1620. if (trHeight != null)
  1621. style.AddIfMissing("height",
  1622. string.Format(NumberFormatInfo.InvariantInfo, "{0:0.00}in", (decimal)trHeight / 1440m));
  1623. var htmlRow = new XElement(Xhtml.tr,
  1624. element.Elements().Select(e => ConvertToHtmlTransform(wordDoc, settings, e, false, currentMarginLeft)));
  1625. if (style.Any())
  1626. htmlRow.AddAnnotation(style);
  1627. return htmlRow;
  1628. }
  1629. private static bool HasStyleSeparator(XElement element)
  1630. {
  1631. return element != null && element.Elements(W.pPr).Elements(W.rPr).Any(e => GetBoolProp(e, W.specVanish));
  1632. }
  1633. private static bool IsBidi(XElement element)
  1634. {
  1635. return element
  1636. .Elements(W.pPr)
  1637. .Elements(W.bidi)
  1638. .Any(b => b.Attribute(W.val) == null || b.Attribute(W.val).ToBoolean() == true);
  1639. }
  1640. private static XName GetParagraphElementName(XElement element, WordprocessingDocument wordDoc)
  1641. {
  1642. var elementName = Xhtml.p;
  1643. var styleId = (string)element.Elements(W.pPr).Elements(W.pStyle).Attributes(W.val).FirstOrDefault();
  1644. if (styleId == null) return elementName;
  1645. var style = GetStyle(styleId, wordDoc);
  1646. if (style == null) return elementName;
  1647. var outlineLevel =
  1648. (int?)style.Elements(W.pPr).Elements(W.outlineLvl).Attributes(W.val).FirstOrDefault();
  1649. if (outlineLevel != null && outlineLevel <= 5)
  1650. {
  1651. elementName = Xhtml.xhtml + string.Format("h{0}", outlineLevel + 1);
  1652. }
  1653. return elementName;
  1654. }
  1655. private static XElement GetStyle(string styleId, WordprocessingDocument wordDoc)
  1656. {
  1657. var stylesPart = wordDoc.MainDocumentPart.StyleDefinitionsPart;
  1658. if (stylesPart == null) return null;
  1659. var styles = stylesPart.GetXDocument().Root;
  1660. return styles != null
  1661. ? styles.Elements(W.style).FirstOrDefault(s => (string)s.Attribute(W.styleId) == styleId)
  1662. : null;
  1663. }
  1664. private static object CreateSectionDivs(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings, XElement element)
  1665. {
  1666. // note: when building a paging html converter, need to attend to new sections with page breaks here.
  1667. // This code conflates adjacent sections if they have identical formatting, which is not an issue
  1668. // for the non-paging transform.
  1669. var groupedIntoDivs = element
  1670. .Elements()
  1671. .GroupAdjacent(e =>
  1672. {
  1673. var sectAnnotation = e.Annotation<SectionAnnotation>();
  1674. return sectAnnotation != null ? sectAnnotation.SectionElement.ToString() : "";
  1675. });
  1676. // note: when creating a paging html converter, need to pay attention to w:rtlGutter element.
  1677. var divList = groupedIntoDivs
  1678. .Select(g =>
  1679. {
  1680. var sectPr = g.First().Annotation<SectionAnnotation>();
  1681. XElement bidi = null;
  1682. if (sectPr != null)
  1683. {
  1684. bidi = sectPr
  1685. .SectionElement
  1686. .Elements(W.bidi)
  1687. .FirstOrDefault(b => b.Attribute(W.val) == null || b.Attribute(W.val).ToBoolean() == true);
  1688. }
  1689. if (sectPr == null || bidi == null)
  1690. {
  1691. var div = new XElement(Xhtml.div, CreateBorderDivs(wordDoc, settings, g));
  1692. return div;
  1693. }
  1694. else
  1695. {
  1696. var div = new XElement(Xhtml.div,
  1697. new XAttribute("dir", "rtl"),
  1698. CreateBorderDivs(wordDoc, settings, g));
  1699. return div;
  1700. }
  1701. });
  1702. return divList;
  1703. }
  1704. private enum BorderType
  1705. {
  1706. Paragraph,
  1707. Cell,
  1708. };
  1709. /*
  1710. * Notes on line spacing
  1711. *
  1712. * the w:line and w:lineRule attributes control spacing between lines - including between lines within a paragraph
  1713. *
  1714. * If w:spacing w:lineRule="auto" then
  1715. * w:spacing w:line is a percentage where 240 == 100%
  1716. *
  1717. * (line value / 240) * 100 = percentage of line
  1718. *
  1719. * If w:spacing w:lineRule="exact" or w:lineRule="atLeast" then
  1720. * w:spacing w:line is in twips
  1721. * 1440 = exactly one inch from line to line
  1722. *
  1723. * Handle
  1724. * - ind
  1725. * - jc
  1726. * - numPr
  1727. * - pBdr
  1728. * - shd
  1729. * - spacing
  1730. * - textAlignment
  1731. *
  1732. * Don't Handle (yet)
  1733. * - adjustRightInd?
  1734. * - autoSpaceDE
  1735. * - autoSpaceDN
  1736. * - bidi
  1737. * - contextualSpacing
  1738. * - divId
  1739. * - framePr
  1740. * - keepLines
  1741. * - keepNext
  1742. * - kinsoku
  1743. * - mirrorIndents
  1744. * - overflowPunct
  1745. * - pageBreakBefore
  1746. * - snapToGrid
  1747. * - suppressAutoHyphens
  1748. * - suppressLineNumbers
  1749. * - suppressOverlap
  1750. * - tabs
  1751. * - textBoxTightWrap
  1752. * - textDirection
  1753. * - topLinePunct
  1754. * - widowControl
  1755. * - wordWrap
  1756. *
  1757. */
  1758. private static object ConvertParagraph(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings,
  1759. XElement paragraph, XName elementName, bool suppressTrailingWhiteSpace, decimal currentMarginLeft, bool isBidi)
  1760. {
  1761. var style = DefineParagraphStyle(paragraph, elementName, suppressTrailingWhiteSpace, currentMarginLeft, isBidi);
  1762. var rtl = isBidi ? new XAttribute("dir", "rtl") : new XAttribute("dir", "ltr");
  1763. var firstMark = isBidi ? new XEntity("#x200f") : null;
  1764. // Analyze initial runs to see whether we have a tab, in which case we will render
  1765. // a span with a defined width and ignore the tab rather than rendering the text
  1766. // preceding the tab and the tab as a span with a computed width.
  1767. var firstTabRun = paragraph
  1768. .Elements(W.r)
  1769. .FirstOrDefault(run => run.Elements(W.tab).Any());
  1770. var elementsPrecedingTab = firstTabRun != null
  1771. ? paragraph.Elements(W.r).TakeWhile(e => e != firstTabRun)
  1772. .Where(e => e.Elements().Any(c => c.Attributes(PtOpenXml.TabWidth).Any())).ToList()
  1773. : Enumerable.Empty<XElement>().ToList();
  1774. // TODO: Revisit
  1775. // For the time being, if a hyperlink field precedes the tab, we'll render it as before.
  1776. var hyperlinkPrecedesTab = elementsPrecedingTab
  1777. .Elements(W.r)
  1778. .Elements(W.instrText)
  1779. .Select(e => e.Value)
  1780. .Any(value => value != null && value.TrimStart().ToUpper().StartsWith("HYPERLINK"));
  1781. if (hyperlinkPrecedesTab)
  1782. {
  1783. var paraElement1 = new XElement(elementName,
  1784. rtl,
  1785. firstMark,
  1786. ConvertContentThatCanContainFields(wordDoc, settings, paragraph.Elements()));
  1787. paraElement1.AddAnnotation(style);
  1788. return paraElement1;
  1789. }
  1790. var txElementsPrecedingTab = TransformElementsPrecedingTab(wordDoc, settings, elementsPrecedingTab, firstTabRun);
  1791. var elementsSucceedingTab = firstTabRun != null
  1792. ? paragraph.Elements().SkipWhile(e => e != firstTabRun).Skip(1)
  1793. : paragraph.Elements();
  1794. var paraElement = new XElement(elementName,
  1795. rtl,
  1796. firstMark,
  1797. txElementsPrecedingTab,
  1798. ConvertContentThatCanContainFields(wordDoc, settings, elementsSucceedingTab));
  1799. paraElement.AddAnnotation(style);
  1800. return paraElement;
  1801. }
  1802. private static List<object> TransformElementsPrecedingTab(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings,
  1803. List<XElement> elementsPrecedingTab, XElement firstTabRun)
  1804. {
  1805. var tabWidth = firstTabRun != null
  1806. ? (decimal?)firstTabRun.Elements(W.tab).Attributes(PtOpenXml.TabWidth).FirstOrDefault() ?? 0m
  1807. : 0m;
  1808. var precedingElementsWidth = elementsPrecedingTab
  1809. .Elements()
  1810. .Where(c => c.Attributes(PtOpenXml.TabWidth).Any())
  1811. .Select(e => (decimal)e.Attribute(PtOpenXml.TabWidth))
  1812. .Sum();
  1813. var totalWidth = precedingElementsWidth + tabWidth;
  1814. var txElementsPrecedingTab = elementsPrecedingTab
  1815. .Select(e => ConvertToHtmlTransform(wordDoc, settings, e, false, 0m))
  1816. .ToList();
  1817. if (txElementsPrecedingTab.Count > 1)
  1818. {
  1819. var span = new XElement(Xhtml.span, txElementsPrecedingTab);
  1820. var spanStyle = new Dictionary<string, string>
  1821. {
  1822. { "display", "inline-block" },
  1823. { "text-indent", "0" },
  1824. { "width", string.Format(NumberFormatInfo.InvariantInfo, "{0:0.000}in", totalWidth) }
  1825. };
  1826. span.AddAnnotation(spanStyle);
  1827. }
  1828. else if (txElementsPrecedingTab.Count == 1)
  1829. {
  1830. var element = txElementsPrecedingTab.First() as XElement;
  1831. if (element != null)
  1832. {
  1833. var spanStyle = element.Annotation<Dictionary<string, string>>();
  1834. spanStyle.AddIfMissing("display", "inline-block");
  1835. spanStyle.AddIfMissing("text-indent", "0");
  1836. spanStyle.AddIfMissing("width", string.Format(NumberFormatInfo.InvariantInfo, "{0:0.000}in", totalWidth));
  1837. }
  1838. }
  1839. return txElementsPrecedingTab;
  1840. }
  1841. private static Dictionary<string, string> DefineParagraphStyle(XElement paragraph, XName elementName,
  1842. bool suppressTrailingWhiteSpace, decimal currentMarginLeft, bool isBidi)
  1843. {
  1844. var style = new Dictionary<string, string>();
  1845. var styleName = (string)paragraph.Attribute(PtOpenXml.StyleName);
  1846. if (styleName != null)
  1847. style.Add("PtStyleName", styleName);
  1848. var pPr = paragraph.Element(W.pPr);
  1849. if (pPr == null) return style;
  1850. CreateStyleFromSpacing(style, pPr.Element(W.spacing), elementName, suppressTrailingWhiteSpace);
  1851. CreateStyleFromInd(style, pPr.Element(W.ind), elementName, currentMarginLeft, isBidi);
  1852. // todo need to handle
  1853. // - both
  1854. // - mediumKashida
  1855. // - distribute
  1856. // - numTab
  1857. // - highKashida
  1858. // - lowKashida
  1859. // - thaiDistribute
  1860. CreateStyleFromJc(style, pPr.Element(W.jc), isBidi);
  1861. CreateStyleFromShd(style, pPr.Element(W.shd));
  1862. // Pt.FontName
  1863. var font = (string)paragraph.Attributes(PtOpenXml.FontName).FirstOrDefault();
  1864. if (font != null)
  1865. CreateFontCssProperty(font, style);
  1866. DefineFontSize(style, paragraph);
  1867. DefineLineHeight(style, paragraph);
  1868. // vertical text alignment as of December 2013 does not work in any major browsers.
  1869. CreateStyleFromTextAlignment(style, pPr.Element(W.textAlignment));
  1870. style.AddIfMissing("margin-top", "0");
  1871. style.AddIfMissing("margin-left", "0");
  1872. style.AddIfMissing("margin-right", "0");
  1873. style.AddIfMissing("margin-bottom", ".001pt");
  1874. return style;
  1875. }
  1876. private static void CreateStyleFromInd(Dictionary<string, string> style, XElement ind, XName elementName,
  1877. decimal currentMarginLeft, bool isBidi)
  1878. {
  1879. if (ind == null) return;
  1880. var left = (decimal?)ind.Attribute(W.left);
  1881. if (left != null && elementName != Xhtml.span)
  1882. {
  1883. var leftInInches = (decimal)left / 1440 - currentMarginLeft;
  1884. style.AddIfMissing(isBidi ? "margin-right" : "margin-left",
  1885. leftInInches > 0m
  1886. ? string.Format(NumberFormatInfo.InvariantInfo, "{0:0.00}in", leftInInches)
  1887. : "0");
  1888. }
  1889. var right = (decimal?)ind.Attribute(W.right);
  1890. if (right != null)
  1891. {
  1892. var rightInInches = (decimal)right / 1440;
  1893. style.AddIfMissing(isBidi ? "margin-left" : "margin-right",
  1894. rightInInches > 0m
  1895. ? string.Format(NumberFormatInfo.InvariantInfo, "{0:0.00}in", rightInInches)
  1896. : "0");
  1897. }
  1898. var firstLine = (decimal?)ind.Attribute(W.firstLine);
  1899. if (firstLine != null && elementName != Xhtml.span)
  1900. {
  1901. var firstLineInInches = (decimal)firstLine / 1440m;
  1902. style.AddIfMissing("text-indent",
  1903. string.Format(NumberFormatInfo.InvariantInfo, "{0:0.00}in", firstLineInInches));
  1904. }
  1905. var hanging = (decimal?)ind.Attribute(W.hanging);
  1906. if (hanging != null && elementName != Xhtml.span)
  1907. {
  1908. var hangingInInches = (decimal)-hanging / 1440m;
  1909. style.AddIfMissing("text-indent",
  1910. string.Format(NumberFormatInfo.InvariantInfo, "{0:0.00}in", hangingInInches));
  1911. }
  1912. }
  1913. private static void CreateStyleFromJc(Dictionary<string, string> style, XElement jc, bool isBidi)
  1914. {
  1915. if (jc != null)
  1916. {
  1917. var jcVal = (string)jc.Attributes(W.val).FirstOrDefault() ?? "left";
  1918. if (jcVal == "left")
  1919. style.AddIfMissing("text-align", isBidi ? "right" : "left");
  1920. else if (jcVal == "right")
  1921. style.AddIfMissing("text-align", isBidi ? "left" : "right");
  1922. else if (jcVal == "center")
  1923. style.AddIfMissing("text-align", "center");
  1924. else if (jcVal == "both")
  1925. style.AddIfMissing("text-align", "justify");
  1926. }
  1927. }
  1928. private static void CreateStyleFromSpacing(Dictionary<string, string> style, XElement spacing, XName elementName,
  1929. bool suppressTrailingWhiteSpace)
  1930. {
  1931. if (spacing == null) return;
  1932. var spacingBefore = (decimal?)spacing.Attribute(W.before);
  1933. if (spacingBefore != null && elementName != Xhtml.span)
  1934. style.AddIfMissing("margin-top",
  1935. spacingBefore > 0m
  1936. ? string.Format(NumberFormatInfo.InvariantInfo, "{0}pt", spacingBefore / 20.0m)
  1937. : "0");
  1938. var lineRule = (string)spacing.Attribute(W.lineRule);
  1939. if (lineRule == "auto")
  1940. {
  1941. var line = (decimal)spacing.Attribute(W.line);
  1942. if (line != 240m)
  1943. {
  1944. var pct = (line / 240m) * 100m;
  1945. style.Add("line-height", string.Format(NumberFormatInfo.InvariantInfo, "{0:0.0}%", pct));
  1946. }
  1947. }
  1948. if (lineRule == "exact")
  1949. {
  1950. var line = (decimal)spacing.Attribute(W.line);
  1951. var points = line / 20m;
  1952. style.Add("line-height", string.Format(NumberFormatInfo.InvariantInfo, "{0:0.0}pt", points));
  1953. }
  1954. if (lineRule == "atLeast")
  1955. {
  1956. var line = (decimal)spacing.Attribute(W.line);
  1957. var points = line / 20m;
  1958. if (points >= 14m)
  1959. style.Add("line-height", string.Format(NumberFormatInfo.InvariantInfo, "{0:0.0}pt", points));
  1960. }
  1961. var spacingAfter = suppressTrailingWhiteSpace ? 0m : (decimal?)spacing.Attribute(W.after);
  1962. if (spacingAfter != null)
  1963. style.AddIfMissing("margin-bottom",
  1964. spacingAfter > 0m
  1965. ? string.Format(NumberFormatInfo.InvariantInfo, "{0}pt", spacingAfter / 20.0m)
  1966. : "0");
  1967. }
  1968. private static void CreateStyleFromTextAlignment(Dictionary<string, string> style, XElement textAlignment)
  1969. {
  1970. if (textAlignment == null) return;
  1971. var verticalTextAlignment = (string)textAlignment.Attributes(W.val).FirstOrDefault();
  1972. if (verticalTextAlignment == null || verticalTextAlignment == "auto") return;
  1973. if (verticalTextAlignment == "top")
  1974. style.AddIfMissing("vertical-align", "top");
  1975. else if (verticalTextAlignment == "center")
  1976. style.AddIfMissing("vertical-align", "middle");
  1977. else if (verticalTextAlignment == "baseline")
  1978. style.AddIfMissing("vertical-align", "baseline");
  1979. else if (verticalTextAlignment == "bottom")
  1980. style.AddIfMissing("vertical-align", "bottom");
  1981. }
  1982. private static void DefineFontSize(Dictionary<string, string> style, XElement paragraph)
  1983. {
  1984. var sz = paragraph
  1985. .DescendantsTrimmed(W.txbxContent)
  1986. .Where(e => e.Name == W.r)
  1987. .Select(r => GetFontSize(r))
  1988. .Max();
  1989. if (sz != null)
  1990. style.AddIfMissing("font-size", string.Format(NumberFormatInfo.InvariantInfo, "{0}pt", sz / 2.0m));
  1991. }
  1992. private static void DefineLineHeight(Dictionary<string, string> style, XElement paragraph)
  1993. {
  1994. var allRunsAreUniDirectional = paragraph
  1995. .DescendantsTrimmed(W.txbxContent)
  1996. .Where(e => e.Name == W.r)
  1997. .Select(run => (string)run.Attribute(PtOpenXml.LanguageType))
  1998. .All(lt => lt != "bidi");
  1999. if (allRunsAreUniDirectional)
  2000. style.AddIfMissing("line-height", "108%");
  2001. }
  2002. /*
  2003. * Handle:
  2004. * - b
  2005. * - bdr
  2006. * - caps
  2007. * - color
  2008. * - dstrike
  2009. * - highlight
  2010. * - i
  2011. * - position
  2012. * - rFonts
  2013. * - shd
  2014. * - smallCaps
  2015. * - spacing
  2016. * - strike
  2017. * - sz
  2018. * - u
  2019. * - vanish
  2020. * - vertAlign
  2021. *
  2022. * Don't handle:
  2023. * - em
  2024. * - emboss
  2025. * - fitText
  2026. * - imprint
  2027. * - kern
  2028. * - outline
  2029. * - shadow
  2030. * - w
  2031. *
  2032. */
  2033. private static object ConvertRun(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings, XElement run)
  2034. {
  2035. var rPr = run.Element(W.rPr);
  2036. if (rPr == null)
  2037. return run.Elements().Select(e => ConvertToHtmlTransform(wordDoc, settings, e, false, 0m));
  2038. // hide all content that contains the w:rPr/w:webHidden element
  2039. if (rPr.Element(W.webHidden) != null)
  2040. return null;
  2041. var style = DefineRunStyle(run);
  2042. object content = run.Elements().Select(e => ConvertToHtmlTransform(wordDoc, settings, e, false, 0m));
  2043. if (rPr.Element(W.em) != null)
  2044. {
  2045. content = new XElement(Xhtml.dot, content);
  2046. }
  2047. // Wrap content in h:sup or h:sub elements as necessary.
  2048. if (rPr.Element(W.vertAlign) != null)
  2049. {
  2050. XElement newContent = null;
  2051. var vertAlignVal = (string)rPr.Elements(W.vertAlign).Attributes(W.val).FirstOrDefault();
  2052. switch (vertAlignVal)
  2053. {
  2054. case "superscript":
  2055. newContent = new XElement(Xhtml.sup, content);
  2056. break;
  2057. case "subscript":
  2058. newContent = new XElement(Xhtml.sub, content);
  2059. break;
  2060. }
  2061. if (newContent != null && newContent.Nodes().Any())
  2062. content = newContent;
  2063. }
  2064. var langAttribute = GetLangAttribute(run);
  2065. XEntity runStartMark;
  2066. XEntity runEndMark;
  2067. DetermineRunMarks(run, rPr, style, out runStartMark, out runEndMark);
  2068. if (style.Any() || langAttribute != null || runStartMark != null)
  2069. {
  2070. style.AddIfMissing("margin", "0");
  2071. style.AddIfMissing("padding", "0");
  2072. var xe = new XElement(Xhtml.span,
  2073. langAttribute,
  2074. runStartMark,
  2075. content,
  2076. runEndMark);
  2077. xe.AddAnnotation(style);
  2078. content = xe;
  2079. }
  2080. return content;
  2081. }
  2082. [SuppressMessage("ReSharper", "FunctionComplexityOverflow")]
  2083. private static Dictionary<string, string> DefineRunStyle(XElement run)
  2084. {
  2085. var style = new Dictionary<string, string>();
  2086. var rPr = run.Elements(W.rPr).First();
  2087. var styleName = (string)run.Attribute(PtOpenXml.StyleName);
  2088. if (styleName != null)
  2089. style.Add("PtStyleName", styleName);
  2090. // W.bdr
  2091. if (rPr.Element(W.bdr) != null && (string)rPr.Elements(W.bdr).Attributes(W.val).FirstOrDefault() != "none")
  2092. {
  2093. style.AddIfMissing("border", "solid windowtext 1.0pt");
  2094. style.AddIfMissing("padding", "0");
  2095. }
  2096. // W.color
  2097. var color = (string)rPr.Elements(W.color).Attributes(W.val).FirstOrDefault();
  2098. if (color != null)
  2099. CreateColorProperty("color", color, style);
  2100. // W.highlight
  2101. var highlight = (string)rPr.Elements(W.highlight).Attributes(W.val).FirstOrDefault();
  2102. if (highlight != null)
  2103. CreateColorProperty("background", highlight, style);
  2104. // W.shd
  2105. var shade = (string)rPr.Elements(W.shd).Attributes(W.fill).FirstOrDefault();
  2106. if (shade != null)
  2107. CreateColorProperty("background", shade, style);
  2108. // Pt.FontName
  2109. var sym = run.Element(W.sym);
  2110. var font = sym != null
  2111. ? (string)sym.Attributes(W.font).FirstOrDefault()
  2112. : (string)run.Attributes(PtOpenXml.FontName).FirstOrDefault();
  2113. if (font != null)
  2114. CreateFontCssProperty(font, style);
  2115. // W.sz
  2116. var languageType = (string)run.Attribute(PtOpenXml.LanguageType);
  2117. var sz = GetFontSize(languageType, rPr);
  2118. if (sz != null)
  2119. style.AddIfMissing("font-size", string.Format(NumberFormatInfo.InvariantInfo, "{0}pt", sz / 2.0m));
  2120. // W.caps
  2121. if (GetBoolProp(rPr, W.caps))
  2122. style.AddIfMissing("text-transform", "uppercase");
  2123. // W.smallCaps
  2124. if (GetBoolProp(rPr, W.smallCaps))
  2125. style.AddIfMissing("font-variant", "small-caps");
  2126. // W.spacing
  2127. var spacingInTwips = (decimal?)rPr.Elements(W.spacing).Attributes(W.val).FirstOrDefault();
  2128. if (spacingInTwips != null)
  2129. style.AddIfMissing("letter-spacing",
  2130. spacingInTwips > 0m
  2131. ? string.Format(NumberFormatInfo.InvariantInfo, "{0}pt", spacingInTwips / 20)
  2132. : "0");
  2133. // W.position
  2134. var position = (decimal?)rPr.Elements(W.position).Attributes(W.val).FirstOrDefault();
  2135. if (position != null)
  2136. {
  2137. style.AddIfMissing("position", "relative");
  2138. style.AddIfMissing("top", string.Format(NumberFormatInfo.InvariantInfo, "{0}pt", -(position / 2)));
  2139. }
  2140. // W.vanish
  2141. if (GetBoolProp(rPr, W.vanish) && !GetBoolProp(rPr, W.specVanish))
  2142. style.AddIfMissing("display", "none");
  2143. // W.u
  2144. if (rPr.Element(W.u) != null && (string)rPr.Elements(W.u).Attributes(W.val).FirstOrDefault() != "none")
  2145. style.AddIfMissing("text-decoration", "underline");
  2146. // W.i
  2147. style.AddIfMissing("font-style", GetBoolProp(rPr, W.i) ? "italic" : "normal");
  2148. // W.b
  2149. style.AddIfMissing("font-weight", GetBoolProp(rPr, W.b) ? "bold" : "normal");
  2150. // W.strike
  2151. if (GetBoolProp(rPr, W.strike) || GetBoolProp(rPr, W.dstrike))
  2152. style.AddIfMissing("text-decoration", "line-through");
  2153. return style;
  2154. }
  2155. private static decimal? GetFontSize(XElement e)
  2156. {
  2157. var languageType = (string)e.Attribute(PtOpenXml.LanguageType);
  2158. if (e.Name == W.p)
  2159. {
  2160. return GetFontSize(languageType, e.Elements(W.pPr).Elements(W.rPr).FirstOrDefault());
  2161. }
  2162. if (e.Name == W.r)
  2163. {
  2164. return GetFontSize(languageType, e.Element(W.rPr));
  2165. }
  2166. return null;
  2167. }
  2168. private static decimal? GetFontSize(string languageType, XElement rPr)
  2169. {
  2170. if (rPr == null) return null;
  2171. return languageType == "bidi"
  2172. ? (decimal?)rPr.Elements(W.szCs).Attributes(W.val).FirstOrDefault()
  2173. : (decimal?)rPr.Elements(W.sz).Attributes(W.val).FirstOrDefault();
  2174. }
  2175. private static void DetermineRunMarks(XElement run, XElement rPr, Dictionary<string, string> style, out XEntity runStartMark, out XEntity runEndMark)
  2176. {
  2177. runStartMark = null;
  2178. runEndMark = null;
  2179. // Only do the following for text runs.
  2180. if (run.Element(W.t) == null) return;
  2181. // Can't add directional marks if the font-family is symbol - they are visible, and display as a ?
  2182. var addDirectionalMarks = true;
  2183. if (style.ContainsKey("font-family"))
  2184. {
  2185. if (style["font-family"].ToLower() == "symbol")
  2186. addDirectionalMarks = false;
  2187. }
  2188. if (!addDirectionalMarks) return;
  2189. var isRtl = rPr.Element(W.rtl) != null;
  2190. if (isRtl)
  2191. {
  2192. runStartMark = new XEntity("#x200f"); // RLM
  2193. runEndMark = new XEntity("#x200f"); // RLM
  2194. }
  2195. else
  2196. {
  2197. var paragraph = run.Ancestors(W.p).First();
  2198. var paraIsBidi = paragraph
  2199. .Elements(W.pPr)
  2200. .Elements(W.bidi)
  2201. .Any(b => b.Attribute(W.val) == null || b.Attribute(W.val).ToBoolean() == true);
  2202. if (paraIsBidi)
  2203. {
  2204. runStartMark = new XEntity("#x200e"); // LRM
  2205. runEndMark = new XEntity("#x200e"); // LRM
  2206. }
  2207. }
  2208. }
  2209. private static XAttribute GetLangAttribute(XElement run)
  2210. {
  2211. const string defaultLanguage = "en-US"; // todo need to get defaultLanguage
  2212. var rPr = run.Elements(W.rPr).First();
  2213. var languageType = (string)run.Attribute(PtOpenXml.LanguageType);
  2214. string lang = null;
  2215. if (languageType == "western")
  2216. lang = (string)rPr.Elements(W.lang).Attributes(W.val).FirstOrDefault();
  2217. else if (languageType == "bidi")
  2218. lang = (string)rPr.Elements(W.lang).Attributes(W.bidi).FirstOrDefault();
  2219. else if (languageType == "eastAsia")
  2220. lang = (string)rPr.Elements(W.lang).Attributes(W.eastAsia).FirstOrDefault();
  2221. if (lang == null)
  2222. lang = defaultLanguage;
  2223. return lang != defaultLanguage ? new XAttribute("lang", lang) : null;
  2224. }
  2225. private static void AdjustTableBorders(WordprocessingDocument wordDoc)
  2226. {
  2227. // Note: when implementing a paging version of the HTML transform, this needs to be done
  2228. // for all content parts, not just the main document part.
  2229. var xd = wordDoc.MainDocumentPart.GetXDocument();
  2230. foreach (var tbl in xd.Descendants(W.tbl))
  2231. AdjustTableBorders(tbl);
  2232. wordDoc.MainDocumentPart.PutXDocument();
  2233. }
  2234. private static void AdjustTableBorders(XElement tbl)
  2235. {
  2236. var ta = tbl
  2237. .Elements(W.tr)
  2238. .Select(r => r
  2239. .Elements(W.tc)
  2240. .SelectMany(c =>
  2241. Enumerable.Repeat(c,
  2242. (int?)c.Elements(W.tcPr).Elements(W.gridSpan).Attributes(W.val).FirstOrDefault() ?? 1))
  2243. .ToArray())
  2244. .ToArray();
  2245. for (var y = 0; y < ta.Length; y++)
  2246. {
  2247. for (var x = 0; x < ta[y].Length; x++)
  2248. {
  2249. var thisCell = ta[y][x];
  2250. FixTopBorder(ta, thisCell, x, y);
  2251. FixLeftBorder(ta, thisCell, x, y);
  2252. FixBottomBorder(ta, thisCell, x, y);
  2253. FixRightBorder(ta, thisCell, x, y);
  2254. }
  2255. }
  2256. }
  2257. private static void FixTopBorder(XElement[][] ta, XElement thisCell, int x, int y)
  2258. {
  2259. if (y > 0)
  2260. {
  2261. var rowAbove = ta[y - 1];
  2262. if (x < rowAbove.Length - 1)
  2263. {
  2264. XElement cellAbove = ta[y - 1][x];
  2265. if (cellAbove != null &&
  2266. thisCell.Elements(W.tcPr).Elements(W.tcBorders).FirstOrDefault() != null &&
  2267. cellAbove.Elements(W.tcPr).Elements(W.tcBorders).FirstOrDefault() != null)
  2268. {
  2269. ResolveCellBorder(
  2270. thisCell.Elements(W.tcPr).Elements(W.tcBorders).Elements(W.top).FirstOrDefault(),
  2271. cellAbove.Elements(W.tcPr).Elements(W.tcBorders).Elements(W.bottom).FirstOrDefault());
  2272. }
  2273. }
  2274. }
  2275. }
  2276. private static void FixLeftBorder(XElement[][] ta, XElement thisCell, int x, int y)
  2277. {
  2278. if (x > 0)
  2279. {
  2280. XElement cellLeft = ta[y][x - 1];
  2281. if (cellLeft != null &&
  2282. thisCell.Elements(W.tcPr).Elements(W.tcBorders).FirstOrDefault() != null &&
  2283. cellLeft.Elements(W.tcPr).Elements(W.tcBorders).FirstOrDefault() != null)
  2284. {
  2285. ResolveCellBorder(
  2286. thisCell.Elements(W.tcPr).Elements(W.tcBorders).Elements(W.left).FirstOrDefault(),
  2287. cellLeft.Elements(W.tcPr).Elements(W.tcBorders).Elements(W.right).FirstOrDefault());
  2288. }
  2289. }
  2290. }
  2291. private static void FixBottomBorder(XElement[][] ta, XElement thisCell, int x, int y)
  2292. {
  2293. if (y < ta.Length - 1)
  2294. {
  2295. var rowBelow = ta[y + 1];
  2296. if (x < rowBelow.Length - 1)
  2297. {
  2298. XElement cellBelow = ta[y + 1][x];
  2299. if (cellBelow != null &&
  2300. thisCell.Elements(W.tcPr).Elements(W.tcBorders).FirstOrDefault() != null &&
  2301. cellBelow.Elements(W.tcPr).Elements(W.tcBorders).FirstOrDefault() != null)
  2302. {
  2303. ResolveCellBorder(
  2304. thisCell.Elements(W.tcPr).Elements(W.tcBorders).Elements(W.bottom).FirstOrDefault(),
  2305. cellBelow.Elements(W.tcPr).Elements(W.tcBorders).Elements(W.top).FirstOrDefault());
  2306. }
  2307. }
  2308. }
  2309. }
  2310. private static void FixRightBorder(XElement[][] ta, XElement thisCell, int x, int y)
  2311. {
  2312. if (x < ta[y].Length - 1)
  2313. {
  2314. XElement cellRight = ta[y][x + 1];
  2315. if (cellRight != null &&
  2316. thisCell.Elements(W.tcPr).Elements(W.tcBorders).FirstOrDefault() != null &&
  2317. cellRight.Elements(W.tcPr).Elements(W.tcBorders).FirstOrDefault() != null)
  2318. {
  2319. ResolveCellBorder(
  2320. thisCell.Elements(W.tcPr).Elements(W.tcBorders).Elements(W.right).FirstOrDefault(),
  2321. cellRight.Elements(W.tcPr).Elements(W.tcBorders).Elements(W.left).FirstOrDefault());
  2322. }
  2323. }
  2324. }
  2325. private static readonly Dictionary<string, int> BorderTypePriority = new Dictionary<string, int>()
  2326. {
  2327. { "single", 1 },
  2328. { "thick", 2 },
  2329. { "double", 3 },
  2330. { "dotted", 4 },
  2331. };
  2332. private static readonly Dictionary<string, int> BorderNumber = new Dictionary<string, int>()
  2333. {
  2334. {"single", 1 },
  2335. {"thick", 2 },
  2336. {"double", 3 },
  2337. {"dotted", 4 },
  2338. {"dashed", 5 },
  2339. {"dotDash", 5 },
  2340. {"dotDotDash", 5 },
  2341. {"triple", 6 },
  2342. {"thinThickSmallGap", 6 },
  2343. {"thickThinSmallGap", 6 },
  2344. {"thinThickThinSmallGap", 6 },
  2345. {"thinThickMediumGap", 6 },
  2346. {"thickThinMediumGap", 6 },
  2347. {"thinThickThinMediumGap", 6 },
  2348. {"thinThickLargeGap", 6 },
  2349. {"thickThinLargeGap", 6 },
  2350. {"thinThickThinLargeGap", 6 },
  2351. {"wave", 7 },
  2352. {"doubleWave", 7 },
  2353. {"dashSmallGap", 5 },
  2354. {"dashDotStroked", 5 },
  2355. {"threeDEmboss", 7 },
  2356. {"threeDEngrave", 7 },
  2357. {"outset", 7 },
  2358. {"inset", 7 },
  2359. };
  2360. private static void ResolveCellBorder(XElement border1, XElement border2)
  2361. {
  2362. if (border1 == null || border2 == null)
  2363. return;
  2364. if ((string)border1.Attribute(W.val) == "nil" || (string)border2.Attribute(W.val) == "nil")
  2365. return;
  2366. if ((string)border1.Attribute(W.sz) == "nil" || (string)border2.Attribute(W.sz) == "nil")
  2367. return;
  2368. var border1Val = (string)border1.Attribute(W.val);
  2369. var border1Weight = 1;
  2370. if (BorderNumber.ContainsKey(border1Val))
  2371. border1Weight = BorderNumber[border1Val];
  2372. var border2Val = (string)border2.Attribute(W.val);
  2373. var border2Weight = 1;
  2374. if (BorderNumber.ContainsKey(border2Val))
  2375. border2Weight = BorderNumber[border2Val];
  2376. if (border1Weight != border2Weight)
  2377. {
  2378. if (border1Weight < border2Weight)
  2379. BorderOverride(border2, border1);
  2380. else
  2381. BorderOverride(border1, border2);
  2382. }
  2383. if ((decimal)border1.Attribute(W.sz) > (decimal)border2.Attribute(W.sz))
  2384. {
  2385. BorderOverride(border1, border2);
  2386. return;
  2387. }
  2388. if ((decimal)border1.Attribute(W.sz) < (decimal)border2.Attribute(W.sz))
  2389. {
  2390. BorderOverride(border2, border1);
  2391. return;
  2392. }
  2393. var border1Type = (string)border1.Attribute(W.val);
  2394. var border2Type = (string)border2.Attribute(W.val);
  2395. if (BorderTypePriority.ContainsKey(border1Type) &&
  2396. BorderTypePriority.ContainsKey(border2Type))
  2397. {
  2398. var border1Pri = BorderTypePriority[border1Type];
  2399. var border2Pri = BorderTypePriority[border2Type];
  2400. if (border1Pri < border2Pri)
  2401. {
  2402. BorderOverride(border2, border1);
  2403. return;
  2404. }
  2405. if (border2Pri < border1Pri)
  2406. {
  2407. BorderOverride(border1, border2);
  2408. return;
  2409. }
  2410. }
  2411. var color1Str = (string)border1.Attribute(W.color);
  2412. if (color1Str == "auto")
  2413. color1Str = "000000";
  2414. var color2Str = (string)border2.Attribute(W.color);
  2415. if (color2Str == "auto")
  2416. color2Str = "000000";
  2417. if (color1Str != null && color2Str != null && color1Str != color2Str)
  2418. {
  2419. try
  2420. {
  2421. var color1 = Convert.ToInt32(color1Str, 16);
  2422. var color2 = Convert.ToInt32(color2Str, 16);
  2423. if (color1 < color2)
  2424. {
  2425. BorderOverride(border1, border2);
  2426. return;
  2427. }
  2428. if (color2 < color1)
  2429. {
  2430. BorderOverride(border2, border1);
  2431. }
  2432. }
  2433. // if the above throws ArgumentException, FormatException, or OverflowException, then abort
  2434. catch (Exception)
  2435. {
  2436. // Ignore
  2437. }
  2438. }
  2439. }
  2440. private static void BorderOverride(XElement fromBorder, XElement toBorder)
  2441. {
  2442. toBorder.Attribute(W.val).Value = fromBorder.Attribute(W.val).Value;
  2443. if (fromBorder.Attribute(W.color) != null)
  2444. toBorder.SetAttributeValue(W.color, fromBorder.Attribute(W.color).Value);
  2445. if (fromBorder.Attribute(W.sz) != null)
  2446. toBorder.SetAttributeValue(W.sz, fromBorder.Attribute(W.sz).Value);
  2447. if (fromBorder.Attribute(W.themeColor) != null)
  2448. toBorder.SetAttributeValue(W.themeColor, fromBorder.Attribute(W.themeColor).Value);
  2449. if (fromBorder.Attribute(W.themeTint) != null)
  2450. toBorder.SetAttributeValue(W.themeTint, fromBorder.Attribute(W.themeTint).Value);
  2451. }
  2452. private static void CalculateSpanWidthForTabs(WordprocessingDocument wordDoc)
  2453. {
  2454. // Note: when implementing a paging version of the HTML transform, this needs to be done
  2455. // for all content parts, not just the main document part.
  2456. // w:defaultTabStop in settings
  2457. var sxd = wordDoc.MainDocumentPart.DocumentSettingsPart.GetXDocument();
  2458. var defaultTabStop = (int?)sxd.Descendants(W.defaultTabStop).Attributes(W.val).FirstOrDefault() ?? 720;
  2459. var pxd = wordDoc.MainDocumentPart.GetXDocument();
  2460. var root = pxd.Root;
  2461. if (root == null) return;
  2462. var newRoot = (XElement)CalculateSpanWidthTransform(root, defaultTabStop);
  2463. root.ReplaceWith(newRoot);
  2464. wordDoc.MainDocumentPart.PutXDocument();
  2465. }
  2466. // TODO: Refactor. This method is way too long.
  2467. [SuppressMessage("ReSharper", "FunctionComplexityOverflow")]
  2468. private static object CalculateSpanWidthTransform(XNode node, int defaultTabStop)
  2469. {
  2470. var element = node as XElement;
  2471. if (element == null) return node;
  2472. // if it is not a paragraph or if there are no tabs in the paragraph,
  2473. // then no need to continue processing.
  2474. if (element.Name != W.p ||
  2475. !element.DescendantsTrimmed(W.txbxContent).Where(d => d.Name == W.r).Elements(W.tab).Any())
  2476. {
  2477. // TODO: Revisit. Can we just return the node if it is a paragraph that does not have any tab?
  2478. return new XElement(element.Name,
  2479. element.Attributes(),
  2480. element.Nodes().Select(n => CalculateSpanWidthTransform(n, defaultTabStop)));
  2481. }
  2482. var clonedPara = new XElement(element);
  2483. var leftInTwips = 0;
  2484. var firstInTwips = 0;
  2485. var ind = clonedPara.Elements(W.pPr).Elements(W.ind).FirstOrDefault();
  2486. if (ind != null)
  2487. {
  2488. // todo need to handle start and end attributes
  2489. var left = (int?)ind.Attribute(W.left);
  2490. if (left != null)
  2491. leftInTwips = (int)left;
  2492. var firstLine = 0;
  2493. var firstLineAtt = (int?)ind.Attribute(W.firstLine);
  2494. if (firstLineAtt != null)
  2495. firstLine = (int)firstLineAtt;
  2496. var hangingAtt = (int?)ind.Attribute(W.hanging);
  2497. if (hangingAtt != null)
  2498. firstLine = -(int)hangingAtt;
  2499. firstInTwips = leftInTwips + firstLine;
  2500. }
  2501. // calculate the tab stops, in twips
  2502. var tabs = clonedPara
  2503. .Elements(W.pPr)
  2504. .Elements(W.tabs)
  2505. .FirstOrDefault();
  2506. if (tabs == null)
  2507. {
  2508. if (leftInTwips == 0)
  2509. {
  2510. tabs = new XElement(W.tabs,
  2511. Enumerable.Range(1, 100)
  2512. .Select(r => new XElement(W.tab,
  2513. new XAttribute(W.val, "left"),
  2514. new XAttribute(W.pos, r * defaultTabStop))));
  2515. }
  2516. else
  2517. {
  2518. tabs = new XElement(W.tabs,
  2519. new XElement(W.tab,
  2520. new XAttribute(W.val, "left"),
  2521. new XAttribute(W.pos, leftInTwips)));
  2522. tabs = AddDefaultTabsAfterLastTab(tabs, defaultTabStop);
  2523. }
  2524. }
  2525. else
  2526. {
  2527. if (leftInTwips != 0)
  2528. {
  2529. tabs.Add(
  2530. new XElement(W.tab,
  2531. new XAttribute(W.val, "left"),
  2532. new XAttribute(W.pos, leftInTwips)));
  2533. }
  2534. tabs = AddDefaultTabsAfterLastTab(tabs, defaultTabStop);
  2535. }
  2536. var twipCounter = firstInTwips;
  2537. var contentToMeasure = element.DescendantsTrimmed(z => z.Name == W.txbxContent || z.Name == W.pPr || z.Name == W.rPr).ToArray();
  2538. var currentElementIdx = 0;
  2539. while (true)
  2540. {
  2541. var currentElement = contentToMeasure[currentElementIdx];
  2542. if (currentElement.Name == W.br)
  2543. {
  2544. twipCounter = leftInTwips;
  2545. currentElement.Add(new XAttribute(PtOpenXml.TabWidth,
  2546. string.Format(NumberFormatInfo.InvariantInfo,
  2547. "{0:0.000}", (decimal)firstInTwips / 1440m)));
  2548. currentElementIdx++;
  2549. if (currentElementIdx >= contentToMeasure.Length)
  2550. break; // we're done
  2551. }
  2552. if (currentElement.Name == W.tab)
  2553. {
  2554. var runContainingTabToReplace = currentElement.Parent;
  2555. var fontNameAtt = runContainingTabToReplace.Attribute(PtOpenXml.pt + "FontName") ??
  2556. runContainingTabToReplace.Ancestors(W.p).First().Attribute(PtOpenXml.pt + "FontName");
  2557. var testAmount = twipCounter;
  2558. var tabAfterText = tabs
  2559. .Elements(W.tab)
  2560. .FirstOrDefault(t => (int)t.Attribute(W.pos) > testAmount);
  2561. if (tabAfterText == null)
  2562. {
  2563. // something has gone wrong, so put 1/2 inch in
  2564. if (currentElement.Attribute(PtOpenXml.TabWidth) == null)
  2565. currentElement.Add(
  2566. new XAttribute(PtOpenXml.TabWidth, 720m));
  2567. break;
  2568. }
  2569. var tabVal = (string)tabAfterText.Attribute(W.val);
  2570. if (tabVal == "right" || tabVal == "end")
  2571. {
  2572. var textAfterElements = contentToMeasure
  2573. .Skip(currentElementIdx + 1);
  2574. // take all the content until another tab, br, or cr
  2575. var textElementsToMeasure = textAfterElements
  2576. .TakeWhile(z =>
  2577. z.Name != W.tab &&
  2578. z.Name != W.br &&
  2579. z.Name != W.cr)
  2580. .ToList();
  2581. var textAfterTab = textElementsToMeasure
  2582. .Where(z => z.Name == W.t)
  2583. .Select(t => (string)t)
  2584. .StringConcatenate();
  2585. var dummyRun2 = new XElement(W.r,
  2586. fontNameAtt,
  2587. runContainingTabToReplace.Elements(W.rPr),
  2588. new XElement(W.t, textAfterTab));
  2589. var widthOfTextAfterTab = CalcWidthOfRunInTwips(dummyRun2);
  2590. var delta2 = (int)tabAfterText.Attribute(W.pos) - widthOfTextAfterTab - twipCounter;
  2591. if (delta2 < 0)
  2592. delta2 = 0;
  2593. currentElement.Add(
  2594. new XAttribute(PtOpenXml.TabWidth,
  2595. string.Format(NumberFormatInfo.InvariantInfo, "{0:0.000}", (decimal)delta2 / 1440m)),
  2596. GetLeader(tabAfterText));
  2597. twipCounter = Math.Max((int)tabAfterText.Attribute(W.pos), twipCounter + widthOfTextAfterTab);
  2598. var lastElement = textElementsToMeasure.LastOrDefault();
  2599. if (lastElement == null)
  2600. break; // we're done
  2601. currentElementIdx = Array.IndexOf(contentToMeasure, lastElement) + 1;
  2602. if (currentElementIdx >= contentToMeasure.Length)
  2603. break; // we're done
  2604. continue;
  2605. }
  2606. if (tabVal == "decimal")
  2607. {
  2608. var textAfterElements = contentToMeasure
  2609. .Skip(currentElementIdx + 1);
  2610. // take all the content until another tab, br, or cr
  2611. var textElementsToMeasure = textAfterElements
  2612. .TakeWhile(z =>
  2613. z.Name != W.tab &&
  2614. z.Name != W.br &&
  2615. z.Name != W.cr)
  2616. .ToList();
  2617. var textAfterTab = textElementsToMeasure
  2618. .Where(z => z.Name == W.t)
  2619. .Select(t => (string)t)
  2620. .StringConcatenate();
  2621. if (textAfterTab.Contains("."))
  2622. {
  2623. var mantissa = textAfterTab.Split('.')[0];
  2624. var dummyRun4 = new XElement(W.r,
  2625. fontNameAtt,
  2626. runContainingTabToReplace.Elements(W.rPr),
  2627. new XElement(W.t, mantissa));
  2628. var widthOfMantissa = CalcWidthOfRunInTwips(dummyRun4);
  2629. var delta2 = (int)tabAfterText.Attribute(W.pos) - widthOfMantissa - twipCounter;
  2630. if (delta2 < 0)
  2631. delta2 = 0;
  2632. currentElement.Add(
  2633. new XAttribute(PtOpenXml.TabWidth,
  2634. string.Format(NumberFormatInfo.InvariantInfo, "{0:0.000}", (decimal)delta2 / 1440m)),
  2635. GetLeader(tabAfterText));
  2636. var decims = textAfterTab.Substring(textAfterTab.IndexOf('.'));
  2637. dummyRun4 = new XElement(W.r,
  2638. fontNameAtt,
  2639. runContainingTabToReplace.Elements(W.rPr),
  2640. new XElement(W.t, decims));
  2641. var widthOfDecims = CalcWidthOfRunInTwips(dummyRun4);
  2642. twipCounter = Math.Max((int)tabAfterText.Attribute(W.pos) + widthOfDecims, twipCounter + widthOfMantissa + widthOfDecims);
  2643. var lastElement = textElementsToMeasure.LastOrDefault();
  2644. if (lastElement == null)
  2645. break; // we're done
  2646. currentElementIdx = Array.IndexOf(contentToMeasure, lastElement) + 1;
  2647. if (currentElementIdx >= contentToMeasure.Length)
  2648. break; // we're done
  2649. continue;
  2650. }
  2651. else
  2652. {
  2653. var dummyRun2 = new XElement(W.r,
  2654. fontNameAtt,
  2655. runContainingTabToReplace.Elements(W.rPr),
  2656. new XElement(W.t, textAfterTab));
  2657. var widthOfTextAfterTab = CalcWidthOfRunInTwips(dummyRun2);
  2658. var delta2 = (int)tabAfterText.Attribute(W.pos) - widthOfTextAfterTab - twipCounter;
  2659. if (delta2 < 0)
  2660. delta2 = 0;
  2661. currentElement.Add(
  2662. new XAttribute(PtOpenXml.TabWidth,
  2663. string.Format(NumberFormatInfo.InvariantInfo, "{0:0.000}", (decimal)delta2 / 1440m)),
  2664. GetLeader(tabAfterText));
  2665. twipCounter = Math.Max((int)tabAfterText.Attribute(W.pos), twipCounter + widthOfTextAfterTab);
  2666. var lastElement = textElementsToMeasure.LastOrDefault();
  2667. if (lastElement == null)
  2668. break; // we're done
  2669. currentElementIdx = Array.IndexOf(contentToMeasure, lastElement) + 1;
  2670. if (currentElementIdx >= contentToMeasure.Length)
  2671. break; // we're done
  2672. continue;
  2673. }
  2674. }
  2675. if ((string)tabAfterText.Attribute(W.val) == "center")
  2676. {
  2677. var textAfterElements = contentToMeasure
  2678. .Skip(currentElementIdx + 1);
  2679. // take all the content until another tab, br, or cr
  2680. var textElementsToMeasure = textAfterElements
  2681. .TakeWhile(z =>
  2682. z.Name != W.tab &&
  2683. z.Name != W.br &&
  2684. z.Name != W.cr)
  2685. .ToList();
  2686. var textAfterTab = textElementsToMeasure
  2687. .Where(z => z.Name == W.t)
  2688. .Select(t => (string)t)
  2689. .StringConcatenate();
  2690. var dummyRun4 = new XElement(W.r,
  2691. fontNameAtt,
  2692. runContainingTabToReplace.Elements(W.rPr),
  2693. new XElement(W.t, textAfterTab));
  2694. var widthOfText = CalcWidthOfRunInTwips(dummyRun4);
  2695. var delta2 = (int)tabAfterText.Attribute(W.pos) - (widthOfText / 2) - twipCounter;
  2696. if (delta2 < 0)
  2697. delta2 = 0;
  2698. currentElement.Add(
  2699. new XAttribute(PtOpenXml.TabWidth,
  2700. string.Format(NumberFormatInfo.InvariantInfo, "{0:0.000}", (decimal)delta2 / 1440m)),
  2701. GetLeader(tabAfterText));
  2702. twipCounter = Math.Max((int)tabAfterText.Attribute(W.pos) + widthOfText / 2, twipCounter + widthOfText);
  2703. var lastElement = textElementsToMeasure.LastOrDefault();
  2704. if (lastElement == null)
  2705. break; // we're done
  2706. currentElementIdx = Array.IndexOf(contentToMeasure, lastElement) + 1;
  2707. if (currentElementIdx >= contentToMeasure.Length)
  2708. break; // we're done
  2709. continue;
  2710. }
  2711. if (tabVal == "left" || tabVal == "start" || tabVal == "num")
  2712. {
  2713. var delta = (int)tabAfterText.Attribute(W.pos) - twipCounter;
  2714. currentElement.Add(
  2715. new XAttribute(PtOpenXml.TabWidth,
  2716. string.Format(NumberFormatInfo.InvariantInfo, "{0:0.000}", (decimal)delta / 1440m)),
  2717. GetLeader(tabAfterText));
  2718. twipCounter = (int)tabAfterText.Attribute(W.pos);
  2719. currentElementIdx++;
  2720. if (currentElementIdx >= contentToMeasure.Length)
  2721. break; // we're done
  2722. continue;
  2723. }
  2724. }
  2725. if (currentElement.Name == W.t)
  2726. {
  2727. // TODO: Revisit. This is a quick fix because it doesn't work on Azure.
  2728. // Given the changes we've made elsewhere, though, this is not required
  2729. // for the first tab at least. We could also enhance that other change
  2730. // to deal with all tabs.
  2731. //var runContainingTabToReplace = currentElement.Parent;
  2732. //var paragraphForRun = runContainingTabToReplace.Ancestors(W.p).First();
  2733. //var fontNameAtt = runContainingTabToReplace.Attribute(PtOpenXml.FontName) ??
  2734. // paragraphForRun.Attribute(PtOpenXml.FontName);
  2735. //var languageTypeAtt = runContainingTabToReplace.Attribute(PtOpenXml.LanguageType) ??
  2736. // paragraphForRun.Attribute(PtOpenXml.LanguageType);
  2737. //var dummyRun3 = new XElement(W.r, fontNameAtt, languageTypeAtt,
  2738. // runContainingTabToReplace.Elements(W.rPr),
  2739. // currentElement);
  2740. //var widthOfText = CalcWidthOfRunInTwips(dummyRun3);
  2741. const int widthOfText = 0;
  2742. currentElement.Add(new XAttribute(PtOpenXml.TabWidth,
  2743. string.Format(NumberFormatInfo.InvariantInfo, "{0:0.000}", (decimal)widthOfText / 1440m)));
  2744. twipCounter += widthOfText;
  2745. currentElementIdx++;
  2746. if (currentElementIdx >= contentToMeasure.Length)
  2747. break; // we're done
  2748. continue;
  2749. }
  2750. currentElementIdx++;
  2751. if (currentElementIdx >= contentToMeasure.Length)
  2752. break; // we're done
  2753. }
  2754. return new XElement(element.Name,
  2755. element.Attributes(),
  2756. element.Nodes().Select(n => CalculateSpanWidthTransform(n, defaultTabStop)));
  2757. }
  2758. private static XAttribute GetLeader(XElement tabAfterText)
  2759. {
  2760. var leader = (string)tabAfterText.Attribute(W.leader);
  2761. if (leader == null)
  2762. return null;
  2763. return new XAttribute(PtOpenXml.Leader, leader);
  2764. }
  2765. private static XElement AddDefaultTabsAfterLastTab(XElement tabs, int defaultTabStop)
  2766. {
  2767. var lastTabElement = tabs
  2768. .Elements(W.tab)
  2769. .Where(t => (string)t.Attribute(W.val) != "clear" && (string)t.Attribute(W.val) != "bar")
  2770. .OrderBy(t => (int)t.Attribute(W.pos))
  2771. .LastOrDefault();
  2772. if (lastTabElement != null)
  2773. {
  2774. if (defaultTabStop == 0)
  2775. defaultTabStop = 720;
  2776. var rangeStart = (int)lastTabElement.Attribute(W.pos) / defaultTabStop + 1;
  2777. var tempTabs = new XElement(W.tabs,
  2778. tabs.Elements().Where(t => (string)t.Attribute(W.val) != "clear" && (string)t.Attribute(W.val) != "bar"),
  2779. Enumerable.Range(rangeStart, 100)
  2780. .Select(r => new XElement(W.tab,
  2781. new XAttribute(W.val, "left"),
  2782. new XAttribute(W.pos, r * defaultTabStop))));
  2783. tempTabs = new XElement(W.tabs,
  2784. tempTabs.Elements().OrderBy(t => (int)t.Attribute(W.pos)));
  2785. return tempTabs;
  2786. }
  2787. else
  2788. {
  2789. tabs = new XElement(W.tabs,
  2790. Enumerable.Range(1, 100)
  2791. .Select(r => new XElement(W.tab,
  2792. new XAttribute(W.val, "left"),
  2793. new XAttribute(W.pos, r * defaultTabStop))));
  2794. }
  2795. return tabs;
  2796. }
  2797. private static readonly HashSet<string> UnknownFonts = new HashSet<string>();
  2798. private static HashSet<string> _knownFamilies;
  2799. private static HashSet<string> KnownFamilies
  2800. {
  2801. get
  2802. {
  2803. if (_knownFamilies == null)
  2804. {
  2805. _knownFamilies = new HashSet<string>();
  2806. var families = FontFamily.Families;
  2807. foreach (var fam in families)
  2808. _knownFamilies.Add(fam.Name);
  2809. }
  2810. return _knownFamilies;
  2811. }
  2812. }
  2813. private static int CalcWidthOfRunInTwips(XElement r)
  2814. {
  2815. var fontName = (string)r.Attribute(PtOpenXml.pt + "FontName") ??
  2816. (string)r.Ancestors(W.p).First().Attribute(PtOpenXml.pt + "FontName");
  2817. if (fontName == null)
  2818. throw new OpenXmlPowerToolsException("Internal Error, should have FontName attribute");
  2819. if (UnknownFonts.Contains(fontName))
  2820. return 0;
  2821. var rPr = r.Element(W.rPr);
  2822. if (rPr == null)
  2823. throw new OpenXmlPowerToolsException("Internal Error, should have run properties");
  2824. var sz = GetFontSize(r) ?? 22m;
  2825. // unknown font families will throw ArgumentException, in which case just return 0
  2826. if (!KnownFamilies.Contains(fontName))
  2827. return 0;
  2828. // in theory, all unknown fonts are found by the above test, but if not...
  2829. FontFamily ff;
  2830. try
  2831. {
  2832. ff = new FontFamily(fontName);
  2833. }
  2834. catch (ArgumentException)
  2835. {
  2836. UnknownFonts.Add(fontName);
  2837. return 0;
  2838. }
  2839. var fs = FontStyle.Regular;
  2840. if (GetBoolProp(rPr, W.b) || GetBoolProp(rPr, W.bCs))
  2841. fs |= FontStyle.Bold;
  2842. if (GetBoolProp(rPr, W.i) || GetBoolProp(rPr, W.iCs))
  2843. fs |= FontStyle.Italic;
  2844. // Appended blank as a quick fix to accommodate &nbsp; that will get
  2845. // appended to some layout-critical runs such as list item numbers.
  2846. // In some cases, this might not be required or even wrong, so this
  2847. // must be revisited.
  2848. // TODO: Revisit.
  2849. var runText = r.DescendantsTrimmed(W.txbxContent)
  2850. .Where(e => e.Name == W.t)
  2851. .Select(t => (string)t)
  2852. .StringConcatenate() + " ";
  2853. var tabLength = r.DescendantsTrimmed(W.txbxContent)
  2854. .Where(e => e.Name == W.tab)
  2855. .Select(t => (decimal)t.Attribute(PtOpenXml.TabWidth))
  2856. .Sum();
  2857. if (runText.Length == 0 && tabLength == 0)
  2858. return 0;
  2859. int multiplier = 1;
  2860. if (runText.Length <= 2)
  2861. multiplier = 100;
  2862. else if (runText.Length <= 4)
  2863. multiplier = 50;
  2864. else if (runText.Length <= 8)
  2865. multiplier = 25;
  2866. else if (runText.Length <= 16)
  2867. multiplier = 12;
  2868. else if (runText.Length <= 32)
  2869. multiplier = 6;
  2870. if (multiplier != 1)
  2871. {
  2872. StringBuilder sb = new StringBuilder();
  2873. for (int i = 0; i < multiplier; i++)
  2874. sb.Append(runText);
  2875. runText = sb.ToString();
  2876. }
  2877. var w = MetricsGetter.GetTextWidth(ff, fs, sz, runText);
  2878. return (int)(w / 96m * 1440m / multiplier + tabLength * 1440m);
  2879. }
  2880. private static void InsertAppropriateNonbreakingSpaces(WordprocessingDocument wordDoc)
  2881. {
  2882. foreach (var part in wordDoc.ContentParts())
  2883. {
  2884. var pxd = part.GetXDocument();
  2885. var root = pxd.Root;
  2886. if (root == null) return;
  2887. var newRoot = (XElement)InsertAppropriateNonbreakingSpacesTransform(root);
  2888. root.ReplaceWith(newRoot);
  2889. part.PutXDocument();
  2890. }
  2891. }
  2892. // Non-breaking spaces are not required if we use appropriate CSS, i.e., "white-space: pre-wrap;".
  2893. // We only need to make sure that empty w:p elements are translated into non-empty h:p elements,
  2894. // because empty h:p elements would be ignored by browsers.
  2895. // Further, in addition to not being required, non-breaking spaces would change the layout behavior
  2896. // of spans having consecutive spaces. Therefore, avoiding non-breaking spaces has the additional
  2897. // benefit of leading to a more faithful representation of the Word document in HTML.
  2898. private static object InsertAppropriateNonbreakingSpacesTransform(XNode node)
  2899. {
  2900. XElement element = node as XElement;
  2901. if (element != null)
  2902. {
  2903. // child content of run to look for
  2904. // W.br
  2905. // W.cr
  2906. // W.dayLong
  2907. // W.dayShort
  2908. // W.drawing
  2909. // W.monthLong
  2910. // W.monthShort
  2911. // W.noBreakHyphen
  2912. // W.object
  2913. // W.pgNum
  2914. // W.pTab
  2915. // W.separator
  2916. // W.softHyphen
  2917. // W.sym
  2918. // W.t
  2919. // W.tab
  2920. // W.yearLong
  2921. // W.yearShort
  2922. if (element.Name == W.p)
  2923. {
  2924. // Translate empty paragraphs to paragraphs having one run with
  2925. // a normal space. A non-breaking space, i.e., \x00A0, is not
  2926. // required if we use appropriate CSS.
  2927. bool hasContent = element
  2928. .Elements()
  2929. .Where(e => e.Name != W.pPr)
  2930. .DescendantsAndSelf()
  2931. .Any(e =>
  2932. e.Name == W.dayLong ||
  2933. e.Name == W.dayShort ||
  2934. e.Name == W.drawing ||
  2935. e.Name == W.monthLong ||
  2936. e.Name == W.monthShort ||
  2937. e.Name == W.noBreakHyphen ||
  2938. e.Name == W._object ||
  2939. e.Name == W.pgNum ||
  2940. e.Name == W.ptab ||
  2941. e.Name == W.separator ||
  2942. e.Name == W.softHyphen ||
  2943. e.Name == W.sym ||
  2944. e.Name == W.t ||
  2945. e.Name == W.tab ||
  2946. e.Name == W.yearLong ||
  2947. e.Name == W.yearShort
  2948. );
  2949. if (hasContent == false)
  2950. return new XElement(element.Name,
  2951. element.Attributes(),
  2952. element.Nodes().Select(n => InsertAppropriateNonbreakingSpacesTransform(n)),
  2953. new XElement(W.r,
  2954. element.Elements(W.pPr).Elements(W.rPr),
  2955. new XElement(W.t, " ")));
  2956. }
  2957. return new XElement(element.Name,
  2958. element.Attributes(),
  2959. element.Nodes().Select(n => InsertAppropriateNonbreakingSpacesTransform(n)));
  2960. }
  2961. return node;
  2962. }
  2963. private class SectionAnnotation
  2964. {
  2965. public XElement SectionElement;
  2966. }
  2967. private static void AnnotateForSections(WordprocessingDocument wordDoc)
  2968. {
  2969. var xd = wordDoc.MainDocumentPart.GetXDocument();
  2970. var document = xd.Root;
  2971. if (document == null) return;
  2972. var body = document.Element(W.body);
  2973. if (body == null) return;
  2974. // move last sectPr into last paragraph
  2975. var lastSectPr = body.Elements(W.sectPr).LastOrDefault();
  2976. if (lastSectPr != null)
  2977. {
  2978. // if the last thing in the document is a table, Word will always insert a paragraph following that.
  2979. var lastPara = body
  2980. .DescendantsTrimmed(W.txbxContent)
  2981. .LastOrDefault(p => p.Name == W.p);
  2982. if (lastPara != null)
  2983. {
  2984. var lastParaProps = lastPara.Element(W.pPr);
  2985. if (lastParaProps != null)
  2986. lastParaProps.Add(lastSectPr);
  2987. else
  2988. lastPara.Add(new XElement(W.pPr, lastSectPr));
  2989. lastSectPr.Remove();
  2990. }
  2991. }
  2992. var reverseDescendants = xd.Descendants().Reverse().ToList();
  2993. var currentSection = InitializeSectionAnnotation(reverseDescendants);
  2994. foreach (var d in reverseDescendants)
  2995. {
  2996. if (d.Name == W.sectPr)
  2997. {
  2998. if (d.Attribute(XNamespace.Xmlns + "w") == null)
  2999. d.Add(new XAttribute(XNamespace.Xmlns + "w", W.w));
  3000. currentSection = new SectionAnnotation()
  3001. {
  3002. SectionElement = d
  3003. };
  3004. }
  3005. else
  3006. d.AddAnnotation(currentSection);
  3007. }
  3008. }
  3009. private static SectionAnnotation InitializeSectionAnnotation(IEnumerable<XElement> reverseDescendants)
  3010. {
  3011. var currentSection = new SectionAnnotation()
  3012. {
  3013. SectionElement = reverseDescendants.FirstOrDefault(e => e.Name == W.sectPr)
  3014. };
  3015. if (currentSection.SectionElement != null &&
  3016. currentSection.SectionElement.Attribute(XNamespace.Xmlns + "w") == null)
  3017. currentSection.SectionElement.Add(new XAttribute(XNamespace.Xmlns + "w", W.w));
  3018. // todo what should the default section props be?
  3019. if (currentSection.SectionElement == null)
  3020. currentSection = new SectionAnnotation()
  3021. {
  3022. SectionElement = new XElement(W.sectPr,
  3023. new XAttribute(XNamespace.Xmlns + "w", W.w),
  3024. new XElement(W.pgSz,
  3025. new XAttribute(W._w, 12240),
  3026. new XAttribute(W.h, 15840)),
  3027. new XElement(W.pgMar,
  3028. new XAttribute(W.top, 1440),
  3029. new XAttribute(W.right, 1440),
  3030. new XAttribute(W.bottom, 1440),
  3031. new XAttribute(W.left, 1440),
  3032. new XAttribute(W.header, 720),
  3033. new XAttribute(W.footer, 720),
  3034. new XAttribute(W.gutter, 0)),
  3035. new XElement(W.cols,
  3036. new XAttribute(W.space, 720)),
  3037. new XElement(W.docGrid,
  3038. new XAttribute(W.linePitch, 360)))
  3039. };
  3040. return currentSection;
  3041. }
  3042. private static object CreateBorderDivs(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings, IEnumerable<XElement> elements)
  3043. {
  3044. return elements.GroupAdjacent(e =>
  3045. {
  3046. var pBdr = e.Elements(W.pPr).Elements(W.pBdr).FirstOrDefault();
  3047. if (pBdr != null)
  3048. {
  3049. var indStr = string.Empty;
  3050. var ind = e.Elements(W.pPr).Elements(W.ind).FirstOrDefault();
  3051. if (ind != null)
  3052. indStr = ind.ToString(SaveOptions.DisableFormatting);
  3053. return pBdr.ToString(SaveOptions.DisableFormatting) + indStr;
  3054. }
  3055. return e.Name == W.tbl ? "table" : string.Empty;
  3056. })
  3057. .Select(g =>
  3058. {
  3059. if (g.Key == string.Empty)
  3060. {
  3061. return (object)GroupAndVerticallySpaceNumberedParagraphs(wordDoc, settings, g, 0m);
  3062. }
  3063. if (g.Key == "table")
  3064. {
  3065. return g.Select(gc => ConvertToHtmlTransform(wordDoc, settings, gc, false, 0));
  3066. }
  3067. var pPr = g.First().Elements(W.pPr).First();
  3068. var pBdr = pPr.Element(W.pBdr);
  3069. var style = new Dictionary<string, string>();
  3070. GenerateBorderStyle(pBdr, W.top, style, BorderType.Paragraph);
  3071. GenerateBorderStyle(pBdr, W.right, style, BorderType.Paragraph);
  3072. GenerateBorderStyle(pBdr, W.bottom, style, BorderType.Paragraph);
  3073. GenerateBorderStyle(pBdr, W.left, style, BorderType.Paragraph);
  3074. var currentMarginLeft = 0m;
  3075. var ind = pPr.Element(W.ind);
  3076. if (ind != null)
  3077. {
  3078. var leftInInches = (decimal?)ind.Attribute(W.left) / 1440m ?? 0;
  3079. var hangingInInches = -(decimal?)ind.Attribute(W.hanging) / 1440m ?? 0;
  3080. currentMarginLeft = leftInInches + hangingInInches;
  3081. style.AddIfMissing("margin-left",
  3082. currentMarginLeft > 0m
  3083. ? string.Format(NumberFormatInfo.InvariantInfo, "{0:0.00}in", currentMarginLeft)
  3084. : "0");
  3085. }
  3086. var div = new XElement(Xhtml.div,
  3087. GroupAndVerticallySpaceNumberedParagraphs(wordDoc, settings, g, currentMarginLeft));
  3088. div.AddAnnotation(style);
  3089. return div;
  3090. })
  3091. .ToList();
  3092. }
  3093. private static IEnumerable<object> GroupAndVerticallySpaceNumberedParagraphs(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings,
  3094. IEnumerable<XElement> elements, decimal currentMarginLeft)
  3095. {
  3096. var grouped = elements
  3097. .GroupAdjacent(e =>
  3098. {
  3099. var abstractNumId = (string)e.Attribute(PtOpenXml.pt + "AbstractNumId");
  3100. if (abstractNumId != null)
  3101. return "num:" + abstractNumId;
  3102. var contextualSpacing = e.Elements(W.pPr).Elements(W.contextualSpacing).FirstOrDefault();
  3103. if (contextualSpacing != null)
  3104. {
  3105. var styleName = (string)e.Elements(W.pPr).Elements(W.pStyle).Attributes(W.val).FirstOrDefault();
  3106. if (styleName == null)
  3107. return "";
  3108. return "sty:" + styleName;
  3109. }
  3110. return "";
  3111. })
  3112. .ToList();
  3113. var newContent = grouped
  3114. .Select(g =>
  3115. {
  3116. if (g.Key == "")
  3117. return g.Select(e => ConvertToHtmlTransform(wordDoc, settings, e, false, currentMarginLeft));
  3118. var last = g.Count() - 1;
  3119. return g.Select((e, i) => ConvertToHtmlTransform(wordDoc, settings, e, i != last, currentMarginLeft));
  3120. });
  3121. return (IEnumerable<object>)newContent;
  3122. }
  3123. private class BorderMappingInfo
  3124. {
  3125. public string CssName;
  3126. public decimal CssSize;
  3127. }
  3128. private static readonly Dictionary<string, BorderMappingInfo> BorderStyleMap = new Dictionary<string, BorderMappingInfo>()
  3129. {
  3130. { "single", new BorderMappingInfo() { CssName = "solid", CssSize = 1.0m }},
  3131. { "dotted", new BorderMappingInfo() { CssName = "dotted", CssSize = 1.0m }},
  3132. { "dashSmallGap", new BorderMappingInfo() { CssName = "dashed", CssSize = 1.0m }},
  3133. { "dashed", new BorderMappingInfo() { CssName = "dashed", CssSize = 1.0m }},
  3134. { "dotDash", new BorderMappingInfo() { CssName = "dashed", CssSize = 1.0m }},
  3135. { "dotDotDash", new BorderMappingInfo() { CssName = "dashed", CssSize = 1.0m }},
  3136. { "double", new BorderMappingInfo() { CssName = "double", CssSize = 2.5m }},
  3137. { "triple", new BorderMappingInfo() { CssName = "double", CssSize = 2.5m }},
  3138. { "thinThickSmallGap", new BorderMappingInfo() { CssName = "double", CssSize = 4.5m }},
  3139. { "thickThinSmallGap", new BorderMappingInfo() { CssName = "double", CssSize = 4.5m }},
  3140. { "thinThickThinSmallGap", new BorderMappingInfo() { CssName = "double", CssSize = 6.0m }},
  3141. { "thickThinMediumGap", new BorderMappingInfo() { CssName = "double", CssSize = 6.0m }},
  3142. { "thinThickMediumGap", new BorderMappingInfo() { CssName = "double", CssSize = 6.0m }},
  3143. { "thinThickThinMediumGap", new BorderMappingInfo() { CssName = "double", CssSize = 9.0m }},
  3144. { "thinThickLargeGap", new BorderMappingInfo() { CssName = "double", CssSize = 5.25m }},
  3145. { "thickThinLargeGap", new BorderMappingInfo() { CssName = "double", CssSize = 5.25m }},
  3146. { "thinThickThinLargeGap", new BorderMappingInfo() { CssName = "double", CssSize = 9.0m }},
  3147. { "wave", new BorderMappingInfo() { CssName = "solid", CssSize = 3.0m }},
  3148. { "doubleWave", new BorderMappingInfo() { CssName = "double", CssSize = 5.25m }},
  3149. { "dashDotStroked", new BorderMappingInfo() { CssName = "solid", CssSize = 3.0m }},
  3150. { "threeDEmboss", new BorderMappingInfo() { CssName = "ridge", CssSize = 6.0m }},
  3151. { "threeDEngrave", new BorderMappingInfo() { CssName = "groove", CssSize = 6.0m }},
  3152. { "outset", new BorderMappingInfo() { CssName = "outset", CssSize = 4.5m }},
  3153. { "inset", new BorderMappingInfo() { CssName = "inset", CssSize = 4.5m }},
  3154. };
  3155. private static void GenerateBorderStyle(XElement pBdr, XName sideXName, Dictionary<string, string> style, BorderType borderType)
  3156. {
  3157. string whichSide;
  3158. if (sideXName == W.top)
  3159. whichSide = "top";
  3160. else if (sideXName == W.right)
  3161. whichSide = "right";
  3162. else if (sideXName == W.bottom)
  3163. whichSide = "bottom";
  3164. else
  3165. whichSide = "left";
  3166. if (pBdr == null)
  3167. {
  3168. style.Add("border-" + whichSide, "none");
  3169. if (borderType == BorderType.Cell &&
  3170. (whichSide == "left" || whichSide == "right"))
  3171. style.Add("padding-" + whichSide, "5.4pt");
  3172. return;
  3173. }
  3174. var side = pBdr.Element(sideXName);
  3175. if (side == null)
  3176. {
  3177. style.Add("border-" + whichSide, "none");
  3178. if (borderType == BorderType.Cell &&
  3179. (whichSide == "left" || whichSide == "right"))
  3180. style.Add("padding-" + whichSide, "5.4pt");
  3181. return;
  3182. }
  3183. var type = (string)side.Attribute(W.val);
  3184. if (type == "nil" || type == "none")
  3185. {
  3186. style.Add("border-" + whichSide + "-style", "none");
  3187. var space = (decimal?)side.Attribute(W.space) ?? 0;
  3188. if (borderType == BorderType.Cell &&
  3189. (whichSide == "left" || whichSide == "right"))
  3190. if (space < 5.4m)
  3191. space = 5.4m;
  3192. style.Add("padding-" + whichSide,
  3193. space == 0 ? "0" : string.Format(NumberFormatInfo.InvariantInfo, "{0:0.0}pt", space));
  3194. }
  3195. else
  3196. {
  3197. var sz = (int)side.Attribute(W.sz);
  3198. var space = (decimal?)side.Attribute(W.space) ?? 0;
  3199. var color = (string)side.Attribute(W.color);
  3200. if (color == null || color == "auto")
  3201. color = "windowtext";
  3202. else
  3203. color = ConvertColor(color);
  3204. decimal borderWidthInPoints = Math.Max(1m, Math.Min(96m, Math.Max(2m, sz)) / 8m);
  3205. var borderStyle = "solid";
  3206. if (BorderStyleMap.ContainsKey(type))
  3207. {
  3208. var borderInfo = BorderStyleMap[type];
  3209. borderStyle = borderInfo.CssName;
  3210. if (type == "double")
  3211. {
  3212. if (sz <= 8)
  3213. borderWidthInPoints = 2.5m;
  3214. else if (sz <= 18)
  3215. borderWidthInPoints = 6.75m;
  3216. else
  3217. borderWidthInPoints = sz / 3m;
  3218. }
  3219. else if (type == "triple")
  3220. {
  3221. if (sz <= 8)
  3222. borderWidthInPoints = 8m;
  3223. else if (sz <= 18)
  3224. borderWidthInPoints = 11.25m;
  3225. else
  3226. borderWidthInPoints = 11.25m;
  3227. }
  3228. else if (type.ToLower().Contains("dash"))
  3229. {
  3230. if (sz <= 4)
  3231. borderWidthInPoints = 1m;
  3232. else if (sz <= 12)
  3233. borderWidthInPoints = 1.5m;
  3234. else
  3235. borderWidthInPoints = 2m;
  3236. }
  3237. else if (type != "single")
  3238. borderWidthInPoints = borderInfo.CssSize;
  3239. }
  3240. if (type == "outset" || type == "inset")
  3241. color = "";
  3242. var borderWidth = string.Format(NumberFormatInfo.InvariantInfo, "{0:0.0}pt", borderWidthInPoints);
  3243. style.Add("border-" + whichSide, borderStyle + " " + color + " " + borderWidth);
  3244. if (borderType == BorderType.Cell &&
  3245. (whichSide == "left" || whichSide == "right"))
  3246. if (space < 5.4m)
  3247. space = 5.4m;
  3248. style.Add("padding-" + whichSide,
  3249. space == 0 ? "0" : string.Format(NumberFormatInfo.InvariantInfo, "{0:0.0}pt", space));
  3250. }
  3251. }
  3252. private static readonly Dictionary<string, Func<string, string, string>> ShadeMapper = new Dictionary<string, Func<string, string, string>>()
  3253. {
  3254. { "auto", (c, f) => c },
  3255. { "clear", (c, f) => f },
  3256. { "nil", (c, f) => f },
  3257. { "solid", (c, f) => c },
  3258. { "diagCross", (c, f) => ConvertColorFillPct(c, f, .75) },
  3259. { "diagStripe", (c, f) => ConvertColorFillPct(c, f, .75) },
  3260. { "horzCross", (c, f) => ConvertColorFillPct(c, f, .5) },
  3261. { "horzStripe", (c, f) => ConvertColorFillPct(c, f, .5) },
  3262. { "pct10", (c, f) => ConvertColorFillPct(c, f, .1) },
  3263. { "pct12", (c, f) => ConvertColorFillPct(c, f, .125) },
  3264. { "pct15", (c, f) => ConvertColorFillPct(c, f, .15) },
  3265. { "pct20", (c, f) => ConvertColorFillPct(c, f, .2) },
  3266. { "pct25", (c, f) => ConvertColorFillPct(c, f, .25) },
  3267. { "pct30", (c, f) => ConvertColorFillPct(c, f, .3) },
  3268. { "pct35", (c, f) => ConvertColorFillPct(c, f, .35) },
  3269. { "pct37", (c, f) => ConvertColorFillPct(c, f, .375) },
  3270. { "pct40", (c, f) => ConvertColorFillPct(c, f, .4) },
  3271. { "pct45", (c, f) => ConvertColorFillPct(c, f, .45) },
  3272. { "pct50", (c, f) => ConvertColorFillPct(c, f, .50) },
  3273. { "pct55", (c, f) => ConvertColorFillPct(c, f, .55) },
  3274. { "pct60", (c, f) => ConvertColorFillPct(c, f, .60) },
  3275. { "pct62", (c, f) => ConvertColorFillPct(c, f, .625) },
  3276. { "pct65", (c, f) => ConvertColorFillPct(c, f, .65) },
  3277. { "pct70", (c, f) => ConvertColorFillPct(c, f, .7) },
  3278. { "pct75", (c, f) => ConvertColorFillPct(c, f, .75) },
  3279. { "pct80", (c, f) => ConvertColorFillPct(c, f, .8) },
  3280. { "pct85", (c, f) => ConvertColorFillPct(c, f, .85) },
  3281. { "pct87", (c, f) => ConvertColorFillPct(c, f, .875) },
  3282. { "pct90", (c, f) => ConvertColorFillPct(c, f, .9) },
  3283. { "pct95", (c, f) => ConvertColorFillPct(c, f, .95) },
  3284. { "reverseDiagStripe", (c, f) => ConvertColorFillPct(c, f, .5) },
  3285. { "thinDiagCross", (c, f) => ConvertColorFillPct(c, f, .5) },
  3286. { "thinDiagStripe", (c, f) => ConvertColorFillPct(c, f, .25) },
  3287. { "thinHorzCross", (c, f) => ConvertColorFillPct(c, f, .3) },
  3288. { "thinHorzStripe", (c, f) => ConvertColorFillPct(c, f, .25) },
  3289. { "thinReverseDiagStripe", (c, f) => ConvertColorFillPct(c, f, .25) },
  3290. { "thinVertStripe", (c, f) => ConvertColorFillPct(c, f, .25) },
  3291. };
  3292. private static readonly Dictionary<string, string> ShadeCache = new Dictionary<string, string>();
  3293. // fill is the background, color is the foreground
  3294. private static string ConvertColorFillPct(string color, string fill, double pct)
  3295. {
  3296. if (color == "auto")
  3297. color = "000000";
  3298. if (fill == "auto")
  3299. fill = "ffffff";
  3300. var key = color + fill + pct.ToString(CultureInfo.InvariantCulture);
  3301. if (ShadeCache.ContainsKey(key))
  3302. return ShadeCache[key];
  3303. var fillRed = Convert.ToInt32(fill.Substring(0, 2), 16);
  3304. var fillGreen = Convert.ToInt32(fill.Substring(2, 2), 16);
  3305. var fillBlue = Convert.ToInt32(fill.Substring(4, 2), 16);
  3306. var colorRed = Convert.ToInt32(color.Substring(0, 2), 16);
  3307. var colorGreen = Convert.ToInt32(color.Substring(2, 2), 16);
  3308. var colorBlue = Convert.ToInt32(color.Substring(4, 2), 16);
  3309. var finalRed = (int)(fillRed - (fillRed - colorRed) * pct);
  3310. var finalGreen = (int)(fillGreen - (fillGreen - colorGreen) * pct);
  3311. var finalBlue = (int)(fillBlue - (fillBlue - colorBlue) * pct);
  3312. var returnValue = string.Format("{0:x2}{1:x2}{2:x2}", finalRed, finalGreen, finalBlue);
  3313. ShadeCache.Add(key, returnValue);
  3314. return returnValue;
  3315. }
  3316. private static void CreateStyleFromShd(Dictionary<string, string> style, XElement shd)
  3317. {
  3318. if (shd == null)
  3319. return;
  3320. var shadeType = (string)shd.Attribute(W.val);
  3321. var color = (string)shd.Attribute(W.color);
  3322. var fill = (string)shd.Attribute(W.fill);
  3323. if (ShadeMapper.ContainsKey(shadeType))
  3324. {
  3325. color = ShadeMapper[shadeType](color, fill);
  3326. }
  3327. if (color != null)
  3328. {
  3329. var cvtColor = ConvertColor(color);
  3330. if (!string.IsNullOrEmpty(cvtColor))
  3331. style.AddIfMissing("background", cvtColor);
  3332. }
  3333. }
  3334. private static readonly Dictionary<string, string> NamedColors = new Dictionary<string, string>()
  3335. {
  3336. {"black", "black"},
  3337. {"blue", "blue" },
  3338. {"cyan", "aqua" },
  3339. {"green", "green" },
  3340. {"magenta", "fuchsia" },
  3341. {"red", "red" },
  3342. {"yellow", "yellow" },
  3343. {"white", "white" },
  3344. {"darkBlue", "#00008B" },
  3345. {"darkCyan", "#008B8B" },
  3346. {"darkGreen", "#006400" },
  3347. {"darkMagenta", "#800080" },
  3348. {"darkRed", "#8B0000" },
  3349. {"darkYellow", "#808000" },
  3350. {"darkGray", "#A9A9A9" },
  3351. {"lightGray", "#D3D3D3" },
  3352. {"none", "" },
  3353. };
  3354. private static void CreateColorProperty(string propertyName, string color, Dictionary<string, string> style)
  3355. {
  3356. if (color == null)
  3357. return;
  3358. // "auto" color is black for "color" and white for "background" property.
  3359. if (color == "auto")
  3360. color = propertyName == "color" ? "black" : "white";
  3361. if (NamedColors.ContainsKey(color))
  3362. {
  3363. var lc = NamedColors[color];
  3364. if (lc == "")
  3365. return;
  3366. style.AddIfMissing(propertyName, lc);
  3367. return;
  3368. }
  3369. style.AddIfMissing(propertyName, "#" + color);
  3370. }
  3371. private static string ConvertColor(string color)
  3372. {
  3373. // "auto" color is black for "color" and white for "background" property.
  3374. // As this method is only called for "background" colors, "auto" is translated
  3375. // to "white" and never "black".
  3376. if (color == "auto")
  3377. color = "white";
  3378. if (NamedColors.ContainsKey(color))
  3379. {
  3380. var lc = NamedColors[color];
  3381. if (lc == "")
  3382. return "black";
  3383. return lc;
  3384. }
  3385. return "#" + color;
  3386. }
  3387. private static readonly Dictionary<string, string> FontFallback = new Dictionary<string, string>()
  3388. {
  3389. { "Arial", @"'{0}', 'sans-serif'" },
  3390. { "Arial Narrow", @"'{0}', 'sans-serif'" },
  3391. { "Arial Rounded MT Bold", @"'{0}', 'sans-serif'" },
  3392. { "Arial Unicode MS", @"'{0}', 'sans-serif'" },
  3393. { "Baskerville Old Face", @"'{0}', 'serif'" },
  3394. { "Berlin Sans FB", @"'{0}', 'sans-serif'" },
  3395. { "Berlin Sans FB Demi", @"'{0}', 'sans-serif'" },
  3396. { "Calibri Light", @"'{0}', 'sans-serif'" },
  3397. { "Gill Sans MT", @"'{0}', 'sans-serif'" },
  3398. { "Gill Sans MT Condensed", @"'{0}', 'sans-serif'" },
  3399. { "Lucida Sans", @"'{0}', 'sans-serif'" },
  3400. { "Lucida Sans Unicode", @"'{0}', 'sans-serif'" },
  3401. { "Segoe UI", @"'{0}', 'sans-serif'" },
  3402. { "Segoe UI Light", @"'{0}', 'sans-serif'" },
  3403. { "Segoe UI Semibold", @"'{0}', 'sans-serif'" },
  3404. { "Tahoma", @"'{0}', 'sans-serif'" },
  3405. { "Trebuchet MS", @"'{0}', 'sans-serif'" },
  3406. { "Verdana", @"'{0}', 'sans-serif'" },
  3407. { "Book Antiqua", @"'{0}', 'serif'" },
  3408. { "Bookman Old Style", @"'{0}', 'serif'" },
  3409. { "Californian FB", @"'{0}', 'serif'" },
  3410. { "Cambria", @"'{0}', 'serif'" },
  3411. { "Constantia", @"'{0}', 'serif'" },
  3412. { "Garamond", @"'{0}', 'serif'" },
  3413. { "Lucida Bright", @"'{0}', 'serif'" },
  3414. { "Lucida Fax", @"'{0}', 'serif'" },
  3415. { "Palatino Linotype", @"'{0}', 'serif'" },
  3416. { "Times New Roman", @"'{0}', 'serif'" },
  3417. { "Wide Latin", @"'{0}', 'serif'" },
  3418. { "Courier New", @"'{0}'" },
  3419. { "Lucida Console", @"'{0}'" },
  3420. };
  3421. private static void CreateFontCssProperty(string font, Dictionary<string, string> style)
  3422. {
  3423. if (FontFallback.ContainsKey(font))
  3424. {
  3425. style.AddIfMissing("font-family", string.Format(FontFallback[font], font));
  3426. return;
  3427. }
  3428. style.AddIfMissing("font-family", font);
  3429. }
  3430. private static bool GetBoolProp(XElement runProps, XName xName)
  3431. {
  3432. var p = runProps.Element(xName);
  3433. if (p == null)
  3434. return false;
  3435. var v = p.Attribute(W.val);
  3436. if (v == null)
  3437. return true;
  3438. var s = v.Value.ToLower();
  3439. if (s == "0" || s == "false")
  3440. return false;
  3441. if (s == "1" || s == "true")
  3442. return true;
  3443. return false;
  3444. }
  3445. private static object ConvertContentThatCanContainFields(WordprocessingDocument wordDoc, WmlToHtmlConverterSettings settings,
  3446. IEnumerable<XElement> elements)
  3447. {
  3448. var grouped = elements
  3449. .GroupAdjacent(e =>
  3450. {
  3451. var stack = e.Annotation<Stack<FieldRetriever.FieldElementTypeInfo>>();
  3452. return stack == null || !stack.Any() ? (int?)null : stack.Select(st => st.Id).Min();
  3453. })
  3454. .ToList();
  3455. var txformed = grouped
  3456. .Select(g =>
  3457. {
  3458. var key = g.Key;
  3459. if (key == null)
  3460. return (object)g.Select(n => ConvertToHtmlTransform(wordDoc, settings, n, false, 0m));
  3461. var instrText = FieldRetriever.InstrText(g.First().Ancestors().Last(), (int)key)
  3462. .TrimStart('{').TrimEnd('}');
  3463. var parsed = FieldRetriever.ParseField(instrText);
  3464. if (parsed.FieldType != "HYPERLINK")
  3465. return g.Select(n => ConvertToHtmlTransform(wordDoc, settings, n, false, 0m));
  3466. var content = g.DescendantsAndSelf(W.r).Select(run => ConvertRun(wordDoc, settings, run));
  3467. var a = parsed.Arguments.Length > 0
  3468. ? new XElement(Xhtml.a, new XAttribute("href", parsed.Arguments[0]), content)
  3469. : new XElement(Xhtml.a, content);
  3470. var a2 = a as XElement;
  3471. if (!a2.Nodes().Any())
  3472. {
  3473. a2.Add(new XText(""));
  3474. return a2;
  3475. }
  3476. return a;
  3477. })
  3478. .ToList();
  3479. return txformed;
  3480. }
  3481. #region Image Processing
  3482. // Don't process wmf files (with contentType == "image/x-wmf") because GDI consumes huge amounts
  3483. // of memory when dealing with wmf perhaps because it loads a DLL to do the rendering?
  3484. // It actually works, but is not recommended.
  3485. public static readonly List<string> ImageContentTypes = new List<string>
  3486. {
  3487. "image/png", "image/gif", "image/tiff", "image/jpeg","image/x-wmf","image/x-emf","application/vnd.openxmlformats-officedocument.oleObject"
  3488. };
  3489. public static XElement ProcessImage(WordprocessingDocument wordDoc,
  3490. XElement element, Func<ImageInfo, XElement> imageHandler)
  3491. {
  3492. if (imageHandler == null)
  3493. {
  3494. return null;
  3495. }
  3496. if (element.Name == W.drawing)
  3497. {
  3498. return ProcessDrawing(wordDoc, element, imageHandler);
  3499. }
  3500. if (element.Name == W.pict || element.Name == W._object)
  3501. {
  3502. return ProcessPictureOrObject(wordDoc, element, imageHandler);
  3503. }
  3504. return null;
  3505. }
  3506. private static XElement ProcessDrawing(WordprocessingDocument wordDoc,
  3507. XElement element, Func<ImageInfo, XElement> imageHandler)
  3508. {
  3509. var containerElement = element.Elements()
  3510. .FirstOrDefault(e => e.Name == WP.inline || e.Name == WP.anchor);
  3511. if (containerElement == null) return null;
  3512. string hyperlinkUri = null;
  3513. var hyperlinkElement = element
  3514. .Elements(WP.inline)
  3515. .Elements(WP.docPr)
  3516. .Elements(A.hlinkClick)
  3517. .FirstOrDefault();
  3518. if (hyperlinkElement != null)
  3519. {
  3520. var rId = (string)hyperlinkElement.Attribute(R.id);
  3521. if (rId != null)
  3522. {
  3523. var hyperlinkRel = wordDoc.MainDocumentPart.HyperlinkRelationships.FirstOrDefault(hlr => hlr.Id == rId);
  3524. if (hyperlinkRel != null)
  3525. {
  3526. hyperlinkUri = hyperlinkRel.Uri.ToString();
  3527. }
  3528. }
  3529. }
  3530. var extentCx = (int?)containerElement.Elements(WP.extent)
  3531. .Attributes(NoNamespace.cx).FirstOrDefault();
  3532. var extentCy = (int?)containerElement.Elements(WP.extent)
  3533. .Attributes(NoNamespace.cy).FirstOrDefault();
  3534. var altText = (string)containerElement.Elements(WP.docPr).Attributes(NoNamespace.descr).FirstOrDefault() ??
  3535. ((string)containerElement.Elements(WP.docPr).Attributes(NoNamespace.name).FirstOrDefault() ?? "");
  3536. if (containerElement.Elements(A.graphic)
  3537. .Elements(A.graphicData) != null)
  3538. {
  3539. XElement rest = null;
  3540. var elms = containerElement.Elements(A.graphic)
  3541. .Elements(A.graphicData).Elements();
  3542. foreach (var elm in elms)
  3543. {
  3544. if (elm.Name == Pic._pic)
  3545. {
  3546. rest = ProcessPic(wordDoc, containerElement, imageHandler, extentCx, extentCy, altText, elm, hyperlinkUri);
  3547. break;
  3548. }
  3549. if (elm.Name == C.chart)
  3550. {
  3551. rest = ProcessChart(wordDoc, containerElement, imageHandler, extentCx, extentCy, altText, elm, hyperlinkUri);
  3552. break;
  3553. }
  3554. if (elm.Name == DGM.relIds)
  3555. {
  3556. rest = ProcessDgm(wordDoc, containerElement, imageHandler, extentCx, extentCy, altText, elm, hyperlinkUri);
  3557. break;
  3558. }
  3559. }
  3560. return rest;
  3561. }
  3562. else
  3563. {
  3564. return null;
  3565. }
  3566. }
  3567. public static XElement ProcessPic(WordprocessingDocument wordDoc, XElement containerElement, Func<ImageInfo, XElement> imageHandler,
  3568. int? extentCx, int? extentCy, string altText, XElement element, string hyperlinkUri)
  3569. {
  3570. var blipFill = containerElement.Elements(A.graphic)
  3571. .Elements(A.graphicData)
  3572. .Elements(Pic._pic).Elements(Pic.blipFill).FirstOrDefault();
  3573. if (blipFill == null) return null;
  3574. var imageRid = (string)blipFill.Elements(A.blip).Attributes(R.embed).FirstOrDefault();
  3575. if (imageRid == null) return null;
  3576. var pp3 = wordDoc.MainDocumentPart.Parts.FirstOrDefault(pp => pp.RelationshipId == imageRid);
  3577. if (pp3 == null) return null;
  3578. var imagePart = (ImagePart)pp3.OpenXmlPart;
  3579. if (imagePart == null) return null;
  3580. // If the image markup points to a NULL image, then following will throw an ArgumentOutOfRangeException
  3581. try
  3582. {
  3583. imagePart = (ImagePart)wordDoc.MainDocumentPart.GetPartById(imageRid);
  3584. }
  3585. catch (ArgumentOutOfRangeException)
  3586. {
  3587. return null;
  3588. }
  3589. var contentType = imagePart.ContentType;
  3590. if (!ImageContentTypes.Contains(contentType))
  3591. return null;
  3592. var uil = imagePart.Uri;
  3593. //WORDHelper.GetWordNodeByPath(wordDoc,"");
  3594. var Stream = imagePart.GetStream();
  3595. try
  3596. {
  3597. if (contentType == "image/x-wmf" || contentType == "image/x-emf")
  3598. {
  3599. return ProcessWmf(Stream, contentType, element, hyperlinkUri, imageHandler);
  3600. }
  3601. else
  3602. {
  3603. using (var bitmap = new Bitmap(Stream))
  3604. {
  3605. if (extentCx != null && extentCy != null)
  3606. {
  3607. var imageInfo = new ImageInfo()
  3608. {
  3609. Bitmap = bitmap,
  3610. ImgStyleAttribute = new XAttribute("style",
  3611. string.Format(NumberFormatInfo.InvariantInfo,
  3612. "width: {0}px; ",
  3613. // "height: {1}in",
  3614. (float)extentCx * 96.0 / (float)ImageInfo.EmusPerInch
  3615. // ,
  3616. //(float)extentCy*96.0 / (float)ImageInfo.EmusPerInch
  3617. )),
  3618. ContentType = contentType,
  3619. DrawingElement = element,
  3620. AltText = altText,
  3621. };
  3622. var imgElement2 = imageHandler(imageInfo);
  3623. if (hyperlinkUri != null)
  3624. {
  3625. return new XElement(XhtmlNoNamespace.a,
  3626. new XAttribute(XhtmlNoNamespace.href, hyperlinkUri),
  3627. imgElement2);
  3628. }
  3629. return imgElement2;
  3630. }
  3631. var imageInfo2 = new ImageInfo()
  3632. {
  3633. Bitmap = bitmap,
  3634. ContentType = contentType,
  3635. DrawingElement = element,
  3636. AltText = altText,
  3637. };
  3638. var imgElement = imageHandler(imageInfo2);
  3639. if (hyperlinkUri != null)
  3640. {
  3641. return new XElement(XhtmlNoNamespace.a,
  3642. new XAttribute(XhtmlNoNamespace.href, hyperlinkUri),
  3643. imgElement);
  3644. }
  3645. return imgElement;
  3646. }
  3647. }
  3648. }
  3649. catch
  3650. {
  3651. XElement img = null;
  3652. byte[] bytes = new byte[Stream.Length];
  3653. Stream.Read(bytes, 0, bytes.Length);
  3654. // 设置当前流的位置为流的开始
  3655. Stream.Seek(0, SeekOrigin.Begin);
  3656. string url = System.Convert.ToBase64String(bytes);
  3657. img = new XElement(Xhtml.img,
  3658. new XAttribute(NoNamespace.src, "data:" + contentType + ";base64," + url),
  3659. new XAttribute(NoNamespace.alt, "wrong image"));
  3660. return img;
  3661. }
  3662. finally
  3663. {
  3664. Stream.Close();
  3665. }
  3666. }
  3667. public static XElement ProcessWmf(Stream stream, string contentType, XElement element, string hyperlinkUri, Func<ImageInfo, XElement> imageHandler)
  3668. {
  3669. XElement img = null;
  3670. string base64 = "";
  3671. string altText = "";
  3672. if (contentType.Equals("image/x-wmf"))
  3673. {
  3674. bool compatible = false;
  3675. bool replaceSymbolFont = false;
  3676. WMFConverter.Wmf.WmfParser parser = new WMFConverter.Wmf.WmfParser();
  3677. WMFConverter.Svg.SvgGdi gdi = new WMFConverter.Svg.SvgGdi(compatible);
  3678. gdi.ReplaceSymbolFont = replaceSymbolFont;
  3679. parser.Parse(stream, gdi);
  3680. base64 = gdi.Document.InnerXml;
  3681. }
  3682. else if (contentType.Equals("image/x-emf"))
  3683. {
  3684. MemoryStream memoryStream = new MemoryStream();
  3685. //Metafile inFile = new Metafile(stream);
  3686. ////Metafile outFile = new Metafile(memoryStream, Graphics.FromHwnd(IntPtr.Zero).GetHdc(), EmfType.EmfOnly);
  3687. //inFile.Save(memoryStream, ImageFormat.Png);
  3688. //var ba = memoryStream.ToArray();
  3689. //mathxml = System.Convert.ToBase64String(ba);
  3690. //System.Drawing.Bitmap inFile = new System.Drawing.Bitmap(stream);
  3691. //Graphics graphics = Graphics.FromImage(inFile);
  3692. //graphics.Save();
  3693. //graphics.Dispose();
  3694. //inFile.Save(memoryStream, System.Drawing.Imaging.ImageFormat.Png);
  3695. //byte[] arr = new byte[memoryStream.Length];
  3696. //memoryStream.Position = 0;
  3697. //memoryStream.Read(arr, 0, (int)memoryStream.Length);
  3698. //memoryStream.Close();
  3699. //memoryStream.Dispose();
  3700. //mathxml = Convert.ToBase64String(arr);
  3701. //mathxml = "EmfBase64ConvertError";
  3702. base64 = Globals.UnsupportedImage;
  3703. contentType = "image/x-emf";
  3704. altText = "UnsupportedImageError";
  3705. }
  3706. try
  3707. {
  3708. var imageInfo = new ImageInfo()
  3709. {
  3710. ContentType = contentType,
  3711. DrawingElement = element,
  3712. AltText = altText,
  3713. Mathxml = base64
  3714. };
  3715. var style = (string)element.Elements(VML.shape).Attributes("style").FirstOrDefault();
  3716. float? widthInPoints = null;
  3717. float? heightInPoints = null;
  3718. if (style != null)
  3719. {
  3720. var tokens = style.Split(';');
  3721. widthInPoints = WidthInPoints(tokens);
  3722. heightInPoints = HeightInPoints(tokens);
  3723. }
  3724. img = imageHandler(imageInfo);
  3725. //var imgElement2 = imageHandler(imageInfo);
  3726. if (widthInPoints != null && heightInPoints != null)
  3727. {
  3728. //imageInfo.ImgStyleAttribute = new XAttribute("style",
  3729. // string.Format(NumberFormatInfo.InvariantInfo,
  3730. // "width: {0}pt; height: {1}pt ; vertical-align: bottom;", widthInPoints, heightInPoints));
  3731. img.SetAttributeValue("style", string.Format(NumberFormatInfo.InvariantInfo,
  3732. "width: {0}pt; height: {1}pt ; vertical-align: bottom;", widthInPoints, heightInPoints));
  3733. }
  3734. else
  3735. {
  3736. //imageInfo.ImgStyleAttribute = new XAttribute("style",
  3737. // string.Format(NumberFormatInfo.InvariantInfo,
  3738. // "vertical-align: middle;"));
  3739. img.SetAttributeValue("style", string.Format(NumberFormatInfo.InvariantInfo,
  3740. "vertical-align: middle;"));
  3741. }
  3742. if (hyperlinkUri != null)
  3743. {
  3744. return new XElement(XhtmlNoNamespace.a,
  3745. new XAttribute(XhtmlNoNamespace.href, hyperlinkUri),
  3746. img);
  3747. }
  3748. return img;
  3749. }
  3750. catch (OutOfMemoryException)
  3751. {
  3752. // the Bitmap class can throw OutOfMemoryException, which means the bitmap is messed up, so punt.
  3753. return null;
  3754. }
  3755. catch (ArgumentException e)
  3756. {
  3757. throw new Exception(e.StackTrace);
  3758. }
  3759. }
  3760. private static XElement ProcessPictureOrObject(WordprocessingDocument wordDoc,
  3761. XElement element, Func<ImageInfo, XElement> imageHandler)
  3762. {
  3763. var imageRid = (string)element.Elements(VML.shape).Elements(VML.imagedata).Attributes(R.id).FirstOrDefault();
  3764. if (imageRid == null) return null;
  3765. var pp = wordDoc.MainDocumentPart.Parts.FirstOrDefault(pp2 => pp2.RelationshipId == imageRid);
  3766. Stream stream = null;
  3767. try
  3768. {
  3769. if (pp == null) return null;
  3770. if (!pp.OpenXmlPart.ContentType.Equals("image/x-wmf") && !pp.OpenXmlPart.ContentType.Equals("image/x-emf"))
  3771. {
  3772. return ProcessObject(wordDoc, element, imageHandler);
  3773. }
  3774. string mathxml = "";
  3775. if (pp.OpenXmlPart.ContentType.Equals("image/x-wmf"))
  3776. {
  3777. stream = pp.OpenXmlPart.GetStream();
  3778. bool compatible = false;
  3779. bool replaceSymbolFont = false;
  3780. WMFConverter.Wmf.WmfParser parser = new WMFConverter.Wmf.WmfParser();
  3781. WMFConverter.Svg.SvgGdi gdi = new WMFConverter.Svg.SvgGdi(compatible);
  3782. gdi.ReplaceSymbolFont = replaceSymbolFont;
  3783. parser.Parse(stream, gdi);
  3784. mathxml = gdi.Document.InnerXml;
  3785. }
  3786. else if (pp.OpenXmlPart.ContentType.Equals("image/x-emf"))
  3787. {
  3788. ////var image= SixLabors.ImageSharp.Image.Load(pp.OpenXmlPart.GetStream());
  3789. ////image.Save(memoryStream, new SixLabors.ImageSharp.Formats.Png.PngEncoder());
  3790. //var stm= pp.OpenXmlPart.GetStream();
  3791. //var parser = new EMF2StringParser();
  3792. //parser.LineBreakCandidates = new[] { EmfPlusRecordType.EmfSelectObject, EmfPlusRecordType.EmfDeleteObject };
  3793. //parser.SpaceCandidates = new[] { EmfPlusRecordType.EmfIntersectClipRect };
  3794. //Metafile inFile = new Metafile(stm);
  3795. //parser.LoadMetaFile(inFile);
  3796. //parser.IsParseFailedLoggingEnabled = true;
  3797. //var expected = parser.GetCombinedStringFromLoadedMetaFile();
  3798. ////MemoryStream memoryStream = new MemoryStream();
  3799. ////byte[] bytes = new byte[stm.Length];
  3800. ////stm.Read(bytes, 0, bytes.Length);
  3801. ////stm.Seek(0, SeekOrigin.Begin);
  3802. ////
  3803. ////Metafile outFile = new Metafile(memoryStream, Graphics.FromHwnd(IntPtr.Zero).GetHdc(), EmfType.EmfOnly);
  3804. //// inFile.Save(memoryStream, ImageFormat.Png);
  3805. ////var ba = memoryStream.ToArray();
  3806. //application.EnablePartialTrustCode = true;
  3807. //mathxml = expected;
  3808. // MemoryStream memoryStream = new MemoryStream();
  3809. //Metafile inFile = new Metafile(pp.OpenXmlPart.GetStream());
  3810. ////Metafile outFile = new Metafile(memoryStream, Graphics.FromHwnd(IntPtr.Zero).GetHdc(), EmfType.EmfOnly);
  3811. //inFile.Save(memoryStream, ImageFormat.Png);
  3812. //var ba = memoryStream.ToArray();
  3813. //mathxml = System.Convert.ToBase64String(ba);
  3814. //System.Drawing.Bitmap inFile = new System.Drawing.Bitmap(pp.OpenXmlPart.GetStream());
  3815. //Graphics graphics = Graphics.FromImage(inFile);
  3816. //graphics.Save();
  3817. //graphics.Dispose();
  3818. //inFile.Save(memoryStream, System.Drawing.Imaging.ImageFormat.Png);
  3819. //byte[] arr = new byte[memoryStream.Length];
  3820. //memoryStream.Position = 0;
  3821. //memoryStream.Read(arr, 0, (int)memoryStream.Length);
  3822. //memoryStream.Close();
  3823. //memoryStream.Dispose();
  3824. // mathxml = Convert.ToBase64String(arr);
  3825. mathxml = "EmfBase64ConvertError";
  3826. }
  3827. string hyperlinkUri = null;
  3828. var contentType = pp.OpenXmlPart.ContentType;
  3829. if (!ImageContentTypes.Contains(contentType))
  3830. return null;
  3831. try
  3832. {
  3833. XElement img = null;
  3834. var imageInfo = new ImageInfo()
  3835. {
  3836. ContentType = contentType,
  3837. DrawingElement = element,
  3838. AltText = "",
  3839. Mathxml = mathxml,
  3840. };
  3841. var style = (string)element.Elements(VML.shape).Attributes("style").FirstOrDefault();
  3842. // if (style == null) {
  3843. // return imageHandler(imageInfo);
  3844. img = imageHandler(imageInfo);
  3845. // }
  3846. var tokens = style.Split(';');
  3847. var widthInPoints = WidthInPoints(tokens);
  3848. var heightInPoints = HeightInPoints(tokens);
  3849. // var imgElement2 = imageHandler(imageInfo);
  3850. if (widthInPoints != null && heightInPoints != null)
  3851. {
  3852. //imageInfo.ImgStyleAttribute = new XAttribute("style",
  3853. // string.Format(NumberFormatInfo.InvariantInfo,
  3854. // "width: {0}pt; height: {1}pt ; vertical-align: bottom;", widthInPoints, heightInPoints));
  3855. img.SetAttributeValue("style", string.Format(NumberFormatInfo.InvariantInfo,
  3856. "width: {0}pt; height: {1}pt ; vertical-align: bottom;", widthInPoints, heightInPoints));
  3857. }
  3858. else
  3859. {
  3860. //imageInfo.ImgStyleAttribute = new XAttribute("style",
  3861. // string.Format(NumberFormatInfo.InvariantInfo,
  3862. // "vertical-align: middle;"));
  3863. img.SetAttributeValue("style", string.Format(NumberFormatInfo.InvariantInfo,
  3864. "vertical-align: middle;"));
  3865. }
  3866. if (hyperlinkUri != null)
  3867. {
  3868. return new XElement(XhtmlNoNamespace.a,
  3869. new XAttribute(XhtmlNoNamespace.href, hyperlinkUri),
  3870. img);
  3871. }
  3872. return img;
  3873. }
  3874. catch (OutOfMemoryException)
  3875. {
  3876. // the Bitmap class can throw OutOfMemoryException, which means the bitmap is messed up, so punt.
  3877. return null;
  3878. }
  3879. catch (ArgumentException e)
  3880. {
  3881. throw new Exception(e.StackTrace);
  3882. }
  3883. }
  3884. catch (ArgumentOutOfRangeException)
  3885. {
  3886. return null;
  3887. }
  3888. finally
  3889. {
  3890. if (stream != null)
  3891. stream.Close();
  3892. }
  3893. }
  3894. private static XElement ProcessObject(WordprocessingDocument wordDoc,
  3895. XElement element, Func<ImageInfo, XElement> imageHandler)
  3896. {
  3897. var imageRid = (string)element.Elements(VML.shape).Elements(VML.imagedata).Attributes(R.id).FirstOrDefault();
  3898. if (imageRid == null) return null;
  3899. try
  3900. {
  3901. var pp = wordDoc.MainDocumentPart.Parts.FirstOrDefault(pp2 => pp2.RelationshipId == imageRid);
  3902. if (pp == null) return null;
  3903. var imagePart = (ImagePart)pp.OpenXmlPart;
  3904. if (imagePart == null) return null;
  3905. string hyperlinkUri = null;
  3906. // WORDHelper.GetWordNodeByPath(wordDoc.MainDocumentPart.GetXDocument(), "//[@name='"+imagePart.Uri+"']");
  3907. using (var partStream = imagePart.GetStream())
  3908. {
  3909. var contentType = imagePart.ContentType;
  3910. if (!ImageContentTypes.Contains(contentType))
  3911. return null;
  3912. try
  3913. {
  3914. using (var bitmap = new Bitmap(partStream))
  3915. {
  3916. var imageInfo = new ImageInfo()
  3917. {
  3918. Bitmap = bitmap,
  3919. ContentType = contentType,
  3920. DrawingElement = element,
  3921. AltText = ""
  3922. };
  3923. var style = (string)element.Elements(VML.shape).Attributes("style").FirstOrDefault();
  3924. if (style == null) return imageHandler(imageInfo);
  3925. var tokens = style.Split(';');
  3926. var widthInPoints = WidthInPoints(tokens);
  3927. var heightInPoints = HeightInPoints(tokens);
  3928. if (widthInPoints != null && heightInPoints != null)
  3929. {
  3930. imageInfo.ImgStyleAttribute = new XAttribute("style",
  3931. string.Format(NumberFormatInfo.InvariantInfo,
  3932. "width: {0}pt; height: {1}pt", widthInPoints, heightInPoints));
  3933. }
  3934. var imgElement2 = imageHandler(imageInfo);
  3935. if (hyperlinkUri != null)
  3936. {
  3937. return new XElement(XhtmlNoNamespace.a,
  3938. new XAttribute(XhtmlNoNamespace.href, hyperlinkUri),
  3939. imgElement2);
  3940. }
  3941. return imgElement2;
  3942. }
  3943. }
  3944. catch (OutOfMemoryException)
  3945. {
  3946. return null;
  3947. }
  3948. catch (ArgumentException e)
  3949. {
  3950. return null; //throw new Exception(e.StackTrace);
  3951. }
  3952. finally { partStream.Close(); }
  3953. }
  3954. }
  3955. catch (ArgumentOutOfRangeException)
  3956. {
  3957. return null;
  3958. }
  3959. }
  3960. private static float? HeightInPoints(IEnumerable<string> tokens)
  3961. {
  3962. return SizeInPoints(tokens, "height");
  3963. }
  3964. private static float? WidthInPoints(IEnumerable<string> tokens)
  3965. {
  3966. return SizeInPoints(tokens, "width");
  3967. }
  3968. private static float? SizeInPoints(IEnumerable<string> tokens, string name)
  3969. {
  3970. var sizeString = tokens
  3971. .Select(t => new
  3972. {
  3973. Name = t.Split(':').First(),
  3974. Value = t.Split(':').Skip(1).Take(1).FirstOrDefault()
  3975. })
  3976. .Where(p => p.Name == name)
  3977. .Select(p => p.Value)
  3978. .FirstOrDefault();
  3979. if (sizeString != null &&
  3980. sizeString.Length > 2 &&
  3981. sizeString.Substring(sizeString.Length - 2) == "pt")
  3982. {
  3983. float size;
  3984. if (float.TryParse(sizeString.Substring(0, sizeString.Length - 2), out size))
  3985. return size;
  3986. }
  3987. return null;
  3988. }
  3989. #endregion
  3990. }
  3991. public static class HtmlConverterExtensions
  3992. {
  3993. public static void AddIfMissing(this Dictionary<string, string> style, string propName, string value)
  3994. {
  3995. if (style.ContainsKey(propName))
  3996. return;
  3997. style.Add(propName, value);
  3998. }
  3999. }
  4000. }