pdf2json 0.1.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (473) hide show
  1. data/README.markdown +9 -0
  2. data/bin/.gitkeep +0 -0
  3. data/ext/extconf.rb +30 -0
  4. data/lib/pdf2json.rb +8 -0
  5. data/pdf2json-0.52-source/AUTHORS +24 -0
  6. data/pdf2json-0.52-source/CHANGES +11 -0
  7. data/pdf2json-0.52-source/Makefile +84 -0
  8. data/pdf2json-0.52-source/Makefile.in +84 -0
  9. data/pdf2json-0.52-source/aclocal.m4 +274 -0
  10. data/pdf2json-0.52-source/aconf-win32.h +86 -0
  11. data/pdf2json-0.52-source/aconf.h +42 -0
  12. data/pdf2json-0.52-source/aconf.h.in +41 -0
  13. data/pdf2json-0.52-source/autom4te.cache/output.0 +6908 -0
  14. data/pdf2json-0.52-source/autom4te.cache/requests +76 -0
  15. data/pdf2json-0.52-source/autom4te.cache/traces.0 +466 -0
  16. data/pdf2json-0.52-source/config.log +1259 -0
  17. data/pdf2json-0.52-source/config.status +1050 -0
  18. data/pdf2json-0.52-source/configure +6908 -0
  19. data/pdf2json-0.52-source/configure.ac +93 -0
  20. data/pdf2json-0.52-source/doc/pdffonts.1 +130 -0
  21. data/pdf2json-0.52-source/doc/pdffonts.cat +107 -0
  22. data/pdf2json-0.52-source/doc/pdffonts.hlp +117 -0
  23. data/pdf2json-0.52-source/doc/pdfimages.1 +102 -0
  24. data/pdf2json-0.52-source/doc/pdfimages.cat +92 -0
  25. data/pdf2json-0.52-source/doc/pdfimages.hlp +101 -0
  26. data/pdf2json-0.52-source/doc/pdfinfo.1 +158 -0
  27. data/pdf2json-0.52-source/doc/pdfinfo.cat +119 -0
  28. data/pdf2json-0.52-source/doc/pdfinfo.hlp +129 -0
  29. data/pdf2json-0.52-source/doc/pdftoppm.1 +115 -0
  30. data/pdf2json-0.52-source/doc/pdftoppm.cat +105 -0
  31. data/pdf2json-0.52-source/doc/pdftoppm.hlp +114 -0
  32. data/pdf2json-0.52-source/doc/pdftops.1 +229 -0
  33. data/pdf2json-0.52-source/doc/pdftops.cat +221 -0
  34. data/pdf2json-0.52-source/doc/pdftops.hlp +231 -0
  35. data/pdf2json-0.52-source/doc/pdftotext.1 +137 -0
  36. data/pdf2json-0.52-source/doc/pdftotext.cat +120 -0
  37. data/pdf2json-0.52-source/doc/pdftotext.hlp +133 -0
  38. data/pdf2json-0.52-source/doc/sample-xpdfrc +91 -0
  39. data/pdf2json-0.52-source/doc/xpdf.1 +513 -0
  40. data/pdf2json-0.52-source/doc/xpdf.cat +476 -0
  41. data/pdf2json-0.52-source/doc/xpdf.hlp +489 -0
  42. data/pdf2json-0.52-source/doc/xpdfrc.5 +480 -0
  43. data/pdf2json-0.52-source/doc/xpdfrc.cat +474 -0
  44. data/pdf2json-0.52-source/doc/xpdfrc.hlp +479 -0
  45. data/pdf2json-0.52-source/fofi/.DS_Store +0 -0
  46. data/pdf2json-0.52-source/fofi/FoFiBase.cc +156 -0
  47. data/pdf2json-0.52-source/fofi/FoFiBase.h +57 -0
  48. data/pdf2json-0.52-source/fofi/FoFiBase.o +0 -0
  49. data/pdf2json-0.52-source/fofi/FoFiEncodings.cc +994 -0
  50. data/pdf2json-0.52-source/fofi/FoFiEncodings.h +36 -0
  51. data/pdf2json-0.52-source/fofi/FoFiEncodings.o +0 -0
  52. data/pdf2json-0.52-source/fofi/FoFiTrueType.cc +2027 -0
  53. data/pdf2json-0.52-source/fofi/FoFiTrueType.h +174 -0
  54. data/pdf2json-0.52-source/fofi/FoFiTrueType.o +0 -0
  55. data/pdf2json-0.52-source/fofi/FoFiType1.cc +252 -0
  56. data/pdf2json-0.52-source/fofi/FoFiType1.h +59 -0
  57. data/pdf2json-0.52-source/fofi/FoFiType1.o +0 -0
  58. data/pdf2json-0.52-source/fofi/FoFiType1C.cc +2603 -0
  59. data/pdf2json-0.52-source/fofi/FoFiType1C.h +233 -0
  60. data/pdf2json-0.52-source/fofi/FoFiType1C.o +0 -0
  61. data/pdf2json-0.52-source/fofi/Makefile +70 -0
  62. data/pdf2json-0.52-source/fofi/Makefile.dep +0 -0
  63. data/pdf2json-0.52-source/fofi/Makefile.in +70 -0
  64. data/pdf2json-0.52-source/fofi/libfofi.a +0 -0
  65. data/pdf2json-0.52-source/fofi/vms_make.com +0 -0
  66. data/pdf2json-0.52-source/freetype.win32/.DS_Store +0 -0
  67. data/pdf2json-0.52-source/freetype.win32/include/.DS_Store +0 -0
  68. data/pdf2json-0.52-source/freetype.win32/include/freetype/config/ftconfig.h +528 -0
  69. data/pdf2json-0.52-source/freetype.win32/include/freetype/config/ftheader.h +780 -0
  70. data/pdf2json-0.52-source/freetype.win32/include/freetype/config/ftmodule.h +32 -0
  71. data/pdf2json-0.52-source/freetype.win32/include/freetype/config/ftoption.h +733 -0
  72. data/pdf2json-0.52-source/freetype.win32/include/freetype/config/ftstdlib.h +173 -0
  73. data/pdf2json-0.52-source/freetype.win32/include/freetype/freetype.h +3919 -0
  74. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftadvanc.h +179 -0
  75. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftbbox.h +94 -0
  76. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftbdf.h +209 -0
  77. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftbitmap.h +227 -0
  78. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftcache.h +1128 -0
  79. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftchapters.h +103 -0
  80. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftcid.h +166 -0
  81. data/pdf2json-0.52-source/freetype.win32/include/freetype/fterrdef.h +244 -0
  82. data/pdf2json-0.52-source/freetype.win32/include/freetype/fterrors.h +206 -0
  83. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftgasp.h +120 -0
  84. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftglyph.h +613 -0
  85. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftgxval.h +358 -0
  86. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftgzip.h +102 -0
  87. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftimage.h +1313 -0
  88. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftincrem.h +353 -0
  89. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftlcdfil.h +213 -0
  90. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftlist.h +277 -0
  91. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftlzw.h +99 -0
  92. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftmac.h +274 -0
  93. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftmm.h +378 -0
  94. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftmodapi.h +483 -0
  95. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftmoderr.h +155 -0
  96. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftotval.h +203 -0
  97. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftoutln.h +537 -0
  98. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftpfr.h +172 -0
  99. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftrender.h +230 -0
  100. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftsizes.h +159 -0
  101. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftsnames.h +200 -0
  102. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftstroke.h +716 -0
  103. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftsynth.h +80 -0
  104. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftsystem.h +347 -0
  105. data/pdf2json-0.52-source/freetype.win32/include/freetype/fttrigon.h +350 -0
  106. data/pdf2json-0.52-source/freetype.win32/include/freetype/fttypes.h +588 -0
  107. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftwinfnt.h +274 -0
  108. data/pdf2json-0.52-source/freetype.win32/include/freetype/ftxf86.h +83 -0
  109. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/autohint.h +231 -0
  110. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/ftcalc.h +179 -0
  111. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/ftdebug.h +250 -0
  112. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/ftdriver.h +422 -0
  113. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/ftgloadr.h +168 -0
  114. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/ftmemory.h +380 -0
  115. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/ftobjs.h +1428 -0
  116. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/ftpic.h +67 -0
  117. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/ftrfork.h +196 -0
  118. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/ftserv.h +620 -0
  119. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/ftstream.h +539 -0
  120. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/fttrace.h +139 -0
  121. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/ftvalid.h +150 -0
  122. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/internal.h +51 -0
  123. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/pcftypes.h +56 -0
  124. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/psaux.h +873 -0
  125. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/pshints.h +712 -0
  126. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svbdf.h +77 -0
  127. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svcid.h +83 -0
  128. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svgldict.h +82 -0
  129. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svgxval.h +72 -0
  130. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svkern.h +51 -0
  131. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svmm.h +104 -0
  132. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svotval.h +55 -0
  133. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svpfr.h +66 -0
  134. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svpostnm.h +79 -0
  135. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svpscmap.h +164 -0
  136. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svpsinfo.h +92 -0
  137. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svsfnt.h +102 -0
  138. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svttcmap.h +106 -0
  139. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svtteng.h +53 -0
  140. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svttglyf.h +67 -0
  141. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svwinfnt.h +50 -0
  142. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/services/svxf86nm.h +55 -0
  143. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/sfnt.h +897 -0
  144. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/t1types.h +270 -0
  145. data/pdf2json-0.52-source/freetype.win32/include/freetype/internal/tttypes.h +1543 -0
  146. data/pdf2json-0.52-source/freetype.win32/include/freetype/t1tables.h +504 -0
  147. data/pdf2json-0.52-source/freetype.win32/include/freetype/ttnameid.h +1247 -0
  148. data/pdf2json-0.52-source/freetype.win32/include/freetype/tttables.h +759 -0
  149. data/pdf2json-0.52-source/freetype.win32/include/freetype/tttags.h +107 -0
  150. data/pdf2json-0.52-source/freetype.win32/include/freetype/ttunpat.h +59 -0
  151. data/pdf2json-0.52-source/freetype.win32/include/ft2build.h +39 -0
  152. data/pdf2json-0.52-source/freetype.win32/lib/freetype_a.lib +0 -0
  153. data/pdf2json-0.52-source/goo/.DS_Store +0 -0
  154. data/pdf2json-0.52-source/goo/FixedPoint.cc +118 -0
  155. data/pdf2json-0.52-source/goo/FixedPoint.h +155 -0
  156. data/pdf2json-0.52-source/goo/FixedPoint.o +0 -0
  157. data/pdf2json-0.52-source/goo/GHash.cc +380 -0
  158. data/pdf2json-0.52-source/goo/GHash.h +78 -0
  159. data/pdf2json-0.52-source/goo/GHash.o +0 -0
  160. data/pdf2json-0.52-source/goo/GList.cc +97 -0
  161. data/pdf2json-0.52-source/goo/GList.h +96 -0
  162. data/pdf2json-0.52-source/goo/GList.o +0 -0
  163. data/pdf2json-0.52-source/goo/GMutex.h +49 -0
  164. data/pdf2json-0.52-source/goo/GString.cc +724 -0
  165. data/pdf2json-0.52-source/goo/GString.cc.fixed +718 -0
  166. data/pdf2json-0.52-source/goo/GString.h +136 -0
  167. data/pdf2json-0.52-source/goo/GString.o +0 -0
  168. data/pdf2json-0.52-source/goo/ImgWriter.o +0 -0
  169. data/pdf2json-0.52-source/goo/JpegWriter.o +0 -0
  170. data/pdf2json-0.52-source/goo/Makefile +72 -0
  171. data/pdf2json-0.52-source/goo/Makefile.dep +0 -0
  172. data/pdf2json-0.52-source/goo/Makefile.in +72 -0
  173. data/pdf2json-0.52-source/goo/PNGWriter.o +0 -0
  174. data/pdf2json-0.52-source/goo/gfile.cc +731 -0
  175. data/pdf2json-0.52-source/goo/gfile.h +138 -0
  176. data/pdf2json-0.52-source/goo/gfile.o +0 -0
  177. data/pdf2json-0.52-source/goo/gmem.cc +264 -0
  178. data/pdf2json-0.52-source/goo/gmem.h +79 -0
  179. data/pdf2json-0.52-source/goo/gmem.o +0 -0
  180. data/pdf2json-0.52-source/goo/gmempp.cc +32 -0
  181. data/pdf2json-0.52-source/goo/gmempp.o +0 -0
  182. data/pdf2json-0.52-source/goo/gtypes.h +29 -0
  183. data/pdf2json-0.52-source/goo/libGoo.a +0 -0
  184. data/pdf2json-0.52-source/goo/parseargs.c +190 -0
  185. data/pdf2json-0.52-source/goo/parseargs.h +71 -0
  186. data/pdf2json-0.52-source/goo/parseargs.o +0 -0
  187. data/pdf2json-0.52-source/goo/vms_directory.c +214 -0
  188. data/pdf2json-0.52-source/goo/vms_dirent.h +67 -0
  189. data/pdf2json-0.52-source/goo/vms_make.com +82 -0
  190. data/pdf2json-0.52-source/goo/vms_sys_dirent.h +54 -0
  191. data/pdf2json-0.52-source/goo/vms_unix_time.h +102 -0
  192. data/pdf2json-0.52-source/goo/vms_unix_times.c +42 -0
  193. data/pdf2json-0.52-source/goo/vms_unlink.c +22 -0
  194. data/pdf2json-0.52-source/ms_make.bat +199 -0
  195. data/pdf2json-0.52-source/splash/.DS_Store +0 -0
  196. data/pdf2json-0.52-source/splash/Makefile +103 -0
  197. data/pdf2json-0.52-source/splash/Makefile.dep +0 -0
  198. data/pdf2json-0.52-source/splash/Makefile.in +103 -0
  199. data/pdf2json-0.52-source/splash/Splash.cc +3310 -0
  200. data/pdf2json-0.52-source/splash/Splash.h +293 -0
  201. data/pdf2json-0.52-source/splash/Splash.o +0 -0
  202. data/pdf2json-0.52-source/splash/SplashBitmap.cc +188 -0
  203. data/pdf2json-0.52-source/splash/SplashBitmap.h +64 -0
  204. data/pdf2json-0.52-source/splash/SplashBitmap.o +0 -0
  205. data/pdf2json-0.52-source/splash/SplashClip.cc +382 -0
  206. data/pdf2json-0.52-source/splash/SplashClip.h +107 -0
  207. data/pdf2json-0.52-source/splash/SplashClip.o +0 -0
  208. data/pdf2json-0.52-source/splash/SplashErrorCodes.h +32 -0
  209. data/pdf2json-0.52-source/splash/SplashFTFont.cc +357 -0
  210. data/pdf2json-0.52-source/splash/SplashFTFont.h +58 -0
  211. data/pdf2json-0.52-source/splash/SplashFTFont.o +0 -0
  212. data/pdf2json-0.52-source/splash/SplashFTFontEngine.cc +179 -0
  213. data/pdf2json-0.52-source/splash/SplashFTFontEngine.h +65 -0
  214. data/pdf2json-0.52-source/splash/SplashFTFontEngine.o +0 -0
  215. data/pdf2json-0.52-source/splash/SplashFTFontFile.cc +114 -0
  216. data/pdf2json-0.52-source/splash/SplashFTFontFile.h +73 -0
  217. data/pdf2json-0.52-source/splash/SplashFTFontFile.o +0 -0
  218. data/pdf2json-0.52-source/splash/SplashFont.cc +176 -0
  219. data/pdf2json-0.52-source/splash/SplashFont.h +104 -0
  220. data/pdf2json-0.52-source/splash/SplashFont.o +0 -0
  221. data/pdf2json-0.52-source/splash/SplashFontEngine.cc +317 -0
  222. data/pdf2json-0.52-source/splash/SplashFontEngine.h +91 -0
  223. data/pdf2json-0.52-source/splash/SplashFontEngine.o +0 -0
  224. data/pdf2json-0.52-source/splash/SplashFontFile.cc +55 -0
  225. data/pdf2json-0.52-source/splash/SplashFontFile.h +60 -0
  226. data/pdf2json-0.52-source/splash/SplashFontFile.o +0 -0
  227. data/pdf2json-0.52-source/splash/SplashFontFileID.cc +23 -0
  228. data/pdf2json-0.52-source/splash/SplashFontFileID.h +30 -0
  229. data/pdf2json-0.52-source/splash/SplashFontFileID.o +0 -0
  230. data/pdf2json-0.52-source/splash/SplashGlyphBitmap.h +26 -0
  231. data/pdf2json-0.52-source/splash/SplashMath.h +89 -0
  232. data/pdf2json-0.52-source/splash/SplashPath.cc +184 -0
  233. data/pdf2json-0.52-source/splash/SplashPath.h +121 -0
  234. data/pdf2json-0.52-source/splash/SplashPath.o +0 -0
  235. data/pdf2json-0.52-source/splash/SplashPattern.cc +40 -0
  236. data/pdf2json-0.52-source/splash/SplashPattern.h +65 -0
  237. data/pdf2json-0.52-source/splash/SplashPattern.o +0 -0
  238. data/pdf2json-0.52-source/splash/SplashScreen.cc +383 -0
  239. data/pdf2json-0.52-source/splash/SplashScreen.h +56 -0
  240. data/pdf2json-0.52-source/splash/SplashScreen.o +0 -0
  241. data/pdf2json-0.52-source/splash/SplashState.cc +165 -0
  242. data/pdf2json-0.52-source/splash/SplashState.h +103 -0
  243. data/pdf2json-0.52-source/splash/SplashState.o +0 -0
  244. data/pdf2json-0.52-source/splash/SplashT1Font.cc +287 -0
  245. data/pdf2json-0.52-source/splash/SplashT1Font.h +57 -0
  246. data/pdf2json-0.52-source/splash/SplashT1Font.o +0 -0
  247. data/pdf2json-0.52-source/splash/SplashT1FontEngine.cc +124 -0
  248. data/pdf2json-0.52-source/splash/SplashT1FontEngine.h +53 -0
  249. data/pdf2json-0.52-source/splash/SplashT1FontEngine.o +0 -0
  250. data/pdf2json-0.52-source/splash/SplashT1FontFile.cc +97 -0
  251. data/pdf2json-0.52-source/splash/SplashT1FontFile.h +58 -0
  252. data/pdf2json-0.52-source/splash/SplashT1FontFile.o +0 -0
  253. data/pdf2json-0.52-source/splash/SplashTypes.h +132 -0
  254. data/pdf2json-0.52-source/splash/SplashXPath.cc +438 -0
  255. data/pdf2json-0.52-source/splash/SplashXPath.h +100 -0
  256. data/pdf2json-0.52-source/splash/SplashXPath.o +0 -0
  257. data/pdf2json-0.52-source/splash/SplashXPathScanner.cc +428 -0
  258. data/pdf2json-0.52-source/splash/SplashXPathScanner.h +87 -0
  259. data/pdf2json-0.52-source/splash/SplashXPathScanner.o +0 -0
  260. data/pdf2json-0.52-source/splash/libsplash.a +0 -0
  261. data/pdf2json-0.52-source/splash/vms_make.com +0 -0
  262. data/pdf2json-0.52-source/src/.DS_Store +0 -0
  263. data/pdf2json-0.52-source/src/GVector.h +101 -0
  264. data/pdf2json-0.52-source/src/ImgOutputDev.cc +1243 -0
  265. data/pdf2json-0.52-source/src/ImgOutputDev.h +307 -0
  266. data/pdf2json-0.52-source/src/ImgOutputDev.o +0 -0
  267. data/pdf2json-0.52-source/src/Makefile +68 -0
  268. data/pdf2json-0.52-source/src/Makefile.in +68 -0
  269. data/pdf2json-0.52-source/src/XmlFonts.cc +367 -0
  270. data/pdf2json-0.52-source/src/XmlFonts.h +91 -0
  271. data/pdf2json-0.52-source/src/XmlFonts.o +0 -0
  272. data/pdf2json-0.52-source/src/XmlLinks.cc +101 -0
  273. data/pdf2json-0.52-source/src/XmlLinks.h +54 -0
  274. data/pdf2json-0.52-source/src/XmlLinks.o +0 -0
  275. data/pdf2json-0.52-source/src/pdf2json +0 -0
  276. data/pdf2json-0.52-source/src/pdf2json.cc +343 -0
  277. data/pdf2json-0.52-source/src/pdf2json.o +0 -0
  278. data/pdf2json-0.52-source/src/pdf2xml.dtd +22 -0
  279. data/pdf2json-0.52-source/src/pdf2xmljson.dtd +9 -0
  280. data/pdf2json-0.52-source/xpdf/.DS_Store +0 -0
  281. data/pdf2json-0.52-source/xpdf/Annot.cc +1556 -0
  282. data/pdf2json-0.52-source/xpdf/Annot.h +142 -0
  283. data/pdf2json-0.52-source/xpdf/Annot.o +0 -0
  284. data/pdf2json-0.52-source/xpdf/Array.cc +73 -0
  285. data/pdf2json-0.52-source/xpdf/Array.h +58 -0
  286. data/pdf2json-0.52-source/xpdf/Array.o +0 -0
  287. data/pdf2json-0.52-source/xpdf/BuiltinFont.cc +65 -0
  288. data/pdf2json-0.52-source/xpdf/BuiltinFont.h +57 -0
  289. data/pdf2json-0.52-source/xpdf/BuiltinFont.o +0 -0
  290. data/pdf2json-0.52-source/xpdf/BuiltinFontTables.cc +4284 -0
  291. data/pdf2json-0.52-source/xpdf/BuiltinFontTables.h +23 -0
  292. data/pdf2json-0.52-source/xpdf/BuiltinFontTables.o +0 -0
  293. data/pdf2json-0.52-source/xpdf/CMap.cc +408 -0
  294. data/pdf2json-0.52-source/xpdf/CMap.h +102 -0
  295. data/pdf2json-0.52-source/xpdf/CMap.o +0 -0
  296. data/pdf2json-0.52-source/xpdf/Catalog.cc +374 -0
  297. data/pdf2json-0.52-source/xpdf/Catalog.h +97 -0
  298. data/pdf2json-0.52-source/xpdf/Catalog.o +0 -0
  299. data/pdf2json-0.52-source/xpdf/CharCodeToUnicode.cc +540 -0
  300. data/pdf2json-0.52-source/xpdf/CharCodeToUnicode.h +117 -0
  301. data/pdf2json-0.52-source/xpdf/CharCodeToUnicode.o +0 -0
  302. data/pdf2json-0.52-source/xpdf/CharTypes.h +24 -0
  303. data/pdf2json-0.52-source/xpdf/CompactFontTables.h +464 -0
  304. data/pdf2json-0.52-source/xpdf/CoreOutputDev.cc +61 -0
  305. data/pdf2json-0.52-source/xpdf/CoreOutputDev.h +61 -0
  306. data/pdf2json-0.52-source/xpdf/Decrypt.cc +776 -0
  307. data/pdf2json-0.52-source/xpdf/Decrypt.h +95 -0
  308. data/pdf2json-0.52-source/xpdf/Decrypt.o +0 -0
  309. data/pdf2json-0.52-source/xpdf/Dict.cc +95 -0
  310. data/pdf2json-0.52-source/xpdf/Dict.h +77 -0
  311. data/pdf2json-0.52-source/xpdf/Dict.o +0 -0
  312. data/pdf2json-0.52-source/xpdf/Error.cc +38 -0
  313. data/pdf2json-0.52-source/xpdf/Error.h +23 -0
  314. data/pdf2json-0.52-source/xpdf/Error.o +0 -0
  315. data/pdf2json-0.52-source/xpdf/ErrorCodes.h +36 -0
  316. data/pdf2json-0.52-source/xpdf/FontEncodingTables.cc +1824 -0
  317. data/pdf2json-0.52-source/xpdf/FontEncodingTables.h +20 -0
  318. data/pdf2json-0.52-source/xpdf/FontEncodingTables.o +0 -0
  319. data/pdf2json-0.52-source/xpdf/Function.cc +1573 -0
  320. data/pdf2json-0.52-source/xpdf/Function.h +229 -0
  321. data/pdf2json-0.52-source/xpdf/Function.o +0 -0
  322. data/pdf2json-0.52-source/xpdf/Gfx.cc +4187 -0
  323. data/pdf2json-0.52-source/xpdf/Gfx.h +312 -0
  324. data/pdf2json-0.52-source/xpdf/Gfx.o +0 -0
  325. data/pdf2json-0.52-source/xpdf/GfxFont.cc +1568 -0
  326. data/pdf2json-0.52-source/xpdf/GfxFont.h +320 -0
  327. data/pdf2json-0.52-source/xpdf/GfxFont.o +0 -0
  328. data/pdf2json-0.52-source/xpdf/GfxState.cc +4137 -0
  329. data/pdf2json-0.52-source/xpdf/GfxState.h +1244 -0
  330. data/pdf2json-0.52-source/xpdf/GfxState.o +0 -0
  331. data/pdf2json-0.52-source/xpdf/GlobalParams.cc +2924 -0
  332. data/pdf2json-0.52-source/xpdf/GlobalParams.cc.old +2908 -0
  333. data/pdf2json-0.52-source/xpdf/GlobalParams.h +466 -0
  334. data/pdf2json-0.52-source/xpdf/GlobalParams.h.old +463 -0
  335. data/pdf2json-0.52-source/xpdf/GlobalParams.o +0 -0
  336. data/pdf2json-0.52-source/xpdf/ImageOutputDev.cc +195 -0
  337. data/pdf2json-0.52-source/xpdf/ImageOutputDev.h +76 -0
  338. data/pdf2json-0.52-source/xpdf/ImageOutputDev.o +0 -0
  339. data/pdf2json-0.52-source/xpdf/JArithmeticDecoder.cc +322 -0
  340. data/pdf2json-0.52-source/xpdf/JArithmeticDecoder.h +109 -0
  341. data/pdf2json-0.52-source/xpdf/JArithmeticDecoder.o +0 -0
  342. data/pdf2json-0.52-source/xpdf/JBIG2Stream.cc +3413 -0
  343. data/pdf2json-0.52-source/xpdf/JBIG2Stream.h +145 -0
  344. data/pdf2json-0.52-source/xpdf/JBIG2Stream.o +0 -0
  345. data/pdf2json-0.52-source/xpdf/JPXStream.cc +3144 -0
  346. data/pdf2json-0.52-source/xpdf/JPXStream.h +351 -0
  347. data/pdf2json-0.52-source/xpdf/JPXStream.o +0 -0
  348. data/pdf2json-0.52-source/xpdf/Lexer.cc +485 -0
  349. data/pdf2json-0.52-source/xpdf/Lexer.h +80 -0
  350. data/pdf2json-0.52-source/xpdf/Lexer.o +0 -0
  351. data/pdf2json-0.52-source/xpdf/Link.cc +806 -0
  352. data/pdf2json-0.52-source/xpdf/Link.cc.old +784 -0
  353. data/pdf2json-0.52-source/xpdf/Link.h +415 -0
  354. data/pdf2json-0.52-source/xpdf/Link.h.old +369 -0
  355. data/pdf2json-0.52-source/xpdf/Link.o +0 -0
  356. data/pdf2json-0.52-source/xpdf/Makefile +232 -0
  357. data/pdf2json-0.52-source/xpdf/Makefile.dep +0 -0
  358. data/pdf2json-0.52-source/xpdf/Makefile.in +232 -0
  359. data/pdf2json-0.52-source/xpdf/NameToCharCode.cc +116 -0
  360. data/pdf2json-0.52-source/xpdf/NameToCharCode.h +42 -0
  361. data/pdf2json-0.52-source/xpdf/NameToCharCode.o +0 -0
  362. data/pdf2json-0.52-source/xpdf/NameToUnicodeTable.h +1097 -0
  363. data/pdf2json-0.52-source/xpdf/Object.cc +231 -0
  364. data/pdf2json-0.52-source/xpdf/Object.h +303 -0
  365. data/pdf2json-0.52-source/xpdf/Object.o +0 -0
  366. data/pdf2json-0.52-source/xpdf/Outline.cc +151 -0
  367. data/pdf2json-0.52-source/xpdf/Outline.h +76 -0
  368. data/pdf2json-0.52-source/xpdf/Outline.o +0 -0
  369. data/pdf2json-0.52-source/xpdf/OutputDev.cc +131 -0
  370. data/pdf2json-0.52-source/xpdf/OutputDev.h +253 -0
  371. data/pdf2json-0.52-source/xpdf/OutputDev.o +0 -0
  372. data/pdf2json-0.52-source/xpdf/PDFCore.cc +2044 -0
  373. data/pdf2json-0.52-source/xpdf/PDFCore.h +321 -0
  374. data/pdf2json-0.52-source/xpdf/PDFDoc.cc +404 -0
  375. data/pdf2json-0.52-source/xpdf/PDFDoc.h +183 -0
  376. data/pdf2json-0.52-source/xpdf/PDFDoc.o +0 -0
  377. data/pdf2json-0.52-source/xpdf/PDFDocEncoding.cc +44 -0
  378. data/pdf2json-0.52-source/xpdf/PDFDocEncoding.h +16 -0
  379. data/pdf2json-0.52-source/xpdf/PDFDocEncoding.o +0 -0
  380. data/pdf2json-0.52-source/xpdf/PSOutputDev.cc +6224 -0
  381. data/pdf2json-0.52-source/xpdf/PSOutputDev.h +395 -0
  382. data/pdf2json-0.52-source/xpdf/PSOutputDev.o +0 -0
  383. data/pdf2json-0.52-source/xpdf/PSTokenizer.cc +135 -0
  384. data/pdf2json-0.52-source/xpdf/PSTokenizer.h +41 -0
  385. data/pdf2json-0.52-source/xpdf/PSTokenizer.o +0 -0
  386. data/pdf2json-0.52-source/xpdf/Page.cc +454 -0
  387. data/pdf2json-0.52-source/xpdf/Page.h +187 -0
  388. data/pdf2json-0.52-source/xpdf/Page.o +0 -0
  389. data/pdf2json-0.52-source/xpdf/Parser.cc +227 -0
  390. data/pdf2json-0.52-source/xpdf/Parser.h +59 -0
  391. data/pdf2json-0.52-source/xpdf/Parser.o +0 -0
  392. data/pdf2json-0.52-source/xpdf/PreScanOutputDev.cc +257 -0
  393. data/pdf2json-0.52-source/xpdf/PreScanOutputDev.h +130 -0
  394. data/pdf2json-0.52-source/xpdf/PreScanOutputDev.o +0 -0
  395. data/pdf2json-0.52-source/xpdf/SecurityHandler.cc +390 -0
  396. data/pdf2json-0.52-source/xpdf/SecurityHandler.h +160 -0
  397. data/pdf2json-0.52-source/xpdf/SecurityHandler.o +0 -0
  398. data/pdf2json-0.52-source/xpdf/SplashOutputDev.cc +2845 -0
  399. data/pdf2json-0.52-source/xpdf/SplashOutputDev.h +247 -0
  400. data/pdf2json-0.52-source/xpdf/SplashOutputDev.o +0 -0
  401. data/pdf2json-0.52-source/xpdf/Stream-CCITT.h +459 -0
  402. data/pdf2json-0.52-source/xpdf/Stream.cc +4627 -0
  403. data/pdf2json-0.52-source/xpdf/Stream.h +858 -0
  404. data/pdf2json-0.52-source/xpdf/Stream.o +0 -0
  405. data/pdf2json-0.52-source/xpdf/TextOutputDev.cc +4090 -0
  406. data/pdf2json-0.52-source/xpdf/TextOutputDev.h +661 -0
  407. data/pdf2json-0.52-source/xpdf/TextOutputDev.o +0 -0
  408. data/pdf2json-0.52-source/xpdf/UTF8.h +56 -0
  409. data/pdf2json-0.52-source/xpdf/UnicodeMap.cc +302 -0
  410. data/pdf2json-0.52-source/xpdf/UnicodeMap.cc.old +293 -0
  411. data/pdf2json-0.52-source/xpdf/UnicodeMap.h +135 -0
  412. data/pdf2json-0.52-source/xpdf/UnicodeMap.h.old +123 -0
  413. data/pdf2json-0.52-source/xpdf/UnicodeMap.o +0 -0
  414. data/pdf2json-0.52-source/xpdf/UnicodeMapTables.h +361 -0
  415. data/pdf2json-0.52-source/xpdf/UnicodeTypeTable.cc +949 -0
  416. data/pdf2json-0.52-source/xpdf/UnicodeTypeTable.h +20 -0
  417. data/pdf2json-0.52-source/xpdf/UnicodeTypeTable.o +0 -0
  418. data/pdf2json-0.52-source/xpdf/XPDFApp.cc +447 -0
  419. data/pdf2json-0.52-source/xpdf/XPDFApp.h +114 -0
  420. data/pdf2json-0.52-source/xpdf/XPDFCore.cc +1655 -0
  421. data/pdf2json-0.52-source/xpdf/XPDFCore.h +251 -0
  422. data/pdf2json-0.52-source/xpdf/XPDFTree.cc +931 -0
  423. data/pdf2json-0.52-source/xpdf/XPDFTree.h +45 -0
  424. data/pdf2json-0.52-source/xpdf/XPDFTreeP.h +87 -0
  425. data/pdf2json-0.52-source/xpdf/XPDFViewer.cc +3488 -0
  426. data/pdf2json-0.52-source/xpdf/XPDFViewer.h +352 -0
  427. data/pdf2json-0.52-source/xpdf/XRef.cc +896 -0
  428. data/pdf2json-0.52-source/xpdf/XRef.h +133 -0
  429. data/pdf2json-0.52-source/xpdf/XRef.o +0 -0
  430. data/pdf2json-0.52-source/xpdf/XpdfPluginAPI.cc +262 -0
  431. data/pdf2json-0.52-source/xpdf/XpdfPluginAPI.h +341 -0
  432. data/pdf2json-0.52-source/xpdf/XpdfPluginAPI.o +0 -0
  433. data/pdf2json-0.52-source/xpdf/about-text.h +48 -0
  434. data/pdf2json-0.52-source/xpdf/about.xbm +6 -0
  435. data/pdf2json-0.52-source/xpdf/backArrow.xbm +6 -0
  436. data/pdf2json-0.52-source/xpdf/backArrowDis.xbm +6 -0
  437. data/pdf2json-0.52-source/xpdf/config.h +112 -0
  438. data/pdf2json-0.52-source/xpdf/dblLeftArrow.xbm +6 -0
  439. data/pdf2json-0.52-source/xpdf/dblLeftArrowDis.xbm +6 -0
  440. data/pdf2json-0.52-source/xpdf/dblRightArrow.xbm +6 -0
  441. data/pdf2json-0.52-source/xpdf/dblRightArrowDis.xbm +6 -0
  442. data/pdf2json-0.52-source/xpdf/find.xbm +6 -0
  443. data/pdf2json-0.52-source/xpdf/findDis.xbm +6 -0
  444. data/pdf2json-0.52-source/xpdf/forwardArrow.xbm +6 -0
  445. data/pdf2json-0.52-source/xpdf/forwardArrowDis.xbm +6 -0
  446. data/pdf2json-0.52-source/xpdf/leftArrow.xbm +5 -0
  447. data/pdf2json-0.52-source/xpdf/leftArrowDis.xbm +5 -0
  448. data/pdf2json-0.52-source/xpdf/libXpdf.a +0 -0
  449. data/pdf2json-0.52-source/xpdf/pdffonts +0 -0
  450. data/pdf2json-0.52-source/xpdf/pdffonts.cc +298 -0
  451. data/pdf2json-0.52-source/xpdf/pdffonts.o +0 -0
  452. data/pdf2json-0.52-source/xpdf/pdfimages +0 -0
  453. data/pdf2json-0.52-source/xpdf/pdfimages.cc +155 -0
  454. data/pdf2json-0.52-source/xpdf/pdfimages.o +0 -0
  455. data/pdf2json-0.52-source/xpdf/pdfinfo +0 -0
  456. data/pdf2json-0.52-source/xpdf/pdfinfo.cc +387 -0
  457. data/pdf2json-0.52-source/xpdf/pdfinfo.o +0 -0
  458. data/pdf2json-0.52-source/xpdf/pdftoppm.cc +203 -0
  459. data/pdf2json-0.52-source/xpdf/pdftops +0 -0
  460. data/pdf2json-0.52-source/xpdf/pdftops.cc +344 -0
  461. data/pdf2json-0.52-source/xpdf/pdftops.o +0 -0
  462. data/pdf2json-0.52-source/xpdf/pdftotext +0 -0
  463. data/pdf2json-0.52-source/xpdf/pdftotext.cc +333 -0
  464. data/pdf2json-0.52-source/xpdf/pdftotext.o +0 -0
  465. data/pdf2json-0.52-source/xpdf/print.xbm +6 -0
  466. data/pdf2json-0.52-source/xpdf/printDis.xbm +6 -0
  467. data/pdf2json-0.52-source/xpdf/rightArrow.xbm +5 -0
  468. data/pdf2json-0.52-source/xpdf/rightArrowDis.xbm +5 -0
  469. data/pdf2json-0.52-source/xpdf/vms_make.com +129 -0
  470. data/pdf2json-0.52-source/xpdf/xpdf.cc +344 -0
  471. data/pdf2json-0.52-source/xpdf/xpdfIcon.xpm +62 -0
  472. data/pdf2json.gemspec +29 -0
  473. metadata +518 -0
@@ -0,0 +1,661 @@
1
+ //========================================================================
2
+ //
3
+ // TextOutputDev.h
4
+ //
5
+ // Copyright 1997-2003 Glyph & Cog, LLC
6
+ //
7
+ //========================================================================
8
+
9
+ #ifndef TEXTOUTPUTDEV_H
10
+ #define TEXTOUTPUTDEV_H
11
+
12
+ #include <aconf.h>
13
+
14
+ #ifdef USE_GCC_PRAGMAS
15
+ #pragma interface
16
+ #endif
17
+
18
+ #include <stdio.h>
19
+ #include "gtypes.h"
20
+ #include "GfxFont.h"
21
+ #include "OutputDev.h"
22
+
23
+ class GString;
24
+ class GList;
25
+ class GfxFont;
26
+ class GfxState;
27
+ class UnicodeMap;
28
+ class Link;
29
+
30
+ class TextWord;
31
+ class TextPool;
32
+ class TextLine;
33
+ class TextLineFrag;
34
+ class TextBlock;
35
+ class TextFlow;
36
+ class TextWordList;
37
+ class TextPage;
38
+
39
+ //------------------------------------------------------------------------
40
+
41
+ typedef void (*TextOutputFunc)(void *stream, char *text, int len);
42
+
43
+ //------------------------------------------------------------------------
44
+ // TextFontInfo
45
+ //------------------------------------------------------------------------
46
+
47
+ class TextFontInfo {
48
+ public:
49
+
50
+ TextFontInfo(GfxState *state);
51
+ ~TextFontInfo();
52
+
53
+ GBool matches(GfxState *state);
54
+
55
+ #if TEXTOUT_WORD_LIST
56
+ // Get the font name (which may be NULL).
57
+ GString *getFontName() { return fontName; }
58
+
59
+ // Get font descriptor flags.
60
+ GBool isFixedWidth() { return flags & fontFixedWidth; }
61
+ GBool isSerif() { return flags & fontSerif; }
62
+ GBool isSymbolic() { return flags & fontSymbolic; }
63
+ GBool isItalic() { return flags & fontItalic; }
64
+ GBool isBold() { return flags & fontBold; }
65
+ #endif
66
+
67
+ private:
68
+
69
+ GfxFont *gfxFont;
70
+ #if TEXTOUT_WORD_LIST
71
+ GString *fontName;
72
+ int flags;
73
+ #endif
74
+
75
+ friend class TextWord;
76
+ friend class TextPage;
77
+ };
78
+
79
+ //------------------------------------------------------------------------
80
+ // TextWord
81
+ //------------------------------------------------------------------------
82
+
83
+ class TextWord {
84
+ public:
85
+
86
+ // Constructor.
87
+ TextWord(GfxState *state, int rotA, double x0, double y0,
88
+ int charPosA, TextFontInfo *fontA, double fontSize);
89
+
90
+ // Destructor.
91
+ ~TextWord();
92
+
93
+ // Add a character to the word.
94
+ void addChar(GfxState *state, double x, double y,
95
+ double dx, double dy, Unicode u);
96
+
97
+ // Merge <word> onto the end of <this>.
98
+ void merge(TextWord *word);
99
+
100
+ // Compares <this> to <word>, returning -1 (<), 0 (=), or +1 (>),
101
+ // based on a primary-axis comparison, e.g., x ordering if rot=0.
102
+ int primaryCmp(TextWord *word);
103
+
104
+ // Return the distance along the primary axis between <this> and
105
+ // <word>.
106
+ double primaryDelta(TextWord *word);
107
+
108
+ static int cmpYX(const void *p1, const void *p2);
109
+
110
+ // Get the TextFontInfo object associated with this word.
111
+ TextFontInfo *getFontInfo() { return font; }
112
+
113
+ // Get the next TextWord on the linked list.
114
+ TextWord *getNext() { return next; }
115
+
116
+ #if TEXTOUT_WORD_LIST
117
+ int getLength() { return len; }
118
+ Unicode getChar(int idx) { return text[idx]; }
119
+ GString *getText();
120
+ GString *getFontName() { return font->fontName; }
121
+ void getColor(double *r, double *g, double *b)
122
+ { *r = colorR; *g = colorG; *b = colorB; }
123
+ void getBBox(double *xMinA, double *yMinA, double *xMaxA, double *yMaxA)
124
+ { *xMinA = xMin; *yMinA = yMin; *xMaxA = xMax; *yMaxA = yMax; }
125
+ void getCharBBox(int charIdx, double *xMinA, double *yMinA,
126
+ double *xMaxA, double *yMaxA);
127
+ double getFontSize() { return fontSize; }
128
+ int getRotation() { return rot; }
129
+ int getCharPos() { return charPos; }
130
+ int getCharLen() { return charLen; }
131
+ GBool getSpaceAfter() { return spaceAfter; }
132
+ #endif
133
+
134
+ GBool isUnderlined() { return underlined; }
135
+ Link *getLink() { return link; }
136
+
137
+ private:
138
+
139
+ int rot; // rotation, multiple of 90 degrees
140
+ // (0, 1, 2, or 3)
141
+ double xMin, xMax; // bounding box x coordinates
142
+ double yMin, yMax; // bounding box y coordinates
143
+ double base; // baseline x or y coordinate
144
+ Unicode *text; // the text
145
+ double *edge; // "near" edge x or y coord of each char
146
+ // (plus one extra entry for the last char)
147
+ int len; // length of text and edge arrays
148
+ int size; // size of text and edge arrays
149
+ int charPos; // character position (within content stream)
150
+ int charLen; // number of content stream characters in
151
+ // this word
152
+ TextFontInfo *font; // font information
153
+ double fontSize; // font size
154
+ GBool spaceAfter; // set if there is a space between this
155
+ // word and the next word on the line
156
+ TextWord *next; // next word in line
157
+
158
+ #if TEXTOUT_WORD_LIST
159
+ double colorR, // word color
160
+ colorG,
161
+ colorB;
162
+ #endif
163
+
164
+ GBool underlined;
165
+ Link *link;
166
+
167
+ friend class TextPool;
168
+ friend class TextLine;
169
+ friend class TextBlock;
170
+ friend class TextFlow;
171
+ friend class TextWordList;
172
+ friend class TextPage;
173
+ };
174
+
175
+ //------------------------------------------------------------------------
176
+ // TextPool
177
+ //------------------------------------------------------------------------
178
+
179
+ class TextPool {
180
+ public:
181
+
182
+ TextPool();
183
+ ~TextPool();
184
+
185
+ TextWord *getPool(int baseIdx) { return pool[baseIdx - minBaseIdx]; }
186
+ void setPool(int baseIdx, TextWord *p) { pool[baseIdx - minBaseIdx] = p; }
187
+
188
+ int getBaseIdx(double base);
189
+
190
+ void addWord(TextWord *word);
191
+
192
+ private:
193
+
194
+ int minBaseIdx; // min baseline bucket index
195
+ int maxBaseIdx; // max baseline bucket index
196
+ TextWord **pool; // array of linked lists, one for each
197
+ // baseline value (multiple of 4 pts)
198
+ TextWord *cursor; // pointer to last-accessed word
199
+ int cursorBaseIdx; // baseline bucket index of last-accessed word
200
+
201
+ friend class TextBlock;
202
+ friend class TextPage;
203
+ };
204
+
205
+ //------------------------------------------------------------------------
206
+ // TextLine
207
+ //------------------------------------------------------------------------
208
+
209
+ class TextLine {
210
+ public:
211
+
212
+ TextLine(TextBlock *blkA, int rotA, double baseA);
213
+ ~TextLine();
214
+
215
+ void addWord(TextWord *word);
216
+
217
+ // Return the distance along the primary axis between <this> and
218
+ // <line>.
219
+ double primaryDelta(TextLine *line);
220
+
221
+ // Compares <this> to <line>, returning -1 (<), 0 (=), or +1 (>),
222
+ // based on a primary-axis comparison, e.g., x ordering if rot=0.
223
+ int primaryCmp(TextLine *line);
224
+
225
+ // Compares <this> to <line>, returning -1 (<), 0 (=), or +1 (>),
226
+ // based on a secondary-axis comparison of the baselines, e.g., y
227
+ // ordering if rot=0.
228
+ int secondaryCmp(TextLine *line);
229
+
230
+ int cmpYX(TextLine *line);
231
+
232
+ static int cmpXY(const void *p1, const void *p2);
233
+
234
+ void coalesce(UnicodeMap *uMap);
235
+
236
+ // Get the head of the linked list of TextWords.
237
+ TextWord *getWords() { return words; }
238
+
239
+ // Get the next TextLine on the linked list.
240
+ TextLine *getNext() { return next; }
241
+
242
+ // Returns true if the last char of the line is a hyphen.
243
+ GBool isHyphenated() { return hyphenated; }
244
+
245
+ private:
246
+
247
+ TextBlock *blk; // parent block
248
+ int rot; // text rotation
249
+ double xMin, xMax; // bounding box x coordinates
250
+ double yMin, yMax; // bounding box y coordinates
251
+ double base; // baseline x or y coordinate
252
+ TextWord *words; // words in this line
253
+ TextWord *lastWord; // last word in this line
254
+ Unicode *text; // Unicode text of the line, including
255
+ // spaces between words
256
+ double *edge; // "near" edge x or y coord of each char
257
+ // (plus one extra entry for the last char)
258
+ int *col; // starting column number of each Unicode char
259
+ int len; // number of Unicode chars
260
+ int convertedLen; // total number of converted characters
261
+ GBool hyphenated; // set if last char is a hyphen
262
+ TextLine *next; // next line in block
263
+
264
+ friend class TextLineFrag;
265
+ friend class TextBlock;
266
+ friend class TextFlow;
267
+ friend class TextWordList;
268
+ friend class TextPage;
269
+ };
270
+
271
+ //------------------------------------------------------------------------
272
+ // TextBlock
273
+ //------------------------------------------------------------------------
274
+
275
+ class TextBlock {
276
+ public:
277
+
278
+ TextBlock(TextPage *pageA, int rotA);
279
+ ~TextBlock();
280
+
281
+ void addWord(TextWord *word);
282
+
283
+ void coalesce(UnicodeMap *uMap);
284
+
285
+ // Update this block's priMin and priMax values, looking at <blk>.
286
+ void updatePriMinMax(TextBlock *blk);
287
+
288
+ static int cmpXYPrimaryRot(const void *p1, const void *p2);
289
+
290
+ static int cmpYXPrimaryRot(const void *p1, const void *p2);
291
+
292
+ int primaryCmp(TextBlock *blk);
293
+
294
+ double secondaryDelta(TextBlock *blk);
295
+
296
+ // Returns true if <this> is below <blk>, relative to the page's
297
+ // primary rotation.
298
+ GBool isBelow(TextBlock *blk);
299
+
300
+ // Get the head of the linked list of TextLines.
301
+ TextLine *getLines() { return lines; }
302
+
303
+ // Get the next TextBlock on the linked list.
304
+ TextBlock *getNext() { return next; }
305
+
306
+ private:
307
+
308
+ TextPage *page; // the parent page
309
+ int rot; // text rotation
310
+ double xMin, xMax; // bounding box x coordinates
311
+ double yMin, yMax; // bounding box y coordinates
312
+ double priMin, priMax; // whitespace bounding box along primary axis
313
+
314
+ TextPool *pool; // pool of words (used only until lines
315
+ // are built)
316
+ TextLine *lines; // linked list of lines
317
+ TextLine *curLine; // most recently added line
318
+ int nLines; // number of lines
319
+ int charCount; // number of characters in the block
320
+ int col; // starting column
321
+ int nColumns; // number of columns in the block
322
+
323
+ TextBlock *next;
324
+ TextBlock *stackNext;
325
+
326
+ friend class TextLine;
327
+ friend class TextLineFrag;
328
+ friend class TextFlow;
329
+ friend class TextWordList;
330
+ friend class TextPage;
331
+ };
332
+
333
+ //------------------------------------------------------------------------
334
+ // TextFlow
335
+ //------------------------------------------------------------------------
336
+
337
+ class TextFlow {
338
+ public:
339
+
340
+ TextFlow(TextPage *pageA, TextBlock *blk);
341
+ ~TextFlow();
342
+
343
+ // Add a block to the end of this flow.
344
+ void addBlock(TextBlock *blk);
345
+
346
+ // Returns true if <blk> fits below <prevBlk> in the flow, i.e., (1)
347
+ // it uses a font no larger than the last block added to the flow,
348
+ // and (2) it fits within the flow's [priMin, priMax] along the
349
+ // primary axis.
350
+ GBool blockFits(TextBlock *blk, TextBlock *prevBlk);
351
+
352
+ // Get the head of the linked list of TextBlocks.
353
+ TextBlock *getBlocks() { return blocks; }
354
+
355
+ // Get the next TextFlow on the linked list.
356
+ TextFlow *getNext() { return next; }
357
+
358
+ private:
359
+
360
+ TextPage *page; // the parent page
361
+ double xMin, xMax; // bounding box x coordinates
362
+ double yMin, yMax; // bounding box y coordinates
363
+ double priMin, priMax; // whitespace bounding box along primary axis
364
+ TextBlock *blocks; // blocks in flow
365
+ TextBlock *lastBlk; // last block in this flow
366
+ TextFlow *next;
367
+
368
+ friend class TextWordList;
369
+ friend class TextPage;
370
+ };
371
+
372
+ #if TEXTOUT_WORD_LIST
373
+
374
+ //------------------------------------------------------------------------
375
+ // TextWordList
376
+ //------------------------------------------------------------------------
377
+
378
+ class TextWordList {
379
+ public:
380
+
381
+ // Build a flat word list, in content stream order (if
382
+ // text->rawOrder is true), physical layout order (if <physLayout>
383
+ // is true and text->rawOrder is false), or reading order (if both
384
+ // flags are false).
385
+ TextWordList(TextPage *text, GBool physLayout);
386
+
387
+ ~TextWordList();
388
+
389
+ // Return the number of words on the list.
390
+ int getLength();
391
+
392
+ // Return the <idx>th word from the list.
393
+ TextWord *get(int idx);
394
+
395
+ private:
396
+
397
+ GList *words; // [TextWord]
398
+ };
399
+
400
+ #endif // TEXTOUT_WORD_LIST
401
+
402
+ //------------------------------------------------------------------------
403
+ // TextPage
404
+ //------------------------------------------------------------------------
405
+
406
+ class TextPage {
407
+ public:
408
+
409
+ // Constructor.
410
+ TextPage(GBool rawOrderA);
411
+
412
+ // Destructor.
413
+ ~TextPage();
414
+
415
+ // Start a new page.
416
+ void startPage(GfxState *state);
417
+
418
+ // End the current page.
419
+ void endPage();
420
+
421
+ // Update the current font.
422
+ void updateFont(GfxState *state);
423
+
424
+ // Begin a new word.
425
+ void beginWord(GfxState *state, double x0, double y0);
426
+
427
+ // Add a character to the current word.
428
+ void addChar(GfxState *state, double x, double y,
429
+ double dx, double dy,
430
+ CharCode c, int nBytes, Unicode *u, int uLen);
431
+
432
+ // End the current word, sorting it into the list of words.
433
+ void endWord();
434
+
435
+ // Add a word, sorting it into the list of words.
436
+ void addWord(TextWord *word);
437
+
438
+ // Add a (potential) underline.
439
+ void addUnderline(double x0, double y0, double x1, double y1);
440
+
441
+ // Add a hyperlink.
442
+ void addLink(int xMin, int yMin, int xMax, int yMax, Link *link);
443
+
444
+ // Coalesce strings that look like parts of the same line.
445
+ void coalesce(GBool physLayout, GBool doHTML);
446
+
447
+ // Find a string. If <startAtTop> is true, starts looking at the
448
+ // top of the page; else if <startAtLast> is true, starts looking
449
+ // immediately after the last find result; else starts looking at
450
+ // <xMin>,<yMin>. If <stopAtBottom> is true, stops looking at the
451
+ // bottom of the page; else if <stopAtLast> is true, stops looking
452
+ // just before the last find result; else stops looking at
453
+ // <xMax>,<yMax>.
454
+ GBool findText(Unicode *s, int len,
455
+ GBool startAtTop, GBool stopAtBottom,
456
+ GBool startAtLast, GBool stopAtLast,
457
+ GBool caseSensitive, GBool backward,
458
+ double *xMin, double *yMin,
459
+ double *xMax, double *yMax);
460
+
461
+ // Get the text which is inside the specified rectangle.
462
+ GString *getText(double xMin, double yMin,
463
+ double xMax, double yMax);
464
+
465
+ // Find a string by character position and length. If found, sets
466
+ // the text bounding rectangle and returns true; otherwise returns
467
+ // false.
468
+ GBool findCharRange(int pos, int length,
469
+ double *xMin, double *yMin,
470
+ double *xMax, double *yMax);
471
+
472
+ // Dump contents of page to a file.
473
+ void dump(void *outputStream, TextOutputFunc outputFunc,
474
+ GBool physLayout);
475
+
476
+ // Get the head of the linked list of TextFlows.
477
+ TextFlow *getFlows() { return flows; }
478
+
479
+ #if TEXTOUT_WORD_LIST
480
+ // Build a flat word list, in content stream order (if
481
+ // this->rawOrder is true), physical layout order (if <physLayout>
482
+ // is true and this->rawOrder is false), or reading order (if both
483
+ // flags are false).
484
+ TextWordList *makeWordList(GBool physLayout);
485
+ #endif
486
+
487
+ private:
488
+
489
+ void clear();
490
+ void assignColumns(TextLineFrag *frags, int nFrags, int rot);
491
+ int dumpFragment(Unicode *text, int len, UnicodeMap *uMap, GString *s);
492
+
493
+ GBool rawOrder; // keep text in content stream order
494
+
495
+ double pageWidth, pageHeight; // width and height of current page
496
+ TextWord *curWord; // currently active string
497
+ int charPos; // next character position (within content
498
+ // stream)
499
+ TextFontInfo *curFont; // current font
500
+ double curFontSize; // current font size
501
+ int nest; // current nesting level (for Type 3 fonts)
502
+ int nTinyChars; // number of "tiny" chars seen so far
503
+ GBool lastCharOverlap; // set if the last added char overlapped the
504
+ // previous char
505
+
506
+ TextPool *pools[4]; // a "pool" of TextWords for each rotation
507
+ TextFlow *flows; // linked list of flows
508
+ TextBlock **blocks; // array of blocks, in yx order
509
+ int nBlocks; // number of blocks
510
+ int primaryRot; // primary rotation
511
+ GBool primaryLR; // primary direction (true means L-to-R,
512
+ // false means R-to-L)
513
+ TextWord *rawWords; // list of words, in raw order (only if
514
+ // rawOrder is set)
515
+ TextWord *rawLastWord; // last word on rawWords list
516
+
517
+ GList *fonts; // all font info objects used on this
518
+ // page [TextFontInfo]
519
+
520
+ double lastFindXMin, // coordinates of the last "find" result
521
+ lastFindYMin;
522
+ GBool haveLastFind;
523
+
524
+ GList *underlines; // [TextUnderline]
525
+ GList *links; // [TextLink]
526
+
527
+ friend class TextLine;
528
+ friend class TextLineFrag;
529
+ friend class TextBlock;
530
+ friend class TextFlow;
531
+ friend class TextWordList;
532
+ };
533
+
534
+ //------------------------------------------------------------------------
535
+ // TextOutputDev
536
+ //------------------------------------------------------------------------
537
+
538
+ class TextOutputDev: public OutputDev {
539
+ public:
540
+
541
+ // Open a text output file. If <fileName> is NULL, no file is
542
+ // written (this is useful, e.g., for searching text). If
543
+ // <physLayoutA> is true, the original physical layout of the text
544
+ // is maintained. If <rawOrder> is true, the text is kept in
545
+ // content stream order.
546
+ TextOutputDev(char *fileName, GBool physLayoutA,
547
+ GBool rawOrderA, GBool append);
548
+
549
+ // Create a TextOutputDev which will write to a generic stream. If
550
+ // <physLayoutA> is true, the original physical layout of the text
551
+ // is maintained. If <rawOrder> is true, the text is kept in
552
+ // content stream order.
553
+ TextOutputDev(TextOutputFunc func, void *stream,
554
+ GBool physLayoutA, GBool rawOrderA);
555
+
556
+ // Destructor.
557
+ virtual ~TextOutputDev();
558
+
559
+ // Check if file was successfully created.
560
+ virtual GBool isOk() { return ok; }
561
+
562
+ //---- get info about output device
563
+
564
+ // Does this device use upside-down coordinates?
565
+ // (Upside-down means (0,0) is the top left corner of the page.)
566
+ virtual GBool upsideDown() { return gTrue; }
567
+
568
+ // Does this device use drawChar() or drawString()?
569
+ virtual GBool useDrawChar() { return gTrue; }
570
+
571
+ // Does this device use beginType3Char/endType3Char? Otherwise,
572
+ // text in Type 3 fonts will be drawn with drawChar/drawString.
573
+ virtual GBool interpretType3Chars() { return gFalse; }
574
+
575
+ // Does this device need non-text content?
576
+ virtual GBool needNonText() { return gFalse; }
577
+
578
+ //----- initialization and control
579
+
580
+ // Start a page.
581
+ virtual void startPage(int pageNum, GfxState *state);
582
+
583
+ // End a page.
584
+ virtual void endPage();
585
+
586
+ //----- update text state
587
+ virtual void updateFont(GfxState *state);
588
+
589
+ //----- text drawing
590
+ virtual void beginString(GfxState *state, GString *s);
591
+ virtual void endString(GfxState *state);
592
+ virtual void drawChar(GfxState *state, double x, double y,
593
+ double dx, double dy,
594
+ double originX, double originY,
595
+ CharCode c, int nBytes, Unicode *u, int uLen);
596
+
597
+ //----- path painting
598
+ virtual void stroke(GfxState *state);
599
+ virtual void fill(GfxState *state);
600
+ virtual void eoFill(GfxState *state);
601
+
602
+ //----- link borders
603
+ virtual void processLink(Link *link, Catalog *catalog);
604
+
605
+ //----- special access
606
+
607
+ // Find a string. If <startAtTop> is true, starts looking at the
608
+ // top of the page; else if <startAtLast> is true, starts looking
609
+ // immediately after the last find result; else starts looking at
610
+ // <xMin>,<yMin>. If <stopAtBottom> is true, stops looking at the
611
+ // bottom of the page; else if <stopAtLast> is true, stops looking
612
+ // just before the last find result; else stops looking at
613
+ // <xMax>,<yMax>.
614
+ GBool findText(Unicode *s, int len,
615
+ GBool startAtTop, GBool stopAtBottom,
616
+ GBool startAtLast, GBool stopAtLast,
617
+ GBool caseSensitive, GBool backward,
618
+ double *xMin, double *yMin,
619
+ double *xMax, double *yMax);
620
+
621
+ // Get the text which is inside the specified rectangle.
622
+ GString *getText(double xMin, double yMin,
623
+ double xMax, double yMax);
624
+
625
+ // Find a string by character position and length. If found, sets
626
+ // the text bounding rectangle and returns true; otherwise returns
627
+ // false.
628
+ GBool findCharRange(int pos, int length,
629
+ double *xMin, double *yMin,
630
+ double *xMax, double *yMax);
631
+
632
+ #if TEXTOUT_WORD_LIST
633
+ // Build a flat word list, in content stream order (if
634
+ // this->rawOrder is true), physical layout order (if
635
+ // this->physLayout is true and this->rawOrder is false), or reading
636
+ // order (if both flags are false).
637
+ TextWordList *makeWordList();
638
+ #endif
639
+
640
+ // Returns the TextPage object for the last rasterized page,
641
+ // transferring ownership to the caller.
642
+ TextPage *takeText();
643
+
644
+ // Turn extra processing for HTML conversion on or off.
645
+ void enableHTMLExtras(GBool doHTMLA) { doHTML = doHTMLA; }
646
+
647
+ private:
648
+
649
+ TextOutputFunc outputFunc; // output function
650
+ void *outputStream; // output stream
651
+ GBool needClose; // need to close the output file?
652
+ // (only if outputStream is a FILE*)
653
+ TextPage *text; // text for the current page
654
+ GBool physLayout; // maintain original physical layout when
655
+ // dumping text
656
+ GBool rawOrder; // keep text in content stream order
657
+ GBool doHTML; // extra processing for HTML conversion
658
+ GBool ok; // set up ok?
659
+ };
660
+
661
+ #endif