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
data/pdf2json-0.52-source/src/ImgOutputDev.h ADDED
@@ -0,0 +1,307 @@
1
+ //========================================================================
2
+ //
3
+ // ImgOutputDev.h
4
+ //
5
+ // Copyright 1997 Derek B. Noonburg
6
+ //
7
+ // Changed 1999 by G.Ovtcharov
8
+ //========================================================================
9
+
10
+ #ifndef HTMLOUTPUTDEV_H
11
+ #define HTMLOUTPUTDEV_H
12
+
13
+ #ifdef __GNUC__
14
+ #pragma interface
15
+ #endif
16
+
17
+ #include <stdio.h>
18
+ #include "gtypes.h"
19
+ #include "GList.h"
20
+ #include "GfxFont.h"
21
+ #include "OutputDev.h"
22
+ #include "XmlLinks.h"
23
+ #include "XmlFonts.h"
24
+ #include "Link.h"
25
+ #include "Catalog.h"
26
+ #include "UnicodeMap.h"
27
+
28
+
29
+ #ifdef WIN32
30
+ # define SLASH '\\'
31
+ #else
32
+ # define SLASH '/'
33
+ #endif
34
+
35
+ #define xoutRound(x) ((int)(x + 0.5))
36
+ #define xoutRoundLower(x) ((int)(x - 0.5))
37
+
38
+ #define DOCTYPE "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\">"
39
+ #define DOCTYPE_FRAMES "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Frameset//EN\"\n\"http://www.w3.org/TR/html4/frameset.dtd\">"
40
+
41
+ class GfxState;
42
+ class GString;
43
+ //------------------------------------------------------------------------
44
+ // HtmlString
45
+ //------------------------------------------------------------------------
46
+
47
+
48
+
49
+ class HtmlString {
50
+ public:
51
+
52
+ // Constructor.
53
+ HtmlString(GfxState *state, double fontSize, double charspace, XmlFontAccu* fonts);
54
+
55
+ // Destructor.
56
+ ~HtmlString();
57
+
58
+ // Add a character to the string.
59
+ void addChar(GfxState *state, double x, double y,
60
+ double dx, double dy,
61
+ Unicode u);
62
+ XmlLink* getLink() { return link; }
63
+ void endString(); // postprocessing
64
+
65
+ private:
66
+ // aender die text variable
67
+ XmlLink *link;
68
+ double xMin, xMax; // bounding box x coordinates
69
+ double yMin, yMax; // bounding box y coordinates
70
+ int col; // starting column
71
+ Unicode *text; // the text
72
+ double *xRight; // right-hand x coord of each char
73
+ HtmlString *yxNext; // next string in y-major order
74
+ HtmlString *xyNext; // next string in x-major order
75
+ int fontpos;
76
+ GString* htext;
77
+ GString* htext2;
78
+ int strSize;
79
+ int len; // length of text and xRight
80
+ int size; // size of text and xRight arrays
81
+ UnicodeTextDirection dir; // direction (left to right/right to left)
82
+
83
+ friend class HtmlPage;
84
+
85
+ };
86
+
87
+
88
+ //------------------------------------------------------------------------
89
+ // HtmlPage
90
+ //------------------------------------------------------------------------
91
+ class HtmlPage {
92
+ public:
93
+
94
+ // Constructor.
95
+ HtmlPage(GBool rawOrder, GBool textAsJSON, GBool compressData, char *imgExtVal);
96
+
97
+ // Destructor.
98
+ ~HtmlPage();
99
+
100
+ // Begin a new string.
101
+ void beginString(GfxState *state, GString *s);
102
+
103
+ // Add a character to the current string.
104
+ void addChar(GfxState *state, double x, double y,
105
+ double dx, double dy,
106
+ double ox, double oy,
107
+ Unicode *u, int uLen); //Guchar c);
108
+
109
+ void updateFont(GfxState *state);
110
+ void updateCharSpace(GfxState *state);
111
+
112
+ // End the current string, sorting it into the list of strings.
113
+ void endString();
114
+
115
+ // Coalesce strings that look like parts of the same line.
116
+ void coalesce();
117
+
118
+ // Find a string. If <top> is true, starts looking at top of page;
119
+ // otherwise starts looking at <xMin>,<yMin>. If <bottom> is true,
120
+ // stops looking at bottom of page; otherwise stops looking at
121
+ // <xMax>,<yMax>. If found, sets the text bounding rectange and
122
+ // returns true; otherwise returns false.
123
+
124
+
125
+ // new functions
126
+ void AddLink(const XmlLink& x){
127
+ links->AddLink(x);
128
+ }
129
+
130
+ void dump(FILE *f, int pageNum, GBool passedFirstPage);
131
+
132
+ // Clear the page.
133
+ void clear();
134
+
135
+ void conv();
136
+ private:
137
+ XmlFont* getFont(HtmlString *hStr) { return fonts->Get(hStr->fontpos); }
138
+
139
+ double fontSize; // current font size
140
+ GBool rawOrder; // keep strings in content stream order
141
+ GBool textAsJSON; // print text data in JSON format
142
+ GBool compressData; // compress data to a minimum format
143
+ double charspace;
144
+ HtmlString *curStr; // currently active string
145
+
146
+ HtmlString *yxStrings; // strings in y-major order
147
+ HtmlString *xyStrings; // strings in x-major order
148
+ HtmlString *yxCur1, *yxCur2; // cursors for yxStrings list
149
+
150
+ void setDocName(char* fname);
151
+ void dumpAsXML(FILE* f,int page, GBool passedFirstPage);
152
+ void dumpComplex(FILE* f, int page);
153
+
154
+ // marks the position of the fonts that belong to current page (for noframes)
155
+ int fontsPageMarker;
156
+ XmlFontAccu *fonts;
157
+ XmlLinks *links;
158
+
159
+ GString *DocName;
160
+ GString *imgExt;
161
+ int pageWidth;
162
+ int pageHeight;
163
+ int X1;
164
+ int X2;
165
+ int Y1;
166
+ int Y2;
167
+ static int pgNum;
168
+ int firstPage; // used to begin the numeration of pages
169
+
170
+ friend class ImgOutputDev;
171
+ };
172
+
173
+ //------------------------------------------------------------------------
174
+ // HtmlMetaVar
175
+ //------------------------------------------------------------------------
176
+ class HtmlMetaVar {
177
+ public:
178
+ HtmlMetaVar(char *_name, char *_content);
179
+ ~HtmlMetaVar();
180
+
181
+ GString* toString();
182
+
183
+ private:
184
+
185
+ GString *name;
186
+ GString *content;
187
+ };
188
+
189
+ //------------------------------------------------------------------------
190
+ // ImgOutputDev
191
+ //------------------------------------------------------------------------
192
+
193
+ class ImgOutputDev: public OutputDev {
194
+ public:
195
+
196
+ // Open a text output file. If <fileName> is NULL, no file is written
197
+ // (this is useful, e.g., for searching text). If <useASCII7> is true,
198
+ // text is converted to 7-bit ASCII; otherwise, text is converted to
199
+ // 8-bit ISO Latin-1. <useASCII7> should also be set for Japanese
200
+ // (EUC-JP) text. If <rawOrder> is true, the text is kept in content
201
+ // stream order.
202
+ ImgOutputDev(char *fileName, char *title,
203
+ char *author,
204
+ char *keywords,
205
+ char *subject,
206
+ char *date,
207
+ char *extension,
208
+ GBool rawOrder,
209
+ GBool textAsJSON,
210
+ GBool compressData,
211
+ int firstPage = 1,
212
+ GBool outline = 0,
213
+ int numPages = 0);
214
+
215
+ // Destructor.
216
+ virtual ~ImgOutputDev();
217
+
218
+ // Check if file was successfully created.
219
+ virtual GBool isOk() { return ok; }
220
+
221
+ //---- get info about output device
222
+
223
+ // Does this device use upside-down coordinates?
224
+ // (Upside-down means (0,0) is the top left corner of the page.)
225
+ virtual GBool upsideDown() { return gTrue; }
226
+
227
+ // Does this device use drawChar() or drawString()?
228
+ virtual GBool useDrawChar() { return gTrue; }
229
+
230
+ // Does this device use beginType3Char/endType3Char? Otherwise,
231
+ // text in Type 3 fonts will be drawn with drawChar/drawString.
232
+ virtual GBool interpretType3Chars() { return gFalse; }
233
+
234
+ // Does this device need non-text content?
235
+ virtual GBool needNonText() { return gFalse; }
236
+
237
+ //----- initialization and control
238
+
239
+ // Start a page.
240
+ virtual void startPage(int pageNum, GfxState *state, double x1,double y1,double x2,double y2);
241
+
242
+ // End a page.
243
+ virtual void endPage();
244
+
245
+ //----- update text state
246
+ virtual void updateFont(GfxState *state);
247
+ virtual void updateCharSpace(GfxState *state);
248
+
249
+ //----- text drawing
250
+ virtual void beginString(GfxState *state, GString *s);
251
+ virtual void endString(GfxState *state);
252
+ virtual void drawChar(GfxState *state, double x, double y,
253
+ double dx, double dy,
254
+ double originX, double originY,
255
+ CharCode code,int nBytes, Unicode *u, int uLen);
256
+
257
+ virtual void drawImageMask(GfxState *state, Object *ref,
258
+ Stream *str,
259
+ int width, int height, GBool invert,
260
+ GBool inlineImg);
261
+ virtual void drawImage(GfxState *state, Object *ref, Stream *str,
262
+ int width, int height, GfxImageColorMap *colorMap,
263
+ int *maskColors, GBool inlineImg);
264
+
265
+ //new feature
266
+ virtual int DevType() {return 1234;}
267
+ virtual void drawLink(Link *link,Catalog *cat);
268
+
269
+ int getPageWidth() { return maxPageWidth; }
270
+ int getPageHeight() { return maxPageHeight; }
271
+
272
+ GBool dumpDocOutline(Catalog* catalog);
273
+
274
+ private:
275
+ // convert encoding into a HTML standard, or encoding->getCString if not
276
+ // recognized
277
+ static char* mapEncodingToHtml(GString* encoding);
278
+ GString* getLinkDest(Link *link,Catalog *catalog);
279
+ void dumpMetaVars(FILE *);
280
+ void doFrame(int firstPage);
281
+ GBool newOutlineLevel(FILE *output, Object *node, Catalog* catalog, int level = 1);
282
+
283
+ FILE *fContentsFrame;
284
+ FILE *page; // html file
285
+ //FILE *tin; // image log file
286
+ //GBool write;
287
+ GBool needClose; // need to close the file?
288
+ HtmlPage *pages; // text for the current page
289
+ GBool rawOrder; // keep text in content stream order
290
+ GBool textAsJSON; // print text data in JSON format
291
+ GBool compressData; // use compressed data mode
292
+ GBool doOutline; // output document outline
293
+ GBool ok; // set up ok?
294
+ GBool dumpJPEG;
295
+ GBool passedFirstPage;
296
+ int pageNum;
297
+ int maxPageWidth;
298
+ int maxPageHeight;
299
+ static int imgNum;
300
+ GString *Docname;
301
+ GString *docTitle;
302
+ GList *glMetaVars;
303
+ int numPages;
304
+ friend class HtmlPage;
305
+ };
306
+
307
+ #endif
data/pdf2json-0.52-source/src/Makefile ADDED
@@ -0,0 +1,68 @@
1
+
2
+ SHELL = /bin/sh
3
+
4
+ SRCDIR = .
5
+ XPDFSRCDIR = ../xpdf
6
+ XPDFLIBDIR = ../xpdf
7
+ GOOSRCDIR = ../goo
8
+ GOOLIBDIR = ../goo
9
+ FOFISRCDIR = ../fofi
10
+ FOFILIBDIR = ../fofi
11
+ SPLASHSRCDIR = ../splash
12
+ SPLASHLIBDIR = ../splash
13
+
14
+ CXXFLAGS = -I/usr/local/include -g -O2 -DHAVE_CONFIG_H -DHAVE_DIRENT_H=1 -I.. -DHAVE_REWINDDIR=1 -DHAVE_POPEN=1 -I.. -I$(GOOSRCDIR) -I$(XPDFSRCDIR) -I$(FOFISRCDIR) -I$(SPLASHSRCDIR) -I$(srcdir) -I/usr/X11R6/include
15
+
16
+ LDFLAGS =
17
+ FTLIBS =
18
+
19
+ OTHERLIBS =
20
+
21
+ CXX ?= c++
22
+
23
+ LIBPREFIX = lib
24
+ EXE =
25
+
26
+ #------------------------------------------------------------------------
27
+
28
+ .SUFFIXES: .cc
29
+
30
+ .cc.o:
31
+ $(CXX) $(CXXFLAGS) -c $<
32
+
33
+ #------------------------------------------------------------------------
34
+
35
+ CXX_SRC = \
36
+ $(SRCDIR)/pdf2json.cc \
37
+ $(SRCDIR)/ImgOutputDev.cc \
38
+ $(SRCDIR)/XmlFonts.cc \
39
+ $(SRCDIR)/XmlLinks.cc
40
+
41
+ #------------------------------------------------------------------------
42
+
43
+ all: pdf2json$(EXE)
44
+
45
+ #-------------------------------------------------------------------------
46
+
47
+ PDF2JSON_OBJS = ImgOutputDev.o XmlFonts.o XmlLinks.o \
48
+ pdf2json.o
49
+ PDF2JSON_LIBS = -L$(GOOLIBDIR) -L$(FOFILIBDIR) -L$(SPLASHLIBDIR) $(FTLIBS) -L$(XPDFLIBDIR) $(OTHERLIBS) -lXpdf -lGoo -lfofi -lsplash -lm
50
+
51
+ pdf2json$(EXE): $(PDF2JSON_OBJS) $(GOOLIBDIR)/$(LIBPREFIX)Goo.a
52
+ $(CXX) $(CXXFLAGS) $(LDFLAGS) -o pdf2json$(EXE) $(PDF2JSON_OBJS) \
53
+ $(PDF2JSON_LIBS)
54
+
55
+ #-------------------------------------------------------------------------
56
+ PDF2JSON_WINOBJS = pdf2json.exe ImgOutPutDev.obj pdf2json.obj XmlFonts.obj XmlLinks.obj
57
+
58
+ clean:
59
+ rm -f $(PDF2JSON_OBJS) pdf2json$(EXE)
60
+ rm -f $(PDF2JSON_WINOBJS)
61
+
62
+ #------------------------------------------------------------------------
63
+
64
+ distdepend:
65
+ cp Makefile.in Makefile.in.bak
66
+ sed '/^#----- dependences -----/q' Makefile.in.bak >Makefile.in
67
+ $(CXX) $(CXXFLAGS) -MM $(CXX_SRC) >>Makefile.in
68
+
data/pdf2json-0.52-source/src/Makefile.in ADDED
@@ -0,0 +1,68 @@
1
+
2
+ SHELL = /bin/sh
3
+
4
+ SRCDIR = .
5
+ XPDFSRCDIR = ../xpdf
6
+ XPDFLIBDIR = ../xpdf
7
+ GOOSRCDIR = ../goo
8
+ GOOLIBDIR = ../goo
9
+ FOFISRCDIR = ../fofi
10
+ FOFILIBDIR = ../fofi
11
+ SPLASHSRCDIR = ../splash
12
+ SPLASHLIBDIR = ../splash
13
+
14
+ CXXFLAGS = -I/usr/local/include -g -O2 -DHAVE_CONFIG_H -DHAVE_DIRENT_H=1 -I.. -DHAVE_REWINDDIR=1 -DHAVE_POPEN=1 -I.. -I$(GOOSRCDIR) -I$(XPDFSRCDIR) -I$(FOFISRCDIR) -I$(SPLASHSRCDIR) -I$(srcdir) -I/usr/X11R6/include
15
+
16
+ LDFLAGS =
17
+ FTLIBS =
18
+
19
+ OTHERLIBS =
20
+
21
+ CXX ?= c++
22
+
23
+ LIBPREFIX = lib
24
+ EXE =
25
+
26
+ #------------------------------------------------------------------------
27
+
28
+ .SUFFIXES: .cc
29
+
30
+ .cc.o:
31
+ $(CXX) $(CXXFLAGS) -c $<
32
+
33
+ #------------------------------------------------------------------------
34
+
35
+ CXX_SRC = \
36
+ $(SRCDIR)/pdf2json.cc \
37
+ $(SRCDIR)/ImgOutputDev.cc \
38
+ $(SRCDIR)/XmlFonts.cc \
39
+ $(SRCDIR)/XmlLinks.cc
40
+
41
+ #------------------------------------------------------------------------
42
+
43
+ all: pdf2json$(EXE)
44
+
45
+ #-------------------------------------------------------------------------
46
+
47
+ PDF2JSON_OBJS = ImgOutputDev.o XmlFonts.o XmlLinks.o \
48
+ pdf2json.o
49
+ PDF2JSON_LIBS = -L$(GOOLIBDIR) -L$(FOFILIBDIR) -L$(SPLASHLIBDIR) $(FTLIBS) -L$(XPDFLIBDIR) $(OTHERLIBS) -lXpdf -lGoo -lfofi -lsplash -lm
50
+
51
+ pdf2json$(EXE): $(PDF2JSON_OBJS) $(GOOLIBDIR)/$(LIBPREFIX)Goo.a
52
+ $(CXX) $(CXXFLAGS) $(LDFLAGS) -o pdf2json$(EXE) $(PDF2JSON_OBJS) \
53
+ $(PDF2JSON_LIBS)
54
+
55
+ #-------------------------------------------------------------------------
56
+ PDF2JSON_WINOBJS = pdf2json.exe ImgOutPutDev.obj pdf2json.obj XmlFonts.obj XmlLinks.obj
57
+
58
+ clean:
59
+ rm -f $(PDF2JSON_OBJS) pdf2json$(EXE)
60
+ rm -f $(PDF2JSON_WINOBJS)
61
+
62
+ #------------------------------------------------------------------------
63
+
64
+ distdepend:
65
+ cp Makefile.in Makefile.in.bak
66
+ sed '/^#----- dependences -----/q' Makefile.in.bak >Makefile.in
67
+ $(CXX) $(CXXFLAGS) -MM $(CXX_SRC) >>Makefile.in
68
+
data/pdf2json-0.52-source/src/XmlFonts.cc ADDED
@@ -0,0 +1,367 @@
1
+ #include "XmlFonts.h"
2
+ #include "GlobalParams.h"
3
+ #include "UnicodeMap.h"
4
+ #include <stdio.h>
5
+
6
+ #ifndef _WIN32
7
+ #include <unistd.h>
8
+ #endif
9
+
10
+ struct Fonts{
11
+ char *Fontname;
12
+ char *name;
13
+ };
14
+
15
+ const int font_num=13;
16
+
17
+ static Fonts fonts[font_num+1]={
18
+ {"Courier", "Courier" },
19
+ {"Courier-Bold", "Courier"},
20
+ {"Courier-BoldOblique", "Courier"},
21
+ {"Courier-Oblique", "Courier"},
22
+ {"Helvetica", "Helvetica"},
23
+ {"Helvetica-Bold", "Helvetica"},
24
+ {"Helvetica-BoldOblique", "Helvetica"},
25
+ {"Helvetica-Oblique", "Helvetica"},
26
+ {"Symbol", "Symbol" },
27
+ {"Times-Bold", "Times" },
28
+ {"Times-BoldItalic", "Times" },
29
+ {"Times-Italic", "Times" },
30
+ {"Times-Roman", "Times" },
31
+ {" " , "Times" },
32
+ };
33
+
34
+ #define xoutRound(x) ((int)(x + 0.5))
35
+ extern GBool xml;
36
+
37
+ GString* XmlFont::DefaultFont=new GString("Times"); // Arial,Helvetica,sans-serif
38
+
39
+ XmlFontColor::XmlFontColor(GfxRGB rgb){
40
+ /*
41
+ r=static_cast<int>(255*rgb.r);
42
+ g=static_cast<int>(255*rgb.g);
43
+ b=static_cast<int>(255*rgb.b);
44
+ */
45
+ r=colToByte(rgb.r);
46
+ g=colToByte(rgb.g);
47
+ b=colToByte(rgb.b);
48
+ if (!(Ok(r)&&Ok(b)&&Ok(g))) {printf("Error : Bad color \n");r=0;g=0;b=0;}
49
+
50
+ }
51
+
52
+ GString *XmlFontColor::convtoX(unsigned int xcol) const{
53
+ GString *xret=new GString();
54
+ char tmp;
55
+ unsigned int k;
56
+ k = (xcol/16);
57
+ if ((k>=0)&&(k<10)) tmp=(char) ('0'+k); else tmp=(char)('a'+k-10);
58
+ xret->append(tmp);
59
+ k = (xcol%16);
60
+ if ((k>=0)&&(k<10)) tmp=(char) ('0'+k); else tmp=(char)('a'+k-10);
61
+ xret->append(tmp);
62
+ return xret;
63
+ }
64
+
65
+ GString *XmlFontColor::toString() const{
66
+ GString *tmp=new GString("#");
67
+ GString *tmpr=convtoX(r);
68
+ GString *tmpg=convtoX(g);
69
+ GString *tmpb=convtoX(b);
70
+ tmp->append(tmpr);
71
+ tmp->append(tmpg);
72
+ tmp->append(tmpb);
73
+ delete tmpr;
74
+ delete tmpg;
75
+ delete tmpb;
76
+ return tmp;
77
+ }
78
+
79
+ XmlFont::XmlFont(GString* ftname,int _size, double _charspace, GfxRGB rgb){
80
+ //if (col) color=XmlFontColor(col);
81
+ //else color=XmlFontColor();
82
+ color=XmlFontColor(rgb);
83
+
84
+ GString *fontname = NULL;
85
+ charspace = _charspace;
86
+
87
+ if( ftname ){
88
+ fontname = new GString(ftname);
89
+ FontName=new GString(ftname);
90
+ }
91
+ else {
92
+ fontname = NULL;
93
+ FontName = NULL;
94
+ }
95
+
96
+ lineSize = -1;
97
+
98
+ size=(_size-1);
99
+ italic = gFalse;
100
+ bold = gFalse;
101
+ oblique = gFalse;
102
+
103
+ if (fontname){
104
+ if (strstr(fontname->lowerCase()->getCString(),"bold")) bold=gTrue;
105
+ if (strstr(fontname->lowerCase()->getCString(),"italic")) italic=gTrue;
106
+ if (strstr(fontname->lowerCase()->getCString(),"oblique")) oblique=gTrue;
107
+ /*||strstr(fontname->lowerCase()->getCString(),"oblique")) italic=gTrue;*/
108
+
109
+ int i=0;
110
+ while (strcmp(ftname->getCString(),fonts[i].Fontname)&&(i<font_num))
111
+ {
112
+ i++;
113
+ }
114
+ pos=i;
115
+ delete fontname;
116
+ }
117
+ if (!DefaultFont) DefaultFont=new GString(fonts[font_num].name);
118
+
119
+ }
120
+
121
+ XmlFont::XmlFont(const XmlFont& x){
122
+ size=x.size;
123
+ lineSize=x.lineSize;
124
+ italic=x.italic;
125
+ oblique=x.oblique;
126
+ bold=x.bold;
127
+ pos=x.pos;
128
+ charspace=x.charspace;
129
+ color=x.color;
130
+ if (x.FontName) FontName=new GString(x.FontName);
131
+ }
132
+
133
+
134
+ XmlFont::~XmlFont(){
135
+ if (FontName) delete FontName;
136
+ }
137
+
138
+ XmlFont& XmlFont::operator=(const XmlFont& x){
139
+ if (this==&x) return *this;
140
+ size=x.size;
141
+ lineSize=x.lineSize;
142
+ italic=x.italic;
143
+ oblique=x.oblique;
144
+ bold=x.bold;
145
+ pos=x.pos;
146
+ color=x.color;
147
+ charspace=x.charspace;
148
+ if (FontName) delete FontName;
149
+ if (x.FontName) FontName=new GString(x.FontName);
150
+ return *this;
151
+ }
152
+
153
+ void XmlFont::clear(){
154
+ if(DefaultFont) delete DefaultFont;
155
+ DefaultFont = NULL;
156
+ }
157
+
158
+
159
+
160
+ /*
161
+ This function is used to compare font uniquily for insertion into
162
+ the list of all encountered fonts
163
+ */
164
+ GBool XmlFont::isEqual(const XmlFont& x) const{
165
+ return ((size==x.size) && (lineSize==x.lineSize) && (charspace==x.charspace) &&
166
+ (pos==x.pos) && (bold==x.bold) && (oblique==x.oblique) && (italic==x.italic) &&
167
+ (color.isEqual(x.getColor())));
168
+ }
169
+
170
+ /*
171
+ This one is used to decide whether two pieces of text can be joined together
172
+ and therefore we don't care about bold/italics properties
173
+ */
174
+ GBool XmlFont::isEqualIgnoreBold(const XmlFont& x) const{
175
+ return ((size==x.size) &&
176
+ (color.isEqual(x.getColor())));
177
+ }
178
+
179
+ GString* XmlFont::getFontName(){
180
+ if (pos!=font_num) return new GString(fonts[pos].name);
181
+ else return new GString(DefaultFont);
182
+ }
183
+
184
+ GString* XmlFont::getFullName(){
185
+ if (FontName)
186
+ return new GString(FontName);
187
+ else return new GString(DefaultFont);
188
+ }
189
+
190
+ void XmlFont::setDefaultFont(GString* defaultFont){
191
+ if (DefaultFont) delete DefaultFont;
192
+ DefaultFont=new GString(defaultFont);
193
+ }
194
+
195
+
196
+ GString* XmlFont::getDefaultFont(){
197
+ return DefaultFont;
198
+ }
199
+
200
+ // this method if plain wrong todo
201
+ GString* XmlFont::HtmlFilter(Unicode* u, int uLen) {
202
+ GString *tmp = new GString();
203
+ UnicodeMap *uMap;
204
+ char buf[8];
205
+ int n;
206
+
207
+ // get the output encoding
208
+ if (!(uMap = globalParams->getTextEncoding())) {
209
+ return tmp;
210
+ }
211
+
212
+ for (int i = 0; i < uLen; ++i) {
213
+ switch (u[i])
214
+ {
215
+ case '"': tmp->append("&quot;"); break;
216
+ case '&': tmp->append("&amp;"); break;
217
+ case '<': tmp->append("&lt;"); break;
218
+ case '>': tmp->append("&gt;"); break;
219
+ default:
220
+ {
221
+ // convert unicode to string
222
+ if ((n = uMap->mapUnicode(u[i], buf, sizeof(buf))) > 0) {
223
+ tmp->append(buf, n);
224
+ }
225
+ }
226
+ }
227
+ }
228
+
229
+ uMap->decRefCnt();
230
+ return tmp;
231
+ }
232
+
233
+ GString* XmlFont::simple(XmlFont* font, Unicode* content, int uLen){
234
+ GString *cont=HtmlFilter (content, uLen);
235
+
236
+ /*if (font.isBold()) {
237
+ cont->insert(0,"<b>",3);
238
+ cont->append("</b>",4);
239
+ }
240
+ if (font.isItalic()) {
241
+ cont->insert(0,"<i>",3);
242
+ cont->append("</i>",4);x.oblique
243
+ } */
244
+
245
+ return cont;
246
+ }
247
+
248
+ XmlFontAccu::XmlFontAccu(){
249
+ accu=new GVector<XmlFont>();
250
+ }
251
+
252
+ XmlFontAccu::~XmlFontAccu(){
253
+ if (accu) delete accu;
254
+ }
255
+
256
+ int XmlFontAccu::AddFont(const XmlFont& font){
257
+ GVector<XmlFont>::iterator i;
258
+ for (i=accu->begin();i!=accu->end();i++)
259
+ {
260
+ if (font.isEqual(*i))
261
+ {
262
+ return (int)(i-(accu->begin()));
263
+ }
264
+ }
265
+
266
+ accu->push_back(font);
267
+ return (accu->size()-1);
268
+ }
269
+
270
+ static GString* EscapeSpecialChars( GString* s)
271
+ {
272
+ GString* tmp = NULL;
273
+ for( int i = 0, j = 0; i < s->getLength(); i++, j++ ){
274
+ const char *replace = NULL;
275
+ switch ( s->getChar(i) ){
276
+ case '"': replace = "\\u0022"; break;
277
+ case '\\': replace = "\\u005C"; break;
278
+ case '<': replace = "\\u003C"; break;
279
+ case '>': replace = "\\u003E"; break;
280
+ case '\'': replace = "\\u0027"; break;
281
+ case '&': replace = "\\u0026"; break;
282
+ case '\n': replace = ""; break;
283
+ case '\r': replace = ""; break;
284
+ case '\t': replace = ""; break;
285
+ default: continue;
286
+ }
287
+ if( replace ){
288
+ if( !tmp ) tmp = new GString( s );
289
+ if( tmp ){
290
+ tmp->del( j, 1 );
291
+ int l = strlen( replace );
292
+ tmp->insert( j, replace, l );
293
+ j += l - 1;
294
+ }
295
+ }
296
+ }
297
+ return tmp ? tmp : s;
298
+ }
299
+
300
+ // get CSS font name for font #i
301
+ GString* XmlFontAccu::getCSStyle(int i, GString* content){
302
+ GString *tmp;
303
+ GString *iStr=GString::fromInt(i);
304
+
305
+ if (!xml) {
306
+ tmp = new GString("<span class=\"ft");
307
+ tmp->append(iStr);
308
+ tmp->append("\">");
309
+ tmp->append(content);
310
+ tmp->append("</span>");
311
+ } else {
312
+ tmp = EscapeSpecialChars(content);
313
+ //tmp->append(content);
314
+ }
315
+
316
+ delete iStr;
317
+ return tmp;
318
+ }
319
+
320
+
321
+ // get CSS font definition for font #i
322
+ GString* XmlFontAccu::CSStyle(int i,GBool textAsJSON){
323
+ GString *tmp=new GString();
324
+ GString *iStr=GString::fromInt(i);
325
+
326
+ GVector<XmlFont>::iterator g=accu->begin();
327
+ g+=i;
328
+ XmlFont font=*g;
329
+ GString *Size=GString::fromInt(font.getSize());
330
+ GString *colorStr=font.getColor().toString();
331
+ GString *fontName=font.getFontName();
332
+ GString *lSize;
333
+ double _charspace = font.getCharSpace();
334
+ char *cspace = new char [20];
335
+ sprintf(cspace,"%0.05f",_charspace);
336
+
337
+ if (xml && !textAsJSON) {
338
+ tmp->append("<fontspec id=\"");
339
+ tmp->append(iStr);
340
+ tmp->append("\" size=\"");
341
+ tmp->append(Size);
342
+ tmp->append("\" family=\"");
343
+ tmp->append(fontName); //font.getFontName());
344
+ tmp->append("\" color=\"");
345
+ tmp->append(colorStr);
346
+ tmp->append("\"/>");
347
+ }else if(textAsJSON){
348
+ tmp->append("{\"fontspec\":\"");
349
+ tmp->append(iStr);
350
+ tmp->append("\",\"size\":\"");
351
+ tmp->append(Size);
352
+ tmp->append("\",\"family\":\"");
353
+ tmp->append(fontName); //font.getFontName());
354
+ tmp->append("\",\"color\":\"");
355
+ tmp->append(colorStr);
356
+ tmp->append("\"}");
357
+ }
358
+
359
+ delete fontName;
360
+ delete colorStr;
361
+ delete iStr;
362
+ delete Size;
363
+ delete cspace;
364
+ return tmp;
365
+ }
366
+
367
+