tesseract_bin 1.0.0

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 (612) hide show
  1. data/.document +5 -0
  2. data/Gemfile +14 -0
  3. data/Gemfile.lock +23 -0
  4. data/LICENSE.txt +20 -0
  5. data/README.rdoc +19 -0
  6. data/Rakefile +46 -0
  7. data/VERSION +1 -0
  8. data/ext/tesseract_bin/extconf.rb +17 -0
  9. data/lib/tesseract_bin.rb +12 -0
  10. data/tesseract_bin.gemspec +660 -0
  11. data/test/helper.rb +18 -0
  12. data/test/test_tesseract_bin.rb +7 -0
  13. data/vendor/tesseract-2.04/AUTHORS +8 -0
  14. data/vendor/tesseract-2.04/COPYING +23 -0
  15. data/vendor/tesseract-2.04/ChangeLog +71 -0
  16. data/vendor/tesseract-2.04/INSTALL +229 -0
  17. data/vendor/tesseract-2.04/Makefile.am +20 -0
  18. data/vendor/tesseract-2.04/Makefile.in +641 -0
  19. data/vendor/tesseract-2.04/NEWS +1 -0
  20. data/vendor/tesseract-2.04/README +138 -0
  21. data/vendor/tesseract-2.04/ReleaseNotes +213 -0
  22. data/vendor/tesseract-2.04/StdAfx.cpp +8 -0
  23. data/vendor/tesseract-2.04/StdAfx.h +24 -0
  24. data/vendor/tesseract-2.04/ccmain/Makefile.am +63 -0
  25. data/vendor/tesseract-2.04/ccmain/Makefile.in +735 -0
  26. data/vendor/tesseract-2.04/ccmain/adaptions.cpp +1082 -0
  27. data/vendor/tesseract-2.04/ccmain/adaptions.h +109 -0
  28. data/vendor/tesseract-2.04/ccmain/applybox.cpp +891 -0
  29. data/vendor/tesseract-2.04/ccmain/applybox.h +73 -0
  30. data/vendor/tesseract-2.04/ccmain/baseapi.cpp +1105 -0
  31. data/vendor/tesseract-2.04/ccmain/baseapi.h +256 -0
  32. data/vendor/tesseract-2.04/ccmain/blobcmp.cpp +76 -0
  33. data/vendor/tesseract-2.04/ccmain/blobcmp.h +29 -0
  34. data/vendor/tesseract-2.04/ccmain/callnet.cpp +93 -0
  35. data/vendor/tesseract-2.04/ccmain/callnet.h +32 -0
  36. data/vendor/tesseract-2.04/ccmain/charcut.cpp +704 -0
  37. data/vendor/tesseract-2.04/ccmain/charcut.h +120 -0
  38. data/vendor/tesseract-2.04/ccmain/charsample.cpp +699 -0
  39. data/vendor/tesseract-2.04/ccmain/control.cpp +1842 -0
  40. data/vendor/tesseract-2.04/ccmain/control.h +198 -0
  41. data/vendor/tesseract-2.04/ccmain/docqual.cpp +1481 -0
  42. data/vendor/tesseract-2.04/ccmain/docqual.h +155 -0
  43. data/vendor/tesseract-2.04/ccmain/expandblob.cpp +82 -0
  44. data/vendor/tesseract-2.04/ccmain/expandblob.h +13 -0
  45. data/vendor/tesseract-2.04/ccmain/fixspace.cpp +989 -0
  46. data/vendor/tesseract-2.04/ccmain/fixspace.h +72 -0
  47. data/vendor/tesseract-2.04/ccmain/fixxht.cpp +825 -0
  48. data/vendor/tesseract-2.04/ccmain/fixxht.h +93 -0
  49. data/vendor/tesseract-2.04/ccmain/imgscale.cpp +154 -0
  50. data/vendor/tesseract-2.04/ccmain/imgscale.h +32 -0
  51. data/vendor/tesseract-2.04/ccmain/matmatch.cpp +391 -0
  52. data/vendor/tesseract-2.04/ccmain/matmatch.h +48 -0
  53. data/vendor/tesseract-2.04/ccmain/output.cpp +1273 -0
  54. data/vendor/tesseract-2.04/ccmain/output.h +116 -0
  55. data/vendor/tesseract-2.04/ccmain/pagewalk.cpp +666 -0
  56. data/vendor/tesseract-2.04/ccmain/pagewalk.h +155 -0
  57. data/vendor/tesseract-2.04/ccmain/paircmp.cpp +107 -0
  58. data/vendor/tesseract-2.04/ccmain/paircmp.h +43 -0
  59. data/vendor/tesseract-2.04/ccmain/pgedit.cpp +1867 -0
  60. data/vendor/tesseract-2.04/ccmain/pgedit.h +181 -0
  61. data/vendor/tesseract-2.04/ccmain/reject.cpp +1775 -0
  62. data/vendor/tesseract-2.04/ccmain/reject.h +181 -0
  63. data/vendor/tesseract-2.04/ccmain/scaleimg.cpp +366 -0
  64. data/vendor/tesseract-2.04/ccmain/scaleimg.h +35 -0
  65. data/vendor/tesseract-2.04/ccmain/tessbox.cpp +375 -0
  66. data/vendor/tesseract-2.04/ccmain/tessbox.h +110 -0
  67. data/vendor/tesseract-2.04/ccmain/tessedit.cpp +278 -0
  68. data/vendor/tesseract-2.04/ccmain/tessedit.h +49 -0
  69. data/vendor/tesseract-2.04/ccmain/tessembedded.cpp +110 -0
  70. data/vendor/tesseract-2.04/ccmain/tessembedded.h +38 -0
  71. data/vendor/tesseract-2.04/ccmain/tesseractfull.cc +37 -0
  72. data/vendor/tesseract-2.04/ccmain/tesseractmain.cpp +387 -0
  73. data/vendor/tesseract-2.04/ccmain/tesseractmain.h +58 -0
  74. data/vendor/tesseract-2.04/ccmain/tessio.h +110 -0
  75. data/vendor/tesseract-2.04/ccmain/tessvars.cpp +38 -0
  76. data/vendor/tesseract-2.04/ccmain/tessvars.h +48 -0
  77. data/vendor/tesseract-2.04/ccmain/tfacep.h +62 -0
  78. data/vendor/tesseract-2.04/ccmain/tfacepp.cpp +443 -0
  79. data/vendor/tesseract-2.04/ccmain/tfacepp.h +85 -0
  80. data/vendor/tesseract-2.04/ccmain/tstruct.cpp +549 -0
  81. data/vendor/tesseract-2.04/ccmain/tstruct.h +108 -0
  82. data/vendor/tesseract-2.04/ccmain/varabled.cpp +346 -0
  83. data/vendor/tesseract-2.04/ccmain/varabled.h +125 -0
  84. data/vendor/tesseract-2.04/ccmain/werdit.cpp +193 -0
  85. data/vendor/tesseract-2.04/ccmain/werdit.h +67 -0
  86. data/vendor/tesseract-2.04/ccstruct/Makefile.am +25 -0
  87. data/vendor/tesseract-2.04/ccstruct/Makefile.in +650 -0
  88. data/vendor/tesseract-2.04/ccstruct/blckerr.h +29 -0
  89. data/vendor/tesseract-2.04/ccstruct/blobbox.cpp +778 -0
  90. data/vendor/tesseract-2.04/ccstruct/blobbox.h +381 -0
  91. data/vendor/tesseract-2.04/ccstruct/blobs.cpp +247 -0
  92. data/vendor/tesseract-2.04/ccstruct/blobs.h +119 -0
  93. data/vendor/tesseract-2.04/ccstruct/blread.cpp +537 -0
  94. data/vendor/tesseract-2.04/ccstruct/blread.h +63 -0
  95. data/vendor/tesseract-2.04/ccstruct/callcpp.cpp +252 -0
  96. data/vendor/tesseract-2.04/ccstruct/coutln.cpp +650 -0
  97. data/vendor/tesseract-2.04/ccstruct/coutln.h +186 -0
  98. data/vendor/tesseract-2.04/ccstruct/crakedge.h +39 -0
  99. data/vendor/tesseract-2.04/ccstruct/genblob.cpp +133 -0
  100. data/vendor/tesseract-2.04/ccstruct/genblob.h +52 -0
  101. data/vendor/tesseract-2.04/ccstruct/hpddef.h +39 -0
  102. data/vendor/tesseract-2.04/ccstruct/hpdsizes.h +8 -0
  103. data/vendor/tesseract-2.04/ccstruct/ipoints.h +479 -0
  104. data/vendor/tesseract-2.04/ccstruct/labls.cpp +188 -0
  105. data/vendor/tesseract-2.04/ccstruct/labls.h +38 -0
  106. data/vendor/tesseract-2.04/ccstruct/linlsq.cpp +249 -0
  107. data/vendor/tesseract-2.04/ccstruct/linlsq.h +102 -0
  108. data/vendor/tesseract-2.04/ccstruct/lmedsq.cpp +453 -0
  109. data/vendor/tesseract-2.04/ccstruct/lmedsq.h +84 -0
  110. data/vendor/tesseract-2.04/ccstruct/mod128.cpp +100 -0
  111. data/vendor/tesseract-2.04/ccstruct/mod128.h +85 -0
  112. data/vendor/tesseract-2.04/ccstruct/normalis.cpp +176 -0
  113. data/vendor/tesseract-2.04/ccstruct/normalis.h +108 -0
  114. data/vendor/tesseract-2.04/ccstruct/ocrblock.cpp +369 -0
  115. data/vendor/tesseract-2.04/ccstruct/ocrblock.h +235 -0
  116. data/vendor/tesseract-2.04/ccstruct/ocrrow.cpp +216 -0
  117. data/vendor/tesseract-2.04/ccstruct/ocrrow.h +133 -0
  118. data/vendor/tesseract-2.04/ccstruct/pageblk.cpp +879 -0
  119. data/vendor/tesseract-2.04/ccstruct/pageblk.h +318 -0
  120. data/vendor/tesseract-2.04/ccstruct/pageres.cpp +330 -0
  121. data/vendor/tesseract-2.04/ccstruct/pageres.h +313 -0
  122. data/vendor/tesseract-2.04/ccstruct/pdblock.cpp +361 -0
  123. data/vendor/tesseract-2.04/ccstruct/pdblock.h +181 -0
  124. data/vendor/tesseract-2.04/ccstruct/pdclass.h +54 -0
  125. data/vendor/tesseract-2.04/ccstruct/points.cpp +102 -0
  126. data/vendor/tesseract-2.04/ccstruct/points.h +299 -0
  127. data/vendor/tesseract-2.04/ccstruct/polyaprx.cpp +588 -0
  128. data/vendor/tesseract-2.04/ccstruct/polyaprx.h +51 -0
  129. data/vendor/tesseract-2.04/ccstruct/polyblk.cpp +398 -0
  130. data/vendor/tesseract-2.04/ccstruct/polyblk.h +122 -0
  131. data/vendor/tesseract-2.04/ccstruct/polyblob.cpp +357 -0
  132. data/vendor/tesseract-2.04/ccstruct/polyblob.h +102 -0
  133. data/vendor/tesseract-2.04/ccstruct/polyvert.cpp +23 -0
  134. data/vendor/tesseract-2.04/ccstruct/polyvert.h +58 -0
  135. data/vendor/tesseract-2.04/ccstruct/poutline.cpp +441 -0
  136. data/vendor/tesseract-2.04/ccstruct/poutline.h +125 -0
  137. data/vendor/tesseract-2.04/ccstruct/quadlsq.cpp +147 -0
  138. data/vendor/tesseract-2.04/ccstruct/quadlsq.h +67 -0
  139. data/vendor/tesseract-2.04/ccstruct/quadratc.cpp +21 -0
  140. data/vendor/tesseract-2.04/ccstruct/quadratc.h +63 -0
  141. data/vendor/tesseract-2.04/ccstruct/quspline.cpp +382 -0
  142. data/vendor/tesseract-2.04/ccstruct/quspline.h +113 -0
  143. data/vendor/tesseract-2.04/ccstruct/ratngs.cpp +372 -0
  144. data/vendor/tesseract-2.04/ccstruct/ratngs.h +198 -0
  145. data/vendor/tesseract-2.04/ccstruct/rect.cpp +229 -0
  146. data/vendor/tesseract-2.04/ccstruct/rect.h +320 -0
  147. data/vendor/tesseract-2.04/ccstruct/rejctmap.cpp +545 -0
  148. data/vendor/tesseract-2.04/ccstruct/rejctmap.h +284 -0
  149. data/vendor/tesseract-2.04/ccstruct/rwpoly.cpp +89 -0
  150. data/vendor/tesseract-2.04/ccstruct/rwpoly.h +45 -0
  151. data/vendor/tesseract-2.04/ccstruct/statistc.cpp +905 -0
  152. data/vendor/tesseract-2.04/ccstruct/statistc.h +135 -0
  153. data/vendor/tesseract-2.04/ccstruct/stepblob.cpp +296 -0
  154. data/vendor/tesseract-2.04/ccstruct/stepblob.h +88 -0
  155. data/vendor/tesseract-2.04/ccstruct/txtregn.cpp +230 -0
  156. data/vendor/tesseract-2.04/ccstruct/txtregn.h +155 -0
  157. data/vendor/tesseract-2.04/ccstruct/vecfuncs.cpp +63 -0
  158. data/vendor/tesseract-2.04/ccstruct/vecfuncs.h +91 -0
  159. data/vendor/tesseract-2.04/ccstruct/werd.cpp +967 -0
  160. data/vendor/tesseract-2.04/ccstruct/werd.h +277 -0
  161. data/vendor/tesseract-2.04/ccutil/Makefile.am +19 -0
  162. data/vendor/tesseract-2.04/ccutil/Makefile.in +626 -0
  163. data/vendor/tesseract-2.04/ccutil/basedir.cpp +118 -0
  164. data/vendor/tesseract-2.04/ccutil/basedir.h +32 -0
  165. data/vendor/tesseract-2.04/ccutil/bits16.cpp +30 -0
  166. data/vendor/tesseract-2.04/ccutil/bits16.h +61 -0
  167. data/vendor/tesseract-2.04/ccutil/boxread.cpp +105 -0
  168. data/vendor/tesseract-2.04/ccutil/boxread.h +44 -0
  169. data/vendor/tesseract-2.04/ccutil/clst.cpp +626 -0
  170. data/vendor/tesseract-2.04/ccutil/clst.h +1085 -0
  171. data/vendor/tesseract-2.04/ccutil/debugwin.cpp +500 -0
  172. data/vendor/tesseract-2.04/ccutil/debugwin.h +103 -0
  173. data/vendor/tesseract-2.04/ccutil/elst.cpp +593 -0
  174. data/vendor/tesseract-2.04/ccutil/elst.h +1125 -0
  175. data/vendor/tesseract-2.04/ccutil/elst2.cpp +606 -0
  176. data/vendor/tesseract-2.04/ccutil/elst2.h +1121 -0
  177. data/vendor/tesseract-2.04/ccutil/errcode.cpp +104 -0
  178. data/vendor/tesseract-2.04/ccutil/errcode.h +104 -0
  179. data/vendor/tesseract-2.04/ccutil/fileerr.h +34 -0
  180. data/vendor/tesseract-2.04/ccutil/globaloc.cpp +115 -0
  181. data/vendor/tesseract-2.04/ccutil/globaloc.h +40 -0
  182. data/vendor/tesseract-2.04/ccutil/hashfn.cpp +57 -0
  183. data/vendor/tesseract-2.04/ccutil/hashfn.h +30 -0
  184. data/vendor/tesseract-2.04/ccutil/host.h +180 -0
  185. data/vendor/tesseract-2.04/ccutil/hosthplb.h +1 -0
  186. data/vendor/tesseract-2.04/ccutil/lsterr.h +43 -0
  187. data/vendor/tesseract-2.04/ccutil/mainblk.cpp +126 -0
  188. data/vendor/tesseract-2.04/ccutil/mainblk.h +39 -0
  189. data/vendor/tesseract-2.04/ccutil/memblk.cpp +1106 -0
  190. data/vendor/tesseract-2.04/ccutil/memblk.h +189 -0
  191. data/vendor/tesseract-2.04/ccutil/memry.cpp +532 -0
  192. data/vendor/tesseract-2.04/ccutil/memry.h +192 -0
  193. data/vendor/tesseract-2.04/ccutil/memryerr.h +38 -0
  194. data/vendor/tesseract-2.04/ccutil/mfcpch.cpp +5 -0
  195. data/vendor/tesseract-2.04/ccutil/mfcpch.h +37 -0
  196. data/vendor/tesseract-2.04/ccutil/ndminx.h +31 -0
  197. data/vendor/tesseract-2.04/ccutil/notdll.h +28 -0
  198. data/vendor/tesseract-2.04/ccutil/nwmain.h +176 -0
  199. data/vendor/tesseract-2.04/ccutil/ocrclass.h +345 -0
  200. data/vendor/tesseract-2.04/ccutil/ocrshell.cpp +772 -0
  201. data/vendor/tesseract-2.04/ccutil/ocrshell.h +191 -0
  202. data/vendor/tesseract-2.04/ccutil/platform.h +18 -0
  203. data/vendor/tesseract-2.04/ccutil/scanutils.cpp +543 -0
  204. data/vendor/tesseract-2.04/ccutil/scanutils.h +55 -0
  205. data/vendor/tesseract-2.04/ccutil/secname.h +9 -0
  206. data/vendor/tesseract-2.04/ccutil/serialis.cpp +117 -0
  207. data/vendor/tesseract-2.04/ccutil/serialis.h +93 -0
  208. data/vendor/tesseract-2.04/ccutil/stderr.h +26 -0
  209. data/vendor/tesseract-2.04/ccutil/strngs.cpp +495 -0
  210. data/vendor/tesseract-2.04/ccutil/strngs.h +138 -0
  211. data/vendor/tesseract-2.04/ccutil/tessclas.h +135 -0
  212. data/vendor/tesseract-2.04/ccutil/tessopt.cpp +61 -0
  213. data/vendor/tesseract-2.04/ccutil/tessopt.h +30 -0
  214. data/vendor/tesseract-2.04/ccutil/tprintf.cpp +122 -0
  215. data/vendor/tesseract-2.04/ccutil/tprintf.h +35 -0
  216. data/vendor/tesseract-2.04/ccutil/unichar.cpp +144 -0
  217. data/vendor/tesseract-2.04/ccutil/unichar.h +84 -0
  218. data/vendor/tesseract-2.04/ccutil/unicharmap.cpp +172 -0
  219. data/vendor/tesseract-2.04/ccutil/unicharmap.h +82 -0
  220. data/vendor/tesseract-2.04/ccutil/unicharset.cpp +307 -0
  221. data/vendor/tesseract-2.04/ccutil/unicharset.h +267 -0
  222. data/vendor/tesseract-2.04/ccutil/varable.cpp +672 -0
  223. data/vendor/tesseract-2.04/ccutil/varable.h +419 -0
  224. data/vendor/tesseract-2.04/classify/Makefile.am +24 -0
  225. data/vendor/tesseract-2.04/classify/Makefile.in +647 -0
  226. data/vendor/tesseract-2.04/classify/adaptive.cpp +535 -0
  227. data/vendor/tesseract-2.04/classify/adaptive.h +199 -0
  228. data/vendor/tesseract-2.04/classify/adaptmatch.cpp +2958 -0
  229. data/vendor/tesseract-2.04/classify/adaptmatch.h +86 -0
  230. data/vendor/tesseract-2.04/classify/baseline.cpp +58 -0
  231. data/vendor/tesseract-2.04/classify/baseline.h +91 -0
  232. data/vendor/tesseract-2.04/classify/blobclass.cpp +123 -0
  233. data/vendor/tesseract-2.04/classify/blobclass.h +49 -0
  234. data/vendor/tesseract-2.04/classify/chartoname.cpp +74 -0
  235. data/vendor/tesseract-2.04/classify/chartoname.h +21 -0
  236. data/vendor/tesseract-2.04/classify/cluster.cpp +2834 -0
  237. data/vendor/tesseract-2.04/classify/cluster.h +158 -0
  238. data/vendor/tesseract-2.04/classify/clusttool.cpp +507 -0
  239. data/vendor/tesseract-2.04/classify/clusttool.h +70 -0
  240. data/vendor/tesseract-2.04/classify/cutoffs.cpp +73 -0
  241. data/vendor/tesseract-2.04/classify/cutoffs.h +49 -0
  242. data/vendor/tesseract-2.04/classify/extern.h +32 -0
  243. data/vendor/tesseract-2.04/classify/extract.cpp +100 -0
  244. data/vendor/tesseract-2.04/classify/extract.h +36 -0
  245. data/vendor/tesseract-2.04/classify/featdefs.cpp +244 -0
  246. data/vendor/tesseract-2.04/classify/featdefs.h +71 -0
  247. data/vendor/tesseract-2.04/classify/flexfx.cpp +87 -0
  248. data/vendor/tesseract-2.04/classify/flexfx.h +34 -0
  249. data/vendor/tesseract-2.04/classify/float2int.cpp +126 -0
  250. data/vendor/tesseract-2.04/classify/float2int.h +65 -0
  251. data/vendor/tesseract-2.04/classify/fpoint.cpp +73 -0
  252. data/vendor/tesseract-2.04/classify/fpoint.h +63 -0
  253. data/vendor/tesseract-2.04/classify/fxdefs.cpp +74 -0
  254. data/vendor/tesseract-2.04/classify/fxdefs.h +93 -0
  255. data/vendor/tesseract-2.04/classify/fxid.h +69 -0
  256. data/vendor/tesseract-2.04/classify/hideedge.cpp +35 -0
  257. data/vendor/tesseract-2.04/classify/hideedge.h +76 -0
  258. data/vendor/tesseract-2.04/classify/intfx.cpp +608 -0
  259. data/vendor/tesseract-2.04/classify/intfx.h +63 -0
  260. data/vendor/tesseract-2.04/classify/intmatcher.cpp +1524 -0
  261. data/vendor/tesseract-2.04/classify/intmatcher.h +199 -0
  262. data/vendor/tesseract-2.04/classify/intproto.cpp +1823 -0
  263. data/vendor/tesseract-2.04/classify/intproto.h +320 -0
  264. data/vendor/tesseract-2.04/classify/kdtree.cpp +884 -0
  265. data/vendor/tesseract-2.04/classify/kdtree.h +118 -0
  266. data/vendor/tesseract-2.04/classify/mf.cpp +106 -0
  267. data/vendor/tesseract-2.04/classify/mf.h +43 -0
  268. data/vendor/tesseract-2.04/classify/mfdefs.cpp +58 -0
  269. data/vendor/tesseract-2.04/classify/mfdefs.h +60 -0
  270. data/vendor/tesseract-2.04/classify/mfoutline.cpp +1087 -0
  271. data/vendor/tesseract-2.04/classify/mfoutline.h +277 -0
  272. data/vendor/tesseract-2.04/classify/mfx.cpp +436 -0
  273. data/vendor/tesseract-2.04/classify/mfx.h +52 -0
  274. data/vendor/tesseract-2.04/classify/normfeat.cpp +132 -0
  275. data/vendor/tesseract-2.04/classify/normfeat.h +63 -0
  276. data/vendor/tesseract-2.04/classify/normmatch.cpp +305 -0
  277. data/vendor/tesseract-2.04/classify/normmatch.h +38 -0
  278. data/vendor/tesseract-2.04/classify/ocrfeatures.cpp +310 -0
  279. data/vendor/tesseract-2.04/classify/ocrfeatures.h +148 -0
  280. data/vendor/tesseract-2.04/classify/outfeat.cpp +262 -0
  281. data/vendor/tesseract-2.04/classify/outfeat.h +76 -0
  282. data/vendor/tesseract-2.04/classify/picofeat.cpp +297 -0
  283. data/vendor/tesseract-2.04/classify/picofeat.h +65 -0
  284. data/vendor/tesseract-2.04/classify/protos.cpp +472 -0
  285. data/vendor/tesseract-2.04/classify/protos.h +258 -0
  286. data/vendor/tesseract-2.04/classify/sigmenu.cpp +225 -0
  287. data/vendor/tesseract-2.04/classify/sigmenu.h +39 -0
  288. data/vendor/tesseract-2.04/classify/speckle.cpp +127 -0
  289. data/vendor/tesseract-2.04/classify/speckle.h +69 -0
  290. data/vendor/tesseract-2.04/classify/xform2d.cpp +120 -0
  291. data/vendor/tesseract-2.04/classify/xform2d.h +60 -0
  292. data/vendor/tesseract-2.04/config/config.guess +1466 -0
  293. data/vendor/tesseract-2.04/config/config.h.in +188 -0
  294. data/vendor/tesseract-2.04/config/config.sub +1579 -0
  295. data/vendor/tesseract-2.04/config/depcomp +530 -0
  296. data/vendor/tesseract-2.04/config/install-sh +269 -0
  297. data/vendor/tesseract-2.04/config/missing +198 -0
  298. data/vendor/tesseract-2.04/config/mkinstalldirs +40 -0
  299. data/vendor/tesseract-2.04/config/stamp-h.in +0 -0
  300. data/vendor/tesseract-2.04/configure +10424 -0
  301. data/vendor/tesseract-2.04/cutil/Makefile.am +14 -0
  302. data/vendor/tesseract-2.04/cutil/Makefile.in +612 -0
  303. data/vendor/tesseract-2.04/cutil/bitvec.cpp +115 -0
  304. data/vendor/tesseract-2.04/cutil/bitvec.h +100 -0
  305. data/vendor/tesseract-2.04/cutil/callcpp.h +190 -0
  306. data/vendor/tesseract-2.04/cutil/const.h +108 -0
  307. data/vendor/tesseract-2.04/cutil/cutil.cpp +92 -0
  308. data/vendor/tesseract-2.04/cutil/cutil.h +159 -0
  309. data/vendor/tesseract-2.04/cutil/danerror.cpp +144 -0
  310. data/vendor/tesseract-2.04/cutil/danerror.h +41 -0
  311. data/vendor/tesseract-2.04/cutil/debug.cpp +97 -0
  312. data/vendor/tesseract-2.04/cutil/debug.h +348 -0
  313. data/vendor/tesseract-2.04/cutil/efio.cpp +62 -0
  314. data/vendor/tesseract-2.04/cutil/efio.h +32 -0
  315. data/vendor/tesseract-2.04/cutil/emalloc.cpp +91 -0
  316. data/vendor/tesseract-2.04/cutil/emalloc.h +44 -0
  317. data/vendor/tesseract-2.04/cutil/freelist.cpp +75 -0
  318. data/vendor/tesseract-2.04/cutil/freelist.h +45 -0
  319. data/vendor/tesseract-2.04/cutil/funcdefs.h +35 -0
  320. data/vendor/tesseract-2.04/cutil/general.h +33 -0
  321. data/vendor/tesseract-2.04/cutil/globals.cpp +69 -0
  322. data/vendor/tesseract-2.04/cutil/globals.h +70 -0
  323. data/vendor/tesseract-2.04/cutil/listio.cpp +68 -0
  324. data/vendor/tesseract-2.04/cutil/listio.h +43 -0
  325. data/vendor/tesseract-2.04/cutil/minmax.h +40 -0
  326. data/vendor/tesseract-2.04/cutil/oldheap.cpp +337 -0
  327. data/vendor/tesseract-2.04/cutil/oldheap.h +126 -0
  328. data/vendor/tesseract-2.04/cutil/oldlist.cpp +393 -0
  329. data/vendor/tesseract-2.04/cutil/oldlist.h +350 -0
  330. data/vendor/tesseract-2.04/cutil/structures.cpp +66 -0
  331. data/vendor/tesseract-2.04/cutil/structures.h +112 -0
  332. data/vendor/tesseract-2.04/cutil/tessarray.cpp +115 -0
  333. data/vendor/tesseract-2.04/cutil/tessarray.h +166 -0
  334. data/vendor/tesseract-2.04/cutil/tordvars.cpp +95 -0
  335. data/vendor/tesseract-2.04/cutil/tordvars.h +61 -0
  336. data/vendor/tesseract-2.04/cutil/variables.cpp +317 -0
  337. data/vendor/tesseract-2.04/cutil/variables.h +170 -0
  338. data/vendor/tesseract-2.04/dict/Makefile.am +13 -0
  339. data/vendor/tesseract-2.04/dict/Makefile.in +609 -0
  340. data/vendor/tesseract-2.04/dict/choicearr.h +96 -0
  341. data/vendor/tesseract-2.04/dict/choices.cpp +210 -0
  342. data/vendor/tesseract-2.04/dict/choices.h +241 -0
  343. data/vendor/tesseract-2.04/dict/context.cpp +270 -0
  344. data/vendor/tesseract-2.04/dict/context.h +82 -0
  345. data/vendor/tesseract-2.04/dict/dawg.cpp +363 -0
  346. data/vendor/tesseract-2.04/dict/dawg.h +394 -0
  347. data/vendor/tesseract-2.04/dict/hyphen.cpp +84 -0
  348. data/vendor/tesseract-2.04/dict/hyphen.h +125 -0
  349. data/vendor/tesseract-2.04/dict/lookdawg.cpp +228 -0
  350. data/vendor/tesseract-2.04/dict/lookdawg.h +76 -0
  351. data/vendor/tesseract-2.04/dict/makedawg.cpp +449 -0
  352. data/vendor/tesseract-2.04/dict/makedawg.h +83 -0
  353. data/vendor/tesseract-2.04/dict/matchdefs.h +145 -0
  354. data/vendor/tesseract-2.04/dict/permdawg.cpp +415 -0
  355. data/vendor/tesseract-2.04/dict/permdawg.h +98 -0
  356. data/vendor/tesseract-2.04/dict/permngram.cpp +358 -0
  357. data/vendor/tesseract-2.04/dict/permngram.h +33 -0
  358. data/vendor/tesseract-2.04/dict/permnum.cpp +522 -0
  359. data/vendor/tesseract-2.04/dict/permnum.h +83 -0
  360. data/vendor/tesseract-2.04/dict/permute.cpp +1704 -0
  361. data/vendor/tesseract-2.04/dict/permute.h +93 -0
  362. data/vendor/tesseract-2.04/dict/reduce.cpp +424 -0
  363. data/vendor/tesseract-2.04/dict/reduce.h +112 -0
  364. data/vendor/tesseract-2.04/dict/states.cpp +382 -0
  365. data/vendor/tesseract-2.04/dict/states.h +111 -0
  366. data/vendor/tesseract-2.04/dict/stopper.cpp +1458 -0
  367. data/vendor/tesseract-2.04/dict/stopper.h +103 -0
  368. data/vendor/tesseract-2.04/dict/trie.cpp +683 -0
  369. data/vendor/tesseract-2.04/dict/trie.h +190 -0
  370. data/vendor/tesseract-2.04/dlltest/Makefile.am +2 -0
  371. data/vendor/tesseract-2.04/dlltest/Makefile.in +388 -0
  372. data/vendor/tesseract-2.04/dlltest/dlltest.cpp +163 -0
  373. data/vendor/tesseract-2.04/dlltest/dlltest.dsp +186 -0
  374. data/vendor/tesseract-2.04/dlltest/dlltest.vcproj +637 -0
  375. data/vendor/tesseract-2.04/eurotext.tif +0 -0
  376. data/vendor/tesseract-2.04/image/Makefile.am +10 -0
  377. data/vendor/tesseract-2.04/image/Makefile.in +596 -0
  378. data/vendor/tesseract-2.04/image/bitstrm.cpp +157 -0
  379. data/vendor/tesseract-2.04/image/bitstrm.h +73 -0
  380. data/vendor/tesseract-2.04/image/img.h +336 -0
  381. data/vendor/tesseract-2.04/image/imgbmp.cpp +223 -0
  382. data/vendor/tesseract-2.04/image/imgbmp.h +50 -0
  383. data/vendor/tesseract-2.04/image/imgerrs.h +35 -0
  384. data/vendor/tesseract-2.04/image/imgio.cpp +321 -0
  385. data/vendor/tesseract-2.04/image/imgio.h +22 -0
  386. data/vendor/tesseract-2.04/image/imgs.cpp +1764 -0
  387. data/vendor/tesseract-2.04/image/imgs.h +102 -0
  388. data/vendor/tesseract-2.04/image/imgtiff.cpp +723 -0
  389. data/vendor/tesseract-2.04/image/imgtiff.h +89 -0
  390. data/vendor/tesseract-2.04/image/imgunpk.h +1377 -0
  391. data/vendor/tesseract-2.04/image/svshowim.cpp +40 -0
  392. data/vendor/tesseract-2.04/image/svshowim.h +25 -0
  393. data/vendor/tesseract-2.04/java/Makefile.am +4 -0
  394. data/vendor/tesseract-2.04/java/Makefile.in +473 -0
  395. data/vendor/tesseract-2.04/java/com/Makefile.am +1 -0
  396. data/vendor/tesseract-2.04/java/com/Makefile.in +470 -0
  397. data/vendor/tesseract-2.04/java/com/google/Makefile.am +1 -0
  398. data/vendor/tesseract-2.04/java/com/google/Makefile.in +470 -0
  399. data/vendor/tesseract-2.04/java/com/google/scrollview/Makefile.am +4 -0
  400. data/vendor/tesseract-2.04/java/com/google/scrollview/Makefile.in +473 -0
  401. data/vendor/tesseract-2.04/java/com/google/scrollview/ScrollView.java +421 -0
  402. data/vendor/tesseract-2.04/java/com/google/scrollview/events/Makefile.am +5 -0
  403. data/vendor/tesseract-2.04/java/com/google/scrollview/events/Makefile.in +474 -0
  404. data/vendor/tesseract-2.04/java/com/google/scrollview/events/SVEvent.java +87 -0
  405. data/vendor/tesseract-2.04/java/com/google/scrollview/events/SVEventHandler.java +296 -0
  406. data/vendor/tesseract-2.04/java/com/google/scrollview/events/SVEventType.java +31 -0
  407. data/vendor/tesseract-2.04/java/com/google/scrollview/ui/Makefile.am +7 -0
  408. data/vendor/tesseract-2.04/java/com/google/scrollview/ui/Makefile.in +476 -0
  409. data/vendor/tesseract-2.04/java/com/google/scrollview/ui/SVAbstractMenuItem.java +58 -0
  410. data/vendor/tesseract-2.04/java/com/google/scrollview/ui/SVCheckboxMenuItem.java +60 -0
  411. data/vendor/tesseract-2.04/java/com/google/scrollview/ui/SVEmptyMenuItem.java +48 -0
  412. data/vendor/tesseract-2.04/java/com/google/scrollview/ui/SVImageHandler.java +228 -0
  413. data/vendor/tesseract-2.04/java/com/google/scrollview/ui/SVMenuBar.java +130 -0
  414. data/vendor/tesseract-2.04/java/com/google/scrollview/ui/SVMenuItem.java +61 -0
  415. data/vendor/tesseract-2.04/java/com/google/scrollview/ui/SVPopupMenu.java +142 -0
  416. data/vendor/tesseract-2.04/java/com/google/scrollview/ui/SVSubMenuItem.java +41 -0
  417. data/vendor/tesseract-2.04/java/com/google/scrollview/ui/SVWindow.java +643 -0
  418. data/vendor/tesseract-2.04/java/makefile +55 -0
  419. data/vendor/tesseract-2.04/pageseg/Makefile.am +13 -0
  420. data/vendor/tesseract-2.04/pageseg/Makefile.in +596 -0
  421. data/vendor/tesseract-2.04/pageseg/leptonica_pageseg.cpp +363 -0
  422. data/vendor/tesseract-2.04/pageseg/leptonica_pageseg.h +90 -0
  423. data/vendor/tesseract-2.04/pageseg/leptonica_pageseg_interface.cpp +82 -0
  424. data/vendor/tesseract-2.04/pageseg/leptonica_pageseg_interface.h +30 -0
  425. data/vendor/tesseract-2.04/pageseg/pageseg.cpp +170 -0
  426. data/vendor/tesseract-2.04/pageseg/pageseg.h +29 -0
  427. data/vendor/tesseract-2.04/phototest.tif +0 -0
  428. data/vendor/tesseract-2.04/tessdata/Makefile.am +31 -0
  429. data/vendor/tesseract-2.04/tessdata/Makefile.in +529 -0
  430. data/vendor/tesseract-2.04/tessdata/configs/Makefile.am +3 -0
  431. data/vendor/tesseract-2.04/tessdata/configs/Makefile.in +344 -0
  432. data/vendor/tesseract-2.04/tessdata/configs/api_config +1 -0
  433. data/vendor/tesseract-2.04/tessdata/configs/box.train +19 -0
  434. data/vendor/tesseract-2.04/tessdata/configs/box.train.stderr +18 -0
  435. data/vendor/tesseract-2.04/tessdata/configs/inter +4 -0
  436. data/vendor/tesseract-2.04/tessdata/configs/kannada +4 -0
  437. data/vendor/tesseract-2.04/tessdata/configs/makebox +1 -0
  438. data/vendor/tesseract-2.04/tessdata/configs/unlv +3 -0
  439. data/vendor/tesseract-2.04/tessdata/confsets +3 -0
  440. data/vendor/tesseract-2.04/tessdata/eng.DangAmbigs +39 -0
  441. data/vendor/tesseract-2.04/tessdata/eng.freq-dawg +0 -0
  442. data/vendor/tesseract-2.04/tessdata/eng.inttemp +0 -0
  443. data/vendor/tesseract-2.04/tessdata/eng.normproto +1247 -0
  444. data/vendor/tesseract-2.04/tessdata/eng.pffmtable +111 -0
  445. data/vendor/tesseract-2.04/tessdata/eng.unicharset +113 -0
  446. data/vendor/tesseract-2.04/tessdata/eng.user-words +921 -0
  447. data/vendor/tesseract-2.04/tessdata/eng.word-dawg +0 -0
  448. data/vendor/tesseract-2.04/tessdata/makedummies +8 -0
  449. data/vendor/tesseract-2.04/tessdata/tessconfigs/Makefile.am +3 -0
  450. data/vendor/tesseract-2.04/tessdata/tessconfigs/Makefile.in +344 -0
  451. data/vendor/tesseract-2.04/tessdata/tessconfigs/batch +2 -0
  452. data/vendor/tesseract-2.04/tessdata/tessconfigs/batch.nochop +2 -0
  453. data/vendor/tesseract-2.04/tessdata/tessconfigs/matdemo +7 -0
  454. data/vendor/tesseract-2.04/tessdata/tessconfigs/msdemo +13 -0
  455. data/vendor/tesseract-2.04/tessdata/tessconfigs/nobatch +2 -0
  456. data/vendor/tesseract-2.04/tessdata/tessconfigs/segdemo +9 -0
  457. data/vendor/tesseract-2.04/tessdll.cpp +351 -0
  458. data/vendor/tesseract-2.04/tessdll.dsp +2050 -0
  459. data/vendor/tesseract-2.04/tessdll.h +143 -0
  460. data/vendor/tesseract-2.04/tessdll.vcproj +5495 -0
  461. data/vendor/tesseract-2.04/tesseract.dsp +2124 -0
  462. data/vendor/tesseract-2.04/tesseract.dsw +116 -0
  463. data/vendor/tesseract-2.04/tesseract.sln +59 -0
  464. data/vendor/tesseract-2.04/tesseract.spec +188 -0
  465. data/vendor/tesseract-2.04/tesseract.vcproj +5859 -0
  466. data/vendor/tesseract-2.04/testing/Makefile.am +2 -0
  467. data/vendor/tesseract-2.04/testing/Makefile.in +312 -0
  468. data/vendor/tesseract-2.04/testing/README +43 -0
  469. data/vendor/tesseract-2.04/testing/counttestset.sh +61 -0
  470. data/vendor/tesseract-2.04/testing/reorgdata.sh +44 -0
  471. data/vendor/tesseract-2.04/testing/reports/1995.bus.3B.sum +1 -0
  472. data/vendor/tesseract-2.04/testing/reports/1995.doe3.3B.sum +1 -0
  473. data/vendor/tesseract-2.04/testing/reports/1995.mag.3B.sum +1 -0
  474. data/vendor/tesseract-2.04/testing/reports/1995.news.3B.sum +1 -0
  475. data/vendor/tesseract-2.04/testing/reports/2.03.summary +9 -0
  476. data/vendor/tesseract-2.04/testing/reports/2.04.summary +9 -0
  477. data/vendor/tesseract-2.04/testing/runalltests.sh +110 -0
  478. data/vendor/tesseract-2.04/testing/runtestset.sh +61 -0
  479. data/vendor/tesseract-2.04/textord/Makefile.am +20 -0
  480. data/vendor/tesseract-2.04/textord/Makefile.in +624 -0
  481. data/vendor/tesseract-2.04/textord/blkocc.cpp +809 -0
  482. data/vendor/tesseract-2.04/textord/blkocc.h +327 -0
  483. data/vendor/tesseract-2.04/textord/blobcmpl.h +31 -0
  484. data/vendor/tesseract-2.04/textord/drawedg.cpp +77 -0
  485. data/vendor/tesseract-2.04/textord/drawedg.h +34 -0
  486. data/vendor/tesseract-2.04/textord/drawtord.cpp +469 -0
  487. data/vendor/tesseract-2.04/textord/drawtord.h +107 -0
  488. data/vendor/tesseract-2.04/textord/edgblob.cpp +412 -0
  489. data/vendor/tesseract-2.04/textord/edgblob.h +100 -0
  490. data/vendor/tesseract-2.04/textord/edgloop.cpp +211 -0
  491. data/vendor/tesseract-2.04/textord/edgloop.h +66 -0
  492. data/vendor/tesseract-2.04/textord/fpchop.cpp +1641 -0
  493. data/vendor/tesseract-2.04/textord/fpchop.h +238 -0
  494. data/vendor/tesseract-2.04/textord/gap_map.cpp +166 -0
  495. data/vendor/tesseract-2.04/textord/gap_map.h +40 -0
  496. data/vendor/tesseract-2.04/textord/makerow.cpp +2628 -0
  497. data/vendor/tesseract-2.04/textord/makerow.h +295 -0
  498. data/vendor/tesseract-2.04/textord/oldbasel.cpp +1761 -0
  499. data/vendor/tesseract-2.04/textord/oldbasel.h +195 -0
  500. data/vendor/tesseract-2.04/textord/pithsync.cpp +696 -0
  501. data/vendor/tesseract-2.04/textord/pithsync.h +134 -0
  502. data/vendor/tesseract-2.04/textord/pitsync1.cpp +425 -0
  503. data/vendor/tesseract-2.04/textord/pitsync1.h +135 -0
  504. data/vendor/tesseract-2.04/textord/scanedg.cpp +452 -0
  505. data/vendor/tesseract-2.04/textord/scanedg.h +74 -0
  506. data/vendor/tesseract-2.04/textord/sortflts.cpp +80 -0
  507. data/vendor/tesseract-2.04/textord/sortflts.h +64 -0
  508. data/vendor/tesseract-2.04/textord/tessout.h +76 -0
  509. data/vendor/tesseract-2.04/textord/topitch.cpp +2019 -0
  510. data/vendor/tesseract-2.04/textord/topitch.h +195 -0
  511. data/vendor/tesseract-2.04/textord/tordmain.cpp +907 -0
  512. data/vendor/tesseract-2.04/textord/tordmain.h +132 -0
  513. data/vendor/tesseract-2.04/textord/tospace.cpp +1939 -0
  514. data/vendor/tesseract-2.04/textord/tospace.h +193 -0
  515. data/vendor/tesseract-2.04/textord/tovars.cpp +87 -0
  516. data/vendor/tesseract-2.04/textord/tovars.h +94 -0
  517. data/vendor/tesseract-2.04/textord/underlin.cpp +312 -0
  518. data/vendor/tesseract-2.04/textord/underlin.h +53 -0
  519. data/vendor/tesseract-2.04/textord/wordseg.cpp +620 -0
  520. data/vendor/tesseract-2.04/textord/wordseg.h +70 -0
  521. data/vendor/tesseract-2.04/training/Makefile.am +54 -0
  522. data/vendor/tesseract-2.04/training/Makefile.in +720 -0
  523. data/vendor/tesseract-2.04/training/cnTraining.cpp +855 -0
  524. data/vendor/tesseract-2.04/training/cntraining.dsp +243 -0
  525. data/vendor/tesseract-2.04/training/cntraining.vcproj +950 -0
  526. data/vendor/tesseract-2.04/training/mergenf.cpp +451 -0
  527. data/vendor/tesseract-2.04/training/mergenf.h +106 -0
  528. data/vendor/tesseract-2.04/training/mfTraining.cpp +1341 -0
  529. data/vendor/tesseract-2.04/training/mftraining.dsp +285 -0
  530. data/vendor/tesseract-2.04/training/mftraining.vcproj +1055 -0
  531. data/vendor/tesseract-2.04/training/name2char.cpp +166 -0
  532. data/vendor/tesseract-2.04/training/name2char.h +38 -0
  533. data/vendor/tesseract-2.04/training/training.cpp +190 -0
  534. data/vendor/tesseract-2.04/training/training.h +130 -0
  535. data/vendor/tesseract-2.04/training/unicharset_extractor.cpp +140 -0
  536. data/vendor/tesseract-2.04/training/unicharset_extractor.dsp +335 -0
  537. data/vendor/tesseract-2.04/training/unicharset_extractor.vcproj +769 -0
  538. data/vendor/tesseract-2.04/training/wordlist2dawg.cpp +69 -0
  539. data/vendor/tesseract-2.04/training/wordlist2dawg.dsp +319 -0
  540. data/vendor/tesseract-2.04/training/wordlist2dawg.vcproj +1113 -0
  541. data/vendor/tesseract-2.04/viewer/Makefile.am +9 -0
  542. data/vendor/tesseract-2.04/viewer/Makefile.in +591 -0
  543. data/vendor/tesseract-2.04/viewer/scrollview.cpp +825 -0
  544. data/vendor/tesseract-2.04/viewer/scrollview.h +414 -0
  545. data/vendor/tesseract-2.04/viewer/svmnode.cpp +140 -0
  546. data/vendor/tesseract-2.04/viewer/svmnode.h +94 -0
  547. data/vendor/tesseract-2.04/viewer/svpaint.cpp +220 -0
  548. data/vendor/tesseract-2.04/viewer/svutil.cpp +347 -0
  549. data/vendor/tesseract-2.04/viewer/svutil.h +138 -0
  550. data/vendor/tesseract-2.04/wordrec/Makefile.am +23 -0
  551. data/vendor/tesseract-2.04/wordrec/Makefile.in +641 -0
  552. data/vendor/tesseract-2.04/wordrec/associate.cpp +62 -0
  553. data/vendor/tesseract-2.04/wordrec/associate.h +93 -0
  554. data/vendor/tesseract-2.04/wordrec/badwords.cpp +106 -0
  555. data/vendor/tesseract-2.04/wordrec/badwords.h +51 -0
  556. data/vendor/tesseract-2.04/wordrec/bestfirst.cpp +526 -0
  557. data/vendor/tesseract-2.04/wordrec/bestfirst.h +203 -0
  558. data/vendor/tesseract-2.04/wordrec/charsample.h +208 -0
  559. data/vendor/tesseract-2.04/wordrec/chop.cpp +458 -0
  560. data/vendor/tesseract-2.04/wordrec/chop.h +153 -0
  561. data/vendor/tesseract-2.04/wordrec/chopper.cpp +750 -0
  562. data/vendor/tesseract-2.04/wordrec/chopper.h +104 -0
  563. data/vendor/tesseract-2.04/wordrec/closed.cpp +136 -0
  564. data/vendor/tesseract-2.04/wordrec/closed.h +65 -0
  565. data/vendor/tesseract-2.04/wordrec/djmenus.cpp +118 -0
  566. data/vendor/tesseract-2.04/wordrec/djmenus.h +33 -0
  567. data/vendor/tesseract-2.04/wordrec/drawfx.cpp +92 -0
  568. data/vendor/tesseract-2.04/wordrec/drawfx.h +33 -0
  569. data/vendor/tesseract-2.04/wordrec/findseam.cpp +566 -0
  570. data/vendor/tesseract-2.04/wordrec/findseam.h +69 -0
  571. data/vendor/tesseract-2.04/wordrec/gradechop.cpp +226 -0
  572. data/vendor/tesseract-2.04/wordrec/gradechop.h +91 -0
  573. data/vendor/tesseract-2.04/wordrec/heuristic.cpp +194 -0
  574. data/vendor/tesseract-2.04/wordrec/heuristic.h +120 -0
  575. data/vendor/tesseract-2.04/wordrec/makechop.cpp +281 -0
  576. data/vendor/tesseract-2.04/wordrec/makechop.h +69 -0
  577. data/vendor/tesseract-2.04/wordrec/matchtab.cpp +191 -0
  578. data/vendor/tesseract-2.04/wordrec/matchtab.h +45 -0
  579. data/vendor/tesseract-2.04/wordrec/matrix.cpp +118 -0
  580. data/vendor/tesseract-2.04/wordrec/matrix.h +104 -0
  581. data/vendor/tesseract-2.04/wordrec/measure.h +135 -0
  582. data/vendor/tesseract-2.04/wordrec/metrics.cpp +363 -0
  583. data/vendor/tesseract-2.04/wordrec/metrics.h +130 -0
  584. data/vendor/tesseract-2.04/wordrec/mfvars.cpp +51 -0
  585. data/vendor/tesseract-2.04/wordrec/mfvars.h +27 -0
  586. data/vendor/tesseract-2.04/wordrec/msmenus.cpp +110 -0
  587. data/vendor/tesseract-2.04/wordrec/msmenus.h +45 -0
  588. data/vendor/tesseract-2.04/wordrec/olutil.cpp +153 -0
  589. data/vendor/tesseract-2.04/wordrec/olutil.h +128 -0
  590. data/vendor/tesseract-2.04/wordrec/outlines.cpp +172 -0
  591. data/vendor/tesseract-2.04/wordrec/outlines.h +148 -0
  592. data/vendor/tesseract-2.04/wordrec/pieces.cpp +410 -0
  593. data/vendor/tesseract-2.04/wordrec/pieces.h +154 -0
  594. data/vendor/tesseract-2.04/wordrec/plotedges.cpp +134 -0
  595. data/vendor/tesseract-2.04/wordrec/plotedges.h +71 -0
  596. data/vendor/tesseract-2.04/wordrec/plotseg.cpp +116 -0
  597. data/vendor/tesseract-2.04/wordrec/plotseg.h +73 -0
  598. data/vendor/tesseract-2.04/wordrec/render.cpp +152 -0
  599. data/vendor/tesseract-2.04/wordrec/render.h +58 -0
  600. data/vendor/tesseract-2.04/wordrec/seam.cpp +482 -0
  601. data/vendor/tesseract-2.04/wordrec/seam.h +136 -0
  602. data/vendor/tesseract-2.04/wordrec/split.cpp +182 -0
  603. data/vendor/tesseract-2.04/wordrec/split.h +115 -0
  604. data/vendor/tesseract-2.04/wordrec/tally.cpp +68 -0
  605. data/vendor/tesseract-2.04/wordrec/tally.h +94 -0
  606. data/vendor/tesseract-2.04/wordrec/tessinit.cpp +108 -0
  607. data/vendor/tesseract-2.04/wordrec/tessinit.h +46 -0
  608. data/vendor/tesseract-2.04/wordrec/tface.cpp +272 -0
  609. data/vendor/tesseract-2.04/wordrec/tface.h +35 -0
  610. data/vendor/tesseract-2.04/wordrec/wordclass.cpp +284 -0
  611. data/vendor/tesseract-2.04/wordrec/wordclass.h +64 -0
  612. metadata +708 -0
@@ -0,0 +1,278 @@
1
+ /**********************************************************************
2
+ * File: tessedit.cpp (Formerly tessedit.c)
3
+ * Description: Main program for merge of tess and editor.
4
+ * Author: Ray Smith
5
+ * Created: Tue Jan 07 15:21:46 GMT 1992
6
+ *
7
+ * (C) Copyright 1992, Hewlett-Packard Ltd.
8
+ ** Licensed under the Apache License, Version 2.0 (the "License");
9
+ ** you may not use this file except in compliance with the License.
10
+ ** You may obtain a copy of the License at
11
+ ** http://www.apache.org/licenses/LICENSE-2.0
12
+ ** Unless required by applicable law or agreed to in writing, software
13
+ ** distributed under the License is distributed on an "AS IS" BASIS,
14
+ ** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
15
+ ** See the License for the specific language governing permissions and
16
+ ** limitations under the License.
17
+ *
18
+ **********************************************************************/
19
+
20
+ #include "mfcpch.h"
21
+ //#include <osfcn.h>
22
+ //#include <signal.h>
23
+ //#include <time.h>
24
+ //#include <unistd.h>
25
+ #include "tfacep.h" //must be before main.h
26
+ //#include "fileerr.h"
27
+ #include "stderr.h"
28
+ #include "basedir.h"
29
+ #include "tessvars.h"
30
+ //#include "debgwin.h"
31
+ //#include "epapdest.h"
32
+ #include "control.h"
33
+ #include "imgs.h"
34
+ #include "reject.h"
35
+ #include "pageres.h"
36
+ //#include "gpapdest.h"
37
+ #include "mainblk.h"
38
+ #include "nwmain.h"
39
+ #include "pgedit.h"
40
+ #include "ocrshell.h"
41
+ #include "tprintf.h"
42
+ //#include "ipeerr.h"
43
+ //#include "restart.h"
44
+ #include "tessedit.h"
45
+ //#include "fontfind.h"
46
+ #include "permute.h"
47
+ #include "permdawg.h"
48
+ #include "permnum.h"
49
+ #include "stopper.h"
50
+ #include "adaptmatch.h"
51
+ #include "intmatcher.h"
52
+ #include "chop.h"
53
+ #include "efio.h"
54
+ #include "danerror.h"
55
+ #include "globals.h"
56
+
57
+ /*
58
+ ** Include automatically generated configuration file if running autoconf
59
+ */
60
+ #ifdef HAVE_CONFIG_H
61
+ #include "config_auto.h"
62
+ #endif
63
+ // Includes libtiff if HAVE_LIBTIFF is defined
64
+ #ifdef HAVE_LIBTIFF
65
+ #include "tiffio.h"
66
+
67
+ #endif
68
+
69
+ //extern "C" {
70
+ #include "callnet.h" //phils nn stuff
71
+ //}
72
+ #include "notdll.h" //phils nn stuff
73
+
74
+ #define VARDIR "configs/" /*variables files */
75
+ //config under api
76
+ #define API_CONFIG "configs/api_config"
77
+ #define EXTERN
78
+
79
+ EXTERN STRING_VAR (tessedit_char_blacklist, "",
80
+ "Blacklist of chars not to recognize");
81
+ EXTERN STRING_VAR (tessedit_char_whitelist, "",
82
+ "Whitelist of chars to recognize");
83
+ EXTERN BOOL_EVAR (tessedit_write_vars, FALSE, "Write all vars to file");
84
+ EXTERN BOOL_VAR (tessedit_tweaking_tess_vars, FALSE,
85
+ "Fiddle tess config values");
86
+
87
+ EXTERN double_VAR (tweak_garbage, 1.5, "Tess VAR");
88
+ EXTERN double_VAR (tweak_ok_word, 1.25, "Tess VAR");
89
+ EXTERN double_VAR (tweak_good_word, 1.1, "Tess VAR");
90
+ EXTERN double_VAR (tweak_freq_word, 1.0, "Tess VAR");
91
+ EXTERN double_VAR (tweak_ok_number, 1.4, "Tess VAR");
92
+ EXTERN double_VAR (tweak_good_number, 1.1, "Tess VAR");
93
+ EXTERN double_VAR (tweak_non_word, 1.25, "Tess VAR");
94
+ EXTERN double_VAR (tweak_CertaintyPerChar, -0.5, "Tess VAR");
95
+ EXTERN double_VAR (tweak_NonDictCertainty, -2.5, "Tess VAR");
96
+ EXTERN double_VAR (tweak_RejectCertaintyOffset, 1.0, "Tess VAR");
97
+ EXTERN double_VAR (tweak_GoodAdaptiveMatch, 0.125, "Tess VAR");
98
+ EXTERN double_VAR (tweak_GreatAdaptiveMatch, 0.10, "Tess VAR");
99
+ EXTERN INT_VAR (tweak_ReliableConfigThreshold, 2, "Tess VAR");
100
+ EXTERN INT_VAR (tweak_AdaptProtoThresh, 230, "Tess VAR");
101
+ EXTERN INT_VAR (tweak_AdaptFeatureThresh, 230, "Tess VAR");
102
+ EXTERN INT_VAR (tweak_min_outline_points, 6, "Tess VAR");
103
+ EXTERN INT_VAR (tweak_min_outline_area, 2000, "Tess VAR");
104
+ EXTERN double_VAR (tweak_good_split, 50.0, "Tess VAR");
105
+ EXTERN double_VAR (tweak_ok_split, 100.0, "Tess VAR");
106
+
107
+ extern inT16 XOFFSET;
108
+ extern inT16 YOFFSET;
109
+ extern int NO_BLOCK;
110
+
111
+ //progress monitor
112
+ ETEXT_DESC *global_monitor = NULL;
113
+
114
+ void init_tesseract_lang_data(const char *arg0,
115
+ const char *textbase,
116
+ const char *language,
117
+ const char *configfile,
118
+ int configc,
119
+ const char *const *configv) {
120
+ FILE *var_file;
121
+ static char c_path[MAX_PATH]; //path for c code
122
+
123
+ // Set the basename, compute the data directory and read C++ configs.
124
+ main_setup(arg0, textbase, configc, configv);
125
+ debug_window_on.set_value (FALSE);
126
+
127
+ if (tessedit_write_vars) {
128
+ var_file = fopen ("edited.cfg", "w");
129
+ if (var_file != NULL) {
130
+ print_variables(var_file);
131
+ fclose(var_file);
132
+ }
133
+ }
134
+ strcpy (c_path, datadir.string ());
135
+ c_path[strlen (c_path) - strlen (m_data_sub_dir.string ())] = '\0';
136
+ demodir = c_path;
137
+
138
+ // Set the language data path prefix
139
+ language_data_path_prefix = datadir;
140
+ if (language != NULL)
141
+ language_data_path_prefix += language;
142
+ else
143
+ language_data_path_prefix += "eng";
144
+ language_data_path_prefix += ".";
145
+
146
+ // Load the unichar set
147
+ STRING unicharpath = language_data_path_prefix;
148
+ unicharpath += "unicharset";
149
+ if (!unicharset.load_from_file(unicharpath.string())) {
150
+ cprintf("Unable to load unicharset file %s\n", unicharpath.string());
151
+ exit(1);
152
+ }
153
+ if (unicharset.size() > MAX_NUM_CLASSES) {
154
+ cprintf("Error: Size of unicharset is greater than MAX_NUM_CLASSES\n");
155
+ exit(1);
156
+ }
157
+ // Set the white and blacklists (if any)
158
+ unicharset.set_black_and_whitelist(tessedit_char_blacklist.string(),
159
+ tessedit_char_whitelist.string());
160
+ }
161
+
162
+ int init_tesseract(const char *arg0,
163
+ const char *textbase,
164
+ const char *language,
165
+ const char *configfile,
166
+ int configc,
167
+ const char *const *configv) {
168
+ init_tesseract_lang_data (arg0, textbase, language,
169
+ configfile, configc, configv);
170
+
171
+ start_recog(configfile, textbase);
172
+
173
+ set_tess_tweak_vars();
174
+
175
+ if (tessedit_use_nn) //phils nn stuff
176
+ init_net();
177
+ return 0; //Normal exit
178
+ }
179
+
180
+ // init the LM component
181
+ int init_tesseract_lm(const char *arg0,
182
+ const char *textbase,
183
+ const char *language,
184
+ const char *configfile,
185
+ int configc,
186
+ const char *const *configv) {
187
+ init_tesseract_lang_data (arg0, textbase, language,
188
+ configfile, configc, configv);
189
+
190
+ init_permute();
191
+
192
+ return 0; //Normal exit
193
+ }
194
+
195
+ void end_tesseract() {
196
+ end_recog();
197
+ }
198
+
199
+ #ifdef _TIFFIO_
200
+ void read_tiff_image(TIFF* tif, IMAGE* image) {
201
+ tdata_t buf;
202
+ uint32 image_width, image_height;
203
+ uint16 photometric;
204
+ inT16 bpp;
205
+ inT16 samples_per_pixel = 0;
206
+ TIFFGetField(tif, TIFFTAG_IMAGEWIDTH, &image_width);
207
+ TIFFGetField(tif, TIFFTAG_IMAGELENGTH, &image_height);
208
+ TIFFGetField(tif, TIFFTAG_BITSPERSAMPLE, &bpp);
209
+ TIFFGetField(tif, TIFFTAG_SAMPLESPERPIXEL, &samples_per_pixel);
210
+ TIFFGetField(tif, TIFFTAG_PHOTOMETRIC, &photometric);
211
+ if (samples_per_pixel > 1)
212
+ bpp *= samples_per_pixel;
213
+ // Tesseract's internal representation is 0-is-black,
214
+ // so if the photometric is 1 (min is black) then high-valued pixels
215
+ // are 1 (white), otherwise they are 0 (black).
216
+ uinT8 high_value = photometric == 1;
217
+ image->create(image_width, image_height, bpp);
218
+ IMAGELINE line;
219
+ line.init(image_width);
220
+
221
+ buf = _TIFFmalloc(TIFFScanlineSize(tif));
222
+ int bytes_per_line = (image_width*bpp + 7)/8;
223
+ uinT8* dest_buf = image->get_buffer();
224
+ // This will go badly wrong with one of the more exotic tiff formats,
225
+ // but the majority will work OK.
226
+ for (int y = 0; y < image_height; ++y) {
227
+ TIFFReadScanline(tif, buf, y);
228
+ memcpy(dest_buf, buf, bytes_per_line);
229
+ dest_buf += bytes_per_line;
230
+ }
231
+ if (high_value == 0)
232
+ invert_image(image);
233
+ _TIFFfree(buf);
234
+ }
235
+ #endif
236
+
237
+ /* Define command type identifiers */
238
+
239
+ enum CMD_EVENTS
240
+ {
241
+ ACTION_1_CMD_EVENT,
242
+ RECOG_WERDS,
243
+ RECOG_PSEUDO,
244
+ ACTION_2_CMD_EVENT
245
+ };
246
+
247
+
248
+ /*************************************************************************
249
+ * set_tess_tweak_vars()
250
+ * Set TESS vars from the tweek value - This is only really of use during search
251
+ * of the space of tess configs - othertimes the default values are set
252
+ *
253
+ *************************************************************************/
254
+ void set_tess_tweak_vars() {
255
+ if (tessedit_tweaking_tess_vars) {
256
+ garbage = tweak_garbage;
257
+ ok_word = tweak_ok_word;
258
+ good_word = tweak_good_word;
259
+ freq_word = tweak_freq_word;
260
+ ok_number = tweak_ok_number;
261
+ good_number = tweak_good_number;
262
+ non_word = tweak_non_word;
263
+ CertaintyPerChar = tweak_CertaintyPerChar;
264
+ NonDictCertainty = tweak_NonDictCertainty;
265
+ RejectCertaintyOffset = tweak_RejectCertaintyOffset;
266
+ GoodAdaptiveMatch = tweak_GoodAdaptiveMatch;
267
+ GreatAdaptiveMatch = tweak_GreatAdaptiveMatch;
268
+ ReliableConfigThreshold = tweak_ReliableConfigThreshold;
269
+ AdaptProtoThresh = tweak_AdaptProtoThresh;
270
+ AdaptFeatureThresh = tweak_AdaptFeatureThresh;
271
+ min_outline_points = tweak_min_outline_points;
272
+ min_outline_area = tweak_min_outline_area;
273
+ good_split = tweak_good_split;
274
+ ok_split = tweak_ok_split;
275
+ }
276
+ // if (expiry_day * 24 * 60 * 60 < time(NULL))
277
+ // err_exit();
278
+ }
@@ -0,0 +1,49 @@
1
+ /**********************************************************************
2
+ * File: tessedit.h (Formerly tessedit.h)
3
+ * Description: Main program for merge of tess and editor.
4
+ * Author: Ray Smith
5
+ * Created: Tue Jan 07 15:21:46 GMT 1992
6
+ *
7
+ * (C) Copyright 1992, Hewlett-Packard Ltd.
8
+ ** Licensed under the Apache License, Version 2.0 (the "License");
9
+ ** you may not use this file except in compliance with the License.
10
+ ** You may obtain a copy of the License at
11
+ ** http://www.apache.org/licenses/LICENSE-2.0
12
+ ** Unless required by applicable law or agreed to in writing, software
13
+ ** distributed under the License is distributed on an "AS IS" BASIS,
14
+ ** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
15
+ ** See the License for the specific language governing permissions and
16
+ ** limitations under the License.
17
+ *
18
+ **********************************************************************/
19
+
20
+ #ifndef TESSEDIT_H
21
+ #define TESSEDIT_H
22
+
23
+ #include "tessclas.h"
24
+ #include "ocrclass.h"
25
+ #include "pgedit.h"
26
+ #include "notdll.h"
27
+
28
+ //progress monitor
29
+ extern ETEXT_DESC *global_monitor;
30
+
31
+ int init_tesseract(const char *arg0,
32
+ const char *textbase,
33
+ const char *language,
34
+ const char *configfile,
35
+ int configc,
36
+ const char *const *configv);
37
+
38
+ int init_tesseract_lm(const char *arg0,
39
+ const char *textbase,
40
+ const char *language,
41
+ const char *configfile,
42
+ int configc,
43
+ const char *const *configv);
44
+
45
+ void recognize_page(STRING& image_name);
46
+ void end_tesseract();
47
+
48
+ void set_tess_tweak_vars();
49
+ #endif
@@ -0,0 +1,110 @@
1
+ /**********************************************************************
2
+ * File: tessembedded.cpp
3
+ * Description: Main program for merge of tess and editor.
4
+ * Author: Marius Renn
5
+ * Created: Sun Oct 21 2006
6
+ *
7
+ * (C) Copyright 1992, Hewlett-Packard Ltd.
8
+ ** Licensed under the Apache License, Version 2.0 (the "License");
9
+ ** you may not use this file except in compliance with the License.
10
+ ** You may obtain a copy of the License at
11
+ ** http://www.apache.org/licenses/LICENSE-2.0
12
+ ** Unless required by applicable law or agreed to in writing, software
13
+ ** distributed under the License is distributed on an "AS IS" BASIS,
14
+ ** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
15
+ ** See the License for the specific language governing permissions and
16
+ ** limitations under the License.
17
+ *
18
+ **********************************************************************/
19
+
20
+ #include "mfcpch.h"
21
+ #include "applybox.h"
22
+ #include "control.h"
23
+ #include "tessvars.h"
24
+ #include "tessedit.h"
25
+ #include "pageres.h"
26
+ #include "imgs.h"
27
+ #include "varabled.h"
28
+ #include "tprintf.h"
29
+ #include "tessembedded.h"
30
+ #include "stderr.h"
31
+ #include "notdll.h"
32
+ #include "mainblk.h"
33
+ #include "globals.h"
34
+ #include "tfacep.h"
35
+ #include "callnet.h"
36
+
37
+ void tessembedded_read_file(STRING &name,
38
+ BLOCK_LIST *blocks) {
39
+ int c; //input character
40
+ FILE *infp; //input file
41
+ BLOCK_IT block_it(blocks); //iterator
42
+ BLOCK *block; //current block
43
+
44
+ ICOORD page_tr; //topright of page
45
+
46
+ char *filename_extension;
47
+
48
+ block_it.move_to_last ();
49
+
50
+ // ptr to last dot
51
+ filename_extension = strrchr (name.string (), '.');
52
+ #ifdef __UNIX__
53
+ if (strcmp (filename_extension, ".pb") == 0) {
54
+ tprintf ("Converting from .pb file format.\n");
55
+ //construct blocks
56
+ read_and_textord (name.string (), blocks);
57
+ }
58
+ else
59
+ #endif
60
+ //xiaofan, a hack here
61
+ if (strcmp (filename_extension, ".tif") == 0) {
62
+ // tprintf( "Interpreting .bl file format.\n" );
63
+ //construct blocks
64
+ edges_and_textord (name.string (), blocks);
65
+ }
66
+ else {
67
+ if ((strcmp (filename_extension, ".pg") == 0) ||
68
+ // read a .pg file
69
+ // or a .sp file
70
+ (strcmp (filename_extension, ".sp") == 0)) {
71
+ tprintf ("Reading %s file format.\n", filename_extension);
72
+ infp = fopen (name.string (), "r");
73
+ if (infp == NULL)
74
+ CANTOPENFILE.error ("pgeditor_read_file", EXIT, name.string ());
75
+ //can't open file
76
+
77
+ while (((c = fgetc (infp)) != EOF) && (ungetc (c, infp) != EOF)) {
78
+ //get one
79
+ block = BLOCK::de_serialise (infp);
80
+ //add to list
81
+ block_it.add_after_then_move (block);
82
+ }
83
+ fclose(infp);
84
+ }
85
+ }
86
+ }
87
+
88
+ int init_tessembedded(const char *arg0,
89
+ const char *textbase,
90
+ const char *configfile,
91
+ int configc,
92
+ const char *const *configv) {
93
+ main_setup(arg0, textbase, configc, configv);
94
+
95
+ debug_window_on.set_value (FALSE);
96
+
97
+ static char c_path[MAX_PATH]; //path for c code
98
+ strcpy (c_path, datadir.string ());
99
+ c_path[strlen (c_path) - strlen (m_data_sub_dir.string ())] = '\0';
100
+ demodir = c_path;
101
+ start_recog(configfile, textbase);
102
+
103
+ init_net();
104
+
105
+ return 0;
106
+ }
107
+
108
+ void end_tessembedded() {
109
+ end_recog();
110
+ }
@@ -0,0 +1,38 @@
1
+ /**********************************************************************
2
+ * File: tessembedded.h
3
+ * Description: Access to initialization functions in embedded environment
4
+ * Author: Marius Renn
5
+ * Created: Sun Oct 21
6
+ *
7
+ * (C) Copyright 1992, Hewlett-Packard Ltd.
8
+ ** Licensed under the Apache License, Version 2.0 (the "License");
9
+ ** you may not use this file except in compliance with the License.
10
+ ** You may obtain a copy of the License at
11
+ ** http://www.apache.org/licenses/LICENSE-2.0
12
+ ** Unless required by applicable law or agreed to in writing, software
13
+ ** distributed under the License is distributed on an "AS IS" BASIS,
14
+ ** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
15
+ ** See the License for the specific language governing permissions and
16
+ ** limitations under the License.
17
+ *
18
+ **********************************************************************/
19
+
20
+ #ifndef TESSEMBEDDED_H
21
+ #define TESSEMBEDDED_H
22
+
23
+ #include "ocrblock.h"
24
+ #include "varable.h"
25
+ #include "notdll.h"
26
+
27
+ int init_tessembedded(const char *arg0,
28
+ const char *textbase,
29
+ const char *configfile,
30
+ int configc,
31
+ const char *const *configv);
32
+
33
+ void tessembedded_read_file(STRING &name,
34
+ BLOCK_LIST *blocks);
35
+
36
+ void end_tessembedded();
37
+
38
+ #endif
@@ -0,0 +1,37 @@
1
+ /**********************************************************************
2
+ * File: tesseractfull.cc
3
+ * Description: Test function to link tesseractfull.a.
4
+ * Author: Ray Smith
5
+ * Created: Tue Jul 17 16:23:46 PDT 2007
6
+ *
7
+ * (C) Copyright 2007 Google Inc.
8
+ ** Licensed under the Apache License, Version 2.0 (the "License");
9
+ ** you may not use this file except in compliance with the License.
10
+ ** You may obtain a copy of the License at
11
+ ** http://www.apache.org/licenses/LICENSE-2.0
12
+ ** Unless required by applicable law or agreed to in writing, software
13
+ ** distributed under the License is distributed on an "AS IS" BASIS,
14
+ ** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
15
+ ** See the License for the specific language governing permissions and
16
+ ** limitations under the License.
17
+ *
18
+ **********************************************************************/
19
+
20
+ #include "baseapi.h"
21
+ #ifndef NULL
22
+ #define NULL 0L
23
+ #endif
24
+
25
+ char* run_tesseract(const char* language,
26
+ const unsigned char* imagedata,
27
+ int bytes_per_pixel,
28
+ int bytes_per_line,
29
+ int width, int height) {
30
+ TessBaseAPI::InitWithLanguage(NULL, NULL, language, NULL, false, 0, NULL);
31
+ char* text = TessBaseAPI::TesseractRect(imagedata, bytes_per_pixel,
32
+ bytes_per_line, 0, 0,
33
+ width, height);
34
+ TessBaseAPI::End();
35
+
36
+ return text; //Normal exit
37
+ }