finishm 0.0.1

Sign up to get free protection for your applications and to get access to all the features.
Files changed (554) hide show
  1. checksums.yaml +7 -0
  2. data/.document +5 -0
  3. data/.gitmodules +3 -0
  4. data/.rspec +1 -0
  5. data/Gemfile +31 -0
  6. data/LICENSE.txt +20 -0
  7. data/README.md +59 -0
  8. data/Rakefile +51 -0
  9. data/VERSION +1 -0
  10. data/bin/assembly_visualiser +106 -0
  11. data/bin/check_primer_combinations.rb +73 -0
  12. data/bin/contig_joiner.rb +244 -0
  13. data/bin/contigs_against_assembly.rb +153 -0
  14. data/bin/finishm +143 -0
  15. data/bin/finishm_assembler +55 -0
  16. data/bin/finishm_gap_closer.rb +241 -0
  17. data/bin/kmer_abundance_file_tool.rb +49 -0
  18. data/bin/kmer_pattern_to_assembly.rb +377 -0
  19. data/bin/kmer_profile_finder.rb +92 -0
  20. data/bin/kmers_count_parse.d +52 -0
  21. data/bin/kmers_count_tabulate.d +123 -0
  22. data/bin/kmers_count_tabulate.rb +84 -0
  23. data/bin/pcr_result_parser.rb +108 -0
  24. data/bin/primer_finder.rb +119 -0
  25. data/bin/read_selection_by_kmer.d +174 -0
  26. data/bin/scaffold_by_pattern.rb +119 -0
  27. data/bin/scaffold_connection_possibilities_to_knowns.rb +193 -0
  28. data/bin/scaffold_end_coverages.rb +69 -0
  29. data/bin/trail_validator.rb +84 -0
  30. data/ext/mkrf_conf.rb +56 -0
  31. data/ext/src/Makefile +140 -0
  32. data/ext/src/src/allocArray.c +305 -0
  33. data/ext/src/src/allocArray.h +86 -0
  34. data/ext/src/src/autoOpen.c +107 -0
  35. data/ext/src/src/autoOpen.h +18 -0
  36. data/ext/src/src/binarySequences.c +813 -0
  37. data/ext/src/src/binarySequences.h +125 -0
  38. data/ext/src/src/concatenatedGraph.c +233 -0
  39. data/ext/src/src/concatenatedGraph.h +30 -0
  40. data/ext/src/src/concatenatedPreGraph.c +262 -0
  41. data/ext/src/src/concatenatedPreGraph.h +29 -0
  42. data/ext/src/src/correctedGraph.c +2643 -0
  43. data/ext/src/src/correctedGraph.h +32 -0
  44. data/ext/src/src/dfib.c +509 -0
  45. data/ext/src/src/dfib.h +69 -0
  46. data/ext/src/src/dfibHeap.c +89 -0
  47. data/ext/src/src/dfibHeap.h +39 -0
  48. data/ext/src/src/dfibpriv.h +105 -0
  49. data/ext/src/src/fib.c +628 -0
  50. data/ext/src/src/fib.h +78 -0
  51. data/ext/src/src/fibHeap.c +79 -0
  52. data/ext/src/src/fibHeap.h +41 -0
  53. data/ext/src/src/fibpriv.h +110 -0
  54. data/ext/src/src/globals.h +154 -0
  55. data/ext/src/src/graph.c +3932 -0
  56. data/ext/src/src/graph.h +233 -0
  57. data/ext/src/src/graphReConstruction.c +1472 -0
  58. data/ext/src/src/graphReConstruction.h +30 -0
  59. data/ext/src/src/graphStats.c +2167 -0
  60. data/ext/src/src/graphStats.h +72 -0
  61. data/ext/src/src/graphStructures.h +52 -0
  62. data/ext/src/src/kmer.c +652 -0
  63. data/ext/src/src/kmer.h +73 -0
  64. data/ext/src/src/kmerOccurenceTable.c +236 -0
  65. data/ext/src/src/kmerOccurenceTable.h +44 -0
  66. data/ext/src/src/kseq.h +223 -0
  67. data/ext/src/src/locallyCorrectedGraph.c +557 -0
  68. data/ext/src/src/locallyCorrectedGraph.h +40 -0
  69. data/ext/src/src/passageMarker.c +677 -0
  70. data/ext/src/src/passageMarker.h +137 -0
  71. data/ext/src/src/preGraph.c +1717 -0
  72. data/ext/src/src/preGraph.h +106 -0
  73. data/ext/src/src/preGraphConstruction.c +990 -0
  74. data/ext/src/src/preGraphConstruction.h +26 -0
  75. data/ext/src/src/probe_node_finder.c +84 -0
  76. data/ext/src/src/probe_node_finder.h +6 -0
  77. data/ext/src/src/readCoherentGraph.c +557 -0
  78. data/ext/src/src/readCoherentGraph.h +30 -0
  79. data/ext/src/src/readSet.c +1734 -0
  80. data/ext/src/src/readSet.h +67 -0
  81. data/ext/src/src/readToNode.c +218 -0
  82. data/ext/src/src/readToNode.h +35 -0
  83. data/ext/src/src/recycleBin.c +199 -0
  84. data/ext/src/src/recycleBin.h +58 -0
  85. data/ext/src/src/roadMap.c +342 -0
  86. data/ext/src/src/roadMap.h +65 -0
  87. data/ext/src/src/run.c +318 -0
  88. data/ext/src/src/run.h +52 -0
  89. data/ext/src/src/run2.c +744 -0
  90. data/ext/src/src/runReadToNode.c +29 -0
  91. data/ext/src/src/scaffold.c +1876 -0
  92. data/ext/src/src/scaffold.h +64 -0
  93. data/ext/src/src/shortReadPairs.c +1243 -0
  94. data/ext/src/src/shortReadPairs.h +32 -0
  95. data/ext/src/src/splay.c +259 -0
  96. data/ext/src/src/splay.h +43 -0
  97. data/ext/src/src/splayTable.c +1315 -0
  98. data/ext/src/src/splayTable.h +31 -0
  99. data/ext/src/src/tightString.c +362 -0
  100. data/ext/src/src/tightString.h +82 -0
  101. data/ext/src/src/utility.c +199 -0
  102. data/ext/src/src/utility.h +98 -0
  103. data/ext/src/third-party/zlib-1.2.3/ChangeLog +855 -0
  104. data/ext/src/third-party/zlib-1.2.3/FAQ +339 -0
  105. data/ext/src/third-party/zlib-1.2.3/INDEX +51 -0
  106. data/ext/src/third-party/zlib-1.2.3/Makefile +154 -0
  107. data/ext/src/third-party/zlib-1.2.3/Makefile.in +154 -0
  108. data/ext/src/third-party/zlib-1.2.3/README +125 -0
  109. data/ext/src/third-party/zlib-1.2.3/adler32.c +149 -0
  110. data/ext/src/third-party/zlib-1.2.3/adler32.o +0 -0
  111. data/ext/src/third-party/zlib-1.2.3/algorithm.txt +209 -0
  112. data/ext/src/third-party/zlib-1.2.3/amiga/Makefile.pup +66 -0
  113. data/ext/src/third-party/zlib-1.2.3/amiga/Makefile.sas +65 -0
  114. data/ext/src/third-party/zlib-1.2.3/as400/bndsrc +132 -0
  115. data/ext/src/third-party/zlib-1.2.3/as400/compile.clp +123 -0
  116. data/ext/src/third-party/zlib-1.2.3/as400/readme.txt +111 -0
  117. data/ext/src/third-party/zlib-1.2.3/as400/zlib.inc +331 -0
  118. data/ext/src/third-party/zlib-1.2.3/compress.c +79 -0
  119. data/ext/src/third-party/zlib-1.2.3/compress.o +0 -0
  120. data/ext/src/third-party/zlib-1.2.3/configure +459 -0
  121. data/ext/src/third-party/zlib-1.2.3/contrib/README.contrib +71 -0
  122. data/ext/src/third-party/zlib-1.2.3/contrib/ada/buffer_demo.adb +106 -0
  123. data/ext/src/third-party/zlib-1.2.3/contrib/ada/mtest.adb +156 -0
  124. data/ext/src/third-party/zlib-1.2.3/contrib/ada/read.adb +156 -0
  125. data/ext/src/third-party/zlib-1.2.3/contrib/ada/readme.txt +65 -0
  126. data/ext/src/third-party/zlib-1.2.3/contrib/ada/test.adb +463 -0
  127. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-streams.adb +225 -0
  128. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-streams.ads +114 -0
  129. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-thin.adb +141 -0
  130. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib-thin.ads +450 -0
  131. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib.adb +701 -0
  132. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib.ads +328 -0
  133. data/ext/src/third-party/zlib-1.2.3/contrib/ada/zlib.gpr +20 -0
  134. data/ext/src/third-party/zlib-1.2.3/contrib/asm586/README.586 +43 -0
  135. data/ext/src/third-party/zlib-1.2.3/contrib/asm586/match.S +364 -0
  136. data/ext/src/third-party/zlib-1.2.3/contrib/asm686/README.686 +34 -0
  137. data/ext/src/third-party/zlib-1.2.3/contrib/asm686/match.S +329 -0
  138. data/ext/src/third-party/zlib-1.2.3/contrib/blast/Makefile +8 -0
  139. data/ext/src/third-party/zlib-1.2.3/contrib/blast/README +4 -0
  140. data/ext/src/third-party/zlib-1.2.3/contrib/blast/blast.c +444 -0
  141. data/ext/src/third-party/zlib-1.2.3/contrib/blast/blast.h +71 -0
  142. data/ext/src/third-party/zlib-1.2.3/contrib/blast/test.pk +0 -0
  143. data/ext/src/third-party/zlib-1.2.3/contrib/blast/test.txt +1 -0
  144. data/ext/src/third-party/zlib-1.2.3/contrib/delphi/ZLib.pas +557 -0
  145. data/ext/src/third-party/zlib-1.2.3/contrib/delphi/ZLibConst.pas +11 -0
  146. data/ext/src/third-party/zlib-1.2.3/contrib/delphi/readme.txt +76 -0
  147. data/ext/src/third-party/zlib-1.2.3/contrib/delphi/zlibd32.mak +93 -0
  148. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib.build +33 -0
  149. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib.chm +0 -0
  150. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib.sln +21 -0
  151. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/AssemblyInfo.cs +58 -0
  152. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/ChecksumImpl.cs +202 -0
  153. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/CircularBuffer.cs +83 -0
  154. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/CodecBase.cs +198 -0
  155. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/Deflater.cs +106 -0
  156. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/DotZLib.cs +288 -0
  157. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/DotZLib.csproj +141 -0
  158. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/GZipStream.cs +301 -0
  159. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/Inflater.cs +105 -0
  160. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/DotZLib/UnitTests.cs +274 -0
  161. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/LICENSE_1_0.txt +23 -0
  162. data/ext/src/third-party/zlib-1.2.3/contrib/dotzlib/readme.txt +58 -0
  163. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/README +1 -0
  164. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/infback9.c +608 -0
  165. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/infback9.h +37 -0
  166. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inffix9.h +107 -0
  167. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inflate9.h +47 -0
  168. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inftree9.c +323 -0
  169. data/ext/src/third-party/zlib-1.2.3/contrib/infback9/inftree9.h +55 -0
  170. data/ext/src/third-party/zlib-1.2.3/contrib/inflate86/inffas86.c +1157 -0
  171. data/ext/src/third-party/zlib-1.2.3/contrib/inflate86/inffast.S +1368 -0
  172. data/ext/src/third-party/zlib-1.2.3/contrib/iostream/test.cpp +24 -0
  173. data/ext/src/third-party/zlib-1.2.3/contrib/iostream/zfstream.cpp +329 -0
  174. data/ext/src/third-party/zlib-1.2.3/contrib/iostream/zfstream.h +128 -0
  175. data/ext/src/third-party/zlib-1.2.3/contrib/iostream2/zstream.h +307 -0
  176. data/ext/src/third-party/zlib-1.2.3/contrib/iostream2/zstream_test.cpp +25 -0
  177. data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/README +35 -0
  178. data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/TODO +17 -0
  179. data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/test.cc +50 -0
  180. data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/zfstream.cc +479 -0
  181. data/ext/src/third-party/zlib-1.2.3/contrib/iostream3/zfstream.h +466 -0
  182. data/ext/src/third-party/zlib-1.2.3/contrib/masm686/match.asm +413 -0
  183. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/bld_ml64.bat +2 -0
  184. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/gvmat64.asm +513 -0
  185. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/gvmat64.obj +0 -0
  186. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/inffas8664.c +186 -0
  187. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/inffasx64.asm +392 -0
  188. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/inffasx64.obj +0 -0
  189. data/ext/src/third-party/zlib-1.2.3/contrib/masmx64/readme.txt +28 -0
  190. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/bld_ml32.bat +2 -0
  191. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/gvmat32.asm +972 -0
  192. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/gvmat32.obj +0 -0
  193. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/gvmat32c.c +62 -0
  194. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/inffas32.asm +1083 -0
  195. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/inffas32.obj +0 -0
  196. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/mkasm.bat +3 -0
  197. data/ext/src/third-party/zlib-1.2.3/contrib/masmx86/readme.txt +21 -0
  198. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/ChangeLogUnzip +67 -0
  199. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/Makefile +25 -0
  200. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/crypt.h +132 -0
  201. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/ioapi.c +177 -0
  202. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/ioapi.h +75 -0
  203. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/iowin32.c +270 -0
  204. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/iowin32.h +21 -0
  205. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/miniunz.c +585 -0
  206. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/minizip.c +420 -0
  207. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/mztools.c +281 -0
  208. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/mztools.h +31 -0
  209. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/unzip.c +1598 -0
  210. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/unzip.h +354 -0
  211. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/zip.c +1219 -0
  212. data/ext/src/third-party/zlib-1.2.3/contrib/minizip/zip.h +235 -0
  213. data/ext/src/third-party/zlib-1.2.3/contrib/pascal/example.pas +599 -0
  214. data/ext/src/third-party/zlib-1.2.3/contrib/pascal/readme.txt +76 -0
  215. data/ext/src/third-party/zlib-1.2.3/contrib/pascal/zlibd32.mak +93 -0
  216. data/ext/src/third-party/zlib-1.2.3/contrib/pascal/zlibpas.pas +236 -0
  217. data/ext/src/third-party/zlib-1.2.3/contrib/puff/Makefile +8 -0
  218. data/ext/src/third-party/zlib-1.2.3/contrib/puff/README +63 -0
  219. data/ext/src/third-party/zlib-1.2.3/contrib/puff/puff.c +837 -0
  220. data/ext/src/third-party/zlib-1.2.3/contrib/puff/puff.h +31 -0
  221. data/ext/src/third-party/zlib-1.2.3/contrib/puff/zeros.raw +0 -0
  222. data/ext/src/third-party/zlib-1.2.3/contrib/testzlib/testzlib.c +275 -0
  223. data/ext/src/third-party/zlib-1.2.3/contrib/testzlib/testzlib.txt +10 -0
  224. data/ext/src/third-party/zlib-1.2.3/contrib/untgz/Makefile +14 -0
  225. data/ext/src/third-party/zlib-1.2.3/contrib/untgz/Makefile.msc +17 -0
  226. data/ext/src/third-party/zlib-1.2.3/contrib/untgz/untgz.c +674 -0
  227. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/readme.txt +73 -0
  228. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/miniunz.vcproj +126 -0
  229. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/minizip.vcproj +126 -0
  230. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/testzlib.vcproj +126 -0
  231. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlib.rc +32 -0
  232. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibstat.vcproj +246 -0
  233. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibvc.def +92 -0
  234. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibvc.sln +78 -0
  235. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc7/zlibvc.vcproj +445 -0
  236. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/miniunz.vcproj +566 -0
  237. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/minizip.vcproj +563 -0
  238. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/testzlib.vcproj +948 -0
  239. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/testzlibdll.vcproj +567 -0
  240. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlib.rc +32 -0
  241. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibstat.vcproj +870 -0
  242. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibvc.def +92 -0
  243. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibvc.sln +144 -0
  244. data/ext/src/third-party/zlib-1.2.3/contrib/vstudio/vc8/zlibvc.vcproj +1219 -0
  245. data/ext/src/third-party/zlib-1.2.3/crc32.c +423 -0
  246. data/ext/src/third-party/zlib-1.2.3/crc32.h +441 -0
  247. data/ext/src/third-party/zlib-1.2.3/crc32.o +0 -0
  248. data/ext/src/third-party/zlib-1.2.3/deflate.c +1736 -0
  249. data/ext/src/third-party/zlib-1.2.3/deflate.h +331 -0
  250. data/ext/src/third-party/zlib-1.2.3/deflate.o +0 -0
  251. data/ext/src/third-party/zlib-1.2.3/example +0 -0
  252. data/ext/src/third-party/zlib-1.2.3/example.c +565 -0
  253. data/ext/src/third-party/zlib-1.2.3/examples/README.examples +42 -0
  254. data/ext/src/third-party/zlib-1.2.3/examples/fitblk.c +233 -0
  255. data/ext/src/third-party/zlib-1.2.3/examples/gun.c +693 -0
  256. data/ext/src/third-party/zlib-1.2.3/examples/gzappend.c +500 -0
  257. data/ext/src/third-party/zlib-1.2.3/examples/gzjoin.c +448 -0
  258. data/ext/src/third-party/zlib-1.2.3/examples/gzlog.c +413 -0
  259. data/ext/src/third-party/zlib-1.2.3/examples/gzlog.h +58 -0
  260. data/ext/src/third-party/zlib-1.2.3/examples/zlib_how.html +523 -0
  261. data/ext/src/third-party/zlib-1.2.3/examples/zpipe.c +191 -0
  262. data/ext/src/third-party/zlib-1.2.3/examples/zran.c +404 -0
  263. data/ext/src/third-party/zlib-1.2.3/gzio.c +1026 -0
  264. data/ext/src/third-party/zlib-1.2.3/gzio.o +0 -0
  265. data/ext/src/third-party/zlib-1.2.3/infback.c +623 -0
  266. data/ext/src/third-party/zlib-1.2.3/infback.o +0 -0
  267. data/ext/src/third-party/zlib-1.2.3/inffast.c +318 -0
  268. data/ext/src/third-party/zlib-1.2.3/inffast.h +11 -0
  269. data/ext/src/third-party/zlib-1.2.3/inffast.o +0 -0
  270. data/ext/src/third-party/zlib-1.2.3/inffixed.h +94 -0
  271. data/ext/src/third-party/zlib-1.2.3/inflate.c +1368 -0
  272. data/ext/src/third-party/zlib-1.2.3/inflate.h +115 -0
  273. data/ext/src/third-party/zlib-1.2.3/inflate.o +0 -0
  274. data/ext/src/third-party/zlib-1.2.3/inftrees.c +329 -0
  275. data/ext/src/third-party/zlib-1.2.3/inftrees.h +55 -0
  276. data/ext/src/third-party/zlib-1.2.3/inftrees.o +0 -0
  277. data/ext/src/third-party/zlib-1.2.3/libz.a +0 -0
  278. data/ext/src/third-party/zlib-1.2.3/make_vms.com +461 -0
  279. data/ext/src/third-party/zlib-1.2.3/minigzip +0 -0
  280. data/ext/src/third-party/zlib-1.2.3/minigzip.c +322 -0
  281. data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.bor +109 -0
  282. data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.dj2 +104 -0
  283. data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.emx +69 -0
  284. data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.msc +106 -0
  285. data/ext/src/third-party/zlib-1.2.3/msdos/Makefile.tc +94 -0
  286. data/ext/src/third-party/zlib-1.2.3/old/Makefile.riscos +151 -0
  287. data/ext/src/third-party/zlib-1.2.3/old/README +3 -0
  288. data/ext/src/third-party/zlib-1.2.3/old/descrip.mms +48 -0
  289. data/ext/src/third-party/zlib-1.2.3/old/os2/Makefile.os2 +136 -0
  290. data/ext/src/third-party/zlib-1.2.3/old/os2/zlib.def +51 -0
  291. data/ext/src/third-party/zlib-1.2.3/old/visual-basic.txt +160 -0
  292. data/ext/src/third-party/zlib-1.2.3/old/zlib.html +971 -0
  293. data/ext/src/third-party/zlib-1.2.3/projects/README.projects +41 -0
  294. data/ext/src/third-party/zlib-1.2.3/projects/visualc6/README.txt +73 -0
  295. data/ext/src/third-party/zlib-1.2.3/projects/visualc6/example.dsp +278 -0
  296. data/ext/src/third-party/zlib-1.2.3/projects/visualc6/minigzip.dsp +278 -0
  297. data/ext/src/third-party/zlib-1.2.3/projects/visualc6/zlib.dsp +609 -0
  298. data/ext/src/third-party/zlib-1.2.3/projects/visualc6/zlib.dsw +59 -0
  299. data/ext/src/third-party/zlib-1.2.3/qnx/package.qpg +141 -0
  300. data/ext/src/third-party/zlib-1.2.3/trees.c +1219 -0
  301. data/ext/src/third-party/zlib-1.2.3/trees.h +128 -0
  302. data/ext/src/third-party/zlib-1.2.3/trees.o +0 -0
  303. data/ext/src/third-party/zlib-1.2.3/uncompr.c +61 -0
  304. data/ext/src/third-party/zlib-1.2.3/uncompr.o +0 -0
  305. data/ext/src/third-party/zlib-1.2.3/win32/DLL_FAQ.txt +397 -0
  306. data/ext/src/third-party/zlib-1.2.3/win32/Makefile.bor +107 -0
  307. data/ext/src/third-party/zlib-1.2.3/win32/Makefile.emx +69 -0
  308. data/ext/src/third-party/zlib-1.2.3/win32/Makefile.gcc +141 -0
  309. data/ext/src/third-party/zlib-1.2.3/win32/Makefile.msc +126 -0
  310. data/ext/src/third-party/zlib-1.2.3/win32/VisualC.txt +3 -0
  311. data/ext/src/third-party/zlib-1.2.3/win32/zlib.def +60 -0
  312. data/ext/src/third-party/zlib-1.2.3/win32/zlib1.rc +39 -0
  313. data/ext/src/third-party/zlib-1.2.3/zconf.h +332 -0
  314. data/ext/src/third-party/zlib-1.2.3/zconf.in.h +332 -0
  315. data/ext/src/third-party/zlib-1.2.3/zlib.3 +159 -0
  316. data/ext/src/third-party/zlib-1.2.3/zlib.h +1357 -0
  317. data/ext/src/third-party/zlib-1.2.3/zutil.c +318 -0
  318. data/ext/src/third-party/zlib-1.2.3/zutil.h +269 -0
  319. data/ext/src/third-party/zlib-1.2.3/zutil.o +0 -0
  320. data/lib/assembly/a_b_visualiser.rb +169 -0
  321. data/lib/assembly/acyclic_connection_finder.rb +81 -0
  322. data/lib/assembly/all_orfs.rb +615 -0
  323. data/lib/assembly/bad_format_writer.rb +46 -0
  324. data/lib/assembly/bam_probe_read_selector.rb +48 -0
  325. data/lib/assembly/bubbly_assembler.rb +842 -0
  326. data/lib/assembly/c_probe_node_finder.rb +38 -0
  327. data/lib/assembly/connection_interpreter.rb +350 -0
  328. data/lib/assembly/contig_printer.rb +400 -0
  329. data/lib/assembly/coverage_based_graph_filter.rb +68 -0
  330. data/lib/assembly/depth_first_search.rb +63 -0
  331. data/lib/assembly/dijkstra.rb +216 -0
  332. data/lib/assembly/fluffer.rb +253 -0
  333. data/lib/assembly/graph_explorer.rb +85 -0
  334. data/lib/assembly/graph_generator.rb +315 -0
  335. data/lib/assembly/height_finder.rb +355 -0
  336. data/lib/assembly/hybrid_velvet_graph.rb +70 -0
  337. data/lib/assembly/input_genome.rb +182 -0
  338. data/lib/assembly/kmer_coverage_based_path_filter.rb +65 -0
  339. data/lib/assembly/node_finder.rb +171 -0
  340. data/lib/assembly/oriented_node_trail.rb +507 -0
  341. data/lib/assembly/paired_end_assembler.rb +53 -0
  342. data/lib/assembly/paired_end_neighbour_finder.rb +176 -0
  343. data/lib/assembly/probed_graph.rb +105 -0
  344. data/lib/assembly/read_input.rb +79 -0
  345. data/lib/assembly/read_to_node.rb +37 -0
  346. data/lib/assembly/scaffold_breaker.rb +126 -0
  347. data/lib/assembly/sequence_hasher.rb +71 -0
  348. data/lib/assembly/single_coherent_paths_between_nodes.rb +533 -0
  349. data/lib/assembly/single_coherent_wanderer.rb +261 -0
  350. data/lib/assembly/single_ended_assembler.rb +441 -0
  351. data/lib/assembly/velvet_c_binding.rb +54 -0
  352. data/lib/assembly/velvet_graph_sequence_extractor.rb +123 -0
  353. data/lib/external/VERSION +1 -0
  354. data/lib/finishm/assemble.rb +224 -0
  355. data/lib/finishm/explore.rb +217 -0
  356. data/lib/finishm/finisher.rb +303 -0
  357. data/lib/finishm/fluff.rb +122 -0
  358. data/lib/finishm/gapfiller.rb +325 -0
  359. data/lib/finishm/orfs_finder.rb +88 -0
  360. data/lib/finishm/path_counter.rb +90 -0
  361. data/lib/finishm/primers.rb +425 -0
  362. data/lib/finishm/primers_check.rb +176 -0
  363. data/lib/finishm/roundup.rb +344 -0
  364. data/lib/finishm/sequence.rb +142 -0
  365. data/lib/finishm/visualise.rb +430 -0
  366. data/lib/finishm/wander.rb +270 -0
  367. data/lib/kmer_abundance_pattern.rb +79 -0
  368. data/lib/kmer_multi_abundance_file.rb +48 -0
  369. data/lib/oligo_designer.rb +88 -0
  370. data/lib/priner.rb +66 -0
  371. data/spec/acyclic_connection_finder_spec.rb +551 -0
  372. data/spec/all_orfs_spec.rb +443 -0
  373. data/spec/assemble_spec.rb +186 -0
  374. data/spec/bubbly_assembler_spec.rb +707 -0
  375. data/spec/c_node_finder_spec.rb +58 -0
  376. data/spec/connection_interpreter_spec.rb +284 -0
  377. data/spec/contig_printer_spec.rb +291 -0
  378. data/spec/coverage_based_graph_filter_spec.rb +102 -0
  379. data/spec/data/6_3e4e5e6e.1vANME.bam +0 -0
  380. data/spec/data/6_3e4e5e6e.1vANME.bam.bai +0 -0
  381. data/spec/data/acyclic_connection_finder/1/probes.fa +5 -0
  382. data/spec/data/acyclic_connection_finder/1/random1.fa +2 -0
  383. data/spec/data/acyclic_connection_finder/1/random1.sammy.fa.gz +0 -0
  384. data/spec/data/acyclic_connection_finder/1/random2.fa +2 -0
  385. data/spec/data/acyclic_connection_finder/1/random2.sammy.fa.gz +0 -0
  386. data/spec/data/assembly/1_simple_bubble_uneven_coverage/random3000.fa +39 -0
  387. data/spec/data/assembly/1_simple_bubble_uneven_coverage/random3000.slightly_changed.fa +39 -0
  388. data/spec/data/assembly/1_simple_bubble_uneven_coverage/reads_combined.fa.gz +0 -0
  389. data/spec/data/assembly_visualiser/Contig_6_1_to_250.fa.kmers31 +220 -0
  390. data/spec/data/assembly_visualiser/Contig_7_1_to_250.fa.kmers31 +220 -0
  391. data/spec/data/assembly_visualiser/Graph +46 -0
  392. data/spec/data/assembly_visualiser/start_kmers1 +2 -0
  393. data/spec/data/bands.csv +1 -0
  394. data/spec/data/c_probe_node_finder/1/CnyUnifiedSeq +0 -0
  395. data/spec/data/c_probe_node_finder/1/CnyUnifiedSeq.names +544 -0
  396. data/spec/data/c_probe_node_finder/1/Graph2 +668 -0
  397. data/spec/data/c_probe_node_finder/1/LastGraph +668 -0
  398. data/spec/data/c_probe_node_finder/1/Log +756 -0
  399. data/spec/data/c_probe_node_finder/1/PreGraph +11 -0
  400. data/spec/data/c_probe_node_finder/1/Roadmaps +2009 -0
  401. data/spec/data/c_probe_node_finder/1/contigs.fa +29 -0
  402. data/spec/data/c_probe_node_finder/1/stats.txt +6 -0
  403. data/spec/data/contig_printer/1/HOWTO_RECREATE +17 -0
  404. data/spec/data/contig_printer/1/contigs.fa +4 -0
  405. data/spec/data/contig_printer/1/seq.fa +2408 -0
  406. data/spec/data/contig_printer/1/seq.fa.svg +153 -0
  407. data/spec/data/contig_printer/1/seq.fa.velvet/Graph2 +2953 -0
  408. data/spec/data/contig_printer/1/seq.fa.velvet/LastGraph +2953 -0
  409. data/spec/data/contig_printer/1/seq.fa.velvet/Log +21 -0
  410. data/spec/data/contig_printer/1/seq.fa.velvet/PreGraph +27 -0
  411. data/spec/data/contig_printer/1/seq.fa.velvet/Roadmaps +5182 -0
  412. data/spec/data/contig_printer/1/seq.fa.velvet/Sequences +3612 -0
  413. data/spec/data/contig_printer/1/seq.fa.velvet/contigs.fa +36 -0
  414. data/spec/data/contig_printer/1/seq.fa.velvet/stats.txt +14 -0
  415. data/spec/data/contig_printer/1/seq.faVseq2_1to550.fa.bam +0 -0
  416. data/spec/data/contig_printer/1/seq.faVseq2_1to550.fa.bam.bai +0 -0
  417. data/spec/data/contig_printer/1/seq.node12.fa +4 -0
  418. data/spec/data/contig_printer/1/seq1_1to550.fa +2 -0
  419. data/spec/data/contig_printer/1/seq2_1to550.fa +2 -0
  420. data/spec/data/contig_printer/1/seq2_1to550.fa.fai +1 -0
  421. data/spec/data/explore/1/2seqs.sammy.fa +12004 -0
  422. data/spec/data/explore/1/HOWTO_RECREATE.txt +6 -0
  423. data/spec/data/explore/1/a.fa +2 -0
  424. data/spec/data/explore/1/seq1_and_a.fa +3 -0
  425. data/spec/data/explore/1/seq2.fa +2 -0
  426. data/spec/data/fluff/1/2seqs.sammy.fa +12004 -0
  427. data/spec/data/fluff/1/HOWTO_RECREATE.txt +5 -0
  428. data/spec/data/fluff/1/seq1.fa +2 -0
  429. data/spec/data/fluff/1/seq2.fa +2 -0
  430. data/spec/data/gapfilling/1/reads.fa +171 -0
  431. data/spec/data/gapfilling/1/trail_with_Ns.fa +5 -0
  432. data/spec/data/gapfilling/1/velvetAssembly/Graph2 +130 -0
  433. data/spec/data/gapfilling/1/velvetAssembly/LastGraph +130 -0
  434. data/spec/data/gapfilling/1/velvetAssembly/Log +199 -0
  435. data/spec/data/gapfilling/1/velvetAssembly/PreGraph +7 -0
  436. data/spec/data/gapfilling/1/velvetAssembly/Roadmaps +239 -0
  437. data/spec/data/gapfilling/1/velvetAssembly/Sequences +281 -0
  438. data/spec/data/gapfilling/1/velvetAssembly/contigs.fa +12 -0
  439. data/spec/data/gapfilling/1/velvetAssembly/stats.txt +4 -0
  440. data/spec/data/gapfilling/2/HOWTO_recreate +17 -0
  441. data/spec/data/gapfilling/2/reference.fa +2 -0
  442. data/spec/data/gapfilling/2/reference_part1.fa +4 -0
  443. data/spec/data/gapfilling/2/reference_part2.fa +4 -0
  444. data/spec/data/gapfilling/2/sammy_reads.fa.gz +0 -0
  445. data/spec/data/gapfilling/2/with_gaps.fa +4 -0
  446. data/spec/data/gapfilling/3/HOWTO_recreate +4 -0
  447. data/spec/data/gapfilling/3/reads.fa.gz +0 -0
  448. data/spec/data/gapfilling/3/reference_part1.fa +4 -0
  449. data/spec/data/gapfilling/3/reference_part2.fa +4 -0
  450. data/spec/data/gapfilling/3/with_gaps.fa +4 -0
  451. data/spec/data/gapfilling/4/HOWTO_recreate +1 -0
  452. data/spec/data/gapfilling/4/reads.fa.gz +0 -0
  453. data/spec/data/gapfilling/5/HOWTO_RECREATE +7 -0
  454. data/spec/data/gapfilling/5/answer.fna +2 -0
  455. data/spec/data/gapfilling/5/gappy.fna +2 -0
  456. data/spec/data/gapfilling/5/reads.fa +17961 -0
  457. data/spec/data/gapfilling/5/velvet51_3.5/LastGraph +8337 -0
  458. data/spec/data/gapfilling/5/velvet51_3.5/Sequences +20921 -0
  459. data/spec/data/gapfilling/6/random1.fa +28 -0
  460. data/spec/data/gapfilling/6/random2.fa +28 -0
  461. data/spec/data/gapfilling/6/random_sequence_length_2000 +0 -0
  462. data/spec/data/gapfilling/6/reads.random1.fa.gz +0 -0
  463. data/spec/data/gapfilling/6/reads.random2.fa.gz +0 -0
  464. data/spec/data/gapfilling/6/to_gapfill.fa +22 -0
  465. data/spec/data/kmer_profile_to_assembly/multiple_abundance_file1.csv +2 -0
  466. data/spec/data/kmers_count1.csv +2 -0
  467. data/spec/data/kmers_count2.csv +3 -0
  468. data/spec/data/out +3 -0
  469. data/spec/data/positive_latching_pair.fa +2 -0
  470. data/spec/data/primers.csv +4 -0
  471. data/spec/data/read_selection_by_kmer/blacklist1.txt +1 -0
  472. data/spec/data/read_selection_by_kmer/input.fasta +6 -0
  473. data/spec/data/read_selection_by_kmer/whitelist1.txt +1 -0
  474. data/spec/data/read_selection_by_kmer/whitelist2.txt +2 -0
  475. data/spec/data/read_to_node/1_a_graph/HOWTO_RECREATE.txt +2 -0
  476. data/spec/data/read_to_node/1_a_graph/LastGraph +6695 -0
  477. data/spec/data/read_to_node/1_a_graph/ReadToNode.bin +0 -0
  478. data/spec/data/read_to_node/2_no_read256_or_259/HOWTO_RECREATE.txt +3 -0
  479. data/spec/data/read_to_node/2_no_read256_or_259/LastGraph +6693 -0
  480. data/spec/data/read_to_node/2_no_read256_or_259/ReadToNode.bin +0 -0
  481. data/spec/data/read_to_node/3_no_last_read/LastGraph +6694 -0
  482. data/spec/data/read_to_node/3_no_last_read/ReadToNode.bin +0 -0
  483. data/spec/data/t/details.txt +5 -0
  484. data/spec/data/t/details.txt.srt +5 -0
  485. data/spec/data/t/location.txt +3 -0
  486. data/spec/data/t/location.txt.srt +3 -0
  487. data/spec/data/tweak/1_gap_then_unscaffolded/answer.fa +2 -0
  488. data/spec/data/tweak/1_gap_then_unscaffolded/reads.fa.gz +0 -0
  489. data/spec/data/tweak/1_gap_then_unscaffolded/scaffolds.fa +6 -0
  490. data/spec/data/tweak/2_second_genome/answer2.fa +2 -0
  491. data/spec/data/tweak/2_second_genome/reads.fa.gz +0 -0
  492. data/spec/data/tweak/3_variant/answer.fa +2 -0
  493. data/spec/data/tweak/3_variant/lesser_answer.fa +2 -0
  494. data/spec/data/tweak/3_variant/reads.fa.gz +0 -0
  495. data/spec/data/tweak/3_variant/with_gaps.fa +2 -0
  496. data/spec/data/velvet_test_trails/Assem/Graph +17 -0
  497. data/spec/data/velvet_test_trails/Assem/Graph2 +40 -0
  498. data/spec/data/velvet_test_trails/Assem/LastGraph +40 -0
  499. data/spec/data/velvet_test_trails/Assem/Log +35 -0
  500. data/spec/data/velvet_test_trails/Assem/PreGraph +9 -0
  501. data/spec/data/velvet_test_trails/Assem/Roadmaps +89 -0
  502. data/spec/data/velvet_test_trails/Assem/Sequences +50 -0
  503. data/spec/data/velvet_test_trails/Assem/a.svg +53 -0
  504. data/spec/data/velvet_test_trails/Assem/contigs.fa +15 -0
  505. data/spec/data/velvet_test_trails/Assem/stats.txt +5 -0
  506. data/spec/data/velvet_test_trails/node_fwds.fa +8 -0
  507. data/spec/data/velvet_test_trails/node_seqs.fa +9 -0
  508. data/spec/data/velvet_test_trails/nodes_fwd_rev.fa +16 -0
  509. data/spec/data/velvet_test_trails/read1.fa +2 -0
  510. data/spec/data/velvet_test_trails/reads.fa +50 -0
  511. data/spec/data/velvet_test_trails_reverse/Assem/LastGraph +17 -0
  512. data/spec/data/velvet_test_trails_reverse/Assem/a.svg +53 -0
  513. data/spec/data/velvet_test_trails_reverse/reads_reversed.fa +10 -0
  514. data/spec/data/visualise/1/LastGraph +6695 -0
  515. data/spec/data/visualise/2_paired_end/HOWTO_RECREATE.txt +10 -0
  516. data/spec/data/visualise/2_paired_end/rand1.fa +2 -0
  517. data/spec/data/visualise/2_paired_end/rand2.fa +2 -0
  518. data/spec/data/visualise/2_paired_end/with_gaps.fa +8 -0
  519. data/spec/data/visualise/2_paired_end/with_gaps.read_pairs.fa.gz +0 -0
  520. data/spec/data/wander/1/random1.fa +2 -0
  521. data/spec/data/wander/1/random1.sammy.fa +804 -0
  522. data/spec/depth_first_search_spec.rb +190 -0
  523. data/spec/dijkstra_spec.rb +143 -0
  524. data/spec/explore_spec.rb +29 -0
  525. data/spec/fluffer_spec.rb +155 -0
  526. data/spec/gapfiller_spec.rb +107 -0
  527. data/spec/graph_explorer_spec.rb +475 -0
  528. data/spec/graph_generator_spec.rb +99 -0
  529. data/spec/height_finder_spec.rb +306 -0
  530. data/spec/kmer_abundance_pattern_spec.rb +56 -0
  531. data/spec/kmer_coverage_based_path_filter_spec.rb +73 -0
  532. data/spec/kmer_profile_finder_spec.rb +38 -0
  533. data/spec/kmers_count_tabulate_spec.rb +120 -0
  534. data/spec/oriented_node_trail_spec.rb +221 -0
  535. data/spec/paired_end_neighbours_spec.rb +126 -0
  536. data/spec/paths_between_nodes_spec.rb +349 -0
  537. data/spec/priner_spec.rb +7 -0
  538. data/spec/read_input_spec.rb +23 -0
  539. data/spec/read_selection_by_kmer_spec.rb +166 -0
  540. data/spec/read_to_node_spec.rb +35 -0
  541. data/spec/roundup_spec.rb +366 -0
  542. data/spec/scaffold_breaker_spec.rb +144 -0
  543. data/spec/sequence_spec.rb +43 -0
  544. data/spec/single_coherent_paths_between_nodes_spec.rb +492 -0
  545. data/spec/single_coherent_wanderer_spec.rb +120 -0
  546. data/spec/single_ended_assembler_spec.rb +398 -0
  547. data/spec/spec_helper.rb +310 -0
  548. data/spec/velvet_graph_sequence_extractor_spec.rb +80 -0
  549. data/spec/visualise_spec.rb +105 -0
  550. data/spec/wander_spec.rb +119 -0
  551. data/spec/watch_for_changes.sh +16 -0
  552. data/validation/fasta_compare.rb +72 -0
  553. data/validation/gapfill_simulate_perfect.rb +108 -0
  554. metadata +899 -0
@@ -0,0 +1,448 @@
1
+ /* gzjoin -- command to join gzip files into one gzip file
2
+
3
+ Copyright (C) 2004 Mark Adler, all rights reserved
4
+ version 1.0, 11 Dec 2004
5
+
6
+ This software is provided 'as-is', without any express or implied
7
+ warranty. In no event will the author be held liable for any damages
8
+ arising from the use of this software.
9
+
10
+ Permission is granted to anyone to use this software for any purpose,
11
+ including commercial applications, and to alter it and redistribute it
12
+ freely, subject to the following restrictions:
13
+
14
+ 1. The origin of this software must not be misrepresented; you must not
15
+ claim that you wrote the original software. If you use this software
16
+ in a product, an acknowledgment in the product documentation would be
17
+ appreciated but is not required.
18
+ 2. Altered source versions must be plainly marked as such, and must not be
19
+ misrepresented as being the original software.
20
+ 3. This notice may not be removed or altered from any source distribution.
21
+
22
+ Mark Adler madler@alumni.caltech.edu
23
+ */
24
+
25
+ /*
26
+ * Change history:
27
+ *
28
+ * 1.0 11 Dec 2004 - First version
29
+ * 1.1 12 Jun 2005 - Changed ssize_t to long for portability
30
+ */
31
+
32
+ /*
33
+ gzjoin takes one or more gzip files on the command line and writes out a
34
+ single gzip file that will uncompress to the concatenation of the
35
+ uncompressed data from the individual gzip files. gzjoin does this without
36
+ having to recompress any of the data and without having to calculate a new
37
+ crc32 for the concatenated uncompressed data. gzjoin does however have to
38
+ decompress all of the input data in order to find the bits in the compressed
39
+ data that need to be modified to concatenate the streams.
40
+
41
+ gzjoin does not do an integrity check on the input gzip files other than
42
+ checking the gzip header and decompressing the compressed data. They are
43
+ otherwise assumed to be complete and correct.
44
+
45
+ Each joint between gzip files removes at least 18 bytes of previous trailer
46
+ and subsequent header, and inserts an average of about three bytes to the
47
+ compressed data in order to connect the streams. The output gzip file
48
+ has a minimal ten-byte gzip header with no file name or modification time.
49
+
50
+ This program was written to illustrate the use of the Z_BLOCK option of
51
+ inflate() and the crc32_combine() function. gzjoin will not compile with
52
+ versions of zlib earlier than 1.2.3.
53
+ */
54
+
55
+ #include <stdio.h> /* fputs(), fprintf(), fwrite(), putc() */
56
+ #include <stdlib.h> /* exit(), malloc(), free() */
57
+ #include <fcntl.h> /* open() */
58
+ #include <unistd.h> /* close(), read(), lseek() */
59
+ #include "zlib.h"
60
+ /* crc32(), crc32_combine(), inflateInit2(), inflate(), inflateEnd() */
61
+
62
+ #define local static
63
+
64
+ /* exit with an error (return a value to allow use in an expression) */
65
+ local int bail(char *why1, char *why2)
66
+ {
67
+ fprintf(stderr, "gzjoin error: %s%s, output incomplete\n", why1, why2);
68
+ exit(1);
69
+ return 0;
70
+ }
71
+
72
+ /* -- simple buffered file input with access to the buffer -- */
73
+
74
+ #define CHUNK 32768 /* must be a power of two and fit in unsigned */
75
+
76
+ /* bin buffered input file type */
77
+ typedef struct {
78
+ char *name; /* name of file for error messages */
79
+ int fd; /* file descriptor */
80
+ unsigned left; /* bytes remaining at next */
81
+ unsigned char *next; /* next byte to read */
82
+ unsigned char *buf; /* allocated buffer of length CHUNK */
83
+ } bin;
84
+
85
+ /* close a buffered file and free allocated memory */
86
+ local void bclose(bin *in)
87
+ {
88
+ if (in != NULL) {
89
+ if (in->fd != -1)
90
+ close(in->fd);
91
+ if (in->buf != NULL)
92
+ free(in->buf);
93
+ free(in);
94
+ }
95
+ }
96
+
97
+ /* open a buffered file for input, return a pointer to type bin, or NULL on
98
+ failure */
99
+ local bin *bopen(char *name)
100
+ {
101
+ bin *in;
102
+
103
+ in = malloc(sizeof(bin));
104
+ if (in == NULL)
105
+ return NULL;
106
+ in->buf = malloc(CHUNK);
107
+ in->fd = open(name, O_RDONLY, 0);
108
+ if (in->buf == NULL || in->fd == -1) {
109
+ bclose(in);
110
+ return NULL;
111
+ }
112
+ in->left = 0;
113
+ in->next = in->buf;
114
+ in->name = name;
115
+ return in;
116
+ }
117
+
118
+ /* load buffer from file, return -1 on read error, 0 or 1 on success, with
119
+ 1 indicating that end-of-file was reached */
120
+ local int bload(bin *in)
121
+ {
122
+ long len;
123
+
124
+ if (in == NULL)
125
+ return -1;
126
+ if (in->left != 0)
127
+ return 0;
128
+ in->next = in->buf;
129
+ do {
130
+ len = (long)read(in->fd, in->buf + in->left, CHUNK - in->left);
131
+ if (len < 0)
132
+ return -1;
133
+ in->left += (unsigned)len;
134
+ } while (len != 0 && in->left < CHUNK);
135
+ return len == 0 ? 1 : 0;
136
+ }
137
+
138
+ /* get a byte from the file, bail if end of file */
139
+ #define bget(in) (in->left ? 0 : bload(in), \
140
+ in->left ? (in->left--, *(in->next)++) : \
141
+ bail("unexpected end of file on ", in->name))
142
+
143
+ /* get a four-byte little-endian unsigned integer from file */
144
+ local unsigned long bget4(bin *in)
145
+ {
146
+ unsigned long val;
147
+
148
+ val = bget(in);
149
+ val += (unsigned long)(bget(in)) << 8;
150
+ val += (unsigned long)(bget(in)) << 16;
151
+ val += (unsigned long)(bget(in)) << 24;
152
+ return val;
153
+ }
154
+
155
+ /* skip bytes in file */
156
+ local void bskip(bin *in, unsigned skip)
157
+ {
158
+ /* check pointer */
159
+ if (in == NULL)
160
+ return;
161
+
162
+ /* easy case -- skip bytes in buffer */
163
+ if (skip <= in->left) {
164
+ in->left -= skip;
165
+ in->next += skip;
166
+ return;
167
+ }
168
+
169
+ /* skip what's in buffer, discard buffer contents */
170
+ skip -= in->left;
171
+ in->left = 0;
172
+
173
+ /* seek past multiples of CHUNK bytes */
174
+ if (skip > CHUNK) {
175
+ unsigned left;
176
+
177
+ left = skip & (CHUNK - 1);
178
+ if (left == 0) {
179
+ /* exact number of chunks: seek all the way minus one byte to check
180
+ for end-of-file with a read */
181
+ lseek(in->fd, skip - 1, SEEK_CUR);
182
+ if (read(in->fd, in->buf, 1) != 1)
183
+ bail("unexpected end of file on ", in->name);
184
+ return;
185
+ }
186
+
187
+ /* skip the integral chunks, update skip with remainder */
188
+ lseek(in->fd, skip - left, SEEK_CUR);
189
+ skip = left;
190
+ }
191
+
192
+ /* read more input and skip remainder */
193
+ bload(in);
194
+ if (skip > in->left)
195
+ bail("unexpected end of file on ", in->name);
196
+ in->left -= skip;
197
+ in->next += skip;
198
+ }
199
+
200
+ /* -- end of buffered input functions -- */
201
+
202
+ /* skip the gzip header from file in */
203
+ local void gzhead(bin *in)
204
+ {
205
+ int flags;
206
+
207
+ /* verify gzip magic header and compression method */
208
+ if (bget(in) != 0x1f || bget(in) != 0x8b || bget(in) != 8)
209
+ bail(in->name, " is not a valid gzip file");
210
+
211
+ /* get and verify flags */
212
+ flags = bget(in);
213
+ if ((flags & 0xe0) != 0)
214
+ bail("unknown reserved bits set in ", in->name);
215
+
216
+ /* skip modification time, extra flags, and os */
217
+ bskip(in, 6);
218
+
219
+ /* skip extra field if present */
220
+ if (flags & 4) {
221
+ unsigned len;
222
+
223
+ len = bget(in);
224
+ len += (unsigned)(bget(in)) << 8;
225
+ bskip(in, len);
226
+ }
227
+
228
+ /* skip file name if present */
229
+ if (flags & 8)
230
+ while (bget(in) != 0)
231
+ ;
232
+
233
+ /* skip comment if present */
234
+ if (flags & 16)
235
+ while (bget(in) != 0)
236
+ ;
237
+
238
+ /* skip header crc if present */
239
+ if (flags & 2)
240
+ bskip(in, 2);
241
+ }
242
+
243
+ /* write a four-byte little-endian unsigned integer to out */
244
+ local void put4(unsigned long val, FILE *out)
245
+ {
246
+ putc(val & 0xff, out);
247
+ putc((val >> 8) & 0xff, out);
248
+ putc((val >> 16) & 0xff, out);
249
+ putc((val >> 24) & 0xff, out);
250
+ }
251
+
252
+ /* Load up zlib stream from buffered input, bail if end of file */
253
+ local void zpull(z_streamp strm, bin *in)
254
+ {
255
+ if (in->left == 0)
256
+ bload(in);
257
+ if (in->left == 0)
258
+ bail("unexpected end of file on ", in->name);
259
+ strm->avail_in = in->left;
260
+ strm->next_in = in->next;
261
+ }
262
+
263
+ /* Write header for gzip file to out and initialize trailer. */
264
+ local void gzinit(unsigned long *crc, unsigned long *tot, FILE *out)
265
+ {
266
+ fwrite("\x1f\x8b\x08\0\0\0\0\0\0\xff", 1, 10, out);
267
+ *crc = crc32(0L, Z_NULL, 0);
268
+ *tot = 0;
269
+ }
270
+
271
+ /* Copy the compressed data from name, zeroing the last block bit of the last
272
+ block if clr is true, and adding empty blocks as needed to get to a byte
273
+ boundary. If clr is false, then the last block becomes the last block of
274
+ the output, and the gzip trailer is written. crc and tot maintains the
275
+ crc and length (modulo 2^32) of the output for the trailer. The resulting
276
+ gzip file is written to out. gzinit() must be called before the first call
277
+ of gzcopy() to write the gzip header and to initialize crc and tot. */
278
+ local void gzcopy(char *name, int clr, unsigned long *crc, unsigned long *tot,
279
+ FILE *out)
280
+ {
281
+ int ret; /* return value from zlib functions */
282
+ int pos; /* where the "last block" bit is in byte */
283
+ int last; /* true if processing the last block */
284
+ bin *in; /* buffered input file */
285
+ unsigned char *start; /* start of compressed data in buffer */
286
+ unsigned char *junk; /* buffer for uncompressed data -- discarded */
287
+ z_off_t len; /* length of uncompressed data (support > 4 GB) */
288
+ z_stream strm; /* zlib inflate stream */
289
+
290
+ /* open gzip file and skip header */
291
+ in = bopen(name);
292
+ if (in == NULL)
293
+ bail("could not open ", name);
294
+ gzhead(in);
295
+
296
+ /* allocate buffer for uncompressed data and initialize raw inflate
297
+ stream */
298
+ junk = malloc(CHUNK);
299
+ strm.zalloc = Z_NULL;
300
+ strm.zfree = Z_NULL;
301
+ strm.opaque = Z_NULL;
302
+ strm.avail_in = 0;
303
+ strm.next_in = Z_NULL;
304
+ ret = inflateInit2(&strm, -15);
305
+ if (junk == NULL || ret != Z_OK)
306
+ bail("out of memory", "");
307
+
308
+ /* inflate and copy compressed data, clear last-block bit if requested */
309
+ len = 0;
310
+ zpull(&strm, in);
311
+ start = strm.next_in;
312
+ last = start[0] & 1;
313
+ if (last && clr)
314
+ start[0] &= ~1;
315
+ strm.avail_out = 0;
316
+ for (;;) {
317
+ /* if input used and output done, write used input and get more */
318
+ if (strm.avail_in == 0 && strm.avail_out != 0) {
319
+ fwrite(start, 1, strm.next_in - start, out);
320
+ start = in->buf;
321
+ in->left = 0;
322
+ zpull(&strm, in);
323
+ }
324
+
325
+ /* decompress -- return early when end-of-block reached */
326
+ strm.avail_out = CHUNK;
327
+ strm.next_out = junk;
328
+ ret = inflate(&strm, Z_BLOCK);
329
+ switch (ret) {
330
+ case Z_MEM_ERROR:
331
+ bail("out of memory", "");
332
+ case Z_DATA_ERROR:
333
+ bail("invalid compressed data in ", in->name);
334
+ }
335
+
336
+ /* update length of uncompressed data */
337
+ len += CHUNK - strm.avail_out;
338
+
339
+ /* check for block boundary (only get this when block copied out) */
340
+ if (strm.data_type & 128) {
341
+ /* if that was the last block, then done */
342
+ if (last)
343
+ break;
344
+
345
+ /* number of unused bits in last byte */
346
+ pos = strm.data_type & 7;
347
+
348
+ /* find the next last-block bit */
349
+ if (pos != 0) {
350
+ /* next last-block bit is in last used byte */
351
+ pos = 0x100 >> pos;
352
+ last = strm.next_in[-1] & pos;
353
+ if (last && clr)
354
+ strm.next_in[-1] &= ~pos;
355
+ }
356
+ else {
357
+ /* next last-block bit is in next unused byte */
358
+ if (strm.avail_in == 0) {
359
+ /* don't have that byte yet -- get it */
360
+ fwrite(start, 1, strm.next_in - start, out);
361
+ start = in->buf;
362
+ in->left = 0;
363
+ zpull(&strm, in);
364
+ }
365
+ last = strm.next_in[0] & 1;
366
+ if (last && clr)
367
+ strm.next_in[0] &= ~1;
368
+ }
369
+ }
370
+ }
371
+
372
+ /* update buffer with unused input */
373
+ in->left = strm.avail_in;
374
+ in->next = strm.next_in;
375
+
376
+ /* copy used input, write empty blocks to get to byte boundary */
377
+ pos = strm.data_type & 7;
378
+ fwrite(start, 1, in->next - start - 1, out);
379
+ last = in->next[-1];
380
+ if (pos == 0 || !clr)
381
+ /* already at byte boundary, or last file: write last byte */
382
+ putc(last, out);
383
+ else {
384
+ /* append empty blocks to last byte */
385
+ last &= ((0x100 >> pos) - 1); /* assure unused bits are zero */
386
+ if (pos & 1) {
387
+ /* odd -- append an empty stored block */
388
+ putc(last, out);
389
+ if (pos == 1)
390
+ putc(0, out); /* two more bits in block header */
391
+ fwrite("\0\0\xff\xff", 1, 4, out);
392
+ }
393
+ else {
394
+ /* even -- append 1, 2, or 3 empty fixed blocks */
395
+ switch (pos) {
396
+ case 6:
397
+ putc(last | 8, out);
398
+ last = 0;
399
+ case 4:
400
+ putc(last | 0x20, out);
401
+ last = 0;
402
+ case 2:
403
+ putc(last | 0x80, out);
404
+ putc(0, out);
405
+ }
406
+ }
407
+ }
408
+
409
+ /* update crc and tot */
410
+ *crc = crc32_combine(*crc, bget4(in), len);
411
+ *tot += (unsigned long)len;
412
+
413
+ /* clean up */
414
+ inflateEnd(&strm);
415
+ free(junk);
416
+ bclose(in);
417
+
418
+ /* write trailer if this is the last gzip file */
419
+ if (!clr) {
420
+ put4(*crc, out);
421
+ put4(*tot, out);
422
+ }
423
+ }
424
+
425
+ /* join the gzip files on the command line, write result to stdout */
426
+ int main(int argc, char **argv)
427
+ {
428
+ unsigned long crc, tot; /* running crc and total uncompressed length */
429
+
430
+ /* skip command name */
431
+ argc--;
432
+ argv++;
433
+
434
+ /* show usage if no arguments */
435
+ if (argc == 0) {
436
+ fputs("gzjoin usage: gzjoin f1.gz [f2.gz [f3.gz ...]] > fjoin.gz\n",
437
+ stderr);
438
+ return 0;
439
+ }
440
+
441
+ /* join gzip files on command line and write to stdout */
442
+ gzinit(&crc, &tot, stdout);
443
+ while (argc--)
444
+ gzcopy(*argv++, argc, &crc, &tot, stdout);
445
+
446
+ /* done */
447
+ return 0;
448
+ }
@@ -0,0 +1,413 @@
1
+ /*
2
+ * gzlog.c
3
+ * Copyright (C) 2004 Mark Adler
4
+ * For conditions of distribution and use, see copyright notice in gzlog.h
5
+ * version 1.0, 26 Nov 2004
6
+ *
7
+ */
8
+
9
+ #include <string.h> /* memcmp() */
10
+ #include <stdlib.h> /* malloc(), free(), NULL */
11
+ #include <sys/types.h> /* size_t, off_t */
12
+ #include <unistd.h> /* read(), close(), sleep(), ftruncate(), */
13
+ /* lseek() */
14
+ #include <fcntl.h> /* open() */
15
+ #include <sys/file.h> /* flock() */
16
+ #include "zlib.h" /* deflateInit2(), deflate(), deflateEnd() */
17
+
18
+ #include "gzlog.h" /* interface */
19
+ #define local static
20
+
21
+ /* log object structure */
22
+ typedef struct {
23
+ int id; /* object identifier */
24
+ int fd; /* log file descriptor */
25
+ off_t extra; /* offset of extra "ap" subfield */
26
+ off_t mark_off; /* offset of marked data */
27
+ off_t last_off; /* offset of last block */
28
+ unsigned long crc; /* uncompressed crc */
29
+ unsigned long len; /* uncompressed length (modulo 2^32) */
30
+ unsigned stored; /* length of current stored block */
31
+ } gz_log;
32
+
33
+ #define GZLOGID 19334 /* gz_log object identifier */
34
+
35
+ #define LOCK_RETRY 1 /* retry lock once a second */
36
+ #define LOCK_PATIENCE 1200 /* try about twenty minutes before forcing */
37
+
38
+ /* acquire a lock on a file */
39
+ local int lock(int fd)
40
+ {
41
+ int patience;
42
+
43
+ /* try to lock every LOCK_RETRY seconds for LOCK_PATIENCE seconds */
44
+ patience = LOCK_PATIENCE;
45
+ do {
46
+ if (flock(fd, LOCK_EX + LOCK_NB) == 0)
47
+ return 0;
48
+ (void)sleep(LOCK_RETRY);
49
+ patience -= LOCK_RETRY;
50
+ } while (patience > 0);
51
+
52
+ /* we've run out of patience -- give up */
53
+ return -1;
54
+ }
55
+
56
+ /* release lock */
57
+ local void unlock(int fd)
58
+ {
59
+ (void)flock(fd, LOCK_UN);
60
+ }
61
+
62
+ /* release a log object */
63
+ local void log_clean(gz_log *log)
64
+ {
65
+ unlock(log->fd);
66
+ (void)close(log->fd);
67
+ free(log);
68
+ }
69
+
70
+ /* read an unsigned long from a byte buffer little-endian */
71
+ local unsigned long make_ulg(unsigned char *buf)
72
+ {
73
+ int n;
74
+ unsigned long val;
75
+
76
+ val = (unsigned long)(*buf++);
77
+ for (n = 8; n < 32; n += 8)
78
+ val += (unsigned long)(*buf++) << n;
79
+ return val;
80
+ }
81
+
82
+ /* read an off_t from a byte buffer little-endian */
83
+ local off_t make_off(unsigned char *buf)
84
+ {
85
+ int n;
86
+ off_t val;
87
+
88
+ val = (off_t)(*buf++);
89
+ for (n = 8; n < 64; n += 8)
90
+ val += (off_t)(*buf++) << n;
91
+ return val;
92
+ }
93
+
94
+ /* write an unsigned long little-endian to byte buffer */
95
+ local void dice_ulg(unsigned long val, unsigned char *buf)
96
+ {
97
+ int n;
98
+
99
+ for (n = 0; n < 4; n++) {
100
+ *buf++ = val & 0xff;
101
+ val >>= 8;
102
+ }
103
+ }
104
+
105
+ /* write an off_t little-endian to byte buffer */
106
+ local void dice_off(off_t val, unsigned char *buf)
107
+ {
108
+ int n;
109
+
110
+ for (n = 0; n < 8; n++) {
111
+ *buf++ = val & 0xff;
112
+ val >>= 8;
113
+ }
114
+ }
115
+
116
+ /* initial, empty gzip file for appending */
117
+ local char empty_gz[] = {
118
+ 0x1f, 0x8b, /* magic gzip id */
119
+ 8, /* compression method is deflate */
120
+ 4, /* there is an extra field */
121
+ 0, 0, 0, 0, /* no modification time provided */
122
+ 0, 0xff, /* no extra flags, no OS */
123
+ 20, 0, 'a', 'p', 16, 0, /* extra field with "ap" subfield */
124
+ 32, 0, 0, 0, 0, 0, 0, 0, /* offset of uncompressed data */
125
+ 32, 0, 0, 0, 0, 0, 0, 0, /* offset of last block */
126
+ 1, 0, 0, 0xff, 0xff, /* empty stored block (last) */
127
+ 0, 0, 0, 0, /* crc */
128
+ 0, 0, 0, 0 /* uncompressed length */
129
+ };
130
+
131
+ /* initialize a log object with locking */
132
+ void *gzlog_open(char *path)
133
+ {
134
+ unsigned xlen;
135
+ unsigned char temp[20];
136
+ unsigned sub_len;
137
+ int good;
138
+ gz_log *log;
139
+
140
+ /* allocate log structure */
141
+ log = malloc(sizeof(gz_log));
142
+ if (log == NULL)
143
+ return NULL;
144
+ log->id = GZLOGID;
145
+
146
+ /* open file, creating it if necessary, and locking it */
147
+ log->fd = open(path, O_RDWR | O_CREAT, 0600);
148
+ if (log->fd < 0) {
149
+ free(log);
150
+ return NULL;
151
+ }
152
+ if (lock(log->fd)) {
153
+ close(log->fd);
154
+ free(log);
155
+ return NULL;
156
+ }
157
+
158
+ /* if file is empty, write new gzip stream */
159
+ if (lseek(log->fd, 0, SEEK_END) == 0) {
160
+ if (write(log->fd, empty_gz, sizeof(empty_gz)) != sizeof(empty_gz)) {
161
+ log_clean(log);
162
+ return NULL;
163
+ }
164
+ }
165
+
166
+ /* check gzip header */
167
+ (void)lseek(log->fd, 0, SEEK_SET);
168
+ if (read(log->fd, temp, 12) != 12 || temp[0] != 0x1f ||
169
+ temp[1] != 0x8b || temp[2] != 8 || (temp[3] & 4) == 0) {
170
+ log_clean(log);
171
+ return NULL;
172
+ }
173
+
174
+ /* process extra field to find "ap" sub-field */
175
+ xlen = temp[10] + (temp[11] << 8);
176
+ good = 0;
177
+ while (xlen) {
178
+ if (xlen < 4 || read(log->fd, temp, 4) != 4)
179
+ break;
180
+ sub_len = temp[2];
181
+ sub_len += temp[3] << 8;
182
+ xlen -= 4;
183
+ if (memcmp(temp, "ap", 2) == 0 && sub_len == 16) {
184
+ good = 1;
185
+ break;
186
+ }
187
+ if (xlen < sub_len)
188
+ break;
189
+ (void)lseek(log->fd, sub_len, SEEK_CUR);
190
+ xlen -= sub_len;
191
+ }
192
+ if (!good) {
193
+ log_clean(log);
194
+ return NULL;
195
+ }
196
+
197
+ /* read in "ap" sub-field */
198
+ log->extra = lseek(log->fd, 0, SEEK_CUR);
199
+ if (read(log->fd, temp, 16) != 16) {
200
+ log_clean(log);
201
+ return NULL;
202
+ }
203
+ log->mark_off = make_off(temp);
204
+ log->last_off = make_off(temp + 8);
205
+
206
+ /* get crc, length of gzip file */
207
+ (void)lseek(log->fd, log->last_off, SEEK_SET);
208
+ if (read(log->fd, temp, 13) != 13 ||
209
+ memcmp(temp, "\001\000\000\377\377", 5) != 0) {
210
+ log_clean(log);
211
+ return NULL;
212
+ }
213
+ log->crc = make_ulg(temp + 5);
214
+ log->len = make_ulg(temp + 9);
215
+
216
+ /* set up to write over empty last block */
217
+ (void)lseek(log->fd, log->last_off + 5, SEEK_SET);
218
+ log->stored = 0;
219
+ return (void *)log;
220
+ }
221
+
222
+ /* maximum amount to put in a stored block before starting a new one */
223
+ #define MAX_BLOCK 16384
224
+
225
+ /* write a block to a log object */
226
+ int gzlog_write(void *obj, char *data, size_t len)
227
+ {
228
+ size_t some;
229
+ unsigned char temp[5];
230
+ gz_log *log;
231
+
232
+ /* check object */
233
+ log = (gz_log *)obj;
234
+ if (log == NULL || log->id != GZLOGID)
235
+ return 1;
236
+
237
+ /* write stored blocks until all of the input is written */
238
+ do {
239
+ some = MAX_BLOCK - log->stored;
240
+ if (some > len)
241
+ some = len;
242
+ if (write(log->fd, data, some) != some)
243
+ return 1;
244
+ log->crc = crc32(log->crc, data, some);
245
+ log->len += some;
246
+ len -= some;
247
+ data += some;
248
+ log->stored += some;
249
+
250
+ /* if the stored block is full, end it and start another */
251
+ if (log->stored == MAX_BLOCK) {
252
+ (void)lseek(log->fd, log->last_off, SEEK_SET);
253
+ temp[0] = 0;
254
+ dice_ulg(log->stored + ((unsigned long)(~log->stored) << 16),
255
+ temp + 1);
256
+ if (write(log->fd, temp, 5) != 5)
257
+ return 1;
258
+ log->last_off = lseek(log->fd, log->stored, SEEK_CUR);
259
+ (void)lseek(log->fd, 5, SEEK_CUR);
260
+ log->stored = 0;
261
+ }
262
+ } while (len);
263
+ return 0;
264
+ }
265
+
266
+ /* recompress the remaining stored deflate data in place */
267
+ local int recomp(gz_log *log)
268
+ {
269
+ z_stream strm;
270
+ size_t len, max;
271
+ unsigned char *in;
272
+ unsigned char *out;
273
+ unsigned char temp[16];
274
+
275
+ /* allocate space and read it all in (it's around 1 MB) */
276
+ len = log->last_off - log->mark_off;
277
+ max = len + (len >> 12) + (len >> 14) + 11;
278
+ out = malloc(max);
279
+ if (out == NULL)
280
+ return 1;
281
+ in = malloc(len);
282
+ if (in == NULL) {
283
+ free(out);
284
+ return 1;
285
+ }
286
+ (void)lseek(log->fd, log->mark_off, SEEK_SET);
287
+ if (read(log->fd, in, len) != len) {
288
+ free(in);
289
+ free(out);
290
+ return 1;
291
+ }
292
+
293
+ /* recompress in memory, decoding stored data as we go */
294
+ /* note: this assumes that unsigned is four bytes or more */
295
+ /* consider not making that assumption */
296
+ strm.zalloc = Z_NULL;
297
+ strm.zfree = Z_NULL;
298
+ strm.opaque = Z_NULL;
299
+ if (deflateInit2(&strm, Z_BEST_COMPRESSION, Z_DEFLATED, -15, 8,
300
+ Z_DEFAULT_STRATEGY) != Z_OK) {
301
+ free(in);
302
+ free(out);
303
+ return 1;
304
+ }
305
+ strm.next_in = in;
306
+ strm.avail_out = max;
307
+ strm.next_out = out;
308
+ while (len >= 5) {
309
+ if (strm.next_in[0] != 0)
310
+ break;
311
+ strm.avail_in = strm.next_in[1] + (strm.next_in[2] << 8);
312
+ strm.next_in += 5;
313
+ len -= 5;
314
+ if (strm.avail_in != 0) {
315
+ if (len < strm.avail_in)
316
+ break;
317
+ len -= strm.avail_in;
318
+ (void)deflate(&strm, Z_NO_FLUSH);
319
+ if (strm.avail_in != 0 || strm.avail_out == 0)
320
+ break;
321
+ }
322
+ }
323
+ (void)deflate(&strm, Z_SYNC_FLUSH);
324
+ (void)deflateEnd(&strm);
325
+ free(in);
326
+ if (len != 0 || strm.avail_out == 0) {
327
+ free(out);
328
+ return 1;
329
+ }
330
+
331
+ /* overwrite stored data with compressed data */
332
+ (void)lseek(log->fd, log->mark_off, SEEK_SET);
333
+ len = max - strm.avail_out;
334
+ if (write(log->fd, out, len) != len) {
335
+ free(out);
336
+ return 1;
337
+ }
338
+ free(out);
339
+
340
+ /* write last empty block, crc, and length */
341
+ log->mark_off = log->last_off = lseek(log->fd, 0, SEEK_CUR);
342
+ temp[0] = 1;
343
+ dice_ulg(0xffffL << 16, temp + 1);
344
+ dice_ulg(log->crc, temp + 5);
345
+ dice_ulg(log->len, temp + 9);
346
+ if (write(log->fd, temp, 13) != 13)
347
+ return 1;
348
+
349
+ /* truncate file to discard remaining stored data and old trailer */
350
+ ftruncate(log->fd, lseek(log->fd, 0, SEEK_CUR));
351
+
352
+ /* update extra field to point to new last empty block */
353
+ (void)lseek(log->fd, log->extra, SEEK_SET);
354
+ dice_off(log->mark_off, temp);
355
+ dice_off(log->last_off, temp + 8);
356
+ if (write(log->fd, temp, 16) != 16)
357
+ return 1;
358
+ return 0;
359
+ }
360
+
361
+ /* maximum accumulation of stored blocks before compressing */
362
+ #define MAX_STORED 1048576
363
+
364
+ /* close log object */
365
+ int gzlog_close(void *obj)
366
+ {
367
+ unsigned char temp[8];
368
+ gz_log *log;
369
+
370
+ /* check object */
371
+ log = (gz_log *)obj;
372
+ if (log == NULL || log->id != GZLOGID)
373
+ return 1;
374
+
375
+ /* go to start of most recent block being written */
376
+ (void)lseek(log->fd, log->last_off, SEEK_SET);
377
+
378
+ /* if some stuff was put there, update block */
379
+ if (log->stored) {
380
+ temp[0] = 0;
381
+ dice_ulg(log->stored + ((unsigned long)(~log->stored) << 16),
382
+ temp + 1);
383
+ if (write(log->fd, temp, 5) != 5)
384
+ return 1;
385
+ log->last_off = lseek(log->fd, log->stored, SEEK_CUR);
386
+ }
387
+
388
+ /* write last block (empty) */
389
+ if (write(log->fd, "\001\000\000\377\377", 5) != 5)
390
+ return 1;
391
+
392
+ /* write updated crc and uncompressed length */
393
+ dice_ulg(log->crc, temp);
394
+ dice_ulg(log->len, temp + 4);
395
+ if (write(log->fd, temp, 8) != 8)
396
+ return 1;
397
+
398
+ /* put offset of that last block in gzip extra block */
399
+ (void)lseek(log->fd, log->extra + 8, SEEK_SET);
400
+ dice_off(log->last_off, temp);
401
+ if (write(log->fd, temp, 8) != 8)
402
+ return 1;
403
+
404
+ /* if more than 1 MB stored, then time to compress it */
405
+ if (log->last_off - log->mark_off > MAX_STORED) {
406
+ if (recomp(log))
407
+ return 1;
408
+ }
409
+
410
+ /* unlock and close file */
411
+ log_clean(log);
412
+ return 0;
413
+ }