ooxml_crypt 0.1.0

Sign up to get free protection for your applications and to get access to all the features.
Files changed (264) hide show
  1. checksums.yaml +7 -0
  2. data/Gemfile +4 -0
  3. data/LICENSE.txt +21 -0
  4. data/README.md +58 -0
  5. data/Rakefile +12 -0
  6. data/bin/console +15 -0
  7. data/bin/setup +8 -0
  8. data/ext/ooxml_crypt/extconf.rb +18 -0
  9. data/ext/ooxml_crypt/ooxml_crypt.c +27 -0
  10. data/ext/ooxml_crypt/ooxml_crypt.h +7 -0
  11. data/lib/ooxml_crypt/version.rb +5 -0
  12. data/lib/ooxml_crypt.rb +75 -0
  13. data/vendor/cybozulib/.github/workflows/main.yml +12 -0
  14. data/vendor/cybozulib/.gitignore +5 -0
  15. data/vendor/cybozulib/CMakeLists.txt +6 -0
  16. data/vendor/cybozulib/COPYRIGHT +27 -0
  17. data/vendor/cybozulib/Makefile +26 -0
  18. data/vendor/cybozulib/bin/libeay32.dll +0 -0
  19. data/vendor/cybozulib/bin/libmecab.dll +0 -0
  20. data/vendor/cybozulib/bin/ssleay32.dll +0 -0
  21. data/vendor/cybozulib/common.mk +116 -0
  22. data/vendor/cybozulib/common.props +25 -0
  23. data/vendor/cybozulib/cybozulib.sln +286 -0
  24. data/vendor/cybozulib/debug.props +14 -0
  25. data/vendor/cybozulib/include/cybozu/array.hpp +197 -0
  26. data/vendor/cybozulib/include/cybozu/atoi.hpp +238 -0
  27. data/vendor/cybozulib/include/cybozu/atomic.hpp +146 -0
  28. data/vendor/cybozulib/include/cybozu/base64.hpp +210 -0
  29. data/vendor/cybozulib/include/cybozu/benchmark.hpp +212 -0
  30. data/vendor/cybozulib/include/cybozu/bfd.hpp +105 -0
  31. data/vendor/cybozulib/include/cybozu/bit_operation.hpp +139 -0
  32. data/vendor/cybozulib/include/cybozu/bitvector.hpp +358 -0
  33. data/vendor/cybozulib/include/cybozu/condition_variable.hpp +113 -0
  34. data/vendor/cybozulib/include/cybozu/condition_variable_cs.hpp +74 -0
  35. data/vendor/cybozulib/include/cybozu/config.hpp +392 -0
  36. data/vendor/cybozulib/include/cybozu/critical_section.hpp +60 -0
  37. data/vendor/cybozulib/include/cybozu/crypto.hpp +321 -0
  38. data/vendor/cybozulib/include/cybozu/csucvector.hpp +624 -0
  39. data/vendor/cybozulib/include/cybozu/csv.hpp +294 -0
  40. data/vendor/cybozulib/include/cybozu/data_type.hpp +27 -0
  41. data/vendor/cybozulib/include/cybozu/endian.hpp +224 -0
  42. data/vendor/cybozulib/include/cybozu/env.hpp +63 -0
  43. data/vendor/cybozulib/include/cybozu/event.hpp +122 -0
  44. data/vendor/cybozulib/include/cybozu/exception.hpp +253 -0
  45. data/vendor/cybozulib/include/cybozu/file.hpp +626 -0
  46. data/vendor/cybozulib/include/cybozu/fmindex.hpp +291 -0
  47. data/vendor/cybozulib/include/cybozu/format.hpp +93 -0
  48. data/vendor/cybozulib/include/cybozu/frequency.hpp +264 -0
  49. data/vendor/cybozulib/include/cybozu/hash.hpp +67 -0
  50. data/vendor/cybozulib/include/cybozu/inttype.hpp +174 -0
  51. data/vendor/cybozulib/include/cybozu/itoa.hpp +336 -0
  52. data/vendor/cybozulib/include/cybozu/json.hpp +120 -0
  53. data/vendor/cybozulib/include/cybozu/line_stream.hpp +149 -0
  54. data/vendor/cybozulib/include/cybozu/link_libeay32.hpp +21 -0
  55. data/vendor/cybozulib/include/cybozu/link_mpir.hpp +18 -0
  56. data/vendor/cybozulib/include/cybozu/link_ssleay32.hpp +19 -0
  57. data/vendor/cybozulib/include/cybozu/log.hpp +237 -0
  58. data/vendor/cybozulib/include/cybozu/minixml.hpp +452 -0
  59. data/vendor/cybozulib/include/cybozu/mmap.hpp +143 -0
  60. data/vendor/cybozulib/include/cybozu/mutex.hpp +144 -0
  61. data/vendor/cybozulib/include/cybozu/nlp/mecab.hpp +96 -0
  62. data/vendor/cybozulib/include/cybozu/nlp/plsi.hpp +315 -0
  63. data/vendor/cybozulib/include/cybozu/nlp/random.hpp +74 -0
  64. data/vendor/cybozulib/include/cybozu/nlp/sparse.hpp +529 -0
  65. data/vendor/cybozulib/include/cybozu/nlp/svd.hpp +486 -0
  66. data/vendor/cybozulib/include/cybozu/nlp/tfidf.hpp +226 -0
  67. data/vendor/cybozulib/include/cybozu/nlp/top_score.hpp +75 -0
  68. data/vendor/cybozulib/include/cybozu/option.hpp +743 -0
  69. data/vendor/cybozulib/include/cybozu/parallel.hpp +88 -0
  70. data/vendor/cybozulib/include/cybozu/pcg.hpp +72 -0
  71. data/vendor/cybozulib/include/cybozu/process.hpp +324 -0
  72. data/vendor/cybozulib/include/cybozu/quit_signal_handler.hpp +66 -0
  73. data/vendor/cybozulib/include/cybozu/random_generator.hpp +144 -0
  74. data/vendor/cybozulib/include/cybozu/regex.hpp +463 -0
  75. data/vendor/cybozulib/include/cybozu/select8.hpp +279 -0
  76. data/vendor/cybozulib/include/cybozu/serializer.hpp +363 -0
  77. data/vendor/cybozulib/include/cybozu/sha1.hpp +209 -0
  78. data/vendor/cybozulib/include/cybozu/sha2.hpp +506 -0
  79. data/vendor/cybozulib/include/cybozu/siphash.hpp +105 -0
  80. data/vendor/cybozulib/include/cybozu/socket.hpp +785 -0
  81. data/vendor/cybozulib/include/cybozu/ssl.hpp +203 -0
  82. data/vendor/cybozulib/include/cybozu/stacktrace.hpp +291 -0
  83. data/vendor/cybozulib/include/cybozu/stream.hpp +269 -0
  84. data/vendor/cybozulib/include/cybozu/string.hpp +1746 -0
  85. data/vendor/cybozulib/include/cybozu/string_operation.hpp +365 -0
  86. data/vendor/cybozulib/include/cybozu/sucvector.hpp +378 -0
  87. data/vendor/cybozulib/include/cybozu/test.hpp +373 -0
  88. data/vendor/cybozulib/include/cybozu/thread.hpp +229 -0
  89. data/vendor/cybozulib/include/cybozu/time.hpp +281 -0
  90. data/vendor/cybozulib/include/cybozu/tls.hpp +115 -0
  91. data/vendor/cybozulib/include/cybozu/unordered_map.hpp +13 -0
  92. data/vendor/cybozulib/include/cybozu/unordered_set.hpp +13 -0
  93. data/vendor/cybozulib/include/cybozu/v128.hpp +376 -0
  94. data/vendor/cybozulib/include/cybozu/wavelet_matrix.hpp +345 -0
  95. data/vendor/cybozulib/include/cybozu/xorshift.hpp +189 -0
  96. data/vendor/cybozulib/include/cybozu/zlib.hpp +325 -0
  97. data/vendor/cybozulib/include/sais.hxx +364 -0
  98. data/vendor/cybozulib/misc/make_select8tbl.cpp +26 -0
  99. data/vendor/cybozulib/mk.bat +37 -0
  100. data/vendor/cybozulib/readme.md +29 -0
  101. data/vendor/cybozulib/release.props +12 -0
  102. data/vendor/cybozulib/sample/Makefile +30 -0
  103. data/vendor/cybozulib/sample/csucvector_smpl.cpp +42 -0
  104. data/vendor/cybozulib/sample/data/svd/org/test1.S +4 -0
  105. data/vendor/cybozulib/sample/data/svd/org/test1.U +4 -0
  106. data/vendor/cybozulib/sample/data/svd/org/test1.V +6 -0
  107. data/vendor/cybozulib/sample/data/svd/test1 +4 -0
  108. data/vendor/cybozulib/sample/data/svd/test2 +4 -0
  109. data/vendor/cybozulib/sample/desymbol.cpp +127 -0
  110. data/vendor/cybozulib/sample/exception_smpl.cpp +46 -0
  111. data/vendor/cybozulib/sample/fmindex_smpl.cpp +231 -0
  112. data/vendor/cybozulib/sample/log_smpl.cpp +19 -0
  113. data/vendor/cybozulib/sample/mecab_smpl.cpp +37 -0
  114. data/vendor/cybozulib/sample/option2_smpl.cpp +68 -0
  115. data/vendor/cybozulib/sample/option_smpl.cpp +42 -0
  116. data/vendor/cybozulib/sample/plsi_smpl.cpp +207 -0
  117. data/vendor/cybozulib/sample/proj/exception_smpl.vcproj +184 -0
  118. data/vendor/cybozulib/sample/proj/mecab_smpl.vcproj +184 -0
  119. data/vendor/cybozulib/sample/proj/ssl_smpl/ssl_smpl.vcxproj +85 -0
  120. data/vendor/cybozulib/sample/proj/ssl_smpl.vcproj +347 -0
  121. data/vendor/cybozulib/sample/proj/stacktrace_smpl/stacktrace_smpl.vcxproj +85 -0
  122. data/vendor/cybozulib/sample/proj/svd_smpl.vcproj +184 -0
  123. data/vendor/cybozulib/sample/quit_signal_handler.cpp +30 -0
  124. data/vendor/cybozulib/sample/serializer_smpl.cpp +196 -0
  125. data/vendor/cybozulib/sample/socket_smpl.cpp +82 -0
  126. data/vendor/cybozulib/sample/ssl_smpl.cpp +39 -0
  127. data/vendor/cybozulib/sample/stacktrace_smpl.cpp +52 -0
  128. data/vendor/cybozulib/sample/svd_bench_smpl.cpp +143 -0
  129. data/vendor/cybozulib/sample/svd_smpl.cpp +94 -0
  130. data/vendor/cybozulib/sample/wm_bench_smpl.cpp +182 -0
  131. data/vendor/cybozulib/sample/zlib_smpl.cpp +41 -0
  132. data/vendor/cybozulib/src/Makefile +8 -0
  133. data/vendor/cybozulib/src/base/Makefile +19 -0
  134. data/vendor/cybozulib/test/Makefile +12 -0
  135. data/vendor/cybozulib/test/base/Makefile +37 -0
  136. data/vendor/cybozulib/test/base/array_test.cpp +173 -0
  137. data/vendor/cybozulib/test/base/atoi_test.cpp +774 -0
  138. data/vendor/cybozulib/test/base/atomic_test.cpp +49 -0
  139. data/vendor/cybozulib/test/base/base64_test.cpp +113 -0
  140. data/vendor/cybozulib/test/base/bit_operation_test.cpp +134 -0
  141. data/vendor/cybozulib/test/base/bitvector_test.cpp +204 -0
  142. data/vendor/cybozulib/test/base/condition_variable_cs_test.cpp +92 -0
  143. data/vendor/cybozulib/test/base/condition_variable_test.cpp +88 -0
  144. data/vendor/cybozulib/test/base/config_test.cpp +236 -0
  145. data/vendor/cybozulib/test/base/crypto_test.cpp +122 -0
  146. data/vendor/cybozulib/test/base/csucvector_test.cpp +63 -0
  147. data/vendor/cybozulib/test/base/csv_test.cpp +182 -0
  148. data/vendor/cybozulib/test/base/data/a.xml +26 -0
  149. data/vendor/cybozulib/test/base/endian_test.cpp +56 -0
  150. data/vendor/cybozulib/test/base/env_test.cpp +22 -0
  151. data/vendor/cybozulib/test/base/event_test.cpp +41 -0
  152. data/vendor/cybozulib/test/base/file_test.cpp +233 -0
  153. data/vendor/cybozulib/test/base/fmindex_test.cpp +118 -0
  154. data/vendor/cybozulib/test/base/format_test.cpp +12 -0
  155. data/vendor/cybozulib/test/base/frequency_test.cpp +104 -0
  156. data/vendor/cybozulib/test/base/itoa_test.cpp +522 -0
  157. data/vendor/cybozulib/test/base/line_stream_test.cpp +208 -0
  158. data/vendor/cybozulib/test/base/mecab_test.cpp +41 -0
  159. data/vendor/cybozulib/test/base/minixml_test.cpp +103 -0
  160. data/vendor/cybozulib/test/base/mmap_test.cpp +15 -0
  161. data/vendor/cybozulib/test/base/option_test.cpp +487 -0
  162. data/vendor/cybozulib/test/base/parallel_test.cpp +48 -0
  163. data/vendor/cybozulib/test/base/proj/array_test/array_test.vcxproj +86 -0
  164. data/vendor/cybozulib/test/base/proj/atoi_test/atoi_test.vcxproj +86 -0
  165. data/vendor/cybozulib/test/base/proj/atomic_test/atomic_test.vcxproj +86 -0
  166. data/vendor/cybozulib/test/base/proj/base64_test/base64_test.vcxproj +86 -0
  167. data/vendor/cybozulib/test/base/proj/condition_variable_cs_test/condition_variable_cs_test.vcxproj +86 -0
  168. data/vendor/cybozulib/test/base/proj/condition_variable_test/condition_variable_test.vcxproj +86 -0
  169. data/vendor/cybozulib/test/base/proj/config_test/config_test.vcxproj +86 -0
  170. data/vendor/cybozulib/test/base/proj/csv_test/csv_test.vcxproj +86 -0
  171. data/vendor/cybozulib/test/base/proj/endian_test/endian_test.vcxproj +86 -0
  172. data/vendor/cybozulib/test/base/proj/env_test/env_test.vcxproj +86 -0
  173. data/vendor/cybozulib/test/base/proj/event_test/event_test.vcxproj +86 -0
  174. data/vendor/cybozulib/test/base/proj/file_test/file_test.vcxproj +86 -0
  175. data/vendor/cybozulib/test/base/proj/itoa_test/itoa_test.vcxproj +86 -0
  176. data/vendor/cybozulib/test/base/proj/mecab_test/mecab_test.vcxproj +88 -0
  177. data/vendor/cybozulib/test/base/proj/minixml_test/minixml_test.vcxproj +86 -0
  178. data/vendor/cybozulib/test/base/proj/mmap_test/mmap_test.vcxproj +86 -0
  179. data/vendor/cybozulib/test/base/proj/serializer_test/serializer_test.vcxproj +86 -0
  180. data/vendor/cybozulib/test/base/proj/sha1_test/sha1_test.vcxproj +86 -0
  181. data/vendor/cybozulib/test/base/proj/stream_test/stream_test.vcxproj +86 -0
  182. data/vendor/cybozulib/test/base/proj/string_operation_test/string_operation_test.vcxproj +86 -0
  183. data/vendor/cybozulib/test/base/proj/string_test/string_test.vcxproj +86 -0
  184. data/vendor/cybozulib/test/base/proj/thread_test/thread_test.vcxproj +86 -0
  185. data/vendor/cybozulib/test/base/proj/time_test/time_test.vcxproj +86 -0
  186. data/vendor/cybozulib/test/base/proj/tls_test/tls_test.vcxproj +86 -0
  187. data/vendor/cybozulib/test/base/proj/zlib_test/zlib_test.vcxproj +86 -0
  188. data/vendor/cybozulib/test/base/random_generator_test.cpp +28 -0
  189. data/vendor/cybozulib/test/base/regex_test.cpp +74 -0
  190. data/vendor/cybozulib/test/base/serializer_test.cpp +483 -0
  191. data/vendor/cybozulib/test/base/sha1_test.cpp +61 -0
  192. data/vendor/cybozulib/test/base/sha2_test.cpp +191 -0
  193. data/vendor/cybozulib/test/base/siphash_test.cpp +33 -0
  194. data/vendor/cybozulib/test/base/socket_test.cpp +76 -0
  195. data/vendor/cybozulib/test/base/stream_test.cpp +101 -0
  196. data/vendor/cybozulib/test/base/string_operation_test.cpp +340 -0
  197. data/vendor/cybozulib/test/base/string_test.cpp +1705 -0
  198. data/vendor/cybozulib/test/base/sucvector_test.cpp +312 -0
  199. data/vendor/cybozulib/test/base/thread_test.cpp +62 -0
  200. data/vendor/cybozulib/test/base/time_test.cpp +164 -0
  201. data/vendor/cybozulib/test/base/tls_test.cpp +50 -0
  202. data/vendor/cybozulib/test/base/wavelet_matrix_test.cpp +145 -0
  203. data/vendor/cybozulib/test/base/zlib_test.cpp +371 -0
  204. data/vendor/cybozulib/test/nlp/Makefile +27 -0
  205. data/vendor/cybozulib/test/nlp/proj/random_test.vcproj +184 -0
  206. data/vendor/cybozulib/test/nlp/proj/sparse_test.vcproj +184 -0
  207. data/vendor/cybozulib/test/nlp/proj/svd_test.vcproj +184 -0
  208. data/vendor/cybozulib/test/nlp/random_test.cpp +62 -0
  209. data/vendor/cybozulib/test/nlp/sparse_test.cpp +347 -0
  210. data/vendor/cybozulib/test/nlp/svd_test.cpp +234 -0
  211. data/vendor/cybozulib/test/nlp/top_score_test.cpp +40 -0
  212. data/vendor/cybozulib/tool/create_vcproj.py +186 -0
  213. data/vendor/cybozulib/tool/vcproj_tmpl.py +185 -0
  214. data/vendor/msoffice/COPYRIGHT +27 -0
  215. data/vendor/msoffice/Makefile +29 -0
  216. data/vendor/msoffice/bin/64/msoc.dll +0 -0
  217. data/vendor/msoffice/bin/64/msocsample.exe +0 -0
  218. data/vendor/msoffice/bin/64/msoffice-crypt.exe +0 -0
  219. data/vendor/msoffice/bin/msoc.dll +0 -0
  220. data/vendor/msoffice/bin/msocsample.exe +0 -0
  221. data/vendor/msoffice/bin/msoffice-crypt.exe +0 -0
  222. data/vendor/msoffice/common.mk +71 -0
  223. data/vendor/msoffice/common.props +26 -0
  224. data/vendor/msoffice/debug.props +14 -0
  225. data/vendor/msoffice/include/attack.hpp +211 -0
  226. data/vendor/msoffice/include/cfb.hpp +777 -0
  227. data/vendor/msoffice/include/crypto_util.hpp +450 -0
  228. data/vendor/msoffice/include/custom_sha1.hpp +342 -0
  229. data/vendor/msoffice/include/decode.hpp +240 -0
  230. data/vendor/msoffice/include/encode.hpp +221 -0
  231. data/vendor/msoffice/include/make_dataspace.hpp +316 -0
  232. data/vendor/msoffice/include/msoc.h +129 -0
  233. data/vendor/msoffice/include/resource.hpp +7 -0
  234. data/vendor/msoffice/include/standard_encryption.hpp +145 -0
  235. data/vendor/msoffice/include/uint32vec.hpp +179 -0
  236. data/vendor/msoffice/include/util.hpp +212 -0
  237. data/vendor/msoffice/lib/.emptydir +0 -0
  238. data/vendor/msoffice/misc/decrypt-xls.vbs +46 -0
  239. data/vendor/msoffice/mk.bat +1 -0
  240. data/vendor/msoffice/mkdll.bat +3 -0
  241. data/vendor/msoffice/msoc.def +13 -0
  242. data/vendor/msoffice/msocsample.py +178 -0
  243. data/vendor/msoffice/msoffice12.sln +31 -0
  244. data/vendor/msoffice/readme.md +110 -0
  245. data/vendor/msoffice/release.props +28 -0
  246. data/vendor/msoffice/src/Makefile +19 -0
  247. data/vendor/msoffice/src/attack.cpp +124 -0
  248. data/vendor/msoffice/src/cfb_test.cpp +77 -0
  249. data/vendor/msoffice/src/minisample.c +54 -0
  250. data/vendor/msoffice/src/msocdll.cpp +276 -0
  251. data/vendor/msoffice/src/msocsample.c +136 -0
  252. data/vendor/msoffice/src/msoffice-crypt.cpp +219 -0
  253. data/vendor/msoffice/src/proj/attack/attack.vcxproj +88 -0
  254. data/vendor/msoffice/src/proj/main/msoffice-crypt.vcxproj +88 -0
  255. data/vendor/msoffice/src/sha1.cpp +234 -0
  256. data/vendor/msoffice/test/Makefile +20 -0
  257. data/vendor/msoffice/test/cfb_test.cpp +74 -0
  258. data/vendor/msoffice/test/hash_test.cpp +59 -0
  259. data/vendor/msoffice/test/proj/cfb/cfb_test.vcxproj +90 -0
  260. data/vendor/msoffice/test/proj/hash/hash_test.vcxproj +90 -0
  261. data/vendor/msoffice/test/sampl.bat +8 -0
  262. data/vendor/msoffice/test_all.py +46 -0
  263. data/vendor/update +4 -0
  264. metadata +351 -0
@@ -0,0 +1,452 @@
1
+ #pragma once
2
+ /**
3
+ @file
4
+ @brief mini XML parser(not strict)
5
+
6
+ @author MITSUNARI Shigeo(@herumi)
7
+ */
8
+
9
+ #include <assert.h>
10
+ #include <cybozu/exception.hpp>
11
+ #include <map>
12
+ #include <istream>
13
+ #include <list>
14
+
15
+ namespace cybozu {
16
+
17
+ namespace minixml {
18
+
19
+ const int indentWidth = 3;
20
+
21
+ inline std::string escape(const std::string& str)
22
+ {
23
+ std::string ret;
24
+ for (size_t i = 0, n = str.size(); i < n; i++) {
25
+ char c = str[i];
26
+ switch (c) {
27
+ case '&':
28
+ ret += "&amp;";
29
+ break;
30
+ case '<':
31
+ ret += "&lt;";
32
+ break;
33
+ case '>':
34
+ ret += "&gt;";
35
+ break;
36
+ case '\'':
37
+ ret += "&apos;";
38
+ break;
39
+ case '"':
40
+ ret += "&quot;";
41
+ break;
42
+ default:
43
+ ret += c;
44
+ break;
45
+ }
46
+ }
47
+ return ret;
48
+ }
49
+
50
+ inline std::string unescape(const std::string& str)
51
+ {
52
+ std::string ret;
53
+ for (size_t i = 0, n = str.size(); i < n; i++) {
54
+ char c = str[i];
55
+ if (c == '&') {
56
+ if (n - i >= 5 && memcmp(&str[i + 1], "amp;", 4) == 0) {
57
+ ret += '&';
58
+ i += 4;
59
+ } else if (n - i >= 4 && memcmp(&str[i + 1], "lt;", 3) == 0) {
60
+ ret += '<';
61
+ i += 3;
62
+ } else if (n - i >= 4 && memcmp(&str[i + 1], "gt;", 3) == 0) {
63
+ ret += '>';
64
+ i += 3;
65
+ } else if (n - i >= 6 && memcmp(&str[i + 1], "apos;", 5) == 0) {
66
+ ret += '\'';
67
+ i += 5;
68
+ } else if (n - i >= 6 && memcmp(&str[i + 1], "quot;", 5) == 0) {
69
+ ret += '"';
70
+ i += 5;
71
+ } else {
72
+ throw cybozu::Exception("xml:unescape") << str.substr(i, 8);
73
+ }
74
+ } else {
75
+ ret += c;
76
+ }
77
+ }
78
+ return ret;
79
+ }
80
+
81
+ class Attributes {
82
+ typedef std::map<std::string, std::string> Map;
83
+ Map m_;
84
+ public:
85
+ const std::string *query(const std::string& key) const
86
+ {
87
+ Map::const_iterator i = m_.find(key);
88
+ if (i != m_.end()) return &(i->second);
89
+ return 0;
90
+ }
91
+ const std::string& operator[](const std::string& key) const
92
+ {
93
+ const std::string *p = query(key);
94
+ if (p) return *p;
95
+ throw cybozu::Exception("xml:Attributes:no key") << key;
96
+ }
97
+ std::string& operator[](const std::string& key)
98
+ {
99
+ return m_[key];
100
+ }
101
+ void put(int level = 0) const
102
+ {
103
+ std::string tabstr(level + indentWidth, ' ');
104
+ const char *tab = tabstr.c_str();
105
+ for (Map::const_iterator i = m_.begin(), ie = m_.end(); i != ie; ++i) {
106
+ printf("%s%s=%s\n", tab, i->first.c_str(), i->second.c_str());
107
+ }
108
+ }
109
+ void swap(Attributes& attr) CYBOZU_NOEXCEPT
110
+ {
111
+ m_.swap(attr.m_);
112
+ }
113
+ friend inline std::ostream& operator<<(std::ostream& os, const Attributes& attr)
114
+ {
115
+ for (Attributes::Map::const_iterator i = attr.m_.begin(), ie = attr.m_.end(); i != ie; ++i) {
116
+ os << " " << i->first << "=\"" << i->second << "\"";
117
+ }
118
+ return os;
119
+ }
120
+ };
121
+
122
+ struct Node;
123
+ typedef std::list<Node*> NodeList;
124
+
125
+ struct Node {
126
+ std::string name;
127
+ Attributes attr;
128
+ std::string content;
129
+ Node *parent;
130
+ NodeList child;
131
+ Node()
132
+ : parent(0)
133
+ {
134
+ }
135
+ ~Node()
136
+ {
137
+ for (NodeList::iterator i = child.begin(); i != child.end(); ++i) {
138
+ delete *i;
139
+ }
140
+ }
141
+ /*
142
+ swap name and attr
143
+ */
144
+ Node *appendChild(std::string& name, Attributes& attr)
145
+ {
146
+ child.push_back(0);
147
+ Node *p = new Node();
148
+ child.back() = p;
149
+ p->parent = this;
150
+ p->name.swap(name);
151
+ p->attr.swap(attr);
152
+ return p;
153
+ }
154
+ const Node *getFirstTagByName(const std::string& v) const
155
+ {
156
+ if (name == v) {
157
+ return this;
158
+ }
159
+ for (NodeList::const_iterator i = child.begin(), ie = child.end(); i != ie; ++i) {
160
+ const Node *p = (*i)->getFirstTagByName(v);
161
+ if (p) return p;
162
+ }
163
+ return 0;
164
+ }
165
+ void put(int level = 0) const
166
+ {
167
+ std::string tabstr(level, ' ');
168
+ const char *tab = tabstr.c_str();
169
+ printf("%s<%s>\n", tab, name.c_str());
170
+ if (!content.empty()) {
171
+ printf("%s[content]\n", tab);
172
+ printf("%s%s\n", tab, content.c_str());
173
+ }
174
+ attr.put(level);
175
+ for (NodeList::const_iterator i = child.begin(), ie = child.end(); i != ie; ++i) {
176
+ (*i)->put(level + indentWidth);
177
+ }
178
+ }
179
+ friend inline std::ostream& operator<<(std::ostream& os, const Node& node)
180
+ {
181
+ os << "<" << node.name << node.attr << ">";
182
+ for (NodeList::const_iterator i = node.child.begin(), ie = node.child.end(); i != ie; ++i) {
183
+ os << **i;
184
+ }
185
+ os << "</" << node.name << ">";
186
+ return os;
187
+ }
188
+ };
189
+
190
+ inline bool isSpace(char c)
191
+ {
192
+ return c == ' ' || c == '\t' || c == '\r' || c == '\n';
193
+ }
194
+
195
+ inline bool isQuote(char c)
196
+ {
197
+ return c == '"' || c == '\'';
198
+ }
199
+
200
+ inline bool isEqual(char c)
201
+ {
202
+ return c == '=';
203
+ }
204
+
205
+ inline bool isName(char c)
206
+ {
207
+ return !isSpace(c) && !isEqual(c) && !isQuote(c);
208
+ }
209
+
210
+ inline bool isName2(char c)
211
+ {
212
+ return !isSpace(c) && !isEqual(c) && !isQuote(c) && c != '>';
213
+ }
214
+
215
+ struct isAttrValue {
216
+ char q_; /* ' or " */
217
+ explicit isAttrValue(char q) : q_(q) { }
218
+ bool operator()(char c) const { return c != q_; }
219
+ };
220
+
221
+ struct isContent {
222
+ bool operator()(char c) const { return c != '<'; }
223
+ };
224
+
225
+ enum TagType {
226
+ BeginTag, // <...>
227
+ EndTag, // </...>
228
+ CompleteTag // <.../>
229
+ };
230
+
231
+ template<class Iter>
232
+ struct InputStream {
233
+ Iter begin_;
234
+ Iter end_;
235
+ InputStream(Iter begin, Iter end)
236
+ : begin_(begin)
237
+ , end_(end)
238
+ {
239
+ }
240
+ bool empty() const { return begin_ == end_; }
241
+ /*
242
+ @note assume not empty
243
+ */
244
+ char query() const
245
+ {
246
+ assert(!empty());
247
+ return *begin_;
248
+ }
249
+ void inc()
250
+ {
251
+ assert(!empty());
252
+ ++begin_;
253
+ }
254
+ char next()
255
+ {
256
+ if (empty()) throw cybozu::Exception("xml:InputStream:empty");
257
+ return *begin_++;
258
+ }
259
+ void skipSpace()
260
+ {
261
+ while (!empty()) {
262
+ if (!isSpace(query())) return;
263
+ inc();
264
+ }
265
+ }
266
+ template<class Cond>
267
+ std::string getWord(Cond f)
268
+ {
269
+ std::string word;
270
+ while (!empty()) {
271
+ char c = query();
272
+ if (!f(c)) break;
273
+ word.push_back(c);
274
+ inc();
275
+ }
276
+ return word;
277
+ }
278
+ /*
279
+ @note assume attributes exists
280
+ */
281
+ void parseAttribute(std::string *key, std::string *val)
282
+ {
283
+ *key = getWord(isName);
284
+ if (key->empty()) throw cybozu::Exception("xml:parseAttribute:bad key");
285
+ skipSpace();
286
+ char c = next();
287
+ if (isEqual(c)) {
288
+ skipSpace();
289
+ char q = next();
290
+ if (isQuote(q)) {
291
+ *val = getWord(isAttrValue(q));
292
+ c = next();
293
+ if (c == q) return;
294
+ }
295
+ }
296
+ throw cybozu::Exception("xml:parseAttribute") << *key;
297
+ }
298
+ /*
299
+ normal : true is last is '/>'
300
+ false if last is '?>'
301
+ @retval true if '/>'
302
+ false if '>'
303
+ */
304
+ bool parseAttributes(Attributes& attr, bool normal = true)
305
+ {
306
+ for (;;) {
307
+ skipSpace();
308
+ if (empty()) throw cybozu::Exception("xml:parseAttributes:tag is not complete");
309
+ char c = query();
310
+ if (normal) {
311
+ if (c == '>') {
312
+ inc();
313
+ return false;
314
+ }
315
+ if (c == '/') goto LAST;
316
+ } else {
317
+ if (c == '?') goto LAST;
318
+ }
319
+ std::string key, val;
320
+ parseAttribute(&key, &val);
321
+ attr[key] = val;
322
+ }
323
+ LAST:
324
+ inc();
325
+ char c = next();
326
+ if (c != '>') throw cybozu::Exception("xml:parseAttributes:bad tag char") << c;
327
+ return true;
328
+ }
329
+ /*
330
+ pointer is next of '<'
331
+ get tag name
332
+ *isEnd = true if </name>
333
+ */
334
+ std::string getTagName(bool *isEnd)
335
+ {
336
+ std::string name = getWord(isName);
337
+ if (name.size() <= 1 || name[name.size() - 1] != '>') throw cybozu::Exception("xml:getTagName:invalid tag");
338
+ name.resize(name.size() - 1);
339
+ if (name[0] == '/') {
340
+ *isEnd = true;
341
+ if (name.size() <= 1) throw cybozu::Exception("xml:getTagName:empty name");
342
+ return name.substr(1);
343
+ }
344
+ return name;
345
+ }
346
+ TagType parseTag(std::string& name, Attributes& attr)
347
+ {
348
+ skipSpace();
349
+ char c = next();
350
+ if (c != '<') throw cybozu::Exception("xml:parseTag:bad begin tag") << c;
351
+ if (empty()) throw cybozu::Exception("xml:parseTag:invalid tag1");
352
+ TagType type = BeginTag;
353
+ c = query();
354
+ if (c == '/') {
355
+ type = EndTag;
356
+ inc();
357
+ }
358
+ name = getWord(isName2);
359
+ if (name.empty()) throw cybozu::Exception("xml:parseTag:invalid tag2");
360
+ if (type == EndTag) {
361
+ char c = next();
362
+ if (c != '>') throw cybozu::Exception("xml:parseTag:invalid tag") << name;
363
+ return EndTag;
364
+ }
365
+ bool ret = parseAttributes(attr);
366
+ return ret ? CompleteTag : BeginTag;
367
+ }
368
+ std::string parseContent()
369
+ {
370
+ return getWord(isContent());
371
+ }
372
+ void parseNode(Node *node)
373
+ {
374
+ for (;;) {
375
+ skipSpace();
376
+ if (empty()) throw cybozu::Exception("xml:parseNode:no end tag1") << node->name;
377
+ if (query() != '<') {
378
+ node->content = parseContent();
379
+ if (empty()) throw cybozu::Exception("xml:parseNode:no end tag2") << node->name;
380
+ }
381
+ std::string name;
382
+ Attributes attr;
383
+ TagType type = parseTag(name, attr);
384
+ if (type == EndTag) {
385
+ if (node->name != name) {
386
+ throw cybozu::Exception("xml:parseNode:") << node->name << name;
387
+ }
388
+ return;
389
+ }
390
+ Node *child = node->appendChild(name, attr);
391
+ if (type == BeginTag) {
392
+ parseNode(child);
393
+ }
394
+ }
395
+ }
396
+ };
397
+
398
+ } // minixml
399
+
400
+ class MiniXml {
401
+ minixml::Attributes xmlAttr_;
402
+ minixml::Node *root_;
403
+ template<class InputStream>
404
+ void readHeader(InputStream& is)
405
+ {
406
+ is.skipSpace();
407
+ std::string w = is.getWord(minixml::isName);
408
+ if (w != "<?xml") throw cybozu::Exception("xml:readHeader:") << w;
409
+ root_ = new minixml::Node();
410
+ is.parseAttributes(xmlAttr_, false);
411
+ }
412
+ MiniXml(const MiniXml&);
413
+ void operator=(const MiniXml&);
414
+ public:
415
+ MiniXml()
416
+ : root_(0)
417
+ {
418
+ }
419
+ MiniXml(const char *begin, const char *end)
420
+ : root_(0)
421
+ {
422
+ parse(begin, end);
423
+ }
424
+ template<class InputStream>
425
+ void parse(InputStream& is)
426
+ {
427
+ readHeader(is);
428
+ minixml::TagType type = is.parseTag(root_->name, root_->attr);
429
+ if (type == minixml::EndTag) throw cybozu::Exception("xml:parse:bad end tag") << root_->name;
430
+ if (type == minixml::BeginTag) {
431
+ is.parseNode(root_);
432
+ }
433
+ }
434
+ void parse(const char *begin, const char *end)
435
+ {
436
+ minixml::InputStream<const char*> is(begin, end);
437
+ parse(is);
438
+ }
439
+ const minixml::Node& get() const { return *root_; }
440
+ ~MiniXml()
441
+ {
442
+ delete root_;
443
+ }
444
+ friend inline std::ostream& operator<<(std::ostream& os, const MiniXml& xml)
445
+ {
446
+ os << "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"yes\"?>";
447
+ os << *xml.root_;
448
+ return os;
449
+ }
450
+ };
451
+
452
+ } // cybozu
@@ -0,0 +1,143 @@
1
+ #pragma once
2
+ /**
3
+ @file
4
+ @brief mmap class
5
+
6
+ @author MITSUNARI Shigeo(@herumi)
7
+ */
8
+ #include <string>
9
+ #include <cybozu/exception.hpp>
10
+
11
+ #ifdef _WIN32
12
+ #ifndef WIN32_LEAN_AND_MEAN
13
+ #define WIN32_LEAN_AND_MEAN
14
+ #endif
15
+ #include <windows.h>
16
+ #else
17
+ #include <sys/types.h>
18
+ #include <sys/stat.h>
19
+ #include <sys/mman.h>
20
+ #include <unistd.h>
21
+ #include <fcntl.h>
22
+ #endif
23
+
24
+ namespace cybozu {
25
+
26
+ class Mmap {
27
+ const char *map_;
28
+ #ifdef _WIN32
29
+ HANDLE hFile_;
30
+ HANDLE hMap_;
31
+ template<class T>
32
+ void subOpen(const T& fileName)
33
+ {
34
+ const char *errMsg = 0;
35
+ if (hFile_ == INVALID_HANDLE_VALUE) {
36
+ errMsg = "CreateFile"; goto ERR_EXIT;
37
+ }
38
+ {
39
+ LARGE_INTEGER size;
40
+ if (GetFileSizeEx(hFile_, &size) == 0) {
41
+ errMsg = "GetFileSizeEx"; goto ERR_EXIT;
42
+ }
43
+ size_ = size.QuadPart;
44
+ }
45
+ if (size_ == 0) {
46
+ CloseHandle(hFile_); hFile_ = INVALID_HANDLE_VALUE;
47
+ return;
48
+ }
49
+
50
+ hMap_ = CreateFileMapping(hFile_, NULL, PAGE_READONLY, 0, 0, NULL);
51
+ if (hMap_ == NULL) {
52
+ errMsg = "CreateFileMapping"; goto ERR_EXIT;
53
+ }
54
+
55
+ map_ = (const char*)MapViewOfFile(hMap_, FILE_MAP_READ, 0, 0, 0);
56
+ if (map_ == 0) {
57
+ errMsg = "MapViewOfFile"; goto ERR_EXIT;
58
+ }
59
+ return;
60
+ ERR_EXIT:
61
+ std::string reason = cybozu::ErrorNo().toString();
62
+ if (hMap_) CloseHandle(hMap_);
63
+ if (hFile_ != INVALID_HANDLE_VALUE) CloseHandle(hFile_);
64
+ throw cybozu::Exception("mmap") << errMsg << fileName << reason;
65
+ }
66
+ #endif
67
+ uint64_t size_;
68
+ public:
69
+ #ifdef _WIN32
70
+ explicit Mmap(const std::string& fileName)
71
+ : map_(0)
72
+ , hFile_(INVALID_HANDLE_VALUE)
73
+ , hMap_(0)
74
+ , size_(0)
75
+ {
76
+ hFile_ = CreateFileA(fileName.c_str(), GENERIC_READ, FILE_SHARE_READ | FILE_SHARE_WRITE, NULL,
77
+ OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL);
78
+ subOpen(fileName);
79
+ }
80
+ explicit Mmap(const std::wstring& fileName)
81
+ : map_(0)
82
+ , hFile_(INVALID_HANDLE_VALUE)
83
+ , hMap_(0)
84
+ , size_(0)
85
+ {
86
+ hFile_ = CreateFileW(fileName.c_str(), GENERIC_READ, FILE_SHARE_READ | FILE_SHARE_WRITE, NULL,
87
+ OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL);
88
+ subOpen(fileName);
89
+ }
90
+ #else
91
+ explicit Mmap(const std::string& fileName)
92
+ : map_(static_cast<const char*>(MAP_FAILED))
93
+ , size_(0)
94
+ {
95
+ const char *errMsg = 0;
96
+ int fd = ::open(fileName.c_str(), O_RDONLY);
97
+ if (fd == -1) {
98
+ errMsg = "open"; goto ERR_EXIT;
99
+ }
100
+
101
+ {
102
+ struct stat st;
103
+ int ret = ::fstat(fd, &st);
104
+ if (ret != 0) {
105
+ errMsg = "fstat"; goto ERR_EXIT;
106
+ }
107
+ size_ = st.st_size;
108
+ }
109
+ if (size_ == 0) {
110
+ ::close(fd);
111
+ return;
112
+ }
113
+
114
+ map_ = (const char*)::mmap(NULL, size_, PROT_READ, MAP_SHARED, fd, 0);
115
+ if (map_ == MAP_FAILED) {
116
+ errMsg = "mmap"; goto ERR_EXIT;
117
+ }
118
+ ::close(fd);
119
+ return;
120
+ ERR_EXIT:
121
+ std::string reason = cybozu::ErrorNo().toString();
122
+ if (fd != -1) close(fd);
123
+ throw cybozu::Exception("mmap") << errMsg << fileName << reason;
124
+ }
125
+ #endif
126
+ ~Mmap()
127
+ {
128
+ #ifdef _WIN32
129
+ if (map_) UnmapViewOfFile(map_);
130
+ if (hMap_) CloseHandle(hMap_);
131
+ if (hFile_ != INVALID_HANDLE_VALUE) CloseHandle(hFile_);
132
+ #else
133
+ if (map_ != MAP_FAILED) ::munmap(const_cast<char*>(map_), size_);
134
+ #endif
135
+ }
136
+ uint64_t size() const { return size_; }
137
+ const char *get() const { return map_; }
138
+ private:
139
+ Mmap(const Mmap &);
140
+ void operator=(const Mmap &);
141
+ };
142
+
143
+ } // cybozu