tpy-lang 0.3.0.dev0__py3-none-any.whl

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 (333) hide show
  1. tpy_lang-0.3.0.dev0.dist-info/METADATA +151 -0
  2. tpy_lang-0.3.0.dev0.dist-info/RECORD +333 -0
  3. tpy_lang-0.3.0.dev0.dist-info/WHEEL +4 -0
  4. tpy_lang-0.3.0.dev0.dist-info/entry_points.txt +3 -0
  5. tpyc/__init__.py +104 -0
  6. tpyc/__main__.py +6 -0
  7. tpyc/_buildinfo.py +1 -0
  8. tpyc/_data/docs/LANGUAGE_FEATURES.md +6278 -0
  9. tpyc/_data/docs/STDLIB_ROADMAP.md +1258 -0
  10. tpyc/_data/docs/TPY_FOR_AGENTS.md +556 -0
  11. tpyc/_data/lib/tpy/_bindings/__init__.py +6 -0
  12. tpyc/_data/lib/tpy/_bindings/pcre2.py +173 -0
  13. tpyc/_data/lib/tpy/_bindings/posix_socket.py +161 -0
  14. tpyc/_data/lib/tpy/_functools_macros.py +80 -0
  15. tpyc/_data/lib/tpy/_macro_helpers.py +161 -0
  16. tpyc/_data/lib/tpy/argparse.py +2062 -0
  17. tpyc/_data/lib/tpy/asyncio/__init__.py +744 -0
  18. tpyc/_data/lib/tpy/asyncio/_executor.py +515 -0
  19. tpyc/_data/lib/tpy/base64.py +410 -0
  20. tpyc/_data/lib/tpy/bisect.py +39 -0
  21. tpyc/_data/lib/tpy/builtins.py +38 -0
  22. tpyc/_data/lib/tpy/dataclasses.py +354 -0
  23. tpyc/_data/lib/tpy/enum.py +23 -0
  24. tpyc/_data/lib/tpy/functools.py +33 -0
  25. tpyc/_data/lib/tpy/hashlib.py +206 -0
  26. tpyc/_data/lib/tpy/heapq.py +118 -0
  27. tpyc/_data/lib/tpy/io.py +395 -0
  28. tpyc/_data/lib/tpy/json.py +221 -0
  29. tpyc/_data/lib/tpy/math.py +406 -0
  30. tpyc/_data/lib/tpy/random.py +597 -0
  31. tpyc/_data/lib/tpy/re.py +467 -0
  32. tpyc/_data/lib/tpy/socket.py +379 -0
  33. tpyc/_data/lib/tpy/struct.py +178 -0
  34. tpyc/_data/lib/tpy/sys.py +40 -0
  35. tpyc/_data/lib/tpy/time.py +39 -0
  36. tpyc/_data/lib/tpy/tpy/__init__.py +78 -0
  37. tpyc/_data/lib/tpy/tpy/_bootstrap/__init__.py +10 -0
  38. tpyc/_data/lib/tpy/tpy/_bootstrap/_decorators.py +37 -0
  39. tpyc/_data/lib/tpy/tpy/_bootstrap/_extern.py +64 -0
  40. tpyc/_data/lib/tpy/tpy/_builtins/__init__.py +11 -0
  41. tpyc/_data/lib/tpy/tpy/_builtins/_bytes.py +378 -0
  42. tpyc/_data/lib/tpy/tpy/_builtins/_dict.py +151 -0
  43. tpyc/_data/lib/tpy/tpy/_builtins/_exceptions.py +125 -0
  44. tpyc/_data/lib/tpy/tpy/_builtins/_funcs.py +681 -0
  45. tpyc/_data/lib/tpy/tpy/_builtins/_io.py +97 -0
  46. tpyc/_data/lib/tpy/tpy/_builtins/_list.py +127 -0
  47. tpyc/_data/lib/tpy/tpy/_builtins/_range.py +52 -0
  48. tpyc/_data/lib/tpy/tpy/_builtins/_set.py +139 -0
  49. tpyc/_data/lib/tpy/tpy/_builtins/_super.py +11 -0
  50. tpyc/_data/lib/tpy/tpy/_builtins/_types.py +661 -0
  51. tpyc/_data/lib/tpy/tpy/_core/__init__.py +23 -0
  52. tpyc/_data/lib/tpy/tpy/_core/_bytes_view.py +129 -0
  53. tpyc/_data/lib/tpy/tpy/_core/_containers.py +137 -0
  54. tpyc/_data/lib/tpy/tpy/_core/_functions.py +40 -0
  55. tpyc/_data/lib/tpy/tpy/_core/_types.py +2061 -0
  56. tpyc/_data/lib/tpy/tpy/_typing/__init__.py +77 -0
  57. tpyc/_data/lib/tpy/tpy/_version.py +29 -0
  58. tpyc/_data/lib/tpy/tpy/bits.py +28 -0
  59. tpyc/_data/lib/tpy/tpy/coro/__init__.py +127 -0
  60. tpyc/_data/lib/tpy/tpy/extern.py +8 -0
  61. tpyc/_data/lib/tpy/tpy/mem.py +49 -0
  62. tpyc/_data/lib/tpy/tpy/unsafe.py +195 -0
  63. tpyc/_data/lib/tpy/tpy/version.py +21 -0
  64. tpyc/_data/lib/tpy/typing.py +13 -0
  65. tpyc/_data/runtime/cpp/include/tpy/any.hpp +461 -0
  66. tpyc/_data/runtime/cpp/include/tpy/as_ostream.hpp +117 -0
  67. tpyc/_data/runtime/cpp/include/tpy/async.hpp +76 -0
  68. tpyc/_data/runtime/cpp/include/tpy/bigint.hpp +1343 -0
  69. tpyc/_data/runtime/cpp/include/tpy/builtins.hpp +400 -0
  70. tpyc/_data/runtime/cpp/include/tpy/bytes_ops.hpp +469 -0
  71. tpyc/_data/runtime/cpp/include/tpy/container_ops.hpp +487 -0
  72. tpyc/_data/runtime/cpp/include/tpy/copy_iter.hpp +82 -0
  73. tpyc/_data/runtime/cpp/include/tpy/core.hpp +558 -0
  74. tpyc/_data/runtime/cpp/include/tpy/dict_ops.hpp +289 -0
  75. tpyc/_data/runtime/cpp/include/tpy/dunder.hpp +750 -0
  76. tpyc/_data/runtime/cpp/include/tpy/dynamic.hpp +44 -0
  77. tpyc/_data/runtime/cpp/include/tpy/enum.hpp +40 -0
  78. tpyc/_data/runtime/cpp/include/tpy/file.hpp +245 -0
  79. tpyc/_data/runtime/cpp/include/tpy/fixed_int.hpp +317 -0
  80. tpyc/_data/runtime/cpp/include/tpy/format.hpp +954 -0
  81. tpyc/_data/runtime/cpp/include/tpy/frame_slot.hpp +120 -0
  82. tpyc/_data/runtime/cpp/include/tpy/generator.hpp +47 -0
  83. tpyc/_data/runtime/cpp/include/tpy/iterable_ops.hpp +122 -0
  84. tpyc/_data/runtime/cpp/include/tpy/itertools.hpp +749 -0
  85. tpyc/_data/runtime/cpp/include/tpy/next_iter.hpp +82 -0
  86. tpyc/_data/runtime/cpp/include/tpy/ordered_map.hpp +518 -0
  87. tpyc/_data/runtime/cpp/include/tpy/ordered_set.hpp +337 -0
  88. tpyc/_data/runtime/cpp/include/tpy/own_iter.hpp +54 -0
  89. tpyc/_data/runtime/cpp/include/tpy/pascal_graph_sdl.hpp +192 -0
  90. tpyc/_data/runtime/cpp/include/tpy/printing.hpp +302 -0
  91. tpyc/_data/runtime/cpp/include/tpy/protocols.hpp +61 -0
  92. tpyc/_data/runtime/cpp/include/tpy/range.hpp +115 -0
  93. tpyc/_data/runtime/cpp/include/tpy/ranges.hpp +212 -0
  94. tpyc/_data/runtime/cpp/include/tpy/set_ops.hpp +265 -0
  95. tpyc/_data/runtime/cpp/include/tpy/slice.hpp +47 -0
  96. tpyc/_data/runtime/cpp/include/tpy/span_iter.hpp +42 -0
  97. tpyc/_data/runtime/cpp/include/tpy/stdlib/math.hpp +41 -0
  98. tpyc/_data/runtime/cpp/include/tpy/stdlib/pcre2_h.hpp +96 -0
  99. tpyc/_data/runtime/cpp/include/tpy/stdlib/random.hpp +25 -0
  100. tpyc/_data/runtime/cpp/include/tpy/stdlib/socket_h.hpp +145 -0
  101. tpyc/_data/runtime/cpp/include/tpy/stdlib/time.hpp +62 -0
  102. tpyc/_data/runtime/cpp/include/tpy/system.hpp +121 -0
  103. tpyc/_data/runtime/cpp/include/tpy/throwable.hpp +55 -0
  104. tpyc/_data/runtime/cpp/include/tpy/tpy.hpp +156 -0
  105. tpyc/_data/runtime/cpp/include/tpy/type_name.hpp +77 -0
  106. tpyc/_data/runtime/cpp/include/tpy/type_traits.hpp +240 -0
  107. tpyc/_data/runtime/cpp/include/tpy/uninit_array_storage.hpp +250 -0
  108. tpyc/_data/runtime/cpp/include/tpy/uninit_heap_storage.hpp +277 -0
  109. tpyc/_data/runtime/cpp/include/tpy/varargs.hpp +174 -0
  110. tpyc/_data/runtime/cpp/include/tpy/variant_ref.hpp +118 -0
  111. tpyc/_data/runtime/cpp/src/stdlib/socket_impl.cpp +104 -0
  112. tpyc/_data/runtime/cpp/third_party/README.md +58 -0
  113. tpyc/_data/runtime/cpp/third_party/pcre2/AUTHORS +36 -0
  114. tpyc/_data/runtime/cpp/third_party/pcre2/CMakeLists.txt +1233 -0
  115. tpyc/_data/runtime/cpp/third_party/pcre2/COPYING +5 -0
  116. tpyc/_data/runtime/cpp/third_party/pcre2/ChangeLog +3097 -0
  117. tpyc/_data/runtime/cpp/third_party/pcre2/HACKING +853 -0
  118. tpyc/_data/runtime/cpp/third_party/pcre2/INSTALL +368 -0
  119. tpyc/_data/runtime/cpp/third_party/pcre2/LICENCE +94 -0
  120. tpyc/_data/runtime/cpp/third_party/pcre2/NEWS +492 -0
  121. tpyc/_data/runtime/cpp/third_party/pcre2/NON-AUTOTOOLS-BUILD +430 -0
  122. tpyc/_data/runtime/cpp/third_party/pcre2/README +956 -0
  123. tpyc/_data/runtime/cpp/third_party/pcre2/cmake/COPYING-CMAKE-SCRIPTS +22 -0
  124. tpyc/_data/runtime/cpp/third_party/pcre2/cmake/FindEditline.cmake +16 -0
  125. tpyc/_data/runtime/cpp/third_party/pcre2/cmake/FindPackageHandleStandardArgs.cmake +58 -0
  126. tpyc/_data/runtime/cpp/third_party/pcre2/cmake/FindReadline.cmake +29 -0
  127. tpyc/_data/runtime/cpp/third_party/pcre2/cmake/pcre2-config-version.cmake.in +15 -0
  128. tpyc/_data/runtime/cpp/third_party/pcre2/cmake/pcre2-config.cmake.in +148 -0
  129. tpyc/_data/runtime/cpp/third_party/pcre2/config-cmake.h.in +56 -0
  130. tpyc/_data/runtime/cpp/third_party/pcre2/libpcre2-16.pc.in +13 -0
  131. tpyc/_data/runtime/cpp/third_party/pcre2/libpcre2-32.pc.in +13 -0
  132. tpyc/_data/runtime/cpp/third_party/pcre2/libpcre2-8.pc.in +13 -0
  133. tpyc/_data/runtime/cpp/third_party/pcre2/libpcre2-posix.pc.in +13 -0
  134. tpyc/_data/runtime/cpp/third_party/pcre2/pcre2-config.in +121 -0
  135. tpyc/_data/runtime/cpp/third_party/pcre2/src/config.h +483 -0
  136. tpyc/_data/runtime/cpp/third_party/pcre2/src/config.h.generic +483 -0
  137. tpyc/_data/runtime/cpp/third_party/pcre2/src/config.h.in +460 -0
  138. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2.h +1010 -0
  139. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2.h.generic +1010 -0
  140. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2.h.in +1010 -0
  141. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_auto_possess.c +1371 -0
  142. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_chartables.c +196 -0
  143. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_chartables.c.dist +196 -0
  144. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_chkdint.c +96 -0
  145. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_compile.c +11001 -0
  146. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_config.c +252 -0
  147. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_context.c +510 -0
  148. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_convert.c +1189 -0
  149. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_dfa_match.c +4119 -0
  150. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_dftables.c +297 -0
  151. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_error.c +345 -0
  152. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_extuni.c +162 -0
  153. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_find_bracket.c +219 -0
  154. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_fuzzsupport.c +792 -0
  155. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_internal.h +2084 -0
  156. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_intmodedep.h +940 -0
  157. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_jit_compile.c +14972 -0
  158. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_jit_match.c +200 -0
  159. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_jit_misc.c +234 -0
  160. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_jit_neon_inc.h +354 -0
  161. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_jit_simd_inc.h +2355 -0
  162. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_jit_test.c +2528 -0
  163. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_maketables.c +165 -0
  164. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_match.c +7777 -0
  165. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_match_data.c +185 -0
  166. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_newline.c +243 -0
  167. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_ord2utf.c +120 -0
  168. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_pattern_info.c +432 -0
  169. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_printint.c +886 -0
  170. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_script_run.c +344 -0
  171. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_serialize.c +286 -0
  172. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_string_utils.c +237 -0
  173. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_study.c +1915 -0
  174. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_substitute.c +1009 -0
  175. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_substring.c +550 -0
  176. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_tables.c +234 -0
  177. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_ucd.c +5460 -0
  178. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_ucp.h +396 -0
  179. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_ucptables.c +1533 -0
  180. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_valid_utf.c +398 -0
  181. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2_xclass.c +308 -0
  182. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2demo.c +497 -0
  183. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2grep.c +4606 -0
  184. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2posix.c +425 -0
  185. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2posix.h +187 -0
  186. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2posix_test.c +209 -0
  187. tpyc/_data/runtime/cpp/third_party/pcre2/src/pcre2test.c +9708 -0
  188. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/allocator_src/sljitExecAllocatorApple.c +137 -0
  189. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/allocator_src/sljitExecAllocatorCore.c +327 -0
  190. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/allocator_src/sljitExecAllocatorFreeBSD.c +89 -0
  191. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/allocator_src/sljitExecAllocatorPosix.c +62 -0
  192. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/allocator_src/sljitExecAllocatorWindows.c +40 -0
  193. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/allocator_src/sljitProtExecAllocatorNetBSD.c +72 -0
  194. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/allocator_src/sljitProtExecAllocatorPosix.c +172 -0
  195. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/allocator_src/sljitWXExecAllocatorPosix.c +141 -0
  196. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/allocator_src/sljitWXExecAllocatorWindows.c +102 -0
  197. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitConfig.h +142 -0
  198. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitConfigCPU.h +188 -0
  199. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitConfigInternal.h +907 -0
  200. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitLir.c +3561 -0
  201. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitLir.h +2466 -0
  202. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeARM_32.c +4636 -0
  203. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeARM_64.c +3491 -0
  204. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeARM_T2_32.c +4302 -0
  205. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeLOONGARCH_64.c +3765 -0
  206. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeMIPS_32.c +472 -0
  207. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeMIPS_64.c +387 -0
  208. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeMIPS_common.c +4259 -0
  209. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativePPC_32.c +485 -0
  210. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativePPC_64.c +719 -0
  211. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativePPC_common.c +3161 -0
  212. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeRISCV_32.c +142 -0
  213. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeRISCV_64.c +222 -0
  214. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeRISCV_common.c +3121 -0
  215. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeS390X.c +4526 -0
  216. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeX86_32.c +1685 -0
  217. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeX86_64.c +1398 -0
  218. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitNativeX86_common.c +5001 -0
  219. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitSerialize.c +516 -0
  220. tpyc/_data/runtime/cpp/third_party/pcre2/src/sljit/sljitUtils.c +344 -0
  221. tpyc/_data/runtime/cpp/third_party/pcre2.sources.txt +54 -0
  222. tpyc/_data/runtime/cpp/third_party/pcre2.vendor.json +7 -0
  223. tpyc/build/__init__.py +7 -0
  224. tpyc/build/pcre2.py +122 -0
  225. tpyc/build/third_party.py +413 -0
  226. tpyc/cli.py +822 -0
  227. tpyc/codegen_cpp/__init__.py +18 -0
  228. tpyc/codegen_cpp/builtins.py +484 -0
  229. tpyc/codegen_cpp/context.py +2064 -0
  230. tpyc/codegen_cpp/expressions.py +5940 -0
  231. tpyc/codegen_cpp/functions.py +1913 -0
  232. tpyc/codegen_cpp/gen_async.py +3258 -0
  233. tpyc/codegen_cpp/gen_generators.py +657 -0
  234. tpyc/codegen_cpp/generator.py +2258 -0
  235. tpyc/codegen_cpp/match.py +1997 -0
  236. tpyc/codegen_cpp/param_const.py +172 -0
  237. tpyc/codegen_cpp/protocols.py +907 -0
  238. tpyc/codegen_cpp/records.py +1654 -0
  239. tpyc/codegen_cpp/resumable_cfg.py +1651 -0
  240. tpyc/codegen_cpp/statements.py +4963 -0
  241. tpyc/codegen_cpp/string_dispatch.py +76 -0
  242. tpyc/codegen_cpp/test_context.py +46 -0
  243. tpyc/codegen_cpp/test_param_const.py +113 -0
  244. tpyc/codegen_cpp/test_resumable_cfg.py +182 -0
  245. tpyc/codegen_cpp/type_resolution.py +53 -0
  246. tpyc/codegen_cpp/types.py +436 -0
  247. tpyc/codegen_cpp/variant_access.py +135 -0
  248. tpyc/coercions.py +749 -0
  249. tpyc/compilation_context.py +57 -0
  250. tpyc/compiler.py +3945 -0
  251. tpyc/cycle_detection.py +358 -0
  252. tpyc/diagnostics.py +135 -0
  253. tpyc/dump_types.py +353 -0
  254. tpyc/frontend_diagnostics.py +47 -0
  255. tpyc/frontend_ir/__init__.py +140 -0
  256. tpyc/frontend_ir/lower.py +1098 -0
  257. tpyc/frontend_ir/nodes.py +718 -0
  258. tpyc/frontend_ir/resolver_adapter.py +151 -0
  259. tpyc/frontend_plugin.py +209 -0
  260. tpyc/install_docs.py +81 -0
  261. tpyc/liveness.py +756 -0
  262. tpyc/macro_api.py +1724 -0
  263. tpyc/macro_loader.py +497 -0
  264. tpyc/module_names.py +64 -0
  265. tpyc/modules/__init__.py +31 -0
  266. tpyc/modules/defs.py +89 -0
  267. tpyc/modules/registry.py +36 -0
  268. tpyc/modules/resolver.py +192 -0
  269. tpyc/modules/type_resolution.py +629 -0
  270. tpyc/namespace.py +172 -0
  271. tpyc/parse/__init__.py +84 -0
  272. tpyc/parse/imports.py +490 -0
  273. tpyc/parse/nodes.py +1732 -0
  274. tpyc/parse/parser.py +4043 -0
  275. tpyc/parse/resolve_refs.py +466 -0
  276. tpyc/parse/type_resolver.py +1060 -0
  277. tpyc/prescan.py +254 -0
  278. tpyc/qnames.py +149 -0
  279. tpyc/repl.py +529 -0
  280. tpyc/repl_backends.py +848 -0
  281. tpyc/sema/__init__.py +21 -0
  282. tpyc/sema/analyzer.py +3625 -0
  283. tpyc/sema/bound_check.py +72 -0
  284. tpyc/sema/builder_trace.py +684 -0
  285. tpyc/sema/calls.py +5406 -0
  286. tpyc/sema/compatibility.py +2107 -0
  287. tpyc/sema/context.py +1243 -0
  288. tpyc/sema/expressions.py +3737 -0
  289. tpyc/sema/flow_facts.py +199 -0
  290. tpyc/sema/init_tracker.py +150 -0
  291. tpyc/sema/list_literals.py +69 -0
  292. tpyc/sema/literal_utils.py +27 -0
  293. tpyc/sema/local_deduction.py +1088 -0
  294. tpyc/sema/macros.py +179 -0
  295. tpyc/sema/match.py +1177 -0
  296. tpyc/sema/method_expansion.py +347 -0
  297. tpyc/sema/methods.py +2197 -0
  298. tpyc/sema/mutation_propagation.py +268 -0
  299. tpyc/sema/narrowing.py +857 -0
  300. tpyc/sema/numeric_lattice.py +160 -0
  301. tpyc/sema/operators.py +402 -0
  302. tpyc/sema/overloads.py +841 -0
  303. tpyc/sema/protocols.py +1209 -0
  304. tpyc/sema/reach_analysis.py +202 -0
  305. tpyc/sema/registration.py +3156 -0
  306. tpyc/sema/scope_tracker.py +193 -0
  307. tpyc/sema/statements.py +4426 -0
  308. tpyc/sema/type_ops.py +1879 -0
  309. tpyc/sema/value_range.py +181 -0
  310. tpyc/symbol_binding.py +259 -0
  311. tpyc/test_c3_mro.py +208 -0
  312. tpyc/test_cli_argv.py +52 -0
  313. tpyc/test_compiler.py +559 -0
  314. tpyc/test_contains_type_param.py +101 -0
  315. tpyc/test_cycle_detection.py +221 -0
  316. tpyc/test_dump_types.py +225 -0
  317. tpyc/test_install_docs.py +65 -0
  318. tpyc/test_local_cpp_form.py +135 -0
  319. tpyc/test_macro_loader.py +76 -0
  320. tpyc/test_method_expansion.py +254 -0
  321. tpyc/test_nominal_identity.py +182 -0
  322. tpyc/test_overloads.py +410 -0
  323. tpyc/test_parse.py +303 -0
  324. tpyc/test_parse_type_ref.py +506 -0
  325. tpyc/test_parse_version_info.py +58 -0
  326. tpyc/test_reach_analysis.py +72 -0
  327. tpyc/test_ref_type.py +216 -0
  328. tpyc/test_send_sync_substitution.py +276 -0
  329. tpyc/test_tuple_mutation_propagation.py +206 -0
  330. tpyc/test_type_def_registry.py +1729 -0
  331. tpyc/test_union_types.py +195 -0
  332. tpyc/type_def_registry.py +975 -0
  333. tpyc/typesys.py +5104 -0
@@ -0,0 +1,792 @@
1
+ /***************************************************************************
2
+ Fuzzer driver for PCRE2. Given an arbitrary string of bytes and a length, it
3
+ tries to compile and match it, deriving options from the string itself. If
4
+ STANDALONE is defined, a main program that calls the driver with the contents
5
+ of specified files is compiled, and commentary on what is happening is output.
6
+ If an argument starts with '=' the rest of it it is taken as a literal string
7
+ rather than a file name. This allows easy testing of short strings.
8
+
9
+ Written by Philip Hazel, October 2016
10
+ Updated February 2024 (Addison Crump added 16-bit/32-bit and JIT support)
11
+ Further updates March/April/May 2024 by PH
12
+ ***************************************************************************/
13
+
14
+ #include <errno.h>
15
+ #include <stdarg.h>
16
+ #include <stdio.h>
17
+ #include <stdlib.h>
18
+ #include <string.h>
19
+ #include <unistd.h>
20
+
21
+ /* stack size adjustment */
22
+ #include <sys/time.h>
23
+ #include <sys/resource.h>
24
+
25
+ #define STACK_SIZE_MB 256
26
+ #define JIT_SIZE_LIMIT (200 * 1024)
27
+
28
+ #ifndef PCRE2_CODE_UNIT_WIDTH
29
+ #define PCRE2_CODE_UNIT_WIDTH 8
30
+ #endif
31
+
32
+ #include "config.h"
33
+ #include "pcre2.h"
34
+ #include "pcre2_internal.h"
35
+
36
+ #define MAX_MATCH_SIZE 1000
37
+
38
+ #define DFA_WORKSPACE_COUNT 100
39
+
40
+ /* When adding new compile or match options, remember to update the functions
41
+ below that output them. */
42
+
43
+ #define ALLOWED_COMPILE_OPTIONS \
44
+ (PCRE2_ANCHORED|PCRE2_ALLOW_EMPTY_CLASS|PCRE2_ALT_BSUX|PCRE2_ALT_CIRCUMFLEX| \
45
+ PCRE2_ALT_VERBNAMES|PCRE2_AUTO_CALLOUT|PCRE2_CASELESS|PCRE2_DOLLAR_ENDONLY| \
46
+ PCRE2_DOTALL|PCRE2_DUPNAMES|PCRE2_ENDANCHORED|PCRE2_EXTENDED| \
47
+ PCRE2_EXTENDED_MORE|PCRE2_FIRSTLINE| \
48
+ PCRE2_MATCH_UNSET_BACKREF|PCRE2_MULTILINE|PCRE2_NEVER_BACKSLASH_C| \
49
+ PCRE2_NO_AUTO_CAPTURE| \
50
+ PCRE2_NO_AUTO_POSSESS|PCRE2_NO_DOTSTAR_ANCHOR|PCRE2_NO_START_OPTIMIZE| \
51
+ PCRE2_UCP|PCRE2_UNGREEDY|PCRE2_USE_OFFSET_LIMIT| \
52
+ PCRE2_UTF)
53
+
54
+ #define ALLOWED_MATCH_OPTIONS \
55
+ (PCRE2_ANCHORED|PCRE2_ENDANCHORED|PCRE2_NOTBOL|PCRE2_NOTEOL|PCRE2_NOTEMPTY| \
56
+ PCRE2_NOTEMPTY_ATSTART|PCRE2_PARTIAL_HARD| \
57
+ PCRE2_PARTIAL_SOFT)
58
+
59
+ #define BASE_MATCH_OPTIONS \
60
+ (PCRE2_NO_JIT|PCRE2_DISABLE_RECURSELOOP_CHECK)
61
+
62
+
63
+ #if defined(SUPPORT_DIFF_FUZZ) || defined(STANDALONE)
64
+ static void print_compile_options(FILE *stream, uint32_t compile_options)
65
+ {
66
+ fprintf(stream, "Compile options %s%.8x =",
67
+ (compile_options == PCRE2_NEVER_BACKSLASH_C)? "(base) " : "",
68
+ compile_options);
69
+
70
+ fprintf(stream, "%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s\n",
71
+ ((compile_options & PCRE2_ALT_BSUX) != 0)? " alt_bsux" : "",
72
+ ((compile_options & PCRE2_ALT_CIRCUMFLEX) != 0)? " alt_circumflex" : "",
73
+ ((compile_options & PCRE2_ALT_VERBNAMES) != 0)? " alt_verbnames" : "",
74
+ ((compile_options & PCRE2_ALLOW_EMPTY_CLASS) != 0)? " allow_empty_class" : "",
75
+ ((compile_options & PCRE2_ANCHORED) != 0)? " anchored" : "",
76
+ ((compile_options & PCRE2_AUTO_CALLOUT) != 0)? " auto_callout" : "",
77
+ ((compile_options & PCRE2_CASELESS) != 0)? " caseless" : "",
78
+ ((compile_options & PCRE2_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "",
79
+ ((compile_options & PCRE2_DOTALL) != 0)? " dotall" : "",
80
+ ((compile_options & PCRE2_DUPNAMES) != 0)? " dupnames" : "",
81
+ ((compile_options & PCRE2_ENDANCHORED) != 0)? " endanchored" : "",
82
+ ((compile_options & PCRE2_EXTENDED) != 0)? " extended" : "",
83
+ ((compile_options & PCRE2_EXTENDED_MORE) != 0)? " extended_more" : "",
84
+ ((compile_options & PCRE2_FIRSTLINE) != 0)? " firstline" : "",
85
+ ((compile_options & PCRE2_MATCH_UNSET_BACKREF) != 0)? " match_unset_backref" : "",
86
+ ((compile_options & PCRE2_MULTILINE) != 0)? " multiline" : "",
87
+ ((compile_options & PCRE2_NEVER_BACKSLASH_C) != 0)? " never_backslash_c" : "",
88
+ ((compile_options & PCRE2_NEVER_UCP) != 0)? " never_ucp" : "",
89
+ ((compile_options & PCRE2_NEVER_UTF) != 0)? " never_utf" : "",
90
+ ((compile_options & PCRE2_NO_AUTO_CAPTURE) != 0)? " no_auto_capture" : "",
91
+ ((compile_options & PCRE2_NO_AUTO_POSSESS) != 0)? " no_auto_possess" : "",
92
+ ((compile_options & PCRE2_NO_DOTSTAR_ANCHOR) != 0)? " no_dotstar_anchor" : "",
93
+ ((compile_options & PCRE2_NO_UTF_CHECK) != 0)? " no_utf_check" : "",
94
+ ((compile_options & PCRE2_NO_START_OPTIMIZE) != 0)? " no_start_optimize" : "",
95
+ ((compile_options & PCRE2_UCP) != 0)? " ucp" : "",
96
+ ((compile_options & PCRE2_UNGREEDY) != 0)? " ungreedy" : "",
97
+ ((compile_options & PCRE2_USE_OFFSET_LIMIT) != 0)? " use_offset_limit" : "",
98
+ ((compile_options & PCRE2_UTF) != 0)? " utf" : "");
99
+ }
100
+
101
+ static void print_match_options(FILE *stream, uint32_t match_options)
102
+ {
103
+ fprintf(stream, "Match options %s%.8x =",
104
+ (match_options == BASE_MATCH_OPTIONS)? "(base) " : "", match_options);
105
+
106
+ fprintf(stream, "%s%s%s%s%s%s%s%s%s%s%s\n",
107
+ ((match_options & PCRE2_ANCHORED) != 0)? " anchored" : "",
108
+ ((match_options & PCRE2_DISABLE_RECURSELOOP_CHECK) != 0)? " disable_recurseloop_check" : "",
109
+ ((match_options & PCRE2_ENDANCHORED) != 0)? " endanchored" : "",
110
+ ((match_options & PCRE2_NO_JIT) != 0)? " no_jit" : "",
111
+ ((match_options & PCRE2_NO_UTF_CHECK) != 0)? " no_utf_check" : "",
112
+ ((match_options & PCRE2_NOTBOL) != 0)? " notbol" : "",
113
+ ((match_options & PCRE2_NOTEMPTY) != 0)? " notempty" : "",
114
+ ((match_options & PCRE2_NOTEMPTY_ATSTART) != 0)? " notempty_atstart" : "",
115
+ ((match_options & PCRE2_NOTEOL) != 0)? " noteol" : "",
116
+ ((match_options & PCRE2_PARTIAL_HARD) != 0)? " partial_hard" : "",
117
+ ((match_options & PCRE2_PARTIAL_SOFT) != 0)? " partial_soft" : "");
118
+ }
119
+
120
+
121
+ /* This function can print an error message at all code unit widths. */
122
+
123
+ static void print_error(FILE *f, int errorcode, const char *text, ...)
124
+ {
125
+ PCRE2_UCHAR buffer[256];
126
+ PCRE2_UCHAR *p = buffer;
127
+ va_list ap;
128
+ va_start(ap, text);
129
+ vfprintf(f, text, ap);
130
+ va_end(ap);
131
+ pcre2_get_error_message(errorcode, buffer, 256);
132
+ while (*p != 0) fprintf(f, "%c", *p++);
133
+ printf("\n");
134
+ }
135
+ #endif /* defined(SUPPORT_DIFF_FUZZ || defined(STANDALONE) */
136
+
137
+
138
+ #ifdef SUPPORT_JIT
139
+ #ifdef SUPPORT_DIFF_FUZZ
140
+ static void dump_matches(FILE *stream, int count, pcre2_match_data *match_data)
141
+ {
142
+ int errorcode;
143
+
144
+ for (int index = 0; index < count; index++)
145
+ {
146
+ PCRE2_UCHAR *bufferptr = NULL;
147
+ PCRE2_SIZE bufflen = 0;
148
+
149
+ errorcode = pcre2_substring_get_bynumber(match_data, index, &bufferptr,
150
+ &bufflen);
151
+
152
+ if (errorcode >= 0)
153
+ {
154
+ fprintf(stream, "Match %d (hex encoded): ", index);
155
+ for (PCRE2_SIZE i = 0; i < bufflen; i++)
156
+ {
157
+ fprintf(stream, "%02x", bufferptr[i]);
158
+ }
159
+ fprintf(stream, "\n");
160
+ }
161
+ else
162
+ {
163
+ print_error(stream, errorcode, "Match %d failed: ", index);
164
+ }
165
+ }
166
+ }
167
+
168
+ /* This function describes the current test case being evaluated, then aborts */
169
+
170
+ static void describe_failure(
171
+ const char *task,
172
+ const unsigned char *data,
173
+ size_t size,
174
+ uint32_t compile_options,
175
+ uint32_t match_options,
176
+ int errorcode,
177
+ int errorcode_jit,
178
+ int matches,
179
+ int matches_jit,
180
+ pcre2_match_data *match_data,
181
+ pcre2_match_data *match_data_jit
182
+ ) {
183
+
184
+ fprintf(stderr, "Encountered failure while performing %s; context:\n", task);
185
+
186
+ fprintf(stderr, "Pattern/sample string (hex encoded): ");
187
+ for (size_t i = 0; i < size; i++)
188
+ {
189
+ fprintf(stderr, "%02x", data[i]);
190
+ }
191
+ fprintf(stderr, "\n");
192
+
193
+ print_compile_options(stderr, compile_options);
194
+ print_match_options(stderr, match_options);
195
+
196
+ if (errorcode < 0)
197
+ {
198
+ print_error(stderr, errorcode, "Non-JIT'd operation emitted an error: ");
199
+ }
200
+
201
+ if (matches >= 0)
202
+ {
203
+ fprintf(stderr, "Non-JIT'd operation did not emit an error.\n");
204
+ if (match_data != NULL)
205
+ {
206
+ fprintf(stderr, "%d matches discovered by non-JIT'd regex:\n", matches);
207
+ dump_matches(stderr, matches, match_data);
208
+ fprintf(stderr, "\n");
209
+ }
210
+ }
211
+
212
+ if (errorcode_jit < 0)
213
+ {
214
+ print_error(stderr, errorcode_jit, "JIT'd operation emitted error %d:",
215
+ errorcode_jit);
216
+ }
217
+
218
+ if (matches_jit >= 0)
219
+ {
220
+ fprintf(stderr, "JIT'd operation did not emit an error.\n");
221
+ if (match_data_jit != NULL)
222
+ {
223
+ fprintf(stderr, "%d matches discovered by JIT'd regex:\n", matches_jit);
224
+ dump_matches(stderr, matches_jit, match_data_jit);
225
+ fprintf(stderr, "\n");
226
+ }
227
+ }
228
+
229
+ abort();
230
+ }
231
+ #endif /* SUPPORT_DIFF_FUZZ */
232
+ #endif /* SUPPORT_JIT */
233
+
234
+ /* This is the callout function. Its only purpose is to halt matching if there
235
+ are more than 100 callouts, as one way of stopping too much time being spent on
236
+ fruitless matches. The callout data is a pointer to the counter. */
237
+
238
+ static int callout_function(pcre2_callout_block *cb, void *callout_data)
239
+ {
240
+ (void)cb; /* Avoid unused parameter warning */
241
+ *((uint32_t *)callout_data) += 1;
242
+ return (*((uint32_t *)callout_data) > 100)? PCRE2_ERROR_CALLOUT : 0;
243
+ }
244
+
245
+ /* Putting in this apparently unnecessary prototype prevents gcc from giving a
246
+ "no previous prototype" warning when compiling at high warning level. */
247
+
248
+ int LLVMFuzzerInitialize(int *, char ***);
249
+
250
+ int LLVMFuzzerTestOneInput(unsigned char *, size_t);
251
+
252
+ int LLVMFuzzerInitialize(int *argc, char ***argv)
253
+ {
254
+ int rc;
255
+ struct rlimit rlim;
256
+ getrlimit(RLIMIT_STACK, &rlim);
257
+ rlim.rlim_cur = STACK_SIZE_MB * 1024 * 1024;
258
+ if (rlim.rlim_cur > rlim.rlim_max)
259
+ {
260
+ fprintf(stderr, "Hard stack size limit is too small (needed 8MiB)!\n");
261
+ _exit(1);
262
+ }
263
+ rc = setrlimit(RLIMIT_STACK, &rlim);
264
+ if (rc != 0)
265
+ {
266
+ fprintf(stderr, "Failed to expand stack size\n");
267
+ _exit(1);
268
+ }
269
+
270
+ (void)argc; /* Avoid "unused parameter" warnings */
271
+ (void)argv;
272
+ return 0;
273
+ }
274
+
275
+ /* Here's the driving function. */
276
+
277
+ int LLVMFuzzerTestOneInput(unsigned char *data, size_t size)
278
+ {
279
+ PCRE2_UCHAR *wdata;
280
+ PCRE2_UCHAR *newwdata = NULL;
281
+ uint32_t compile_options;
282
+ uint32_t match_options;
283
+ uint64_t random_options;
284
+ pcre2_match_data *match_data = NULL;
285
+ #ifdef SUPPORT_JIT
286
+ pcre2_match_data *match_data_jit = NULL;
287
+ #endif
288
+ pcre2_compile_context *compile_context = NULL;
289
+ pcre2_match_context *match_context = NULL;
290
+ size_t match_size;
291
+ int dfa_workspace[DFA_WORKSPACE_COUNT];
292
+
293
+ if (size < sizeof(random_options)) return -1;
294
+
295
+ random_options = *(uint64_t *)(data);
296
+ data += sizeof(random_options);
297
+ wdata = (PCRE2_UCHAR *)data;
298
+ size -= sizeof(random_options);
299
+ size /= PCRE2_CODE_UNIT_WIDTH / 8;
300
+
301
+ /* PCRE2 compiles quantified groups by replicating them. In certain cases of
302
+ very large quantifiers this can lead to unacceptably long JIT compile times. To
303
+ get around this, we scan the data string for large quantifiers that follow a
304
+ closing parenthesis, and reduce the value of the quantifier to 10, assuming
305
+ that this will make minimal difference to the detection of bugs.
306
+
307
+ Do the same for quantifiers that follow a closing square bracket, because
308
+ classes that contain a number of non-ascii characters can take a lot of time
309
+ when matching.
310
+
311
+ We have to make a copy of the input because oss-fuzz complains if we overwrite
312
+ the original. Start the scan at the second character so there can be a
313
+ lookbehind for a backslash, and end it before the end so that the next
314
+ character can be checked for an opening brace. */
315
+
316
+ if (size > 3)
317
+ {
318
+ newwdata = malloc(size * sizeof(PCRE2_UCHAR));
319
+ memcpy(newwdata, wdata, size * sizeof(PCRE2_UCHAR));
320
+ wdata = newwdata;
321
+
322
+ for (size_t i = 1; i < size - 2; i++)
323
+ {
324
+ size_t j;
325
+
326
+ if ((wdata[i] != ')' && wdata[i] != ']') || wdata[i-1] == '\\' ||
327
+ wdata[i+1] != '{')
328
+ continue;
329
+ i++; /* Points to '{' */
330
+
331
+ /* Loop for two values a quantifier. Offset i points to brace or comma at the
332
+ start of the loop.*/
333
+
334
+ for (int ii = 0; ii < 2; ii++)
335
+ {
336
+ int q = 0;
337
+
338
+ if (i >= size - 1) goto END_QSCAN; /* Can happen for , */
339
+
340
+ /* Ignore leading spaces */
341
+
342
+ while (wdata[i+1] == ' ' || wdata[i+1] == '\t')
343
+ {
344
+ i++;
345
+ if (i >= size - 1) goto END_QSCAN;
346
+ }
347
+
348
+ /* Scan for a number ending in brace or comma in the first iteration,
349
+ optionally preceded by space. */
350
+
351
+ for (j = i + 1; j < size && j < i + 7; j++)
352
+ {
353
+ if (wdata[j] == ' ' || wdata[j] == '\t')
354
+ {
355
+ j++;
356
+ while (j < size && (wdata[j] == ' ' || wdata[j] == '\t')) j++;
357
+ if (j >= size) goto OUTERLOOP;
358
+ if (wdata[j] != '}' && wdata[j] != ',') goto OUTERLOOP;
359
+ }
360
+ if (wdata[j] == '}' || (ii == 0 && wdata[j] == ',')) break;
361
+ if (wdata[j] < '0' || wdata[j] > '9')
362
+ {
363
+ j--; /* Ensure this character is checked next. The */
364
+ goto OUTERLOOP; /* string might be (e.g.) "){9){234}" */
365
+ }
366
+ q = q * 10 + wdata[j] - '0';
367
+ }
368
+
369
+ if (j >= size) goto END_QSCAN; /* End of data */
370
+
371
+ /* Hit ',' or '}' or read 6 digits. Six digits is a number > 65536 which is
372
+ the maximum quantifier. Leave such numbers alone. */
373
+
374
+ if (j >= i + 7 || q > 65535) goto OUTERLOOP;
375
+
376
+ /* Limit the quantifier size to 10 */
377
+
378
+ if (q > 10)
379
+ {
380
+ #ifdef STANDALONE
381
+ printf("Reduced quantifier value %d to 10.\n", q);
382
+ #endif
383
+ for (size_t k = i + 1; k < j; k++) wdata[k] = '0';
384
+ wdata[j - 2] = '1';
385
+ }
386
+
387
+ /* Advance to end of number and break if reached closing brace (continue
388
+ after comma, which is only valid in the first time round this loop). */
389
+
390
+ i = j;
391
+ if (wdata[i] == '}') break;
392
+ }
393
+
394
+ /* Continue along the data string */
395
+
396
+ OUTERLOOP:
397
+ i = j;
398
+ continue;
399
+ }
400
+ }
401
+ END_QSCAN:
402
+
403
+ /* Limiting the length of the subject for matching stops fruitless searches
404
+ in large trees taking too much time. */
405
+
406
+ match_size = (size > MAX_MATCH_SIZE)? MAX_MATCH_SIZE : size;
407
+
408
+ /* Create a compile context, and set a limit on the size of the compiled
409
+ pattern. This stops the fuzzer using vast amounts of memory. */
410
+
411
+ compile_context = pcre2_compile_context_create(NULL);
412
+ if (compile_context == NULL)
413
+ {
414
+ #ifdef STANDALONE
415
+ fprintf(stderr, "** Failed to create compile context block\n");
416
+ #endif
417
+ abort();
418
+ }
419
+ pcre2_set_max_pattern_compiled_length(compile_context, 10*1024*1024);
420
+
421
+ /* Ensure that all undefined option bits are zero (waste of time trying them)
422
+ and also that PCRE2_NO_UTF_CHECK is unset, as there is no guarantee that the
423
+ input is valid UTF. Also unset PCRE2_NEVER_UTF and PCRE2_NEVER_UCP as there is
424
+ no reason to disallow UTF and UCP. Force PCRE2_NEVER_BACKSLASH_C to be set
425
+ because \C in random patterns is highly likely to cause a crash. */
426
+
427
+ compile_options = ((random_options >> 32) & ALLOWED_COMPILE_OPTIONS) |
428
+ PCRE2_NEVER_BACKSLASH_C;
429
+ match_options = (((uint32_t)random_options) & ALLOWED_MATCH_OPTIONS) |
430
+ BASE_MATCH_OPTIONS;
431
+
432
+ /* Discard partial matching if PCRE2_ENDANCHORED is set, because they are not
433
+ allowed together and just give an immediate error return. */
434
+
435
+ if (((compile_options|match_options) & PCRE2_ENDANCHORED) != 0)
436
+ match_options &= ~(PCRE2_PARTIAL_HARD|PCRE2_PARTIAL_SOFT);
437
+
438
+ /* Do the compile with and without the options, and after a successful compile,
439
+ likewise do the match with and without the options. */
440
+
441
+ for (int i = 0; i < 2; i++)
442
+ {
443
+ uint32_t callout_count;
444
+ int errorcode;
445
+ #ifdef SUPPORT_JIT
446
+ int errorcode_jit;
447
+ #ifdef SUPPORT_DIFF_FUZZ
448
+ int matches = 0;
449
+ int matches_jit = 0;
450
+ #endif
451
+ #endif
452
+ PCRE2_SIZE erroroffset;
453
+ pcre2_code *code;
454
+
455
+ #ifdef STANDALONE
456
+ printf("\n");
457
+ print_compile_options(stdout, compile_options);
458
+ #endif
459
+
460
+ code = pcre2_compile((PCRE2_SPTR)wdata, (PCRE2_SIZE)size, compile_options,
461
+ &errorcode, &erroroffset, compile_context);
462
+
463
+ /* Compilation succeeded */
464
+
465
+ if (code != NULL)
466
+ {
467
+ int j;
468
+ uint32_t save_match_options = match_options;
469
+
470
+ /* Call JIT compile only if the compiled pattern is not too big. */
471
+
472
+ #ifdef SUPPORT_JIT
473
+ int jit_ret = -1;
474
+ if (((struct pcre2_real_code *)code)->blocksize <= JIT_SIZE_LIMIT)
475
+ {
476
+ #ifdef STANDALONE
477
+ printf("Compile succeeded; calling JIT compile\n");
478
+ #endif
479
+ jit_ret = pcre2_jit_compile(code, PCRE2_JIT_COMPLETE);
480
+ #ifdef STANDALONE
481
+ if (jit_ret < 0) printf("JIT compile error %d\n", jit_ret);
482
+ #endif
483
+ }
484
+ else
485
+ {
486
+ #ifdef STANDALONE
487
+ printf("Not calling JIT: compiled pattern is too long "
488
+ "(%ld bytes; limit=%d)\n",
489
+ ((struct pcre2_real_code *)code)->blocksize, JIT_SIZE_LIMIT);
490
+ #endif
491
+ }
492
+ #endif /* SUPPORT_JIT */
493
+
494
+ /* Create match data and context blocks only when we first need them. Set
495
+ low match and depth limits to avoid wasting too much searching large
496
+ pattern trees. Almost all matches are going to fail. */
497
+
498
+ if (match_data == NULL)
499
+ {
500
+ match_data = pcre2_match_data_create(32, NULL);
501
+ #ifdef SUPPORT_JIT
502
+ match_data_jit = pcre2_match_data_create(32, NULL);
503
+ if (match_data == NULL || match_data_jit == NULL)
504
+ #else
505
+ if (match_data == NULL)
506
+ #endif
507
+ {
508
+ #ifdef STANDALONE
509
+ fprintf(stderr, "** Failed to create match data block\n");
510
+ #endif
511
+ abort();
512
+ }
513
+ }
514
+
515
+ if (match_context == NULL)
516
+ {
517
+ match_context = pcre2_match_context_create(NULL);
518
+ if (match_context == NULL)
519
+ {
520
+ #ifdef STANDALONE
521
+ fprintf(stderr, "** Failed to create match context block\n");
522
+ #endif
523
+ abort();
524
+ }
525
+ (void)pcre2_set_match_limit(match_context, 100);
526
+ (void)pcre2_set_depth_limit(match_context, 100);
527
+ (void)pcre2_set_callout(match_context, callout_function, &callout_count);
528
+ }
529
+
530
+ /* Match twice, with and without options. */
531
+
532
+ #ifdef STANDALONE
533
+ printf("\n");
534
+ #endif
535
+ for (j = 0; j < 2; j++)
536
+ {
537
+ #ifdef STANDALONE
538
+ print_match_options(stdout, match_options);
539
+ #endif
540
+
541
+ callout_count = 0;
542
+ errorcode = pcre2_match(code, (PCRE2_SPTR)wdata, (PCRE2_SIZE)match_size, 0,
543
+ match_options, match_data, match_context);
544
+
545
+ #ifdef STANDALONE
546
+ if (errorcode >= 0) printf("Match returned %d\n", errorcode); else
547
+ print_error(stdout, errorcode, "Match failed: error %d: ", errorcode);
548
+ #endif
549
+
550
+ /* If JIT is enabled, do a JIT match and, if appropriately compiled, compare
551
+ with the interpreter. */
552
+
553
+ #ifdef SUPPORT_JIT
554
+ if (jit_ret >= 0)
555
+ {
556
+ #ifdef STANDALONE
557
+ printf("Matching with JIT\n");
558
+ #endif
559
+ callout_count = 0;
560
+ errorcode_jit = pcre2_match(code, (PCRE2_SPTR)wdata, (PCRE2_SIZE)match_size, 0,
561
+ match_options & ~PCRE2_NO_JIT, match_data_jit, match_context);
562
+
563
+ #ifdef STANDALONE
564
+ if (errorcode_jit >= 0)
565
+ printf("Match returned %d\n", errorcode_jit);
566
+ else
567
+ print_error(stdout, errorcode_jit, "JIT match failed: error %d: ",
568
+ errorcode_jit);
569
+ #else
570
+ (void)errorcode_jit; /* Avoid compiler warning */
571
+ #endif /* STANDALONE */
572
+
573
+ /* With differential matching enabled, compare with interpreter. */
574
+
575
+ #ifdef SUPPORT_DIFF_FUZZ
576
+ matches = errorcode;
577
+ matches_jit = errorcode_jit;
578
+
579
+ if (errorcode_jit != errorcode)
580
+ {
581
+ if (!(errorcode < 0 && errorcode_jit < 0) &&
582
+ errorcode != PCRE2_ERROR_MATCHLIMIT && errorcode != PCRE2_ERROR_CALLOUT &&
583
+ errorcode_jit != PCRE2_ERROR_MATCHLIMIT && errorcode_jit != PCRE2_ERROR_JIT_STACKLIMIT && errorcode_jit != PCRE2_ERROR_CALLOUT)
584
+ {
585
+ describe_failure("match errorcode comparison", wdata, size, compile_options, match_options, errorcode, errorcode_jit, matches, matches_jit, match_data, match_data_jit);
586
+ }
587
+ }
588
+ else
589
+ {
590
+ for (int index = 0; index < errorcode; index++)
591
+ {
592
+ PCRE2_UCHAR *bufferptr, *bufferptr_jit;
593
+ PCRE2_SIZE bufflen, bufflen_jit;
594
+
595
+ bufferptr = bufferptr_jit = NULL;
596
+ bufflen = bufflen_jit = 0;
597
+
598
+ errorcode = pcre2_substring_get_bynumber(match_data, (uint32_t) index, &bufferptr, &bufflen);
599
+ errorcode_jit = pcre2_substring_get_bynumber(match_data_jit, (uint32_t) index, &bufferptr_jit, &bufflen_jit);
600
+
601
+ if (errorcode != errorcode_jit)
602
+ {
603
+ describe_failure("match entry errorcode comparison", wdata, size,
604
+ compile_options, match_options, errorcode, errorcode_jit,
605
+ matches, matches_jit, match_data, match_data_jit);
606
+ }
607
+
608
+ if (errorcode >= 0)
609
+ {
610
+ if (bufflen != bufflen_jit)
611
+ {
612
+ describe_failure("match entry length comparison", wdata, size,
613
+ compile_options, match_options, errorcode, errorcode_jit,
614
+ matches, matches_jit, match_data, match_data_jit);
615
+ }
616
+
617
+ if (memcmp(bufferptr, bufferptr_jit, bufflen) != 0)
618
+ {
619
+ describe_failure("match entry content comparison", wdata, size,
620
+ compile_options, match_options, errorcode, errorcode_jit,
621
+ matches, matches_jit, match_data, match_data_jit);
622
+ }
623
+ }
624
+
625
+ pcre2_substring_free(bufferptr);
626
+ pcre2_substring_free(bufferptr_jit);
627
+ }
628
+ }
629
+ #endif /* SUPPORT_DIFF_FUZZ */
630
+ }
631
+ #endif /* SUPPORT_JIT */
632
+
633
+ if (match_options == BASE_MATCH_OPTIONS) break; /* Don't do same twice */
634
+ match_options = BASE_MATCH_OPTIONS; /* For second time */
635
+ }
636
+
637
+ /* Match with DFA twice, with and without options, but remove options that
638
+ are not allowed with DFA. */
639
+
640
+ match_options = save_match_options & ~BASE_MATCH_OPTIONS;
641
+
642
+ #ifdef STANDALONE
643
+ printf("\n");
644
+ #endif
645
+
646
+ for (j = 0; j < 2; j++)
647
+ {
648
+ #ifdef STANDALONE
649
+ printf("DFA match options %.8x =", match_options);
650
+ printf("%s%s%s%s%s%s%s%s%s\n",
651
+ ((match_options & PCRE2_ANCHORED) != 0)? " anchored" : "",
652
+ ((match_options & PCRE2_ENDANCHORED) != 0)? " endanchored" : "",
653
+ ((match_options & PCRE2_NO_UTF_CHECK) != 0)? " no_utf_check" : "",
654
+ ((match_options & PCRE2_NOTBOL) != 0)? " notbol" : "",
655
+ ((match_options & PCRE2_NOTEMPTY) != 0)? " notempty" : "",
656
+ ((match_options & PCRE2_NOTEMPTY_ATSTART) != 0)? " notempty_atstart" : "",
657
+ ((match_options & PCRE2_NOTEOL) != 0)? " noteol" : "",
658
+ ((match_options & PCRE2_PARTIAL_HARD) != 0)? " partial_hard" : "",
659
+ ((match_options & PCRE2_PARTIAL_SOFT) != 0)? " partial_soft" : "");
660
+ #endif
661
+
662
+ callout_count = 0;
663
+ errorcode = pcre2_dfa_match(code, (PCRE2_SPTR)wdata,
664
+ (PCRE2_SIZE)match_size, 0, match_options, match_data,
665
+ match_context, dfa_workspace, DFA_WORKSPACE_COUNT);
666
+
667
+ #ifdef STANDALONE
668
+ if (errorcode >= 0)
669
+ printf("Match returned %d\n", errorcode);
670
+ else
671
+ print_error(stdout, errorcode, "DFA match failed: error %d: ", errorcode);
672
+ #endif
673
+
674
+ if (match_options == 0) break; /* No point doing same twice */
675
+ match_options = 0; /* For second time */
676
+ }
677
+
678
+ match_options = save_match_options; /* Reset for the second compile */
679
+ pcre2_code_free(code);
680
+ }
681
+
682
+ /* Compilation failed */
683
+
684
+ else
685
+ {
686
+ #ifdef STANDALONE
687
+ print_error(stdout, errorcode, "Error %d at offset %lu: ", errorcode,
688
+ erroroffset);
689
+ #else
690
+ if (errorcode == PCRE2_ERROR_INTERNAL) abort();
691
+ #endif
692
+ }
693
+
694
+ if (compile_options == PCRE2_NEVER_BACKSLASH_C) break; /* Avoid same twice */
695
+ compile_options = PCRE2_NEVER_BACKSLASH_C; /* For second time */
696
+ }
697
+
698
+ /* Tidy up before exiting */
699
+
700
+ if (match_data != NULL) pcre2_match_data_free(match_data);
701
+ #ifdef SUPPORT_JIT
702
+ if (match_data_jit != NULL) pcre2_match_data_free(match_data_jit);
703
+ free(newwdata);
704
+ #endif
705
+ if (match_context != NULL) pcre2_match_context_free(match_context);
706
+ if (compile_context != NULL) pcre2_compile_context_free(compile_context);
707
+ return 0;
708
+ }
709
+
710
+
711
+ /* Optional main program. */
712
+
713
+ #ifdef STANDALONE
714
+ int main(int argc, char **argv)
715
+ {
716
+ LLVMFuzzerInitialize(&argc, &argv);
717
+
718
+ if (argc < 2)
719
+ {
720
+ printf("** No arguments given\n");
721
+ return 0;
722
+ }
723
+
724
+ for (int i = 1; i < argc; i++)
725
+ {
726
+ size_t filelen;
727
+ size_t readsize;
728
+ unsigned char *buffer;
729
+ FILE *f;
730
+
731
+ /* Handle a literal string. Copy to an exact size buffer so that checks for
732
+ overrunning work. */
733
+
734
+ if (argv[i][0] == '=')
735
+ {
736
+ readsize = strlen(argv[i]) - 1;
737
+ printf("------ <Literal> ------\n");
738
+ printf("Length = %lu\n", readsize);
739
+ printf("%.*s\n", (int)readsize, argv[i]+1);
740
+ buffer = (unsigned char *)malloc(readsize);
741
+ if (buffer == NULL)
742
+ printf("** Failed to allocate %lu bytes of memory\n", readsize);
743
+ else
744
+ {
745
+ memcpy(buffer, argv[i]+1, readsize);
746
+ LLVMFuzzerTestOneInput(buffer, readsize);
747
+ free(buffer);
748
+ }
749
+ continue;
750
+ }
751
+
752
+ /* Handle a string given in a file */
753
+
754
+ f = fopen(argv[i], "rb");
755
+ if (f == NULL)
756
+ {
757
+ printf("** Failed to open %s: %s\n", argv[i], strerror(errno));
758
+ continue;
759
+ }
760
+
761
+ printf("------ %s ------\n", argv[i]);
762
+
763
+ fseek(f, 0, SEEK_END);
764
+ filelen = ftell(f);
765
+ fseek(f, 0, SEEK_SET);
766
+
767
+ buffer = (unsigned char *)malloc(filelen);
768
+ if (buffer == NULL)
769
+ {
770
+ printf("** Failed to allocate %lu bytes of memory\n", filelen);
771
+ fclose(f);
772
+ continue;
773
+ }
774
+
775
+ readsize = fread(buffer, 1, filelen, f);
776
+ fclose(f);
777
+
778
+ if (readsize != filelen)
779
+ printf("** File size is %lu but fread() returned %lu\n", filelen, readsize);
780
+ else
781
+ {
782
+ printf("Length = %lu\n", filelen);
783
+ LLVMFuzzerTestOneInput(buffer, filelen);
784
+ }
785
+ free(buffer);
786
+ }
787
+
788
+ return 0;
789
+ }
790
+ #endif /* STANDALONE */
791
+
792
+ /* End */