sre_constants.py 6.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222
  1. #
  2. # Secret Labs' Regular Expression Engine
  3. #
  4. # various symbols used by the regular expression engine.
  5. # run this script to update the _sre include files!
  6. #
  7. # Copyright (c) 1998-2001 by Secret Labs AB. All rights reserved.
  8. #
  9. # See the sre.py file for information on usage and redistribution.
  10. #
  11. """Internal support module for sre"""
  12. # update when constants are added or removed
  13. MAGIC = 20140917
  14. from _sre import MAXREPEAT, MAXGROUPS
  15. # SRE standard exception (access as sre.error)
  16. # should this really be here?
  17. class error(Exception):
  18. def __init__(self, msg, pattern=None, pos=None):
  19. self.msg = msg
  20. self.pattern = pattern
  21. self.pos = pos
  22. if pattern is not None and pos is not None:
  23. msg = '%s at position %d' % (msg, pos)
  24. if isinstance(pattern, str):
  25. newline = '\n'
  26. else:
  27. newline = b'\n'
  28. self.lineno = pattern.count(newline, 0, pos) + 1
  29. self.colno = pos - pattern.rfind(newline, 0, pos)
  30. if newline in pattern:
  31. msg = '%s (line %d, column %d)' % (msg, self.lineno, self.colno)
  32. else:
  33. self.lineno = self.colno = None
  34. super().__init__(msg)
  35. class _NamedIntConstant(int):
  36. def __new__(cls, value, name):
  37. self = super(_NamedIntConstant, cls).__new__(cls, value)
  38. self.name = name
  39. return self
  40. def __str__(self):
  41. return self.name
  42. __repr__ = __str__
  43. MAXREPEAT = _NamedIntConstant(MAXREPEAT, 'MAXREPEAT')
  44. def _makecodes(names):
  45. names = names.strip().split()
  46. items = [_NamedIntConstant(i, name) for i, name in enumerate(names)]
  47. globals().update({item.name: item for item in items})
  48. return items
  49. # operators
  50. # failure=0 success=1 (just because it looks better that way :-)
  51. OPCODES = _makecodes("""
  52. FAILURE SUCCESS
  53. ANY ANY_ALL
  54. ASSERT ASSERT_NOT
  55. AT
  56. BRANCH
  57. CALL
  58. CATEGORY
  59. CHARSET BIGCHARSET
  60. GROUPREF GROUPREF_EXISTS GROUPREF_IGNORE
  61. IN IN_IGNORE
  62. INFO
  63. JUMP
  64. LITERAL LITERAL_IGNORE
  65. MARK
  66. MAX_UNTIL
  67. MIN_UNTIL
  68. NOT_LITERAL NOT_LITERAL_IGNORE
  69. NEGATE
  70. RANGE
  71. REPEAT
  72. REPEAT_ONE
  73. SUBPATTERN
  74. MIN_REPEAT_ONE
  75. RANGE_IGNORE
  76. MIN_REPEAT MAX_REPEAT
  77. """)
  78. del OPCODES[-2:] # remove MIN_REPEAT and MAX_REPEAT
  79. # positions
  80. ATCODES = _makecodes("""
  81. AT_BEGINNING AT_BEGINNING_LINE AT_BEGINNING_STRING
  82. AT_BOUNDARY AT_NON_BOUNDARY
  83. AT_END AT_END_LINE AT_END_STRING
  84. AT_LOC_BOUNDARY AT_LOC_NON_BOUNDARY
  85. AT_UNI_BOUNDARY AT_UNI_NON_BOUNDARY
  86. """)
  87. # categories
  88. CHCODES = _makecodes("""
  89. CATEGORY_DIGIT CATEGORY_NOT_DIGIT
  90. CATEGORY_SPACE CATEGORY_NOT_SPACE
  91. CATEGORY_WORD CATEGORY_NOT_WORD
  92. CATEGORY_LINEBREAK CATEGORY_NOT_LINEBREAK
  93. CATEGORY_LOC_WORD CATEGORY_LOC_NOT_WORD
  94. CATEGORY_UNI_DIGIT CATEGORY_UNI_NOT_DIGIT
  95. CATEGORY_UNI_SPACE CATEGORY_UNI_NOT_SPACE
  96. CATEGORY_UNI_WORD CATEGORY_UNI_NOT_WORD
  97. CATEGORY_UNI_LINEBREAK CATEGORY_UNI_NOT_LINEBREAK
  98. """)
  99. # replacement operations for "ignore case" mode
  100. OP_IGNORE = {
  101. GROUPREF: GROUPREF_IGNORE,
  102. IN: IN_IGNORE,
  103. LITERAL: LITERAL_IGNORE,
  104. NOT_LITERAL: NOT_LITERAL_IGNORE,
  105. RANGE: RANGE_IGNORE,
  106. }
  107. AT_MULTILINE = {
  108. AT_BEGINNING: AT_BEGINNING_LINE,
  109. AT_END: AT_END_LINE
  110. }
  111. AT_LOCALE = {
  112. AT_BOUNDARY: AT_LOC_BOUNDARY,
  113. AT_NON_BOUNDARY: AT_LOC_NON_BOUNDARY
  114. }
  115. AT_UNICODE = {
  116. AT_BOUNDARY: AT_UNI_BOUNDARY,
  117. AT_NON_BOUNDARY: AT_UNI_NON_BOUNDARY
  118. }
  119. CH_LOCALE = {
  120. CATEGORY_DIGIT: CATEGORY_DIGIT,
  121. CATEGORY_NOT_DIGIT: CATEGORY_NOT_DIGIT,
  122. CATEGORY_SPACE: CATEGORY_SPACE,
  123. CATEGORY_NOT_SPACE: CATEGORY_NOT_SPACE,
  124. CATEGORY_WORD: CATEGORY_LOC_WORD,
  125. CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD,
  126. CATEGORY_LINEBREAK: CATEGORY_LINEBREAK,
  127. CATEGORY_NOT_LINEBREAK: CATEGORY_NOT_LINEBREAK
  128. }
  129. CH_UNICODE = {
  130. CATEGORY_DIGIT: CATEGORY_UNI_DIGIT,
  131. CATEGORY_NOT_DIGIT: CATEGORY_UNI_NOT_DIGIT,
  132. CATEGORY_SPACE: CATEGORY_UNI_SPACE,
  133. CATEGORY_NOT_SPACE: CATEGORY_UNI_NOT_SPACE,
  134. CATEGORY_WORD: CATEGORY_UNI_WORD,
  135. CATEGORY_NOT_WORD: CATEGORY_UNI_NOT_WORD,
  136. CATEGORY_LINEBREAK: CATEGORY_UNI_LINEBREAK,
  137. CATEGORY_NOT_LINEBREAK: CATEGORY_UNI_NOT_LINEBREAK
  138. }
  139. # flags
  140. SRE_FLAG_TEMPLATE = 1 # template mode (disable backtracking)
  141. SRE_FLAG_IGNORECASE = 2 # case insensitive
  142. SRE_FLAG_LOCALE = 4 # honour system locale
  143. SRE_FLAG_MULTILINE = 8 # treat target as multiline string
  144. SRE_FLAG_DOTALL = 16 # treat target as a single string
  145. SRE_FLAG_UNICODE = 32 # use unicode "locale"
  146. SRE_FLAG_VERBOSE = 64 # ignore whitespace and comments
  147. SRE_FLAG_DEBUG = 128 # debugging
  148. SRE_FLAG_ASCII = 256 # use ascii "locale"
  149. # flags for INFO primitive
  150. SRE_INFO_PREFIX = 1 # has prefix
  151. SRE_INFO_LITERAL = 2 # entire pattern is literal (given by prefix)
  152. SRE_INFO_CHARSET = 4 # pattern starts with character from given set
  153. if __name__ == "__main__":
  154. def dump(f, d, prefix):
  155. items = sorted(d)
  156. for item in items:
  157. f.write("#define %s_%s %d\n" % (prefix, item, item))
  158. with open("sre_constants.h", "w") as f:
  159. f.write("""\
  160. /*
  161. * Secret Labs' Regular Expression Engine
  162. *
  163. * regular expression matching engine
  164. *
  165. * NOTE: This file is generated by sre_constants.py. If you need
  166. * to change anything in here, edit sre_constants.py and run it.
  167. *
  168. * Copyright (c) 1997-2001 by Secret Labs AB. All rights reserved.
  169. *
  170. * See the _sre.c file for information on usage and redistribution.
  171. */
  172. """)
  173. f.write("#define SRE_MAGIC %d\n" % MAGIC)
  174. dump(f, OPCODES, "SRE_OP")
  175. dump(f, ATCODES, "SRE")
  176. dump(f, CHCODES, "SRE")
  177. f.write("#define SRE_FLAG_TEMPLATE %d\n" % SRE_FLAG_TEMPLATE)
  178. f.write("#define SRE_FLAG_IGNORECASE %d\n" % SRE_FLAG_IGNORECASE)
  179. f.write("#define SRE_FLAG_LOCALE %d\n" % SRE_FLAG_LOCALE)
  180. f.write("#define SRE_FLAG_MULTILINE %d\n" % SRE_FLAG_MULTILINE)
  181. f.write("#define SRE_FLAG_DOTALL %d\n" % SRE_FLAG_DOTALL)
  182. f.write("#define SRE_FLAG_UNICODE %d\n" % SRE_FLAG_UNICODE)
  183. f.write("#define SRE_FLAG_VERBOSE %d\n" % SRE_FLAG_VERBOSE)
  184. f.write("#define SRE_FLAG_DEBUG %d\n" % SRE_FLAG_DEBUG)
  185. f.write("#define SRE_FLAG_ASCII %d\n" % SRE_FLAG_ASCII)
  186. f.write("#define SRE_INFO_PREFIX %d\n" % SRE_INFO_PREFIX)
  187. f.write("#define SRE_INFO_LITERAL %d\n" % SRE_INFO_LITERAL)
  188. f.write("#define SRE_INFO_CHARSET %d\n" % SRE_INFO_CHARSET)
  189. print("done")