Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 1 | # |
| 2 | # Secret Labs' Regular Expression Engine |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 3 | # |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 4 | # convert re-style regular expression to sre pattern |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 5 | # |
Fredrik Lundh | 770617b | 2001-01-14 15:06:11 +0000 | [diff] [blame] | 6 | # Copyright (c) 1998-2001 by Secret Labs AB. All rights reserved. |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 7 | # |
Fredrik Lundh | 29c4ba9 | 2000-08-01 18:20:07 +0000 | [diff] [blame] | 8 | # See the sre.py file for information on usage and redistribution. |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 9 | # |
| 10 | |
Fred Drake | b8f2274 | 2001-09-04 19:10:20 +0000 | [diff] [blame] | 11 | """Internal support module for sre""" |
| 12 | |
Fredrik Lundh | 470ea5a | 2001-01-14 21:00:44 +0000 | [diff] [blame] | 13 | # XXX: show string offset and offending character for all errors |
| 14 | |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 15 | from sre_constants import * |
| 16 | |
| 17 | SPECIAL_CHARS = ".\\[{()*+?^$|" |
Fredrik Lundh | 143328b | 2000-09-02 11:03:34 +0000 | [diff] [blame] | 18 | REPEAT_CHARS = "*+?{" |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 19 | |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 20 | DIGITS = frozenset("0123456789") |
Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 21 | |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 22 | OCTDIGITS = frozenset("01234567") |
| 23 | HEXDIGITS = frozenset("0123456789abcdefABCDEF") |
Serhiy Storchaka | a54aae0 | 2015-03-24 22:58:14 +0200 | [diff] [blame] | 24 | ASCIILETTERS = frozenset("abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ") |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 25 | |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 26 | WHITESPACE = frozenset(" \t\n\r\v\f") |
| 27 | |
Raymond Hettinger | df1b699 | 2014-11-09 15:56:33 -0800 | [diff] [blame] | 28 | _REPEATCODES = frozenset({MIN_REPEAT, MAX_REPEAT}) |
| 29 | _UNITCODES = frozenset({ANY, RANGE, IN, LITERAL, NOT_LITERAL, CATEGORY}) |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 30 | |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 31 | ESCAPES = { |
Fredrik Lundh | f2989b2 | 2001-02-18 12:05:16 +0000 | [diff] [blame] | 32 | r"\a": (LITERAL, ord("\a")), |
| 33 | r"\b": (LITERAL, ord("\b")), |
| 34 | r"\f": (LITERAL, ord("\f")), |
| 35 | r"\n": (LITERAL, ord("\n")), |
| 36 | r"\r": (LITERAL, ord("\r")), |
| 37 | r"\t": (LITERAL, ord("\t")), |
| 38 | r"\v": (LITERAL, ord("\v")), |
Fredrik Lundh | 0640e11 | 2000-06-30 13:55:15 +0000 | [diff] [blame] | 39 | r"\\": (LITERAL, ord("\\")) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 40 | } |
| 41 | |
| 42 | CATEGORIES = { |
Fredrik Lundh | 770617b | 2001-01-14 15:06:11 +0000 | [diff] [blame] | 43 | r"\A": (AT, AT_BEGINNING_STRING), # start of string |
Fredrik Lundh | 01016fe | 2000-06-30 00:27:46 +0000 | [diff] [blame] | 44 | r"\b": (AT, AT_BOUNDARY), |
| 45 | r"\B": (AT, AT_NON_BOUNDARY), |
| 46 | r"\d": (IN, [(CATEGORY, CATEGORY_DIGIT)]), |
| 47 | r"\D": (IN, [(CATEGORY, CATEGORY_NOT_DIGIT)]), |
| 48 | r"\s": (IN, [(CATEGORY, CATEGORY_SPACE)]), |
| 49 | r"\S": (IN, [(CATEGORY, CATEGORY_NOT_SPACE)]), |
| 50 | r"\w": (IN, [(CATEGORY, CATEGORY_WORD)]), |
| 51 | r"\W": (IN, [(CATEGORY, CATEGORY_NOT_WORD)]), |
Fredrik Lundh | 770617b | 2001-01-14 15:06:11 +0000 | [diff] [blame] | 52 | r"\Z": (AT, AT_END_STRING), # end of string |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 53 | } |
| 54 | |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 55 | FLAGS = { |
Fredrik Lundh | 436c3d58 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 56 | # standard flags |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 57 | "i": SRE_FLAG_IGNORECASE, |
| 58 | "L": SRE_FLAG_LOCALE, |
| 59 | "m": SRE_FLAG_MULTILINE, |
| 60 | "s": SRE_FLAG_DOTALL, |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 61 | "x": SRE_FLAG_VERBOSE, |
Fredrik Lundh | 436c3d58 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 62 | # extensions |
Antoine Pitrou | fd03645 | 2008-08-19 17:56:33 +0000 | [diff] [blame] | 63 | "a": SRE_FLAG_ASCII, |
Fredrik Lundh | 436c3d58 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 64 | "t": SRE_FLAG_TEMPLATE, |
| 65 | "u": SRE_FLAG_UNICODE, |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 66 | } |
| 67 | |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 68 | GLOBAL_FLAGS = (SRE_FLAG_ASCII | SRE_FLAG_LOCALE | SRE_FLAG_UNICODE | |
| 69 | SRE_FLAG_DEBUG | SRE_FLAG_TEMPLATE) |
| 70 | |
| 71 | class Verbose(Exception): |
| 72 | pass |
| 73 | |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 74 | class Pattern: |
| 75 | # master pattern object. keeps track of global attributes |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 76 | def __init__(self): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 77 | self.flags = 0 |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 78 | self.groupdict = {} |
Serhiy Storchaka | b5d0a21 | 2015-11-05 17:49:26 +0200 | [diff] [blame] | 79 | self.groupwidths = [None] # group 0 |
Serhiy Storchaka | 4eea62f | 2015-02-21 10:07:35 +0200 | [diff] [blame] | 80 | self.lookbehindgroups = None |
| 81 | @property |
| 82 | def groups(self): |
Serhiy Storchaka | b5d0a21 | 2015-11-05 17:49:26 +0200 | [diff] [blame] | 83 | return len(self.groupwidths) |
Fredrik Lundh | ebc37b2 | 2000-10-28 19:30:41 +0000 | [diff] [blame] | 84 | def opengroup(self, name=None): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 85 | gid = self.groups |
Serhiy Storchaka | b5d0a21 | 2015-11-05 17:49:26 +0200 | [diff] [blame] | 86 | self.groupwidths.append(None) |
Serhiy Storchaka | 9baa5b2 | 2014-09-29 22:49:23 +0300 | [diff] [blame] | 87 | if self.groups > MAXGROUPS: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 88 | raise error("too many groups") |
Raymond Hettinger | f13eb55 | 2002-06-02 00:40:05 +0000 | [diff] [blame] | 89 | if name is not None: |
Tim Peters | 7533587 | 2001-11-03 19:35:43 +0000 | [diff] [blame] | 90 | ogid = self.groupdict.get(name, None) |
| 91 | if ogid is not None: |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 92 | raise error("redefinition of group name %r as group %d; " |
| 93 | "was group %d" % (name, gid, ogid)) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 94 | self.groupdict[name] = gid |
| 95 | return gid |
Serhiy Storchaka | 4eea62f | 2015-02-21 10:07:35 +0200 | [diff] [blame] | 96 | def closegroup(self, gid, p): |
Serhiy Storchaka | b5d0a21 | 2015-11-05 17:49:26 +0200 | [diff] [blame] | 97 | self.groupwidths[gid] = p.getwidth() |
Fredrik Lundh | ebc37b2 | 2000-10-28 19:30:41 +0000 | [diff] [blame] | 98 | def checkgroup(self, gid): |
Serhiy Storchaka | b5d0a21 | 2015-11-05 17:49:26 +0200 | [diff] [blame] | 99 | return gid < self.groups and self.groupwidths[gid] is not None |
Serhiy Storchaka | 4eea62f | 2015-02-21 10:07:35 +0200 | [diff] [blame] | 100 | |
| 101 | def checklookbehindgroup(self, gid, source): |
| 102 | if self.lookbehindgroups is not None: |
| 103 | if not self.checkgroup(gid): |
| 104 | raise source.error('cannot refer to an open group') |
| 105 | if gid >= self.lookbehindgroups: |
| 106 | raise source.error('cannot refer to group defined in the same ' |
| 107 | 'lookbehind subpattern') |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 108 | |
| 109 | class SubPattern: |
| 110 | # a subpattern, in intermediate form |
| 111 | def __init__(self, pattern, data=None): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 112 | self.pattern = pattern |
Raymond Hettinger | f13eb55 | 2002-06-02 00:40:05 +0000 | [diff] [blame] | 113 | if data is None: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 114 | data = [] |
| 115 | self.data = data |
| 116 | self.width = None |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 117 | |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 118 | def dump(self, level=0): |
Serhiy Storchaka | 44dae8b | 2014-09-21 22:47:55 +0300 | [diff] [blame] | 119 | nl = True |
Guido van Rossum | 1325790 | 2007-06-07 23:15:56 +0000 | [diff] [blame] | 120 | seqtypes = (tuple, list) |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 121 | for op, av in self.data: |
Serhiy Storchaka | c7f7d38 | 2014-11-09 20:48:36 +0200 | [diff] [blame] | 122 | print(level*" " + str(op), end='') |
Serhiy Storchaka | ab14088 | 2014-11-11 21:13:28 +0200 | [diff] [blame] | 123 | if op is IN: |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 124 | # member sublanguage |
Serhiy Storchaka | 44dae8b | 2014-09-21 22:47:55 +0300 | [diff] [blame] | 125 | print() |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 126 | for op, a in av: |
Serhiy Storchaka | c7f7d38 | 2014-11-09 20:48:36 +0200 | [diff] [blame] | 127 | print((level+1)*" " + str(op), a) |
Serhiy Storchaka | ab14088 | 2014-11-11 21:13:28 +0200 | [diff] [blame] | 128 | elif op is BRANCH: |
Serhiy Storchaka | 44dae8b | 2014-09-21 22:47:55 +0300 | [diff] [blame] | 129 | print() |
| 130 | for i, a in enumerate(av[1]): |
| 131 | if i: |
Serhiy Storchaka | c7f7d38 | 2014-11-09 20:48:36 +0200 | [diff] [blame] | 132 | print(level*" " + "OR") |
Serhiy Storchaka | 44dae8b | 2014-09-21 22:47:55 +0300 | [diff] [blame] | 133 | a.dump(level+1) |
Serhiy Storchaka | ab14088 | 2014-11-11 21:13:28 +0200 | [diff] [blame] | 134 | elif op is GROUPREF_EXISTS: |
Serhiy Storchaka | 44dae8b | 2014-09-21 22:47:55 +0300 | [diff] [blame] | 135 | condgroup, item_yes, item_no = av |
| 136 | print('', condgroup) |
| 137 | item_yes.dump(level+1) |
| 138 | if item_no: |
Serhiy Storchaka | c7f7d38 | 2014-11-09 20:48:36 +0200 | [diff] [blame] | 139 | print(level*" " + "ELSE") |
Serhiy Storchaka | 44dae8b | 2014-09-21 22:47:55 +0300 | [diff] [blame] | 140 | item_no.dump(level+1) |
Guido van Rossum | 1325790 | 2007-06-07 23:15:56 +0000 | [diff] [blame] | 141 | elif isinstance(av, seqtypes): |
Serhiy Storchaka | 44dae8b | 2014-09-21 22:47:55 +0300 | [diff] [blame] | 142 | nl = False |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 143 | for a in av: |
| 144 | if isinstance(a, SubPattern): |
Serhiy Storchaka | 44dae8b | 2014-09-21 22:47:55 +0300 | [diff] [blame] | 145 | if not nl: |
| 146 | print() |
| 147 | a.dump(level+1) |
| 148 | nl = True |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 149 | else: |
Serhiy Storchaka | 44dae8b | 2014-09-21 22:47:55 +0300 | [diff] [blame] | 150 | if not nl: |
| 151 | print(' ', end='') |
| 152 | print(a, end='') |
| 153 | nl = False |
| 154 | if not nl: |
| 155 | print() |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 156 | else: |
Serhiy Storchaka | 44dae8b | 2014-09-21 22:47:55 +0300 | [diff] [blame] | 157 | print('', av) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 158 | def __repr__(self): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 159 | return repr(self.data) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 160 | def __len__(self): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 161 | return len(self.data) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 162 | def __delitem__(self, index): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 163 | del self.data[index] |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 164 | def __getitem__(self, index): |
Thomas Wouters | 902d6eb | 2007-01-09 23:18:33 +0000 | [diff] [blame] | 165 | if isinstance(index, slice): |
| 166 | return SubPattern(self.pattern, self.data[index]) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 167 | return self.data[index] |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 168 | def __setitem__(self, index, code): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 169 | self.data[index] = code |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 170 | def insert(self, index, code): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 171 | self.data.insert(index, code) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 172 | def append(self, code): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 173 | self.data.append(code) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 174 | def getwidth(self): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 175 | # determine the width (min, max) for this subpattern |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 176 | if self.width is not None: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 177 | return self.width |
Guido van Rossum | e2a383d | 2007-01-15 16:59:06 +0000 | [diff] [blame] | 178 | lo = hi = 0 |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 179 | for op, av in self.data: |
| 180 | if op is BRANCH: |
Serhiy Storchaka | 9d96542 | 2013-08-19 22:50:54 +0300 | [diff] [blame] | 181 | i = MAXREPEAT - 1 |
Fredrik Lundh | 2f2c67d | 2000-08-01 21:05:41 +0000 | [diff] [blame] | 182 | j = 0 |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 183 | for av in av[1]: |
Fredrik Lundh | 2f2c67d | 2000-08-01 21:05:41 +0000 | [diff] [blame] | 184 | l, h = av.getwidth() |
| 185 | i = min(i, l) |
Fredrik Lundh | e186983 | 2000-08-01 22:47:49 +0000 | [diff] [blame] | 186 | j = max(j, h) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 187 | lo = lo + i |
| 188 | hi = hi + j |
| 189 | elif op is CALL: |
| 190 | i, j = av.getwidth() |
| 191 | lo = lo + i |
| 192 | hi = hi + j |
| 193 | elif op is SUBPATTERN: |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 194 | i, j = av[-1].getwidth() |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 195 | lo = lo + i |
| 196 | hi = hi + j |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 197 | elif op in _REPEATCODES: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 198 | i, j = av[2].getwidth() |
Serhiy Storchaka | 9d96542 | 2013-08-19 22:50:54 +0300 | [diff] [blame] | 199 | lo = lo + i * av[0] |
| 200 | hi = hi + j * av[1] |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 201 | elif op in _UNITCODES: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 202 | lo = lo + 1 |
| 203 | hi = hi + 1 |
Serhiy Storchaka | 4eea62f | 2015-02-21 10:07:35 +0200 | [diff] [blame] | 204 | elif op is GROUPREF: |
Serhiy Storchaka | b5d0a21 | 2015-11-05 17:49:26 +0200 | [diff] [blame] | 205 | i, j = self.pattern.groupwidths[av] |
Serhiy Storchaka | 4eea62f | 2015-02-21 10:07:35 +0200 | [diff] [blame] | 206 | lo = lo + i |
| 207 | hi = hi + j |
| 208 | elif op is GROUPREF_EXISTS: |
| 209 | i, j = av[1].getwidth() |
| 210 | if av[2] is not None: |
| 211 | l, h = av[2].getwidth() |
| 212 | i = min(i, l) |
| 213 | j = max(j, h) |
| 214 | else: |
| 215 | i = 0 |
| 216 | lo = lo + i |
| 217 | hi = hi + j |
| 218 | elif op is SUCCESS: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 219 | break |
Serhiy Storchaka | 9d96542 | 2013-08-19 22:50:54 +0300 | [diff] [blame] | 220 | self.width = min(lo, MAXREPEAT - 1), min(hi, MAXREPEAT) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 221 | return self.width |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 222 | |
| 223 | class Tokenizer: |
| 224 | def __init__(self, string): |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 225 | self.istext = isinstance(string, str) |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 226 | self.string = string |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 227 | if not self.istext: |
| 228 | string = str(string, 'latin1') |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 229 | self.decoded_string = string |
Fredrik Lundh | c13222c | 2000-07-01 23:49:14 +0000 | [diff] [blame] | 230 | self.index = 0 |
Serhiy Storchaka | b99c132 | 2014-11-10 14:38:16 +0200 | [diff] [blame] | 231 | self.next = None |
Fredrik Lundh | c13222c | 2000-07-01 23:49:14 +0000 | [diff] [blame] | 232 | self.__next() |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 233 | def __next(self): |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 234 | index = self.index |
| 235 | try: |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 236 | char = self.decoded_string[index] |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 237 | except IndexError: |
Fredrik Lundh | c13222c | 2000-07-01 23:49:14 +0000 | [diff] [blame] | 238 | self.next = None |
| 239 | return |
Guido van Rossum | 75a902d | 2007-10-19 22:06:24 +0000 | [diff] [blame] | 240 | if char == "\\": |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 241 | index += 1 |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 242 | try: |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 243 | char += self.decoded_string[index] |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 244 | except IndexError: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 245 | raise error("bad escape (end of pattern)", |
Serhiy Storchaka | 1b2004f | 2014-11-10 18:28:53 +0200 | [diff] [blame] | 246 | self.string, len(self.string) - 1) from None |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 247 | self.index = index + 1 |
Fredrik Lundh | c13222c | 2000-07-01 23:49:14 +0000 | [diff] [blame] | 248 | self.next = char |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 249 | def match(self, char): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 250 | if char == self.next: |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 251 | self.__next() |
| 252 | return True |
| 253 | return False |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 254 | def get(self): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 255 | this = self.next |
Fredrik Lundh | c13222c | 2000-07-01 23:49:14 +0000 | [diff] [blame] | 256 | self.__next() |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 257 | return this |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 258 | def getwhile(self, n, charset): |
| 259 | result = '' |
| 260 | for _ in range(n): |
| 261 | c = self.next |
| 262 | if c not in charset: |
| 263 | break |
| 264 | result += c |
| 265 | self.__next() |
| 266 | return result |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 267 | def getuntil(self, terminator): |
| 268 | result = '' |
| 269 | while True: |
| 270 | c = self.next |
| 271 | self.__next() |
| 272 | if c is None: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 273 | if not result: |
| 274 | raise self.error("missing group name") |
| 275 | raise self.error("missing %s, unterminated name" % terminator, |
| 276 | len(result)) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 277 | if c == terminator: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 278 | if not result: |
| 279 | raise self.error("missing group name", 1) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 280 | break |
| 281 | result += c |
| 282 | return result |
Serhiy Storchaka | bd48d27 | 2016-09-11 12:50:02 +0300 | [diff] [blame] | 283 | @property |
| 284 | def pos(self): |
| 285 | return self.index - len(self.next or '') |
Fredrik Lundh | c13222c | 2000-07-01 23:49:14 +0000 | [diff] [blame] | 286 | def tell(self): |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 287 | return self.index - len(self.next or '') |
Fredrik Lundh | c13222c | 2000-07-01 23:49:14 +0000 | [diff] [blame] | 288 | def seek(self, index): |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 289 | self.index = index |
| 290 | self.__next() |
| 291 | |
| 292 | def error(self, msg, offset=0): |
| 293 | return error(msg, self.string, self.tell() - offset) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 294 | |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 295 | def _class_escape(source, escape): |
| 296 | # handle escape code inside character class |
| 297 | code = ESCAPES.get(escape) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 298 | if code: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 299 | return code |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 300 | code = CATEGORIES.get(escape) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 301 | if code and code[0] is IN: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 302 | return code |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 303 | try: |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 304 | c = escape[1:2] |
| 305 | if c == "x": |
Fredrik Lundh | 0c4fdba | 2000-08-31 22:57:55 +0000 | [diff] [blame] | 306 | # hexadecimal escape (exactly two digits) |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 307 | escape += source.getwhile(2, HEXDIGITS) |
| 308 | if len(escape) != 4: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 309 | raise source.error("incomplete escape %s" % escape, len(escape)) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 310 | return LITERAL, int(escape[2:], 16) |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 311 | elif c == "u" and source.istext: |
| 312 | # unicode escape (exactly four digits) |
| 313 | escape += source.getwhile(4, HEXDIGITS) |
| 314 | if len(escape) != 6: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 315 | raise source.error("incomplete escape %s" % escape, len(escape)) |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 316 | return LITERAL, int(escape[2:], 16) |
| 317 | elif c == "U" and source.istext: |
| 318 | # unicode escape (exactly eight digits) |
| 319 | escape += source.getwhile(8, HEXDIGITS) |
| 320 | if len(escape) != 10: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 321 | raise source.error("incomplete escape %s" % escape, len(escape)) |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 322 | c = int(escape[2:], 16) |
| 323 | chr(c) # raise ValueError for invalid code |
| 324 | return LITERAL, c |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 325 | elif c in OCTDIGITS: |
Fredrik Lundh | 0c4fdba | 2000-08-31 22:57:55 +0000 | [diff] [blame] | 326 | # octal escape (up to three digits) |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 327 | escape += source.getwhile(2, OCTDIGITS) |
Serhiy Storchaka | c563caf | 2014-09-23 23:22:41 +0300 | [diff] [blame] | 328 | c = int(escape[1:], 8) |
| 329 | if c > 0o377: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 330 | raise source.error('octal escape value %s outside of ' |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 331 | 'range 0-0o377' % escape, len(escape)) |
Serhiy Storchaka | c563caf | 2014-09-23 23:22:41 +0300 | [diff] [blame] | 332 | return LITERAL, c |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 333 | elif c in DIGITS: |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 334 | raise ValueError |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 335 | if len(escape) == 2: |
Serhiy Storchaka | a54aae0 | 2015-03-24 22:58:14 +0200 | [diff] [blame] | 336 | if c in ASCIILETTERS: |
Serhiy Storchaka | 9bd85b8 | 2016-06-11 19:15:00 +0300 | [diff] [blame] | 337 | raise source.error('bad escape %s' % escape, len(escape)) |
Fredrik Lundh | 0640e11 | 2000-06-30 13:55:15 +0000 | [diff] [blame] | 338 | return LITERAL, ord(escape[1]) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 339 | except ValueError: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 340 | pass |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 341 | raise source.error("bad escape %s" % escape, len(escape)) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 342 | |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 343 | def _escape(source, escape, state): |
| 344 | # handle escape code in expression |
| 345 | code = CATEGORIES.get(escape) |
| 346 | if code: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 347 | return code |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 348 | code = ESCAPES.get(escape) |
| 349 | if code: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 350 | return code |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 351 | try: |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 352 | c = escape[1:2] |
| 353 | if c == "x": |
Fredrik Lundh | 0c4fdba | 2000-08-31 22:57:55 +0000 | [diff] [blame] | 354 | # hexadecimal escape |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 355 | escape += source.getwhile(2, HEXDIGITS) |
Fredrik Lundh | 143328b | 2000-09-02 11:03:34 +0000 | [diff] [blame] | 356 | if len(escape) != 4: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 357 | raise source.error("incomplete escape %s" % escape, len(escape)) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 358 | return LITERAL, int(escape[2:], 16) |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 359 | elif c == "u" and source.istext: |
| 360 | # unicode escape (exactly four digits) |
| 361 | escape += source.getwhile(4, HEXDIGITS) |
| 362 | if len(escape) != 6: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 363 | raise source.error("incomplete escape %s" % escape, len(escape)) |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 364 | return LITERAL, int(escape[2:], 16) |
| 365 | elif c == "U" and source.istext: |
| 366 | # unicode escape (exactly eight digits) |
| 367 | escape += source.getwhile(8, HEXDIGITS) |
| 368 | if len(escape) != 10: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 369 | raise source.error("incomplete escape %s" % escape, len(escape)) |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 370 | c = int(escape[2:], 16) |
| 371 | chr(c) # raise ValueError for invalid code |
| 372 | return LITERAL, c |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 373 | elif c == "0": |
Fredrik Lundh | 0c4fdba | 2000-08-31 22:57:55 +0000 | [diff] [blame] | 374 | # octal escape |
Antoine Pitrou | 463badf | 2012-06-23 13:29:19 +0200 | [diff] [blame] | 375 | escape += source.getwhile(2, OCTDIGITS) |
Serhiy Storchaka | c563caf | 2014-09-23 23:22:41 +0300 | [diff] [blame] | 376 | return LITERAL, int(escape[1:], 8) |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 377 | elif c in DIGITS: |
Fredrik Lundh | 0c4fdba | 2000-08-31 22:57:55 +0000 | [diff] [blame] | 378 | # octal escape *or* decimal group reference (sigh) |
Fredrik Lundh | 0c4fdba | 2000-08-31 22:57:55 +0000 | [diff] [blame] | 379 | if source.next in DIGITS: |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 380 | escape += source.get() |
Fredrik Lundh | 143328b | 2000-09-02 11:03:34 +0000 | [diff] [blame] | 381 | if (escape[1] in OCTDIGITS and escape[2] in OCTDIGITS and |
| 382 | source.next in OCTDIGITS): |
Fredrik Lundh | 0c4fdba | 2000-08-31 22:57:55 +0000 | [diff] [blame] | 383 | # got three octal digits; this is an octal escape |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 384 | escape += source.get() |
Serhiy Storchaka | c563caf | 2014-09-23 23:22:41 +0300 | [diff] [blame] | 385 | c = int(escape[1:], 8) |
| 386 | if c > 0o377: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 387 | raise source.error('octal escape value %s outside of ' |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 388 | 'range 0-0o377' % escape, |
| 389 | len(escape)) |
Serhiy Storchaka | c563caf | 2014-09-23 23:22:41 +0300 | [diff] [blame] | 390 | return LITERAL, c |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 391 | # not an octal escape, so this is a group reference |
| 392 | group = int(escape[1:]) |
| 393 | if group < state.groups: |
Fredrik Lundh | ebc37b2 | 2000-10-28 19:30:41 +0000 | [diff] [blame] | 394 | if not state.checkgroup(group): |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 395 | raise source.error("cannot refer to an open group", |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 396 | len(escape)) |
Serhiy Storchaka | 4eea62f | 2015-02-21 10:07:35 +0200 | [diff] [blame] | 397 | state.checklookbehindgroup(group, source) |
Fredrik Lundh | 0c4fdba | 2000-08-31 22:57:55 +0000 | [diff] [blame] | 398 | return GROUPREF, group |
Serhiy Storchaka | 662cef6 | 2016-10-23 12:11:19 +0300 | [diff] [blame] | 399 | raise source.error("invalid group reference %d" % group, len(escape) - 1) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 400 | if len(escape) == 2: |
Serhiy Storchaka | a54aae0 | 2015-03-24 22:58:14 +0200 | [diff] [blame] | 401 | if c in ASCIILETTERS: |
Serhiy Storchaka | 9bd85b8 | 2016-06-11 19:15:00 +0300 | [diff] [blame] | 402 | raise source.error("bad escape %s" % escape, len(escape)) |
Fredrik Lundh | 0640e11 | 2000-06-30 13:55:15 +0000 | [diff] [blame] | 403 | return LITERAL, ord(escape[1]) |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 404 | except ValueError: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 405 | pass |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 406 | raise source.error("bad escape %s" % escape, len(escape)) |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 407 | |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 408 | def _uniq(items): |
| 409 | if len(set(items)) == len(items): |
| 410 | return items |
| 411 | newitems = [] |
| 412 | for item in items: |
| 413 | if item not in newitems: |
| 414 | newitems.append(item) |
| 415 | return newitems |
| 416 | |
Serhiy Storchaka | c7ac728 | 2017-05-16 15:16:15 +0300 | [diff] [blame] | 417 | def _parse_sub(source, state, verbose, nested): |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 418 | # parse an alternation: a|b|c |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 419 | |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 420 | items = [] |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 421 | itemsappend = items.append |
| 422 | sourcematch = source.match |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 423 | start = source.tell() |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 424 | while True: |
Serhiy Storchaka | c7ac728 | 2017-05-16 15:16:15 +0300 | [diff] [blame] | 425 | itemsappend(_parse(source, state, verbose, nested + 1, |
| 426 | not nested and not items)) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 427 | if not sourcematch("|"): |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 428 | break |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 429 | |
| 430 | if len(items) == 1: |
| 431 | return items[0] |
| 432 | |
| 433 | subpattern = SubPattern(state) |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 434 | |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 435 | # check if all items share a common prefix |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 436 | while True: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 437 | prefix = None |
| 438 | for item in items: |
| 439 | if not item: |
| 440 | break |
| 441 | if prefix is None: |
| 442 | prefix = item[0] |
| 443 | elif item[0] != prefix: |
| 444 | break |
| 445 | else: |
| 446 | # all subitems start with a common "prefix". |
| 447 | # move it out of the branch |
| 448 | for item in items: |
| 449 | del item[0] |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 450 | subpattern.append(prefix) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 451 | continue # check next one |
| 452 | break |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 453 | |
| 454 | # check if the branch can be replaced by a character set |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 455 | set = [] |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 456 | for item in items: |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 457 | if len(item) != 1: |
| 458 | break |
| 459 | op, av = item[0] |
| 460 | if op is LITERAL: |
| 461 | set.append((op, av)) |
| 462 | elif op is IN and av[0][0] is not NEGATE: |
| 463 | set.extend(av) |
| 464 | else: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 465 | break |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 466 | else: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 467 | # we can store this as a character set instead of a |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 468 | # branch (the compiler may optimize this even more) |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 469 | subpattern.append((IN, _uniq(set))) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 470 | return subpattern |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 471 | |
| 472 | subpattern.append((BRANCH, (None, items))) |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 473 | return subpattern |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 474 | |
Serhiy Storchaka | c7ac728 | 2017-05-16 15:16:15 +0300 | [diff] [blame] | 475 | def _parse(source, state, verbose, nested, first=False): |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 476 | # parse a simple pattern |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 477 | subpattern = SubPattern(state) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 478 | |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 479 | # precompute constants into local variables |
| 480 | subpatternappend = subpattern.append |
| 481 | sourceget = source.get |
| 482 | sourcematch = source.match |
| 483 | _len = len |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 484 | _ord = ord |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 485 | |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 486 | while True: |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 487 | |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 488 | this = source.next |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 489 | if this is None: |
| 490 | break # end of pattern |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 491 | if this in "|)": |
| 492 | break # end of subpattern |
| 493 | sourceget() |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 494 | |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 495 | if verbose: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 496 | # skip whitespace and comments |
| 497 | if this in WHITESPACE: |
| 498 | continue |
| 499 | if this == "#": |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 500 | while True: |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 501 | this = sourceget() |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 502 | if this is None or this == "\n": |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 503 | break |
| 504 | continue |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 505 | |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 506 | if this[0] == "\\": |
| 507 | code = _escape(source, this, state) |
| 508 | subpatternappend(code) |
| 509 | |
| 510 | elif this not in SPECIAL_CHARS: |
| 511 | subpatternappend((LITERAL, _ord(this))) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 512 | |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 513 | elif this == "[": |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 514 | here = source.tell() - 1 |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 515 | # character set |
| 516 | set = [] |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 517 | setappend = set.append |
| 518 | ## if sourcematch(":"): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 519 | ## pass # handle character classes |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 520 | negate = sourcematch("^") |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 521 | # check remaining characters |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 522 | while True: |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 523 | this = sourceget() |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 524 | if this is None: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 525 | raise source.error("unterminated character set", |
| 526 | source.tell() - here) |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 527 | if this == "]" and set: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 528 | break |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 529 | elif this[0] == "\\": |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 530 | code1 = _class_escape(source, this) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 531 | else: |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 532 | code1 = LITERAL, _ord(this) |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 533 | if sourcematch("-"): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 534 | # potential range |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 535 | that = sourceget() |
| 536 | if that is None: |
| 537 | raise source.error("unterminated character set", |
| 538 | source.tell() - here) |
| 539 | if that == "]": |
Fredrik Lundh | 025468d | 2000-10-07 10:16:19 +0000 | [diff] [blame] | 540 | if code1[0] is IN: |
| 541 | code1 = code1[1][0] |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 542 | setappend(code1) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 543 | setappend((LITERAL, _ord("-"))) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 544 | break |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 545 | if that[0] == "\\": |
| 546 | code2 = _class_escape(source, that) |
Guido van Rossum | 41c99e7 | 2003-04-14 17:59:34 +0000 | [diff] [blame] | 547 | else: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 548 | code2 = LITERAL, _ord(that) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 549 | if code1[0] != LITERAL or code2[0] != LITERAL: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 550 | msg = "bad character range %s-%s" % (this, that) |
| 551 | raise source.error(msg, len(this) + 1 + len(that)) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 552 | lo = code1[1] |
| 553 | hi = code2[1] |
| 554 | if hi < lo: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 555 | msg = "bad character range %s-%s" % (this, that) |
| 556 | raise source.error(msg, len(this) + 1 + len(that)) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 557 | setappend((RANGE, (lo, hi))) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 558 | else: |
| 559 | if code1[0] is IN: |
| 560 | code1 = code1[1][0] |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 561 | setappend(code1) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 562 | |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 563 | set = _uniq(set) |
Fredrik Lundh | 770617b | 2001-01-14 15:06:11 +0000 | [diff] [blame] | 564 | # XXX: <fl> should move set optimization to compiler! |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 565 | if _len(set) == 1 and set[0][0] is LITERAL: |
| 566 | # optimization |
| 567 | if negate: |
| 568 | subpatternappend((NOT_LITERAL, set[0][1])) |
| 569 | else: |
| 570 | subpatternappend(set[0]) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 571 | else: |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 572 | if negate: |
| 573 | set.insert(0, (NEGATE, None)) |
| 574 | # charmap optimization can't be added here because |
| 575 | # global flags still are not known |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 576 | subpatternappend((IN, set)) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 577 | |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 578 | elif this in REPEAT_CHARS: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 579 | # repeat previous item |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 580 | here = source.tell() |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 581 | if this == "?": |
| 582 | min, max = 0, 1 |
| 583 | elif this == "*": |
| 584 | min, max = 0, MAXREPEAT |
Fredrik Lundh | c0c7ee3 | 2001-02-18 21:04:48 +0000 | [diff] [blame] | 585 | |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 586 | elif this == "+": |
| 587 | min, max = 1, MAXREPEAT |
| 588 | elif this == "{": |
Gustavo Niemeyer | 6fa0c5a | 2005-09-14 08:54:39 +0000 | [diff] [blame] | 589 | if source.next == "}": |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 590 | subpatternappend((LITERAL, _ord(this))) |
Gustavo Niemeyer | 6fa0c5a | 2005-09-14 08:54:39 +0000 | [diff] [blame] | 591 | continue |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 592 | |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 593 | min, max = 0, MAXREPEAT |
| 594 | lo = hi = "" |
| 595 | while source.next in DIGITS: |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 596 | lo += sourceget() |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 597 | if sourcematch(","): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 598 | while source.next in DIGITS: |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 599 | hi += sourceget() |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 600 | else: |
| 601 | hi = lo |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 602 | if not sourcematch("}"): |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 603 | subpatternappend((LITERAL, _ord(this))) |
Fredrik Lundh | c13222c | 2000-07-01 23:49:14 +0000 | [diff] [blame] | 604 | source.seek(here) |
| 605 | continue |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 606 | |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 607 | if lo: |
Barry Warsaw | 8bee761 | 2004-08-25 02:22:30 +0000 | [diff] [blame] | 608 | min = int(lo) |
Serhiy Storchaka | 70ca021 | 2013-02-16 16:47:47 +0200 | [diff] [blame] | 609 | if min >= MAXREPEAT: |
| 610 | raise OverflowError("the repetition number is too large") |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 611 | if hi: |
Barry Warsaw | 8bee761 | 2004-08-25 02:22:30 +0000 | [diff] [blame] | 612 | max = int(hi) |
Serhiy Storchaka | 70ca021 | 2013-02-16 16:47:47 +0200 | [diff] [blame] | 613 | if max >= MAXREPEAT: |
| 614 | raise OverflowError("the repetition number is too large") |
| 615 | if max < min: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 616 | raise source.error("min repeat greater than max repeat", |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 617 | source.tell() - here) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 618 | else: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 619 | raise AssertionError("unsupported quantifier %r" % (char,)) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 620 | # figure out which item to repeat |
| 621 | if subpattern: |
| 622 | item = subpattern[-1:] |
| 623 | else: |
Fredrik Lundh | c0c7ee3 | 2001-02-18 21:04:48 +0000 | [diff] [blame] | 624 | item = None |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 625 | if not item or item[0][0] is AT: |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 626 | raise source.error("nothing to repeat", |
| 627 | source.tell() - here + len(this)) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 628 | if item[0][0] in _REPEATCODES: |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 629 | raise source.error("multiple repeat", |
| 630 | source.tell() - here + len(this)) |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 631 | if item[0][0] is SUBPATTERN: |
| 632 | group, add_flags, del_flags, p = item[0][1] |
| 633 | if group is None and not add_flags and not del_flags: |
| 634 | item = p |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 635 | if sourcematch("?"): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 636 | subpattern[-1] = (MIN_REPEAT, (min, max, item)) |
| 637 | else: |
| 638 | subpattern[-1] = (MAX_REPEAT, (min, max, item)) |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 639 | |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 640 | elif this == ".": |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 641 | subpatternappend((ANY, None)) |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 642 | |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 643 | elif this == "(": |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 644 | start = source.tell() - 1 |
| 645 | group = True |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 646 | name = None |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 647 | add_flags = 0 |
| 648 | del_flags = 0 |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 649 | if sourcematch("?"): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 650 | # options |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 651 | char = sourceget() |
| 652 | if char is None: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 653 | raise source.error("unexpected end of pattern") |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 654 | if char == "P": |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 655 | # python extensions |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 656 | if sourcematch("<"): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 657 | # named group: skip forward to end of name |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 658 | name = source.getuntil(">") |
Georg Brandl | 1d472b7 | 2013-04-14 11:40:00 +0200 | [diff] [blame] | 659 | if not name.isidentifier(): |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 660 | msg = "bad character in group name %r" % name |
| 661 | raise source.error(msg, len(name) + 1) |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 662 | elif sourcematch("="): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 663 | # named backreference |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 664 | name = source.getuntil(")") |
Georg Brandl | 1d472b7 | 2013-04-14 11:40:00 +0200 | [diff] [blame] | 665 | if not name.isidentifier(): |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 666 | msg = "bad character in group name %r" % name |
| 667 | raise source.error(msg, len(name) + 1) |
Fredrik Lundh | b71624e | 2000-06-30 09:13:06 +0000 | [diff] [blame] | 668 | gid = state.groupdict.get(name) |
| 669 | if gid is None: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 670 | msg = "unknown group name %r" % name |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 671 | raise source.error(msg, len(name) + 1) |
Serhiy Storchaka | 485407c | 2015-07-18 23:27:00 +0300 | [diff] [blame] | 672 | if not state.checkgroup(gid): |
| 673 | raise source.error("cannot refer to an open group", |
| 674 | len(name) + 1) |
Serhiy Storchaka | 4eea62f | 2015-02-21 10:07:35 +0200 | [diff] [blame] | 675 | state.checklookbehindgroup(gid, source) |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 676 | subpatternappend((GROUPREF, gid)) |
Fredrik Lundh | 7cafe4d | 2000-07-02 17:33:27 +0000 | [diff] [blame] | 677 | continue |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 678 | |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 679 | else: |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 680 | char = sourceget() |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 681 | if char is None: |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 682 | raise source.error("unexpected end of pattern") |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 683 | raise source.error("unknown extension ?P" + char, |
| 684 | len(char) + 2) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 685 | elif char == ":": |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 686 | # non-capturing group |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 687 | group = None |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 688 | elif char == "#": |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 689 | # comment |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 690 | while True: |
| 691 | if source.next is None: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 692 | raise source.error("missing ), unterminated comment", |
| 693 | source.tell() - start) |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 694 | if sourceget() == ")": |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 695 | break |
Fredrik Lundh | 0c4fdba | 2000-08-31 22:57:55 +0000 | [diff] [blame] | 696 | continue |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 697 | |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 698 | elif char in "=!<": |
Fredrik Lundh | 43b3b49 | 2000-06-30 10:41:31 +0000 | [diff] [blame] | 699 | # lookahead assertions |
Fredrik Lundh | 6f01398 | 2000-07-03 18:44:21 +0000 | [diff] [blame] | 700 | dir = 1 |
| 701 | if char == "<": |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 702 | char = sourceget() |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 703 | if char is None: |
| 704 | raise source.error("unexpected end of pattern") |
| 705 | if char not in "=!": |
| 706 | raise source.error("unknown extension ?<" + char, |
| 707 | len(char) + 2) |
Fredrik Lundh | 6f01398 | 2000-07-03 18:44:21 +0000 | [diff] [blame] | 708 | dir = -1 # lookbehind |
Serhiy Storchaka | 4eea62f | 2015-02-21 10:07:35 +0200 | [diff] [blame] | 709 | lookbehindgroups = state.lookbehindgroups |
| 710 | if lookbehindgroups is None: |
| 711 | state.lookbehindgroups = state.groups |
Serhiy Storchaka | c7ac728 | 2017-05-16 15:16:15 +0300 | [diff] [blame] | 712 | p = _parse_sub(source, state, verbose, nested + 1) |
Serhiy Storchaka | 4eea62f | 2015-02-21 10:07:35 +0200 | [diff] [blame] | 713 | if dir < 0: |
| 714 | if lookbehindgroups is None: |
| 715 | state.lookbehindgroups = None |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 716 | if not sourcematch(")"): |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 717 | raise source.error("missing ), unterminated subpattern", |
| 718 | source.tell() - start) |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 719 | if char == "=": |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 720 | subpatternappend((ASSERT, (dir, p))) |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 721 | else: |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 722 | subpatternappend((ASSERT_NOT, (dir, p))) |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 723 | continue |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 724 | |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 725 | elif char == "(": |
Gustavo Niemeyer | ad3fc44 | 2003-10-17 22:13:16 +0000 | [diff] [blame] | 726 | # conditional backreference group |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 727 | condname = source.getuntil(")") |
Georg Brandl | 1d472b7 | 2013-04-14 11:40:00 +0200 | [diff] [blame] | 728 | if condname.isidentifier(): |
Gustavo Niemeyer | ad3fc44 | 2003-10-17 22:13:16 +0000 | [diff] [blame] | 729 | condgroup = state.groupdict.get(condname) |
| 730 | if condgroup is None: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 731 | msg = "unknown group name %r" % condname |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 732 | raise source.error(msg, len(condname) + 1) |
Gustavo Niemeyer | ad3fc44 | 2003-10-17 22:13:16 +0000 | [diff] [blame] | 733 | else: |
| 734 | try: |
Barry Warsaw | 8bee761 | 2004-08-25 02:22:30 +0000 | [diff] [blame] | 735 | condgroup = int(condname) |
Serhiy Storchaka | 9baa5b2 | 2014-09-29 22:49:23 +0300 | [diff] [blame] | 736 | if condgroup < 0: |
| 737 | raise ValueError |
Gustavo Niemeyer | ad3fc44 | 2003-10-17 22:13:16 +0000 | [diff] [blame] | 738 | except ValueError: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 739 | msg = "bad character in group name %r" % condname |
| 740 | raise source.error(msg, len(condname) + 1) from None |
Serhiy Storchaka | 9baa5b2 | 2014-09-29 22:49:23 +0300 | [diff] [blame] | 741 | if not condgroup: |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 742 | raise source.error("bad group number", |
| 743 | len(condname) + 1) |
Serhiy Storchaka | 9baa5b2 | 2014-09-29 22:49:23 +0300 | [diff] [blame] | 744 | if condgroup >= MAXGROUPS: |
Serhiy Storchaka | 662cef6 | 2016-10-23 12:11:19 +0300 | [diff] [blame] | 745 | msg = "invalid group reference %d" % condgroup |
| 746 | raise source.error(msg, len(condname) + 1) |
Serhiy Storchaka | 4eea62f | 2015-02-21 10:07:35 +0200 | [diff] [blame] | 747 | state.checklookbehindgroup(condgroup, source) |
Serhiy Storchaka | c7ac728 | 2017-05-16 15:16:15 +0300 | [diff] [blame] | 748 | item_yes = _parse(source, state, verbose, nested + 1) |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 749 | if source.match("|"): |
Serhiy Storchaka | c7ac728 | 2017-05-16 15:16:15 +0300 | [diff] [blame] | 750 | item_no = _parse(source, state, verbose, nested + 1) |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 751 | if source.next == "|": |
| 752 | raise source.error("conditional backref with more than two branches") |
| 753 | else: |
| 754 | item_no = None |
| 755 | if not source.match(")"): |
| 756 | raise source.error("missing ), unterminated subpattern", |
| 757 | source.tell() - start) |
| 758 | subpatternappend((GROUPREF_EXISTS, (condgroup, item_yes, item_no))) |
| 759 | continue |
| 760 | |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 761 | elif char in FLAGS or char == "-": |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 762 | # flags |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 763 | flags = _parse_flags(source, state, char) |
| 764 | if flags is None: # global flags |
Serhiy Storchaka | 305ccbe | 2017-05-10 06:05:20 +0300 | [diff] [blame] | 765 | if not first or subpattern: |
Serhiy Storchaka | bd48d27 | 2016-09-11 12:50:02 +0300 | [diff] [blame] | 766 | import warnings |
Serhiy Storchaka | abf275a | 2016-09-17 01:29:58 +0300 | [diff] [blame] | 767 | warnings.warn( |
Roy Williams | 171b9a3 | 2017-06-09 22:01:16 -0700 | [diff] [blame] | 768 | 'Flags not at the start of the expression %r%s' % ( |
Serhiy Storchaka | abf275a | 2016-09-17 01:29:58 +0300 | [diff] [blame] | 769 | source.string[:20], # truncate long regexes |
| 770 | ' (truncated)' if len(source.string) > 20 else '', |
| 771 | ), |
Serhiy Storchaka | c7ac728 | 2017-05-16 15:16:15 +0300 | [diff] [blame] | 772 | DeprecationWarning, stacklevel=nested + 6 |
Serhiy Storchaka | abf275a | 2016-09-17 01:29:58 +0300 | [diff] [blame] | 773 | ) |
Serhiy Storchaka | 305ccbe | 2017-05-10 06:05:20 +0300 | [diff] [blame] | 774 | if (state.flags & SRE_FLAG_VERBOSE) and not verbose: |
| 775 | raise Verbose |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 776 | continue |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 777 | |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 778 | add_flags, del_flags = flags |
| 779 | group = None |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 780 | else: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 781 | raise source.error("unknown extension ?" + char, |
| 782 | len(char) + 1) |
| 783 | |
| 784 | # parse group contents |
| 785 | if group is not None: |
| 786 | try: |
| 787 | group = state.opengroup(name) |
| 788 | except error as err: |
| 789 | raise source.error(err.msg, len(name) + 1) from None |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 790 | sub_verbose = ((verbose or (add_flags & SRE_FLAG_VERBOSE)) and |
| 791 | not (del_flags & SRE_FLAG_VERBOSE)) |
Serhiy Storchaka | c7ac728 | 2017-05-16 15:16:15 +0300 | [diff] [blame] | 792 | p = _parse_sub(source, state, sub_verbose, nested + 1) |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 793 | if not source.match(")"): |
| 794 | raise source.error("missing ), unterminated subpattern", |
| 795 | source.tell() - start) |
| 796 | if group is not None: |
| 797 | state.closegroup(group, p) |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 798 | subpatternappend((SUBPATTERN, (group, add_flags, del_flags, p))) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 799 | |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 800 | elif this == "^": |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 801 | subpatternappend((AT, AT_BEGINNING)) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 802 | |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 803 | elif this == "$": |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 804 | subpatternappend((AT, AT_END)) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 805 | |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 806 | else: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 807 | raise AssertionError("unsupported special character %r" % (char,)) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 808 | |
Serhiy Storchaka | 821a9d1 | 2017-05-14 08:32:33 +0300 | [diff] [blame] | 809 | # unpack non-capturing groups |
| 810 | for i in range(len(subpattern))[::-1]: |
| 811 | op, av = subpattern[i] |
| 812 | if op is SUBPATTERN: |
| 813 | group, add_flags, del_flags, p = av |
| 814 | if group is None and not add_flags and not del_flags: |
| 815 | subpattern[i: i+1] = p |
| 816 | |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 817 | return subpattern |
| 818 | |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 819 | def _parse_flags(source, state, char): |
| 820 | sourceget = source.get |
| 821 | add_flags = 0 |
| 822 | del_flags = 0 |
| 823 | if char != "-": |
| 824 | while True: |
| 825 | add_flags |= FLAGS[char] |
| 826 | char = sourceget() |
| 827 | if char is None: |
| 828 | raise source.error("missing -, : or )") |
| 829 | if char in ")-:": |
| 830 | break |
| 831 | if char not in FLAGS: |
| 832 | msg = "unknown flag" if char.isalpha() else "missing -, : or )" |
| 833 | raise source.error(msg, len(char)) |
| 834 | if char == ")": |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 835 | state.flags |= add_flags |
| 836 | return None |
| 837 | if add_flags & GLOBAL_FLAGS: |
| 838 | raise source.error("bad inline flags: cannot turn on global flag", 1) |
| 839 | if char == "-": |
| 840 | char = sourceget() |
| 841 | if char is None: |
| 842 | raise source.error("missing flag") |
| 843 | if char not in FLAGS: |
| 844 | msg = "unknown flag" if char.isalpha() else "missing flag" |
| 845 | raise source.error(msg, len(char)) |
| 846 | while True: |
| 847 | del_flags |= FLAGS[char] |
| 848 | char = sourceget() |
| 849 | if char is None: |
| 850 | raise source.error("missing :") |
| 851 | if char == ":": |
| 852 | break |
| 853 | if char not in FLAGS: |
| 854 | msg = "unknown flag" if char.isalpha() else "missing :" |
| 855 | raise source.error(msg, len(char)) |
| 856 | assert char == ":" |
| 857 | if del_flags & GLOBAL_FLAGS: |
| 858 | raise source.error("bad inline flags: cannot turn off global flag", 1) |
| 859 | if add_flags & del_flags: |
| 860 | raise source.error("bad inline flags: flag turned on and off", 1) |
| 861 | return add_flags, del_flags |
| 862 | |
Antoine Pitrou | fd03645 | 2008-08-19 17:56:33 +0000 | [diff] [blame] | 863 | def fix_flags(src, flags): |
| 864 | # Check and fix flags according to the type of pattern (str or bytes) |
| 865 | if isinstance(src, str): |
Serhiy Storchaka | 22a309a | 2014-12-01 11:50:07 +0200 | [diff] [blame] | 866 | if flags & SRE_FLAG_LOCALE: |
Serhiy Storchaka | 9bd85b8 | 2016-06-11 19:15:00 +0300 | [diff] [blame] | 867 | raise ValueError("cannot use LOCALE flag with a str pattern") |
Antoine Pitrou | fd03645 | 2008-08-19 17:56:33 +0000 | [diff] [blame] | 868 | if not flags & SRE_FLAG_ASCII: |
| 869 | flags |= SRE_FLAG_UNICODE |
| 870 | elif flags & SRE_FLAG_UNICODE: |
| 871 | raise ValueError("ASCII and UNICODE flags are incompatible") |
| 872 | else: |
| 873 | if flags & SRE_FLAG_UNICODE: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 874 | raise ValueError("cannot use UNICODE flag with a bytes pattern") |
Serhiy Storchaka | 22a309a | 2014-12-01 11:50:07 +0200 | [diff] [blame] | 875 | if flags & SRE_FLAG_LOCALE and flags & SRE_FLAG_ASCII: |
Serhiy Storchaka | 9bd85b8 | 2016-06-11 19:15:00 +0300 | [diff] [blame] | 876 | raise ValueError("ASCII and LOCALE flags are incompatible") |
Antoine Pitrou | fd03645 | 2008-08-19 17:56:33 +0000 | [diff] [blame] | 877 | return flags |
| 878 | |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 879 | def parse(str, flags=0, pattern=None): |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 880 | # parse 're' pattern into list of (opcode, argument) tuples |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 881 | |
| 882 | source = Tokenizer(str) |
| 883 | |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 884 | if pattern is None: |
| 885 | pattern = Pattern() |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 886 | pattern.flags = flags |
Fredrik Lundh | 470ea5a | 2001-01-14 21:00:44 +0000 | [diff] [blame] | 887 | pattern.str = str |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 888 | |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 889 | try: |
Serhiy Storchaka | c7ac728 | 2017-05-16 15:16:15 +0300 | [diff] [blame] | 890 | p = _parse_sub(source, pattern, flags & SRE_FLAG_VERBOSE, 0) |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 891 | except Verbose: |
| 892 | # the VERBOSE flag was switched on inside the pattern. to be |
| 893 | # on the safe side, we'll parse the whole thing again... |
| 894 | pattern = Pattern() |
| 895 | pattern.flags = flags | SRE_FLAG_VERBOSE |
| 896 | pattern.str = str |
Serhiy Storchaka | d65cd09 | 2016-09-11 01:39:01 +0300 | [diff] [blame] | 897 | source.seek(0) |
Serhiy Storchaka | c7ac728 | 2017-05-16 15:16:15 +0300 | [diff] [blame] | 898 | p = _parse_sub(source, pattern, True, 0) |
Serhiy Storchaka | be9a4e5 | 2016-09-10 00:57:55 +0300 | [diff] [blame] | 899 | |
Antoine Pitrou | fd03645 | 2008-08-19 17:56:33 +0000 | [diff] [blame] | 900 | p.pattern.flags = fix_flags(str, p.pattern.flags) |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 901 | |
Serhiy Storchaka | e2ccf56 | 2014-10-10 11:14:49 +0300 | [diff] [blame] | 902 | if source.next is not None: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 903 | assert source.next == ")" |
| 904 | raise source.error("unbalanced parenthesis") |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 905 | |
Serhiy Storchaka | a01a144 | 2016-03-06 09:15:47 +0200 | [diff] [blame] | 906 | if flags & SRE_FLAG_DEBUG: |
| 907 | p.dump() |
| 908 | |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 909 | return p |
| 910 | |
Fredrik Lundh | 436c3d58 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 911 | def parse_template(source, pattern): |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 912 | # parse 're' replacement string into list of literals and |
| 913 | # group references |
| 914 | s = Tokenizer(source) |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 915 | sget = s.get |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 916 | groups = [] |
| 917 | literals = [] |
| 918 | literal = [] |
| 919 | lappend = literal.append |
Serhiy Storchaka | 662cef6 | 2016-10-23 12:11:19 +0300 | [diff] [blame] | 920 | def addgroup(index, pos): |
| 921 | if index > pattern.groups: |
| 922 | raise s.error("invalid group reference %d" % index, pos) |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 923 | if literal: |
| 924 | literals.append(''.join(literal)) |
| 925 | del literal[:] |
| 926 | groups.append((len(literals), index)) |
| 927 | literals.append(None) |
Serhiy Storchaka | 07360df | 2015-03-30 01:01:48 +0300 | [diff] [blame] | 928 | groupindex = pattern.groupindex |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 929 | while True: |
Raymond Hettinger | 968c56a | 2004-03-26 23:24:00 +0000 | [diff] [blame] | 930 | this = sget() |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 931 | if this is None: |
| 932 | break # end of replacement string |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 933 | if this[0] == "\\": |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 934 | # group |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 935 | c = this[1] |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 936 | if c == "g": |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 937 | name = "" |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 938 | if not s.match("<"): |
| 939 | raise s.error("missing <") |
| 940 | name = s.getuntil(">") |
| 941 | if name.isidentifier(): |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 942 | try: |
Serhiy Storchaka | 07360df | 2015-03-30 01:01:48 +0300 | [diff] [blame] | 943 | index = groupindex[name] |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 944 | except KeyError: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 945 | raise IndexError("unknown group name %r" % name) |
| 946 | else: |
| 947 | try: |
| 948 | index = int(name) |
| 949 | if index < 0: |
| 950 | raise ValueError |
| 951 | except ValueError: |
| 952 | raise s.error("bad character in group name %r" % name, |
| 953 | len(name) + 1) from None |
| 954 | if index >= MAXGROUPS: |
Serhiy Storchaka | 662cef6 | 2016-10-23 12:11:19 +0300 | [diff] [blame] | 955 | raise s.error("invalid group reference %d" % index, |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 956 | len(name) + 1) |
Serhiy Storchaka | 662cef6 | 2016-10-23 12:11:19 +0300 | [diff] [blame] | 957 | addgroup(index, len(name) + 1) |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 958 | elif c == "0": |
| 959 | if s.next in OCTDIGITS: |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 960 | this += sget() |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 961 | if s.next in OCTDIGITS: |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 962 | this += sget() |
| 963 | lappend(chr(int(this[1:], 8) & 0xff)) |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 964 | elif c in DIGITS: |
| 965 | isoctal = False |
| 966 | if s.next in DIGITS: |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 967 | this += sget() |
Gustavo Niemeyer | f5a1599 | 2004-09-03 20:15:56 +0000 | [diff] [blame] | 968 | if (c in OCTDIGITS and this[2] in OCTDIGITS and |
| 969 | s.next in OCTDIGITS): |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 970 | this += sget() |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 971 | isoctal = True |
Serhiy Storchaka | c563caf | 2014-09-23 23:22:41 +0300 | [diff] [blame] | 972 | c = int(this[1:], 8) |
| 973 | if c > 0o377: |
Serhiy Storchaka | 632a77e | 2015-03-25 21:03:47 +0200 | [diff] [blame] | 974 | raise s.error('octal escape value %s outside of ' |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 975 | 'range 0-0o377' % this, len(this)) |
Serhiy Storchaka | c563caf | 2014-09-23 23:22:41 +0300 | [diff] [blame] | 976 | lappend(chr(c)) |
Gustavo Niemeyer | a01a2ee | 2004-09-03 17:06:10 +0000 | [diff] [blame] | 977 | if not isoctal: |
Serhiy Storchaka | 662cef6 | 2016-10-23 12:11:19 +0300 | [diff] [blame] | 978 | addgroup(int(this[1:]), len(this) - 1) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 979 | else: |
| 980 | try: |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 981 | this = chr(ESCAPES[this][1]) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 982 | except KeyError: |
Serhiy Storchaka | a54aae0 | 2015-03-24 22:58:14 +0200 | [diff] [blame] | 983 | if c in ASCIILETTERS: |
Serhiy Storchaka | 9bd85b8 | 2016-06-11 19:15:00 +0300 | [diff] [blame] | 984 | raise s.error('bad escape %s' % this, len(this)) |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 985 | lappend(this) |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 986 | else: |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 987 | lappend(this) |
| 988 | if literal: |
| 989 | literals.append(''.join(literal)) |
| 990 | if not isinstance(source, str): |
Ezio Melotti | b92ed7c | 2010-03-06 15:24:08 +0000 | [diff] [blame] | 991 | # The tokenizer implicitly decodes bytes objects as latin-1, we must |
| 992 | # therefore re-encode the final representation. |
Serhiy Storchaka | 9c15ec1 | 2013-10-23 22:27:52 +0300 | [diff] [blame] | 993 | literals = [None if s is None else s.encode('latin-1') for s in literals] |
Fredrik Lundh | b25e1ad | 2001-03-22 15:50:10 +0000 | [diff] [blame] | 994 | return groups, literals |
Andrew M. Kuchling | 815d5b9 | 2000-06-09 14:08:07 +0000 | [diff] [blame] | 995 | |
Fredrik Lundh | 436c3d58 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 996 | def expand_template(template, match): |
Fredrik Lundh | b25e1ad | 2001-03-22 15:50:10 +0000 | [diff] [blame] | 997 | g = match.group |
Serhiy Storchaka | 7438e4b | 2014-10-10 11:06:31 +0300 | [diff] [blame] | 998 | empty = match.string[:0] |
Fredrik Lundh | b25e1ad | 2001-03-22 15:50:10 +0000 | [diff] [blame] | 999 | groups, literals = template |
| 1000 | literals = literals[:] |
| 1001 | try: |
| 1002 | for index, group in groups: |
Serhiy Storchaka | 7438e4b | 2014-10-10 11:06:31 +0300 | [diff] [blame] | 1003 | literals[index] = g(group) or empty |
Fredrik Lundh | b25e1ad | 2001-03-22 15:50:10 +0000 | [diff] [blame] | 1004 | except IndexError: |
Serhiy Storchaka | 662cef6 | 2016-10-23 12:11:19 +0300 | [diff] [blame] | 1005 | raise error("invalid group reference %d" % index) |
Serhiy Storchaka | 7438e4b | 2014-10-10 11:06:31 +0300 | [diff] [blame] | 1006 | return empty.join(literals) |