| 1 | /* dfa.c - deterministic extended regexp routines for GNU
|
|---|
| 2 | Copyright (C) 1988, 1998, 2000, 2002, 2004-2005, 2007-2012 Free Software
|
|---|
| 3 | Foundation, Inc.
|
|---|
| 4 |
|
|---|
| 5 | This program is free software; you can redistribute it and/or modify
|
|---|
| 6 | it under the terms of the GNU General Public License as published by
|
|---|
| 7 | the Free Software Foundation; either version 3, or (at your option)
|
|---|
| 8 | any later version.
|
|---|
| 9 |
|
|---|
| 10 | This program is distributed in the hope that it will be useful,
|
|---|
| 11 | but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|---|
| 12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|---|
| 13 | GNU General Public License for more details.
|
|---|
| 14 |
|
|---|
| 15 | You should have received a copy of the GNU General Public License
|
|---|
| 16 | along with this program; if not, write to the Free Software
|
|---|
| 17 | Foundation, Inc.,
|
|---|
| 18 | 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA */
|
|---|
| 19 |
|
|---|
| 20 | /* Written June, 1988 by Mike Haertel
|
|---|
| 21 | Modified July, 1988 by Arthur David Olson to assist BMG speedups */
|
|---|
| 22 |
|
|---|
| 23 | #include <config.h>
|
|---|
| 24 | #include <assert.h>
|
|---|
| 25 | #include <ctype.h>
|
|---|
| 26 | #include <stdio.h>
|
|---|
| 27 | #include <sys/types.h>
|
|---|
| 28 | #include <stdlib.h>
|
|---|
| 29 | #include <limits.h>
|
|---|
| 30 | #include <string.h>
|
|---|
| 31 | #include <locale.h>
|
|---|
| 32 |
|
|---|
| 33 | #define STREQ(a, b) (strcmp (a, b) == 0)
|
|---|
| 34 |
|
|---|
| 35 | /* ISASCIIDIGIT differs from isdigit, as follows:
|
|---|
| 36 | - Its arg may be any int or unsigned int; it need not be an unsigned char.
|
|---|
| 37 | - It's guaranteed to evaluate its argument exactly once.
|
|---|
| 38 | - It's typically faster.
|
|---|
| 39 | Posix 1003.2-1992 section 2.5.2.1 page 50 lines 1556-1558 says that
|
|---|
| 40 | only '0' through '9' are digits. Prefer ISASCIIDIGIT to isdigit unless
|
|---|
| 41 | it's important to use the locale's definition of `digit' even when the
|
|---|
| 42 | host does not conform to Posix. */
|
|---|
| 43 | #define ISASCIIDIGIT(c) ((unsigned) (c) - '0' <= 9)
|
|---|
| 44 |
|
|---|
| 45 | /* gettext.h ensures that we don't use gettext if ENABLE_NLS is not defined */
|
|---|
| 46 | #include "gettext.h"
|
|---|
| 47 | #define _(str) gettext (str)
|
|---|
| 48 |
|
|---|
| 49 | #include "mbsupport.h" /* defines MBS_SUPPORT if appropriate */
|
|---|
| 50 | #include <wchar.h>
|
|---|
| 51 | #include <wctype.h>
|
|---|
| 52 |
|
|---|
| 53 | #if HAVE_LANGINFO_CODESET
|
|---|
| 54 | # include <langinfo.h>
|
|---|
| 55 | #endif
|
|---|
| 56 |
|
|---|
| 57 | #include "regex.h"
|
|---|
| 58 | #include "dfa.h"
|
|---|
| 59 | #include "hard-locale.h"
|
|---|
| 60 | #include "xalloc.h"
|
|---|
| 61 |
|
|---|
| 62 | /* HPUX, define those as macros in sys/param.h */
|
|---|
| 63 | #ifdef setbit
|
|---|
| 64 | # undef setbit
|
|---|
| 65 | #endif
|
|---|
| 66 | #ifdef clrbit
|
|---|
| 67 | # undef clrbit
|
|---|
| 68 | #endif
|
|---|
| 69 |
|
|---|
| 70 | /* Number of bits in an unsigned char. */
|
|---|
| 71 | #ifndef CHARBITS
|
|---|
| 72 | # define CHARBITS 8
|
|---|
| 73 | #endif
|
|---|
| 74 |
|
|---|
| 75 | /* First integer value that is greater than any character code. */
|
|---|
| 76 | #define NOTCHAR (1 << CHARBITS)
|
|---|
| 77 |
|
|---|
| 78 | /* INTBITS need not be exact, just a lower bound. */
|
|---|
| 79 | #ifndef INTBITS
|
|---|
| 80 | # define INTBITS (CHARBITS * sizeof (int))
|
|---|
| 81 | #endif
|
|---|
| 82 |
|
|---|
| 83 | /* Number of ints required to hold a bit for every character. */
|
|---|
| 84 | #define CHARCLASS_INTS ((NOTCHAR + INTBITS - 1) / INTBITS)
|
|---|
| 85 |
|
|---|
| 86 | /* Sets of unsigned characters are stored as bit vectors in arrays of ints. */
|
|---|
| 87 | typedef int charclass[CHARCLASS_INTS];
|
|---|
| 88 |
|
|---|
| 89 | /* Convert a possibly-signed character to an unsigned character. This is
|
|---|
| 90 | a bit safer than casting to unsigned char, since it catches some type
|
|---|
| 91 | errors that the cast doesn't. */
|
|---|
| 92 | static inline unsigned char
|
|---|
| 93 | to_uchar (char ch)
|
|---|
| 94 | {
|
|---|
| 95 | return ch;
|
|---|
| 96 | }
|
|---|
| 97 |
|
|---|
| 98 | /* Contexts tell us whether a character is a newline or a word constituent.
|
|---|
| 99 | Word-constituent characters are those that satisfy iswalnum(), plus '_'.
|
|---|
| 100 | Each character has a single CTX_* value; bitmasks of CTX_* values denote
|
|---|
| 101 | a particular character class.
|
|---|
| 102 |
|
|---|
| 103 | A state also stores a context value, which is a bitmask of CTX_* values.
|
|---|
| 104 | A state's context represents a set of characters that the state's
|
|---|
| 105 | predecessors must match. For example, a state whose context does not
|
|---|
| 106 | include CTX_LETTER will never have transitions where the previous
|
|---|
| 107 | character is a word constituent. A state whose context is CTX_ANY
|
|---|
| 108 | might have transitions from any character. */
|
|---|
| 109 |
|
|---|
| 110 | #define CTX_NONE 1
|
|---|
| 111 | #define CTX_LETTER 2
|
|---|
| 112 | #define CTX_NEWLINE 4
|
|---|
| 113 | #define CTX_ANY 7
|
|---|
| 114 |
|
|---|
| 115 | /* Sometimes characters can only be matched depending on the surrounding
|
|---|
| 116 | context. Such context decisions depend on what the previous character
|
|---|
| 117 | was, and the value of the current (lookahead) character. Context
|
|---|
| 118 | dependent constraints are encoded as 8 bit integers. Each bit that
|
|---|
| 119 | is set indicates that the constraint succeeds in the corresponding
|
|---|
| 120 | context.
|
|---|
| 121 |
|
|---|
| 122 | bit 8-11 - valid contexts when next character is CTX_NEWLINE
|
|---|
| 123 | bit 4-7 - valid contexts when next character is CTX_LETTER
|
|---|
| 124 | bit 0-3 - valid contexts when next character is CTX_NONE
|
|---|
| 125 |
|
|---|
| 126 | The macro SUCCEEDS_IN_CONTEXT determines whether a given constraint
|
|---|
| 127 | succeeds in a particular context. Prev is a bitmask of possible
|
|---|
| 128 | context values for the previous character, curr is the (single-bit)
|
|---|
| 129 | context value for the lookahead character. */
|
|---|
| 130 | #define NEWLINE_CONSTRAINT(constraint) (((constraint) >> 8) & 0xf)
|
|---|
| 131 | #define LETTER_CONSTRAINT(constraint) (((constraint) >> 4) & 0xf)
|
|---|
| 132 | #define OTHER_CONSTRAINT(constraint) ((constraint) & 0xf)
|
|---|
| 133 |
|
|---|
| 134 | #define SUCCEEDS_IN_CONTEXT(constraint, prev, curr) \
|
|---|
| 135 | ((((curr) & CTX_NONE ? OTHER_CONSTRAINT(constraint) : 0) \
|
|---|
| 136 | | ((curr) & CTX_LETTER ? LETTER_CONSTRAINT(constraint) : 0) \
|
|---|
| 137 | | ((curr) & CTX_NEWLINE ? NEWLINE_CONSTRAINT(constraint) : 0)) & (prev))
|
|---|
| 138 |
|
|---|
| 139 | /* The following macros give information about what a constraint depends on. */
|
|---|
| 140 | #define PREV_NEWLINE_CONSTRAINT(constraint) (((constraint) >> 2) & 0x111)
|
|---|
| 141 | #define PREV_LETTER_CONSTRAINT(constraint) (((constraint) >> 1) & 0x111)
|
|---|
| 142 | #define PREV_OTHER_CONSTRAINT(constraint) ((constraint) & 0x111)
|
|---|
| 143 |
|
|---|
| 144 | #define PREV_NEWLINE_DEPENDENT(constraint) \
|
|---|
| 145 | (PREV_NEWLINE_CONSTRAINT (constraint) != PREV_OTHER_CONSTRAINT (constraint))
|
|---|
| 146 | #define PREV_LETTER_DEPENDENT(constraint) \
|
|---|
| 147 | (PREV_LETTER_CONSTRAINT (constraint) != PREV_OTHER_CONSTRAINT (constraint))
|
|---|
| 148 |
|
|---|
| 149 | /* Tokens that match the empty string subject to some constraint actually
|
|---|
| 150 | work by applying that constraint to determine what may follow them,
|
|---|
| 151 | taking into account what has gone before. The following values are
|
|---|
| 152 | the constraints corresponding to the special tokens previously defined. */
|
|---|
| 153 | #define NO_CONSTRAINT 0x777
|
|---|
| 154 | #define BEGLINE_CONSTRAINT 0x444
|
|---|
| 155 | #define ENDLINE_CONSTRAINT 0x700
|
|---|
| 156 | #define BEGWORD_CONSTRAINT 0x050
|
|---|
| 157 | #define ENDWORD_CONSTRAINT 0x202
|
|---|
| 158 | #define LIMWORD_CONSTRAINT 0x252
|
|---|
| 159 | #define NOTLIMWORD_CONSTRAINT 0x525
|
|---|
| 160 |
|
|---|
| 161 | /* The regexp is parsed into an array of tokens in postfix form. Some tokens
|
|---|
| 162 | are operators and others are terminal symbols. Most (but not all) of these
|
|---|
| 163 | codes are returned by the lexical analyzer. */
|
|---|
| 164 |
|
|---|
| 165 | typedef ptrdiff_t token;
|
|---|
| 166 |
|
|---|
| 167 | /* Predefined token values. */
|
|---|
| 168 | enum
|
|---|
| 169 | {
|
|---|
| 170 | END = -1, /* END is a terminal symbol that matches the
|
|---|
| 171 | end of input; any value of END or less in
|
|---|
| 172 | the parse tree is such a symbol. Accepting
|
|---|
| 173 | states of the DFA are those that would have
|
|---|
| 174 | a transition on END. */
|
|---|
| 175 |
|
|---|
| 176 | /* Ordinary character values are terminal symbols that match themselves. */
|
|---|
| 177 |
|
|---|
| 178 | EMPTY = NOTCHAR, /* EMPTY is a terminal symbol that matches
|
|---|
| 179 | the empty string. */
|
|---|
| 180 |
|
|---|
| 181 | BACKREF, /* BACKREF is generated by \<digit>; it
|
|---|
| 182 | is not completely handled. If the scanner
|
|---|
| 183 | detects a transition on backref, it returns
|
|---|
| 184 | a kind of "semi-success" indicating that
|
|---|
| 185 | the match will have to be verified with
|
|---|
| 186 | a backtracking matcher. */
|
|---|
| 187 |
|
|---|
| 188 | BEGLINE, /* BEGLINE is a terminal symbol that matches
|
|---|
| 189 | the empty string if it is at the beginning
|
|---|
| 190 | of a line. */
|
|---|
| 191 |
|
|---|
| 192 | ENDLINE, /* ENDLINE is a terminal symbol that matches
|
|---|
| 193 | the empty string if it is at the end of
|
|---|
| 194 | a line. */
|
|---|
| 195 |
|
|---|
| 196 | BEGWORD, /* BEGWORD is a terminal symbol that matches
|
|---|
| 197 | the empty string if it is at the beginning
|
|---|
| 198 | of a word. */
|
|---|
| 199 |
|
|---|
| 200 | ENDWORD, /* ENDWORD is a terminal symbol that matches
|
|---|
| 201 | the empty string if it is at the end of
|
|---|
| 202 | a word. */
|
|---|
| 203 |
|
|---|
| 204 | LIMWORD, /* LIMWORD is a terminal symbol that matches
|
|---|
| 205 | the empty string if it is at the beginning
|
|---|
| 206 | or the end of a word. */
|
|---|
| 207 |
|
|---|
| 208 | NOTLIMWORD, /* NOTLIMWORD is a terminal symbol that
|
|---|
| 209 | matches the empty string if it is not at
|
|---|
| 210 | the beginning or end of a word. */
|
|---|
| 211 |
|
|---|
| 212 | QMARK, /* QMARK is an operator of one argument that
|
|---|
| 213 | matches zero or one occurrences of its
|
|---|
| 214 | argument. */
|
|---|
| 215 |
|
|---|
| 216 | STAR, /* STAR is an operator of one argument that
|
|---|
| 217 | matches the Kleene closure (zero or more
|
|---|
| 218 | occurrences) of its argument. */
|
|---|
| 219 |
|
|---|
| 220 | PLUS, /* PLUS is an operator of one argument that
|
|---|
| 221 | matches the positive closure (one or more
|
|---|
| 222 | occurrences) of its argument. */
|
|---|
| 223 |
|
|---|
| 224 | REPMN, /* REPMN is a lexical token corresponding
|
|---|
| 225 | to the {m,n} construct. REPMN never
|
|---|
| 226 | appears in the compiled token vector. */
|
|---|
| 227 |
|
|---|
| 228 | CAT, /* CAT is an operator of two arguments that
|
|---|
| 229 | matches the concatenation of its
|
|---|
| 230 | arguments. CAT is never returned by the
|
|---|
| 231 | lexical analyzer. */
|
|---|
| 232 |
|
|---|
| 233 | OR, /* OR is an operator of two arguments that
|
|---|
| 234 | matches either of its arguments. */
|
|---|
| 235 |
|
|---|
| 236 | LPAREN, /* LPAREN never appears in the parse tree,
|
|---|
| 237 | it is only a lexeme. */
|
|---|
| 238 |
|
|---|
| 239 | RPAREN, /* RPAREN never appears in the parse tree. */
|
|---|
| 240 |
|
|---|
| 241 | ANYCHAR, /* ANYCHAR is a terminal symbol that matches
|
|---|
| 242 | any multibyte (or single byte) characters.
|
|---|
| 243 | It is used only if MB_CUR_MAX > 1. */
|
|---|
| 244 |
|
|---|
| 245 | MBCSET, /* MBCSET is similar to CSET, but for
|
|---|
| 246 | multibyte characters. */
|
|---|
| 247 |
|
|---|
| 248 | WCHAR, /* Only returned by lex. wctok contains
|
|---|
| 249 | the wide character representation. */
|
|---|
| 250 |
|
|---|
| 251 | CSET /* CSET and (and any value greater) is a
|
|---|
| 252 | terminal symbol that matches any of a
|
|---|
| 253 | class of characters. */
|
|---|
| 254 | };
|
|---|
| 255 |
|
|---|
| 256 |
|
|---|
| 257 | /* States of the recognizer correspond to sets of positions in the parse
|
|---|
| 258 | tree, together with the constraints under which they may be matched.
|
|---|
| 259 | So a position is encoded as an index into the parse tree together with
|
|---|
| 260 | a constraint. */
|
|---|
| 261 | typedef struct
|
|---|
| 262 | {
|
|---|
| 263 | size_t index; /* Index into the parse array. */
|
|---|
| 264 | unsigned int constraint; /* Constraint for matching this position. */
|
|---|
| 265 | } position;
|
|---|
| 266 |
|
|---|
| 267 | /* Sets of positions are stored as arrays. */
|
|---|
| 268 | typedef struct
|
|---|
| 269 | {
|
|---|
| 270 | position *elems; /* Elements of this position set. */
|
|---|
| 271 | size_t nelem; /* Number of elements in this set. */
|
|---|
| 272 | size_t alloc; /* Number of elements allocated in ELEMS. */
|
|---|
| 273 | } position_set;
|
|---|
| 274 |
|
|---|
| 275 | /* Sets of leaves are also stored as arrays. */
|
|---|
| 276 | typedef struct
|
|---|
| 277 | {
|
|---|
| 278 | size_t *elems; /* Elements of this position set. */
|
|---|
| 279 | size_t nelem; /* Number of elements in this set. */
|
|---|
| 280 | } leaf_set;
|
|---|
| 281 |
|
|---|
| 282 | /* A state of the dfa consists of a set of positions, some flags,
|
|---|
| 283 | and the token value of the lowest-numbered position of the state that
|
|---|
| 284 | contains an END token. */
|
|---|
| 285 | typedef struct
|
|---|
| 286 | {
|
|---|
| 287 | size_t hash; /* Hash of the positions of this state. */
|
|---|
| 288 | position_set elems; /* Positions this state could match. */
|
|---|
| 289 | unsigned char context; /* Context from previous state. */
|
|---|
| 290 | char backref; /* True if this state matches a \<digit>. */
|
|---|
| 291 | unsigned short constraint; /* Constraint for this state to accept. */
|
|---|
| 292 | token first_end; /* Token value of the first END in elems. */
|
|---|
| 293 | position_set mbps; /* Positions which can match multibyte
|
|---|
| 294 | characters. e.g. period.
|
|---|
| 295 | These staff are used only if
|
|---|
| 296 | MB_CUR_MAX > 1. */
|
|---|
| 297 | } dfa_state;
|
|---|
| 298 |
|
|---|
| 299 | /* States are indexed by state_num values. These are normally
|
|---|
| 300 | nonnegative but -1 is used as a special value. */
|
|---|
| 301 | typedef ptrdiff_t state_num;
|
|---|
| 302 |
|
|---|
| 303 | /* A bracket operator.
|
|---|
| 304 | e.g. [a-c], [[:alpha:]], etc. */
|
|---|
| 305 | struct mb_char_classes
|
|---|
| 306 | {
|
|---|
| 307 | ptrdiff_t cset;
|
|---|
| 308 | int invert;
|
|---|
| 309 | wchar_t *chars; /* Normal characters. */
|
|---|
| 310 | size_t nchars;
|
|---|
| 311 | wctype_t *ch_classes; /* Character classes. */
|
|---|
| 312 | size_t nch_classes;
|
|---|
| 313 | wchar_t *range_sts; /* Range characters (start of the range). */
|
|---|
| 314 | wchar_t *range_ends; /* Range characters (end of the range). */
|
|---|
| 315 | size_t nranges;
|
|---|
| 316 | char **equivs; /* Equivalence classes. */
|
|---|
| 317 | size_t nequivs;
|
|---|
| 318 | char **coll_elems;
|
|---|
| 319 | size_t ncoll_elems; /* Collating elements. */
|
|---|
| 320 | };
|
|---|
| 321 |
|
|---|
| 322 | /* A compiled regular expression. */
|
|---|
| 323 | struct dfa
|
|---|
| 324 | {
|
|---|
| 325 | /* Fields filled by the scanner. */
|
|---|
| 326 | charclass *charclasses; /* Array of character sets for CSET tokens. */
|
|---|
| 327 | size_t cindex; /* Index for adding new charclasses. */
|
|---|
| 328 | size_t calloc; /* Number of charclasses currently allocated. */
|
|---|
| 329 |
|
|---|
| 330 | /* Fields filled by the parser. */
|
|---|
| 331 | token *tokens; /* Postfix parse array. */
|
|---|
| 332 | size_t tindex; /* Index for adding new tokens. */
|
|---|
| 333 | size_t talloc; /* Number of tokens currently allocated. */
|
|---|
| 334 | size_t depth; /* Depth required of an evaluation stack
|
|---|
| 335 | used for depth-first traversal of the
|
|---|
| 336 | parse tree. */
|
|---|
| 337 | size_t nleaves; /* Number of leaves on the parse tree. */
|
|---|
| 338 | size_t nregexps; /* Count of parallel regexps being built
|
|---|
| 339 | with dfaparse(). */
|
|---|
| 340 | unsigned int mb_cur_max; /* Cached value of MB_CUR_MAX. */
|
|---|
| 341 | token utf8_anychar_classes[5]; /* To lower ANYCHAR in UTF-8 locales. */
|
|---|
| 342 |
|
|---|
| 343 | /* The following are used only if MB_CUR_MAX > 1. */
|
|---|
| 344 |
|
|---|
| 345 | /* The value of multibyte_prop[i] is defined by following rule.
|
|---|
| 346 | if tokens[i] < NOTCHAR
|
|---|
| 347 | bit 0 : tokens[i] is the first byte of a character, including
|
|---|
| 348 | single-byte characters.
|
|---|
| 349 | bit 1 : tokens[i] is the last byte of a character, including
|
|---|
| 350 | single-byte characters.
|
|---|
| 351 |
|
|---|
| 352 | if tokens[i] = MBCSET
|
|---|
| 353 | ("the index of mbcsets corresponding to this operator" << 2) + 3
|
|---|
| 354 |
|
|---|
| 355 | e.g.
|
|---|
| 356 | tokens
|
|---|
| 357 | = 'single_byte_a', 'multi_byte_A', single_byte_b'
|
|---|
| 358 | = 'sb_a', 'mb_A(1st byte)', 'mb_A(2nd byte)', 'mb_A(3rd byte)', 'sb_b'
|
|---|
| 359 | multibyte_prop
|
|---|
| 360 | = 3 , 1 , 0 , 2 , 3
|
|---|
| 361 | */
|
|---|
| 362 | size_t nmultibyte_prop;
|
|---|
| 363 | int *multibyte_prop;
|
|---|
| 364 |
|
|---|
| 365 | /* Array of the bracket expression in the DFA. */
|
|---|
| 366 | struct mb_char_classes *mbcsets;
|
|---|
| 367 | size_t nmbcsets;
|
|---|
| 368 | size_t mbcsets_alloc;
|
|---|
| 369 |
|
|---|
| 370 | /* Fields filled by the state builder. */
|
|---|
| 371 | dfa_state *states; /* States of the dfa. */
|
|---|
| 372 | state_num sindex; /* Index for adding new states. */
|
|---|
| 373 | state_num salloc; /* Number of states currently allocated. */
|
|---|
| 374 |
|
|---|
| 375 | /* Fields filled by the parse tree->NFA conversion. */
|
|---|
| 376 | position_set *follows; /* Array of follow sets, indexed by position
|
|---|
| 377 | index. The follow of a position is the set
|
|---|
| 378 | of positions containing characters that
|
|---|
| 379 | could conceivably follow a character
|
|---|
| 380 | matching the given position in a string
|
|---|
| 381 | matching the regexp. Allocated to the
|
|---|
| 382 | maximum possible position index. */
|
|---|
| 383 | int searchflag; /* True if we are supposed to build a searching
|
|---|
| 384 | as opposed to an exact matcher. A searching
|
|---|
| 385 | matcher finds the first and shortest string
|
|---|
| 386 | matching a regexp anywhere in the buffer,
|
|---|
| 387 | whereas an exact matcher finds the longest
|
|---|
| 388 | string matching, but anchored to the
|
|---|
| 389 | beginning of the buffer. */
|
|---|
| 390 |
|
|---|
| 391 | /* Fields filled by dfaexec. */
|
|---|
| 392 | state_num tralloc; /* Number of transition tables that have
|
|---|
| 393 | slots so far. */
|
|---|
| 394 | int trcount; /* Number of transition tables that have
|
|---|
| 395 | actually been built. */
|
|---|
| 396 | state_num **trans; /* Transition tables for states that can
|
|---|
| 397 | never accept. If the transitions for a
|
|---|
| 398 | state have not yet been computed, or the
|
|---|
| 399 | state could possibly accept, its entry in
|
|---|
| 400 | this table is NULL. */
|
|---|
| 401 | state_num **realtrans; /* Trans always points to realtrans + 1; this
|
|---|
| 402 | is so trans[-1] can contain NULL. */
|
|---|
| 403 | state_num **fails; /* Transition tables after failing to accept
|
|---|
| 404 | on a state that potentially could do so. */
|
|---|
| 405 | int *success; /* Table of acceptance conditions used in
|
|---|
| 406 | dfaexec and computed in build_state. */
|
|---|
| 407 | state_num *newlines; /* Transitions on newlines. The entry for a
|
|---|
| 408 | newline in any transition table is always
|
|---|
| 409 | -1 so we can count lines without wasting
|
|---|
| 410 | too many cycles. The transition for a
|
|---|
| 411 | newline is stored separately and handled
|
|---|
| 412 | as a special case. Newline is also used
|
|---|
| 413 | as a sentinel at the end of the buffer. */
|
|---|
| 414 | struct dfamust *musts; /* List of strings, at least one of which
|
|---|
| 415 | is known to appear in any r.e. matching
|
|---|
| 416 | the dfa. */
|
|---|
| 417 | };
|
|---|
| 418 |
|
|---|
| 419 | /* Some macros for user access to dfa internals. */
|
|---|
| 420 |
|
|---|
| 421 | /* ACCEPTING returns true if s could possibly be an accepting state of r. */
|
|---|
| 422 | #define ACCEPTING(s, r) ((r).states[s].constraint)
|
|---|
| 423 |
|
|---|
| 424 | /* ACCEPTS_IN_CONTEXT returns true if the given state accepts in the
|
|---|
| 425 | specified context. */
|
|---|
| 426 | #define ACCEPTS_IN_CONTEXT(prev, curr, state, dfa) \
|
|---|
| 427 | SUCCEEDS_IN_CONTEXT ((dfa).states[state].constraint, prev, curr)
|
|---|
| 428 |
|
|---|
| 429 | static void dfamust (struct dfa *dfa);
|
|---|
| 430 | static void regexp (void);
|
|---|
| 431 |
|
|---|
| 432 | /* These two macros are identical to the ones in gnulib's xalloc.h,
|
|---|
| 433 | except that they not to case the result to "(t *)", and thus may
|
|---|
| 434 | be used via type-free CALLOC and MALLOC macros. */
|
|---|
| 435 | #undef XNMALLOC
|
|---|
| 436 | #undef XCALLOC
|
|---|
| 437 |
|
|---|
| 438 | /* Allocate memory for N elements of type T, with error checking. */
|
|---|
| 439 | /* extern t *XNMALLOC (size_t n, typename t); */
|
|---|
| 440 | # define XNMALLOC(n, t) \
|
|---|
| 441 | (sizeof (t) == 1 ? xmalloc (n) : xnmalloc (n, sizeof (t)))
|
|---|
| 442 |
|
|---|
| 443 | /* Allocate memory for N elements of type T, with error checking,
|
|---|
| 444 | and zero it. */
|
|---|
| 445 | /* extern t *XCALLOC (size_t n, typename t); */
|
|---|
| 446 | # define XCALLOC(n, t) \
|
|---|
| 447 | (sizeof (t) == 1 ? xzalloc (n) : xcalloc (n, sizeof (t)))
|
|---|
| 448 |
|
|---|
| 449 | #define CALLOC(p, n) do { (p) = XCALLOC (n, *(p)); } while (0)
|
|---|
| 450 | #define MALLOC(p, n) do { (p) = XNMALLOC (n, *(p)); } while (0)
|
|---|
| 451 | #define REALLOC(p, n) do {(p) = xnrealloc (p, n, sizeof (*(p))); } while (0)
|
|---|
| 452 |
|
|---|
| 453 | /* Reallocate an array of type *P if N_ALLOC is <= N_REQUIRED. */
|
|---|
| 454 | #define REALLOC_IF_NECESSARY(p, n_alloc, n_required) \
|
|---|
| 455 | do \
|
|---|
| 456 | { \
|
|---|
| 457 | if ((n_alloc) <= (n_required)) \
|
|---|
| 458 | { \
|
|---|
| 459 | size_t new_n_alloc = (n_required) + !(p); \
|
|---|
| 460 | (p) = x2nrealloc (p, &new_n_alloc, sizeof (*(p))); \
|
|---|
| 461 | (n_alloc) = new_n_alloc; \
|
|---|
| 462 | } \
|
|---|
| 463 | } \
|
|---|
| 464 | while (false)
|
|---|
| 465 |
|
|---|
| 466 |
|
|---|
| 467 | #ifdef DEBUG
|
|---|
| 468 |
|
|---|
| 469 | static void
|
|---|
| 470 | prtok (token t)
|
|---|
| 471 | {
|
|---|
| 472 | char const *s;
|
|---|
| 473 |
|
|---|
| 474 | if (t < 0)
|
|---|
| 475 | fprintf (stderr, "END");
|
|---|
| 476 | else if (t < NOTCHAR)
|
|---|
| 477 | {
|
|---|
| 478 | int ch = t;
|
|---|
| 479 | fprintf (stderr, "%c", ch);
|
|---|
| 480 | }
|
|---|
| 481 | else
|
|---|
| 482 | {
|
|---|
| 483 | switch (t)
|
|---|
| 484 | {
|
|---|
| 485 | case EMPTY:
|
|---|
| 486 | s = "EMPTY";
|
|---|
| 487 | break;
|
|---|
| 488 | case BACKREF:
|
|---|
| 489 | s = "BACKREF";
|
|---|
| 490 | break;
|
|---|
| 491 | case BEGLINE:
|
|---|
| 492 | s = "BEGLINE";
|
|---|
| 493 | break;
|
|---|
| 494 | case ENDLINE:
|
|---|
| 495 | s = "ENDLINE";
|
|---|
| 496 | break;
|
|---|
| 497 | case BEGWORD:
|
|---|
| 498 | s = "BEGWORD";
|
|---|
| 499 | break;
|
|---|
| 500 | case ENDWORD:
|
|---|
| 501 | s = "ENDWORD";
|
|---|
| 502 | break;
|
|---|
| 503 | case LIMWORD:
|
|---|
| 504 | s = "LIMWORD";
|
|---|
| 505 | break;
|
|---|
| 506 | case NOTLIMWORD:
|
|---|
| 507 | s = "NOTLIMWORD";
|
|---|
| 508 | break;
|
|---|
| 509 | case QMARK:
|
|---|
| 510 | s = "QMARK";
|
|---|
| 511 | break;
|
|---|
| 512 | case STAR:
|
|---|
| 513 | s = "STAR";
|
|---|
| 514 | break;
|
|---|
| 515 | case PLUS:
|
|---|
| 516 | s = "PLUS";
|
|---|
| 517 | break;
|
|---|
| 518 | case CAT:
|
|---|
| 519 | s = "CAT";
|
|---|
| 520 | break;
|
|---|
| 521 | case OR:
|
|---|
| 522 | s = "OR";
|
|---|
| 523 | break;
|
|---|
| 524 | case LPAREN:
|
|---|
| 525 | s = "LPAREN";
|
|---|
| 526 | break;
|
|---|
| 527 | case RPAREN:
|
|---|
| 528 | s = "RPAREN";
|
|---|
| 529 | break;
|
|---|
| 530 | case ANYCHAR:
|
|---|
| 531 | s = "ANYCHAR";
|
|---|
| 532 | break;
|
|---|
| 533 | case MBCSET:
|
|---|
| 534 | s = "MBCSET";
|
|---|
| 535 | break;
|
|---|
| 536 | default:
|
|---|
| 537 | s = "CSET";
|
|---|
| 538 | break;
|
|---|
| 539 | }
|
|---|
| 540 | fprintf (stderr, "%s", s);
|
|---|
| 541 | }
|
|---|
| 542 | }
|
|---|
| 543 | #endif /* DEBUG */
|
|---|
| 544 |
|
|---|
| 545 | /* Stuff pertaining to charclasses. */
|
|---|
| 546 |
|
|---|
| 547 | static int
|
|---|
| 548 | tstbit (unsigned int b, charclass const c)
|
|---|
| 549 | {
|
|---|
| 550 | return c[b / INTBITS] & 1 << b % INTBITS;
|
|---|
| 551 | }
|
|---|
| 552 |
|
|---|
| 553 | static void
|
|---|
| 554 | setbit (unsigned int b, charclass c)
|
|---|
| 555 | {
|
|---|
| 556 | c[b / INTBITS] |= 1 << b % INTBITS;
|
|---|
| 557 | }
|
|---|
| 558 |
|
|---|
| 559 | static void
|
|---|
| 560 | clrbit (unsigned int b, charclass c)
|
|---|
| 561 | {
|
|---|
| 562 | c[b / INTBITS] &= ~(1 << b % INTBITS);
|
|---|
| 563 | }
|
|---|
| 564 |
|
|---|
| 565 | static void
|
|---|
| 566 | copyset (charclass const src, charclass dst)
|
|---|
| 567 | {
|
|---|
| 568 | memcpy (dst, src, sizeof (charclass));
|
|---|
| 569 | }
|
|---|
| 570 |
|
|---|
| 571 | static void
|
|---|
| 572 | zeroset (charclass s)
|
|---|
| 573 | {
|
|---|
| 574 | memset (s, 0, sizeof (charclass));
|
|---|
| 575 | }
|
|---|
| 576 |
|
|---|
| 577 | static void
|
|---|
| 578 | notset (charclass s)
|
|---|
| 579 | {
|
|---|
| 580 | int i;
|
|---|
| 581 |
|
|---|
| 582 | for (i = 0; i < CHARCLASS_INTS; ++i)
|
|---|
| 583 | s[i] = ~s[i];
|
|---|
| 584 | }
|
|---|
| 585 |
|
|---|
| 586 | static int
|
|---|
| 587 | equal (charclass const s1, charclass const s2)
|
|---|
| 588 | {
|
|---|
| 589 | return memcmp (s1, s2, sizeof (charclass)) == 0;
|
|---|
| 590 | }
|
|---|
| 591 |
|
|---|
| 592 | /* A pointer to the current dfa is kept here during parsing. */
|
|---|
| 593 | static struct dfa *dfa;
|
|---|
| 594 |
|
|---|
| 595 | /* Find the index of charclass s in dfa->charclasses, or allocate a new charclass. */
|
|---|
| 596 | static size_t
|
|---|
| 597 | charclass_index (charclass const s)
|
|---|
| 598 | {
|
|---|
| 599 | size_t i;
|
|---|
| 600 |
|
|---|
| 601 | for (i = 0; i < dfa->cindex; ++i)
|
|---|
| 602 | if (equal (s, dfa->charclasses[i]))
|
|---|
| 603 | return i;
|
|---|
| 604 | REALLOC_IF_NECESSARY (dfa->charclasses, dfa->calloc, dfa->cindex + 1);
|
|---|
| 605 | ++dfa->cindex;
|
|---|
| 606 | copyset (s, dfa->charclasses[i]);
|
|---|
| 607 | return i;
|
|---|
| 608 | }
|
|---|
| 609 |
|
|---|
| 610 | /* Syntax bits controlling the behavior of the lexical analyzer. */
|
|---|
| 611 | static reg_syntax_t syntax_bits, syntax_bits_set;
|
|---|
| 612 |
|
|---|
| 613 | /* Flag for case-folding letters into sets. */
|
|---|
| 614 | static int case_fold;
|
|---|
| 615 |
|
|---|
| 616 | /* End-of-line byte in data. */
|
|---|
| 617 | static unsigned char eolbyte;
|
|---|
| 618 |
|
|---|
| 619 | /* Cache of char-context values. */
|
|---|
| 620 | static int sbit[NOTCHAR];
|
|---|
| 621 |
|
|---|
| 622 | /* Set of characters considered letters. */
|
|---|
| 623 | static charclass letters;
|
|---|
| 624 |
|
|---|
| 625 | /* Set of characters that are newline. */
|
|---|
| 626 | static charclass newline;
|
|---|
| 627 |
|
|---|
| 628 | /* Add this to the test for whether a byte is word-constituent, since on
|
|---|
| 629 | BSD-based systems, many values in the 128..255 range are classified as
|
|---|
| 630 | alphabetic, while on glibc-based systems, they are not. */
|
|---|
| 631 | #ifdef __GLIBC__
|
|---|
| 632 | # define is_valid_unibyte_character(c) 1
|
|---|
| 633 | #else
|
|---|
| 634 | # define is_valid_unibyte_character(c) (! (MBS_SUPPORT && btowc (c) == WEOF))
|
|---|
| 635 | #endif
|
|---|
| 636 |
|
|---|
| 637 | /* Return non-zero if C is a 'word-constituent' byte; zero otherwise. */
|
|---|
| 638 | #define IS_WORD_CONSTITUENT(C) \
|
|---|
| 639 | (is_valid_unibyte_character (C) && (isalnum (C) || (C) == '_'))
|
|---|
| 640 |
|
|---|
| 641 | static int
|
|---|
| 642 | char_context (unsigned char c)
|
|---|
| 643 | {
|
|---|
| 644 | if (c == eolbyte || c == 0)
|
|---|
| 645 | return CTX_NEWLINE;
|
|---|
| 646 | if (IS_WORD_CONSTITUENT (c))
|
|---|
| 647 | return CTX_LETTER;
|
|---|
| 648 | return CTX_NONE;
|
|---|
| 649 | }
|
|---|
| 650 |
|
|---|
| 651 | static int
|
|---|
| 652 | wchar_context (wint_t wc)
|
|---|
| 653 | {
|
|---|
| 654 | if (wc == (wchar_t) eolbyte || wc == 0)
|
|---|
| 655 | return CTX_NEWLINE;
|
|---|
| 656 | if (wc == L'_' || iswalnum (wc))
|
|---|
| 657 | return CTX_LETTER;
|
|---|
| 658 | return CTX_NONE;
|
|---|
| 659 | }
|
|---|
| 660 |
|
|---|
| 661 | /* Entry point to set syntax options. */
|
|---|
| 662 | void
|
|---|
| 663 | dfasyntax (reg_syntax_t bits, int fold, unsigned char eol)
|
|---|
| 664 | {
|
|---|
| 665 | unsigned int i;
|
|---|
| 666 |
|
|---|
| 667 | syntax_bits_set = 1;
|
|---|
| 668 | syntax_bits = bits;
|
|---|
| 669 | case_fold = fold;
|
|---|
| 670 | eolbyte = eol;
|
|---|
| 671 |
|
|---|
| 672 | for (i = 0; i < NOTCHAR; ++i)
|
|---|
| 673 | {
|
|---|
| 674 | sbit[i] = char_context (i);
|
|---|
| 675 | switch (sbit[i])
|
|---|
| 676 | {
|
|---|
| 677 | case CTX_LETTER:
|
|---|
| 678 | setbit (i, letters);
|
|---|
| 679 | break;
|
|---|
| 680 | case CTX_NEWLINE:
|
|---|
| 681 | setbit (i, newline);
|
|---|
| 682 | break;
|
|---|
| 683 | }
|
|---|
| 684 | }
|
|---|
| 685 | }
|
|---|
| 686 |
|
|---|
| 687 | /* Set a bit in the charclass for the given wchar_t. Do nothing if WC
|
|---|
| 688 | is represented by a multi-byte sequence. Even for MB_CUR_MAX == 1,
|
|---|
| 689 | this may happen when folding case in weird Turkish locales where
|
|---|
| 690 | dotless i/dotted I are not included in the chosen character set.
|
|---|
| 691 | Return whether a bit was set in the charclass. */
|
|---|
| 692 | #if MBS_SUPPORT
|
|---|
| 693 | static bool
|
|---|
| 694 | setbit_wc (wint_t wc, charclass c)
|
|---|
| 695 | {
|
|---|
| 696 | int b = wctob (wc);
|
|---|
| 697 | if (b == EOF)
|
|---|
| 698 | return false;
|
|---|
| 699 |
|
|---|
| 700 | setbit (b, c);
|
|---|
| 701 | return true;
|
|---|
| 702 | }
|
|---|
| 703 |
|
|---|
| 704 | /* Set a bit in the charclass for the given single byte character,
|
|---|
| 705 | if it is valid in the current character set. */
|
|---|
| 706 | static void
|
|---|
| 707 | setbit_c (int b, charclass c)
|
|---|
| 708 | {
|
|---|
| 709 | /* Do nothing if b is invalid in this character set. */
|
|---|
| 710 | if (MB_CUR_MAX > 1 && btowc (b) == WEOF)
|
|---|
| 711 | return;
|
|---|
| 712 | setbit (b, c);
|
|---|
| 713 | }
|
|---|
| 714 | #else
|
|---|
| 715 | # define setbit_c setbit
|
|---|
| 716 | static inline bool
|
|---|
| 717 | setbit_wc (wint_t wc, charclass c)
|
|---|
| 718 | {
|
|---|
| 719 | abort ();
|
|---|
| 720 | /*NOTREACHED*/ return false;
|
|---|
| 721 | }
|
|---|
| 722 | #endif
|
|---|
| 723 |
|
|---|
| 724 | /* Like setbit_c, but if case is folded, set both cases of a letter. For
|
|---|
| 725 | MB_CUR_MAX > 1, the resulting charset is only used as an optimization,
|
|---|
| 726 | and the caller takes care of setting the appropriate field of struct
|
|---|
| 727 | mb_char_classes. */
|
|---|
| 728 | static void
|
|---|
| 729 | setbit_case_fold_c (int b, charclass c)
|
|---|
| 730 | {
|
|---|
| 731 | if (MB_CUR_MAX > 1)
|
|---|
| 732 | {
|
|---|
| 733 | wint_t wc = btowc (b);
|
|---|
| 734 | if (wc == WEOF)
|
|---|
| 735 | return;
|
|---|
| 736 | setbit (b, c);
|
|---|
| 737 | if (case_fold && iswalpha (wc))
|
|---|
| 738 | setbit_wc (iswupper (wc) ? towlower (wc) : towupper (wc), c);
|
|---|
| 739 | }
|
|---|
| 740 | else
|
|---|
| 741 | {
|
|---|
| 742 | setbit (b, c);
|
|---|
| 743 | if (case_fold && isalpha (b))
|
|---|
| 744 | setbit_c (isupper (b) ? tolower (b) : toupper (b), c);
|
|---|
| 745 | }
|
|---|
| 746 | }
|
|---|
| 747 |
|
|---|
| 748 |
|
|---|
| 749 |
|
|---|
| 750 | /* UTF-8 encoding allows some optimizations that we can't otherwise
|
|---|
| 751 | assume in a multibyte encoding. */
|
|---|
| 752 | static inline int
|
|---|
| 753 | using_utf8 (void)
|
|---|
| 754 | {
|
|---|
| 755 | static int utf8 = -1;
|
|---|
| 756 | if (utf8 == -1)
|
|---|
| 757 | {
|
|---|
| 758 | #if defined HAVE_LANGINFO_CODESET && MBS_SUPPORT
|
|---|
| 759 | utf8 = (STREQ (nl_langinfo (CODESET), "UTF-8"));
|
|---|
| 760 | #else
|
|---|
| 761 | utf8 = 0;
|
|---|
| 762 | #endif
|
|---|
| 763 | }
|
|---|
| 764 |
|
|---|
| 765 | return utf8;
|
|---|
| 766 | }
|
|---|
| 767 |
|
|---|
| 768 | /* Lexical analyzer. All the dross that deals with the obnoxious
|
|---|
| 769 | GNU Regex syntax bits is located here. The poor, suffering
|
|---|
| 770 | reader is referred to the GNU Regex documentation for the
|
|---|
| 771 | meaning of the @#%!@#%^!@ syntax bits. */
|
|---|
| 772 |
|
|---|
| 773 | static char const *lexptr; /* Pointer to next input character. */
|
|---|
| 774 | static size_t lexleft; /* Number of characters remaining. */
|
|---|
| 775 | static token lasttok; /* Previous token returned; initially END. */
|
|---|
| 776 | static int laststart; /* True if we're separated from beginning or (, |
|
|---|
| 777 | only by zero-width characters. */
|
|---|
| 778 | static size_t parens; /* Count of outstanding left parens. */
|
|---|
| 779 | static int minrep, maxrep; /* Repeat counts for {m,n}. */
|
|---|
| 780 | static int hard_LC_COLLATE; /* Nonzero if LC_COLLATE is hard. */
|
|---|
| 781 |
|
|---|
| 782 | static int cur_mb_len = 1; /* Length of the multibyte representation of
|
|---|
| 783 | wctok. */
|
|---|
| 784 | /* These variables are used only if (MB_CUR_MAX > 1). */
|
|---|
| 785 | static mbstate_t mbs; /* Mbstate for mbrlen(). */
|
|---|
| 786 | static wchar_t wctok; /* Wide character representation of the current
|
|---|
| 787 | multibyte character. */
|
|---|
| 788 | static unsigned char *mblen_buf; /* Correspond to the input buffer in dfaexec().
|
|---|
| 789 | Each element store the amount of remain
|
|---|
| 790 | byte of corresponding multibyte character
|
|---|
| 791 | in the input string. A element's value
|
|---|
| 792 | is 0 if corresponding character is a
|
|---|
| 793 | single byte character.
|
|---|
| 794 | e.g. input : 'a', <mb(0)>, <mb(1)>, <mb(2)>
|
|---|
| 795 | mblen_buf : 0, 3, 2, 1
|
|---|
| 796 | */
|
|---|
| 797 | static wchar_t *inputwcs; /* Wide character representation of input
|
|---|
| 798 | string in dfaexec().
|
|---|
| 799 | The length of this array is same as
|
|---|
| 800 | the length of input string(char array).
|
|---|
| 801 | inputstring[i] is a single-byte char,
|
|---|
| 802 | or 1st byte of a multibyte char.
|
|---|
| 803 | And inputwcs[i] is the codepoint. */
|
|---|
| 804 | static unsigned char const *buf_begin; /* reference to begin in dfaexec(). */
|
|---|
| 805 | static unsigned char const *buf_end; /* reference to end in dfaexec(). */
|
|---|
| 806 |
|
|---|
| 807 |
|
|---|
| 808 | #if MBS_SUPPORT
|
|---|
| 809 | /* Note that characters become unsigned here. */
|
|---|
| 810 | # define FETCH_WC(c, wc, eoferr) \
|
|---|
| 811 | do { \
|
|---|
| 812 | if (! lexleft) \
|
|---|
| 813 | { \
|
|---|
| 814 | if ((eoferr) != 0) \
|
|---|
| 815 | dfaerror (eoferr); \
|
|---|
| 816 | else \
|
|---|
| 817 | return lasttok = END; \
|
|---|
| 818 | } \
|
|---|
| 819 | else \
|
|---|
| 820 | { \
|
|---|
| 821 | wchar_t _wc; \
|
|---|
| 822 | cur_mb_len = mbrtowc (&_wc, lexptr, lexleft, &mbs); \
|
|---|
| 823 | if (cur_mb_len <= 0) \
|
|---|
| 824 | { \
|
|---|
| 825 | cur_mb_len = 1; \
|
|---|
| 826 | --lexleft; \
|
|---|
| 827 | (wc) = (c) = to_uchar (*lexptr++); \
|
|---|
| 828 | } \
|
|---|
| 829 | else \
|
|---|
| 830 | { \
|
|---|
| 831 | lexptr += cur_mb_len; \
|
|---|
| 832 | lexleft -= cur_mb_len; \
|
|---|
| 833 | (wc) = _wc; \
|
|---|
| 834 | (c) = wctob (wc); \
|
|---|
| 835 | } \
|
|---|
| 836 | } \
|
|---|
| 837 | } while(0)
|
|---|
| 838 |
|
|---|
| 839 | # define FETCH(c, eoferr) \
|
|---|
| 840 | do { \
|
|---|
| 841 | wint_t wc; \
|
|---|
| 842 | FETCH_WC (c, wc, eoferr); \
|
|---|
| 843 | } while (0)
|
|---|
| 844 |
|
|---|
| 845 | #else
|
|---|
| 846 | /* Note that characters become unsigned here. */
|
|---|
| 847 | # define FETCH(c, eoferr) \
|
|---|
| 848 | do { \
|
|---|
| 849 | if (! lexleft) \
|
|---|
| 850 | { \
|
|---|
| 851 | if ((eoferr) != 0) \
|
|---|
| 852 | dfaerror (eoferr); \
|
|---|
| 853 | else \
|
|---|
| 854 | return lasttok = END; \
|
|---|
| 855 | } \
|
|---|
| 856 | (c) = to_uchar (*lexptr++); \
|
|---|
| 857 | --lexleft; \
|
|---|
| 858 | } while(0)
|
|---|
| 859 |
|
|---|
| 860 | # define FETCH_WC(c, unused, eoferr) FETCH (c, eoferr)
|
|---|
| 861 |
|
|---|
| 862 | #endif /* MBS_SUPPORT */
|
|---|
| 863 |
|
|---|
| 864 | #ifndef MIN
|
|---|
| 865 | # define MIN(a,b) ((a) < (b) ? (a) : (b))
|
|---|
| 866 | #endif
|
|---|
| 867 |
|
|---|
| 868 | typedef int predicate (int);
|
|---|
| 869 |
|
|---|
| 870 | /* The following list maps the names of the Posix named character classes
|
|---|
| 871 | to predicate functions that determine whether a given character is in
|
|---|
| 872 | the class. The leading [ has already been eaten by the lexical analyzer. */
|
|---|
| 873 | struct dfa_ctype
|
|---|
| 874 | {
|
|---|
| 875 | const char *name;
|
|---|
| 876 | predicate *func;
|
|---|
| 877 | bool single_byte_only;
|
|---|
| 878 | };
|
|---|
| 879 |
|
|---|
| 880 | static const struct dfa_ctype prednames[] = {
|
|---|
| 881 | {"alpha", isalpha, false},
|
|---|
| 882 | {"upper", isupper, false},
|
|---|
| 883 | {"lower", islower, false},
|
|---|
| 884 | {"digit", isdigit, true},
|
|---|
| 885 | {"xdigit", isxdigit, true},
|
|---|
| 886 | {"space", isspace, false},
|
|---|
| 887 | {"punct", ispunct, false},
|
|---|
| 888 | {"alnum", isalnum, false},
|
|---|
| 889 | {"print", isprint, false},
|
|---|
| 890 | {"graph", isgraph, false},
|
|---|
| 891 | {"cntrl", iscntrl, false},
|
|---|
| 892 | {"blank", isblank, false},
|
|---|
| 893 | {NULL, NULL, false}
|
|---|
| 894 | };
|
|---|
| 895 |
|
|---|
| 896 | static const struct dfa_ctype *_GL_ATTRIBUTE_PURE
|
|---|
| 897 | find_pred (const char *str)
|
|---|
| 898 | {
|
|---|
| 899 | unsigned int i;
|
|---|
| 900 | for (i = 0; prednames[i].name; ++i)
|
|---|
| 901 | if (STREQ (str, prednames[i].name))
|
|---|
| 902 | break;
|
|---|
| 903 |
|
|---|
| 904 | return &prednames[i];
|
|---|
| 905 | }
|
|---|
| 906 |
|
|---|
| 907 | /* Multibyte character handling sub-routine for lex.
|
|---|
| 908 | This function parse a bracket expression and build a struct
|
|---|
| 909 | mb_char_classes. */
|
|---|
| 910 | static token
|
|---|
| 911 | parse_bracket_exp (void)
|
|---|
| 912 | {
|
|---|
| 913 | int invert;
|
|---|
| 914 | int c, c1, c2;
|
|---|
| 915 | charclass ccl;
|
|---|
| 916 |
|
|---|
| 917 | /* Used to warn about [:space:].
|
|---|
| 918 | Bit 0 = first character is a colon.
|
|---|
| 919 | Bit 1 = last character is a colon.
|
|---|
| 920 | Bit 2 = includes any other character but a colon.
|
|---|
| 921 | Bit 3 = includes ranges, char/equiv classes or collation elements. */
|
|---|
| 922 | int colon_warning_state;
|
|---|
| 923 |
|
|---|
| 924 | wint_t wc;
|
|---|
| 925 | wint_t wc2;
|
|---|
| 926 | wint_t wc1 = 0;
|
|---|
| 927 |
|
|---|
| 928 | /* Work area to build a mb_char_classes. */
|
|---|
| 929 | struct mb_char_classes *work_mbc;
|
|---|
| 930 | size_t chars_al, range_sts_al, range_ends_al, ch_classes_al,
|
|---|
| 931 | equivs_al, coll_elems_al;
|
|---|
| 932 |
|
|---|
| 933 | chars_al = 0;
|
|---|
| 934 | range_sts_al = range_ends_al = 0;
|
|---|
| 935 | ch_classes_al = equivs_al = coll_elems_al = 0;
|
|---|
| 936 | if (MB_CUR_MAX > 1)
|
|---|
| 937 | {
|
|---|
| 938 | REALLOC_IF_NECESSARY (dfa->mbcsets, dfa->mbcsets_alloc,
|
|---|
| 939 | dfa->nmbcsets + 1);
|
|---|
| 940 |
|
|---|
| 941 | /* dfa->multibyte_prop[] hold the index of dfa->mbcsets.
|
|---|
| 942 | We will update dfa->multibyte_prop[] in addtok(), because we can't
|
|---|
| 943 | decide the index in dfa->tokens[]. */
|
|---|
| 944 |
|
|---|
| 945 | /* Initialize work area. */
|
|---|
| 946 | work_mbc = &(dfa->mbcsets[dfa->nmbcsets++]);
|
|---|
| 947 | memset (work_mbc, 0, sizeof *work_mbc);
|
|---|
| 948 | }
|
|---|
| 949 | else
|
|---|
| 950 | work_mbc = NULL;
|
|---|
| 951 |
|
|---|
| 952 | memset (ccl, 0, sizeof ccl);
|
|---|
| 953 | FETCH_WC (c, wc, _("unbalanced ["));
|
|---|
| 954 | if (c == '^')
|
|---|
| 955 | {
|
|---|
| 956 | FETCH_WC (c, wc, _("unbalanced ["));
|
|---|
| 957 | invert = 1;
|
|---|
| 958 | }
|
|---|
| 959 | else
|
|---|
| 960 | invert = 0;
|
|---|
| 961 |
|
|---|
| 962 | colon_warning_state = (c == ':');
|
|---|
| 963 | do
|
|---|
| 964 | {
|
|---|
| 965 | c1 = EOF; /* mark c1 is not initialized". */
|
|---|
| 966 | colon_warning_state &= ~2;
|
|---|
| 967 |
|
|---|
| 968 | /* Note that if we're looking at some other [:...:] construct,
|
|---|
| 969 | we just treat it as a bunch of ordinary characters. We can do
|
|---|
| 970 | this because we assume regex has checked for syntax errors before
|
|---|
| 971 | dfa is ever called. */
|
|---|
| 972 | if (c == '[' && (syntax_bits & RE_CHAR_CLASSES))
|
|---|
| 973 | {
|
|---|
| 974 | #define BRACKET_BUFFER_SIZE 128
|
|---|
| 975 | char str[BRACKET_BUFFER_SIZE];
|
|---|
| 976 | FETCH_WC (c1, wc1, _("unbalanced ["));
|
|---|
| 977 |
|
|---|
| 978 | /* If pattern contains `[[:', `[[.', or `[[='. */
|
|---|
| 979 | if (c1 == ':'
|
|---|
| 980 | /* TODO: handle `[[.' and `[[=' also for MB_CUR_MAX == 1. */
|
|---|
| 981 | || (MB_CUR_MAX > 1 && (c1 == '.' || c1 == '=')))
|
|---|
| 982 | {
|
|---|
| 983 | size_t len = 0;
|
|---|
| 984 | for (;;)
|
|---|
| 985 | {
|
|---|
| 986 | FETCH_WC (c, wc, _("unbalanced ["));
|
|---|
| 987 | if ((c == c1 && *lexptr == ']') || lexleft == 0)
|
|---|
| 988 | break;
|
|---|
| 989 | if (len < BRACKET_BUFFER_SIZE)
|
|---|
| 990 | str[len++] = c;
|
|---|
| 991 | else
|
|---|
| 992 | /* This is in any case an invalid class name. */
|
|---|
| 993 | str[0] = '\0';
|
|---|
| 994 | }
|
|---|
| 995 | str[len] = '\0';
|
|---|
| 996 |
|
|---|
| 997 | /* Fetch bracket. */
|
|---|
| 998 | FETCH_WC (c, wc, _("unbalanced ["));
|
|---|
| 999 | if (c1 == ':')
|
|---|
| 1000 | /* build character class. */
|
|---|
| 1001 | {
|
|---|
| 1002 | char const *class
|
|---|
| 1003 | = (case_fold && (STREQ (str, "upper")
|
|---|
| 1004 | || STREQ (str, "lower")) ? "alpha" : str);
|
|---|
| 1005 | const struct dfa_ctype *pred = find_pred (class);
|
|---|
| 1006 | if (!pred)
|
|---|
| 1007 | dfaerror (_("invalid character class"));
|
|---|
| 1008 |
|
|---|
| 1009 | if (MB_CUR_MAX > 1 && !pred->single_byte_only)
|
|---|
| 1010 | {
|
|---|
| 1011 | /* Store the character class as wctype_t. */
|
|---|
| 1012 | wctype_t wt = wctype (class);
|
|---|
| 1013 |
|
|---|
| 1014 | REALLOC_IF_NECESSARY (work_mbc->ch_classes,
|
|---|
| 1015 | ch_classes_al,
|
|---|
| 1016 | work_mbc->nch_classes + 1);
|
|---|
| 1017 | work_mbc->ch_classes[work_mbc->nch_classes++] = wt;
|
|---|
| 1018 | }
|
|---|
| 1019 |
|
|---|
| 1020 | for (c2 = 0; c2 < NOTCHAR; ++c2)
|
|---|
| 1021 | if (pred->func (c2))
|
|---|
| 1022 | setbit_case_fold_c (c2, ccl);
|
|---|
| 1023 | }
|
|---|
| 1024 |
|
|---|
| 1025 | else if (MBS_SUPPORT && (c1 == '=' || c1 == '.'))
|
|---|
| 1026 | {
|
|---|
| 1027 | char *elem = xmemdup (str, len + 1);
|
|---|
| 1028 |
|
|---|
| 1029 | if (c1 == '=')
|
|---|
| 1030 | /* build equivalence class. */
|
|---|
| 1031 | {
|
|---|
| 1032 | REALLOC_IF_NECESSARY (work_mbc->equivs,
|
|---|
| 1033 | equivs_al, work_mbc->nequivs + 1);
|
|---|
| 1034 | work_mbc->equivs[work_mbc->nequivs++] = elem;
|
|---|
| 1035 | }
|
|---|
| 1036 |
|
|---|
| 1037 | if (c1 == '.')
|
|---|
| 1038 | /* build collating element. */
|
|---|
| 1039 | {
|
|---|
| 1040 | REALLOC_IF_NECESSARY (work_mbc->coll_elems,
|
|---|
| 1041 | coll_elems_al,
|
|---|
| 1042 | work_mbc->ncoll_elems + 1);
|
|---|
| 1043 | work_mbc->coll_elems[work_mbc->ncoll_elems++] = elem;
|
|---|
| 1044 | }
|
|---|
| 1045 | }
|
|---|
| 1046 | colon_warning_state |= 8;
|
|---|
| 1047 |
|
|---|
| 1048 | /* Fetch new lookahead character. */
|
|---|
| 1049 | FETCH_WC (c1, wc1, _("unbalanced ["));
|
|---|
| 1050 | continue;
|
|---|
| 1051 | }
|
|---|
| 1052 |
|
|---|
| 1053 | /* We treat '[' as a normal character here. c/c1/wc/wc1
|
|---|
| 1054 | are already set up. */
|
|---|
| 1055 | }
|
|---|
| 1056 |
|
|---|
| 1057 | if (c == '\\' && (syntax_bits & RE_BACKSLASH_ESCAPE_IN_LISTS))
|
|---|
| 1058 | FETCH_WC (c, wc, _("unbalanced ["));
|
|---|
| 1059 |
|
|---|
| 1060 | if (c1 == EOF)
|
|---|
| 1061 | FETCH_WC (c1, wc1, _("unbalanced ["));
|
|---|
| 1062 |
|
|---|
| 1063 | if (c1 == '-')
|
|---|
| 1064 | /* build range characters. */
|
|---|
| 1065 | {
|
|---|
| 1066 | FETCH_WC (c2, wc2, _("unbalanced ["));
|
|---|
| 1067 | if (c2 == ']')
|
|---|
| 1068 | {
|
|---|
| 1069 | /* In the case [x-], the - is an ordinary hyphen,
|
|---|
| 1070 | which is left in c1, the lookahead character. */
|
|---|
| 1071 | lexptr -= cur_mb_len;
|
|---|
| 1072 | lexleft += cur_mb_len;
|
|---|
| 1073 | }
|
|---|
| 1074 | }
|
|---|
| 1075 |
|
|---|
| 1076 | if (c1 == '-' && c2 != ']')
|
|---|
| 1077 | {
|
|---|
| 1078 | if (c2 == '\\' && (syntax_bits & RE_BACKSLASH_ESCAPE_IN_LISTS))
|
|---|
| 1079 | FETCH_WC (c2, wc2, _("unbalanced ["));
|
|---|
| 1080 |
|
|---|
| 1081 | if (MB_CUR_MAX > 1)
|
|---|
| 1082 | {
|
|---|
| 1083 | /* When case folding map a range, say [m-z] (or even [M-z])
|
|---|
| 1084 | to the pair of ranges, [m-z] [M-Z]. */
|
|---|
| 1085 | REALLOC_IF_NECESSARY (work_mbc->range_sts,
|
|---|
| 1086 | range_sts_al, work_mbc->nranges + 1);
|
|---|
| 1087 | REALLOC_IF_NECESSARY (work_mbc->range_ends,
|
|---|
| 1088 | range_ends_al, work_mbc->nranges + 1);
|
|---|
| 1089 | work_mbc->range_sts[work_mbc->nranges] =
|
|---|
| 1090 | case_fold ? towlower (wc) : (wchar_t) wc;
|
|---|
| 1091 | work_mbc->range_ends[work_mbc->nranges++] =
|
|---|
| 1092 | case_fold ? towlower (wc2) : (wchar_t) wc2;
|
|---|
| 1093 |
|
|---|
| 1094 | #ifndef GREP
|
|---|
| 1095 | if (case_fold && (iswalpha (wc) || iswalpha (wc2)))
|
|---|
| 1096 | {
|
|---|
| 1097 | REALLOC_IF_NECESSARY (work_mbc->range_sts,
|
|---|
| 1098 | range_sts_al, work_mbc->nranges + 1);
|
|---|
| 1099 | work_mbc->range_sts[work_mbc->nranges] = towupper (wc);
|
|---|
| 1100 | REALLOC_IF_NECESSARY (work_mbc->range_ends,
|
|---|
| 1101 | range_ends_al, work_mbc->nranges + 1);
|
|---|
| 1102 | work_mbc->range_ends[work_mbc->nranges++] = towupper (wc2);
|
|---|
| 1103 | }
|
|---|
| 1104 | #endif
|
|---|
| 1105 | }
|
|---|
| 1106 | else
|
|---|
| 1107 | {
|
|---|
| 1108 | c1 = c;
|
|---|
| 1109 | if (case_fold)
|
|---|
| 1110 | {
|
|---|
| 1111 | c1 = tolower (c1);
|
|---|
| 1112 | c2 = tolower (c2);
|
|---|
| 1113 | }
|
|---|
| 1114 | if (!hard_LC_COLLATE)
|
|---|
| 1115 | for (c = c1; c <= c2; c++)
|
|---|
| 1116 | setbit_case_fold_c (c, ccl);
|
|---|
| 1117 | else
|
|---|
| 1118 | {
|
|---|
| 1119 | /* Defer to the system regex library about the meaning
|
|---|
| 1120 | of range expressions. */
|
|---|
| 1121 | regex_t re;
|
|---|
| 1122 | char pattern[6] = { '[', c1, '-', c2, ']', 0 };
|
|---|
| 1123 | char subject[2] = { 0, 0 };
|
|---|
| 1124 | regcomp (&re, pattern, REG_NOSUB);
|
|---|
| 1125 | for (c = 0; c < NOTCHAR; ++c)
|
|---|
| 1126 | {
|
|---|
| 1127 | subject[0] = c;
|
|---|
| 1128 | if (!(case_fold && isupper (c))
|
|---|
| 1129 | && regexec (&re, subject, 0, NULL, 0) != REG_NOMATCH)
|
|---|
| 1130 | setbit_case_fold_c (c, ccl);
|
|---|
| 1131 | }
|
|---|
| 1132 | regfree (&re);
|
|---|
| 1133 | }
|
|---|
| 1134 | }
|
|---|
| 1135 |
|
|---|
| 1136 | colon_warning_state |= 8;
|
|---|
| 1137 | FETCH_WC (c1, wc1, _("unbalanced ["));
|
|---|
| 1138 | continue;
|
|---|
| 1139 | }
|
|---|
| 1140 |
|
|---|
| 1141 | colon_warning_state |= (c == ':') ? 2 : 4;
|
|---|
| 1142 |
|
|---|
| 1143 | if (MB_CUR_MAX == 1)
|
|---|
| 1144 | {
|
|---|
| 1145 | setbit_case_fold_c (c, ccl);
|
|---|
| 1146 | continue;
|
|---|
| 1147 | }
|
|---|
| 1148 |
|
|---|
| 1149 | if (case_fold && iswalpha (wc))
|
|---|
| 1150 | {
|
|---|
| 1151 | wc = towlower (wc);
|
|---|
| 1152 | if (!setbit_wc (wc, ccl))
|
|---|
| 1153 | {
|
|---|
| 1154 | REALLOC_IF_NECESSARY (work_mbc->chars, chars_al,
|
|---|
| 1155 | work_mbc->nchars + 1);
|
|---|
| 1156 | work_mbc->chars[work_mbc->nchars++] = wc;
|
|---|
| 1157 | }
|
|---|
| 1158 | #ifdef GREP
|
|---|
| 1159 | continue;
|
|---|
| 1160 | #else
|
|---|
| 1161 | wc = towupper (wc);
|
|---|
| 1162 | #endif
|
|---|
| 1163 | }
|
|---|
| 1164 | if (!setbit_wc (wc, ccl))
|
|---|
| 1165 | {
|
|---|
| 1166 | REALLOC_IF_NECESSARY (work_mbc->chars, chars_al,
|
|---|
| 1167 | work_mbc->nchars + 1);
|
|---|
| 1168 | work_mbc->chars[work_mbc->nchars++] = wc;
|
|---|
| 1169 | }
|
|---|
| 1170 | }
|
|---|
| 1171 | while ((wc = wc1, (c = c1) != ']'));
|
|---|
| 1172 |
|
|---|
| 1173 | if (colon_warning_state == 7)
|
|---|
| 1174 | dfawarn (_("character class syntax is [[:space:]], not [:space:]"));
|
|---|
| 1175 |
|
|---|
| 1176 | if (MB_CUR_MAX > 1)
|
|---|
| 1177 | {
|
|---|
| 1178 | static charclass zeroclass;
|
|---|
| 1179 | work_mbc->invert = invert;
|
|---|
| 1180 | work_mbc->cset = equal (ccl, zeroclass) ? -1 : charclass_index (ccl);
|
|---|
| 1181 | return MBCSET;
|
|---|
| 1182 | }
|
|---|
| 1183 |
|
|---|
| 1184 | if (invert)
|
|---|
| 1185 | {
|
|---|
| 1186 | assert (MB_CUR_MAX == 1);
|
|---|
| 1187 | notset (ccl);
|
|---|
| 1188 | if (syntax_bits & RE_HAT_LISTS_NOT_NEWLINE)
|
|---|
| 1189 | clrbit (eolbyte, ccl);
|
|---|
| 1190 | }
|
|---|
| 1191 |
|
|---|
| 1192 | return CSET + charclass_index (ccl);
|
|---|
| 1193 | }
|
|---|
| 1194 |
|
|---|
| 1195 | static token
|
|---|
| 1196 | lex (void)
|
|---|
| 1197 | {
|
|---|
| 1198 | unsigned int c, c2;
|
|---|
| 1199 | int backslash = 0;
|
|---|
| 1200 | charclass ccl;
|
|---|
| 1201 | int i;
|
|---|
| 1202 |
|
|---|
| 1203 | /* Basic plan: We fetch a character. If it's a backslash,
|
|---|
| 1204 | we set the backslash flag and go through the loop again.
|
|---|
| 1205 | On the plus side, this avoids having a duplicate of the
|
|---|
| 1206 | main switch inside the backslash case. On the minus side,
|
|---|
| 1207 | it means that just about every case begins with
|
|---|
| 1208 | "if (backslash) ...". */
|
|---|
| 1209 | for (i = 0; i < 2; ++i)
|
|---|
| 1210 | {
|
|---|
| 1211 | if (MB_CUR_MAX > 1)
|
|---|
| 1212 | {
|
|---|
| 1213 | FETCH_WC (c, wctok, NULL);
|
|---|
| 1214 | if ((int) c == EOF)
|
|---|
| 1215 | goto normal_char;
|
|---|
| 1216 | }
|
|---|
| 1217 | else
|
|---|
| 1218 | FETCH (c, NULL);
|
|---|
| 1219 |
|
|---|
| 1220 | switch (c)
|
|---|
| 1221 | {
|
|---|
| 1222 | case '\\':
|
|---|
| 1223 | if (backslash)
|
|---|
| 1224 | goto normal_char;
|
|---|
| 1225 | if (lexleft == 0)
|
|---|
| 1226 | dfaerror (_("unfinished \\ escape"));
|
|---|
| 1227 | backslash = 1;
|
|---|
| 1228 | break;
|
|---|
| 1229 |
|
|---|
| 1230 | case '^':
|
|---|
| 1231 | if (backslash)
|
|---|
| 1232 | goto normal_char;
|
|---|
| 1233 | if (syntax_bits & RE_CONTEXT_INDEP_ANCHORS
|
|---|
| 1234 | || lasttok == END || lasttok == LPAREN || lasttok == OR)
|
|---|
| 1235 | return lasttok = BEGLINE;
|
|---|
| 1236 | goto normal_char;
|
|---|
| 1237 |
|
|---|
| 1238 | case '$':
|
|---|
| 1239 | if (backslash)
|
|---|
| 1240 | goto normal_char;
|
|---|
| 1241 | if (syntax_bits & RE_CONTEXT_INDEP_ANCHORS
|
|---|
| 1242 | || lexleft == 0
|
|---|
| 1243 | || (syntax_bits & RE_NO_BK_PARENS
|
|---|
| 1244 | ? lexleft > 0 && *lexptr == ')'
|
|---|
| 1245 | : lexleft > 1 && lexptr[0] == '\\' && lexptr[1] == ')')
|
|---|
| 1246 | || (syntax_bits & RE_NO_BK_VBAR
|
|---|
| 1247 | ? lexleft > 0 && *lexptr == '|'
|
|---|
| 1248 | : lexleft > 1 && lexptr[0] == '\\' && lexptr[1] == '|')
|
|---|
| 1249 | || ((syntax_bits & RE_NEWLINE_ALT)
|
|---|
| 1250 | && lexleft > 0 && *lexptr == '\n'))
|
|---|
| 1251 | return lasttok = ENDLINE;
|
|---|
| 1252 | goto normal_char;
|
|---|
| 1253 |
|
|---|
| 1254 | case '1':
|
|---|
| 1255 | case '2':
|
|---|
| 1256 | case '3':
|
|---|
| 1257 | case '4':
|
|---|
| 1258 | case '5':
|
|---|
| 1259 | case '6':
|
|---|
| 1260 | case '7':
|
|---|
| 1261 | case '8':
|
|---|
| 1262 | case '9':
|
|---|
| 1263 | if (backslash && !(syntax_bits & RE_NO_BK_REFS))
|
|---|
| 1264 | {
|
|---|
| 1265 | laststart = 0;
|
|---|
| 1266 | return lasttok = BACKREF;
|
|---|
| 1267 | }
|
|---|
| 1268 | goto normal_char;
|
|---|
| 1269 |
|
|---|
| 1270 | case '`':
|
|---|
| 1271 | if (backslash && !(syntax_bits & RE_NO_GNU_OPS))
|
|---|
| 1272 | return lasttok = BEGLINE; /* FIXME: should be beginning of string */
|
|---|
| 1273 | goto normal_char;
|
|---|
| 1274 |
|
|---|
| 1275 | case '\'':
|
|---|
| 1276 | if (backslash && !(syntax_bits & RE_NO_GNU_OPS))
|
|---|
| 1277 | return lasttok = ENDLINE; /* FIXME: should be end of string */
|
|---|
| 1278 | goto normal_char;
|
|---|
| 1279 |
|
|---|
| 1280 | case '<':
|
|---|
| 1281 | if (backslash && !(syntax_bits & RE_NO_GNU_OPS))
|
|---|
| 1282 | return lasttok = BEGWORD;
|
|---|
| 1283 | goto normal_char;
|
|---|
| 1284 |
|
|---|
| 1285 | case '>':
|
|---|
| 1286 | if (backslash && !(syntax_bits & RE_NO_GNU_OPS))
|
|---|
| 1287 | return lasttok = ENDWORD;
|
|---|
| 1288 | goto normal_char;
|
|---|
| 1289 |
|
|---|
| 1290 | case 'b':
|
|---|
| 1291 | if (backslash && !(syntax_bits & RE_NO_GNU_OPS))
|
|---|
| 1292 | return lasttok = LIMWORD;
|
|---|
| 1293 | goto normal_char;
|
|---|
| 1294 |
|
|---|
| 1295 | case 'B':
|
|---|
| 1296 | if (backslash && !(syntax_bits & RE_NO_GNU_OPS))
|
|---|
| 1297 | return lasttok = NOTLIMWORD;
|
|---|
| 1298 | goto normal_char;
|
|---|
| 1299 |
|
|---|
| 1300 | case '?':
|
|---|
| 1301 | if (syntax_bits & RE_LIMITED_OPS)
|
|---|
| 1302 | goto normal_char;
|
|---|
| 1303 | if (backslash != ((syntax_bits & RE_BK_PLUS_QM) != 0))
|
|---|
| 1304 | goto normal_char;
|
|---|
| 1305 | if (!(syntax_bits & RE_CONTEXT_INDEP_OPS) && laststart)
|
|---|
| 1306 | goto normal_char;
|
|---|
| 1307 | return lasttok = QMARK;
|
|---|
| 1308 |
|
|---|
| 1309 | case '*':
|
|---|
| 1310 | if (backslash)
|
|---|
| 1311 | goto normal_char;
|
|---|
| 1312 | if (!(syntax_bits & RE_CONTEXT_INDEP_OPS) && laststart)
|
|---|
| 1313 | goto normal_char;
|
|---|
| 1314 | return lasttok = STAR;
|
|---|
| 1315 |
|
|---|
| 1316 | case '+':
|
|---|
| 1317 | if (syntax_bits & RE_LIMITED_OPS)
|
|---|
| 1318 | goto normal_char;
|
|---|
| 1319 | if (backslash != ((syntax_bits & RE_BK_PLUS_QM) != 0))
|
|---|
| 1320 | goto normal_char;
|
|---|
| 1321 | if (!(syntax_bits & RE_CONTEXT_INDEP_OPS) && laststart)
|
|---|
| 1322 | goto normal_char;
|
|---|
| 1323 | return lasttok = PLUS;
|
|---|
| 1324 |
|
|---|
| 1325 | case '{':
|
|---|
| 1326 | if (!(syntax_bits & RE_INTERVALS))
|
|---|
| 1327 | goto normal_char;
|
|---|
| 1328 | if (backslash != ((syntax_bits & RE_NO_BK_BRACES) == 0))
|
|---|
| 1329 | goto normal_char;
|
|---|
| 1330 | if (!(syntax_bits & RE_CONTEXT_INDEP_OPS) && laststart)
|
|---|
| 1331 | goto normal_char;
|
|---|
| 1332 |
|
|---|
| 1333 | /* Cases:
|
|---|
| 1334 | {M} - exact count
|
|---|
| 1335 | {M,} - minimum count, maximum is infinity
|
|---|
| 1336 | {,N} - 0 through N
|
|---|
| 1337 | {,} - 0 to infinity (same as '*')
|
|---|
| 1338 | {M,N} - M through N */
|
|---|
| 1339 | {
|
|---|
| 1340 | char const *p = lexptr;
|
|---|
| 1341 | char const *lim = p + lexleft;
|
|---|
| 1342 | minrep = maxrep = -1;
|
|---|
| 1343 | for (; p != lim && ISASCIIDIGIT (*p); p++)
|
|---|
| 1344 | {
|
|---|
| 1345 | if (minrep < 0)
|
|---|
| 1346 | minrep = *p - '0';
|
|---|
| 1347 | else
|
|---|
| 1348 | minrep = MIN (RE_DUP_MAX + 1, minrep * 10 + *p - '0');
|
|---|
| 1349 | }
|
|---|
| 1350 | if (p != lim)
|
|---|
| 1351 | {
|
|---|
| 1352 | if (*p != ',')
|
|---|
| 1353 | maxrep = minrep;
|
|---|
| 1354 | else
|
|---|
| 1355 | {
|
|---|
| 1356 | if (minrep < 0)
|
|---|
| 1357 | minrep = 0;
|
|---|
| 1358 | while (++p != lim && ISASCIIDIGIT (*p))
|
|---|
| 1359 | {
|
|---|
| 1360 | if (maxrep < 0)
|
|---|
| 1361 | maxrep = *p - '0';
|
|---|
| 1362 | else
|
|---|
| 1363 | maxrep = MIN (RE_DUP_MAX + 1, maxrep * 10 + *p - '0');
|
|---|
| 1364 | }
|
|---|
| 1365 | }
|
|---|
| 1366 | }
|
|---|
| 1367 | if (! ((! backslash || (p != lim && *p++ == '\\'))
|
|---|
| 1368 | && p != lim && *p++ == '}'
|
|---|
| 1369 | && 0 <= minrep && (maxrep < 0 || minrep <= maxrep)))
|
|---|
| 1370 | {
|
|---|
| 1371 | if (syntax_bits & RE_INVALID_INTERVAL_ORD)
|
|---|
| 1372 | goto normal_char;
|
|---|
| 1373 | dfaerror (_("Invalid content of \\{\\}"));
|
|---|
| 1374 | }
|
|---|
| 1375 | if (RE_DUP_MAX < maxrep)
|
|---|
| 1376 | dfaerror (_("Regular expression too big"));
|
|---|
| 1377 | lexptr = p;
|
|---|
| 1378 | lexleft = lim - p;
|
|---|
| 1379 | }
|
|---|
| 1380 | laststart = 0;
|
|---|
| 1381 | return lasttok = REPMN;
|
|---|
| 1382 |
|
|---|
| 1383 | case '|':
|
|---|
| 1384 | if (syntax_bits & RE_LIMITED_OPS)
|
|---|
| 1385 | goto normal_char;
|
|---|
| 1386 | if (backslash != ((syntax_bits & RE_NO_BK_VBAR) == 0))
|
|---|
| 1387 | goto normal_char;
|
|---|
| 1388 | laststart = 1;
|
|---|
| 1389 | return lasttok = OR;
|
|---|
| 1390 |
|
|---|
| 1391 | case '\n':
|
|---|
| 1392 | if (syntax_bits & RE_LIMITED_OPS
|
|---|
| 1393 | || backslash || !(syntax_bits & RE_NEWLINE_ALT))
|
|---|
| 1394 | goto normal_char;
|
|---|
| 1395 | laststart = 1;
|
|---|
| 1396 | return lasttok = OR;
|
|---|
| 1397 |
|
|---|
| 1398 | case '(':
|
|---|
| 1399 | if (backslash != ((syntax_bits & RE_NO_BK_PARENS) == 0))
|
|---|
| 1400 | goto normal_char;
|
|---|
| 1401 | ++parens;
|
|---|
| 1402 | laststart = 1;
|
|---|
| 1403 | return lasttok = LPAREN;
|
|---|
| 1404 |
|
|---|
| 1405 | case ')':
|
|---|
| 1406 | if (backslash != ((syntax_bits & RE_NO_BK_PARENS) == 0))
|
|---|
| 1407 | goto normal_char;
|
|---|
| 1408 | if (parens == 0 && syntax_bits & RE_UNMATCHED_RIGHT_PAREN_ORD)
|
|---|
| 1409 | goto normal_char;
|
|---|
| 1410 | --parens;
|
|---|
| 1411 | laststart = 0;
|
|---|
| 1412 | return lasttok = RPAREN;
|
|---|
| 1413 |
|
|---|
| 1414 | case '.':
|
|---|
| 1415 | if (backslash)
|
|---|
| 1416 | goto normal_char;
|
|---|
| 1417 | if (MB_CUR_MAX > 1)
|
|---|
| 1418 | {
|
|---|
| 1419 | /* In multibyte environment period must match with a single
|
|---|
| 1420 | character not a byte. So we use ANYCHAR. */
|
|---|
| 1421 | laststart = 0;
|
|---|
| 1422 | return lasttok = ANYCHAR;
|
|---|
| 1423 | }
|
|---|
| 1424 | zeroset (ccl);
|
|---|
| 1425 | notset (ccl);
|
|---|
| 1426 | if (!(syntax_bits & RE_DOT_NEWLINE))
|
|---|
| 1427 | clrbit (eolbyte, ccl);
|
|---|
| 1428 | if (syntax_bits & RE_DOT_NOT_NULL)
|
|---|
| 1429 | clrbit ('\0', ccl);
|
|---|
| 1430 | laststart = 0;
|
|---|
| 1431 | return lasttok = CSET + charclass_index (ccl);
|
|---|
| 1432 |
|
|---|
| 1433 | case 's':
|
|---|
| 1434 | case 'S':
|
|---|
| 1435 | if (!backslash || (syntax_bits & RE_NO_GNU_OPS))
|
|---|
| 1436 | goto normal_char;
|
|---|
| 1437 | zeroset (ccl);
|
|---|
| 1438 | for (c2 = 0; c2 < NOTCHAR; ++c2)
|
|---|
| 1439 | if (isspace (c2))
|
|---|
| 1440 | setbit (c2, ccl);
|
|---|
| 1441 | if (c == 'S')
|
|---|
| 1442 | notset (ccl);
|
|---|
| 1443 | laststart = 0;
|
|---|
| 1444 | return lasttok = CSET + charclass_index (ccl);
|
|---|
| 1445 |
|
|---|
| 1446 | case 'w':
|
|---|
| 1447 | case 'W':
|
|---|
| 1448 | if (!backslash || (syntax_bits & RE_NO_GNU_OPS))
|
|---|
| 1449 | goto normal_char;
|
|---|
| 1450 | zeroset (ccl);
|
|---|
| 1451 | for (c2 = 0; c2 < NOTCHAR; ++c2)
|
|---|
| 1452 | if (IS_WORD_CONSTITUENT (c2))
|
|---|
| 1453 | setbit (c2, ccl);
|
|---|
| 1454 | if (c == 'W')
|
|---|
| 1455 | notset (ccl);
|
|---|
| 1456 | laststart = 0;
|
|---|
| 1457 | return lasttok = CSET + charclass_index (ccl);
|
|---|
| 1458 |
|
|---|
| 1459 | case '[':
|
|---|
| 1460 | if (backslash)
|
|---|
| 1461 | goto normal_char;
|
|---|
| 1462 | laststart = 0;
|
|---|
| 1463 | return lasttok = parse_bracket_exp ();
|
|---|
| 1464 |
|
|---|
| 1465 | default:
|
|---|
| 1466 | normal_char:
|
|---|
| 1467 | laststart = 0;
|
|---|
| 1468 | /* For multibyte character sets, folding is done in atom. Always
|
|---|
| 1469 | return WCHAR. */
|
|---|
| 1470 | if (MB_CUR_MAX > 1)
|
|---|
| 1471 | return lasttok = WCHAR;
|
|---|
| 1472 |
|
|---|
| 1473 | if (case_fold && isalpha (c))
|
|---|
| 1474 | {
|
|---|
| 1475 | zeroset (ccl);
|
|---|
| 1476 | setbit_case_fold_c (c, ccl);
|
|---|
| 1477 | return lasttok = CSET + charclass_index (ccl);
|
|---|
| 1478 | }
|
|---|
| 1479 |
|
|---|
| 1480 | return lasttok = c;
|
|---|
| 1481 | }
|
|---|
| 1482 | }
|
|---|
| 1483 |
|
|---|
| 1484 | /* The above loop should consume at most a backslash
|
|---|
| 1485 | and some other character. */
|
|---|
| 1486 | abort ();
|
|---|
| 1487 | return END; /* keeps pedantic compilers happy. */
|
|---|
| 1488 | }
|
|---|
| 1489 |
|
|---|
| 1490 | /* Recursive descent parser for regular expressions. */
|
|---|
| 1491 |
|
|---|
| 1492 | static token tok; /* Lookahead token. */
|
|---|
| 1493 | static size_t depth; /* Current depth of a hypothetical stack
|
|---|
| 1494 | holding deferred productions. This is
|
|---|
| 1495 | used to determine the depth that will be
|
|---|
| 1496 | required of the real stack later on in
|
|---|
| 1497 | dfaanalyze(). */
|
|---|
| 1498 |
|
|---|
| 1499 | static void
|
|---|
| 1500 | addtok_mb (token t, int mbprop)
|
|---|
| 1501 | {
|
|---|
| 1502 | if (MB_CUR_MAX > 1)
|
|---|
| 1503 | {
|
|---|
| 1504 | REALLOC_IF_NECESSARY (dfa->multibyte_prop, dfa->nmultibyte_prop,
|
|---|
| 1505 | dfa->tindex + 1);
|
|---|
| 1506 | dfa->multibyte_prop[dfa->tindex] = mbprop;
|
|---|
| 1507 | }
|
|---|
| 1508 |
|
|---|
| 1509 | REALLOC_IF_NECESSARY (dfa->tokens, dfa->talloc, dfa->tindex + 1);
|
|---|
| 1510 | dfa->tokens[dfa->tindex++] = t;
|
|---|
| 1511 |
|
|---|
| 1512 | switch (t)
|
|---|
| 1513 | {
|
|---|
| 1514 | case QMARK:
|
|---|
| 1515 | case STAR:
|
|---|
| 1516 | case PLUS:
|
|---|
| 1517 | break;
|
|---|
| 1518 |
|
|---|
| 1519 | case CAT:
|
|---|
| 1520 | case OR:
|
|---|
| 1521 | --depth;
|
|---|
| 1522 | break;
|
|---|
| 1523 |
|
|---|
| 1524 | default:
|
|---|
| 1525 | ++dfa->nleaves;
|
|---|
| 1526 | case EMPTY:
|
|---|
| 1527 | ++depth;
|
|---|
| 1528 | break;
|
|---|
| 1529 | }
|
|---|
| 1530 | if (depth > dfa->depth)
|
|---|
| 1531 | dfa->depth = depth;
|
|---|
| 1532 | }
|
|---|
| 1533 |
|
|---|
| 1534 | static void addtok_wc (wint_t wc);
|
|---|
| 1535 |
|
|---|
| 1536 | /* Add the given token to the parse tree, maintaining the depth count and
|
|---|
| 1537 | updating the maximum depth if necessary. */
|
|---|
| 1538 | static void
|
|---|
| 1539 | addtok (token t)
|
|---|
| 1540 | {
|
|---|
| 1541 | if (MB_CUR_MAX > 1 && t == MBCSET)
|
|---|
| 1542 | {
|
|---|
| 1543 | bool need_or = false;
|
|---|
| 1544 | struct mb_char_classes *work_mbc = &dfa->mbcsets[dfa->nmbcsets - 1];
|
|---|
| 1545 |
|
|---|
| 1546 | /* Extract wide characters into alternations for better performance.
|
|---|
| 1547 | This does not require UTF-8. */
|
|---|
| 1548 | if (!work_mbc->invert)
|
|---|
| 1549 | {
|
|---|
| 1550 | size_t i;
|
|---|
| 1551 | for (i = 0; i < work_mbc->nchars; i++)
|
|---|
| 1552 | {
|
|---|
| 1553 | addtok_wc (work_mbc->chars[i]);
|
|---|
| 1554 | if (need_or)
|
|---|
| 1555 | addtok (OR);
|
|---|
| 1556 | need_or = true;
|
|---|
| 1557 | }
|
|---|
| 1558 | work_mbc->nchars = 0;
|
|---|
| 1559 | }
|
|---|
| 1560 |
|
|---|
| 1561 | /* UTF-8 allows treating a simple, non-inverted MBCSET like a CSET. */
|
|---|
| 1562 | if (work_mbc->invert
|
|---|
| 1563 | || (!using_utf8 () && work_mbc->cset != -1)
|
|---|
| 1564 | || work_mbc->nchars != 0
|
|---|
| 1565 | || work_mbc->nch_classes != 0
|
|---|
| 1566 | || work_mbc->nranges != 0
|
|---|
| 1567 | || work_mbc->nequivs != 0 || work_mbc->ncoll_elems != 0)
|
|---|
| 1568 | {
|
|---|
| 1569 | addtok_mb (MBCSET, ((dfa->nmbcsets - 1) << 2) + 3);
|
|---|
| 1570 | if (need_or)
|
|---|
| 1571 | addtok (OR);
|
|---|
| 1572 | }
|
|---|
| 1573 | else
|
|---|
| 1574 | {
|
|---|
| 1575 | /* Characters have been handled above, so it is possible
|
|---|
| 1576 | that the mbcset is empty now. Do nothing in that case. */
|
|---|
| 1577 | if (work_mbc->cset != -1)
|
|---|
| 1578 | {
|
|---|
| 1579 | assert (using_utf8 ());
|
|---|
| 1580 | addtok (CSET + work_mbc->cset);
|
|---|
| 1581 | if (need_or)
|
|---|
| 1582 | addtok (OR);
|
|---|
| 1583 | }
|
|---|
| 1584 | }
|
|---|
| 1585 | }
|
|---|
| 1586 | else
|
|---|
| 1587 | {
|
|---|
| 1588 | addtok_mb (t, 3);
|
|---|
| 1589 | }
|
|---|
| 1590 | }
|
|---|
| 1591 |
|
|---|
| 1592 | #if MBS_SUPPORT
|
|---|
| 1593 | /* We treat a multibyte character as a single atom, so that DFA
|
|---|
| 1594 | can treat a multibyte character as a single expression.
|
|---|
| 1595 |
|
|---|
| 1596 | e.g. We construct following tree from "<mb1><mb2>".
|
|---|
| 1597 | <mb1(1st-byte)><mb1(2nd-byte)><CAT><mb1(3rd-byte)><CAT>
|
|---|
| 1598 | <mb2(1st-byte)><mb2(2nd-byte)><CAT><mb2(3rd-byte)><CAT><CAT> */
|
|---|
| 1599 | static void
|
|---|
| 1600 | addtok_wc (wint_t wc)
|
|---|
| 1601 | {
|
|---|
| 1602 | unsigned char buf[MB_LEN_MAX];
|
|---|
| 1603 | mbstate_t s;
|
|---|
| 1604 | int i;
|
|---|
| 1605 | memset (&s, 0, sizeof s);
|
|---|
| 1606 | cur_mb_len = wcrtomb ((char *) buf, wc, &s);
|
|---|
| 1607 |
|
|---|
| 1608 | /* This is merely stop-gap. When cur_mb_len is 0 or negative,
|
|---|
| 1609 | buf[0] is undefined, yet skipping the addtok_mb call altogether
|
|---|
| 1610 | can result in heap corruption. */
|
|---|
| 1611 | if (cur_mb_len <= 0)
|
|---|
| 1612 | buf[0] = 0;
|
|---|
| 1613 |
|
|---|
| 1614 | addtok_mb (buf[0], cur_mb_len == 1 ? 3 : 1);
|
|---|
| 1615 | for (i = 1; i < cur_mb_len; i++)
|
|---|
| 1616 | {
|
|---|
| 1617 | addtok_mb (buf[i], i == cur_mb_len - 1 ? 2 : 0);
|
|---|
| 1618 | addtok (CAT);
|
|---|
| 1619 | }
|
|---|
| 1620 | }
|
|---|
| 1621 | #else
|
|---|
| 1622 | static void
|
|---|
| 1623 | addtok_wc (wint_t wc)
|
|---|
| 1624 | {
|
|---|
| 1625 | }
|
|---|
| 1626 | #endif
|
|---|
| 1627 |
|
|---|
| 1628 | static void
|
|---|
| 1629 | add_utf8_anychar (void)
|
|---|
| 1630 | {
|
|---|
| 1631 | #if MBS_SUPPORT
|
|---|
| 1632 | static const charclass utf8_classes[5] = {
|
|---|
| 1633 | {0, 0, 0, 0, ~0, ~0, 0, 0}, /* 80-bf: non-lead bytes */
|
|---|
| 1634 | {~0, ~0, ~0, ~0, 0, 0, 0, 0}, /* 00-7f: 1-byte sequence */
|
|---|
| 1635 | {0, 0, 0, 0, 0, 0, 0xfffffffcU, 0}, /* c2-df: 2-byte sequence */
|
|---|
| 1636 | {0, 0, 0, 0, 0, 0, 0, 0xffff}, /* e0-ef: 3-byte sequence */
|
|---|
| 1637 | {0, 0, 0, 0, 0, 0, 0, 0xff0000} /* f0-f7: 4-byte sequence */
|
|---|
| 1638 | };
|
|---|
| 1639 | const unsigned int n = sizeof (utf8_classes) / sizeof (utf8_classes[0]);
|
|---|
| 1640 | unsigned int i;
|
|---|
| 1641 |
|
|---|
| 1642 | /* Define the five character classes that are needed below. */
|
|---|
| 1643 | if (dfa->utf8_anychar_classes[0] == 0)
|
|---|
| 1644 | for (i = 0; i < n; i++)
|
|---|
| 1645 | {
|
|---|
| 1646 | charclass c;
|
|---|
| 1647 | copyset (utf8_classes[i], c);
|
|---|
| 1648 | if (i == 1)
|
|---|
| 1649 | {
|
|---|
| 1650 | if (!(syntax_bits & RE_DOT_NEWLINE))
|
|---|
| 1651 | clrbit (eolbyte, c);
|
|---|
| 1652 | if (syntax_bits & RE_DOT_NOT_NULL)
|
|---|
| 1653 | clrbit ('\0', c);
|
|---|
| 1654 | }
|
|---|
| 1655 | dfa->utf8_anychar_classes[i] = CSET + charclass_index (c);
|
|---|
| 1656 | }
|
|---|
| 1657 |
|
|---|
| 1658 | /* A valid UTF-8 character is
|
|---|
| 1659 |
|
|---|
| 1660 | ([0x00-0x7f]
|
|---|
| 1661 | |[0xc2-0xdf][0x80-0xbf]
|
|---|
| 1662 | |[0xe0-0xef[0x80-0xbf][0x80-0xbf]
|
|---|
| 1663 | |[0xf0-f7][0x80-0xbf][0x80-0xbf][0x80-0xbf])
|
|---|
| 1664 |
|
|---|
| 1665 | which I'll write more concisely "B|CA|DAA|EAAA". Factor the [0x00-0x7f]
|
|---|
| 1666 | and you get "B|(C|(D|EA)A)A". And since the token buffer is in reverse
|
|---|
| 1667 | Polish notation, you get "B C D E A CAT OR A CAT OR A CAT OR". */
|
|---|
| 1668 | for (i = 1; i < n; i++)
|
|---|
| 1669 | addtok (dfa->utf8_anychar_classes[i]);
|
|---|
| 1670 | while (--i > 1)
|
|---|
| 1671 | {
|
|---|
| 1672 | addtok (dfa->utf8_anychar_classes[0]);
|
|---|
| 1673 | addtok (CAT);
|
|---|
| 1674 | addtok (OR);
|
|---|
| 1675 | }
|
|---|
| 1676 | #endif
|
|---|
| 1677 | }
|
|---|
| 1678 |
|
|---|
| 1679 | /* The grammar understood by the parser is as follows.
|
|---|
| 1680 |
|
|---|
| 1681 | regexp:
|
|---|
| 1682 | regexp OR branch
|
|---|
| 1683 | branch
|
|---|
| 1684 |
|
|---|
| 1685 | branch:
|
|---|
| 1686 | branch closure
|
|---|
| 1687 | closure
|
|---|
| 1688 |
|
|---|
| 1689 | closure:
|
|---|
| 1690 | closure QMARK
|
|---|
| 1691 | closure STAR
|
|---|
| 1692 | closure PLUS
|
|---|
| 1693 | closure REPMN
|
|---|
| 1694 | atom
|
|---|
| 1695 |
|
|---|
| 1696 | atom:
|
|---|
| 1697 | <normal character>
|
|---|
| 1698 | <multibyte character>
|
|---|
| 1699 | ANYCHAR
|
|---|
| 1700 | MBCSET
|
|---|
| 1701 | CSET
|
|---|
| 1702 | BACKREF
|
|---|
| 1703 | BEGLINE
|
|---|
| 1704 | ENDLINE
|
|---|
| 1705 | BEGWORD
|
|---|
| 1706 | ENDWORD
|
|---|
| 1707 | LIMWORD
|
|---|
| 1708 | NOTLIMWORD
|
|---|
| 1709 | LPAREN regexp RPAREN
|
|---|
| 1710 | <empty>
|
|---|
| 1711 |
|
|---|
| 1712 | The parser builds a parse tree in postfix form in an array of tokens. */
|
|---|
| 1713 |
|
|---|
| 1714 | static void
|
|---|
| 1715 | atom (void)
|
|---|
| 1716 | {
|
|---|
| 1717 | if (0)
|
|---|
| 1718 | {
|
|---|
| 1719 | /* empty */
|
|---|
| 1720 | }
|
|---|
| 1721 | else if (MBS_SUPPORT && tok == WCHAR)
|
|---|
| 1722 | {
|
|---|
| 1723 | addtok_wc (case_fold ? towlower (wctok) : wctok);
|
|---|
| 1724 | #ifndef GREP
|
|---|
| 1725 | if (case_fold && iswalpha (wctok))
|
|---|
| 1726 | {
|
|---|
| 1727 | addtok_wc (towupper (wctok));
|
|---|
| 1728 | addtok (OR);
|
|---|
| 1729 | }
|
|---|
| 1730 | #endif
|
|---|
| 1731 |
|
|---|
| 1732 | tok = lex ();
|
|---|
| 1733 | }
|
|---|
| 1734 | else if (MBS_SUPPORT && tok == ANYCHAR && using_utf8 ())
|
|---|
| 1735 | {
|
|---|
| 1736 | /* For UTF-8 expand the period to a series of CSETs that define a valid
|
|---|
| 1737 | UTF-8 character. This avoids using the slow multibyte path. I'm
|
|---|
| 1738 | pretty sure it would be both profitable and correct to do it for
|
|---|
| 1739 | any encoding; however, the optimization must be done manually as
|
|---|
| 1740 | it is done above in add_utf8_anychar. So, let's start with
|
|---|
| 1741 | UTF-8: it is the most used, and the structure of the encoding
|
|---|
| 1742 | makes the correctness more obvious. */
|
|---|
| 1743 | add_utf8_anychar ();
|
|---|
| 1744 | tok = lex ();
|
|---|
| 1745 | }
|
|---|
| 1746 | else if ((tok >= 0 && tok < NOTCHAR) || tok >= CSET || tok == BACKREF
|
|---|
| 1747 | || tok == BEGLINE || tok == ENDLINE || tok == BEGWORD
|
|---|
| 1748 | #if MBS_SUPPORT
|
|---|
| 1749 | || tok == ANYCHAR || tok == MBCSET
|
|---|
| 1750 | #endif /* MBS_SUPPORT */
|
|---|
| 1751 | || tok == ENDWORD || tok == LIMWORD || tok == NOTLIMWORD)
|
|---|
| 1752 | {
|
|---|
| 1753 | addtok (tok);
|
|---|
| 1754 | tok = lex ();
|
|---|
| 1755 | }
|
|---|
| 1756 | else if (tok == LPAREN)
|
|---|
| 1757 | {
|
|---|
| 1758 | tok = lex ();
|
|---|
| 1759 | regexp ();
|
|---|
| 1760 | if (tok != RPAREN)
|
|---|
| 1761 | dfaerror (_("unbalanced ("));
|
|---|
| 1762 | tok = lex ();
|
|---|
| 1763 | }
|
|---|
| 1764 | else
|
|---|
| 1765 | addtok (EMPTY);
|
|---|
| 1766 | }
|
|---|
| 1767 |
|
|---|
| 1768 | /* Return the number of tokens in the given subexpression. */
|
|---|
| 1769 | static size_t _GL_ATTRIBUTE_PURE
|
|---|
| 1770 | nsubtoks (size_t tindex)
|
|---|
| 1771 | {
|
|---|
| 1772 | size_t ntoks1;
|
|---|
| 1773 |
|
|---|
| 1774 | switch (dfa->tokens[tindex - 1])
|
|---|
| 1775 | {
|
|---|
| 1776 | default:
|
|---|
| 1777 | return 1;
|
|---|
| 1778 | case QMARK:
|
|---|
| 1779 | case STAR:
|
|---|
| 1780 | case PLUS:
|
|---|
| 1781 | return 1 + nsubtoks (tindex - 1);
|
|---|
| 1782 | case CAT:
|
|---|
| 1783 | case OR:
|
|---|
| 1784 | ntoks1 = nsubtoks (tindex - 1);
|
|---|
| 1785 | return 1 + ntoks1 + nsubtoks (tindex - 1 - ntoks1);
|
|---|
| 1786 | }
|
|---|
| 1787 | }
|
|---|
| 1788 |
|
|---|
| 1789 | /* Copy the given subexpression to the top of the tree. */
|
|---|
| 1790 | static void
|
|---|
| 1791 | copytoks (size_t tindex, size_t ntokens)
|
|---|
| 1792 | {
|
|---|
| 1793 | size_t i;
|
|---|
| 1794 |
|
|---|
| 1795 | for (i = 0; i < ntokens; ++i)
|
|---|
| 1796 | {
|
|---|
| 1797 | addtok (dfa->tokens[tindex + i]);
|
|---|
| 1798 | /* Update index into multibyte csets. */
|
|---|
| 1799 | if (MB_CUR_MAX > 1 && dfa->tokens[tindex + i] == MBCSET)
|
|---|
| 1800 | dfa->multibyte_prop[dfa->tindex - 1] = dfa->multibyte_prop[tindex + i];
|
|---|
| 1801 | }
|
|---|
| 1802 | }
|
|---|
| 1803 |
|
|---|
| 1804 | static void
|
|---|
| 1805 | closure (void)
|
|---|
| 1806 | {
|
|---|
| 1807 | int i;
|
|---|
| 1808 | size_t tindex, ntokens;
|
|---|
| 1809 |
|
|---|
| 1810 | atom ();
|
|---|
| 1811 | while (tok == QMARK || tok == STAR || tok == PLUS || tok == REPMN)
|
|---|
| 1812 | if (tok == REPMN && (minrep || maxrep))
|
|---|
| 1813 | {
|
|---|
| 1814 | ntokens = nsubtoks (dfa->tindex);
|
|---|
| 1815 | tindex = dfa->tindex - ntokens;
|
|---|
| 1816 | if (maxrep < 0)
|
|---|
| 1817 | addtok (PLUS);
|
|---|
| 1818 | if (minrep == 0)
|
|---|
| 1819 | addtok (QMARK);
|
|---|
| 1820 | for (i = 1; i < minrep; ++i)
|
|---|
| 1821 | {
|
|---|
| 1822 | copytoks (tindex, ntokens);
|
|---|
| 1823 | addtok (CAT);
|
|---|
| 1824 | }
|
|---|
| 1825 | for (; i < maxrep; ++i)
|
|---|
| 1826 | {
|
|---|
| 1827 | copytoks (tindex, ntokens);
|
|---|
| 1828 | addtok (QMARK);
|
|---|
| 1829 | addtok (CAT);
|
|---|
| 1830 | }
|
|---|
| 1831 | tok = lex ();
|
|---|
| 1832 | }
|
|---|
| 1833 | else if (tok == REPMN)
|
|---|
| 1834 | {
|
|---|
| 1835 | dfa->tindex -= nsubtoks (dfa->tindex);
|
|---|
| 1836 | tok = lex ();
|
|---|
| 1837 | closure ();
|
|---|
| 1838 | }
|
|---|
| 1839 | else
|
|---|
| 1840 | {
|
|---|
| 1841 | addtok (tok);
|
|---|
| 1842 | tok = lex ();
|
|---|
| 1843 | }
|
|---|
| 1844 | }
|
|---|
| 1845 |
|
|---|
| 1846 | static void
|
|---|
| 1847 | branch (void)
|
|---|
| 1848 | {
|
|---|
| 1849 | closure ();
|
|---|
| 1850 | while (tok != RPAREN && tok != OR && tok >= 0)
|
|---|
| 1851 | {
|
|---|
| 1852 | closure ();
|
|---|
| 1853 | addtok (CAT);
|
|---|
| 1854 | }
|
|---|
| 1855 | }
|
|---|
| 1856 |
|
|---|
| 1857 | static void
|
|---|
| 1858 | regexp (void)
|
|---|
| 1859 | {
|
|---|
| 1860 | branch ();
|
|---|
| 1861 | while (tok == OR)
|
|---|
| 1862 | {
|
|---|
| 1863 | tok = lex ();
|
|---|
| 1864 | branch ();
|
|---|
| 1865 | addtok (OR);
|
|---|
| 1866 | }
|
|---|
| 1867 | }
|
|---|
| 1868 |
|
|---|
| 1869 | /* Main entry point for the parser. S is a string to be parsed, len is the
|
|---|
| 1870 | length of the string, so s can include NUL characters. D is a pointer to
|
|---|
| 1871 | the struct dfa to parse into. */
|
|---|
| 1872 | void
|
|---|
| 1873 | dfaparse (char const *s, size_t len, struct dfa *d)
|
|---|
| 1874 | {
|
|---|
| 1875 | dfa = d;
|
|---|
| 1876 | lexptr = s;
|
|---|
| 1877 | lexleft = len;
|
|---|
| 1878 | lasttok = END;
|
|---|
| 1879 | laststart = 1;
|
|---|
| 1880 | parens = 0;
|
|---|
| 1881 | #ifdef LC_COLLATE
|
|---|
| 1882 | hard_LC_COLLATE = hard_locale (LC_COLLATE);
|
|---|
| 1883 | #endif
|
|---|
| 1884 | if (MB_CUR_MAX > 1)
|
|---|
| 1885 | {
|
|---|
| 1886 | cur_mb_len = 0;
|
|---|
| 1887 | memset (&mbs, 0, sizeof mbs);
|
|---|
| 1888 | }
|
|---|
| 1889 |
|
|---|
| 1890 | if (!syntax_bits_set)
|
|---|
| 1891 | dfaerror (_("no syntax specified"));
|
|---|
| 1892 |
|
|---|
| 1893 | tok = lex ();
|
|---|
| 1894 | depth = d->depth;
|
|---|
| 1895 |
|
|---|
| 1896 | regexp ();
|
|---|
| 1897 |
|
|---|
| 1898 | if (tok != END)
|
|---|
| 1899 | dfaerror (_("unbalanced )"));
|
|---|
| 1900 |
|
|---|
| 1901 | addtok (END - d->nregexps);
|
|---|
| 1902 | addtok (CAT);
|
|---|
| 1903 |
|
|---|
| 1904 | if (d->nregexps)
|
|---|
| 1905 | addtok (OR);
|
|---|
| 1906 |
|
|---|
| 1907 | ++d->nregexps;
|
|---|
| 1908 | }
|
|---|
| 1909 |
|
|---|
| 1910 | /* Some primitives for operating on sets of positions. */
|
|---|
| 1911 |
|
|---|
| 1912 | /* Copy one set to another; the destination must be large enough. */
|
|---|
| 1913 | static void
|
|---|
| 1914 | copy (position_set const *src, position_set * dst)
|
|---|
| 1915 | {
|
|---|
| 1916 | REALLOC_IF_NECESSARY (dst->elems, dst->alloc, src->nelem);
|
|---|
| 1917 | memcpy (dst->elems, src->elems, sizeof (dst->elems[0]) * src->nelem);
|
|---|
| 1918 | dst->nelem = src->nelem;
|
|---|
| 1919 | }
|
|---|
| 1920 |
|
|---|
| 1921 | static void
|
|---|
| 1922 | alloc_position_set (position_set * s, size_t size)
|
|---|
| 1923 | {
|
|---|
| 1924 | MALLOC (s->elems, size);
|
|---|
| 1925 | s->alloc = size;
|
|---|
| 1926 | s->nelem = 0;
|
|---|
| 1927 | }
|
|---|
| 1928 |
|
|---|
| 1929 | /* Insert position P in set S. S is maintained in sorted order on
|
|---|
| 1930 | decreasing index. If there is already an entry in S with P.index
|
|---|
| 1931 | then merge (logically-OR) P's constraints into the one in S.
|
|---|
| 1932 | S->elems must point to an array large enough to hold the resulting set. */
|
|---|
| 1933 | static void
|
|---|
| 1934 | insert (position p, position_set * s)
|
|---|
| 1935 | {
|
|---|
| 1936 | size_t count = s->nelem;
|
|---|
| 1937 | size_t lo = 0, hi = count;
|
|---|
| 1938 | size_t i;
|
|---|
| 1939 | while (lo < hi)
|
|---|
| 1940 | {
|
|---|
| 1941 | size_t mid = (lo + hi) >> 1;
|
|---|
| 1942 | if (s->elems[mid].index > p.index)
|
|---|
| 1943 | lo = mid + 1;
|
|---|
| 1944 | else
|
|---|
| 1945 | hi = mid;
|
|---|
| 1946 | }
|
|---|
| 1947 |
|
|---|
| 1948 | if (lo < count && p.index == s->elems[lo].index)
|
|---|
| 1949 | {
|
|---|
| 1950 | s->elems[lo].constraint |= p.constraint;
|
|---|
| 1951 | return;
|
|---|
| 1952 | }
|
|---|
| 1953 |
|
|---|
| 1954 | REALLOC_IF_NECESSARY (s->elems, s->alloc, count + 1);
|
|---|
| 1955 | for (i = count; i > lo; i--)
|
|---|
| 1956 | s->elems[i] = s->elems[i - 1];
|
|---|
| 1957 | s->elems[lo] = p;
|
|---|
| 1958 | ++s->nelem;
|
|---|
| 1959 | }
|
|---|
| 1960 |
|
|---|
| 1961 | /* Merge two sets of positions into a third. The result is exactly as if
|
|---|
| 1962 | the positions of both sets were inserted into an initially empty set. */
|
|---|
| 1963 | static void
|
|---|
| 1964 | merge (position_set const *s1, position_set const *s2, position_set * m)
|
|---|
| 1965 | {
|
|---|
| 1966 | size_t i = 0, j = 0;
|
|---|
| 1967 |
|
|---|
| 1968 | REALLOC_IF_NECESSARY (m->elems, m->alloc, s1->nelem + s2->nelem);
|
|---|
| 1969 | m->nelem = 0;
|
|---|
| 1970 | while (i < s1->nelem && j < s2->nelem)
|
|---|
| 1971 | if (s1->elems[i].index > s2->elems[j].index)
|
|---|
| 1972 | m->elems[m->nelem++] = s1->elems[i++];
|
|---|
| 1973 | else if (s1->elems[i].index < s2->elems[j].index)
|
|---|
| 1974 | m->elems[m->nelem++] = s2->elems[j++];
|
|---|
| 1975 | else
|
|---|
| 1976 | {
|
|---|
| 1977 | m->elems[m->nelem] = s1->elems[i++];
|
|---|
| 1978 | m->elems[m->nelem++].constraint |= s2->elems[j++].constraint;
|
|---|
| 1979 | }
|
|---|
| 1980 | while (i < s1->nelem)
|
|---|
| 1981 | m->elems[m->nelem++] = s1->elems[i++];
|
|---|
| 1982 | while (j < s2->nelem)
|
|---|
| 1983 | m->elems[m->nelem++] = s2->elems[j++];
|
|---|
| 1984 | }
|
|---|
| 1985 |
|
|---|
| 1986 | /* Delete a position from a set. */
|
|---|
| 1987 | static void
|
|---|
| 1988 | delete (position p, position_set * s)
|
|---|
| 1989 | {
|
|---|
| 1990 | size_t i;
|
|---|
| 1991 |
|
|---|
| 1992 | for (i = 0; i < s->nelem; ++i)
|
|---|
| 1993 | if (p.index == s->elems[i].index)
|
|---|
| 1994 | break;
|
|---|
| 1995 | if (i < s->nelem)
|
|---|
| 1996 | for (--s->nelem; i < s->nelem; ++i)
|
|---|
| 1997 | s->elems[i] = s->elems[i + 1];
|
|---|
| 1998 | }
|
|---|
| 1999 |
|
|---|
| 2000 | /* Find the index of the state corresponding to the given position set with
|
|---|
| 2001 | the given preceding context, or create a new state if there is no such
|
|---|
| 2002 | state. Context tells whether we got here on a newline or letter. */
|
|---|
| 2003 | static state_num
|
|---|
| 2004 | state_index (struct dfa *d, position_set const *s, int context)
|
|---|
| 2005 | {
|
|---|
| 2006 | size_t hash = 0;
|
|---|
| 2007 | int constraint;
|
|---|
| 2008 | state_num i, j;
|
|---|
| 2009 |
|
|---|
| 2010 | for (i = 0; i < s->nelem; ++i)
|
|---|
| 2011 | hash ^= s->elems[i].index + s->elems[i].constraint;
|
|---|
| 2012 |
|
|---|
| 2013 | /* Try to find a state that exactly matches the proposed one. */
|
|---|
| 2014 | for (i = 0; i < d->sindex; ++i)
|
|---|
| 2015 | {
|
|---|
| 2016 | if (hash != d->states[i].hash || s->nelem != d->states[i].elems.nelem
|
|---|
| 2017 | || context != d->states[i].context)
|
|---|
| 2018 | continue;
|
|---|
| 2019 | for (j = 0; j < s->nelem; ++j)
|
|---|
| 2020 | if (s->elems[j].constraint
|
|---|
| 2021 | != d->states[i].elems.elems[j].constraint
|
|---|
| 2022 | || s->elems[j].index != d->states[i].elems.elems[j].index)
|
|---|
| 2023 | break;
|
|---|
| 2024 | if (j == s->nelem)
|
|---|
| 2025 | return i;
|
|---|
| 2026 | }
|
|---|
| 2027 |
|
|---|
| 2028 | /* We'll have to create a new state. */
|
|---|
| 2029 | REALLOC_IF_NECESSARY (d->states, d->salloc, d->sindex + 1);
|
|---|
| 2030 | d->states[i].hash = hash;
|
|---|
| 2031 | alloc_position_set (&d->states[i].elems, s->nelem);
|
|---|
| 2032 | copy (s, &d->states[i].elems);
|
|---|
| 2033 | d->states[i].context = context;
|
|---|
| 2034 | d->states[i].backref = 0;
|
|---|
| 2035 | d->states[i].constraint = 0;
|
|---|
| 2036 | d->states[i].first_end = 0;
|
|---|
| 2037 | if (MBS_SUPPORT)
|
|---|
| 2038 | {
|
|---|
| 2039 | d->states[i].mbps.nelem = 0;
|
|---|
| 2040 | d->states[i].mbps.elems = NULL;
|
|---|
| 2041 | }
|
|---|
| 2042 | for (j = 0; j < s->nelem; ++j)
|
|---|
| 2043 | if (d->tokens[s->elems[j].index] < 0)
|
|---|
| 2044 | {
|
|---|
| 2045 | constraint = s->elems[j].constraint;
|
|---|
| 2046 | if (SUCCEEDS_IN_CONTEXT (constraint, context, CTX_ANY))
|
|---|
| 2047 | d->states[i].constraint |= constraint;
|
|---|
| 2048 | if (!d->states[i].first_end)
|
|---|
| 2049 | d->states[i].first_end = d->tokens[s->elems[j].index];
|
|---|
| 2050 | }
|
|---|
| 2051 | else if (d->tokens[s->elems[j].index] == BACKREF)
|
|---|
| 2052 | {
|
|---|
| 2053 | d->states[i].constraint = NO_CONSTRAINT;
|
|---|
| 2054 | d->states[i].backref = 1;
|
|---|
| 2055 | }
|
|---|
| 2056 |
|
|---|
| 2057 | ++d->sindex;
|
|---|
| 2058 |
|
|---|
| 2059 | return i;
|
|---|
| 2060 | }
|
|---|
| 2061 |
|
|---|
| 2062 | /* Find the epsilon closure of a set of positions. If any position of the set
|
|---|
| 2063 | contains a symbol that matches the empty string in some context, replace
|
|---|
| 2064 | that position with the elements of its follow labeled with an appropriate
|
|---|
| 2065 | constraint. Repeat exhaustively until no funny positions are left.
|
|---|
| 2066 | S->elems must be large enough to hold the result. */
|
|---|
| 2067 | static void
|
|---|
| 2068 | epsclosure (position_set * s, struct dfa const *d)
|
|---|
| 2069 | {
|
|---|
| 2070 | size_t i, j;
|
|---|
| 2071 | char *visited; /* array of booleans, enough to use char, not int */
|
|---|
| 2072 | position p, old;
|
|---|
| 2073 |
|
|---|
| 2074 | CALLOC (visited, d->tindex);
|
|---|
| 2075 |
|
|---|
| 2076 | for (i = 0; i < s->nelem; ++i)
|
|---|
| 2077 | if (d->tokens[s->elems[i].index] >= NOTCHAR
|
|---|
| 2078 | && d->tokens[s->elems[i].index] != BACKREF
|
|---|
| 2079 | #if MBS_SUPPORT
|
|---|
| 2080 | && d->tokens[s->elems[i].index] != ANYCHAR
|
|---|
| 2081 | && d->tokens[s->elems[i].index] != MBCSET
|
|---|
| 2082 | #endif
|
|---|
| 2083 | && d->tokens[s->elems[i].index] < CSET)
|
|---|
| 2084 | {
|
|---|
| 2085 | old = s->elems[i];
|
|---|
| 2086 | p.constraint = old.constraint;
|
|---|
| 2087 | delete (s->elems[i], s);
|
|---|
| 2088 | if (visited[old.index])
|
|---|
| 2089 | {
|
|---|
| 2090 | --i;
|
|---|
| 2091 | continue;
|
|---|
| 2092 | }
|
|---|
| 2093 | visited[old.index] = 1;
|
|---|
| 2094 | switch (d->tokens[old.index])
|
|---|
| 2095 | {
|
|---|
| 2096 | case BEGLINE:
|
|---|
| 2097 | p.constraint &= BEGLINE_CONSTRAINT;
|
|---|
| 2098 | break;
|
|---|
| 2099 | case ENDLINE:
|
|---|
| 2100 | p.constraint &= ENDLINE_CONSTRAINT;
|
|---|
| 2101 | break;
|
|---|
| 2102 | case BEGWORD:
|
|---|
| 2103 | p.constraint &= BEGWORD_CONSTRAINT;
|
|---|
| 2104 | break;
|
|---|
| 2105 | case ENDWORD:
|
|---|
| 2106 | p.constraint &= ENDWORD_CONSTRAINT;
|
|---|
| 2107 | break;
|
|---|
| 2108 | case LIMWORD:
|
|---|
| 2109 | p.constraint &= LIMWORD_CONSTRAINT;
|
|---|
| 2110 | break;
|
|---|
| 2111 | case NOTLIMWORD:
|
|---|
| 2112 | p.constraint &= NOTLIMWORD_CONSTRAINT;
|
|---|
| 2113 | break;
|
|---|
| 2114 | default:
|
|---|
| 2115 | break;
|
|---|
| 2116 | }
|
|---|
| 2117 | for (j = 0; j < d->follows[old.index].nelem; ++j)
|
|---|
| 2118 | {
|
|---|
| 2119 | p.index = d->follows[old.index].elems[j].index;
|
|---|
| 2120 | insert (p, s);
|
|---|
| 2121 | }
|
|---|
| 2122 | /* Force rescan to start at the beginning. */
|
|---|
| 2123 | i = -1;
|
|---|
| 2124 | }
|
|---|
| 2125 |
|
|---|
| 2126 | free (visited);
|
|---|
| 2127 | }
|
|---|
| 2128 |
|
|---|
| 2129 | /* Returns the set of contexts for which there is at least one
|
|---|
| 2130 | character included in C. */
|
|---|
| 2131 |
|
|---|
| 2132 | static int
|
|---|
| 2133 | charclass_context (charclass c)
|
|---|
| 2134 | {
|
|---|
| 2135 | int context = 0;
|
|---|
| 2136 | unsigned int j;
|
|---|
| 2137 |
|
|---|
| 2138 | if (tstbit (eolbyte, c))
|
|---|
| 2139 | context |= CTX_NEWLINE;
|
|---|
| 2140 |
|
|---|
| 2141 | for (j = 0; j < CHARCLASS_INTS; ++j)
|
|---|
| 2142 | {
|
|---|
| 2143 | if (c[j] & letters[j])
|
|---|
| 2144 | context |= CTX_LETTER;
|
|---|
| 2145 | if (c[j] & ~(letters[j] | newline[j]))
|
|---|
| 2146 | context |= CTX_NONE;
|
|---|
| 2147 | }
|
|---|
| 2148 |
|
|---|
| 2149 | return context;
|
|---|
| 2150 | }
|
|---|
| 2151 |
|
|---|
| 2152 | /* Returns the contexts on which the position set S depends. Each context
|
|---|
| 2153 | in the set of returned contexts (let's call it SC) may have a different
|
|---|
| 2154 | follow set than other contexts in SC, and also different from the
|
|---|
| 2155 | follow set of the complement set (sc ^ CTX_ANY). However, all contexts
|
|---|
| 2156 | in the complement set will have the same follow set. */
|
|---|
| 2157 |
|
|---|
| 2158 | static int _GL_ATTRIBUTE_PURE
|
|---|
| 2159 | state_separate_contexts (position_set const *s)
|
|---|
| 2160 | {
|
|---|
| 2161 | int separate_contexts = 0;
|
|---|
| 2162 | size_t j;
|
|---|
| 2163 |
|
|---|
| 2164 | for (j = 0; j < s->nelem; ++j)
|
|---|
| 2165 | {
|
|---|
| 2166 | if (PREV_NEWLINE_DEPENDENT (s->elems[j].constraint))
|
|---|
| 2167 | separate_contexts |= CTX_NEWLINE;
|
|---|
| 2168 | if (PREV_LETTER_DEPENDENT (s->elems[j].constraint))
|
|---|
| 2169 | separate_contexts |= CTX_LETTER;
|
|---|
| 2170 | }
|
|---|
| 2171 |
|
|---|
| 2172 | return separate_contexts;
|
|---|
| 2173 | }
|
|---|
| 2174 |
|
|---|
| 2175 |
|
|---|
| 2176 | /* Perform bottom-up analysis on the parse tree, computing various functions.
|
|---|
| 2177 | Note that at this point, we're pretending constructs like \< are real
|
|---|
| 2178 | characters rather than constraints on what can follow them.
|
|---|
| 2179 |
|
|---|
| 2180 | Nullable: A node is nullable if it is at the root of a regexp that can
|
|---|
| 2181 | match the empty string.
|
|---|
| 2182 | * EMPTY leaves are nullable.
|
|---|
| 2183 | * No other leaf is nullable.
|
|---|
| 2184 | * A QMARK or STAR node is nullable.
|
|---|
| 2185 | * A PLUS node is nullable if its argument is nullable.
|
|---|
| 2186 | * A CAT node is nullable if both its arguments are nullable.
|
|---|
| 2187 | * An OR node is nullable if either argument is nullable.
|
|---|
| 2188 |
|
|---|
| 2189 | Firstpos: The firstpos of a node is the set of positions (nonempty leaves)
|
|---|
| 2190 | that could correspond to the first character of a string matching the
|
|---|
| 2191 | regexp rooted at the given node.
|
|---|
| 2192 | * EMPTY leaves have empty firstpos.
|
|---|
| 2193 | * The firstpos of a nonempty leaf is that leaf itself.
|
|---|
| 2194 | * The firstpos of a QMARK, STAR, or PLUS node is the firstpos of its
|
|---|
| 2195 | argument.
|
|---|
| 2196 | * The firstpos of a CAT node is the firstpos of the left argument, union
|
|---|
| 2197 | the firstpos of the right if the left argument is nullable.
|
|---|
| 2198 | * The firstpos of an OR node is the union of firstpos of each argument.
|
|---|
| 2199 |
|
|---|
| 2200 | Lastpos: The lastpos of a node is the set of positions that could
|
|---|
| 2201 | correspond to the last character of a string matching the regexp at
|
|---|
| 2202 | the given node.
|
|---|
| 2203 | * EMPTY leaves have empty lastpos.
|
|---|
| 2204 | * The lastpos of a nonempty leaf is that leaf itself.
|
|---|
| 2205 | * The lastpos of a QMARK, STAR, or PLUS node is the lastpos of its
|
|---|
| 2206 | argument.
|
|---|
| 2207 | * The lastpos of a CAT node is the lastpos of its right argument, union
|
|---|
| 2208 | the lastpos of the left if the right argument is nullable.
|
|---|
| 2209 | * The lastpos of an OR node is the union of the lastpos of each argument.
|
|---|
| 2210 |
|
|---|
| 2211 | Follow: The follow of a position is the set of positions that could
|
|---|
| 2212 | correspond to the character following a character matching the node in
|
|---|
| 2213 | a string matching the regexp. At this point we consider special symbols
|
|---|
| 2214 | that match the empty string in some context to be just normal characters.
|
|---|
| 2215 | Later, if we find that a special symbol is in a follow set, we will
|
|---|
| 2216 | replace it with the elements of its follow, labeled with an appropriate
|
|---|
| 2217 | constraint.
|
|---|
| 2218 | * Every node in the firstpos of the argument of a STAR or PLUS node is in
|
|---|
| 2219 | the follow of every node in the lastpos.
|
|---|
| 2220 | * Every node in the firstpos of the second argument of a CAT node is in
|
|---|
| 2221 | the follow of every node in the lastpos of the first argument.
|
|---|
| 2222 |
|
|---|
| 2223 | Because of the postfix representation of the parse tree, the depth-first
|
|---|
| 2224 | analysis is conveniently done by a linear scan with the aid of a stack.
|
|---|
| 2225 | Sets are stored as arrays of the elements, obeying a stack-like allocation
|
|---|
| 2226 | scheme; the number of elements in each set deeper in the stack can be
|
|---|
| 2227 | used to determine the address of a particular set's array. */
|
|---|
| 2228 | void
|
|---|
| 2229 | dfaanalyze (struct dfa *d, int searchflag)
|
|---|
| 2230 | {
|
|---|
| 2231 | int *nullable; /* Nullable stack. */
|
|---|
| 2232 | size_t *nfirstpos; /* Element count stack for firstpos sets. */
|
|---|
| 2233 | position *firstpos; /* Array where firstpos elements are stored. */
|
|---|
| 2234 | size_t *nlastpos; /* Element count stack for lastpos sets. */
|
|---|
| 2235 | position *lastpos; /* Array where lastpos elements are stored. */
|
|---|
| 2236 | position_set tmp; /* Temporary set for merging sets. */
|
|---|
| 2237 | position_set merged; /* Result of merging sets. */
|
|---|
| 2238 | int separate_contexts; /* Context wanted by some position. */
|
|---|
| 2239 | int *o_nullable;
|
|---|
| 2240 | size_t *o_nfirst, *o_nlast;
|
|---|
| 2241 | position *o_firstpos, *o_lastpos;
|
|---|
| 2242 | size_t i, j;
|
|---|
| 2243 | position *pos;
|
|---|
| 2244 |
|
|---|
| 2245 | #ifdef DEBUG
|
|---|
| 2246 | fprintf (stderr, "dfaanalyze:\n");
|
|---|
| 2247 | for (i = 0; i < d->tindex; ++i)
|
|---|
| 2248 | {
|
|---|
| 2249 | fprintf (stderr, " %zd:", i);
|
|---|
| 2250 | prtok (d->tokens[i]);
|
|---|
| 2251 | }
|
|---|
| 2252 | putc ('\n', stderr);
|
|---|
| 2253 | #endif
|
|---|
| 2254 |
|
|---|
| 2255 | d->searchflag = searchflag;
|
|---|
| 2256 |
|
|---|
| 2257 | MALLOC (nullable, d->depth);
|
|---|
| 2258 | o_nullable = nullable;
|
|---|
| 2259 | MALLOC (nfirstpos, d->depth);
|
|---|
| 2260 | o_nfirst = nfirstpos;
|
|---|
| 2261 | MALLOC (firstpos, d->nleaves);
|
|---|
| 2262 | o_firstpos = firstpos, firstpos += d->nleaves;
|
|---|
| 2263 | MALLOC (nlastpos, d->depth);
|
|---|
| 2264 | o_nlast = nlastpos;
|
|---|
| 2265 | MALLOC (lastpos, d->nleaves);
|
|---|
| 2266 | o_lastpos = lastpos, lastpos += d->nleaves;
|
|---|
| 2267 | alloc_position_set (&merged, d->nleaves);
|
|---|
| 2268 |
|
|---|
| 2269 | CALLOC (d->follows, d->tindex);
|
|---|
| 2270 |
|
|---|
| 2271 | for (i = 0; i < d->tindex; ++i)
|
|---|
| 2272 | {
|
|---|
| 2273 | switch (d->tokens[i])
|
|---|
| 2274 | {
|
|---|
| 2275 | case EMPTY:
|
|---|
| 2276 | /* The empty set is nullable. */
|
|---|
| 2277 | *nullable++ = 1;
|
|---|
| 2278 |
|
|---|
| 2279 | /* The firstpos and lastpos of the empty leaf are both empty. */
|
|---|
| 2280 | *nfirstpos++ = *nlastpos++ = 0;
|
|---|
| 2281 | break;
|
|---|
| 2282 |
|
|---|
| 2283 | case STAR:
|
|---|
| 2284 | case PLUS:
|
|---|
| 2285 | /* Every element in the firstpos of the argument is in the follow
|
|---|
| 2286 | of every element in the lastpos. */
|
|---|
| 2287 | tmp.nelem = nfirstpos[-1];
|
|---|
| 2288 | tmp.elems = firstpos;
|
|---|
| 2289 | pos = lastpos;
|
|---|
| 2290 | for (j = 0; j < nlastpos[-1]; ++j)
|
|---|
| 2291 | {
|
|---|
| 2292 | merge (&tmp, &d->follows[pos[j].index], &merged);
|
|---|
| 2293 | copy (&merged, &d->follows[pos[j].index]);
|
|---|
| 2294 | }
|
|---|
| 2295 |
|
|---|
| 2296 | case QMARK:
|
|---|
| 2297 | /* A QMARK or STAR node is automatically nullable. */
|
|---|
| 2298 | if (d->tokens[i] != PLUS)
|
|---|
| 2299 | nullable[-1] = 1;
|
|---|
| 2300 | break;
|
|---|
| 2301 |
|
|---|
| 2302 | case CAT:
|
|---|
| 2303 | /* Every element in the firstpos of the second argument is in the
|
|---|
| 2304 | follow of every element in the lastpos of the first argument. */
|
|---|
| 2305 | tmp.nelem = nfirstpos[-1];
|
|---|
| 2306 | tmp.elems = firstpos;
|
|---|
| 2307 | pos = lastpos + nlastpos[-1];
|
|---|
| 2308 | for (j = 0; j < nlastpos[-2]; ++j)
|
|---|
| 2309 | {
|
|---|
| 2310 | merge (&tmp, &d->follows[pos[j].index], &merged);
|
|---|
| 2311 | copy (&merged, &d->follows[pos[j].index]);
|
|---|
| 2312 | }
|
|---|
| 2313 |
|
|---|
| 2314 | /* The firstpos of a CAT node is the firstpos of the first argument,
|
|---|
| 2315 | union that of the second argument if the first is nullable. */
|
|---|
| 2316 | if (nullable[-2])
|
|---|
| 2317 | nfirstpos[-2] += nfirstpos[-1];
|
|---|
| 2318 | else
|
|---|
| 2319 | firstpos += nfirstpos[-1];
|
|---|
| 2320 | --nfirstpos;
|
|---|
| 2321 |
|
|---|
| 2322 | /* The lastpos of a CAT node is the lastpos of the second argument,
|
|---|
| 2323 | union that of the first argument if the second is nullable. */
|
|---|
| 2324 | if (nullable[-1])
|
|---|
| 2325 | nlastpos[-2] += nlastpos[-1];
|
|---|
| 2326 | else
|
|---|
| 2327 | {
|
|---|
| 2328 | pos = lastpos + nlastpos[-2];
|
|---|
| 2329 | for (j = nlastpos[-1]; j-- > 0;)
|
|---|
| 2330 | pos[j] = lastpos[j];
|
|---|
| 2331 | lastpos += nlastpos[-2];
|
|---|
| 2332 | nlastpos[-2] = nlastpos[-1];
|
|---|
| 2333 | }
|
|---|
| 2334 | --nlastpos;
|
|---|
| 2335 |
|
|---|
| 2336 | /* A CAT node is nullable if both arguments are nullable. */
|
|---|
| 2337 | nullable[-2] = nullable[-1] && nullable[-2];
|
|---|
| 2338 | --nullable;
|
|---|
| 2339 | break;
|
|---|
| 2340 |
|
|---|
| 2341 | case OR:
|
|---|
| 2342 | /* The firstpos is the union of the firstpos of each argument. */
|
|---|
| 2343 | nfirstpos[-2] += nfirstpos[-1];
|
|---|
| 2344 | --nfirstpos;
|
|---|
| 2345 |
|
|---|
| 2346 | /* The lastpos is the union of the lastpos of each argument. */
|
|---|
| 2347 | nlastpos[-2] += nlastpos[-1];
|
|---|
| 2348 | --nlastpos;
|
|---|
| 2349 |
|
|---|
| 2350 | /* An OR node is nullable if either argument is nullable. */
|
|---|
| 2351 | nullable[-2] = nullable[-1] || nullable[-2];
|
|---|
| 2352 | --nullable;
|
|---|
| 2353 | break;
|
|---|
| 2354 |
|
|---|
| 2355 | default:
|
|---|
| 2356 | /* Anything else is a nonempty position. (Note that special
|
|---|
| 2357 | constructs like \< are treated as nonempty strings here;
|
|---|
| 2358 | an "epsilon closure" effectively makes them nullable later.
|
|---|
| 2359 | Backreferences have to get a real position so we can detect
|
|---|
| 2360 | transitions on them later. But they are nullable. */
|
|---|
| 2361 | *nullable++ = d->tokens[i] == BACKREF;
|
|---|
| 2362 |
|
|---|
| 2363 | /* This position is in its own firstpos and lastpos. */
|
|---|
| 2364 | *nfirstpos++ = *nlastpos++ = 1;
|
|---|
| 2365 | --firstpos, --lastpos;
|
|---|
| 2366 | firstpos->index = lastpos->index = i;
|
|---|
| 2367 | firstpos->constraint = lastpos->constraint = NO_CONSTRAINT;
|
|---|
| 2368 |
|
|---|
| 2369 | /* Allocate the follow set for this position. */
|
|---|
| 2370 | alloc_position_set (&d->follows[i], 1);
|
|---|
| 2371 | break;
|
|---|
| 2372 | }
|
|---|
| 2373 | #ifdef DEBUG
|
|---|
| 2374 | /* ... balance the above nonsyntactic #ifdef goo... */
|
|---|
| 2375 | fprintf (stderr, "node %zd:", i);
|
|---|
| 2376 | prtok (d->tokens[i]);
|
|---|
| 2377 | putc ('\n', stderr);
|
|---|
| 2378 | fprintf (stderr, nullable[-1] ? " nullable: yes\n" : " nullable: no\n");
|
|---|
| 2379 | fprintf (stderr, " firstpos:");
|
|---|
| 2380 | for (j = nfirstpos[-1]; j-- > 0;)
|
|---|
| 2381 | {
|
|---|
| 2382 | fprintf (stderr, " %zd:", firstpos[j].index);
|
|---|
| 2383 | prtok (d->tokens[firstpos[j].index]);
|
|---|
| 2384 | }
|
|---|
| 2385 | fprintf (stderr, "\n lastpos:");
|
|---|
| 2386 | for (j = nlastpos[-1]; j-- > 0;)
|
|---|
| 2387 | {
|
|---|
| 2388 | fprintf (stderr, " %zd:", lastpos[j].index);
|
|---|
| 2389 | prtok (d->tokens[lastpos[j].index]);
|
|---|
| 2390 | }
|
|---|
| 2391 | putc ('\n', stderr);
|
|---|
| 2392 | #endif
|
|---|
| 2393 | }
|
|---|
| 2394 |
|
|---|
| 2395 | /* For each follow set that is the follow set of a real position, replace
|
|---|
| 2396 | it with its epsilon closure. */
|
|---|
| 2397 | for (i = 0; i < d->tindex; ++i)
|
|---|
| 2398 | if (d->tokens[i] < NOTCHAR || d->tokens[i] == BACKREF
|
|---|
| 2399 | #if MBS_SUPPORT
|
|---|
| 2400 | || d->tokens[i] == ANYCHAR || d->tokens[i] == MBCSET
|
|---|
| 2401 | #endif
|
|---|
| 2402 | || d->tokens[i] >= CSET)
|
|---|
| 2403 | {
|
|---|
| 2404 | #ifdef DEBUG
|
|---|
| 2405 | fprintf (stderr, "follows(%zd:", i);
|
|---|
| 2406 | prtok (d->tokens[i]);
|
|---|
| 2407 | fprintf (stderr, "):");
|
|---|
| 2408 | for (j = d->follows[i].nelem; j-- > 0;)
|
|---|
| 2409 | {
|
|---|
| 2410 | fprintf (stderr, " %zd:", d->follows[i].elems[j].index);
|
|---|
| 2411 | prtok (d->tokens[d->follows[i].elems[j].index]);
|
|---|
| 2412 | }
|
|---|
| 2413 | putc ('\n', stderr);
|
|---|
| 2414 | #endif
|
|---|
| 2415 | copy (&d->follows[i], &merged);
|
|---|
| 2416 | epsclosure (&merged, d);
|
|---|
| 2417 | copy (&merged, &d->follows[i]);
|
|---|
| 2418 | }
|
|---|
| 2419 |
|
|---|
| 2420 | /* Get the epsilon closure of the firstpos of the regexp. The result will
|
|---|
| 2421 | be the set of positions of state 0. */
|
|---|
| 2422 | merged.nelem = 0;
|
|---|
| 2423 | for (i = 0; i < nfirstpos[-1]; ++i)
|
|---|
| 2424 | insert (firstpos[i], &merged);
|
|---|
| 2425 | epsclosure (&merged, d);
|
|---|
| 2426 |
|
|---|
| 2427 | /* Build the initial state. */
|
|---|
| 2428 | d->salloc = 1;
|
|---|
| 2429 | d->sindex = 0;
|
|---|
| 2430 | MALLOC (d->states, d->salloc);
|
|---|
| 2431 |
|
|---|
| 2432 | separate_contexts = state_separate_contexts (&merged);
|
|---|
| 2433 | state_index (d, &merged,
|
|---|
| 2434 | (separate_contexts & CTX_NEWLINE
|
|---|
| 2435 | ? CTX_NEWLINE : separate_contexts ^ CTX_ANY));
|
|---|
| 2436 |
|
|---|
| 2437 | free (o_nullable);
|
|---|
| 2438 | free (o_nfirst);
|
|---|
| 2439 | free (o_firstpos);
|
|---|
| 2440 | free (o_nlast);
|
|---|
| 2441 | free (o_lastpos);
|
|---|
| 2442 | free (merged.elems);
|
|---|
| 2443 | }
|
|---|
| 2444 |
|
|---|
| 2445 |
|
|---|
| 2446 | /* Find, for each character, the transition out of state s of d, and store
|
|---|
| 2447 | it in the appropriate slot of trans.
|
|---|
| 2448 |
|
|---|
| 2449 | We divide the positions of s into groups (positions can appear in more
|
|---|
| 2450 | than one group). Each group is labeled with a set of characters that
|
|---|
| 2451 | every position in the group matches (taking into account, if necessary,
|
|---|
| 2452 | preceding context information of s). For each group, find the union
|
|---|
| 2453 | of the its elements' follows. This set is the set of positions of the
|
|---|
| 2454 | new state. For each character in the group's label, set the transition
|
|---|
| 2455 | on this character to be to a state corresponding to the set's positions,
|
|---|
| 2456 | and its associated backward context information, if necessary.
|
|---|
| 2457 |
|
|---|
| 2458 | If we are building a searching matcher, we include the positions of state
|
|---|
| 2459 | 0 in every state.
|
|---|
| 2460 |
|
|---|
| 2461 | The collection of groups is constructed by building an equivalence-class
|
|---|
| 2462 | partition of the positions of s.
|
|---|
| 2463 |
|
|---|
| 2464 | For each position, find the set of characters C that it matches. Eliminate
|
|---|
| 2465 | any characters from C that fail on grounds of backward context.
|
|---|
| 2466 |
|
|---|
| 2467 | Search through the groups, looking for a group whose label L has nonempty
|
|---|
| 2468 | intersection with C. If L - C is nonempty, create a new group labeled
|
|---|
| 2469 | L - C and having the same positions as the current group, and set L to
|
|---|
| 2470 | the intersection of L and C. Insert the position in this group, set
|
|---|
| 2471 | C = C - L, and resume scanning.
|
|---|
| 2472 |
|
|---|
| 2473 | If after comparing with every group there are characters remaining in C,
|
|---|
| 2474 | create a new group labeled with the characters of C and insert this
|
|---|
| 2475 | position in that group. */
|
|---|
| 2476 | void
|
|---|
| 2477 | dfastate (state_num s, struct dfa *d, state_num trans[])
|
|---|
| 2478 | {
|
|---|
| 2479 | leaf_set *grps; /* As many as will ever be needed. */
|
|---|
| 2480 | charclass *labels; /* Labels corresponding to the groups. */
|
|---|
| 2481 | size_t ngrps = 0; /* Number of groups actually used. */
|
|---|
| 2482 | position pos; /* Current position being considered. */
|
|---|
| 2483 | charclass matches; /* Set of matching characters. */
|
|---|
| 2484 | int matchesf; /* True if matches is nonempty. */
|
|---|
| 2485 | charclass intersect; /* Intersection with some label set. */
|
|---|
| 2486 | int intersectf; /* True if intersect is nonempty. */
|
|---|
| 2487 | charclass leftovers; /* Stuff in the label that didn't match. */
|
|---|
| 2488 | int leftoversf; /* True if leftovers is nonempty. */
|
|---|
| 2489 | position_set follows; /* Union of the follows of some group. */
|
|---|
| 2490 | position_set tmp; /* Temporary space for merging sets. */
|
|---|
| 2491 | int possible_contexts; /* Contexts that this group can match. */
|
|---|
| 2492 | int separate_contexts; /* Context that new state wants to know. */
|
|---|
| 2493 | state_num state; /* New state. */
|
|---|
| 2494 | state_num state_newline; /* New state on a newline transition. */
|
|---|
| 2495 | state_num state_letter; /* New state on a letter transition. */
|
|---|
| 2496 | int next_isnt_1st_byte = 0; /* Flag if we can't add state0. */
|
|---|
| 2497 | size_t i, j, k;
|
|---|
| 2498 |
|
|---|
| 2499 | MALLOC (grps, NOTCHAR);
|
|---|
| 2500 | MALLOC (labels, NOTCHAR);
|
|---|
| 2501 |
|
|---|
| 2502 | zeroset (matches);
|
|---|
| 2503 |
|
|---|
| 2504 | for (i = 0; i < d->states[s].elems.nelem; ++i)
|
|---|
| 2505 | {
|
|---|
| 2506 | pos = d->states[s].elems.elems[i];
|
|---|
| 2507 | if (d->tokens[pos.index] >= 0 && d->tokens[pos.index] < NOTCHAR)
|
|---|
| 2508 | setbit (d->tokens[pos.index], matches);
|
|---|
| 2509 | else if (d->tokens[pos.index] >= CSET)
|
|---|
| 2510 | copyset (d->charclasses[d->tokens[pos.index] - CSET], matches);
|
|---|
| 2511 | else if (MBS_SUPPORT
|
|---|
| 2512 | && (d->tokens[pos.index] == ANYCHAR
|
|---|
| 2513 | || d->tokens[pos.index] == MBCSET))
|
|---|
| 2514 | /* MB_CUR_MAX > 1 */
|
|---|
| 2515 | {
|
|---|
| 2516 | /* ANYCHAR and MBCSET must match with a single character, so we
|
|---|
| 2517 | must put it to d->states[s].mbps, which contains the positions
|
|---|
| 2518 | which can match with a single character not a byte. */
|
|---|
| 2519 | if (d->states[s].mbps.nelem == 0)
|
|---|
| 2520 | alloc_position_set (&d->states[s].mbps, 1);
|
|---|
| 2521 | insert (pos, &(d->states[s].mbps));
|
|---|
| 2522 | continue;
|
|---|
| 2523 | }
|
|---|
| 2524 | else
|
|---|
| 2525 | continue;
|
|---|
| 2526 |
|
|---|
| 2527 | /* Some characters may need to be eliminated from matches because
|
|---|
| 2528 | they fail in the current context. */
|
|---|
| 2529 | if (pos.constraint != NO_CONSTRAINT)
|
|---|
| 2530 | {
|
|---|
| 2531 | if (!SUCCEEDS_IN_CONTEXT (pos.constraint,
|
|---|
| 2532 | d->states[s].context, CTX_NEWLINE))
|
|---|
| 2533 | for (j = 0; j < CHARCLASS_INTS; ++j)
|
|---|
| 2534 | matches[j] &= ~newline[j];
|
|---|
| 2535 | if (!SUCCEEDS_IN_CONTEXT (pos.constraint,
|
|---|
| 2536 | d->states[s].context, CTX_LETTER))
|
|---|
| 2537 | for (j = 0; j < CHARCLASS_INTS; ++j)
|
|---|
| 2538 | matches[j] &= ~letters[j];
|
|---|
| 2539 | if (!SUCCEEDS_IN_CONTEXT (pos.constraint,
|
|---|
| 2540 | d->states[s].context, CTX_NONE))
|
|---|
| 2541 | for (j = 0; j < CHARCLASS_INTS; ++j)
|
|---|
| 2542 | matches[j] &= letters[j] | newline[j];
|
|---|
| 2543 |
|
|---|
| 2544 | /* If there are no characters left, there's no point in going on. */
|
|---|
| 2545 | for (j = 0; j < CHARCLASS_INTS && !matches[j]; ++j)
|
|---|
| 2546 | continue;
|
|---|
| 2547 | if (j == CHARCLASS_INTS)
|
|---|
| 2548 | continue;
|
|---|
| 2549 | }
|
|---|
| 2550 |
|
|---|
| 2551 | for (j = 0; j < ngrps; ++j)
|
|---|
| 2552 | {
|
|---|
| 2553 | /* If matches contains a single character only, and the current
|
|---|
| 2554 | group's label doesn't contain that character, go on to the
|
|---|
| 2555 | next group. */
|
|---|
| 2556 | if (d->tokens[pos.index] >= 0 && d->tokens[pos.index] < NOTCHAR
|
|---|
| 2557 | && !tstbit (d->tokens[pos.index], labels[j]))
|
|---|
| 2558 | continue;
|
|---|
| 2559 |
|
|---|
| 2560 | /* Check if this group's label has a nonempty intersection with
|
|---|
| 2561 | matches. */
|
|---|
| 2562 | intersectf = 0;
|
|---|
| 2563 | for (k = 0; k < CHARCLASS_INTS; ++k)
|
|---|
| 2564 | (intersect[k] = matches[k] & labels[j][k]) ? (intersectf = 1) : 0;
|
|---|
| 2565 | if (!intersectf)
|
|---|
| 2566 | continue;
|
|---|
| 2567 |
|
|---|
| 2568 | /* It does; now find the set differences both ways. */
|
|---|
| 2569 | leftoversf = matchesf = 0;
|
|---|
| 2570 | for (k = 0; k < CHARCLASS_INTS; ++k)
|
|---|
| 2571 | {
|
|---|
| 2572 | /* Even an optimizing compiler can't know this for sure. */
|
|---|
| 2573 | int match = matches[k], label = labels[j][k];
|
|---|
| 2574 |
|
|---|
| 2575 | (leftovers[k] = ~match & label) ? (leftoversf = 1) : 0;
|
|---|
| 2576 | (matches[k] = match & ~label) ? (matchesf = 1) : 0;
|
|---|
| 2577 | }
|
|---|
| 2578 |
|
|---|
| 2579 | /* If there were leftovers, create a new group labeled with them. */
|
|---|
| 2580 | if (leftoversf)
|
|---|
| 2581 | {
|
|---|
| 2582 | copyset (leftovers, labels[ngrps]);
|
|---|
| 2583 | copyset (intersect, labels[j]);
|
|---|
| 2584 | MALLOC (grps[ngrps].elems, d->nleaves);
|
|---|
| 2585 | memcpy (grps[ngrps].elems, grps[j].elems,
|
|---|
| 2586 | sizeof (grps[j].elems[0]) * grps[j].nelem);
|
|---|
| 2587 | grps[ngrps].nelem = grps[j].nelem;
|
|---|
| 2588 | ++ngrps;
|
|---|
| 2589 | }
|
|---|
| 2590 |
|
|---|
| 2591 | /* Put the position in the current group. The constraint is
|
|---|
| 2592 | irrelevant here. */
|
|---|
| 2593 | grps[j].elems[grps[j].nelem++] = pos.index;
|
|---|
| 2594 |
|
|---|
| 2595 | /* If every character matching the current position has been
|
|---|
| 2596 | accounted for, we're done. */
|
|---|
| 2597 | if (!matchesf)
|
|---|
| 2598 | break;
|
|---|
| 2599 | }
|
|---|
| 2600 |
|
|---|
| 2601 | /* If we've passed the last group, and there are still characters
|
|---|
| 2602 | unaccounted for, then we'll have to create a new group. */
|
|---|
| 2603 | if (j == ngrps)
|
|---|
| 2604 | {
|
|---|
| 2605 | copyset (matches, labels[ngrps]);
|
|---|
| 2606 | zeroset (matches);
|
|---|
| 2607 | MALLOC (grps[ngrps].elems, d->nleaves);
|
|---|
| 2608 | grps[ngrps].nelem = 1;
|
|---|
| 2609 | grps[ngrps].elems[0] = pos.index;
|
|---|
| 2610 | ++ngrps;
|
|---|
| 2611 | }
|
|---|
| 2612 | }
|
|---|
| 2613 |
|
|---|
| 2614 | alloc_position_set (&follows, d->nleaves);
|
|---|
| 2615 | alloc_position_set (&tmp, d->nleaves);
|
|---|
| 2616 |
|
|---|
| 2617 | /* If we are a searching matcher, the default transition is to a state
|
|---|
| 2618 | containing the positions of state 0, otherwise the default transition
|
|---|
| 2619 | is to fail miserably. */
|
|---|
| 2620 | if (d->searchflag)
|
|---|
| 2621 | {
|
|---|
| 2622 | /* Find the state(s) corresponding to the positions of state 0. */
|
|---|
| 2623 | copy (&d->states[0].elems, &follows);
|
|---|
| 2624 | separate_contexts = state_separate_contexts (&follows);
|
|---|
| 2625 | state = state_index (d, &follows, separate_contexts ^ CTX_ANY);
|
|---|
| 2626 | if (separate_contexts & CTX_NEWLINE)
|
|---|
| 2627 | state_newline = state_index (d, &follows, CTX_NEWLINE);
|
|---|
| 2628 | else
|
|---|
| 2629 | state_newline = state;
|
|---|
| 2630 | if (separate_contexts & CTX_LETTER)
|
|---|
| 2631 | state_letter = state_index (d, &follows, CTX_LETTER);
|
|---|
| 2632 | else
|
|---|
| 2633 | state_letter = state;
|
|---|
| 2634 |
|
|---|
| 2635 | for (i = 0; i < NOTCHAR; ++i)
|
|---|
| 2636 | trans[i] = (IS_WORD_CONSTITUENT (i)) ? state_letter : state;
|
|---|
| 2637 | trans[eolbyte] = state_newline;
|
|---|
| 2638 | }
|
|---|
| 2639 | else
|
|---|
| 2640 | for (i = 0; i < NOTCHAR; ++i)
|
|---|
| 2641 | trans[i] = -1;
|
|---|
| 2642 |
|
|---|
| 2643 | for (i = 0; i < ngrps; ++i)
|
|---|
| 2644 | {
|
|---|
| 2645 | follows.nelem = 0;
|
|---|
| 2646 |
|
|---|
| 2647 | /* Find the union of the follows of the positions of the group.
|
|---|
| 2648 | This is a hideously inefficient loop. Fix it someday. */
|
|---|
| 2649 | for (j = 0; j < grps[i].nelem; ++j)
|
|---|
| 2650 | for (k = 0; k < d->follows[grps[i].elems[j]].nelem; ++k)
|
|---|
| 2651 | insert (d->follows[grps[i].elems[j]].elems[k], &follows);
|
|---|
| 2652 |
|
|---|
| 2653 | if (d->mb_cur_max > 1)
|
|---|
| 2654 | {
|
|---|
| 2655 | /* If a token in follows.elems is not 1st byte of a multibyte
|
|---|
| 2656 | character, or the states of follows must accept the bytes
|
|---|
| 2657 | which are not 1st byte of the multibyte character.
|
|---|
| 2658 | Then, if a state of follows encounter a byte, it must not be
|
|---|
| 2659 | a 1st byte of a multibyte character nor single byte character.
|
|---|
| 2660 | We cansel to add state[0].follows to next state, because
|
|---|
| 2661 | state[0] must accept 1st-byte
|
|---|
| 2662 |
|
|---|
| 2663 | For example, we assume <sb a> is a certain single byte
|
|---|
| 2664 | character, <mb A> is a certain multibyte character, and the
|
|---|
| 2665 | codepoint of <sb a> equals the 2nd byte of the codepoint of
|
|---|
| 2666 | <mb A>.
|
|---|
| 2667 | When state[0] accepts <sb a>, state[i] transit to state[i+1]
|
|---|
| 2668 | by accepting accepts 1st byte of <mb A>, and state[i+1]
|
|---|
| 2669 | accepts 2nd byte of <mb A>, if state[i+1] encounter the
|
|---|
| 2670 | codepoint of <sb a>, it must not be <sb a> but 2nd byte of
|
|---|
| 2671 | <mb A>, so we cannot add state[0]. */
|
|---|
| 2672 |
|
|---|
| 2673 | next_isnt_1st_byte = 0;
|
|---|
| 2674 | for (j = 0; j < follows.nelem; ++j)
|
|---|
| 2675 | {
|
|---|
| 2676 | if (!(d->multibyte_prop[follows.elems[j].index] & 1))
|
|---|
| 2677 | {
|
|---|
| 2678 | next_isnt_1st_byte = 1;
|
|---|
| 2679 | break;
|
|---|
| 2680 | }
|
|---|
| 2681 | }
|
|---|
| 2682 | }
|
|---|
| 2683 |
|
|---|
| 2684 | /* If we are building a searching matcher, throw in the positions
|
|---|
| 2685 | of state 0 as well. */
|
|---|
| 2686 | if (d->searchflag
|
|---|
| 2687 | && (!MBS_SUPPORT || (d->mb_cur_max == 1 || !next_isnt_1st_byte)))
|
|---|
| 2688 | for (j = 0; j < d->states[0].elems.nelem; ++j)
|
|---|
| 2689 | insert (d->states[0].elems.elems[j], &follows);
|
|---|
| 2690 |
|
|---|
| 2691 | /* Find out if the new state will want any context information. */
|
|---|
| 2692 | possible_contexts = charclass_context (labels[i]);
|
|---|
| 2693 | separate_contexts = state_separate_contexts (&follows);
|
|---|
| 2694 |
|
|---|
| 2695 | /* Find the state(s) corresponding to the union of the follows. */
|
|---|
| 2696 | if ((separate_contexts & possible_contexts) != possible_contexts)
|
|---|
| 2697 | state = state_index (d, &follows, separate_contexts ^ CTX_ANY);
|
|---|
| 2698 | else
|
|---|
| 2699 | state = -1;
|
|---|
| 2700 | if (separate_contexts & possible_contexts & CTX_NEWLINE)
|
|---|
| 2701 | state_newline = state_index (d, &follows, CTX_NEWLINE);
|
|---|
| 2702 | else
|
|---|
| 2703 | state_newline = state;
|
|---|
| 2704 | if (separate_contexts & possible_contexts & CTX_LETTER)
|
|---|
| 2705 | state_letter = state_index (d, &follows, CTX_LETTER);
|
|---|
| 2706 | else
|
|---|
| 2707 | state_letter = state;
|
|---|
| 2708 |
|
|---|
| 2709 | /* Set the transitions for each character in the current label. */
|
|---|
| 2710 | for (j = 0; j < CHARCLASS_INTS; ++j)
|
|---|
| 2711 | for (k = 0; k < INTBITS; ++k)
|
|---|
| 2712 | if (labels[i][j] & 1 << k)
|
|---|
| 2713 | {
|
|---|
| 2714 | int c = j * INTBITS + k;
|
|---|
| 2715 |
|
|---|
| 2716 | if (c == eolbyte)
|
|---|
| 2717 | trans[c] = state_newline;
|
|---|
| 2718 | else if (IS_WORD_CONSTITUENT (c))
|
|---|
| 2719 | trans[c] = state_letter;
|
|---|
| 2720 | else if (c < NOTCHAR)
|
|---|
| 2721 | trans[c] = state;
|
|---|
| 2722 | }
|
|---|
| 2723 | }
|
|---|
| 2724 |
|
|---|
| 2725 | for (i = 0; i < ngrps; ++i)
|
|---|
| 2726 | free (grps[i].elems);
|
|---|
| 2727 | free (follows.elems);
|
|---|
| 2728 | free (tmp.elems);
|
|---|
| 2729 | free (grps);
|
|---|
| 2730 | free (labels);
|
|---|
| 2731 | }
|
|---|
| 2732 |
|
|---|
| 2733 | /* Some routines for manipulating a compiled dfa's transition tables.
|
|---|
| 2734 | Each state may or may not have a transition table; if it does, and it
|
|---|
| 2735 | is a non-accepting state, then d->trans[state] points to its table.
|
|---|
| 2736 | If it is an accepting state then d->fails[state] points to its table.
|
|---|
| 2737 | If it has no table at all, then d->trans[state] is NULL.
|
|---|
| 2738 | TODO: Improve this comment, get rid of the unnecessary redundancy. */
|
|---|
| 2739 |
|
|---|
| 2740 | static void
|
|---|
| 2741 | build_state (state_num s, struct dfa *d)
|
|---|
| 2742 | {
|
|---|
| 2743 | state_num *trans; /* The new transition table. */
|
|---|
| 2744 | state_num i;
|
|---|
| 2745 |
|
|---|
| 2746 | /* Set an upper limit on the number of transition tables that will ever
|
|---|
| 2747 | exist at once. 1024 is arbitrary. The idea is that the frequently
|
|---|
| 2748 | used transition tables will be quickly rebuilt, whereas the ones that
|
|---|
| 2749 | were only needed once or twice will be cleared away. */
|
|---|
| 2750 | if (d->trcount >= 1024)
|
|---|
| 2751 | {
|
|---|
| 2752 | for (i = 0; i < d->tralloc; ++i)
|
|---|
| 2753 | {
|
|---|
| 2754 | free (d->trans[i]);
|
|---|
| 2755 | free (d->fails[i]);
|
|---|
| 2756 | d->trans[i] = d->fails[i] = NULL;
|
|---|
| 2757 | }
|
|---|
| 2758 | d->trcount = 0;
|
|---|
| 2759 | }
|
|---|
| 2760 |
|
|---|
| 2761 | ++d->trcount;
|
|---|
| 2762 |
|
|---|
| 2763 | /* Set up the success bits for this state. */
|
|---|
| 2764 | d->success[s] = 0;
|
|---|
| 2765 | if (ACCEPTS_IN_CONTEXT (d->states[s].context, CTX_NEWLINE, s, *d))
|
|---|
| 2766 | d->success[s] |= CTX_NEWLINE;
|
|---|
| 2767 | if (ACCEPTS_IN_CONTEXT (d->states[s].context, CTX_LETTER, s, *d))
|
|---|
| 2768 | d->success[s] |= CTX_LETTER;
|
|---|
| 2769 | if (ACCEPTS_IN_CONTEXT (d->states[s].context, CTX_NONE, s, *d))
|
|---|
| 2770 | d->success[s] |= CTX_NONE;
|
|---|
| 2771 |
|
|---|
| 2772 | MALLOC (trans, NOTCHAR);
|
|---|
| 2773 | dfastate (s, d, trans);
|
|---|
| 2774 |
|
|---|
| 2775 | /* Now go through the new transition table, and make sure that the trans
|
|---|
| 2776 | and fail arrays are allocated large enough to hold a pointer for the
|
|---|
| 2777 | largest state mentioned in the table. */
|
|---|
| 2778 | for (i = 0; i < NOTCHAR; ++i)
|
|---|
| 2779 | if (trans[i] >= d->tralloc)
|
|---|
| 2780 | {
|
|---|
| 2781 | state_num oldalloc = d->tralloc;
|
|---|
| 2782 |
|
|---|
| 2783 | while (trans[i] >= d->tralloc)
|
|---|
| 2784 | d->tralloc *= 2;
|
|---|
| 2785 | REALLOC (d->realtrans, d->tralloc + 1);
|
|---|
| 2786 | d->trans = d->realtrans + 1;
|
|---|
| 2787 | REALLOC (d->fails, d->tralloc);
|
|---|
| 2788 | REALLOC (d->success, d->tralloc);
|
|---|
| 2789 | REALLOC (d->newlines, d->tralloc);
|
|---|
| 2790 | while (oldalloc < d->tralloc)
|
|---|
| 2791 | {
|
|---|
| 2792 | d->trans[oldalloc] = NULL;
|
|---|
| 2793 | d->fails[oldalloc++] = NULL;
|
|---|
| 2794 | }
|
|---|
| 2795 | }
|
|---|
| 2796 |
|
|---|
| 2797 | /* Keep the newline transition in a special place so we can use it as
|
|---|
| 2798 | a sentinel. */
|
|---|
| 2799 | d->newlines[s] = trans[eolbyte];
|
|---|
| 2800 | trans[eolbyte] = -1;
|
|---|
| 2801 |
|
|---|
| 2802 | if (ACCEPTING (s, *d))
|
|---|
| 2803 | d->fails[s] = trans;
|
|---|
| 2804 | else
|
|---|
| 2805 | d->trans[s] = trans;
|
|---|
| 2806 | }
|
|---|
| 2807 |
|
|---|
| 2808 | static void
|
|---|
| 2809 | build_state_zero (struct dfa *d)
|
|---|
| 2810 | {
|
|---|
| 2811 | d->tralloc = 1;
|
|---|
| 2812 | d->trcount = 0;
|
|---|
| 2813 | CALLOC (d->realtrans, d->tralloc + 1);
|
|---|
| 2814 | d->trans = d->realtrans + 1;
|
|---|
| 2815 | CALLOC (d->fails, d->tralloc);
|
|---|
| 2816 | MALLOC (d->success, d->tralloc);
|
|---|
| 2817 | MALLOC (d->newlines, d->tralloc);
|
|---|
| 2818 | build_state (0, d);
|
|---|
| 2819 | }
|
|---|
| 2820 |
|
|---|
| 2821 | /* Multibyte character handling sub-routines for dfaexec. */
|
|---|
| 2822 |
|
|---|
| 2823 | /* Initial state may encounter the byte which is not a single byte character
|
|---|
| 2824 | nor 1st byte of a multibyte character. But it is incorrect for initial
|
|---|
| 2825 | state to accept such a byte.
|
|---|
| 2826 | For example, in sjis encoding the regular expression like "\\" accepts
|
|---|
| 2827 | the codepoint 0x5c, but should not accept the 2nd byte of the codepoint
|
|---|
| 2828 | 0x815c. Then Initial state must skip the bytes which are not a single byte
|
|---|
| 2829 | character nor 1st byte of a multibyte character. */
|
|---|
| 2830 | #define SKIP_REMAINS_MB_IF_INITIAL_STATE(s, p) \
|
|---|
| 2831 | if (s == 0) \
|
|---|
| 2832 | { \
|
|---|
| 2833 | while (inputwcs[p - buf_begin] == 0 \
|
|---|
| 2834 | && mblen_buf[p - buf_begin] > 0 \
|
|---|
| 2835 | && (unsigned char const *) p < buf_end) \
|
|---|
| 2836 | ++p; \
|
|---|
| 2837 | if ((char *) p >= end) \
|
|---|
| 2838 | { \
|
|---|
| 2839 | free (mblen_buf); \
|
|---|
| 2840 | free (inputwcs); \
|
|---|
| 2841 | *end = saved_end; \
|
|---|
| 2842 | return NULL; \
|
|---|
| 2843 | } \
|
|---|
| 2844 | }
|
|---|
| 2845 |
|
|---|
| 2846 | static void
|
|---|
| 2847 | realloc_trans_if_necessary (struct dfa *d, state_num new_state)
|
|---|
| 2848 | {
|
|---|
| 2849 | /* Make sure that the trans and fail arrays are allocated large enough
|
|---|
| 2850 | to hold a pointer for the new state. */
|
|---|
| 2851 | if (new_state >= d->tralloc)
|
|---|
| 2852 | {
|
|---|
| 2853 | state_num oldalloc = d->tralloc;
|
|---|
| 2854 |
|
|---|
| 2855 | while (new_state >= d->tralloc)
|
|---|
| 2856 | d->tralloc *= 2;
|
|---|
| 2857 | REALLOC (d->realtrans, d->tralloc + 1);
|
|---|
| 2858 | d->trans = d->realtrans + 1;
|
|---|
| 2859 | REALLOC (d->fails, d->tralloc);
|
|---|
| 2860 | REALLOC (d->success, d->tralloc);
|
|---|
| 2861 | REALLOC (d->newlines, d->tralloc);
|
|---|
| 2862 | while (oldalloc < d->tralloc)
|
|---|
| 2863 | {
|
|---|
| 2864 | d->trans[oldalloc] = NULL;
|
|---|
| 2865 | d->fails[oldalloc++] = NULL;
|
|---|
| 2866 | }
|
|---|
| 2867 | }
|
|---|
| 2868 | }
|
|---|
| 2869 |
|
|---|
| 2870 | /* Return values of transit_state_singlebyte(), and
|
|---|
| 2871 | transit_state_consume_1char. */
|
|---|
| 2872 | typedef enum
|
|---|
| 2873 | {
|
|---|
| 2874 | TRANSIT_STATE_IN_PROGRESS, /* State transition has not finished. */
|
|---|
| 2875 | TRANSIT_STATE_DONE, /* State transition has finished. */
|
|---|
| 2876 | TRANSIT_STATE_END_BUFFER /* Reach the end of the buffer. */
|
|---|
| 2877 | } status_transit_state;
|
|---|
| 2878 |
|
|---|
| 2879 | /* Consume a single byte and transit state from 's' to '*next_state'.
|
|---|
| 2880 | This function is almost same as the state transition routin in dfaexec().
|
|---|
| 2881 | But state transition is done just once, otherwise matching succeed or
|
|---|
| 2882 | reach the end of the buffer. */
|
|---|
| 2883 | static status_transit_state
|
|---|
| 2884 | transit_state_singlebyte (struct dfa *d, state_num s, unsigned char const *p,
|
|---|
| 2885 | state_num * next_state)
|
|---|
| 2886 | {
|
|---|
| 2887 | state_num *t;
|
|---|
| 2888 | state_num works = s;
|
|---|
| 2889 |
|
|---|
| 2890 | status_transit_state rval = TRANSIT_STATE_IN_PROGRESS;
|
|---|
| 2891 |
|
|---|
| 2892 | while (rval == TRANSIT_STATE_IN_PROGRESS)
|
|---|
| 2893 | {
|
|---|
| 2894 | if ((t = d->trans[works]) != NULL)
|
|---|
| 2895 | {
|
|---|
| 2896 | works = t[*p];
|
|---|
| 2897 | rval = TRANSIT_STATE_DONE;
|
|---|
| 2898 | if (works < 0)
|
|---|
| 2899 | works = 0;
|
|---|
| 2900 | }
|
|---|
| 2901 | else if (works < 0)
|
|---|
| 2902 | {
|
|---|
| 2903 | if (p == buf_end)
|
|---|
| 2904 | {
|
|---|
| 2905 | /* At the moment, it must not happen. */
|
|---|
| 2906 | abort ();
|
|---|
| 2907 | }
|
|---|
| 2908 | works = 0;
|
|---|
| 2909 | }
|
|---|
| 2910 | else if (d->fails[works])
|
|---|
| 2911 | {
|
|---|
| 2912 | works = d->fails[works][*p];
|
|---|
| 2913 | rval = TRANSIT_STATE_DONE;
|
|---|
| 2914 | }
|
|---|
| 2915 | else
|
|---|
| 2916 | {
|
|---|
| 2917 | build_state (works, d);
|
|---|
| 2918 | }
|
|---|
| 2919 | }
|
|---|
| 2920 | *next_state = works;
|
|---|
| 2921 | return rval;
|
|---|
| 2922 | }
|
|---|
| 2923 |
|
|---|
| 2924 | /* Match a "." against the current context. buf_begin[IDX] is the
|
|---|
| 2925 | current position. Return the length of the match, in bytes.
|
|---|
| 2926 | POS is the position of the ".". */
|
|---|
| 2927 | static int
|
|---|
| 2928 | match_anychar (struct dfa *d, state_num s, position pos, size_t idx)
|
|---|
| 2929 | {
|
|---|
| 2930 | int context;
|
|---|
| 2931 | wchar_t wc;
|
|---|
| 2932 | int mbclen;
|
|---|
| 2933 |
|
|---|
| 2934 | wc = inputwcs[idx];
|
|---|
| 2935 | mbclen = (mblen_buf[idx] == 0) ? 1 : mblen_buf[idx];
|
|---|
| 2936 |
|
|---|
| 2937 | /* Check syntax bits. */
|
|---|
| 2938 | if (wc == (wchar_t) eolbyte)
|
|---|
| 2939 | {
|
|---|
| 2940 | if (!(syntax_bits & RE_DOT_NEWLINE))
|
|---|
| 2941 | return 0;
|
|---|
| 2942 | }
|
|---|
| 2943 | else if (wc == (wchar_t) '\0')
|
|---|
| 2944 | {
|
|---|
| 2945 | if (syntax_bits & RE_DOT_NOT_NULL)
|
|---|
| 2946 | return 0;
|
|---|
| 2947 | }
|
|---|
| 2948 |
|
|---|
| 2949 | context = wchar_context (wc);
|
|---|
| 2950 | if (!SUCCEEDS_IN_CONTEXT (pos.constraint, d->states[s].context, context))
|
|---|
| 2951 | return 0;
|
|---|
| 2952 |
|
|---|
| 2953 | return mbclen;
|
|---|
| 2954 | }
|
|---|
| 2955 |
|
|---|
| 2956 | /* Match a bracket expression against the current context.
|
|---|
| 2957 | buf_begin[IDX] is the current position.
|
|---|
| 2958 | Return the length of the match, in bytes.
|
|---|
| 2959 | POS is the position of the bracket expression. */
|
|---|
| 2960 | static int
|
|---|
| 2961 | match_mb_charset (struct dfa *d, state_num s, position pos, size_t idx)
|
|---|
| 2962 | {
|
|---|
| 2963 | size_t i;
|
|---|
| 2964 | int match; /* Flag which represent that matching succeed. */
|
|---|
| 2965 | int match_len; /* Length of the character (or collating element)
|
|---|
| 2966 | with which this operator match. */
|
|---|
| 2967 | int op_len; /* Length of the operator. */
|
|---|
| 2968 | char buffer[128];
|
|---|
| 2969 | wchar_t wcbuf[6];
|
|---|
| 2970 |
|
|---|
| 2971 | /* Pointer to the structure to which we are currently referring. */
|
|---|
| 2972 | struct mb_char_classes *work_mbc;
|
|---|
| 2973 |
|
|---|
| 2974 | int context;
|
|---|
| 2975 | wchar_t wc; /* Current referring character. */
|
|---|
| 2976 |
|
|---|
| 2977 | wc = inputwcs[idx];
|
|---|
| 2978 |
|
|---|
| 2979 | /* Check syntax bits. */
|
|---|
| 2980 | if (wc == (wchar_t) eolbyte)
|
|---|
| 2981 | {
|
|---|
| 2982 | if (!(syntax_bits & RE_DOT_NEWLINE))
|
|---|
| 2983 | return 0;
|
|---|
| 2984 | }
|
|---|
| 2985 | else if (wc == (wchar_t) '\0')
|
|---|
| 2986 | {
|
|---|
| 2987 | if (syntax_bits & RE_DOT_NOT_NULL)
|
|---|
| 2988 | return 0;
|
|---|
| 2989 | }
|
|---|
| 2990 |
|
|---|
| 2991 | context = wchar_context (wc);
|
|---|
| 2992 | if (!SUCCEEDS_IN_CONTEXT (pos.constraint, d->states[s].context, context))
|
|---|
| 2993 | return 0;
|
|---|
| 2994 |
|
|---|
| 2995 | /* Assign the current referring operator to work_mbc. */
|
|---|
| 2996 | work_mbc = &(d->mbcsets[(d->multibyte_prop[pos.index]) >> 2]);
|
|---|
| 2997 | match = !work_mbc->invert;
|
|---|
| 2998 | match_len = (mblen_buf[idx] == 0) ? 1 : mblen_buf[idx];
|
|---|
| 2999 |
|
|---|
| 3000 | /* Match in range 0-255? */
|
|---|
| 3001 | if (wc < NOTCHAR && work_mbc->cset != -1
|
|---|
| 3002 | && tstbit ((unsigned char) wc, d->charclasses[work_mbc->cset]))
|
|---|
| 3003 | goto charset_matched;
|
|---|
| 3004 |
|
|---|
| 3005 | /* match with a character class? */
|
|---|
| 3006 | for (i = 0; i < work_mbc->nch_classes; i++)
|
|---|
| 3007 | {
|
|---|
| 3008 | if (iswctype ((wint_t) wc, work_mbc->ch_classes[i]))
|
|---|
| 3009 | goto charset_matched;
|
|---|
| 3010 | }
|
|---|
| 3011 |
|
|---|
| 3012 | strncpy (buffer, (char const *) buf_begin + idx, match_len);
|
|---|
| 3013 | buffer[match_len] = '\0';
|
|---|
| 3014 |
|
|---|
| 3015 | /* match with an equivalence class? */
|
|---|
| 3016 | for (i = 0; i < work_mbc->nequivs; i++)
|
|---|
| 3017 | {
|
|---|
| 3018 | op_len = strlen (work_mbc->equivs[i]);
|
|---|
| 3019 | strncpy (buffer, (char const *) buf_begin + idx, op_len);
|
|---|
| 3020 | buffer[op_len] = '\0';
|
|---|
| 3021 | if (strcoll (work_mbc->equivs[i], buffer) == 0)
|
|---|
| 3022 | {
|
|---|
| 3023 | match_len = op_len;
|
|---|
| 3024 | goto charset_matched;
|
|---|
| 3025 | }
|
|---|
| 3026 | }
|
|---|
| 3027 |
|
|---|
| 3028 | /* match with a collating element? */
|
|---|
| 3029 | for (i = 0; i < work_mbc->ncoll_elems; i++)
|
|---|
| 3030 | {
|
|---|
| 3031 | op_len = strlen (work_mbc->coll_elems[i]);
|
|---|
| 3032 | strncpy (buffer, (char const *) buf_begin + idx, op_len);
|
|---|
| 3033 | buffer[op_len] = '\0';
|
|---|
| 3034 |
|
|---|
| 3035 | if (strcoll (work_mbc->coll_elems[i], buffer) == 0)
|
|---|
| 3036 | {
|
|---|
| 3037 | match_len = op_len;
|
|---|
| 3038 | goto charset_matched;
|
|---|
| 3039 | }
|
|---|
| 3040 | }
|
|---|
| 3041 |
|
|---|
| 3042 | wcbuf[0] = wc;
|
|---|
| 3043 | wcbuf[1] = wcbuf[3] = wcbuf[5] = '\0';
|
|---|
| 3044 |
|
|---|
| 3045 | /* match with a range? */
|
|---|
| 3046 | for (i = 0; i < work_mbc->nranges; i++)
|
|---|
| 3047 | {
|
|---|
| 3048 | wcbuf[2] = work_mbc->range_sts[i];
|
|---|
| 3049 | wcbuf[4] = work_mbc->range_ends[i];
|
|---|
| 3050 |
|
|---|
| 3051 | if (wcscoll (wcbuf, wcbuf + 2) >= 0 && wcscoll (wcbuf + 4, wcbuf) >= 0)
|
|---|
| 3052 | goto charset_matched;
|
|---|
| 3053 | }
|
|---|
| 3054 |
|
|---|
| 3055 | /* match with a character? */
|
|---|
| 3056 | for (i = 0; i < work_mbc->nchars; i++)
|
|---|
| 3057 | {
|
|---|
| 3058 | if (wc == work_mbc->chars[i])
|
|---|
| 3059 | goto charset_matched;
|
|---|
| 3060 | }
|
|---|
| 3061 |
|
|---|
| 3062 | match = !match;
|
|---|
| 3063 |
|
|---|
| 3064 | charset_matched:
|
|---|
| 3065 | return match ? match_len : 0;
|
|---|
| 3066 | }
|
|---|
| 3067 |
|
|---|
| 3068 | /* Check each of `d->states[s].mbps.elem' can match or not. Then return the
|
|---|
| 3069 | array which corresponds to `d->states[s].mbps.elem' and each element of
|
|---|
| 3070 | the array contains the amount of the bytes with which the element can
|
|---|
| 3071 | match.
|
|---|
| 3072 | `idx' is the index from the buf_begin, and it is the current position
|
|---|
| 3073 | in the buffer.
|
|---|
| 3074 | Caller MUST free the array which this function return. */
|
|---|
| 3075 | static int *
|
|---|
| 3076 | check_matching_with_multibyte_ops (struct dfa *d, state_num s, size_t idx)
|
|---|
| 3077 | {
|
|---|
| 3078 | size_t i;
|
|---|
| 3079 | int *rarray;
|
|---|
| 3080 |
|
|---|
| 3081 | MALLOC (rarray, d->states[s].mbps.nelem);
|
|---|
| 3082 | for (i = 0; i < d->states[s].mbps.nelem; ++i)
|
|---|
| 3083 | {
|
|---|
| 3084 | position pos = d->states[s].mbps.elems[i];
|
|---|
| 3085 | switch (d->tokens[pos.index])
|
|---|
| 3086 | {
|
|---|
| 3087 | case ANYCHAR:
|
|---|
| 3088 | rarray[i] = match_anychar (d, s, pos, idx);
|
|---|
| 3089 | break;
|
|---|
| 3090 | case MBCSET:
|
|---|
| 3091 | rarray[i] = match_mb_charset (d, s, pos, idx);
|
|---|
| 3092 | break;
|
|---|
| 3093 | default:
|
|---|
| 3094 | break; /* cannot happen. */
|
|---|
| 3095 | }
|
|---|
| 3096 | }
|
|---|
| 3097 | return rarray;
|
|---|
| 3098 | }
|
|---|
| 3099 |
|
|---|
| 3100 | /* Consume a single character and enumerate all of the positions which can
|
|---|
| 3101 | be next position from the state `s'.
|
|---|
| 3102 | `match_lens' is the input. It can be NULL, but it can also be the output
|
|---|
| 3103 | of check_matching_with_multibyte_ops() for optimization.
|
|---|
| 3104 | `mbclen' and `pps' are the output. `mbclen' is the length of the
|
|---|
| 3105 | character consumed, and `pps' is the set this function enumerate. */
|
|---|
| 3106 | static status_transit_state
|
|---|
| 3107 | transit_state_consume_1char (struct dfa *d, state_num s,
|
|---|
| 3108 | unsigned char const **pp,
|
|---|
| 3109 | int *match_lens, int *mbclen, position_set * pps)
|
|---|
| 3110 | {
|
|---|
| 3111 | size_t i, j;
|
|---|
| 3112 | int k;
|
|---|
| 3113 | state_num s1, s2;
|
|---|
| 3114 | int *work_mbls;
|
|---|
| 3115 | status_transit_state rs = TRANSIT_STATE_DONE;
|
|---|
| 3116 |
|
|---|
| 3117 | /* Calculate the length of the (single/multi byte) character
|
|---|
| 3118 | to which p points. */
|
|---|
| 3119 | *mbclen = (mblen_buf[*pp - buf_begin] == 0) ? 1 : mblen_buf[*pp - buf_begin];
|
|---|
| 3120 |
|
|---|
| 3121 | /* Calculate the state which can be reached from the state `s' by
|
|---|
| 3122 | consuming `*mbclen' single bytes from the buffer. */
|
|---|
| 3123 | s1 = s;
|
|---|
| 3124 | for (k = 0; k < *mbclen; k++)
|
|---|
| 3125 | {
|
|---|
| 3126 | s2 = s1;
|
|---|
| 3127 | rs = transit_state_singlebyte (d, s2, (*pp)++, &s1);
|
|---|
| 3128 | }
|
|---|
| 3129 | /* Copy the positions contained by `s1' to the set `pps'. */
|
|---|
| 3130 | copy (&(d->states[s1].elems), pps);
|
|---|
| 3131 |
|
|---|
| 3132 | /* Check (input) match_lens, and initialize if it is NULL. */
|
|---|
| 3133 | if (match_lens == NULL && d->states[s].mbps.nelem != 0)
|
|---|
| 3134 | work_mbls = check_matching_with_multibyte_ops (d, s, *pp - buf_begin);
|
|---|
| 3135 | else
|
|---|
| 3136 | work_mbls = match_lens;
|
|---|
| 3137 |
|
|---|
| 3138 | /* Add all of the positions which can be reached from `s' by consuming
|
|---|
| 3139 | a single character. */
|
|---|
| 3140 | for (i = 0; i < d->states[s].mbps.nelem; i++)
|
|---|
| 3141 | {
|
|---|
| 3142 | if (work_mbls[i] == *mbclen)
|
|---|
| 3143 | for (j = 0; j < d->follows[d->states[s].mbps.elems[i].index].nelem;
|
|---|
| 3144 | j++)
|
|---|
| 3145 | insert (d->follows[d->states[s].mbps.elems[i].index].elems[j], pps);
|
|---|
| 3146 | }
|
|---|
| 3147 |
|
|---|
| 3148 | if (match_lens == NULL && work_mbls != NULL)
|
|---|
| 3149 | free (work_mbls);
|
|---|
| 3150 |
|
|---|
| 3151 | /* FIXME: this return value is always ignored. */
|
|---|
| 3152 | return rs;
|
|---|
| 3153 | }
|
|---|
| 3154 |
|
|---|
| 3155 | /* Transit state from s, then return new state and update the pointer of the
|
|---|
| 3156 | buffer. This function is for some operator which can match with a multi-
|
|---|
| 3157 | byte character or a collating element (which may be multi characters). */
|
|---|
| 3158 | static state_num
|
|---|
| 3159 | transit_state (struct dfa *d, state_num s, unsigned char const **pp)
|
|---|
| 3160 | {
|
|---|
| 3161 | state_num s1;
|
|---|
| 3162 | int mbclen; /* The length of current input multibyte character. */
|
|---|
| 3163 | int maxlen = 0;
|
|---|
| 3164 | size_t i, j;
|
|---|
| 3165 | int *match_lens = NULL;
|
|---|
| 3166 | size_t nelem = d->states[s].mbps.nelem; /* Just a alias. */
|
|---|
| 3167 | position_set follows;
|
|---|
| 3168 | unsigned char const *p1 = *pp;
|
|---|
| 3169 | wchar_t wc;
|
|---|
| 3170 |
|
|---|
| 3171 | if (nelem > 0)
|
|---|
| 3172 | /* This state has (a) multibyte operator(s).
|
|---|
| 3173 | We check whether each of them can match or not. */
|
|---|
| 3174 | {
|
|---|
| 3175 | /* Note: caller must free the return value of this function. */
|
|---|
| 3176 | match_lens = check_matching_with_multibyte_ops (d, s, *pp - buf_begin);
|
|---|
| 3177 |
|
|---|
| 3178 | for (i = 0; i < nelem; i++)
|
|---|
| 3179 | /* Search the operator which match the longest string,
|
|---|
| 3180 | in this state. */
|
|---|
| 3181 | {
|
|---|
| 3182 | if (match_lens[i] > maxlen)
|
|---|
| 3183 | maxlen = match_lens[i];
|
|---|
| 3184 | }
|
|---|
| 3185 | }
|
|---|
| 3186 |
|
|---|
| 3187 | if (nelem == 0 || maxlen == 0)
|
|---|
| 3188 | /* This state has no multibyte operator which can match.
|
|---|
| 3189 | We need to check only one single byte character. */
|
|---|
| 3190 | {
|
|---|
| 3191 | status_transit_state rs;
|
|---|
| 3192 | rs = transit_state_singlebyte (d, s, *pp, &s1);
|
|---|
| 3193 |
|
|---|
| 3194 | /* We must update the pointer if state transition succeeded. */
|
|---|
| 3195 | if (rs == TRANSIT_STATE_DONE)
|
|---|
| 3196 | ++*pp;
|
|---|
| 3197 |
|
|---|
| 3198 | free (match_lens);
|
|---|
| 3199 | return s1;
|
|---|
| 3200 | }
|
|---|
| 3201 |
|
|---|
| 3202 | /* This state has some operators which can match a multibyte character. */
|
|---|
| 3203 | alloc_position_set (&follows, d->nleaves);
|
|---|
| 3204 |
|
|---|
| 3205 | /* `maxlen' may be longer than the length of a character, because it may
|
|---|
| 3206 | not be a character but a (multi character) collating element.
|
|---|
| 3207 | We enumerate all of the positions which `s' can reach by consuming
|
|---|
| 3208 | `maxlen' bytes. */
|
|---|
| 3209 | transit_state_consume_1char (d, s, pp, match_lens, &mbclen, &follows);
|
|---|
| 3210 |
|
|---|
| 3211 | wc = inputwcs[*pp - mbclen - buf_begin];
|
|---|
| 3212 | s1 = state_index (d, &follows, wchar_context (wc));
|
|---|
| 3213 | realloc_trans_if_necessary (d, s1);
|
|---|
| 3214 |
|
|---|
| 3215 | while (*pp - p1 < maxlen)
|
|---|
| 3216 | {
|
|---|
| 3217 | transit_state_consume_1char (d, s1, pp, NULL, &mbclen, &follows);
|
|---|
| 3218 |
|
|---|
| 3219 | for (i = 0; i < nelem; i++)
|
|---|
| 3220 | {
|
|---|
| 3221 | if (match_lens[i] == *pp - p1)
|
|---|
| 3222 | for (j = 0;
|
|---|
| 3223 | j < d->follows[d->states[s1].mbps.elems[i].index].nelem; j++)
|
|---|
| 3224 | insert (d->follows[d->states[s1].mbps.elems[i].index].elems[j],
|
|---|
| 3225 | &follows);
|
|---|
| 3226 | }
|
|---|
| 3227 |
|
|---|
| 3228 | wc = inputwcs[*pp - mbclen - buf_begin];
|
|---|
| 3229 | s1 = state_index (d, &follows, wchar_context (wc));
|
|---|
| 3230 | realloc_trans_if_necessary (d, s1);
|
|---|
| 3231 | }
|
|---|
| 3232 | free (match_lens);
|
|---|
| 3233 | free (follows.elems);
|
|---|
| 3234 | return s1;
|
|---|
| 3235 | }
|
|---|
| 3236 |
|
|---|
| 3237 |
|
|---|
| 3238 | /* Initialize mblen_buf and inputwcs with data from the next line. */
|
|---|
| 3239 |
|
|---|
| 3240 | static void
|
|---|
| 3241 | prepare_wc_buf (const char *begin, const char *end)
|
|---|
| 3242 | {
|
|---|
| 3243 | #if MBS_SUPPORT
|
|---|
| 3244 | unsigned char eol = eolbyte;
|
|---|
| 3245 | size_t remain_bytes, i;
|
|---|
| 3246 |
|
|---|
| 3247 | buf_begin = (unsigned char *) begin;
|
|---|
| 3248 |
|
|---|
| 3249 | remain_bytes = 0;
|
|---|
| 3250 | for (i = 0; i < end - begin + 1; i++)
|
|---|
| 3251 | {
|
|---|
| 3252 | if (remain_bytes == 0)
|
|---|
| 3253 | {
|
|---|
| 3254 | remain_bytes
|
|---|
| 3255 | = mbrtowc (inputwcs + i, begin + i, end - begin - i + 1, &mbs);
|
|---|
| 3256 | if (remain_bytes < 1
|
|---|
| 3257 | || remain_bytes == (size_t) -1
|
|---|
| 3258 | || remain_bytes == (size_t) -2
|
|---|
| 3259 | || (remain_bytes == 1 && inputwcs[i] == (wchar_t) begin[i]))
|
|---|
| 3260 | {
|
|---|
| 3261 | remain_bytes = 0;
|
|---|
| 3262 | inputwcs[i] = (wchar_t) begin[i];
|
|---|
| 3263 | mblen_buf[i] = 0;
|
|---|
| 3264 | if (begin[i] == eol)
|
|---|
| 3265 | break;
|
|---|
| 3266 | }
|
|---|
| 3267 | else
|
|---|
| 3268 | {
|
|---|
| 3269 | mblen_buf[i] = remain_bytes;
|
|---|
| 3270 | remain_bytes--;
|
|---|
| 3271 | }
|
|---|
| 3272 | }
|
|---|
| 3273 | else
|
|---|
| 3274 | {
|
|---|
| 3275 | mblen_buf[i] = remain_bytes;
|
|---|
| 3276 | inputwcs[i] = 0;
|
|---|
| 3277 | remain_bytes--;
|
|---|
| 3278 | }
|
|---|
| 3279 | }
|
|---|
| 3280 |
|
|---|
| 3281 | buf_end = (unsigned char *) (begin + i);
|
|---|
| 3282 | mblen_buf[i] = 0;
|
|---|
| 3283 | inputwcs[i] = 0; /* sentinel */
|
|---|
| 3284 | #endif /* MBS_SUPPORT */
|
|---|
| 3285 | }
|
|---|
| 3286 |
|
|---|
| 3287 | /* Search through a buffer looking for a match to the given struct dfa.
|
|---|
| 3288 | Find the first occurrence of a string matching the regexp in the
|
|---|
| 3289 | buffer, and the shortest possible version thereof. Return a pointer to
|
|---|
| 3290 | the first character after the match, or NULL if none is found. BEGIN
|
|---|
| 3291 | points to the beginning of the buffer, and END points to the first byte
|
|---|
| 3292 | after its end. Note however that we store a sentinel byte (usually
|
|---|
| 3293 | newline) in *END, so the actual buffer must be one byte longer.
|
|---|
| 3294 | When ALLOW_NL is nonzero, newlines may appear in the matching string.
|
|---|
| 3295 | If COUNT is non-NULL, increment *COUNT once for each newline processed.
|
|---|
| 3296 | Finally, if BACKREF is non-NULL set *BACKREF to indicate whether we
|
|---|
| 3297 | encountered a back-reference (1) or not (0). The caller may use this
|
|---|
| 3298 | to decide whether to fall back on a backtracking matcher. */
|
|---|
| 3299 | char *
|
|---|
| 3300 | dfaexec (struct dfa *d, char const *begin, char *end,
|
|---|
| 3301 | int allow_nl, size_t *count, int *backref)
|
|---|
| 3302 | {
|
|---|
| 3303 | state_num s, s1; /* Current state. */
|
|---|
| 3304 | unsigned char const *p; /* Current input character. */
|
|---|
| 3305 | state_num **trans, *t; /* Copy of d->trans so it can be optimized
|
|---|
| 3306 | into a register. */
|
|---|
| 3307 | unsigned char eol = eolbyte; /* Likewise for eolbyte. */
|
|---|
| 3308 | unsigned char saved_end;
|
|---|
| 3309 |
|
|---|
| 3310 | if (!d->tralloc)
|
|---|
| 3311 | build_state_zero (d);
|
|---|
| 3312 |
|
|---|
| 3313 | s = s1 = 0;
|
|---|
| 3314 | p = (unsigned char const *) begin;
|
|---|
| 3315 | trans = d->trans;
|
|---|
| 3316 | saved_end = *(unsigned char *) end;
|
|---|
| 3317 | *end = eol;
|
|---|
| 3318 |
|
|---|
| 3319 | if (d->mb_cur_max > 1)
|
|---|
| 3320 | {
|
|---|
| 3321 | MALLOC (mblen_buf, end - begin + 2);
|
|---|
| 3322 | MALLOC (inputwcs, end - begin + 2);
|
|---|
| 3323 | memset (&mbs, 0, sizeof (mbstate_t));
|
|---|
| 3324 | prepare_wc_buf ((const char *) p, end);
|
|---|
| 3325 | }
|
|---|
| 3326 |
|
|---|
| 3327 | for (;;)
|
|---|
| 3328 | {
|
|---|
| 3329 | if (d->mb_cur_max > 1)
|
|---|
| 3330 | while ((t = trans[s]) != NULL)
|
|---|
| 3331 | {
|
|---|
| 3332 | if (p > buf_end)
|
|---|
| 3333 | break;
|
|---|
| 3334 | s1 = s;
|
|---|
| 3335 | SKIP_REMAINS_MB_IF_INITIAL_STATE (s, p);
|
|---|
| 3336 |
|
|---|
| 3337 | if (d->states[s].mbps.nelem == 0)
|
|---|
| 3338 | {
|
|---|
| 3339 | s = t[*p++];
|
|---|
| 3340 | continue;
|
|---|
| 3341 | }
|
|---|
| 3342 |
|
|---|
| 3343 | /* Falling back to the glibc matcher in this case gives
|
|---|
| 3344 | better performance (up to 25% better on [a-z], for
|
|---|
| 3345 | example) and enables support for collating symbols and
|
|---|
| 3346 | equivalence classes. */
|
|---|
| 3347 | if (backref)
|
|---|
| 3348 | {
|
|---|
| 3349 | *backref = 1;
|
|---|
| 3350 | free (mblen_buf);
|
|---|
| 3351 | free (inputwcs);
|
|---|
| 3352 | *end = saved_end;
|
|---|
| 3353 | return (char *) p;
|
|---|
| 3354 | }
|
|---|
| 3355 |
|
|---|
| 3356 | /* Can match with a multibyte character (and multi character
|
|---|
| 3357 | collating element). Transition table might be updated. */
|
|---|
| 3358 | s = transit_state (d, s, &p);
|
|---|
| 3359 | trans = d->trans;
|
|---|
| 3360 | }
|
|---|
| 3361 | else
|
|---|
| 3362 | {
|
|---|
| 3363 | while ((t = trans[s]) != NULL)
|
|---|
| 3364 | {
|
|---|
| 3365 | s1 = t[*p++];
|
|---|
| 3366 | if ((t = trans[s1]) == NULL)
|
|---|
| 3367 | {
|
|---|
| 3368 | state_num tmp = s;
|
|---|
| 3369 | s = s1;
|
|---|
| 3370 | s1 = tmp; /* swap */
|
|---|
| 3371 | break;
|
|---|
| 3372 | }
|
|---|
| 3373 | s = t[*p++];
|
|---|
| 3374 | }
|
|---|
| 3375 | }
|
|---|
| 3376 |
|
|---|
| 3377 | if (s >= 0 && (char *) p <= end && d->fails[s])
|
|---|
| 3378 | {
|
|---|
| 3379 | if (d->success[s] & sbit[*p])
|
|---|
| 3380 | {
|
|---|
| 3381 | if (backref)
|
|---|
| 3382 | *backref = (d->states[s].backref != 0);
|
|---|
| 3383 | if (d->mb_cur_max > 1)
|
|---|
| 3384 | {
|
|---|
| 3385 | free (mblen_buf);
|
|---|
| 3386 | free (inputwcs);
|
|---|
| 3387 | }
|
|---|
| 3388 | *end = saved_end;
|
|---|
| 3389 | return (char *) p;
|
|---|
| 3390 | }
|
|---|
| 3391 |
|
|---|
| 3392 | s1 = s;
|
|---|
| 3393 | if (d->mb_cur_max > 1)
|
|---|
| 3394 | {
|
|---|
| 3395 | /* Can match with a multibyte character (and multicharacter
|
|---|
| 3396 | collating element). Transition table might be updated. */
|
|---|
| 3397 | s = transit_state (d, s, &p);
|
|---|
| 3398 | trans = d->trans;
|
|---|
| 3399 | }
|
|---|
| 3400 | else
|
|---|
| 3401 | s = d->fails[s][*p++];
|
|---|
| 3402 | continue;
|
|---|
| 3403 | }
|
|---|
| 3404 |
|
|---|
| 3405 | /* If the previous character was a newline, count it. */
|
|---|
| 3406 | if ((char *) p <= end && p[-1] == eol)
|
|---|
| 3407 | {
|
|---|
| 3408 | if (count)
|
|---|
| 3409 | ++*count;
|
|---|
| 3410 |
|
|---|
| 3411 | if (d->mb_cur_max > 1)
|
|---|
| 3412 | prepare_wc_buf ((const char *) p, end);
|
|---|
| 3413 | }
|
|---|
| 3414 |
|
|---|
| 3415 | /* Check if we've run off the end of the buffer. */
|
|---|
| 3416 | if ((char *) p > end)
|
|---|
| 3417 | {
|
|---|
| 3418 | if (d->mb_cur_max > 1)
|
|---|
| 3419 | {
|
|---|
| 3420 | free (mblen_buf);
|
|---|
| 3421 | free (inputwcs);
|
|---|
| 3422 | }
|
|---|
| 3423 | *end = saved_end;
|
|---|
| 3424 | return NULL;
|
|---|
| 3425 | }
|
|---|
| 3426 |
|
|---|
| 3427 | if (s >= 0)
|
|---|
| 3428 | {
|
|---|
| 3429 | build_state (s, d);
|
|---|
| 3430 | trans = d->trans;
|
|---|
| 3431 | continue;
|
|---|
| 3432 | }
|
|---|
| 3433 |
|
|---|
| 3434 | if (p[-1] == eol && allow_nl)
|
|---|
| 3435 | {
|
|---|
| 3436 | s = d->newlines[s1];
|
|---|
| 3437 | continue;
|
|---|
| 3438 | }
|
|---|
| 3439 |
|
|---|
| 3440 | s = 0;
|
|---|
| 3441 | }
|
|---|
| 3442 | }
|
|---|
| 3443 |
|
|---|
| 3444 | static void
|
|---|
| 3445 | free_mbdata (struct dfa *d)
|
|---|
| 3446 | {
|
|---|
| 3447 | size_t i;
|
|---|
| 3448 |
|
|---|
| 3449 | free (d->multibyte_prop);
|
|---|
| 3450 | d->multibyte_prop = NULL;
|
|---|
| 3451 |
|
|---|
| 3452 | for (i = 0; i < d->nmbcsets; ++i)
|
|---|
| 3453 | {
|
|---|
| 3454 | size_t j;
|
|---|
| 3455 | struct mb_char_classes *p = &(d->mbcsets[i]);
|
|---|
| 3456 | free (p->chars);
|
|---|
| 3457 | free (p->ch_classes);
|
|---|
| 3458 | free (p->range_sts);
|
|---|
| 3459 | free (p->range_ends);
|
|---|
| 3460 |
|
|---|
| 3461 | for (j = 0; j < p->nequivs; ++j)
|
|---|
| 3462 | free (p->equivs[j]);
|
|---|
| 3463 | free (p->equivs);
|
|---|
| 3464 |
|
|---|
| 3465 | for (j = 0; j < p->ncoll_elems; ++j)
|
|---|
| 3466 | free (p->coll_elems[j]);
|
|---|
| 3467 | free (p->coll_elems);
|
|---|
| 3468 | }
|
|---|
| 3469 |
|
|---|
| 3470 | free (d->mbcsets);
|
|---|
| 3471 | d->mbcsets = NULL;
|
|---|
| 3472 | d->nmbcsets = 0;
|
|---|
| 3473 | }
|
|---|
| 3474 |
|
|---|
| 3475 | /* Initialize the components of a dfa that the other routines don't
|
|---|
| 3476 | initialize for themselves. */
|
|---|
| 3477 | void
|
|---|
| 3478 | dfainit (struct dfa *d)
|
|---|
| 3479 | {
|
|---|
| 3480 | memset (d, 0, sizeof *d);
|
|---|
| 3481 |
|
|---|
| 3482 | d->calloc = 1;
|
|---|
| 3483 | MALLOC (d->charclasses, d->calloc);
|
|---|
| 3484 |
|
|---|
| 3485 | d->talloc = 1;
|
|---|
| 3486 | MALLOC (d->tokens, d->talloc);
|
|---|
| 3487 |
|
|---|
| 3488 | d->mb_cur_max = MB_CUR_MAX;
|
|---|
| 3489 |
|
|---|
| 3490 | if (d->mb_cur_max > 1)
|
|---|
| 3491 | {
|
|---|
| 3492 | d->nmultibyte_prop = 1;
|
|---|
| 3493 | MALLOC (d->multibyte_prop, d->nmultibyte_prop);
|
|---|
| 3494 | d->mbcsets_alloc = 1;
|
|---|
| 3495 | MALLOC (d->mbcsets, d->mbcsets_alloc);
|
|---|
| 3496 | }
|
|---|
| 3497 | }
|
|---|
| 3498 |
|
|---|
| 3499 | static void
|
|---|
| 3500 | dfaoptimize (struct dfa *d)
|
|---|
| 3501 | {
|
|---|
| 3502 | size_t i;
|
|---|
| 3503 |
|
|---|
| 3504 | if (!MBS_SUPPORT || !using_utf8 ())
|
|---|
| 3505 | return;
|
|---|
| 3506 |
|
|---|
| 3507 | for (i = 0; i < d->tindex; ++i)
|
|---|
| 3508 | {
|
|---|
| 3509 | switch (d->tokens[i])
|
|---|
| 3510 | {
|
|---|
| 3511 | case ANYCHAR:
|
|---|
| 3512 | /* Lowered. */
|
|---|
| 3513 | abort ();
|
|---|
| 3514 | case MBCSET:
|
|---|
| 3515 | /* Requires multi-byte algorithm. */
|
|---|
| 3516 | return;
|
|---|
| 3517 | default:
|
|---|
| 3518 | break;
|
|---|
| 3519 | }
|
|---|
| 3520 | }
|
|---|
| 3521 |
|
|---|
| 3522 | free_mbdata (d);
|
|---|
| 3523 | d->mb_cur_max = 1;
|
|---|
| 3524 | }
|
|---|
| 3525 |
|
|---|
| 3526 | /* Parse and analyze a single string of the given length. */
|
|---|
| 3527 | void
|
|---|
| 3528 | dfacomp (char const *s, size_t len, struct dfa *d, int searchflag)
|
|---|
| 3529 | {
|
|---|
| 3530 | dfainit (d);
|
|---|
| 3531 | dfaparse (s, len, d);
|
|---|
| 3532 | dfamust (d);
|
|---|
| 3533 | dfaoptimize (d);
|
|---|
| 3534 | dfaanalyze (d, searchflag);
|
|---|
| 3535 | }
|
|---|
| 3536 |
|
|---|
| 3537 | /* Free the storage held by the components of a dfa. */
|
|---|
| 3538 | void
|
|---|
| 3539 | dfafree (struct dfa *d)
|
|---|
| 3540 | {
|
|---|
| 3541 | size_t i;
|
|---|
| 3542 | struct dfamust *dm, *ndm;
|
|---|
| 3543 |
|
|---|
| 3544 | free (d->charclasses);
|
|---|
| 3545 | free (d->tokens);
|
|---|
| 3546 |
|
|---|
| 3547 | if (d->mb_cur_max > 1)
|
|---|
| 3548 | free_mbdata (d);
|
|---|
| 3549 |
|
|---|
| 3550 | for (i = 0; i < d->sindex; ++i)
|
|---|
| 3551 | {
|
|---|
| 3552 | free (d->states[i].elems.elems);
|
|---|
| 3553 | if (MBS_SUPPORT)
|
|---|
| 3554 | free (d->states[i].mbps.elems);
|
|---|
| 3555 | }
|
|---|
| 3556 | free (d->states);
|
|---|
| 3557 | for (i = 0; i < d->tindex; ++i)
|
|---|
| 3558 | free (d->follows[i].elems);
|
|---|
| 3559 | free (d->follows);
|
|---|
| 3560 | for (i = 0; i < d->tralloc; ++i)
|
|---|
| 3561 | {
|
|---|
| 3562 | free (d->trans[i]);
|
|---|
| 3563 | free (d->fails[i]);
|
|---|
| 3564 | }
|
|---|
| 3565 | free (d->realtrans);
|
|---|
| 3566 | free (d->fails);
|
|---|
| 3567 | free (d->newlines);
|
|---|
| 3568 | free (d->success);
|
|---|
| 3569 | for (dm = d->musts; dm; dm = ndm)
|
|---|
| 3570 | {
|
|---|
| 3571 | ndm = dm->next;
|
|---|
| 3572 | free (dm->must);
|
|---|
| 3573 | free (dm);
|
|---|
| 3574 | }
|
|---|
| 3575 | }
|
|---|
| 3576 |
|
|---|
| 3577 | /* Having found the postfix representation of the regular expression,
|
|---|
| 3578 | try to find a long sequence of characters that must appear in any line
|
|---|
| 3579 | containing the r.e.
|
|---|
| 3580 | Finding a "longest" sequence is beyond the scope here;
|
|---|
| 3581 | we take an easy way out and hope for the best.
|
|---|
| 3582 | (Take "(ab|a)b"--please.)
|
|---|
| 3583 |
|
|---|
| 3584 | We do a bottom-up calculation of sequences of characters that must appear
|
|---|
| 3585 | in matches of r.e.'s represented by trees rooted at the nodes of the postfix
|
|---|
| 3586 | representation:
|
|---|
| 3587 | sequences that must appear at the left of the match ("left")
|
|---|
| 3588 | sequences that must appear at the right of the match ("right")
|
|---|
| 3589 | lists of sequences that must appear somewhere in the match ("in")
|
|---|
| 3590 | sequences that must constitute the match ("is")
|
|---|
| 3591 |
|
|---|
| 3592 | When we get to the root of the tree, we use one of the longest of its
|
|---|
| 3593 | calculated "in" sequences as our answer. The sequence we find is returned in
|
|---|
| 3594 | d->must (where "d" is the single argument passed to "dfamust");
|
|---|
| 3595 | the length of the sequence is returned in d->mustn.
|
|---|
| 3596 |
|
|---|
| 3597 | The sequences calculated for the various types of node (in pseudo ANSI c)
|
|---|
| 3598 | are shown below. "p" is the operand of unary operators (and the left-hand
|
|---|
| 3599 | operand of binary operators); "q" is the right-hand operand of binary
|
|---|
| 3600 | operators.
|
|---|
| 3601 |
|
|---|
| 3602 | "ZERO" means "a zero-length sequence" below.
|
|---|
| 3603 |
|
|---|
| 3604 | Type left right is in
|
|---|
| 3605 | ---- ---- ----- -- --
|
|---|
| 3606 | char c # c # c # c # c
|
|---|
| 3607 |
|
|---|
| 3608 | ANYCHAR ZERO ZERO ZERO ZERO
|
|---|
| 3609 |
|
|---|
| 3610 | MBCSET ZERO ZERO ZERO ZERO
|
|---|
| 3611 |
|
|---|
| 3612 | CSET ZERO ZERO ZERO ZERO
|
|---|
| 3613 |
|
|---|
| 3614 | STAR ZERO ZERO ZERO ZERO
|
|---|
| 3615 |
|
|---|
| 3616 | QMARK ZERO ZERO ZERO ZERO
|
|---|
| 3617 |
|
|---|
| 3618 | PLUS p->left p->right ZERO p->in
|
|---|
| 3619 |
|
|---|
| 3620 | CAT (p->is==ZERO)? (q->is==ZERO)? (p->is!=ZERO && p->in plus
|
|---|
| 3621 | p->left : q->right : q->is!=ZERO) ? q->in plus
|
|---|
| 3622 | p->is##q->left p->right##q->is p->is##q->is : p->right##q->left
|
|---|
| 3623 | ZERO
|
|---|
| 3624 |
|
|---|
| 3625 | OR longest common longest common (do p->is and substrings common to
|
|---|
| 3626 | leading trailing q->is have same p->in and q->in
|
|---|
| 3627 | (sub)sequence (sub)sequence length and
|
|---|
| 3628 | of p->left of p->right content) ?
|
|---|
| 3629 | and q->left and q->right p->is : NULL
|
|---|
| 3630 |
|
|---|
| 3631 | If there's anything else we recognize in the tree, all four sequences get set
|
|---|
| 3632 | to zero-length sequences. If there's something we don't recognize in the tree,
|
|---|
| 3633 | we just return a zero-length sequence.
|
|---|
| 3634 |
|
|---|
| 3635 | Break ties in favor of infrequent letters (choosing 'zzz' in preference to
|
|---|
| 3636 | 'aaa')?
|
|---|
| 3637 |
|
|---|
| 3638 | And. . .is it here or someplace that we might ponder "optimizations" such as
|
|---|
| 3639 | egrep 'psi|epsilon' -> egrep 'psi'
|
|---|
| 3640 | egrep 'pepsi|epsilon' -> egrep 'epsi'
|
|---|
| 3641 | (Yes, we now find "epsi" as a "string
|
|---|
| 3642 | that must occur", but we might also
|
|---|
| 3643 | simplify the *entire* r.e. being sought)
|
|---|
| 3644 | grep '[c]' -> grep 'c'
|
|---|
| 3645 | grep '(ab|a)b' -> grep 'ab'
|
|---|
| 3646 | grep 'ab*' -> grep 'a'
|
|---|
| 3647 | grep 'a*b' -> grep 'b'
|
|---|
| 3648 |
|
|---|
| 3649 | There are several issues:
|
|---|
| 3650 |
|
|---|
| 3651 | Is optimization easy (enough)?
|
|---|
| 3652 |
|
|---|
| 3653 | Does optimization actually accomplish anything,
|
|---|
| 3654 | or is the automaton you get from "psi|epsilon" (for example)
|
|---|
| 3655 | the same as the one you get from "psi" (for example)?
|
|---|
| 3656 |
|
|---|
| 3657 | Are optimizable r.e.'s likely to be used in real-life situations
|
|---|
| 3658 | (something like 'ab*' is probably unlikely; something like is
|
|---|
| 3659 | 'psi|epsilon' is likelier)? */
|
|---|
| 3660 |
|
|---|
| 3661 | static char *
|
|---|
| 3662 | icatalloc (char *old, char const *new)
|
|---|
| 3663 | {
|
|---|
| 3664 | char *result;
|
|---|
| 3665 | size_t oldsize = old == NULL ? 0 : strlen (old);
|
|---|
| 3666 | size_t newsize = new == NULL ? 0 : strlen (new);
|
|---|
| 3667 | if (newsize == 0)
|
|---|
| 3668 | return old;
|
|---|
| 3669 | result = xrealloc (old, oldsize + newsize + 1);
|
|---|
| 3670 | memcpy (result + oldsize, new, newsize + 1);
|
|---|
| 3671 | return result;
|
|---|
| 3672 | }
|
|---|
| 3673 |
|
|---|
| 3674 | static char *
|
|---|
| 3675 | icpyalloc (char const *string)
|
|---|
| 3676 | {
|
|---|
| 3677 | return icatalloc (NULL, string);
|
|---|
| 3678 | }
|
|---|
| 3679 |
|
|---|
| 3680 | static char *_GL_ATTRIBUTE_PURE
|
|---|
| 3681 | istrstr (char const *lookin, char const *lookfor)
|
|---|
| 3682 | {
|
|---|
| 3683 | char const *cp;
|
|---|
| 3684 | size_t len;
|
|---|
| 3685 |
|
|---|
| 3686 | len = strlen (lookfor);
|
|---|
| 3687 | for (cp = lookin; *cp != '\0'; ++cp)
|
|---|
| 3688 | if (strncmp (cp, lookfor, len) == 0)
|
|---|
| 3689 | return (char *) cp;
|
|---|
| 3690 | return NULL;
|
|---|
| 3691 | }
|
|---|
| 3692 |
|
|---|
| 3693 | static void
|
|---|
| 3694 | freelist (char **cpp)
|
|---|
| 3695 | {
|
|---|
| 3696 | size_t i;
|
|---|
| 3697 |
|
|---|
| 3698 | if (cpp == NULL)
|
|---|
| 3699 | return;
|
|---|
| 3700 | for (i = 0; cpp[i] != NULL; ++i)
|
|---|
| 3701 | {
|
|---|
| 3702 | free (cpp[i]);
|
|---|
| 3703 | cpp[i] = NULL;
|
|---|
| 3704 | }
|
|---|
| 3705 | }
|
|---|
| 3706 |
|
|---|
| 3707 | static char **
|
|---|
| 3708 | enlist (char **cpp, char *new, size_t len)
|
|---|
| 3709 | {
|
|---|
| 3710 | size_t i, j;
|
|---|
| 3711 |
|
|---|
| 3712 | if (cpp == NULL)
|
|---|
| 3713 | return NULL;
|
|---|
| 3714 | if ((new = icpyalloc (new)) == NULL)
|
|---|
| 3715 | {
|
|---|
| 3716 | freelist (cpp);
|
|---|
| 3717 | return NULL;
|
|---|
| 3718 | }
|
|---|
| 3719 | new[len] = '\0';
|
|---|
| 3720 | /* Is there already something in the list that's new (or longer)? */
|
|---|
| 3721 | for (i = 0; cpp[i] != NULL; ++i)
|
|---|
| 3722 | if (istrstr (cpp[i], new) != NULL)
|
|---|
| 3723 | {
|
|---|
| 3724 | free (new);
|
|---|
| 3725 | return cpp;
|
|---|
| 3726 | }
|
|---|
| 3727 | /* Eliminate any obsoleted strings. */
|
|---|
| 3728 | j = 0;
|
|---|
| 3729 | while (cpp[j] != NULL)
|
|---|
| 3730 | if (istrstr (new, cpp[j]) == NULL)
|
|---|
| 3731 | ++j;
|
|---|
| 3732 | else
|
|---|
| 3733 | {
|
|---|
| 3734 | free (cpp[j]);
|
|---|
| 3735 | if (--i == j)
|
|---|
| 3736 | break;
|
|---|
| 3737 | cpp[j] = cpp[i];
|
|---|
| 3738 | cpp[i] = NULL;
|
|---|
| 3739 | }
|
|---|
| 3740 | /* Add the new string. */
|
|---|
| 3741 | REALLOC (cpp, i + 2);
|
|---|
| 3742 | cpp[i] = new;
|
|---|
| 3743 | cpp[i + 1] = NULL;
|
|---|
| 3744 | return cpp;
|
|---|
| 3745 | }
|
|---|
| 3746 |
|
|---|
| 3747 | /* Given pointers to two strings, return a pointer to an allocated
|
|---|
| 3748 | list of their distinct common substrings. Return NULL if something
|
|---|
| 3749 | seems wild. */
|
|---|
| 3750 | static char **
|
|---|
| 3751 | comsubs (char *left, char const *right)
|
|---|
| 3752 | {
|
|---|
| 3753 | char **cpp;
|
|---|
| 3754 | char *lcp;
|
|---|
| 3755 | char *rcp;
|
|---|
| 3756 | size_t i, len;
|
|---|
| 3757 |
|
|---|
| 3758 | if (left == NULL || right == NULL)
|
|---|
| 3759 | return NULL;
|
|---|
| 3760 | cpp = malloc (sizeof *cpp);
|
|---|
| 3761 | if (cpp == NULL)
|
|---|
| 3762 | return NULL;
|
|---|
| 3763 | cpp[0] = NULL;
|
|---|
| 3764 | for (lcp = left; *lcp != '\0'; ++lcp)
|
|---|
| 3765 | {
|
|---|
| 3766 | len = 0;
|
|---|
| 3767 | rcp = strchr (right, *lcp);
|
|---|
| 3768 | while (rcp != NULL)
|
|---|
| 3769 | {
|
|---|
| 3770 | for (i = 1; lcp[i] != '\0' && lcp[i] == rcp[i]; ++i)
|
|---|
| 3771 | continue;
|
|---|
| 3772 | if (i > len)
|
|---|
| 3773 | len = i;
|
|---|
| 3774 | rcp = strchr (rcp + 1, *lcp);
|
|---|
| 3775 | }
|
|---|
| 3776 | if (len == 0)
|
|---|
| 3777 | continue;
|
|---|
| 3778 | {
|
|---|
| 3779 | char **p = enlist (cpp, lcp, len);
|
|---|
| 3780 | if (p == NULL)
|
|---|
| 3781 | {
|
|---|
| 3782 | freelist (cpp);
|
|---|
| 3783 | cpp = NULL;
|
|---|
| 3784 | break;
|
|---|
| 3785 | }
|
|---|
| 3786 | cpp = p;
|
|---|
| 3787 | }
|
|---|
| 3788 | }
|
|---|
| 3789 | return cpp;
|
|---|
| 3790 | }
|
|---|
| 3791 |
|
|---|
| 3792 | static char **
|
|---|
| 3793 | addlists (char **old, char **new)
|
|---|
| 3794 | {
|
|---|
| 3795 | size_t i;
|
|---|
| 3796 |
|
|---|
| 3797 | if (old == NULL || new == NULL)
|
|---|
| 3798 | return NULL;
|
|---|
| 3799 | for (i = 0; new[i] != NULL; ++i)
|
|---|
| 3800 | {
|
|---|
| 3801 | old = enlist (old, new[i], strlen (new[i]));
|
|---|
| 3802 | if (old == NULL)
|
|---|
| 3803 | break;
|
|---|
| 3804 | }
|
|---|
| 3805 | return old;
|
|---|
| 3806 | }
|
|---|
| 3807 |
|
|---|
| 3808 | /* Given two lists of substrings, return a new list giving substrings
|
|---|
| 3809 | common to both. */
|
|---|
| 3810 | static char **
|
|---|
| 3811 | inboth (char **left, char **right)
|
|---|
| 3812 | {
|
|---|
| 3813 | char **both;
|
|---|
| 3814 | char **temp;
|
|---|
| 3815 | size_t lnum, rnum;
|
|---|
| 3816 |
|
|---|
| 3817 | if (left == NULL || right == NULL)
|
|---|
| 3818 | return NULL;
|
|---|
| 3819 | both = malloc (sizeof *both);
|
|---|
| 3820 | if (both == NULL)
|
|---|
| 3821 | return NULL;
|
|---|
| 3822 | both[0] = NULL;
|
|---|
| 3823 | for (lnum = 0; left[lnum] != NULL; ++lnum)
|
|---|
| 3824 | {
|
|---|
| 3825 | for (rnum = 0; right[rnum] != NULL; ++rnum)
|
|---|
| 3826 | {
|
|---|
| 3827 | temp = comsubs (left[lnum], right[rnum]);
|
|---|
| 3828 | if (temp == NULL)
|
|---|
| 3829 | {
|
|---|
| 3830 | freelist (both);
|
|---|
| 3831 | return NULL;
|
|---|
| 3832 | }
|
|---|
| 3833 | both = addlists (both, temp);
|
|---|
| 3834 | freelist (temp);
|
|---|
| 3835 | free (temp);
|
|---|
| 3836 | if (both == NULL)
|
|---|
| 3837 | return NULL;
|
|---|
| 3838 | }
|
|---|
| 3839 | }
|
|---|
| 3840 | return both;
|
|---|
| 3841 | }
|
|---|
| 3842 |
|
|---|
| 3843 | typedef struct
|
|---|
| 3844 | {
|
|---|
| 3845 | char **in;
|
|---|
| 3846 | char *left;
|
|---|
| 3847 | char *right;
|
|---|
| 3848 | char *is;
|
|---|
| 3849 | } must;
|
|---|
| 3850 |
|
|---|
| 3851 | static void
|
|---|
| 3852 | resetmust (must * mp)
|
|---|
| 3853 | {
|
|---|
| 3854 | mp->left[0] = mp->right[0] = mp->is[0] = '\0';
|
|---|
| 3855 | freelist (mp->in);
|
|---|
| 3856 | }
|
|---|
| 3857 |
|
|---|
| 3858 | static void
|
|---|
| 3859 | dfamust (struct dfa *d)
|
|---|
| 3860 | {
|
|---|
| 3861 | must *musts;
|
|---|
| 3862 | must *mp;
|
|---|
| 3863 | char *result;
|
|---|
| 3864 | size_t ri;
|
|---|
| 3865 | size_t i;
|
|---|
| 3866 | int exact;
|
|---|
| 3867 | token t;
|
|---|
| 3868 | static must must0;
|
|---|
| 3869 | struct dfamust *dm;
|
|---|
| 3870 | static char empty_string[] = "";
|
|---|
| 3871 |
|
|---|
| 3872 | result = empty_string;
|
|---|
| 3873 | exact = 0;
|
|---|
| 3874 | MALLOC (musts, d->tindex + 1);
|
|---|
| 3875 | mp = musts;
|
|---|
| 3876 | for (i = 0; i <= d->tindex; ++i)
|
|---|
| 3877 | mp[i] = must0;
|
|---|
| 3878 | for (i = 0; i <= d->tindex; ++i)
|
|---|
| 3879 | {
|
|---|
| 3880 | mp[i].in = xmalloc (sizeof *mp[i].in);
|
|---|
| 3881 | mp[i].left = xmalloc (2);
|
|---|
| 3882 | mp[i].right = xmalloc (2);
|
|---|
| 3883 | mp[i].is = xmalloc (2);
|
|---|
| 3884 | mp[i].left[0] = mp[i].right[0] = mp[i].is[0] = '\0';
|
|---|
| 3885 | mp[i].in[0] = NULL;
|
|---|
| 3886 | }
|
|---|
| 3887 | #ifdef DEBUG
|
|---|
| 3888 | fprintf (stderr, "dfamust:\n");
|
|---|
| 3889 | for (i = 0; i < d->tindex; ++i)
|
|---|
| 3890 | {
|
|---|
| 3891 | fprintf (stderr, " %zd:", i);
|
|---|
| 3892 | prtok (d->tokens[i]);
|
|---|
| 3893 | }
|
|---|
| 3894 | putc ('\n', stderr);
|
|---|
| 3895 | #endif
|
|---|
| 3896 | for (ri = 0; ri < d->tindex; ++ri)
|
|---|
| 3897 | {
|
|---|
| 3898 | switch (t = d->tokens[ri])
|
|---|
| 3899 | {
|
|---|
| 3900 | case LPAREN:
|
|---|
| 3901 | case RPAREN:
|
|---|
| 3902 | assert (!"neither LPAREN nor RPAREN may appear here");
|
|---|
| 3903 | case EMPTY:
|
|---|
| 3904 | case BEGLINE:
|
|---|
| 3905 | case ENDLINE:
|
|---|
| 3906 | case BEGWORD:
|
|---|
| 3907 | case ENDWORD:
|
|---|
| 3908 | case LIMWORD:
|
|---|
| 3909 | case NOTLIMWORD:
|
|---|
| 3910 | case BACKREF:
|
|---|
| 3911 | resetmust (mp);
|
|---|
| 3912 | break;
|
|---|
| 3913 | case STAR:
|
|---|
| 3914 | case QMARK:
|
|---|
| 3915 | assert (musts < mp);
|
|---|
| 3916 | --mp;
|
|---|
| 3917 | resetmust (mp);
|
|---|
| 3918 | break;
|
|---|
| 3919 | case OR:
|
|---|
| 3920 | assert (&musts[2] <= mp);
|
|---|
| 3921 | {
|
|---|
| 3922 | char **new;
|
|---|
| 3923 | must *lmp;
|
|---|
| 3924 | must *rmp;
|
|---|
| 3925 | size_t j, ln, rn, n;
|
|---|
| 3926 |
|
|---|
| 3927 | rmp = --mp;
|
|---|
| 3928 | lmp = --mp;
|
|---|
| 3929 | /* Guaranteed to be. Unlikely, but. . . */
|
|---|
| 3930 | if (!STREQ (lmp->is, rmp->is))
|
|---|
| 3931 | lmp->is[0] = '\0';
|
|---|
| 3932 | /* Left side--easy */
|
|---|
| 3933 | i = 0;
|
|---|
| 3934 | while (lmp->left[i] != '\0' && lmp->left[i] == rmp->left[i])
|
|---|
| 3935 | ++i;
|
|---|
| 3936 | lmp->left[i] = '\0';
|
|---|
| 3937 | /* Right side */
|
|---|
| 3938 | ln = strlen (lmp->right);
|
|---|
| 3939 | rn = strlen (rmp->right);
|
|---|
| 3940 | n = ln;
|
|---|
| 3941 | if (n > rn)
|
|---|
| 3942 | n = rn;
|
|---|
| 3943 | for (i = 0; i < n; ++i)
|
|---|
| 3944 | if (lmp->right[ln - i - 1] != rmp->right[rn - i - 1])
|
|---|
| 3945 | break;
|
|---|
| 3946 | for (j = 0; j < i; ++j)
|
|---|
| 3947 | lmp->right[j] = lmp->right[(ln - i) + j];
|
|---|
| 3948 | lmp->right[j] = '\0';
|
|---|
| 3949 | new = inboth (lmp->in, rmp->in);
|
|---|
| 3950 | if (new == NULL)
|
|---|
| 3951 | goto done;
|
|---|
| 3952 | freelist (lmp->in);
|
|---|
| 3953 | free (lmp->in);
|
|---|
| 3954 | lmp->in = new;
|
|---|
| 3955 | }
|
|---|
| 3956 | break;
|
|---|
| 3957 | case PLUS:
|
|---|
| 3958 | assert (musts < mp);
|
|---|
| 3959 | --mp;
|
|---|
| 3960 | mp->is[0] = '\0';
|
|---|
| 3961 | break;
|
|---|
| 3962 | case END:
|
|---|
| 3963 | assert (mp == &musts[1]);
|
|---|
| 3964 | for (i = 0; musts[0].in[i] != NULL; ++i)
|
|---|
| 3965 | if (strlen (musts[0].in[i]) > strlen (result))
|
|---|
| 3966 | result = musts[0].in[i];
|
|---|
| 3967 | if (STREQ (result, musts[0].is))
|
|---|
| 3968 | exact = 1;
|
|---|
| 3969 | goto done;
|
|---|
| 3970 | case CAT:
|
|---|
| 3971 | assert (&musts[2] <= mp);
|
|---|
| 3972 | {
|
|---|
| 3973 | must *lmp;
|
|---|
| 3974 | must *rmp;
|
|---|
| 3975 |
|
|---|
| 3976 | rmp = --mp;
|
|---|
| 3977 | lmp = --mp;
|
|---|
| 3978 | /* In. Everything in left, plus everything in
|
|---|
| 3979 | right, plus concatenation of
|
|---|
| 3980 | left's right and right's left. */
|
|---|
| 3981 | lmp->in = addlists (lmp->in, rmp->in);
|
|---|
| 3982 | if (lmp->in == NULL)
|
|---|
| 3983 | goto done;
|
|---|
| 3984 | if (lmp->right[0] != '\0' && rmp->left[0] != '\0')
|
|---|
| 3985 | {
|
|---|
| 3986 | char *tp;
|
|---|
| 3987 |
|
|---|
| 3988 | tp = icpyalloc (lmp->right);
|
|---|
| 3989 | tp = icatalloc (tp, rmp->left);
|
|---|
| 3990 | lmp->in = enlist (lmp->in, tp, strlen (tp));
|
|---|
| 3991 | free (tp);
|
|---|
| 3992 | if (lmp->in == NULL)
|
|---|
| 3993 | goto done;
|
|---|
| 3994 | }
|
|---|
| 3995 | /* Left-hand */
|
|---|
| 3996 | if (lmp->is[0] != '\0')
|
|---|
| 3997 | {
|
|---|
| 3998 | lmp->left = icatalloc (lmp->left, rmp->left);
|
|---|
| 3999 | if (lmp->left == NULL)
|
|---|
| 4000 | goto done;
|
|---|
| 4001 | }
|
|---|
| 4002 | /* Right-hand */
|
|---|
| 4003 | if (rmp->is[0] == '\0')
|
|---|
| 4004 | lmp->right[0] = '\0';
|
|---|
| 4005 | lmp->right = icatalloc (lmp->right, rmp->right);
|
|---|
| 4006 | if (lmp->right == NULL)
|
|---|
| 4007 | goto done;
|
|---|
| 4008 | /* Guaranteed to be */
|
|---|
| 4009 | if (lmp->is[0] != '\0' && rmp->is[0] != '\0')
|
|---|
| 4010 | {
|
|---|
| 4011 | lmp->is = icatalloc (lmp->is, rmp->is);
|
|---|
| 4012 | if (lmp->is == NULL)
|
|---|
| 4013 | goto done;
|
|---|
| 4014 | }
|
|---|
| 4015 | else
|
|---|
| 4016 | lmp->is[0] = '\0';
|
|---|
| 4017 | }
|
|---|
| 4018 | break;
|
|---|
| 4019 | default:
|
|---|
| 4020 | if (t < END)
|
|---|
| 4021 | {
|
|---|
| 4022 | assert (!"oops! t >= END");
|
|---|
| 4023 | }
|
|---|
| 4024 | else if (t == '\0')
|
|---|
| 4025 | {
|
|---|
| 4026 | /* not on *my* shift */
|
|---|
| 4027 | goto done;
|
|---|
| 4028 | }
|
|---|
| 4029 | else if (t >= CSET || !MBS_SUPPORT || t == ANYCHAR || t == MBCSET)
|
|---|
| 4030 | {
|
|---|
| 4031 | /* easy enough */
|
|---|
| 4032 | resetmust (mp);
|
|---|
| 4033 | }
|
|---|
| 4034 | else
|
|---|
| 4035 | {
|
|---|
| 4036 | /* plain character */
|
|---|
| 4037 | resetmust (mp);
|
|---|
| 4038 | mp->is[0] = mp->left[0] = mp->right[0] = t;
|
|---|
| 4039 | mp->is[1] = mp->left[1] = mp->right[1] = '\0';
|
|---|
| 4040 | mp->in = enlist (mp->in, mp->is, (size_t) 1);
|
|---|
| 4041 | if (mp->in == NULL)
|
|---|
| 4042 | goto done;
|
|---|
| 4043 | }
|
|---|
| 4044 | break;
|
|---|
| 4045 | }
|
|---|
| 4046 | #ifdef DEBUG
|
|---|
| 4047 | fprintf (stderr, " node: %zd:", ri);
|
|---|
| 4048 | prtok (d->tokens[ri]);
|
|---|
| 4049 | fprintf (stderr, "\n in:");
|
|---|
| 4050 | for (i = 0; mp->in[i]; ++i)
|
|---|
| 4051 | fprintf (stderr, " \"%s\"", mp->in[i]);
|
|---|
| 4052 | fprintf (stderr, "\n is: \"%s\"\n", mp->is);
|
|---|
| 4053 | fprintf (stderr, " left: \"%s\"\n", mp->left);
|
|---|
| 4054 | fprintf (stderr, " right: \"%s\"\n", mp->right);
|
|---|
| 4055 | #endif
|
|---|
| 4056 | ++mp;
|
|---|
| 4057 | }
|
|---|
| 4058 | done:
|
|---|
| 4059 | if (strlen (result))
|
|---|
| 4060 | {
|
|---|
| 4061 | MALLOC (dm, 1);
|
|---|
| 4062 | dm->exact = exact;
|
|---|
| 4063 | dm->must = xmemdup (result, strlen (result) + 1);
|
|---|
| 4064 | dm->next = d->musts;
|
|---|
| 4065 | d->musts = dm;
|
|---|
| 4066 | }
|
|---|
| 4067 | mp = musts;
|
|---|
| 4068 | for (i = 0; i <= d->tindex; ++i)
|
|---|
| 4069 | {
|
|---|
| 4070 | freelist (mp[i].in);
|
|---|
| 4071 | free (mp[i].in);
|
|---|
| 4072 | free (mp[i].left);
|
|---|
| 4073 | free (mp[i].right);
|
|---|
| 4074 | free (mp[i].is);
|
|---|
| 4075 | }
|
|---|
| 4076 | free (mp);
|
|---|
| 4077 | }
|
|---|
| 4078 |
|
|---|
| 4079 | struct dfa *
|
|---|
| 4080 | dfaalloc (void)
|
|---|
| 4081 | {
|
|---|
| 4082 | return xmalloc (sizeof (struct dfa));
|
|---|
| 4083 | }
|
|---|
| 4084 |
|
|---|
| 4085 | struct dfamust *_GL_ATTRIBUTE_PURE
|
|---|
| 4086 | dfamusts (struct dfa const *d)
|
|---|
| 4087 | {
|
|---|
| 4088 | return d->musts;
|
|---|
| 4089 | }
|
|---|
| 4090 |
|
|---|
| 4091 | /* vim:set shiftwidth=2: */
|
|---|