X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regcomp.h;h=3aa5c1e18bb66bbc533b8b477b8f1d421f64f76c;hb=d6d3e8bddad8c105fc1972d4d9a8298ad3f73f11;hp=be6e5b1c33c5c05cf40b04bc6031b4a2cf43b5be;hpb=e91177edb0beae74fd07fb1ec08a528ab1fb63c2;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regcomp.h b/regcomp.h index be6e5b1..3aa5c1e 100644 --- a/regcomp.h +++ b/regcomp.h @@ -1,4 +1,11 @@ /* regcomp.h + * + * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, + * 2000, 2001, 2002, 2003, by Larry Wall and others + * + * You may distribute under the terms of either the GNU General Public + * License or the Artistic License, as specified in the README file. + * */ typedef OP OP_4tree; /* Will be redefined later. */ @@ -49,174 +56,6 @@ typedef OP OP_4tree; /* Will be redefined later. */ * to the thing following the set of BRANCHes.) The opcodes are: */ -/* definition number opnd? meaning */ -#define END 0 /* no End of program. */ -#define BOL 1 /* no Match "" at beginning of line. */ -#define MBOL 2 /* no Same, assuming multiline. */ -#define SBOL 3 /* no Same, assuming singleline. */ -#define EOL 4 /* no Match "" at end of line. */ -#define MEOL 5 /* no Same, assuming multiline. */ -#define SEOL 6 /* no Same, assuming singleline. */ -#define ANY 7 /* no Match any one character (except newline). */ -#define SANY 8 /* no Match any one character. */ -#define ANYOF 9 /* sv Match character in (or not in) this class. */ -#define CURLY 10 /* sv Match this simple thing {n,m} times. */ -#define CURLYX 11 /* sv Match this complex thing {n,m} times. */ -#define BRANCH 12 /* node Match this alternative, or the next... */ -#define BACK 13 /* no Match "", "next" ptr points backward. */ -#define EXACT 14 /* sv Match this string (preceded by length). */ -#define EXACTF 15 /* sv Match this string, folded (prec. by length). */ -#define EXACTFL 16 /* sv Match this string, folded in locale (w/len). */ -#define NOTHING 17 /* no Match empty string. */ -#define STAR 18 /* node Match this (simple) thing 0 or more times. */ -#define PLUS 19 /* node Match this (simple) thing 1 or more times. */ -#define BOUND 20 /* no Match "" at any word boundary */ -#define BOUNDL 21 /* no Match "" at any word boundary */ -#define NBOUND 22 /* no Match "" at any word non-boundary */ -#define NBOUNDL 23 /* no Match "" at any word non-boundary */ -#define REF 24 /* num Match some already matched string */ -#define OPEN 25 /* num Mark this point in input as start of #n. */ -#define CLOSE 26 /* num Analogous to OPEN. */ -#define MINMOD 27 /* no Next operator is not greedy. */ -#define GPOS 28 /* no Matches where last m//g left off. */ -#define IFMATCH 29 /* off Succeeds if the following matches. */ -#define UNLESSM 30 /* off Fails if the following matches. */ -#define SUCCEED 31 /* no Return from a subroutine, basically. */ -#define WHILEM 32 /* no Do curly processing and see if rest matches. */ -#define ALNUM 33 /* no Match any alphanumeric character */ -#define ALNUML 34 /* no Match any alphanumeric char in locale */ -#define NALNUM 35 /* no Match any non-alphanumeric character */ -#define NALNUML 36 /* no Match any non-alphanumeric char in locale */ -#define SPACE 37 /* no Match any whitespace character */ -#define SPACEL 38 /* no Match any whitespace char in locale */ -#define NSPACE 39 /* no Match any non-whitespace character */ -#define NSPACEL 40 /* no Match any non-whitespace char in locale */ -#define DIGIT 41 /* no Match any numeric character */ -#define NDIGIT 42 /* no Match any non-numeric character */ -#define CURLYM 43 /* no Match this medium-complex thing {n,m} times. */ -#define CURLYN 44 /* no Match next-after-this simple thing - {n,m} times, set parenths. */ -#define TAIL 45 /* no Match empty string. Can jump here from outside. */ -#define REFF 46 /* num Match already matched string, folded */ -#define REFFL 47 /* num Match already matched string, folded in loc. */ -#define EVAL 48 /* evl Execute some Perl code. */ -#define LONGJMP 49 /* off Jump far away. */ -#define BRANCHJ 50 /* off BRANCH with long offset. */ -#define IFTHEN 51 /* off Switch, should be preceeded by switcher . */ -#define GROUPP 52 /* num Whether the group matched. */ -#define LOGICAL 53 /* no Next opcode should set the flag only. */ -#define SUSPEND 54 /* off "Independent" sub-RE. */ -#define RENUM 55 /* off Group with independently numbered parens. */ -#define OPTIMIZED 56 /* off Placeholder for dump. */ - -/* - * Opcode notes: - * - * BRANCH The set of branches constituting a single choice are hooked - * together with their "next" pointers, since precedence prevents - * anything being concatenated to any individual branch. The - * "next" pointer of the last BRANCH in a choice points to the - * thing following the whole choice. This is also where the - * final "next" pointer of each individual branch points; each - * branch starts with the operand node of a BRANCH node. - * - * BACK Normal "next" pointers all implicitly point forward; BACK - * exists to make loop structures possible. - * - * STAR,PLUS '?', and complex '*' and '+', are implemented as circular - * BRANCH structures using BACK. Simple cases (one character - * per match) are implemented with STAR and PLUS for speed - * and to minimize recursive plunges. - * - * OPEN,CLOSE,GROUPP ...are numbered at compile time. - */ - -#ifndef DOINIT -EXTCONST U8 regkind[]; -#else -EXTCONST U8 regkind[] = { - END, - BOL, - BOL, - BOL, - EOL, - EOL, - EOL, - ANY, - ANY, - ANYOF, - CURLY, - CURLY, - BRANCH, - BACK, - EXACT, - EXACT, - EXACT, - NOTHING, - STAR, - PLUS, - BOUND, - BOUND, - NBOUND, - NBOUND, - REF, - OPEN, - CLOSE, - MINMOD, - GPOS, - BRANCHJ, - BRANCHJ, - END, - WHILEM, - ALNUM, - ALNUM, - NALNUM, - NALNUM, - SPACE, - SPACE, - NSPACE, - NSPACE, - DIGIT, - NDIGIT, - CURLY, - CURLY, - NOTHING, - REF, - REF, - EVAL, - LONGJMP, - BRANCHJ, - BRANCHJ, - GROUPP, - LOGICAL, - BRANCHJ, - BRANCHJ, - NOTHING, -}; -#endif - -/* The following have no fixed length. char* since we do strchr on it. */ -#ifndef DOINIT -EXTCONST char varies[]; -#else -EXTCONST char varies[] = { - BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL, - WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, 0 -}; -#endif - -/* The following always have a length of 1. char* since we do strchr on it. */ -#ifndef DOINIT -EXTCONST char simple[]; -#else -EXTCONST char simple[] = { - ANY, SANY, ANYOF, - ALNUM, ALNUML, NALNUM, NALNUML, - SPACE, SPACEL, NSPACE, NSPACEL, - DIGIT, NDIGIT, 0 -}; -#endif - /* * A node is one char of opcode followed by two chars of "next" pointer. * "Next" pointers are stored as two 8-bit pieces, high order first. The @@ -234,10 +73,10 @@ EXTCONST char simple[] = { */ struct regnode_string { - U8 flags; + U8 str_len; U8 type; U16 next_off; - U8 string[1]; + char string[1]; }; struct regnode_1 { @@ -255,6 +94,26 @@ struct regnode_2 { U16 arg2; }; +#define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */ +#define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */ + +struct regnode_charclass { + U8 flags; + U8 type; + U16 next_off; + U32 arg1; + char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */ +}; + +struct regnode_charclass_class { /* has [[:blah:]] classes */ + U8 flags; /* should have ANYOF_CLASS here */ + U8 type; + U16 next_off; + U32 arg1; + char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */ + char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */ +}; + /* XXX fix this description. Impose a limit of REG_INFTY on various pattern matching operations to limit stack growth and to avoid "infinite" recursions. @@ -280,6 +139,10 @@ struct regnode_2 { #define ARG_VALUE(arg) (arg) #define ARG__SET(arg,val) ((arg) = (val)) +#undef ARG +#undef ARG1 +#undef ARG2 + #define ARG(p) ARG_VALUE(ARG_LOC(p)) #define ARG1(p) ARG_VALUE(ARG1_LOC(p)) #define ARG2(p) ARG_VALUE(ARG2_LOC(p)) @@ -287,6 +150,9 @@ struct regnode_2 { #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val)) #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val)) +#undef NEXT_OFF +#undef NODE_ALIGN + #ifndef lint # define NEXT_OFF(p) ((p)->next_off) # define NODE_ALIGN(node) @@ -299,8 +165,24 @@ struct regnode_2 { #define SIZE_ALIGN NODE_ALIGN +#undef OP +#undef OPERAND +#undef MASK +#undef STRING + #define OP(p) ((p)->type) #define OPERAND(p) (((struct regnode_string *)p)->string) +#define MASK(p) ((char*)OPERAND(p)) +#define STR_LEN(p) (((struct regnode_string *)p)->str_len) +#define STRING(p) (((struct regnode_string *)p)->string) +#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode)) +#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1) + +#undef NODE_ALIGN +#undef ARG_LOC +#undef NEXTOPER +#undef PREVOPER + #define NODE_ALIGN(node) #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1) #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1) @@ -318,85 +200,209 @@ struct regnode_2 { #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \ ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END -#define MAGIC 0234 - -#define SIZE_ONLY (regcode == ®dummy) - -/* Flags for first parameter byte of ANYOF */ -#define ANYOF_INVERT 0x40 -#define ANYOF_FOLD 0x20 -#define ANYOF_LOCALE 0x10 -#define ANYOF_ISA 0x0F -#define ANYOF_ALNUML 0x08 -#define ANYOF_NALNUML 0x04 -#define ANYOF_SPACEL 0x02 -#define ANYOF_NSPACEL 0x01 - -/* Utility macros for bitmap of ANYOF */ -#define ANYOF_BYTE(p,c) (p)[1 + (((c) >> 3) & 31)] -#define ANYOF_BIT(c) (1 << ((c) & 7)) -#define ANYOF_SET(p,c) (ANYOF_BYTE(p,c) |= ANYOF_BIT(c)) -#define ANYOF_CLEAR(p,c) (ANYOF_BYTE(p,c) &= ~ANYOF_BIT(c)) -#define ANYOF_TEST(p,c) (ANYOF_BYTE(p,c) & ANYOF_BIT(c)) - -#define ANY_SKIP ((33 - 1)/sizeof(regnode) + 1) +#define REG_MAGIC 0234 + +#define SIZE_ONLY (RExC_emit == &PL_regdummy) + +/* Flags for node->flags of ANYOF */ + +#define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */ +#define ANYOF_INVERT 0x04 +#define ANYOF_FOLD 0x02 +#define ANYOF_LOCALE 0x01 + +/* Used for regstclass only */ +#define ANYOF_EOS 0x10 /* Can match an empty string too */ + +/* There is a character or a range past 0xff */ +#define ANYOF_UNICODE 0x20 +#define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */ + +/* size of node is large (includes class pointer) */ +#define ANYOF_LARGE 0x80 + +/* Are there any runtime flags on in this node? */ +#define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f) + +#define ANYOF_FLAGS_ALL 0xff + +/* Character classes for node->classflags of ANYOF */ +/* Should be synchronized with a table in regprop() */ +/* 2n should pair with 2n+1 */ + +#define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */ +#define ANYOF_NALNUM 1 +#define ANYOF_SPACE 2 /* \s */ +#define ANYOF_NSPACE 3 +#define ANYOF_DIGIT 4 +#define ANYOF_NDIGIT 5 +#define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */ +#define ANYOF_NALNUMC 7 +#define ANYOF_ALPHA 8 +#define ANYOF_NALPHA 9 +#define ANYOF_ASCII 10 +#define ANYOF_NASCII 11 +#define ANYOF_CNTRL 12 +#define ANYOF_NCNTRL 13 +#define ANYOF_GRAPH 14 +#define ANYOF_NGRAPH 15 +#define ANYOF_LOWER 16 +#define ANYOF_NLOWER 17 +#define ANYOF_PRINT 18 +#define ANYOF_NPRINT 19 +#define ANYOF_PUNCT 20 +#define ANYOF_NPUNCT 21 +#define ANYOF_UPPER 22 +#define ANYOF_NUPPER 23 +#define ANYOF_XDIGIT 24 +#define ANYOF_NXDIGIT 25 +#define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */ +#define ANYOF_NPSXSPC 27 +#define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */ +#define ANYOF_NBLANK 29 + +#define ANYOF_MAX 32 + +/* Backward source code compatibility. */ + +#define ANYOF_ALNUML ANYOF_ALNUM +#define ANYOF_NALNUML ANYOF_NALNUM +#define ANYOF_SPACEL ANYOF_SPACE +#define ANYOF_NSPACEL ANYOF_NSPACE + +/* Utility macros for the bitmap and classes of ANYOF */ + +#define ANYOF_SIZE (sizeof(struct regnode_charclass)) +#define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class)) + +#define ANYOF_FLAGS(p) ((p)->flags) + +#define ANYOF_BIT(c) (1 << ((c) & 7)) + +#define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3]) +#define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c)) +#define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c)) +#define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c)) + +#define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char) +#define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char) + +#define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap) +#define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[((c) >> 3) & 31]) +#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c)) +#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c)) +#define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c)) + +#define ANYOF_BITMAP_SETALL(p) \ + memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE) +#define ANYOF_BITMAP_CLEARALL(p) \ + Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE) +/* Check that all 256 bits are all set. Used in S_cl_is_anything() */ +#define ANYOF_BITMAP_TESTALLSET(p) \ + memEQ (ANYOF_BITMAP(p), "\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377", ANYOF_BITMAP_SIZE) + +#define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode)) +#define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode)) +#define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP) /* * Utility definitions. */ #ifndef lint #ifndef CHARMASK -#define UCHARAT(p) ((int)*(unsigned char *)(p)) +#define UCHARAT(p) ((int)*(U8*)(p)) #else #define UCHARAT(p) ((int)*(p)&CHARMASK) #endif #else /* lint */ -#define UCHARAT(p) regdummy +#define UCHARAT(p) PL_regdummy #endif /* lint */ -#define FAIL(m) croak ("/%.127s/: %s", regprecomp,m) -#define FAIL2(pat,m) re_croak2("/%.127s/: ",pat,regprecomp,m) - #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode)) -#ifdef REG_COMP_C -const static U8 regarglen[] = { - 0,0,0,0,0,0,0,0,0,0, - /*CURLY*/ EXTRA_SIZE(struct regnode_2), - /*CURLYX*/ EXTRA_SIZE(struct regnode_2), - 0,0,0,0,0,0,0,0,0,0,0,0, - /*REF*/ EXTRA_SIZE(struct regnode_1), - /*OPEN*/ EXTRA_SIZE(struct regnode_1), - /*CLOSE*/ EXTRA_SIZE(struct regnode_1), - 0,0, - /*IFMATCH*/ EXTRA_SIZE(struct regnode_1), - /*UNLESSM*/ EXTRA_SIZE(struct regnode_1), - 0,0,0,0,0,0,0,0,0,0,0,0, - /*CURLYM*/ EXTRA_SIZE(struct regnode_2), - /*CURLYN*/ EXTRA_SIZE(struct regnode_2), - 0, - /*REFF*/ EXTRA_SIZE(struct regnode_1), - /*REFFL*/ EXTRA_SIZE(struct regnode_1), - /*EVAL*/ EXTRA_SIZE(struct regnode_1), - /*LONGJMP*/ EXTRA_SIZE(struct regnode_1), - /*BRANCHJ*/ EXTRA_SIZE(struct regnode_1), - /*IFTHEN*/ EXTRA_SIZE(struct regnode_1), - /*GROUPP*/ EXTRA_SIZE(struct regnode_1), - /*LOGICAL*/ 0, - /*SUSPEND*/ EXTRA_SIZE(struct regnode_1), - /*RENUM*/ EXTRA_SIZE(struct regnode_1), 0, +#define REG_SEEN_ZERO_LEN 1 +#define REG_SEEN_LOOKBEHIND 2 +#define REG_SEEN_GPOS 4 +#define REG_SEEN_EVAL 8 +#define REG_SEEN_CANY 16 +#define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */ + +START_EXTERN_C + +#include "regnodes.h" + +/* The following have no fixed length. U8 so we can do strchr() on it. */ +#ifndef DOINIT +EXTCONST U8 PL_varies[]; +#else +EXTCONST U8 PL_varies[] = { + BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL, + WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP, 0 }; +#endif -const static char reg_off_by_arg[] = { - 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0 .. 15 */ - 0,0,0,0,0,0,0,0,0,0,0,0,0, /*IFMATCH*/ 2, /*UNLESSM*/ 2, 0, - 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 32 .. 47 */ - 0, /*LONGJMP*/ 1, /*BRANCHJ*/ 1, /*IFTHEN*/ 1, 0, 0, - /*RENUM*/ 1, /*RENUM*/ 1,0, +/* The following always have a length of 1. U8 we can do strchr() on it. */ +/* (Note that length 1 means "one character" under UTF8, not "one octet".) */ +#ifndef DOINIT +EXTCONST U8 PL_simple[]; +#else +EXTCONST U8 PL_simple[] = { + REG_ANY, SANY, CANY, + ANYOF, + ALNUM, ALNUML, + NALNUM, NALNUML, + SPACE, SPACEL, + NSPACE, NSPACEL, + DIGIT, NDIGIT, + 0 }; #endif -#define REG_SEEN_ZERO_LEN 1 -#define REG_SEEN_LOOKBEHIND 2 -#define REG_SEEN_GPOS 4 -#define REG_SEEN_EVAL 8 +END_EXTERN_C + +typedef struct re_scream_pos_data_s +{ + char **scream_olds; /* match pos */ + I32 *scream_pos; /* Internal iterator of scream. */ +} re_scream_pos_data; + +/* .what is a character array with one character for each member of .data + * The character describes the function of the corresponding .data item: + * f - start-class data for regstclass optimization + * n - Root of op tree for (?{EVAL}) item + * o - Start op for (?{EVAL}) item + * p - Pad for (?{EVAL} item + * s - swash for unicode-style character class, and the multicharacter + * strings resulting from casefolding the single-character entries + * in the character class + * 20010712 mjd@plover.com + * (Remember to update re_dup() and pregfree() if you add any items.) + */ +struct reg_data { + U32 count; + U8 *what; + void* data[1]; +}; + +struct reg_substr_datum { + I32 min_offset; + I32 max_offset; + SV *substr; /* non-utf8 variant */ + SV *utf8_substr; /* utf8 variant */ +}; + +struct reg_substr_data { + struct reg_substr_datum data[3]; /* Actual array */ +}; + +#define anchored_substr substrs->data[0].substr +#define anchored_utf8 substrs->data[0].utf8_substr +#define anchored_offset substrs->data[0].min_offset +#define float_substr substrs->data[1].substr +#define float_utf8 substrs->data[1].utf8_substr +#define float_min_offset substrs->data[1].min_offset +#define float_max_offset substrs->data[1].max_offset +#define check_substr substrs->data[2].substr +#define check_utf8 substrs->data[2].utf8_substr +#define check_offset_min substrs->data[2].min_offset +#define check_offset_max substrs->data[2].max_offset