X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/a687059cbaf2c6fdccb5e0fae2aee80ec15625a8..389b72ffba57effeed96b76240b691bbd83408f7:/regcomp.h diff --git a/regcomp.h b/regcomp.h index c4c6520..e30c8e7 100644 --- a/regcomp.h +++ b/regcomp.h @@ -1,30 +1,27 @@ -/* $Header: regcomp.h,v 3.0 89/10/18 15:22:39 lwall Locked $ - * - * $Log: regcomp.h,v $ - * Revision 3.0 89/10/18 15:22:39 lwall - * 3.0 baseline - * +/* regcomp.h */ +typedef OP OP_4tree; /* Will be redefined later. */ + /* * The "internal use only" fields in regexp.h are present to pass info from * compile to execute that permits the execute phase to run lots faster on * simple cases. They are: * - * regstart str that must begin a match; Nullch if none obvious + * regstart sv that must begin a match; Nullch if none obvious * reganch is the match anchored (at beginning-of-line only)? * regmust string (pointer into program) that match must include, or NULL - * [regmust changed to STR* for bminstr()--law] + * [regmust changed to SV* for bminstr()--law] * regmlen length of regmust string * [regmlen not used currently] * * Regstart and reganch permit very fast decisions on suitable starting points * for a match, cutting down the work a lot. Regmust permits fast rejection * of lines that cannot possibly match. The regmust tests are costly enough - * that regcomp() supplies a regmust only if the r.e. contains something + * that pregcomp() supplies a regmust only if the r.e. contains something * potentially expensive (at present, the only such thing detected is * or + * at the start of the r.e., which can involve a lot of backup). Regmlen is - * supplied because the test in regexec() needs it and regcomp() is computing + * supplied because the test in pregexec() needs it and pregcomp() is computing * it anyway. * [regmust is now supplied always. The tests that use regmust have a * heuristic that disables the test if it usually matches.] @@ -52,72 +49,6 @@ * 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 EOL 2 /* no Match "" at end of line. */ -#define ANY 3 /* no Match any one character. */ -#define ANYOF 4 /* str Match any character in this string. */ -#define ANYBUT 5 /* str Match any character not in this string. */ -#define BRANCH 6 /* node Match this alternative, or the next... */ -#define BACK 7 /* no Match "", "next" ptr points backward. */ -#define EXACTLY 8 /* str Match this string (preceded by length). */ -#define NOTHING 9 /* no Match empty string. */ -#define STAR 10 /* node Match this (simple) thing 0 or more times. */ -#define PLUS 11 /* node Match this (simple) thing 1 or more times. */ -#define ALNUM 12 /* no Match any alphanumeric character */ -#define NALNUM 13 /* no Match any non-alphanumeric character */ -#define BOUND 14 /* no Match "" at any word boundary */ -#define NBOUND 15 /* no Match "" at any word non-boundary */ -#define SPACE 16 /* no Match any whitespace character */ -#define NSPACE 17 /* no Match any non-whitespace character */ -#define DIGIT 18 /* no Match any numeric character */ -#define NDIGIT 19 /* no Match any non-numeric character */ -#define REF 20 /* no Match some already matched string */ -#define OPEN 30 /* no Mark this point in input as start of #n. */ - /* OPEN+1 is number 1, etc. */ -#define CLOSE 40 /* no Analogous to OPEN. */ -/* CLOSE must be last one! see regmust finder */ - -/* - * 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 ...are numbered at compile time. - */ - -/* The following have no fixed length. */ -#ifndef DOINIT -extern char varies[]; -#else -char varies[] = {BRANCH,BACK,STAR,PLUS, - REF+1,REF+2,REF+3,REF+4,REF+5,REF+6,REF+7,REF+8,REF+9,0}; -#endif - -/* The following always have a length of 1. */ -#ifndef DOINIT -extern char simple[]; -#else -char simple[] = {ANY,ANYOF,ANYBUT,ALNUM,NALNUM,SPACE,NSPACE,DIGIT,NDIGIT,0}; -#endif - -EXT char regdummy; - /* * 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 @@ -128,58 +59,270 @@ EXT char regdummy; * Using two bytes for the "next" pointer is vast overkill for most things, * but allows patterns to get big without disasters. * - * [If REGALIGN is defined, the "next" pointer is always aligned on an even + * [The "next" pointer is always aligned on an even * boundary, and reads the offset directly as a short. Also, there is no * special test to reverse the sign of BACK pointers since the offset is * stored negative.] */ -#ifndef gould -#ifndef cray -#define REGALIGN +struct regnode_string { + U8 str_len; + U8 type; + U16 next_off; + char string[1]; +}; + +struct regnode_1 { + U8 flags; + U8 type; + U16 next_off; + U32 arg1; +}; + +struct regnode_2 { + U8 flags; + U8 type; + U16 next_off; + U16 arg1; + U16 arg2; +}; + +/* XXX fix this description. + Impose a limit of REG_INFTY on various pattern matching operations + to limit stack growth and to avoid "infinite" recursions. +*/ +/* The default size for REG_INFTY is I16_MAX, which is the same as + SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits + (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is + ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is + ((1<<63)-1). To limit stack growth to reasonable sizes, supply a + smaller default. + --Andy Dougherty 11 June 1998 +*/ +#if SHORTSIZE > 2 +# ifndef REG_INFTY +# define REG_INFTY ((1<<15)-1) +# endif #endif + +#ifndef REG_INFTY +# define REG_INFTY I16_MAX #endif -#define OP(p) (*(p)) +#define ARG_VALUE(arg) (arg) +#define ARG__SET(arg,val) ((arg) = (val)) + +#define ARG(p) ARG_VALUE(ARG_LOC(p)) +#define ARG1(p) ARG_VALUE(ARG1_LOC(p)) +#define ARG2(p) ARG_VALUE(ARG2_LOC(p)) +#define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val)) +#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val)) +#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val)) #ifndef lint -#ifdef REGALIGN -#define NEXT(p) (*(short*)(p+1)) -#else -#define NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377)) -#endif +# define NEXT_OFF(p) ((p)->next_off) +# define NODE_ALIGN(node) +# define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */ #else /* lint */ -#define NEXT(p) 0 +# define NEXT_OFF(p) 0 +# define NODE_ALIGN(node) +# define NODE_ALIGN_FILL(node) #endif /* lint */ -#define OPERAND(p) ((p) + 3) +#define SIZE_ALIGN NODE_ALIGN -#ifdef REGALIGN -#define NEXTOPER(p) ((p) + 4) -#else -#define NEXTOPER(p) ((p) + 3) -#endif +#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) + +#define NODE_ALIGN(node) +#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1) +#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1) +#define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2) +#define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */ +#define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2) + +#define NODE_STEP_B 4 + +#define NEXTOPER(p) ((p) + NODE_STEP_REGNODE) +#define PREVOPER(p) ((p) - NODE_STEP_REGNODE) + +#define FILL_ADVANCE_NODE(ptr, op) STMT_START { \ + (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END +#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 REG_MAGIC 0234 + +#define SIZE_ONLY (PL_regcode == &PL_regdummy) + +/* Flags for first parameter byte [0] of ANYOF */ + +#define ANYOF_CLASS 0x08 +#define ANYOF_INVERT 0x04 +#define ANYOF_FOLD 0x02 +#define ANYOF_LOCALE 0x01 + +/* Character classes for bytes [1..4] of ANYOF */ + +#define ANYOF_ALNUM 0 /* \w, utf8::IsWord, isALNUM() */ +#define ANYOF_NALNUM 1 +#define ANYOF_SPACE 2 +#define ANYOF_NSPACE 3 +#define ANYOF_DIGIT 4 +#define ANYOF_NDIGIT 5 +#define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, isALNUMC() */ +#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_MAX 31 + +/* 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_OPND_SIZE 1 +#define ANYOF_CLASS_SIZE 4 +#define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */ +#define ANYOF_SIZE (ANYOF_OPND_SIZE+ANYOF_CLASS_SIZE+ANYOF_BITMAP_SIZE) + +#define ANYOF_FLAGS(p) ((p)[0]) +#define ANYOF_FLAGS_ALL 0xff + +#define ANYOF_BIT(c) (1 << ((c) & 7)) + +#define ANYOF_CLASS_OFFSET ANYOF_OPND_SIZE +#define ANYOF_CLASS_BYTE(p, c) ((p)[ANYOF_CLASS_OFFSET + (((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_BITMAP_OFFSET (ANYOF_CLASS_OFFSET+ANYOF_CLASS_SIZE) +#define ANYOF_BITMAP_BYTE(p, c) ((p)[ANYOF_BITMAP_OFFSET + (((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 ANY_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode) + 1) /* * Utility definitions. */ #ifndef lint -#ifndef CHARBITS -#define UCHARAT(p) ((int)*(unsigned char *)(p)) +#ifndef CHARMASK +#define UCHARAT(p) ((int)*(U8*)(p)) #else -#define UCHARAT(p) ((int)*(p)&CHARBITS) +#define UCHARAT(p) ((int)*(p)&CHARMASK) #endif #else /* lint */ -#define UCHARAT(p) regdummy +#define UCHARAT(p) PL_regdummy #endif /* lint */ -#define FAIL(m) fatal("/%s/: %s",regprecomp,m) +#define FAIL(m) \ + STMT_START { \ + if (!SIZE_ONLY) \ + SAVEDESTRUCTOR(clear_re,(void*)PL_regcomp_rx); \ + Perl_croak(aTHX_ "/%.127s/: %s", PL_regprecomp,m); \ + } STMT_END + +#define FAIL2(pat,m) \ + STMT_START { \ + if (!SIZE_ONLY) \ + SAVEDESTRUCTOR(clear_re,(void*)PL_regcomp_rx); \ + S_re_croak2(aTHX_ "/%.127s/: ",pat,PL_regprecomp,m); \ + } STMT_END -char *regnext(); -#ifdef DEBUGGING -void regdump(); -char *regprop(); +#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode)) + +#define REG_SEEN_ZERO_LEN 1 +#define REG_SEEN_LOOKBEHIND 2 +#define REG_SEEN_GPOS 4 +#define REG_SEEN_EVAL 8 + +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 +/* 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, ANYUTF8, SANY, SANYUTF8, ANYOF, ANYOFUTF8, + ALNUM, ALNUMUTF8, ALNUML, ALNUMLUTF8, + NALNUM, NALNUMUTF8, NALNUML, NALNUMLUTF8, + SPACE, SPACEUTF8, SPACEL, SPACELUTF8, + NSPACE, NSPACEUTF8, NSPACEL, NSPACELUTF8, + DIGIT, DIGITUTF8, NDIGIT, NDIGITUTF8, 0 +}; +#endif + +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; + +struct reg_data { + U32 count; + U8 *what; + void* data[1]; +}; + +struct reg_substr_datum { + I32 min_offset; + I32 max_offset; + SV *substr; +}; + +struct reg_substr_data { + struct reg_substr_datum data[3]; /* Actual array */ +}; + +#define anchored_substr substrs->data[0].substr +#define anchored_offset substrs->data[0].min_offset +#define float_substr substrs->data[1].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_offset_min substrs->data[2].min_offset +#define check_offset_max substrs->data[2].max_offset