Magic-Mirror/gecko-dev
1
0
Fork 0
mirror of https://github.com/mozilla/gecko-dev.git synced 2024-11-26 14:22:01 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
89d641ff37
gecko-dev/js/ref/jsregexp.c
brendan%netscape.com 347aaac8d3 js.c jsemit.c jsemit.h jsgc.c jsinterp.c jsopcode.c jsopcode.def 
- Switch improvements:
  - JSOP_CONDSWITCH is a 1 byte nop, not variable length with the same kind
    of immediate operand as JSOP_LOOKUPSWITCH (which is useless except for
    decompilation).  New scheme uses SRC_COMMA notes on each JSOP_CASE opcode,
    usually 2 bytes per note, and a typically-1-byte 2nd offset on SRC_SWITCH:
      1 + 2 * ncases
    vs. the previous JSOP_LOOKUPSWITCH immediate, which consumed:
      4 * ncases
    bytes after the switch opcode just for decompilation.
  - SRC_SWITCH has two offsets, first to end of switch as before, the second
    to first case if JSOP_CONDSWITCH, for decompilation.
  - Optimize switches with all-constant cases using JSOP_TABLESWITH, or if
    that can't be used, JSOP_LOOKUPSWITCH, before falling back on ECMAv2's
    JSOP_CONDSWITCH.
  - Use cx->gcDisabled when evaluating case exprs at compile time for old,
    pre-ECMAv2 switches, to prevent branch-callback-based GC invocations
    from ripping apart the unrooted temporary script for each case expr.
  - Fixed up stale SRC_SWITCH comments in jsemit.h.

jsemit.c jsemit.h
  - TREE_CONTEXT_INIT to match ATOM_LIST_INIT, not English word order.
  - Reorganized JSCodeGenerator to sort of match argument order to
    js_InitCodeGenerator.
  - Got rid of confusing CG_RESET* macros and used memset(cg, 0, sizeof *cg)
    and non-zero-default init in js_InitCodeGenerator.  js_ResetCodeGenerator
    just releases the code and temp arena pools and leaves the cg in a state
    where it must be re-initialized (as before, but more obvious).
  - In the same spirit, don't do partial "resets" or src and trynotes in their
    js_FinishTaking*Notes functions -- those are friends of jsscript.c and are
    not general, idempotent functions.

jsapi.c jsapi.h jsarray.c jsatom.c jsatom.h jscntxt.c jsemit.c jsmsg.def
jsnum.c jsobj.c jsopcode.c jsregexp.c jsscan.c jsstr.c jsxdrapi.
  - Use PR_snprintf rather than sprintf always, so we don't have to worry
    about systems with 64-bit longs that overflow 12-byte buffers and open
    Morris-Worm-type security holes.
  - Trim extra spaces, fix hanging indentation, and similar anal retention.
  - Renamed JSMSG_BAD_PROTO_SORT to JSMSG_BAD_SORT_ARG cuz that's what it
    is complaining about.
  - SRC_CATCHGUARD still lived in comments, but it's SRC_CATCH in code.

jscntxt.c jscntxt.h jsinterp.c
  - Packed nearby JSPackedBools and added a new one: gcDisabled, for use by
    jsemit.c's pre-ECMAv2 switch case expr eval.
  - Rip out old js_InterpreterHooks stuff from original liveconnect (moja).
  - Remove javaData and savedErrors from JSContext.  Leaving it to fur or
    shaver to remove javaData from jsscript.h.
1998-09-08 05:39:51 +00:00

3312 lines
84 KiB
C
Raw Blame History

/* -*- Mode: C; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 4 -*-
*
* The contents of this file are subject to the Netscape Public License
* Version 1.0 (the "NPL"); you may not use this file except in
* compliance with the NPL. You may obtain a copy of the NPL at
* http://www.mozilla.org/NPL/
*
* Software distributed under the NPL is distributed on an "AS IS" basis,
* WITHOUT WARRANTY OF ANY KIND, either express or implied. See the NPL
* for the specific language governing rights and limitations under the
* NPL.
*
* The Initial Developer of this code under the NPL is Netscape
* Communications Corporation. Portions created by Netscape are
* Copyright (C) 1998 Netscape Communications Corporation. All Rights
* Reserved.
*/
/*
* JS regular expressions, after Perl.
*/
#include "jsstddef.h"
#include <stdlib.h>
#include <string.h>
#include "prtypes.h"
#include "prarena.h"
#include "prassert.h"
#include "jsapi.h"
#include "jsarray.h"
#include "jsatom.h"
#include "jscntxt.h"
#include "jsconfig.h"
#include "jsfun.h"
#include "jsgc.h"
#include "jsinterp.h"
#include "jslock.h"
#include "jsnum.h"
#include "jsobj.h"
#include "jsopcode.h"
#include "jsregexp.h"
#include "jsstr.h"
#if JS_HAS_REGEXPS
typedef struct RENode RENode;
typedef enum REOp {
REOP_EMPTY = 0, /* match rest of input against rest of r.e. */
REOP_ALT = 1, /* alternative subexpressions in kid and next */
REOP_BOL = 2, /* beginning of input (or line if multiline) */
REOP_EOL = 3, /* end of input (or line if multiline) */
REOP_WBDRY = 4, /* match "" at word boundary */
REOP_WNONBDRY = 5, /* match "" at word non-boundary */
REOP_QUANT = 6, /* quantified atom: atom{1,2} */
REOP_STAR = 7, /* zero or more occurrences of kid */
REOP_PLUS = 8, /* one or more occurrences of kid */
REOP_OPT = 9, /* optional subexpression in kid */
REOP_LPAREN = 10, /* left paren bytecode: kid is u.num'th sub-regexp */
REOP_RPAREN = 11, /* right paren bytecode */
REOP_DOT = 12, /* stands for any character */
REOP_CCLASS = 13, /* character class: [a-f] */
REOP_DIGIT = 14, /* match a digit char: [0-9] */
REOP_NONDIGIT = 15, /* match a non-digit char: [^0-9] */
REOP_ALNUM = 16, /* match an alphanumeric char: [0-9a-z_A-Z] */
REOP_NONALNUM = 17, /* match a non-alphanumeric char: [^0-9a-z_A-Z] */
REOP_SPACE = 18, /* match a whitespace char */
REOP_NONSPACE = 19, /* match a non-whitespace char */
REOP_BACKREF = 20, /* back-reference (e.g., \1) to a parenthetical */
REOP_FLAT = 21, /* match a flat string */
REOP_FLAT1 = 22, /* match a single char */
REOP_JUMP = 23, /* for deoptimized closure loops */
REOP_DOTSTAR = 24, /* optimize .* to use a single opcode */
REOP_ANCHOR = 25, /* like .* but skips left context to unanchored r.e. */
REOP_EOLONLY = 26, /* $ not preceded by any pattern */
REOP_UCFLAT = 27, /* flat Unicode string; len immediate counts chars */
REOP_UCFLAT1 = 28, /* single Unicode char */
REOP_UCCLASS = 29, /* Unicode character class, vector of chars to match */
REOP_NUCCLASS = 30, /* negated Unicode character class */
REOP_BACKREFi = 31, /* case-independent REOP_BACKREF */
REOP_FLATi = 32, /* case-independent REOP_FLAT */
REOP_FLAT1i = 33, /* case-independent REOP_FLAT1 */
REOP_UCFLATi = 34, /* case-independent REOP_UCFLAT */
REOP_UCFLAT1i = 35, /* case-independent REOP_UCFLAT1 */
REOP_ANCHOR1 = 36, /* first-char discriminating REOP_ANCHOR */
REOP_NCCLASS = 37, /* negated 8-bit character class */
REOP_END
} REOp;
#define CCLASS_CHARSET_SIZE 256 /* ISO-Latin-1 */
uint8 reopsize[] = {
/* EMPTY */ 1,
/* ALT */ 3,
/* BOL */ 1,
/* EOL */ 1,
/* WBDRY */ 1,
/* WNONBDRY */ 1,
/* QUANT */ 7,
/* STAR */ 1,
/* PLUS */ 1,
/* OPT */ 1,
/* LPAREN */ 3,
/* RPAREN */ 3,
/* DOT */ 1,
/* CCLASS */ 1 + (CCLASS_CHARSET_SIZE / PR_BITS_PER_BYTE),
/* DIGIT */ 1,
/* NONDIGIT */ 1,
/* ALNUM */ 1,
/* NONALNUM */ 1,
/* SPACE */ 1,
/* NONSPACE */ 1,
/* BACKREF */ 2,
/* FLAT */ 2, /* (2 = op + len) + [len bytes] */
/* FLAT1 */ 2,
/* JUMP */ 3,
/* DOTSTAR */ 1,
/* ANCHOR */ 1,
/* EOLONLY */ 1,
/* UCFLAT */ 2, /* (2 = op + len) + [len 2-byte chars] */
/* UCFLAT1 */ 3, /* op + hibyte + lobyte */
/* UCCLASS */ 3, /* (3 = op + 2-byte len) + [len bytes] */
/* NUCCLASS */ 3, /* (3 = op + 2-byte len) + [len bytes] */
/* BACKREFi */ 2,
/* FLATi */ 2, /* (2 = op + len) + [len bytes] */
/* FLAT1i */ 2,
/* UCFLATi */ 2, /* (2 = op + len) + [len 2-byte chars] */
/* UCFLAT1i */ 3, /* op + hibyte + lobyte */
/* ANCHOR1 */ 1,
/* NCCLASS */ 1 + (CCLASS_CHARSET_SIZE / PR_BITS_PER_BYTE),
/* END */ 0,
};
#define REOP_FLATLEN_MAX 255 /* maximum length of FLAT string */
struct RENode {
uint8 op; /* packed r.e. op bytecode */
uint8 flags; /* flags, see below */
uint16 offset; /* bytecode offset */
RENode *next; /* next in concatenation order */
void *kid; /* first operand */
union {
void *kid2; /* second operand */
jsint num; /* could be a number */
jschar chr; /* or a character */
struct { /* or a quantifier range */
uint16 min;
uint16 max;
} range;
struct { /* or a Unicode character class */
uint16 kidlen; /* length of string at kid, in jschars */
uint16 bmsize; /* bitmap size, based on max char code */
} ucclass;
} u;
};
#define REOP(ren) ((REOp)(ren)->op)
#define RENODE_ANCHORED 0x01 /* anchored at the front */
#define RENODE_SINGLE 0x02 /* matches a single char */
#define RENODE_NONEMPTY 0x04 /* does not match empty string */
#define RENODE_ISNEXT 0x08 /* ren is next after at least one node */
#define RENODE_GOODNEXT 0x10 /* ren->next is a tree-like edge in the graph */
#define RENODE_ISJOIN 0x20 /* ren is a join point in the graph */
#define RENODE_REALLOK 0x40 /* REOP_FLAT owns tempPool space to realloc */
typedef struct CompilerState {
JSContext *context;
const jschar *cpbegin;
const jschar *cp;
uintN flags;
uintN parenCount;
size_t progLength;
} CompilerState;
static RENode *
NewRENode(CompilerState *state, REOp op, void *kid)
{
JSContext *cx;
RENode *ren;
cx = state->context;
PR_ARENA_ALLOCATE(ren, &cx->tempPool, sizeof *ren);
if (!ren) {
JS_ReportOutOfMemory(cx);
return NULL;
}
ren->op = (uint8)op;
ren->flags = 0;
ren->offset = 0;
ren->next = NULL;
ren->kid = kid;
return ren;
}
#ifdef DEBUG
#include <stdio.h>
static char *reopname[] = {
"empty",
"alt",
"bol",
"eol",
"wbdry",
"wnonbdry",
"quant",
"star",
"plus",
"opt",
"lparen",
"rparen",
"dot",
"cclass",
"digit",
"nondigit",
"alnum",
"nonalnum",
"space",
"nonspace",
"backref",
"flat",
"flat1",
"jump",
"dotstar",
"anchor",
"eolonly",
"ucflat",
"ucflat1",
"ucclass",
"nucclass",
"backrefi",
"flati",
"flat1i",
"ucflati",
"ucflat1i",
"anchor1",
"ncclass",
"end"
};
static void
PrintChar(jschar c)
{
if (c >> 8)
printf("\\u%04X", c);
else
#if !defined XP_PC || !defined _MSC_VER || _MSC_VER > 800
putchar((char)c);
#else
/* XXX is there a better way with MSVC1.52? */
printf("%c", c);
#endif
}
static JSBool
DumpRegExp(JSContext *cx, RENode *ren)
{
static int level;
JSBool ok;
int i, len;
jschar *cp;
char *cstr;
if (level == 0)
printf("level offset flags description\n");
level++;
ok = JS_TRUE;
do {
printf("%5d %6d %c%c%c%c%c%c %s",
level,
(int)ren->offset,
(ren->flags & RENODE_ANCHORED) ? 'A' : '-',
(ren->flags & RENODE_SINGLE) ? 'S' : '-',
(ren->flags & RENODE_NONEMPTY) ? 'F' : '-', /* F for full */
(ren->flags & RENODE_ISNEXT) ? 'N' : '-', /* N for next */
(ren->flags & RENODE_GOODNEXT) ? 'G' : '-',
(ren->flags & RENODE_ISJOIN) ? 'J' : '-',
reopname[ren->op]);
switch (REOP(ren)) {
case REOP_ALT:
printf(" %d\n", ren->next->offset);
ok = DumpRegExp(cx, ren->kid);
if (!ok)
goto out;
break;
case REOP_STAR:
case REOP_PLUS:
case REOP_OPT:
case REOP_ANCHOR1:
printf("\n");
ok = DumpRegExp(cx, ren->kid);
if (!ok)
goto out;
break;
case REOP_QUANT:
printf(" next %d min %d max %d\n",
ren->next->offset, ren->u.range.min, ren->u.range.max);
ok = DumpRegExp(cx, ren->kid);
if (!ok)
goto out;
break;
case REOP_LPAREN:
printf(" num %d\n", (int)ren->u.num);
ok = DumpRegExp(cx, ren->kid);
if (!ok)
goto out;
break;
case REOP_RPAREN:
printf(" num %d\n", (int)ren->u.num);
break;
case REOP_CCLASS:
len = (jschar *)ren->u.kid2 - (jschar *)ren->kid;
cstr = js_DeflateString(cx, (jschar *)ren->kid, len);
if (!cstr) {
ok = JS_FALSE;
goto out;
}
printf(" [%s]\n", cstr);
JS_free(cx, cstr);
break;
case REOP_BACKREF:
printf(" num %d\n", (int)ren->u.num);
break;
case REOP_FLAT:
len = (jschar *)ren->u.kid2 - (jschar *)ren->kid;
cstr = js_DeflateString(cx, (jschar *)ren->kid, len);
if (!cstr) {
ok = JS_FALSE;
goto out;
}
printf(" %s (%d)\n", cstr, len);
JS_free(cx, cstr);
break;
case REOP_FLAT1:
printf(" %c ('\\%o')\n", (char)ren->u.chr, ren->u.chr);
break;
case REOP_JUMP:
printf(" %d\n", ren->next->offset);
break;
case REOP_UCFLAT:
cp = ren->kid;
len = (jschar *)ren->u.kid2 - cp;
for (i = 0; i < len; i++)
PrintChar(cp[i]);
break;
case REOP_UCFLAT1:
PrintChar(ren->u.chr);
break;
case REOP_UCCLASS:
cp = ren->kid;
len = (jschar *)ren->u.kid2 - cp;
printf(" [");
for (i = 0; i < len; i++)
PrintChar(cp[i]);
printf("]\n");
break;
default:
printf("\n");
break;
}
if (!(ren->flags & RENODE_GOODNEXT))
break;
} while ((ren = ren->next) != NULL);
out:
level--;
return ok;
}
#endif /* DEBUG */
static JSBool
FixNext(CompilerState *state, RENode *ren1, RENode *ren2, RENode *oldnext)
{
JSBool goodnext;
RENode *next, *kid, *ren;
goodnext = ren2 && !(ren2->flags & RENODE_ISNEXT);
/*
* Find the final node in a list of alternatives, or concatenations, or
* even a concatenation of alternatives followed by non-alternatives (e.g.
* ((x|y)z)w where ((x|y)z) is ren1 and w is ren2).
*/
for (; (next = ren1->next) != NULL && next != oldnext; ren1 = next) {
if (REOP(ren1) == REOP_ALT) {
/* Find the end of this alternative's operand list. */
kid = ren1->kid;
if (REOP(kid) == REOP_JUMP)
continue;
for (ren = kid; ren->next; ren = ren->next)
PR_ASSERT(REOP(ren) != REOP_ALT);
/* Append a jump node to all but the last alternative. */
ren->next = NewRENode(state, REOP_JUMP, NULL);
if (!ren->next)
return JS_FALSE;
ren->next->flags |= RENODE_ISNEXT;
ren->flags |= RENODE_GOODNEXT;
/* Recur to fix all descendent nested alternatives. */
if (!FixNext(state, kid, ren2, oldnext))
return JS_FALSE;
}
}
/*
* Now ren1 points to the last alternative, or to the final node on a
* concatenation list. Set its next link to ren2, flagging a join point
* if appropriate.
*/
if (ren2) {
if (!(ren2->flags & RENODE_ISNEXT))
ren2->flags |= RENODE_ISNEXT;
else
ren2->flags |= RENODE_ISJOIN;
}
ren1->next = ren2;
if (goodnext)
ren1->flags |= RENODE_GOODNEXT;
/*
* The following ops have a kid subtree through which to recur. Here is
* where we fix the next links under the final ALT node's kid.
*/
switch (REOP(ren1)) {
case REOP_ALT:
case REOP_QUANT:
case REOP_STAR:
case REOP_PLUS:
case REOP_OPT:
case REOP_LPAREN:
if (!FixNext(state, ren1->kid, ren2, oldnext))
return JS_FALSE;
break;
default:;
}
return JS_TRUE;
}
static JSBool
SetNext(CompilerState *state, RENode *ren1, RENode *ren2)
{
return FixNext(state, ren1, ren2, NULL);
}
/*
* Parser forward declarations.
*/
typedef RENode *REParser(CompilerState *state);
static REParser ParseRegExp;
static REParser ParseAltern;
static REParser ParseItem;
static REParser ParseQuantAtom;
static REParser ParseAtom;
/*
* Top-down regular expression grammar, based closely on Perl4.
*
* regexp: altern A regular expression is one or more
* altern '|' regexp alternatives separated by vertical bar.
*/
static RENode *
ParseRegExp(CompilerState *state)
{
RENode *ren, *kid, *ren1, *ren2;
const jschar *cp;
ren = ParseAltern(state);
if (!ren)
return NULL;
cp = state->cp;
if (*cp == '|') {
kid = ren;
ren = NewRENode(state, REOP_ALT, kid);
if (!ren)
return NULL;
ren->flags = kid->flags & (RENODE_ANCHORED | RENODE_NONEMPTY);
ren1 = ren;
do {
/* (balance: */
state->cp = ++cp;
if (*cp == '|' || *cp == ')') {
kid = NewRENode(state, REOP_EMPTY, NULL);
} else {
kid = ParseAltern(state);
cp = state->cp;
}
if (!kid)
return NULL;
ren2 = NewRENode(state, REOP_ALT, kid);
if (!ren2)
return NULL;
ren1->next = ren2;
ren1->flags |= RENODE_GOODNEXT;
ren2->flags = (kid->flags & (RENODE_ANCHORED | RENODE_NONEMPTY))
| RENODE_ISNEXT;
ren1 = ren2;
} while (*cp == '|');
}
return ren;
}
/*
* altern: item An alternative is one or more items,
* item altern concatenated together.
*/
static RENode *
ParseAltern(CompilerState *state)
{
RENode *ren, *ren1, *ren2;
uintN flags;
const jschar *cp;
jschar c;
ren = ren1 = ParseItem(state);
if (!ren)
return NULL;
flags = 0;
cp = state->cp;
/* (balance: */
while ((c = *cp) != 0 && c != '|' && c != ')') {
ren2 = ParseItem(state);
if (!ren2)
return NULL;
if (!SetNext(state, ren1, ren2))
return NULL;
flags |= ren2->flags;
ren1 = ren2;
cp = state->cp;
}
/*
* Propagate NONEMPTY to the front of a concatenation list, so that the
* first alternative in (^a|b) is considered non-empty. The first node
* in a list may match the empty string (as ^ does), but if the list is
* non-empty, then the first node's NONEMPTY flag must be set.
*/
ren->flags |= flags & RENODE_NONEMPTY;
return ren;
}
/*
* item: assertion An item is either an assertion or
* quantatom a quantified atom.
*
* assertion: '^' Assertions match beginning of string
* (or line if the class static property
* RegExp.multiline is true).
* '$' End of string (or line if the class
* static property RegExp.multiline is
* true).
* '\b' Word boundary (between \w and \W).
* '\B' Word non-boundary.
*/
static RENode *
ParseItem(CompilerState *state)
{
const jschar *cp;
RENode *ren;
REOp op;
cp = state->cp;
switch (*cp) {
case '^':
state->cp = cp + 1;
ren = NewRENode(state, REOP_BOL, NULL);
if (ren)
ren->flags |= RENODE_ANCHORED;
return ren;
case '$':
state->cp = cp + 1;
return NewRENode(state,
(cp == state->cpbegin ||
((cp[-1] == '(' || cp[-1] == '|') && /*balance)*/
(cp - 1 == state->cpbegin || cp[-2] != '\\')))
? REOP_EOLONLY
: REOP_EOL,
NULL);
case '\\':
switch (*++cp) {
case 'b':
op = REOP_WBDRY;
break;
case 'B':
op = REOP_WNONBDRY;
break;
default:
return ParseQuantAtom(state);
}
/*
* Word boundaries and non-boundaries are flagged as non-empty so they
* will be prefixed by an anchoring node.
*/
state->cp = cp + 1;
ren = NewRENode(state, op, NULL);
if (ren)
ren->flags |= RENODE_NONEMPTY;
return ren;
default:;
}
return ParseQuantAtom(state);
}
/*
* quantatom: atom An unquantified atom.
* quantatom '{' n ',' m '}'
* Atom must occur between n and m times.
* quantatom '{' n ',' '}' Atom must occur at least n times.
* quantatom '{' n '}' Atom must occur exactly n times.
* quantatom '*' Zero or more times (same as {0,}).
* quantatom '+' One or more times (same as {1,}).
* quantatom '?' Zero or one time (same as {0,1}).
*/
static RENode *
ParseQuantAtom(CompilerState *state)
{
RENode *ren, *ren2;
const jschar *cp, *up;
jschar c;
uint32 min, max;
ren = ParseAtom(state);
if (!ren)
return NULL;
cp = state->cp;
loop:
switch (*cp) {
case '{':
c = *++cp;
if (!JS7_ISDEC(c)) {
JS_ReportErrorNumber(state->context, js_GetErrorMessage, NULL,
JSMSG_BAD_QUANTIFIER, state->cp);
return NULL;
}
min = (uint32)JS7_UNDEC(c);
for (c = *++cp; JS7_ISDEC(c); c = *++cp) {
min = 10 * min + (uint32)JS7_UNDEC(c);
if (min >> 16) {
JS_ReportErrorNumber(state->context, js_GetErrorMessage, NULL,
JSMSG_MIN_TOO_BIG, state->cp);
return NULL;
}
}
if (*cp == ',') {
up = ++cp;
if (JS7_ISDEC(*cp)) {
max = (uint32)JS7_UNDEC(*cp);
for (c = *++cp; JS7_ISDEC(c); c = *++cp) {
max = 10 * max + (uint32)JS7_UNDEC(c);
if (max >> 16) {
JS_ReportErrorNumber(state->context,
js_GetErrorMessage, NULL,
JSMSG_MAX_TOO_BIG, up);
return NULL;
}
}
if (max == 0)
goto zero_quant;
if (min > max) {
JS_ReportErrorNumber(state->context,
js_GetErrorMessage, NULL,
JSMSG_OUT_OF_ORDER, up);
return NULL;
}
} else {
/* 0 means no upper bound. */
max = 0;
}
} else {
/* Exactly n times. */
if (min == 0) {
zero_quant:
JS_ReportErrorNumber(state->context, js_GetErrorMessage, NULL,
JSMSG_ZERO_QUANTIFIER, state->cp);
return NULL;
}
max = min;
}
if (*cp != '}') {
JS_ReportErrorNumber(state->context, js_GetErrorMessage, NULL,
JSMSG_UNTERM_QUANTIFIER, state->cp);
return NULL;
}
cp++;
ren2 = NewRENode(state, REOP_QUANT, ren);
if (!ren2)
return NULL;
if (min > 0 && (ren->flags & RENODE_NONEMPTY))
ren2->flags |= RENODE_NONEMPTY;
ren2->u.range.min = (uint16)min;
ren2->u.range.max = (uint16)max;
ren = ren2;
goto loop;
case '*':
if (!(ren->flags & RENODE_NONEMPTY)) {
JS_ReportErrorNumber(state->context, js_GetErrorMessage, NULL,
JSMSG_EMPTY_BEFORE_STAR);
return NULL;
}
cp++;
ren = NewRENode(state, REOP_STAR, ren);
goto loop;
case '+':
if (!(ren->flags & RENODE_NONEMPTY)) {
JS_ReportErrorNumber(state->context, js_GetErrorMessage, NULL,
JSMSG_EMPTY_BEFORE_PLUS);
return NULL;
}
cp++;
ren2 = NewRENode(state, REOP_PLUS, ren);
if (!ren2)
return NULL;
if (ren->flags & RENODE_NONEMPTY)
ren2->flags |= RENODE_NONEMPTY;
ren = ren2;
goto loop;
case '?':
cp++;
ren = NewRENode(state, REOP_OPT, ren);
goto loop;
}
state->cp = cp;
return ren;
}
/*
* atom: '(' regexp ')' A parenthesized regexp (what matched
* can be addressed using a backreference,
* see '\' n below).
* '.' Matches any char except '\n'.
* '[' classlist ']' A character class.
* '[' '^' classlist ']' A negated character class.
* '\f' Form Feed.
* '\n' Newline (Line Feed).
* '\r' Carriage Return.
* '\t' Horizontal Tab.
* '\v' Vertical Tab.
* '\d' A digit (same as [0-9]).
* '\D' A non-digit.
* '\w' A word character, [0-9a-z_A-Z].
* '\W' A non-word character.
* '\s' A whitespace character, [ \b\f\n\r\t\v].
* '\S' A non-whitespace character.
* '\' n A backreference to the nth (n decimal
* and positive) parenthesized expression.
* '\' octal An octal escape sequence (octal must be
* two or three digits long, unless it is
* 0 for the null character).
* '\x' hex A hex escape (hex must be two digits).
* '\c' ctrl A control character, ctrl is a letter.
* '\' literalatomchar Any character except one of the above
* that follow '\' in an atom.
* otheratomchar Any character not first among the other
* atom right-hand sides.
*/
static jschar metachars[] = {
'|', '^', '$', '{', '*', '+', '?', '(', ')', '.', '[', '\\', '}', 0
};
static jschar closurechars[] = {
'{', '*', '+', '?', 0 /* balance} */
};
static RENode *
ParseAtom(CompilerState *state)
{
const jschar *cp, *ocp;
uintN num, tmp, len;
RENode *ren, *ren2;
jschar c;
cp = ocp = state->cp;
switch (*cp) {
case 0:
ren = NewRENode(state, REOP_EMPTY, NULL);
break;
case '(':
num = state->parenCount++; /* \1 is numbered 0, etc. */
state->cp = cp + 1;
ren2 = ParseRegExp(state);
if (!ren2)
return NULL;
cp = state->cp;
if (*cp != ')') {
JS_ReportErrorNumber(state->context, js_GetErrorMessage, NULL,
JSMSG_MISSING_PAREN, ocp);
return NULL;
}
cp++;
ren = NewRENode(state, REOP_LPAREN, ren2);
if (!ren)
return NULL;
ren->flags = ren2->flags & (RENODE_ANCHORED | RENODE_NONEMPTY);
ren->u.num = num;
ren2 = NewRENode(state, REOP_RPAREN, NULL);
if (!ren2 || !SetNext(state, ren, ren2))
return NULL;
ren2->u.num = num;
break;
case '.':
cp++;
if ((c = *cp) == '*')
cp++;
ren = NewRENode(state, (c == '*') ? REOP_DOTSTAR : REOP_DOT, NULL);
if (ren && REOP(ren) == REOP_DOT)
ren->flags = RENODE_SINGLE | RENODE_NONEMPTY;
break;
case '[':
/* A char class must have at least one char in it. */
if ((c = *++cp) == 0)
goto bad_cclass;
ren = NewRENode(state, REOP_CCLASS, (void *)cp);
if (!ren)
return NULL;
/* A negated class must have at least one char in it after the ^. */
if (c == '^' && *++cp == 0)
goto bad_cclass;
while ((c = *++cp) != ']') {
if (c == 0) {
bad_cclass:
JS_ReportErrorNumber(state->context, js_GetErrorMessage, NULL,
JSMSG_UNTERM_CLASS, ocp);
return NULL;
}
if (c == '\\' && cp[1] != 0)
cp++;
}
ren->u.kid2 = (void *)cp++;
/* Since we rule out [] and [^], we can set the non-empty flag. */
ren->flags = RENODE_SINGLE | RENODE_NONEMPTY;
break;
case '\\':
c = *++cp;
switch (c) {
case 0:
JS_ReportErrorNumber(state->context, js_GetErrorMessage, NULL,
JSMSG_TRAILING_SLASH);
return NULL;
case 'f':
case 'n':
case 'r':
case 't':
case 'v':
ren = NewRENode(state, REOP_FLAT1, NULL);
c = js_strchr(js_EscapeMap, c)[-1];
break;
case 'd':
ren = NewRENode(state, REOP_DIGIT, NULL);
break;
case 'D':
ren = NewRENode(state, REOP_NONDIGIT, NULL);
break;
case 'w':
ren = NewRENode(state, REOP_ALNUM, NULL);
break;
case 'W':
ren = NewRENode(state, REOP_NONALNUM, NULL);
break;
case 's':
ren = NewRENode(state, REOP_SPACE, NULL);
break;
case 'S':
ren = NewRENode(state, REOP_NONSPACE, NULL);
break;
case '0':
do_octal:
num = 0;
while ('0' <= (c = *++cp) && c <= '7') {
tmp = 8 * num + (uintN)JS7_UNDEC(c);
if (tmp > 0377)
break;
num = tmp;
}
cp--;
ren = NewRENode(state, REOP_FLAT1, NULL);
c = (jschar)num;
break;
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
case '8':
case '9':
num = (uintN)JS7_UNDEC(c);
for (c = *++cp; JS7_ISDEC(c); c = *++cp)
num = 10 * num - (uintN)JS7_UNDEC(c);
if (num > 9 && num > state->parenCount) {
cp = ocp;
goto do_octal;
}
cp--;
ren = NewRENode(state, REOP_BACKREF, NULL);
if (!ren)
return NULL;
ren->u.num = num - 1; /* \1 is numbered 0, etc. */
/* Avoid common chr- and flags-setting code after switch. */
ren->flags = RENODE_NONEMPTY;
goto bump_cp;
case 'x':
ocp = cp;
c = *++cp;
if (JS7_ISHEX(c)) {
num = JS7_UNHEX(c);
c = *++cp;
if (JS7_ISHEX(c)) {
num <<= 4;
num += JS7_UNHEX(c);
} else {
cp--; /* back up so cp points to last hex char */
}
} else {
cp = ocp; /* \xZZ is xZZ (Perl does \0ZZ!) */
num = 'x';
}
ren = NewRENode(state, REOP_FLAT1, NULL);
c = (jschar)num;
break;
case 'c':
c = *++cp;
if (!JS7_ISLET(c)) {
cp -= 2;
ocp = cp;
goto do_flat;
}
c = JS_TOUPPER(c);
c = JS_TOCTRL(c);
ren = NewRENode(state, REOP_FLAT1, NULL);
break;
case 'u':
if (JS7_ISHEX(cp[1]) && JS7_ISHEX(cp[2]) &&
JS7_ISHEX(cp[3]) && JS7_ISHEX(cp[4])) {
c = (((((JS7_UNHEX(cp[1]) << 4) + JS7_UNHEX(cp[2])) << 4)
+ JS7_UNHEX(cp[3])) << 4) + JS7_UNHEX(cp[4]);
cp += 4;
ren = NewRENode(state, REOP_FLAT1, NULL);
break;
}
/* Unlike Perl \xZZ, we take \uZZZ to be literal-u then ZZZ. */
ocp = cp;
goto do_flat;
default:
ocp = cp;
goto do_flat;
}
/* Common chr- and flags-setting code for escape opcodes. */
if (ren) {
ren->u.chr = c;
ren->flags = RENODE_SINGLE | RENODE_NONEMPTY;
}
bump_cp:
/* Skip to next unparsed char. */
cp++;
break;
default:
do_flat:
while ((c = *++cp) != 0 && !js_strchr(metachars, c))
;
len = (uintN)(cp - ocp);
if (c != 0 && len > 1 && js_strchr(closurechars, c)) {
cp--;
len--;
}
if (len > REOP_FLATLEN_MAX) {
len = REOP_FLATLEN_MAX;
cp = ocp + len;
}
ren = NewRENode(state, (len == 1) ? REOP_FLAT1 : REOP_FLAT,
(void *)ocp);
if (!ren)
return NULL;
ren->flags = RENODE_NONEMPTY;
if (len > 1) {
ren->u.kid2 = (void *)cp;
} else {
ren->flags |= RENODE_SINGLE;
ren->u.chr = *ocp;
}
break;
}
state->cp = cp;
return ren;
}
static ptrdiff_t
CountFirstChars(RENode *alt)
{
ptrdiff_t len, sublen;
RENode *kid;
jschar c, *ccp, *ccend;
len = 0;
do {
for (kid = alt->kid; REOP(kid) == REOP_LPAREN; kid = kid->kid)
;
switch (REOP(kid)) {
case REOP_QUANT:
if (kid->u.range.min == 0)
return -1;
/* FALL THROUGH */
case REOP_PLUS:
case REOP_ALT:
sublen = CountFirstChars(kid);
if (sublen < 0)
return sublen;
len += sublen;
break;
case REOP_FLAT:
c = *(jschar *)kid->kid;
goto count_char;
case REOP_FLAT1:
c = kid->u.chr;
count_char:
/* Only '\\' and '-' need escaping within a character class. */
if (c == '\\' || c == '-')
len += 2;
else
len++;
break;
case REOP_CCLASS:
ccp = kid->kid;
if (*ccp == '^')
return -1;
ccend = kid->u.kid2;
len += ccend - ccp;
break;
case REOP_DIGIT:
case REOP_NONDIGIT:
case REOP_ALNUM:
case REOP_NONALNUM:
case REOP_SPACE:
case REOP_NONSPACE:
len += 2;
break;
default:
return -1;
}
/* Test for non-alt so quant and plus execute to here only. */
if (REOP(alt) != REOP_ALT)
break;
alt = alt->next;
} while (alt && REOP(alt) == REOP_ALT);
return len;
}
static ptrdiff_t
StoreChar(jschar *cp, ptrdiff_t i, jschar c, JSBool escape)
{
ptrdiff_t j;
/* Suppress dups to avoid making a flat1 into a cclass. */
for (j = 0; j < i; j++) {
if (cp[j] == '\\')
j++;
if (cp[j] == c && (!escape || (j > 0 && cp[j-1] == '\\')))
return i;
}
/* Only '\\' and '-' need escaping within a character class. */
if (escape || c == '\\' || c == '-')
cp[i++] = '\\';
cp[i++] = c;
return i;
}
static ptrdiff_t
StoreFirstChars(RENode *alt, jschar *cp, ptrdiff_t i)
{
RENode *kid;
jschar *ccp, *ccend;
do {
for (kid = alt->kid; REOP(kid) == REOP_LPAREN; kid = kid->kid)
;
switch (REOP(kid)) {
case REOP_QUANT:
PR_ASSERT(kid->u.range.min != 0);
/* FALL THROUGH */
case REOP_PLUS:
case REOP_ALT:
i = StoreFirstChars(kid, cp, i);
break;
case REOP_FLAT:
i = StoreChar(cp, i, *(jschar *)kid->kid, JS_FALSE);
break;
case REOP_FLAT1:
i = StoreChar(cp, i, kid->u.chr, JS_FALSE);
break;
case REOP_CCLASS:
ccend = kid->u.kid2;
for (ccp = kid->kid; ccp < ccend; ccp++)
cp[i++] = *ccp;
break;
case REOP_DIGIT:
i = StoreChar(cp, i, 'd', JS_TRUE);
break;
case REOP_NONDIGIT:
i = StoreChar(cp, i, 'D', JS_TRUE);
break;
case REOP_ALNUM:
i = StoreChar(cp, i, 'w', JS_TRUE);
break;
case REOP_NONALNUM:
i = StoreChar(cp, i, 'W', JS_TRUE);
break;
case REOP_SPACE:
i = StoreChar(cp, i, 's', JS_TRUE);
break;
case REOP_NONSPACE:
i = StoreChar(cp, i, 'S', JS_TRUE);
break;
default:
PR_ASSERT(0);
}
/* Test for non-alt so quant and plus execute to here only. */
if (REOP(alt) != REOP_ALT)
break;
alt = alt->next;
} while (alt && REOP(alt) == REOP_ALT);
return i;
}
static JSBool
AnchorRegExp(CompilerState *state, RENode *ren)
{
RENode *ren2, *kid;
ptrdiff_t len;
jschar *cp;
REOp op;
for (ren2 = ren; REOP(ren2) == REOP_LPAREN; ren2 = ren2->kid)
;
switch (REOP(ren2)) {
case REOP_ALT:
len = CountFirstChars(ren2);
if (len <= 0)
goto do_anchor;
PR_ARENA_ALLOCATE(cp, &state->context->tempPool, len * sizeof(jschar));
if (!cp) {
JS_ReportOutOfMemory(state->context);
return JS_FALSE;
}
len = StoreFirstChars(ren2, cp, 0);
if (len == 1) {
op = REOP_FLAT1;
} else if (len == 2 && *cp == '\\') {
switch (cp[1]) {
case '\\':
case '-':
/* No need for a character class if just '\\' or '-'. */
cp++;
op = REOP_FLAT1;
break;
case 'd':
op = REOP_DIGIT;
break;
case 'D':
op = REOP_NONDIGIT;
break;
case 'w':
op = REOP_ALNUM;
break;
case 'W':
op = REOP_NONALNUM;
break;
case 's':
op = REOP_SPACE;
break;
case 'S':
op = REOP_NONSPACE;
break;
default:
op = REOP_CCLASS;
break;
}
} else {
op = REOP_CCLASS;
}
do_first_char:
kid = NewRENode(state, op, cp);
if (!kid)
return JS_FALSE;
kid->flags = RENODE_SINGLE | RENODE_NONEMPTY;
if (op == REOP_FLAT1)
kid->u.chr = *cp;
else if (op == REOP_CCLASS)
kid->u.kid2 = cp + len;
ren2 = NewRENode(state, REOP(ren), ren->kid);
if (!ren2)
return JS_FALSE;
ren2->flags = ren->flags | RENODE_ISNEXT;
ren2->next = ren->next;
ren2->u = ren->u;
ren->op = REOP_ANCHOR1;
ren->flags = RENODE_GOODNEXT;
ren->next = ren2;
ren->kid = kid;
ren->u.kid2 = NULL;
break;
case REOP_FLAT:
cp = ren2->kid;
op = REOP_FLAT1;
goto do_first_char;
case REOP_FLAT1:
cp = &ren2->u.chr;
op = REOP_FLAT1;
goto do_first_char;
case REOP_DOTSTAR:
/*
* ".*" is anchored by definition when at the front of a list.
*/
break;
default:
do_anchor:
/*
* Any node other than dotstar that's unanchored and nonempty must be
* prefixed by REOP_ANCHOR.
*/
PR_ASSERT(REOP(ren2) != REOP_ANCHOR);
PR_ASSERT(!(ren2->flags & RENODE_ISNEXT));
if ((ren2->flags & (RENODE_ANCHORED | RENODE_NONEMPTY))
== RENODE_NONEMPTY) {
ren2 = NewRENode(state, REOP(ren), ren->kid);
if (!ren2)
return JS_FALSE;
ren2->flags = ren->flags | RENODE_ISNEXT;
ren2->next = ren->next;
ren2->u = ren->u;
ren->op = REOP_ANCHOR;
ren->flags = RENODE_GOODNEXT;
ren->next = ren2;
ren->kid = ren->u.kid2 = NULL;
}
break;
}
return JS_TRUE;
}
static RENode *
CloseTail(CompilerState *state, RENode *alt1, RENode *next)
{
RENode *alt2, *empty;
empty = NewRENode(state, REOP_EMPTY, NULL);
alt2 = NewRENode(state, REOP_ALT, empty);
if (!alt2 || !empty)
return NULL;
alt1->next = alt2;
alt2->next = empty->next = next;
if (alt1->flags & RENODE_GOODNEXT)
alt2->flags |= RENODE_GOODNEXT;
else
alt1->flags |= RENODE_GOODNEXT;
alt2->flags |= RENODE_ISNEXT;
return alt2;
}
static JSBool
OptimizeRegExp(CompilerState *state, RENode *ren)
{
RENode *kid, *next, *jump, *alt1;
uintN flag;
jschar c, c2, maxc, *cp, *cp2;
ptrdiff_t len, len2;
size_t size, incr;
JSContext *cx;
JSBool reallok;
do {
switch (REOP(ren)) {
case REOP_STAR:
kid = ren->kid;
if (!(kid->flags & RENODE_SINGLE)) {
/*
* If kid is not simple, deoptimize <kid>* as follows (the |__|
* are byte placeholders for next/jump offsets):
*
* FROM: |STAR|<kid>|
*
* +-----------------------+
* V |
* TO: |ALT|__|__|<kid>|JUMP|__|__|ALT|__|__|EMPTY|
* | ^ | ^
* +-------------------+ +--------+
*/
ren->op = REOP_ALT;
next = ren->next;
jump = NewRENode(state, REOP_JUMP, NULL);
if (!jump || !FixNext(state, kid, jump, next))
return JS_FALSE;
jump->next = ren;
if (ren->flags & RENODE_ISNEXT)
ren->flags |= RENODE_ISJOIN;
if (!CloseTail(state, ren, next))
return JS_FALSE;
}
break;
case REOP_PLUS:
kid = ren->kid;
if (!(kid->flags & RENODE_SINGLE)) {
/*
* FROM: |PLUS|<kid>|
*
* +-----------------------+
* V |
* TO: |<kid>|ALT|__|__|JUMP|__|__|ALT|__|__|EMPTY|
* | ^ | ^
* +-------------+ +--------+
*/
next = ren->next;
flag = (ren->flags & RENODE_GOODNEXT);
*ren = *kid;
jump = NewRENode(state, REOP_JUMP, NULL);
alt1 = NewRENode(state, REOP_ALT, jump);
if (!alt1 || !jump || !FixNext(state, ren, alt1, next))
return JS_FALSE;
alt1->flags |= flag;
jump->next = ren;
if (ren->flags & RENODE_ISNEXT)
ren->flags |= RENODE_ISJOIN;
if (!CloseTail(state, alt1, next))
return JS_FALSE;
}
break;
case REOP_OPT:
kid = ren->kid;
if (!(kid->flags & RENODE_SINGLE)) {
/*
* FROM: |OPT|<kid>|
*
* +------------------+
* | v
* TO: |ALT|__|__|<kid>|JUMP|__|__|ALT|__|__|EMPTY|
* | ^ | ^
* +-------------------+ +--------+
*/
ren->op = REOP_ALT;
next = ren->next;
jump = NewRENode(state, REOP_JUMP, NULL);
if (!jump || !FixNext(state, kid, jump, next))
return JS_FALSE;
jump->next = next;
if (!CloseTail(state, ren, next))
return JS_FALSE;
next->flags |= RENODE_ISJOIN;
}
break;
case REOP_FLAT:
/*
* Coalesce adjacent FLAT and FLAT1 nodes. Also coalesce FLAT and
* FLAT, which can result from deleting a coalesced FLAT1.
*/
while ((next = ren->next) != NULL &&
!(next->flags & RENODE_ISJOIN) &&
(REOP(next) == REOP_FLAT || REOP(next) == REOP_FLAT1)) {
if (REOP(next) == REOP_FLAT) {
cp2 = next->kid;
len2 = PTRDIFF((jschar *)next->u.kid2, cp2, jschar);
} else {
cp2 = &next->u.chr;
len2 = 1;
}
cp = ren->kid;
len = PTRDIFF((jschar *)ren->u.kid2, cp, jschar);
if (len + len2 > REOP_FLATLEN_MAX)
break;
cx = state->context;
reallok = ren->flags & RENODE_REALLOK;
if (reallok) {
/* Try to extend the last alloc, to fuse FLAT,FLAT1,... */
size = (len + 1) * sizeof(jschar);
incr = len2 * sizeof(jschar);
PR_ARENA_GROW(cp, &cx->tempPool, size, incr);
} else {
size = (len + len2 + 1) * sizeof(jschar);
PR_ARENA_ALLOCATE(cp, &cx->tempPool, size);
}
if (!cp) {
JS_ReportOutOfMemory(cx);
return JS_FALSE;
}
if (!reallok) {
js_strncpy(cp, ren->kid, len);
ren->flags |= RENODE_REALLOK;
}
js_strncpy(&cp[len], cp2, len2);
len += len2;
cp[len] = 0;
end_coalesce:
ren->kid = cp;
PR_ASSERT(ren->flags & RENODE_GOODNEXT);
if (!(next->flags & RENODE_GOODNEXT))
ren->flags &= ~RENODE_GOODNEXT;
ren->u.kid2 = cp + len;
ren->next = next->next;
next->op = REOP_EMPTY; /* next should be unreachable! */
}
break;
case REOP_FLAT1:
/*
* Coalesce adjacent FLAT1 nodes. Also coalesce FLAT1 and FLAT.
* After a single coalesce, we reuse the REOP_FLAT case's code by
* jumping into the bottom of its while loop.
*/
next = ren->next;
if (next &&
!(next->flags & RENODE_ISJOIN) &&
(REOP(next) == REOP_FLAT || REOP(next) == REOP_FLAT1)) {
if (REOP(next) == REOP_FLAT) {
cp2 = next->kid;
len = PTRDIFF((jschar *)next->u.kid2, cp2, jschar);
} else {
cp2 = &next->u.chr;
len = 1;
}
cx = state->context;
PR_ARENA_ALLOCATE(cp, &cx->tempPool, len + 2);
if (!cp) {
JS_ReportOutOfMemory(cx);
return JS_FALSE;
}
cp[0] = ren->u.chr;
js_strncpy(&cp[1], cp2, len);
cp[++len] = 0;
ren->op = REOP_FLAT;
ren->flags |= RENODE_REALLOK;
goto end_coalesce;
}
break;
default:;
}
/*
* Set ren's offset and advance progLength by ren's base size.
*/
ren->offset = (uint16) state->progLength;
state->progLength += reopsize[ren->op];
switch (REOP(ren)) {
case REOP_ALT:
case REOP_QUANT:
case REOP_STAR:
case REOP_PLUS:
case REOP_OPT:
case REOP_LPAREN:
case REOP_ANCHOR1:
/*
* Recur for nodes that have kid links to other nodes.
*/
if (!OptimizeRegExp(state, ren->kid))
return JS_FALSE;
break;
case REOP_CCLASS:
/*
* Check for a nonzero high byte or a \uXXXX escape sequence.
*/
cp = ren->kid;
cp2 = ren->u.kid2;
len = PTRDIFF(cp2, cp, jschar);
maxc = 0;
while (cp < cp2) {
c = *cp++;
if (c == '\\') {
if (cp + 5 <= cp2 && *cp == 'u' &&
JS7_ISHEX(cp[1]) && JS7_ISHEX(cp[2]) &&
JS7_ISHEX(cp[3]) && JS7_ISHEX(cp[4])) {
c = (((((JS7_UNHEX(cp[1]) << 4)
+ JS7_UNHEX(cp[2])) << 4)
+ JS7_UNHEX(cp[3])) << 4)
+ JS7_UNHEX(cp[4]);
cp += 5;
} else {
/*
* Octal and hex escapes can't be > 255. Skip this
* backslash and let the loop pass over the remaining
* escape sequence as if it were text to match.
*/
continue;
}
}
if (state->flags & JSREG_FOLD) {
/*
* Don't assume that lowercase are above uppercase, or
* that c is either even when c has upper and lowercase
* versions.
*/
if ((c2 = JS_TOUPPER(c)) > maxc)
maxc = c2;
if ((c2 = JS_TOLOWER(c2)) > maxc)
maxc = c2;
}
if (c > maxc)
maxc = c;
}
if (maxc >= CCLASS_CHARSET_SIZE) {
ren->op = (uint8)REOP_UCCLASS;
size = (size_t)(maxc + PR_BITS_PER_BYTE) / PR_BITS_PER_BYTE;
ren->u.ucclass.kidlen = (uint16)len;
ren->u.ucclass.bmsize = (uint16)size;
state->progLength -= reopsize[REOP_CCLASS];
state->progLength += reopsize[REOP_UCCLASS] + size;
}
break;
case REOP_FLAT:
/*
* FLAT takes 2 bytes plus the bytes in the string to match.
* If any character has a non-zero high byte, switch to UCFLAT
* and double the immediate operand length.
*/
cp = ren->kid;
cp2 = ren->u.kid2;
len = PTRDIFF(cp2, cp, jschar);
while (cp < cp2) {
c = *cp++;
if (c >> 8) {
ren->op = (uint8)REOP_UCFLAT;
len *= 2;
break;
}
}
state->progLength += len;
break;
case REOP_FLAT1:
c = ren->u.chr;
if (c >> 8) {
ren->op = (uint8)REOP_UCFLAT1;
state->progLength++;
}
break;
case REOP_JUMP:
/*
* Eliminate jumps to jumps.
*/
while ((next = ren->next) != NULL && REOP(next) == REOP_JUMP)
ren->next = next->next;
break;
case REOP_END:
/*
* End of program.
*/
return JS_TRUE;
default:;
}
if (!(ren->flags & RENODE_GOODNEXT))
break;
} while ((ren = ren->next) != NULL);
return JS_TRUE;
}
static JSBool
EmitRegExp(CompilerState *state, RENode *ren, JSRegExp *re)
{
REOp op;
jsbytecode *pc, fill;
RENode *next;
ptrdiff_t diff;
jschar *cp, *end, *ocp;
uintN b, c, i, j, n, lastc, foldc, nchars;
JSBool inrange;
do {
op = REOP(ren);
if (op == REOP_END)
return JS_TRUE;
pc = &re->program[state->progLength];
state->progLength += reopsize[ren->op];
pc[0] = ren->op;
next = ren->next;
switch (op) {
case REOP_ALT:
diff = next->offset - ren->offset;
SET_JUMP_OFFSET(pc, diff);
if (!EmitRegExp(state, ren->kid, re))
return JS_FALSE;
break;
case REOP_QUANT:
diff = next->offset - ren->offset;
SET_JUMP_OFFSET(pc, diff);
pc += 2;
SET_ARGNO(pc, ren->u.range.min);
pc += 2;
SET_ARGNO(pc, ren->u.range.max);
if (!EmitRegExp(state, ren->kid, re))
return JS_FALSE;
break;
case REOP_STAR:
case REOP_PLUS:
case REOP_OPT:
case REOP_ANCHOR1:
if (!EmitRegExp(state, ren->kid, re))
return JS_FALSE;
break;
case REOP_LPAREN:
SET_ARGNO(pc, ren->u.num);
if (!EmitRegExp(state, ren->kid, re))
return JS_FALSE;
break;
case REOP_RPAREN:
SET_ARGNO(pc, ren->u.num);
break;
case REOP_CCLASS:
case REOP_UCCLASS:
cp = ren->kid;
if (*cp == '^') {
pc[0] = (jsbytecode)
((op == REOP_CCLASS) ? REOP_NCCLASS : REOP_NUCCLASS);
fill = 0xff;
cp++;
} else {
fill = 0;
}
pc++;
if (op == REOP_CCLASS) {
end = ren->u.kid2;
for (i = 0; i < CCLASS_CHARSET_SIZE / PR_BITS_PER_BYTE; i++)
pc[i] = fill;
nchars = CCLASS_CHARSET_SIZE;
} else {
end = cp + ren->u.ucclass.kidlen;
n = (uintN)ren->u.ucclass.bmsize;
*pc++ = (jsbytecode)(n >> 8);
*pc++ = (jsbytecode)n;
state->progLength += n;
for (i = 0; i < n; i++)
pc[i] = fill;
nchars = n * PR_BITS_PER_BYTE;
}
/* Split ops up into statements to keep MSVC1.52 from crashing. */
#define MATCH_BIT(c) { i = (c) >> 3; b = (c) & 7; b = 1 << b; \
if (fill) pc[i] &= ~b; else pc[i] |= b; }
lastc = nchars;
inrange = JS_FALSE;
while (cp < end) {
c = (uintN) *cp++;
if (c == '\\') {
c = *cp++;
switch (c) {
case 'b':
case 'f':
case 'n':
case 'r':
case 't':
case 'v':
c = js_strchr(js_EscapeMap, (jschar)c)[-1];
break;
#define CHECK_RANGE() if (inrange) { MATCH_BIT(lastc); MATCH_BIT('-'); \
inrange = JS_FALSE; }
case 'd':
CHECK_RANGE();
for (c = '0'; c <= '9'; c++)
MATCH_BIT(c);
continue;
case 'D':
CHECK_RANGE();
for (c = 0; c < '0'; c++)
MATCH_BIT(c);
for (c = '9' + 1; c < nchars; c++)
MATCH_BIT(c);
continue;
case 'w':
CHECK_RANGE();
for (c = 0; c < nchars; c++)
if (JS_ISWORD(c))
MATCH_BIT(c);
continue;
case 'W':
CHECK_RANGE();
for (c = 0; c < nchars; c++)
if (!JS_ISWORD(c))
MATCH_BIT(c);
continue;
case 's':
CHECK_RANGE();
for (c = 0; c < nchars; c++)
if (JS_ISSPACE(c))
MATCH_BIT(c);
continue;
case 'S':
CHECK_RANGE();
for (c = 0; c < nchars; c++)
if (!JS_ISSPACE(c))
MATCH_BIT(c);
continue;
#undef CHECK_RANGE
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
n = JS7_UNDEC(c);
ocp = cp - 2;
c = *cp;
if ('0' <= c && c <= '7') {
cp++;
n = 8 * n + JS7_UNDEC(c);
c = *cp;
if ('0' <= c && c <= '7') {
cp++;
i = 8 * n + JS7_UNDEC(c);
if (i <= 0377)
n = i;
else
cp--;
}
}
c = n;
break;
case 'x':
ocp = cp;
c = *cp++;
if (JS7_ISHEX(c)) {
n = JS7_UNHEX(c);
c = *cp++;
if (JS7_ISHEX(c)) {
n <<= 4;
n += JS7_UNHEX(c);
}
} else {
cp = ocp; /* \xZZ is xZZ (Perl does \0ZZ!) */
n = 'x';
}
c = n;
break;
case 'u':
if (JS7_ISHEX(cp[0]) && JS7_ISHEX(cp[1]) &&
JS7_ISHEX(cp[2]) && JS7_ISHEX(cp[3])) {
n = (((((JS7_UNHEX(cp[0]) << 4)
+ JS7_UNHEX(cp[1])) << 4)
+ JS7_UNHEX(cp[2])) << 4)
+ JS7_UNHEX(cp[3]);
c = n;
cp += 4;
}
break;
case 'c':
c = *cp++;
c = JS_TOUPPER(c);
c = JS_TOCTRL(c);
break;
}
}
if (inrange) {
if (lastc > c) {
JS_ReportErrorNumber(state->context,
js_GetErrorMessage, NULL,
JSMSG_BAD_CLASS_RANGE);
return JS_FALSE;
}
inrange = JS_FALSE;
} else {
/* Set lastc so we match just c's bit in the for loop. */
lastc = c;
/* [balance: */
if (*cp == '-' && cp + 1 < end && cp[1] != ']') {
cp++;
inrange = JS_TRUE;
continue;
}
}
/* Match characters in the range [lastc, c]. */
for (; lastc <= c; lastc++) {
MATCH_BIT(lastc);
if (state->flags & JSREG_FOLD) {
/*
* Must do both upper and lower for Turkish dotless i,
* Georgian, etc.
*/
foldc = JS_TOUPPER(lastc);
MATCH_BIT(foldc);
foldc = JS_TOLOWER(foldc);
MATCH_BIT(foldc);
}
}
lastc = c;
}
#undef MATCH_BIT
break;
case REOP_BACKREF:
if (state->flags & JSREG_FOLD)
pc[0] = (jsbytecode)REOP_BACKREFi;
pc[1] = (jsbytecode)ren->u.num;
break;
case REOP_FLAT:
if (state->flags & JSREG_FOLD)
pc[0] = (jsbytecode)REOP_FLATi;
goto emit_flat;
case REOP_UCFLAT:
if (state->flags & JSREG_FOLD)
pc[0] = (jsbytecode)REOP_UCFLATi;
emit_flat:
cp = ren->kid;
diff = (jschar *)ren->u.kid2 - cp;
pc[1] = (jsbytecode)diff;
pc += 2;
state->progLength += diff;
if (op == REOP_UCFLAT)
state->progLength += diff;
for (i = j = 0; i < (uintN)diff; i++, j++) {
c = (uintN)cp[i];
/*
* Lay down immediate chars in native byte order so memcmp
* with a JSString's chars works.
*/
#if IS_BIG_ENDIAN
if (op == REOP_UCFLAT)
pc[j++] = (jsbytecode)(c >> 8);
#endif
pc[j] = (jsbytecode)c;
#if IS_LITTLE_ENDIAN
if (op == REOP_UCFLAT)
pc[j++] = (jsbytecode)(c >> 8);
#endif
}
break;
case REOP_FLAT1:
if (state->flags & JSREG_FOLD)
pc[0] = (jsbytecode)REOP_FLAT1i;
pc[1] = (jsbytecode)ren->u.chr;
break;
case REOP_UCFLAT1:
if (state->flags & JSREG_FOLD)
pc[0] = (jsbytecode)REOP_UCFLAT1i;
c = (uintN)ren->u.chr;
pc[1] = (jsbytecode)(c >> 8);
pc[2] = (jsbytecode)c;
break;
case REOP_JUMP:
diff = next->offset - ren->offset;
SET_JUMP_OFFSET(pc, diff);
break;
default:;
}
if (!(ren->flags & RENODE_GOODNEXT))
break;
} while ((ren = next) != NULL);
return JS_TRUE;
}
JSRegExp *
js_NewRegExp(JSContext *cx, JSString *str, uintN flags)
{
JSRegExp *re;
void *mark;
CompilerState state;
RENode *ren, *end;
size_t resize;
re = NULL;
mark = PR_ARENA_MARK(&cx->tempPool);
state.context = cx;
state.cpbegin = state.cp = str->chars;
state.flags = flags;
state.parenCount = 0;
state.progLength = 0;
ren = ParseRegExp(&state);
if (!ren)
goto out;
end = NewRENode(&state, REOP_END, NULL);
if (!end || !SetNext(&state, ren, end))
goto out;
if (!AnchorRegExp(&state, ren))
goto out;
if (!OptimizeRegExp(&state, ren))
goto out;
#ifdef DEBUG_notme
if (!DumpRegExp(cx, ren))
goto out;
#endif
resize = sizeof *re + state.progLength - 1;
re = JS_malloc(cx, PR_ROUNDUP(resize, sizeof(prword)));
if (!re)
goto out;
re->source = str;
re->length = state.progLength;
re->lastIndex = 0;
re->parenCount = state.parenCount;
re->flags = flags;
state.progLength = 0;
if (!EmitRegExp(&state, ren, re)) {
js_DestroyRegExp(cx, re);
re = NULL;
goto out;
}
#ifdef DEBUG_notme
{
/* print the compiled regexp program bytecode */
size_t i;
for (i = 0; i < state.progLength; i++) {
int b = (int) re->program[i];
fprintf(stderr, "%d", b);
if ((i > 0 && i % 8 == 0) || i == state.progLength-1)
fprintf(stderr, "\n");
else
fprintf(stderr, ", ");
}
fprintf(stderr, "\n");
}
#endif
/* Success: lock re->source string. */
(void) js_LockGCThing(cx, str);
out:
PR_ARENA_RELEASE(&cx->tempPool, mark);
return re;
}
JSRegExp *
js_NewRegExpOpt(JSContext *cx, JSString *str, JSString *opt)
{
uintN flags;
jschar *cp;
flags = 0;
if (opt) {
for (cp = opt->chars; *cp; cp++) {
switch (*cp) {
case 'g':
flags |= JSREG_GLOB;
break;
case 'i':
flags |= JSREG_FOLD;
break;
default: {
char charBuf[2] = " ";
charBuf[0] = (char)*cp;
JS_ReportErrorNumber(cx, js_GetErrorMessage, NULL,
JSMSG_BAD_FLAG, charBuf);
return NULL;
}
}
}
}
return js_NewRegExp(cx, str, flags);
}
void
js_DestroyRegExp(JSContext *cx, JSRegExp *re)
{
js_UnlockGCThing(cx, re->source);
JS_free(cx, re);
}
typedef struct MatchState {
JSContext *context; /* for access to regExpStatics */
JSBool anchoring; /* true if multiline anchoring ^/$ */
jsbytecode *pcend; /* pc limit (fencepost) */
const jschar *cpbegin, *cpend; /* cp base address and limit */
size_t start; /* offset from cpbegin to start at */
ptrdiff_t skipped; /* chars skipped anchoring this r.e. */
uintN parenCount; /* number of paren substring matches */
JSSubString *maybeParens; /* possible paren substring pointers */
JSSubString *parens; /* certain paren substring matches */
} MatchState;
/*
* Returns updated cp on match, null on mismatch.
*/
static const jschar *
MatchRegExp(MatchState *state, jsbytecode *pc, const jschar *cp)
{
jsbytecode *pc2, *pcend;
const jschar *cp2, *cp3, *cpbegin, *cpend;
REOp op;
JSBool matched;
ptrdiff_t i, oplen, altlen, matchlen;
uintN min, max, num;
JSSubString *parsub;
const jschar *parstr;
size_t parlen;
jschar c, c2;
uintN bit, byte, size;
pcend = state->pcend;
cpbegin = state->cpbegin;
cpend = state->cpend;
while (pc < pcend) {
op = (REOp) *pc;
oplen = reopsize[op];
switch (op) {
case REOP_EMPTY:
pc += oplen;
continue;
case REOP_ALT:
altlen = GET_JUMP_OFFSET(pc);
pc2 = pc + oplen;
if ((REOp)pc[altlen] != REOP_ALT) {
pc = pc2;
continue;
}
cp2 = MatchRegExp(state, pc2, cp);
if (cp2)
return cp2;
pc += altlen;
continue;
case REOP_BOL:
matched = (cp == cpbegin);
if (state->context->regExpStatics.multiline) {
/* Anchor-search only if RegExp.multiline is true. */
if (state->anchoring) {
if (!matched)
matched = (cp[-1] == '\n');
} else {
state->anchoring = JS_TRUE;
for (cp2 = cp; cp2 < cpend; cp2++) {
if (cp2 == cpbegin || cp2[-1] == '\n') {
cp3 = MatchRegExp(state, pc, cp2);
if (cp3) {
state->skipped = cp2 - cp;
state->anchoring = JS_FALSE;
return cp3;
}
}
}
state->anchoring = JS_FALSE;
}
}
matchlen = 0;
break;
case REOP_EOL:
case REOP_EOLONLY:
matched = (cp == cpend);
if (op == REOP_EOL || state->anchoring) {
if (!matched && state->context->regExpStatics.multiline)
matched = (*cp == '\n');
} else {
/* Always anchor-search EOLONLY, which has no BOL analogue. */
state->anchoring = JS_TRUE;
for (cp2 = cp; cp2 <= cpend; cp2++) {
if (cp2 == cpend || *cp2 == '\n') {
cp3 = MatchRegExp(state, pc, cp2);
if (cp3) {
state->anchoring = JS_FALSE;
state->skipped = cp2 - cp;
return cp3;
}
}
}
state->anchoring = JS_FALSE;
}
matchlen = 0;
break;
case REOP_WBDRY:
matched = (cp == cpbegin || !JS_ISWORD(cp[-1])) ^ !JS_ISWORD(*cp);
matchlen = 0;
break;
case REOP_WNONBDRY:
matched = (cp == cpbegin || !JS_ISWORD(cp[-1])) ^ JS_ISWORD(*cp);
matchlen = 0;
break;
case REOP_QUANT:
pc2 = pc;
oplen = GET_JUMP_OFFSET(pc2);
pc2 += 2;
min = GET_ARGNO(pc2);
pc2 += 2;
max = GET_ARGNO(pc2);
pc2 += 3;
/* Reduce state->pcend so we match only the quantified regexp. */
state->pcend = pc + oplen;
/* If min is non-zero, insist on at least that many matches. */
for (num = 0; num < min; num++) {
cp = MatchRegExp(state, pc2, cp);
if (!cp) {
state->pcend = pcend;
return NULL;
}
}
/* Try matches from min to max, or forever if max == 0. */
for (; !max || num < max; num++) {
cp2 = MatchRegExp(state, pc2, cp);
if (!cp2)
break;
cp = cp2;
}
/* Restore state->pcend and set match and matchlen. */
state->pcend = pcend;
matched = (min <= num && (!max || num <= max));
matchlen = 0;
break;
case REOP_LPAREN:
num = GET_ARGNO(pc);
parsub = &state->maybeParens[num];
parstr = parsub->chars;
parsub->chars = cp;
pc += oplen;
cp3 = MatchRegExp(state, pc, cp);
if (!cp3) {
/* Restore so later backrefs work, unlike Perl4. */
parsub->chars = parstr;
return NULL;
}
parsub = &state->parens[num];
if (!parsub->chars) {
cp2 = cpbegin + state->start + state->skipped;
if (cp < cp2) {
parsub->chars = cp2;
parsub->length -= cp2 - cp;
} else {
parsub->chars = cp;
}
}
return cp3;
case REOP_RPAREN:
num = GET_ARGNO(pc);
parsub = &state->maybeParens[num];
parsub->length = parlen = cp - parsub->chars;
pc += oplen;
cp = MatchRegExp(state, pc, cp);
if (cp) {
parsub = &state->parens[num];
if (!parsub->chars)
parsub->length = parlen;
if (num >= state->parenCount)
state->parenCount = num + 1;
}
return cp;
case REOP_BACKREF:
num = (uintN)pc[1];
parsub = &state->maybeParens[num];
matchlen = (ptrdiff_t)parsub->length;
matched = (cp + matchlen <= cpend &&
!memcmp(cp, parsub->chars, matchlen * sizeof(jschar)));
break;
/*
* See java.lang.String for more on why both toupper and tolower are needed, in
* comments for equalsIgnoreCase and regionMatches(boolean ignoreCase, ...).
*/
#define MATCH_CHARS_IGNORING_CASE(c, c2) \
((c) == (c2) || \
(c = JS_TOUPPER(c)) == (c2 = JS_TOUPPER(c2)) || \
JS_TOLOWER(c) == JS_TOLOWER(c2))
case REOP_BACKREFi:
num = (uintN)pc[1];
parsub = &state->maybeParens[num];
matchlen = (ptrdiff_t)parsub->length;
matched = (cp + matchlen <= cpend);
if (matched) {
for (i = 0; i < matchlen; i++) {
c = cp[i];
c2 = parsub->chars[i];
matched = MATCH_CHARS_IGNORING_CASE(c, c2);
if (!matched)
break;
}
}
break;
#define SINGLE_CASES \
case REOP_DOT: \
matched = (cp != cpend && *cp != '\n'); \
matchlen = 1; \
break; \
\
NONDOT_SINGLE_CASES \
/* END SINGLE_CASES */
#define NONDOT_SINGLE_CASES \
case REOP_CCLASS: \
case REOP_NCCLASS: \
c = *cp; \
if (c >= CCLASS_CHARSET_SIZE) { \
matched = (op == REOP_NCCLASS); \
} else { \
byte = (uintN)c >> 3; \
bit = c & 7; \
bit = 1 << bit; \
matched = pc[1 + byte] & bit; \
} \
matchlen = 1; \
break; \
\
case REOP_DIGIT: \
matched = JS_ISDIGIT(*cp); \
matchlen = 1; \
break; \
\
case REOP_NONDIGIT: \
matched = !JS_ISDIGIT(*cp); \
matchlen = 1; \
break; \
\
case REOP_ALNUM: \
matched = JS_ISWORD(*cp); \
matchlen = 1; \
break; \
\
case REOP_NONALNUM: \
matched = !JS_ISWORD(*cp); \
matchlen = 1; \
break; \
\
case REOP_SPACE: \
matched = JS_ISSPACE(*cp); \
matchlen = 1; \
break; \
\
case REOP_NONSPACE: \
matched = !JS_ISSPACE(*cp); \
matchlen = 1; \
break; \
\
case REOP_FLAT1: \
c = *cp; \
c2 = (jschar)pc[1]; \
matched = (c == c2); \
matchlen = 1; \
break; \
\
case REOP_FLAT1i: \
c = *cp; \
c2 = (jschar)pc[1]; \
matched = MATCH_CHARS_IGNORING_CASE(c, c2); \
matchlen = 1; \
break; \
\
case REOP_UCFLAT1: \
c = *cp; \
c2 = ((pc[1] << 8) | pc[2]); \
matched = (c == c2); \
matchlen = 1; \
break; \
\
case REOP_UCFLAT1i: \
c = *cp; \
c2 = ((pc[1] << 8) | pc[2]); \
matched = MATCH_CHARS_IGNORING_CASE(c, c2); \
matchlen = 1; \
break; \
\
case REOP_UCCLASS: \
case REOP_NUCCLASS: \
size = (pc[1] << 8) | pc[2]; \
oplen += size; \
c = *cp; \
byte = (uintN)c >> 3; \
if (byte >= size) { \
matched = (op == REOP_NUCCLASS); \
} else { \
bit = c & 7; \
bit = 1 << bit; \
matched = pc[3 + byte] & bit; \
} \
matchlen = 1; \
break; \
/* END NONDOT_SINGLE_CASES */
/*
* Macro-expand single-char/single-opcode cases here and below.
*/
SINGLE_CASES
case REOP_STAR:
op = (REOp) *++pc;
oplen = reopsize[op];
for (cp2 = cp; cp < cpend; cp++) {
switch (op) {
NONDOT_SINGLE_CASES
default:
PR_ASSERT(0);
}
if (!matched)
break;
}
backtracker:
pc += oplen;
do {
cp3 = MatchRegExp(state, pc, cp);
if (cp3)
return cp3;
} while (--cp >= cp2);
return NULL;
case REOP_PLUS:
op = (REOp) *++pc;
oplen = reopsize[op];
for (cp2 = cp; cp < cpend; cp++) {
switch (op) {
SINGLE_CASES
default:
PR_ASSERT(0);
}
if (!matched)
break;
}
if (cp == cp2) {
/* Did not match once, hope for an alternative. */
return NULL;
}
/* Matched one or more times, try rest of regexp. */
cp2++;
goto backtracker;
case REOP_OPT:
op = (REOp) *++pc;
oplen = reopsize[op];
switch (op) {
SINGLE_CASES
default:
PR_ASSERT(0);
}
pc += oplen;
if (matched) {
cp2 = MatchRegExp(state, pc, cp + 1);
if (cp2)
return cp2;
}
continue;
case REOP_FLAT:
matchlen = (ptrdiff_t)pc[1];
oplen += matchlen;
matched = (cp + matchlen <= cpend);
if (matched) {
pc2 = pc + 2;
for (i = 0; i < matchlen; i++) {
matched = (cp[i] == (jschar)pc2[i]);
if (!matched)
break;
}
}
break;
case REOP_FLATi:
matchlen = (ptrdiff_t)pc[1];
oplen += matchlen;
matched = (cp + matchlen <= cpend);
if (matched) {
pc2 = pc + 2;
for (i = 0; i < matchlen; i++) {
c = cp[i];
c2 = (jschar)pc2[i];
matched = MATCH_CHARS_IGNORING_CASE(c, c2);
if (!matched)
break;
}
}
break;
case REOP_UCFLAT:
matchlen = (ptrdiff_t)pc[1];
oplen += 2 * matchlen;
matched = (cp + matchlen <= cpend &&
!memcmp(cp, pc + 2, matchlen * sizeof(jschar)));
break;
case REOP_UCFLATi:
matchlen = (ptrdiff_t)pc[1];
oplen += matchlen;
matched = (cp + matchlen <= cpend);
if (matched) {
pc2 = pc + 2;
for (i = 0; i < matchlen; i++) {
c = cp[i];
#if IS_BIG_ENDIAN
c2 = *pc2++ << 8;
c2 |= *pc2++;
#endif
#if IS_LITTLE_ENDIAN
c2 = *pc2++;
c2 |= *pc2++ << 8;
#endif
matched = MATCH_CHARS_IGNORING_CASE(c, c2);
if (!matched)
break;
}
}
break;
case REOP_JUMP:
oplen = GET_JUMP_OFFSET(pc);
pc += oplen;
continue;
case REOP_DOTSTAR:
for (cp2 = cp; cp2 < cpend; cp2++)
if (*cp2 == '\n')
break;
for (pc2 = pc + oplen; cp2 >= cp; cp2--) {
cp3 = MatchRegExp(state, pc2, cp2);
if (cp3)
return cp3;
}
return NULL;
case REOP_ANCHOR:
pc2 = pc + oplen;
if (pc2 == pcend)
break;
for (cp2 = cp; cp2 < cpend; cp2++) {
cp3 = MatchRegExp(state, pc2, cp2);
if (cp3) {
state->skipped = cp2 - cp;
return cp3;
}
}
return NULL;
case REOP_ANCHOR1:
op = (REOp) *++pc;
oplen = reopsize[op];
pc2 = pc + oplen;
PR_ASSERT(pc2 < pcend);
for (cp2 = cp; cp < cpend; cp++) {
switch (op) {
NONDOT_SINGLE_CASES
default:
PR_ASSERT(0);
}
if (matched) {
cp3 = MatchRegExp(state, pc2, cp);
if (cp3) {
state->skipped = cp - cp2;
return cp3;
}
}
}
return NULL;
#undef MATCH_CHARS_IGNORING_CASE
#undef SINGLE_CASES
#undef NONDOT_SINGLE_CASES
default:
PR_ASSERT(0);
return NULL;
}
if (!matched)
return NULL;
pc += oplen;
if (matchlen) {
cp += matchlen;
if (cp > cpend)
cp = cpend;
}
}
return cp;
}
JSBool
js_ExecuteRegExp(JSContext *cx, JSRegExp *re, JSString *str, size_t *indexp,
JSBool test, jsval *rval)
{
MatchState state;
jsbytecode *pc;
const jschar *cp, *ep;
size_t i, length, start;
void *mark;
JSSubString *parsub, *morepar;
JSBool ok;
JSRegExpStatics *res;
ptrdiff_t matchlen;
uintN num, morenum;
JSString *parstr, *matchstr;
JSObject *obj;
/*
* Initialize a state struct to minimize recursive argument traffic.
*/
state.context = cx;
state.anchoring = JS_FALSE;
pc = re->program;
state.pcend = pc + re->length;
/*
* It's safe to load from cp because JSStrings have a zero at the end,
* and we never let cp get beyond cpend.
*/
start = *indexp;
if (start > str->length)
start = str->length;
cp = str->chars + start;
state.cpbegin = str->chars;
state.cpend = str->chars + str->length;
state.start = start;
state.skipped = 0;
/*
* Use the temporary arena pool to grab space for parenthetical matches.
* After the PR_ARENA_ALLOCATE early return on error, goto out to be sure
* to free this memory.
*/
length = 2 * sizeof(JSSubString) * re->parenCount;
mark = PR_ARENA_MARK(&cx->tempPool);
PR_ARENA_ALLOCATE(parsub, &cx->tempPool, length);
if (!parsub) {
JS_ReportOutOfMemory(cx);
return JS_FALSE;
}
memset(parsub, 0, length);
state.parenCount = 0;
state.maybeParens = parsub;
state.parens = parsub + re->parenCount;
ok = JS_TRUE;
/*
* Call the recursive matcher to do the real work. Return null on mismatch
* whether testing or not. On match, return an extended Array object.
*/
cp = MatchRegExp(&state, pc, cp);
if (!cp) {
*rval = JSVAL_NULL;
goto out;
}
i = PTRDIFF(cp, state.cpbegin, jschar);
*indexp = i;
matchlen = i - (start + state.skipped);
ep = cp;
cp -= matchlen;
if (test) {
/*
* Testing for a match and updating cx->regExpStatics: don't allocate
* an array object, do return true.
*/
*rval = JSVAL_TRUE;
} else {
/*
* The array returned on match has element 0 bound to the matched
* string, elements 1 through state.parenCount bound to the paren
* matches, an index property telling the length of the left context,
* and an input property referring to the input string.
*/
obj = js_NewArrayObject(cx, 0, NULL);
if (!obj) {
ok = JS_FALSE;
goto out;
}
*rval = OBJECT_TO_JSVAL(obj);
#define DEFVAL(val, id) { \
ok = js_DefineProperty(cx, obj, id, val, \
JS_PropertyStub, JS_PropertyStub, \
JSPROP_ENUMERATE, NULL); \
if (!ok) { \
cx->newborn[GCX_OBJECT] = NULL; \
cx->newborn[GCX_STRING] = NULL; \
goto out; \
} \
}
matchstr = js_NewStringCopyN(cx, cp, matchlen, 0);
if (!matchstr) {
cx->newborn[GCX_OBJECT] = NULL;
ok = JS_FALSE;
goto out;
}
DEFVAL(STRING_TO_JSVAL(matchstr), INT_TO_JSVAL(0));
}
res = &cx->regExpStatics;
PR_ASSERT(state.parenCount <= re->parenCount);
if (state.parenCount == 0) {
res->parenCount = 0;
res->lastParen = js_EmptySubString;
} else {
for (num = 0; num < state.parenCount; num++) {
parsub = &state.parens[num];
if (num < 9) {
res->parens[num] = *parsub;
} else {
morenum = num - 9;
morepar = res->moreParens;
if (!morepar) {
res->moreLength = 10;
morepar = JS_malloc(cx, 10 * sizeof(JSSubString));
} else if (morenum > res->moreLength) {
res->moreLength += 10;
morepar = JS_realloc(cx, morepar,
res->moreLength * sizeof(JSSubString));
}
if (!morepar) {
cx->newborn[GCX_OBJECT] = NULL;
cx->newborn[GCX_STRING] = NULL;
ok = JS_FALSE;
goto out;
}
res->moreParens = morepar;
morepar[morenum] = *parsub;
}
if (test)
continue;
parstr = js_NewStringCopyN(cx, parsub->chars, parsub->length, 0);
if (!parstr) {
cx->newborn[GCX_OBJECT] = NULL;
cx->newborn[GCX_STRING] = NULL;
ok = JS_FALSE;
goto out;
}
ok = js_DefineProperty(cx, obj, INT_TO_JSVAL(num + 1),
STRING_TO_JSVAL(parstr), NULL, NULL,
JSPROP_ENUMERATE, NULL);
if (!ok) {
cx->newborn[GCX_OBJECT] = NULL;
cx->newborn[GCX_STRING] = NULL;
goto out;
}
}
res->parenCount = num;
res->lastParen = *parsub;
}
if (!test) {
/*
* Define the index and input properties last for better for/in loop
* order (so they come after the elements).
*/
DEFVAL(INT_TO_JSVAL(start + state.skipped),
(jsid)cx->runtime->atomState.indexAtom);
DEFVAL(STRING_TO_JSVAL(str),
(jsid)cx->runtime->atomState.inputAtom);
}
#undef DEFVAL
res->lastMatch.chars = cp;
res->lastMatch.length = matchlen;
if (cx->version == JSVERSION_1_2) {
/*
* JS1.2 emulated Perl4.0.1.8 (patch level 36) for global regexps used
* in scalar contexts, and unintentionally for the string.match "list"
* psuedo-context. On "hi there bye", the following would result:
*
* Language while(/ /g){print("$`");} s/ /$`/g
* perl4.036 "hi", "there" "hihitherehi therebye"
* perl5 "hi", "hi there" "hihitherehi therebye"
* js1.2 "hi", "there" "hihitheretherebye"
*/
res->leftContext.chars = str->chars + start;
res->leftContext.length = state.skipped;
} else {
/*
* For JS1.3 and ECMAv2, emulate Perl5 exactly:
*
* js1.3 "hi", "hi there" "hihitherehi therebye"
*/
res->leftContext.chars = str->chars;
res->leftContext.length = start + state.skipped;
}
res->rightContext.chars = ep;
res->rightContext.length = state.cpend - ep;
out:
PR_ARENA_RELEASE(&cx->tempPool, mark);
return ok;
}
/************************************************************************/
enum regexp_tinyid {
REGEXP_SOURCE = -1,
REGEXP_GLOBAL = -2,
REGEXP_IGNORE_CASE = -3,
REGEXP_LAST_INDEX = -4
};
static JSPropertySpec regexp_props[] = {
{"source", REGEXP_SOURCE, JSPROP_ENUMERATE | JSPROP_READONLY},
{"global", REGEXP_GLOBAL, JSPROP_ENUMERATE | JSPROP_READONLY},
{"ignoreCase", REGEXP_IGNORE_CASE, JSPROP_ENUMERATE | JSPROP_READONLY},
{"lastIndex", REGEXP_LAST_INDEX, JSPROP_ENUMERATE},
{0}
};
static JSBool
regexp_getProperty(JSContext *cx, JSObject *obj, jsval id, jsval *vp)
{
jsint slot;
JSRegExp *re;
if (!JSVAL_IS_INT(id))
return JS_TRUE;
slot = JSVAL_TO_INT(id);
JS_LOCK_OBJ(cx, obj);
re = JS_GetInstancePrivate(cx, obj, &js_RegExpClass, NULL);
if (re) {
switch (slot) {
case REGEXP_SOURCE:
*vp = STRING_TO_JSVAL(re->source);
break;
case REGEXP_GLOBAL:
*vp = BOOLEAN_TO_JSVAL((re->flags & JSREG_GLOB) != 0);
break;
case REGEXP_IGNORE_CASE:
*vp = BOOLEAN_TO_JSVAL((re->flags & JSREG_FOLD) != 0);
break;
case REGEXP_LAST_INDEX:
*vp = INT_TO_JSVAL((jsint)re->lastIndex);
break;
}
}
JS_UNLOCK_OBJ(cx, obj);
return JS_TRUE;
}
static JSBool
regexp_setProperty(JSContext *cx, JSObject *obj, jsval id, jsval *vp)
{
jsint slot;
JSRegExp *re;
jsdouble d;
if (!JSVAL_IS_INT(id))
return JS_TRUE;
slot = JSVAL_TO_INT(id);
JS_LOCK_OBJ(cx, obj);
re = JS_GetInstancePrivate(cx, obj, &js_RegExpClass, NULL);
if (re && slot == REGEXP_LAST_INDEX) {
if (!js_ValueToNumber(cx, *vp, &d))
return JS_FALSE;
re->lastIndex = (size_t)d;
}
JS_UNLOCK_OBJ(cx, obj);
return JS_TRUE;
}
/*
* RegExp class static properties and their Perl counterparts:
*
* RegExp.input $_
* RegExp.multiline $*
* RegExp.lastMatch $&
* RegExp.lastParen $+
* RegExp.leftContext $`
* RegExp.rightContext $'
*/
enum regexp_static_tinyid {
REGEXP_STATIC_INPUT = -1,
REGEXP_STATIC_MULTILINE = -2,
REGEXP_STATIC_LAST_MATCH = -3,
REGEXP_STATIC_LAST_PAREN = -4,
REGEXP_STATIC_LEFT_CONTEXT = -5,
REGEXP_STATIC_RIGHT_CONTEXT = -6
};
JSBool
js_InitRegExpStatics(JSContext *cx, JSRegExpStatics *res)
{
JS_ClearRegExpStatics(cx);
return js_AddRoot(cx, &res->input, "res->input");
}
void
js_FreeRegExpStatics(JSContext *cx, JSRegExpStatics *res)
{
if (res->moreParens) {
JS_free(cx, res->moreParens);
res->moreParens = NULL;
}
js_RemoveRoot(cx, &res->input);
}
static JSBool
regexp_static_getProperty(JSContext *cx, JSObject *obj, jsval id, jsval *vp)
{
jsint slot;
JSRegExpStatics *res;
JSString *str;
JSSubString *sub;
res = &cx->regExpStatics;
if (!JSVAL_IS_INT(id))
return JS_TRUE;
slot = JSVAL_TO_INT(id);
switch (slot) {
case REGEXP_STATIC_INPUT:
*vp = res->input ? STRING_TO_JSVAL(res->input)
: JS_GetEmptyStringValue(cx);
return JS_TRUE;
case REGEXP_STATIC_MULTILINE:
*vp = BOOLEAN_TO_JSVAL(res->multiline);
return JS_TRUE;
case REGEXP_STATIC_LAST_MATCH:
sub = &res->lastMatch;
break;
case REGEXP_STATIC_LAST_PAREN:
sub = &res->lastParen;
break;
case REGEXP_STATIC_LEFT_CONTEXT:
sub = &res->leftContext;
break;
case REGEXP_STATIC_RIGHT_CONTEXT:
sub = &res->rightContext;
break;
default:
sub = REGEXP_PAREN_SUBSTRING(res, slot);
break;
}
str = js_NewStringCopyN(cx, sub->chars, sub->length, 0);
if (!str)
return JS_FALSE;
*vp = STRING_TO_JSVAL(str);
return JS_TRUE;
}
static JSBool
regexp_static_setProperty(JSContext *cx, JSObject *obj, jsval id, jsval *vp)
{
JSRegExpStatics *res;
if (!JSVAL_IS_INT(id))
return JS_TRUE;
res = &cx->regExpStatics;
/* XXX use if-else rather than switch to keep MSVC1.52 from crashing */
if (JSVAL_TO_INT(id) == REGEXP_STATIC_INPUT) {
if (!JSVAL_IS_STRING(*vp) &&
!JS_ConvertValue(cx, *vp, JSTYPE_STRING, vp)) {
return JS_FALSE;
}
res->input = JSVAL_TO_STRING(*vp);
} else if (JSVAL_TO_INT(id) == REGEXP_STATIC_MULTILINE) {
if (!JSVAL_IS_BOOLEAN(*vp) &&
!JS_ConvertValue(cx, *vp, JSTYPE_BOOLEAN, vp)) {
return JS_FALSE;
}
res->multiline = JSVAL_TO_BOOLEAN(*vp);
}
return JS_TRUE;
}
static JSPropertySpec regexp_static_props[] = {
{"input",
REGEXP_STATIC_INPUT, JSPROP_ENUMERATE,
regexp_static_getProperty, regexp_static_setProperty},
{"multiline",
REGEXP_STATIC_MULTILINE, JSPROP_ENUMERATE,
regexp_static_getProperty, regexp_static_setProperty},
{"lastMatch",
REGEXP_STATIC_LAST_MATCH, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
{"lastParen",
REGEXP_STATIC_LAST_PAREN, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
{"leftContext",
REGEXP_STATIC_LEFT_CONTEXT, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
{"rightContext",
REGEXP_STATIC_RIGHT_CONTEXT, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
/* XXX should have block scope and local $1, etc. */
{"$1", 0, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
{"$2", 1, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
{"$3", 2, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
{"$4", 3, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
{"$5", 4, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
{"$6", 5, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
{"$7", 6, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
{"$8", 7, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
{"$9", 8, JSPROP_ENUMERATE|JSPROP_READONLY,
regexp_static_getProperty, regexp_static_getProperty},
{0}
};
static void
regexp_finalize(JSContext *cx, JSObject *obj)
{
JSRegExp *re;
re = JS_GetPrivate(cx, obj);
if (!re)
return;
js_DestroyRegExp(cx, re);
}
/* Forward static prototype. */
static JSBool
regexp_exec(JSContext *cx, JSObject *obj, uintN argc, jsval *argv,
jsval *rval);
static JSBool
regexp_call(JSContext *cx, JSObject *obj, uintN argc, jsval *argv, jsval *rval)
{
return regexp_exec(cx, JSVAL_TO_OBJECT(argv[-2]), argc, argv, rval);
}
#if JS_HAS_XDR
#include "jsxdrapi.h"
static JSBool
regexp_xdrObject(JSXDRState *xdr, JSObject **objp)
{
JSRegExp *re;
JSString *source;
uint8 flags;
if (xdr->mode == JSXDR_ENCODE) {
re = JS_GetPrivate(xdr->cx, *objp);
if (!re)
return JS_FALSE;
source = re->source;
flags = re->flags;
}
if (!JS_XDRString(xdr, &source) ||
!JS_XDRUint8(xdr, &flags)) {
return JS_FALSE;
}
if (xdr->mode == JSXDR_DECODE) {
*objp = js_NewObject(xdr->cx, &js_RegExpClass, NULL, NULL);
if (!*objp)
return JS_FALSE;
re = js_NewRegExp(xdr->cx, source, flags);
if (!re)
return JS_FALSE;
if (!JS_SetPrivate(xdr->cx, *objp, re)) {
js_DestroyRegExp(xdr->cx, re);
return JS_FALSE;
}
}
return JS_TRUE;
}
#else /* !JS_HAS_XDR */
#define regexp_xdrObject NULL
#endif /* !JS_HAS_XDR */
JSClass js_RegExpClass = {
"RegExp",
JSCLASS_HAS_PRIVATE,
JS_PropertyStub, JS_PropertyStub, regexp_getProperty, regexp_setProperty,
JS_EnumerateStub, JS_ResolveStub, JS_ConvertStub, regexp_finalize,
NULL, NULL, regexp_call, NULL,
regexp_xdrObject,
};
static JSBool
regexp_toString(JSContext *cx, JSObject *obj, uintN argc, jsval *argv,
jsval *rval)
{
JSBool ok;
JSRegExp *re;
jschar *chars;
size_t length, nflags;
uintN flags;
JSString *str;
if (!JS_InstanceOf(cx, obj, &js_RegExpClass, argv))
return JS_FALSE;
ok = JS_TRUE;
JS_LOCK_OBJ(cx, obj);
re = JS_GetPrivate(cx, obj);
if (!re) {
*rval = STRING_TO_JSVAL(cx->runtime->emptyString);
goto out;
}
length = re->source->length + 2;
nflags = 0;
for (flags = re->flags; flags != 0; flags &= flags - 1)
nflags++;
chars = JS_malloc(cx, (length + nflags + 1) * sizeof(jschar));
if (!chars) {
ok = JS_FALSE;
goto out;
}
chars[0] = '/';
js_strncpy(&chars[1], re->source->chars, length - 2);
chars[length-1] = '/';
if (nflags) {
if (re->flags & JSREG_GLOB)
chars[length++] = 'g';
if (re->flags & JSREG_FOLD)
chars[length++] = 'i';
}
chars[length] = 0;
str = js_NewString(cx, chars, length, 0);
if (!str) {
JS_free(cx, chars);
ok = JS_FALSE;
goto out;
}
*rval = STRING_TO_JSVAL(str);
out:
JS_UNLOCK_OBJ(cx, obj);
return ok;
}
static JSBool
regexp_compile(JSContext *cx, JSObject *obj, uintN argc, jsval *argv,
jsval *rval)
{
JSString *opt, *str;
JSRegExp *oldre, *re;
JSBool ok;
if (!JS_InstanceOf(cx, obj, &js_RegExpClass, argv))
return JS_FALSE;
opt = NULL;
JS_LOCK_OBJ(cx, obj);
if (argc == 0) {
str = cx->runtime->emptyString;
} else {
str = js_ValueToString(cx, argv[0]);
if (!str) {
ok = JS_FALSE;
goto out;
}
argv[0] = STRING_TO_JSVAL(str);
if (argc > 1) {
opt = js_ValueToString(cx, argv[1]);
if (!opt) {
ok = JS_FALSE;
goto out;
}
argv[1] = STRING_TO_JSVAL(opt);
}
}
re = js_NewRegExpOpt(cx, str, opt);
if (!re) {
ok = JS_FALSE;
goto out;
}
oldre = JS_GetPrivate(cx, obj);
ok = JS_SetPrivate(cx, obj, re);
if (!ok) {
js_DestroyRegExp(cx, re);
goto out;
}
if (oldre)
js_DestroyRegExp(cx, oldre);
*rval = OBJECT_TO_JSVAL(obj);
out:
JS_UNLOCK_OBJ(cx, obj);
return ok;
}
static JSBool
regexp_exec_sub(JSContext *cx, JSObject *obj, uintN argc, jsval *argv,
JSBool test, jsval *rval)
{
JSBool ok, locked;
JSRegExp *re;
JSString *str;
size_t i;
if (!JS_InstanceOf(cx, obj, &js_RegExpClass, argv))
return JS_FALSE;
re = JS_GetPrivate(cx, obj);
if (!re)
return JS_TRUE;
ok = locked = JS_FALSE;
if (argc == 0) {
str = cx->regExpStatics.input;
if (!str) {
JS_ReportErrorNumber(cx, js_GetErrorMessage, NULL,
JSMSG_NO_INPUT,
JS_GetStringBytes(re->source),
(re->flags & JSREG_GLOB) ? "g" : "",
(re->flags & JSREG_FOLD) ? "i" : "");
goto out;
}
} else {
str = js_ValueToString(cx, argv[0]);
if (!str)
goto out;
argv[0] = STRING_TO_JSVAL(str);
}
if (re->flags & JSREG_GLOB) {
JS_LOCK_OBJ(cx, obj);
locked = JS_TRUE;
i = re->lastIndex;
} else {
i = 0;
}
ok = js_ExecuteRegExp(cx, re, str, &i, test, rval);
if (re->flags & JSREG_GLOB)
re->lastIndex = (*rval == JSVAL_NULL) ? 0 : i;
out:
if (locked)
JS_UNLOCK_OBJ(cx, obj);
return ok;
}
static JSBool
regexp_exec(JSContext *cx, JSObject *obj, uintN argc, jsval *argv, jsval *rval)
{
return regexp_exec_sub(cx, obj, argc, argv, JS_FALSE, rval);
}
static JSBool
regexp_test(JSContext *cx, JSObject *obj, uintN argc, jsval *argv, jsval *rval)
{
if (!regexp_exec_sub(cx, obj, argc, argv, JS_TRUE, rval))
return JS_FALSE;
if (*rval != JSVAL_TRUE)
*rval = JSVAL_FALSE;
return JS_TRUE;
}
static JSFunctionSpec regexp_methods[] = {
#if JS_HAS_TOSOURCE
{js_toSource_str, regexp_toString, 0},
#endif
{js_toString_str, regexp_toString, 0},
{"compile", regexp_compile, 1},
{"exec", regexp_exec, 0},
{"test", regexp_test, 0},
{0}
};
static JSBool
RegExp(JSContext *cx, JSObject *obj, uintN argc, jsval *argv, jsval *rval)
{
/* If not constructing, replace obj with a new RegExp object. */
if (!cx->fp->constructing) {
obj = js_NewObject(cx, &js_RegExpClass, NULL, NULL);
if (!obj)
return JS_FALSE;
}
return regexp_compile(cx, obj, argc, argv, rval);
}
JSObject *
js_InitRegExpClass(JSContext *cx, JSObject *obj)
{
JSObject *proto, *ctor;
proto = JS_InitClass(cx, obj, NULL, &js_RegExpClass, RegExp, 1,
regexp_props, regexp_methods,
regexp_static_props, NULL);
if (!proto || !(ctor = JS_GetConstructor(cx, proto)))
return NULL;
if (!JS_AliasProperty(cx, ctor, "input", "$_") ||
!JS_AliasProperty(cx, ctor, "multiline", "$*") ||
!JS_AliasProperty(cx, ctor, "lastMatch", "$&") ||
!JS_AliasProperty(cx, ctor, "lastParen", "$+") ||
!JS_AliasProperty(cx, ctor, "leftContext", "$`") ||
!JS_AliasProperty(cx, ctor, "rightContext", "$'")) {
goto bad;
}
return proto;
bad:
JS_DeleteProperty(cx, obj, js_RegExpClass.name);
return NULL;
}
JSObject *
js_NewRegExpObject(JSContext *cx, jschar *chars, size_t length, uintN flags)
{
JSString *str;
JSObject *obj;
JSRegExp *re;
str = js_NewStringCopyN(cx, chars, length, 0);
if (!str)
return NULL;
re = js_NewRegExp(cx, str, flags);
if (!re)
return NULL;
obj = js_NewObject(cx, &js_RegExpClass, NULL, NULL);
if (!obj || !JS_SetPrivate(cx, obj, re)) {
js_DestroyRegExp(cx, re);
return NULL;
}
return obj;
}
#endif /* JS_HAS_REGEXPS */
Reference in New Issue View Git Blame Copy Permalink