DUK_INTERNAL void duk_regexp_compile(duk_hthread *thr) {
duk_context *ctx = (duk_context *) thr;
duk_re_compiler_ctx re_ctx;
duk_lexer_point lex_point;
duk_hstring *h_pattern;
duk_hstring *h_flags;
duk__re_disjunction_info ign_disj;
DUK_ASSERT(thr != NULL);
DUK_ASSERT(ctx != NULL);
/*
* Args validation
*/
/* TypeError if fails */
h_pattern = duk_require_hstring(ctx, -2);
h_flags = duk_require_hstring(ctx, -1);
/*
* Create normalized 'source' property (E5 Section 15.10.3).
*/
/* [ ... pattern flags ] */
duk__create_escaped_source(thr, -2);
/* [ ... pattern flags escaped_source ] */
/*
* Init compilation context
*/
/* [ ... pattern flags escaped_source buffer ] */
DUK_MEMZERO(&re_ctx, sizeof(re_ctx));
DUK_LEXER_INITCTX(&re_ctx.lex); /* duplicate zeroing, expect for (possible) NULL inits */
re_ctx.thr = thr;
re_ctx.lex.thr = thr;
re_ctx.lex.input = DUK_HSTRING_GET_DATA(h_pattern);
re_ctx.lex.input_length = DUK_HSTRING_GET_BYTELEN(h_pattern);
re_ctx.lex.token_limit = DUK_RE_COMPILE_TOKEN_LIMIT;
re_ctx.recursion_limit = DUK_USE_REGEXP_COMPILER_RECLIMIT;
re_ctx.re_flags = duk__parse_regexp_flags(thr, h_flags);
DUK_BW_INIT_PUSHBUF(thr, &re_ctx.bw, DUK__RE_INITIAL_BUFSIZE);
DUK_DD(DUK_DDPRINT("regexp compiler ctx initialized, flags=0x%08lx, recursion_limit=%ld",
(unsigned long) re_ctx.re_flags, (long) re_ctx.recursion_limit));
/*
* Init lexer
*/
lex_point.offset = 0; /* expensive init, just want to fill window */
lex_point.line = 1;
DUK_LEXER_SETPOINT(&re_ctx.lex, &lex_point);
/*
* Compilation
*/
DUK_DD(DUK_DDPRINT("starting regexp compilation"));
duk__append_u32(&re_ctx, DUK_REOP_SAVE);
duk__append_u32(&re_ctx, 0);
duk__parse_disjunction(&re_ctx, 1 /*expect_eof*/, &ign_disj);
duk__append_u32(&re_ctx, DUK_REOP_SAVE);
duk__append_u32(&re_ctx, 1);
duk__append_u32(&re_ctx, DUK_REOP_MATCH);
/*
* Check for invalid backreferences; note that it is NOT an error
* to back-reference a capture group which has not yet been introduced
* in the pattern (as in /\1(foo)/); in fact, the backreference will
* always match! It IS an error to back-reference a capture group
* which will never be introduced in the pattern. Thus, we can check
* for such references only after parsing is complete.
*/
if (re_ctx.highest_backref > re_ctx.captures) {
DUK_ERROR_SYNTAX(thr, DUK_STR_INVALID_BACKREFS);
}
/*
* Emit compiled regexp header: flags, ncaptures
* (insertion order inverted on purpose)
*/
duk__insert_u32(&re_ctx, 0, (re_ctx.captures + 1) * 2);
duk__insert_u32(&re_ctx, 0, re_ctx.re_flags);
/* [ ... pattern flags escaped_source buffer ] */
DUK_BW_COMPACT(thr, &re_ctx.bw);
duk_to_string(ctx, -1); /* coerce to string */
/* [ ... pattern flags escaped_source bytecode ] */
/*
* Finalize stack
*/
duk_remove(ctx, -4); /* -> [ ... flags escaped_source bytecode ] */
duk_remove(ctx, -3); /* -> [ ... escaped_source bytecode ] */
DUK_DD(DUK_DDPRINT("regexp compilation successful, bytecode: %!T, escaped source: %!T",
(duk_tval *) duk_get_tval(ctx, -1), (duk_tval *) duk_get_tval(ctx, -2)));
}
void duk_regexp_compile(duk_hthread *thr) {
duk_context *ctx = (duk_context *) thr;
duk_re_compiler_ctx re_ctx;
duk_lexer_point lex_point;
duk_hstring *h_pattern;
duk_hstring *h_flags;
duk_hbuffer_dynamic *h_buffer;
DUK_ASSERT(thr != NULL);
DUK_ASSERT(ctx != NULL);
/*
* Args validation
*/
/* TypeError if fails */
h_pattern = duk_require_hstring(ctx, -2);
h_flags = duk_require_hstring(ctx, -1);
/*
* Create normalized 'source' property (E5 Section 15.10.3).
*/
/* [ ... pattern flags ] */
create_escaped_source(thr, -2);
/* [ ... pattern flags escaped_source ] */
/*
* Init compilation context
*/
duk_push_dynamic_buffer(ctx, 0);
h_buffer = (duk_hbuffer_dynamic *) duk_require_hbuffer(ctx, -1);
DUK_ASSERT(DUK_HBUFFER_HAS_DYNAMIC(h_buffer));
/* [ ... pattern flags escaped_source buffer ] */
DUK_MEMSET(&re_ctx, 0, sizeof(re_ctx));
DUK_LEXER_INITCTX(&re_ctx.lex); /* duplicate zeroing, expect for (possible) NULL inits */
re_ctx.thr = thr;
re_ctx.lex.thr = thr;
re_ctx.lex.input = DUK_HSTRING_GET_DATA(h_pattern);
re_ctx.lex.input_length = DUK_HSTRING_GET_BYTELEN(h_pattern);
re_ctx.buf = h_buffer;
re_ctx.recursion_limit = DUK_RE_COMPILE_RECURSION_LIMIT;
re_ctx.re_flags = parse_regexp_flags(thr, h_flags);
DUK_DDPRINT("regexp compiler ctx initialized, flags=0x%08x, recursion_limit=%d",
(unsigned int) re_ctx.re_flags, (int) re_ctx.recursion_limit);
/*
* Init lexer
*/
lex_point.offset = 0; /* expensive init, just want to fill window */
lex_point.line = 1;
DUK_LEXER_SETPOINT(&re_ctx.lex, &lex_point);
/*
* Compilation
*/
DUK_DPRINT("starting regexp compilation");
append_u32(&re_ctx, DUK_REOP_SAVE);
append_u32(&re_ctx, 0);
(void) parse_disjunction(&re_ctx, 1); /* 1 = expect eof */
append_u32(&re_ctx, DUK_REOP_SAVE);
append_u32(&re_ctx, 1);
append_u32(&re_ctx, DUK_REOP_MATCH);
DUK_DPRINT("regexp bytecode size (before header) is %d bytes",
(int) DUK_HBUFFER_GET_SIZE(re_ctx.buf));
/*
* Check for invalid backreferences; note that it is NOT an error
* to back-reference a capture group which has not yet been introduced
* in the pattern (as in /\1(foo)/); in fact, the backreference will
* always match! It IS an error to back-reference a capture group
* which will never be introduced in the pattern. Thus, we can check
* for such references only after parsing is complete.
*/
if (re_ctx.highest_backref > re_ctx.captures) {
DUK_ERROR(thr, DUK_ERR_SYNTAX_ERROR, "invalid backreference(s)");
}
/*
* Emit compiled regexp header: flags, ncaptures
* (insertion order inverted on purpose)
*/
insert_u32(&re_ctx, 0, (re_ctx.captures + 1) * 2);
insert_u32(&re_ctx, 0, re_ctx.re_flags);
DUK_DPRINT("regexp bytecode size (after header) is %d bytes",
(int) DUK_HBUFFER_GET_SIZE(re_ctx.buf));
DUK_DDDPRINT("compiled regexp: %!xO", re_ctx.buf);
/* [ ... pattern flags escaped_source buffer ] */
duk_to_string(ctx, -1); /* coerce to string */
/* [ ... pattern flags escaped_source bytecode ] */
/*
* Finalize stack
*/
duk_remove(ctx, -4); /* -> [ ... flags escaped_source bytecode ] */
duk_remove(ctx, -3); /* -> [ ... escaped_source bytecode ] */
DUK_DPRINT("regexp compilation successful, bytecode: %!T, escaped source: %!T",
duk_get_tval(ctx, -1), duk_get_tval(ctx, -2));
}
