4  *  See doc/regexp.rst for a discussion of the compilation approach and
16  *  input string size (assuming an upper bound can be computed for number
40 	/* Number of characters that the atom matches (e.g. 3 for 'abc'),
69  * larger than expected (or at least assert for it).  Many things in the
82 /* XXX: return type should probably be duk_size_t, or explicit checks are needed for
113 /* special helper for emitting u16 lists (used for character ranges for built-in char classes) */
145  *  variable length UTF-8 encoding.  See doc/regexp.rst for discussion.
167  *  duk_re_range_callback for generating character class ranges.
170  *  We don't, for instance, eliminate duplicates or overlapping
178  *  time execution time (see doc/regexp.rst for discussion).
211 		for (i = r1 + 1; i <= r2; i++) {  in duk__generate_ranges()
251  *  Appends the bytecode for the disjunction matcher to the end of the temp
268  *      disjunction are separate matters.  For instance, the disjunction
274  *      and it would be possible to expand it.  For instance, captures
279  *      main for loop collectively for atoms.  Assertions, quantifiers,
307 	for (;;) {  in duk__parse_disjunction()
311 		 * new_atom_char_length etc are for the atom parsed on this round;  in duk__parse_disjunction()
337 			 *  we have a pending split1 and a pending jump (for a previous match).  These  in duk__parse_disjunction()
338 			 *  need to be back-patched carefully.  See docs for a detailed example.  in duk__parse_disjunction()
360 			unpatched_disjunction_split = entry_offset + 1;   /* +1 for opcode */  in duk__parse_disjunction()
362 			/* add a new pending match jump for latest finished alternative */  in duk__parse_disjunction()
381 				 *  If atom_char_length is zero, we'll have unbounded execution time for e.g.  in duk__parse_disjunction()
383 				 *  side effects (for instance, if we allowed captures in simple atoms, the  in duk__parse_disjunction()
430 				 *  NOTE: there is no current solution for empty atoms in complex  in duk__parse_disjunction()
476 				/* insert code for matching the remainder - infinite or finite */  in duk__parse_disjunction()
478 					/* reuse last emitted atom for remaining 'infinite' quantifier */  in duk__parse_disjunction()
493 					duk__append_jump_offset(re_ctx, -atom_code_length - 1);  /* -1 for opcode */  in duk__parse_disjunction()
520 						                        offset + 1,   /* +1 for opcode */  in duk__parse_disjunction()
562 			                               offset + 1,   /* +1 for opcode */  in duk__parse_disjunction()
579 			 * but this is not trivial (consider e.g. '/xyz+/); see docs for  in duk__parse_disjunction()
661 			 *  us for every range parsed.  This is different from how rest of  in duk__parse_disjunction()
666 			 *  case insensitive matching is used (see docs for discussion).  in duk__parse_disjunction()
738 	/* finish up pending jump and split for last alternative */  in duk__parse_disjunction()
821  *  Note that normalization does not seem to be necessary for
826  *  for expressions like 'new Regexp("...", "")' however.
861 	for (i = 0; i < n; i++) {  in duk__create_escaped_source()
889  *  An escaped version of the regexp source, suitable for use as a RegExp instance
932 	DUK_LEXER_INITCTX(&re_ctx.lex);  /* duplicate zeroing, expect for (possible) NULL inits */  in duk_regexp_compile()
968 	 *  Check for invalid backreferences; note that it is NOT an error  in duk_regexp_compile()
973 	 *  for such references only after parsing is complete.  in duk_regexp_compile()