From: kkosako@... Date: 2004-05-24T15:25:34+09:00 Subject: [ruby-dev:23596] Re: RegexpError: undefined bytecode (bug) > -----Original Message----- > From: Masahiro Sakai (酒井政裕) [mailto:sakai@tom.sfc.keio.ac.jp] > たまたま気が付いたのですが、1.9で正規表現のマッチングを行うと、 > 以下のようなエラーが発生することがあるようです。 > > % ruby -ruri -ve 'URI::ABS_URI =~ "http://example.org/Andr\xC3\xA9"' > -e:1:in `=~': undefined bytecode (bug): /^ (RegexpError) 設計ミスです。 Ruby 1.9.0(Oniguruma 2.2.8)へのパッチです。 -- 小迫@出向中 --- regint.h.orig 2004-04-28 01:20:02.000000000 +0900 +++ regint.h 2004-05-24 14:51:42.000000000 +0900 @@ -500,6 +500,8 @@ enum OpCode { OP_REPEAT_NG, /* {n,m}? (non greedy) */ OP_REPEAT_INC, OP_REPEAT_INC_NG, /* non greedy */ + OP_REPEAT_INC_SG, /* search and get in stack */ + OP_REPEAT_INC_NG_SG, /* search and get in stack (non greedy) */ OP_NULL_CHECK_START, /* null loop checker start */ OP_NULL_CHECK_END, /* null loop checker end */ OP_NULL_CHECK_END_MEMST, /* null loop checker end (with capture status) */ --- regparse.h.orig 2004-03-06 00:31:54.000000000 +0900 +++ regparse.h 2004-05-24 14:32:30.000000000 +0900 @@ -96,6 +96,7 @@ typedef struct { } CClassNode; typedef struct { + int state; struct _Node* target; int lower; int upper; @@ -121,6 +122,7 @@ typedef struct { #define NST_ADDR_FIXED (1<<9) #define NST_NAMED_GROUP (1<<10) #define NST_NAME_REF (1<<11) +#define NST_IN_REPEAT (1<<12) /* STK_REPEAT is nested in match stack. */ #define SET_EFFECT_STATUS(node,f) (node)->u.effect.state |= (f) #define CLEAR_EFFECT_STATUS(node,f) (node)->u.effect.state &= ~(f) @@ -140,6 +142,7 @@ typedef struct { #define IS_CALL_RECURSION(cn) (((cn)->state & NST_RECURSION) != 0) #define IS_CALL_NAME_REF(cn) (((cn)->state & NST_NAME_REF) != 0) #define IS_BACKREF_NAME_REF(bn) (((bn)->state & NST_NAME_REF) != 0) +#define IS_QUALIFIER_IN_REPEAT(qn) (((qn)->state & NST_IN_REPEAT) != 0) typedef struct { int state; --- regparse.c.orig 2004-04-28 01:37:56.000000000 +0900 +++ regparse.c 2004-05-24 14:33:11.000000000 +0900 @@ -1058,6 +1058,7 @@ node_new_qualifier(int lower, int upper, Node* node = node_new(); CHECK_NULL_RETURN(node); node->type = N_QUALIFIER; + NQUALIFIER(node).state = 0; NQUALIFIER(node).target = NULL; NQUALIFIER(node).lower = lower; NQUALIFIER(node).upper = upper; --- regcomp.c.orig 2004-04-28 01:20:01.000000000 +0900 +++ regcomp.c 2004-05-24 14:53:31.000000000 +0900 @@ -684,7 +684,16 @@ compile_range_repeat_node(QualifierNode* r = compile_tree_empty_check(qn->target, reg, empty_info); if (r) return r; - r = add_opcode(reg, qn->greedy ? OP_REPEAT_INC : OP_REPEAT_INC_NG); + if ( +#ifdef USE_SUBEXP_CALL + reg->num_call > 0 || +#endif + IS_QUALIFIER_IN_REPEAT(qn)) { + r = add_opcode(reg, qn->greedy ? OP_REPEAT_INC_SG : OP_REPEAT_INC_NG_SG); + } + else { + r = add_opcode(reg, qn->greedy ? OP_REPEAT_INC : OP_REPEAT_INC_NG); + } if (r) return r; r = add_mem_num(reg, num_repeat); /* OP_REPEAT ID */ return r; @@ -3057,6 +3066,10 @@ setup_tree(Node* node, regex_t* reg, int QualifierNode* qn = &(NQUALIFIER(node)); Node* target = qn->target; + if ((state & IN_REPEAT) != 0) { + qn->state |= NST_IN_REPEAT; + } + if (IS_REPEAT_INFINITE(qn->upper) || qn->upper >= 1) { r = get_min_match_length(target, &d, env); if (r) break; --- regexec.c.orig 2004-04-28 01:20:02.000000000 +0900 +++ regexec.c 2004-05-24 14:59:40.000000000 +0900 @@ -1053,6 +1053,7 @@ match_at(regex_t* reg, UChar* str, UChar char *alloca_base; StackType *stk_alloc, *stk_base, *stk, *stk_end; StackType *stkp; /* used as any purpose. */ + StackIndex si; StackIndex *repeat_stk; StackIndex *mem_start_stk, *mem_end_stk; n = reg->num_repeat + reg->num_mem * 2; @@ -2170,79 +2171,67 @@ match_at(regex_t* reg, UChar* str, UChar break; case OP_REPEAT_INC: STAT_OP_IN(OP_REPEAT_INC); - { - StackIndex si; - - GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */ -#ifdef USE_SUBEXP_CALL - if (reg->num_call > 0) { - STACK_GET_REPEAT(mem, stkp); - si = GET_STACK_INDEX(stkp); - } - else { - si = repeat_stk[mem]; - stkp = STACK_AT(si); - } -#else - si = repeat_stk[mem]; - stkp = STACK_AT(si); -#endif - stkp->u.repeat.count++; - if (stkp->u.repeat.count == reg->repeat_range[mem].upper) { - /* end of repeat. Nothing to do. */ - } - else if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) { - STACK_PUSH_ALT(p, s, sprev); - p = stkp->u.repeat.pcode; - } - else { - p = stkp->u.repeat.pcode; - } - STACK_PUSH_REPEAT_INC(si); + GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */ + si = repeat_stk[mem]; + stkp = STACK_AT(si); + + repeat_inc: + stkp->u.repeat.count++; + if (stkp->u.repeat.count == reg->repeat_range[mem].upper) { + /* end of repeat. Nothing to do. */ + } + else if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) { + STACK_PUSH_ALT(p, s, sprev); + p = stkp->u.repeat.pcode; } + else { + p = stkp->u.repeat.pcode; + } + STACK_PUSH_REPEAT_INC(si); STAT_OP_OUT; continue; break; - case OP_REPEAT_INC_NG: STAT_OP_IN(OP_REPEAT_INC_NG); - { - StackIndex si; - - GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */ -#ifdef USE_SUBEXP_CALL - if (reg->num_call > 0) { - STACK_GET_REPEAT(mem, stkp); - si = GET_STACK_INDEX(stkp); - } - else { - si = repeat_stk[mem]; - stkp = STACK_AT(si); - } -#else - si = repeat_stk[mem]; - stkp = STACK_AT(si); -#endif - stkp->u.repeat.count++; - if (stkp->u.repeat.count < reg->repeat_range[mem].upper) { - if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) { - UChar* pcode = stkp->u.repeat.pcode; + case OP_REPEAT_INC_SG: STAT_OP_IN(OP_REPEAT_INC_SG); + GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */ + STACK_GET_REPEAT(mem, stkp); + si = GET_STACK_INDEX(stkp); + goto repeat_inc; + break; - STACK_PUSH_REPEAT_INC(si); - STACK_PUSH_ALT(pcode, s, sprev); - } - else { - p = stkp->u.repeat.pcode; - STACK_PUSH_REPEAT_INC(si); - } - } - else if (stkp->u.repeat.count == reg->repeat_range[mem].upper) { - STACK_PUSH_REPEAT_INC(si); - } + case OP_REPEAT_INC_NG: STAT_OP_IN(OP_REPEAT_INC_NG); + GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */ + si = repeat_stk[mem]; + stkp = STACK_AT(si); + + repeat_inc_ng: + stkp->u.repeat.count++; + if (stkp->u.repeat.count < reg->repeat_range[mem].upper) { + if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) { + UChar* pcode = stkp->u.repeat.pcode; + + STACK_PUSH_REPEAT_INC(si); + STACK_PUSH_ALT(pcode, s, sprev); + } + else { + p = stkp->u.repeat.pcode; + STACK_PUSH_REPEAT_INC(si); + } + } + else if (stkp->u.repeat.count == reg->repeat_range[mem].upper) { + STACK_PUSH_REPEAT_INC(si); } STAT_OP_OUT; continue; break; + case OP_REPEAT_INC_NG_SG: STAT_OP_IN(OP_REPEAT_INC_NG_SG); + GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */ + STACK_GET_REPEAT(mem, stkp); + si = GET_STACK_INDEX(stkp); + goto repeat_inc_ng; + break; + case OP_PUSH_POS: STAT_OP_IN(OP_PUSH_POS); STACK_PUSH_POS(s, sprev); STAT_OP_OUT;