diff options
| author | drh <> | 2022-07-18 13:55:21 +0000 |
|---|---|---|
| committer | drh <> | 2022-07-18 13:55:21 +0000 |
| commit | 449b473ee6dacd93d31672396bdd4742dd916d35 (patch) | |
| tree | eaabdbbefcc98ddf5dbcf48397b44b6f72776f64 /ext/misc/regexp.c | |
| parent | 18934137ddfd765025e73b2eb1a1bb83ee5fc03b (diff) | |
| download | sqlite-449b473ee6dacd93d31672396bdd4742dd916d35.tar.gz sqlite-449b473ee6dacd93d31672396bdd4742dd916d35.zip | |
Enhance the REGEXP extension so that it will accept the start-of-input
mark ("^") in the middle of parentheses.
[forum:/forumpost/0d6a9160f81ef1a8|Forum post 0d6a9160f81ef1a8].
FossilOrigin-Name: ed8a8ebd62a319b5dabbdf67ee27141153b9899d7c8f08eeb4bdf35271015c71
Diffstat (limited to 'ext/misc/regexp.c')
| -rw-r--r-- | ext/misc/regexp.c | 14 |
1 files changed, 13 insertions, 1 deletions
diff --git a/ext/misc/regexp.c b/ext/misc/regexp.c index 8a3e13115..7413ab80e 100644 --- a/ext/misc/regexp.c +++ b/ext/misc/regexp.c @@ -72,6 +72,7 @@ SQLITE_EXTENSION_INIT1 /* The end-of-input character */ #define RE_EOF 0 /* End of input */ +#define RE_START 0xfffffff /* Start of input - larger than an UTF-8 */ /* The NFA is implemented as sequence of opcodes taken from the following ** set. Each opcode has a single integer argument. @@ -93,6 +94,7 @@ SQLITE_EXTENSION_INIT1 #define RE_OP_SPACE 15 /* space: [ \t\n\r\v\f] */ #define RE_OP_NOTSPACE 16 /* Not a digit */ #define RE_OP_BOUNDARY 17 /* Boundary between word and non-word */ +#define RE_OP_ATSTART 18 /* Currently at the start of the string */ #if defined(SQLITE_DEBUG) /* Opcode names used for symbolic debugging */ @@ -115,6 +117,7 @@ static const char *ReOpName[] = { "SPACE", "NOTSPACE", "BOUNDARY", + "ATSTART", }; #endif /* SQLITE_DEBUG */ @@ -225,7 +228,7 @@ static int re_match(ReCompiled *pRe, const unsigned char *zIn, int nIn){ ReStateNumber *pToFree; unsigned int i = 0; unsigned int iSwap = 0; - int c = RE_EOF+1; + int c = RE_START; int cPrev = 0; int rc = 0; ReInput in; @@ -244,6 +247,7 @@ static int re_match(ReCompiled *pRe, const unsigned char *zIn, int nIn){ in.i++; } if( in.i+pRe->nInit>in.mx ) return 0; + c = RE_START-1; } if( pRe->nState<=(sizeof(aSpace)/(sizeof(aSpace[0])*2)) ){ @@ -272,6 +276,10 @@ static int re_match(ReCompiled *pRe, const unsigned char *zIn, int nIn){ if( pRe->aArg[x]==c ) re_add_state(pNext, x+1); break; } + case RE_OP_ATSTART: { + if( cPrev==RE_START ) re_add_state(pThis, x+1); + break; + } case RE_OP_ANY: { if( c!=0 ) re_add_state(pNext, x+1); break; @@ -550,6 +558,10 @@ static const char *re_subcompile_string(ReCompiled *p){ re_append(p, RE_OP_MATCH, RE_EOF); break; } + case '^': { + re_append(p, RE_OP_ATSTART, 0); + break; + } case '{': { int m = 0, n = 0; int sz, j; |