Enhance the REGEXP extension so that the end-of-input indicate ("$") is

allowed to occur on one branch of an OR ("|").
[forum:/forumpost/0107d5d40dd273e2|Forum post 0107d5d40dd273e2], second
issue.

FossilOrigin-Name: 3c04d21e6c632feb3bea8d1fa76bedcbfe254b0dc59865633d158a3f1bddefba

diff --git a/ext/misc/regexp.c b/ext/misc/regexp.c
index 52973cc73fbfdd3815b8fabe29808f331959fee4..d5b3872058c621a7876ef77c084c9841a2709778 100644 (file)
--- a/ext/misc/regexp.c
+++ b/ext/misc/regexp.c
@@ -328,7 +328,9 @@ static int re_match(ReCompiled *pRe, const unsigned char *zIn, int nIn){
     }
   }
   for(i=0; i<pNext->nState; i++){
-    if( pRe->aOp[pNext->aState[i]]==RE_OP_ACCEPT ){ rc = 1; break; }
+    int x = pNext->aState[i];
+    while( pRe->aOp[x]==RE_OP_GOTO ) x += pRe->aArg[x];
+    if( pRe->aOp[x]==RE_OP_ACCEPT ){ rc = 1; break; }
   }
 re_match_end:
   sqlite3_free(pToFree);
@@ -483,7 +485,6 @@ static const char *re_subcompile_string(ReCompiled *p){
     iStart = p->nState;
     switch( c ){
       case '|':
-      case '$':
       case ')': {
         p->sIn.i--;
         return 0;
@@ -520,6 +521,10 @@ static const char *re_subcompile_string(ReCompiled *p){
         re_insert(p, iPrev, RE_OP_FORK, p->nState - iPrev+1);
         break;
       }
+      case '$': {
+        re_append(p, RE_OP_MATCH, RE_EOF);
+        break;
+      }
       case '{': {
         int m = 0, n = 0;
         int sz, j;
@@ -656,11 +661,7 @@ static const char *re_compile(ReCompiled **ppRe, const char *zIn, int noCase){
     re_free(pRe);
     return zErr;
   }
-  if( rePeek(pRe)=='$' && pRe->sIn.i+1>=pRe->sIn.mx ){
-    re_append(pRe, RE_OP_MATCH, RE_EOF);
-    re_append(pRe, RE_OP_ACCEPT, 0);
-    *ppRe = pRe;
-  }else if( pRe->sIn.i>=pRe->sIn.mx ){
+  if( pRe->sIn.i>=pRe->sIn.mx ){
     re_append(pRe, RE_OP_ACCEPT, 0);
     *ppRe = pRe;
   }else{

diff --git a/manifest b/manifest
index 018a75e20b8bf8a5e5b1aaf148c22ac31afba7d1..64f106cefe177571a3007d91ce512e5147b8b8c1 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Fix\sthe\sinitial-prefix\soptimization\sfor\sthe\sREGEXP\sextension\ssuch\sthat\sit\nworks\seven\sif\sthe\sprefix\scontains\scharacters\sthat\srequire\sa\s3-byte\sUTF8\nencoding.\s\sThis\sshould\sfix\sthe\sproblem\sreported\sby\n[forum:/forumpost/96692f8ba5|forum\spost\s96692f8ba5].
-D 2022-07-03T14:25:47.812
+C Enhance\sthe\sREGEXP\sextension\sso\sthat\sthe\send-of-input\sindicate\s("$")\sis\nallowed\sto\soccur\son\sone\sbranch\sof\san\sOR\s("|").\n[forum:/forumpost/0107d5d40dd273e2|Forum\spost\s0107d5d40dd273e2],\ssecond\nissue.
+D 2022-07-03T18:12:43.089
 F .fossil-settings/empty-dirs dbb81e8fc0401ac46a1491ab34a7f2c7c0452f2f06b54ebb845d024ca8283ef1
 F .fossil-settings/ignore-glob 35175cdfcf539b2318cb04a9901442804be81cd677d8b889fcc9149c21f239ea
 F LICENSE.md df5091916dbb40e6e9686186587125e1b2ff51f022cc334e886c19a0e9982724
@@ -334,7 +334,7 @@ F ext/misc/normalize.c bd84355c118e297522aba74de34a4fd286fc775524e0499b14473918d
 F ext/misc/percentile.c b9086e223d583bdaf8cb73c98a6539d501a2fc4282654adbfea576453d82e691
 F ext/misc/prefixes.c 0f4f8cff5aebc00a7e3ac4021fd59cfe1a8e17c800ceaf592859ecb9cbc38196
 F ext/misc/qpvtab.c 09738419e25f603a35c0ac8bd0a04daab794f48d08a9bc07a6085b9057b99009
-F ext/misc/regexp.c 03e483711534c437b2e29648d2a4b7730f5cb781a434ac8150907376bc4489f6
+F ext/misc/regexp.c c1fb4f0639ad70f180fa7ebca3b9830c1c648cdcd072e28747f727699f3a5071
 F ext/misc/remember.c add730f0f7e7436cd15ea3fd6a90fd83c3f706ab44169f7f048438b7d6baa69c
 F ext/misc/rot13.c 51ac5f51e9d5fd811db58a9c23c628ad5f333c173f1fc53c8491a3603d38556c
 F ext/misc/scrub.c 2a44b0d44c69584c0580ad2553f6290a307a49df4668941d2812135bfb96a946
@@ -1342,7 +1342,7 @@ F test/randexpr1.test eda062a97e60f9c38ae8d806b03b0ddf23d796df
 F test/rbu.test 168573d353cd0fd10196b87b0caa322c144ef736
 F test/rdonly.test 64e2696c322e3538df0b1ed624e21f9a23ed9ff8
 F test/recover.test ccb8c2623902a92ebb76770edd075cb4f75a4760bb7afde38026572c6e79070d
-F test/regexp1.test 4a44e014664a109bbb1c37d29d9b61ca5aa5a7f49cc564c95208a80f818e3377
+F test/regexp1.test 611adedda9ab00c86a14e2a4becf4086ed712e83d86f576a884d21795213f16b
 F test/regexp2.test 55ed41da802b0e284ac7e2fe944be3948f93ff25abbca0361a609acfed1368b5
 F test/reindex.test cd9d6021729910ece82267b4f5e1b5ac2911a7566c43b43c176a6a4732e2118d
 F test/releasetest_data.tcl 11ba48a21ed1c808147b0e77c6e93d204577f4327ffe6d7c3b34cd3c01eac3a2
@@ -1978,8 +1978,8 @@ F vsixtest/vsixtest.tcl 6a9a6ab600c25a91a7acc6293828957a386a8a93
 F vsixtest/vsixtest.vcxproj.data 2ed517e100c66dc455b492e1a33350c1b20fbcdc
 F vsixtest/vsixtest.vcxproj.filters 37e51ffedcdb064aad6ff33b6148725226cd608e
 F vsixtest/vsixtest_TemporaryKey.pfx e5b1b036facdb453873e7084e1cae9102ccc67a0
-P 6a8e4fb72a9e3dea9e5752c3d54fddba8878b355bd43f3c879f042f247a6610f
-R 3b0d1d13c75a6aec06b5d58ce72f012e
+P c94595a6e15490b432f099fefbe2429fa19287f7bdc86332cba0fd1e08f65bd6
+R 5a847080949684e34540bd8fbb3bcdb5
 U drh
-Z 9a117fdde370084441c6305b15600e26
+Z 843de43aa11ad4fe844e55becdb579a4
 # Remove this line to create a well-formed Fossil manifest.

diff --git a/manifest.uuid b/manifest.uuid
index 141113cd8097a5bac27d471f4a90dab8b80ce0c8..7ce3a33e2a15d7900ec11b711431da6239a61405 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-c94595a6e15490b432f099fefbe2429fa19287f7bdc86332cba0fd1e08f65bd6
\ No newline at end of file
+3c04d21e6c632feb3bea8d1fa76bedcbfe254b0dc59865633d158a3f1bddefba
\ No newline at end of file

diff --git a/test/regexp1.test b/test/regexp1.test
index 569dd66c2a2879d204b79cbee20b5f3dab52801b..fa941af53b38939f55f9b288fbe3b3e0c7a4f579 100644 (file)
--- a/test/regexp1.test
+++ b/test/regexp1.test
@@ -261,4 +261,23 @@ do_execsql_test regexp1-3.4 {
   SELECT * FROM t1 wHERE a REGEXP '日本語';
 } {1 日本語}
 
+# 2022-07-03
+# https://sqlite.org/forum/forumpost/96692f8ba5  Issue #2
+# The '$' token in REGEXP contained within other elements.
+#
+do_execsql_test regexp1-4.1 {SELECT 'xab' REGEXP 'a(b$|cd)';} {1}
+do_execsql_test regexp1-4.1b {SELECT 'xab' REGEXP '(b$|cd)';} {1}
+do_execsql_test regexp1-4.2 {SELECT 'xaby' REGEXP 'a(b$|cd)';} {0}
+do_execsql_test regexp1-4.3 {SELECT 'xacd' REGEXP 'a(b$|cd)';} {1}
+do_execsql_test regexp1-4.4 {SELECT 'xacdy' REGEXP 'a(b$|cd)';} {1}
+do_execsql_test regexp1-4.5 {SELECT 'xab' REGEXP 'a(cd|b$)';} {1}
+do_execsql_test regexp1-4.6 {SELECT 'xaby' REGEXP 'a(cd|b$)';} {0}
+do_execsql_test regexp1-4.7 {SELECT 'xacd' REGEXP 'a(cd|b$)';} {1}
+do_execsql_test regexp1-4.8 {SELECT 'xacdy' REGEXP 'a(cd|b$)';} {1}
+do_execsql_test regexp1-4.9 {SELECT 'xab' REGEXP 'a(cd|b$|e)';} {1}
+do_execsql_test regexp1-4.10 {SELECT 'xaby' REGEXP 'a(cd|b$|e)';} {0}
+do_execsql_test regexp1-4.11 {SELECT 'xacd' REGEXP 'a(cd|b$|e)';} {1}
+do_execsql_test regexp1-4.12 {SELECT 'xacdy' REGEXP 'a(cd|b$|e)';} {1}
+
+
 finish_test