Repair recently-introduced error in makeIndexable for LIKE:

a non-leading % would be put into the >=/<= patterns.  Also, repair
longstanding confusion about whether %% means a literal %%.  The SQL92
doesn't say any such thing, and textlike() knows that, but gram.y didn't.

diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index c038f43..bb36c6f 100644 (file)
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -10,7 +10,7 @@
  *
  *
  * IDENTIFICATION
- *       $Header: /cvsroot/pgsql/src/backend/parser/gram.y,v 2.83 1999/05/22 05:06:43 momjian Exp $
+ *       $Header: /cvsroot/pgsql/src/backend/parser/gram.y,v 2.84 1999/06/07 14:28:25 tgl Exp $
  *
  * HISTORY
  *       AUTHOR                        DATE                    MAJOR EVENT
@@ -5357,6 +5357,7 @@ static Node *makeIndexable(char *opname, Node *lexpr, Node *rexpr)
                        int pos, match_pos=0;
                        bool found_special = false;
 
+                       /* Cannot optimize if unquoted | { } is present in pattern */
                        for (pos = 1; n->val.val.str[pos]; pos++)
                        {
                                if (n->val.val.str[pos] == '|' ||
@@ -5367,12 +5368,16 @@ static Node *makeIndexable(char *opname, Node *lexpr, Node *rexpr)
                                        break;
                                }
                        if (n->val.val.str[pos] == '\\')
+                               {
                                        pos++;
+                                       if (n->val.val.str[pos] == '\0')
+                                               break;
+                               }
                        }
 
-                       /* skip leading ^ */
                        if (!found_special)
                        {
+                               /* note start at pos 1 to skip leading ^ */
                                for (pos = 1; n->val.val.str[pos]; pos++)
                                {
                                        if (n->val.val.str[pos] == '.' ||
@@ -5383,9 +5388,11 @@ static Node *makeIndexable(char *opname, Node *lexpr, Node *rexpr)
                                                (strcmp(opname,"~*") == 0 && isalpha(n->val.val.str[pos])))
                                        break;
                                if (n->val.val.str[pos] == '\\')
+                                       {
                                                pos++;
-                                       if (n->val.val.str[pos] == '\0')
-                                               break;
+                                               if (n->val.val.str[pos] == '\0')
+                                                       break;
+                                       }
                                        match_least[match_pos] = n->val.val.str[pos];
                                        match_most[match_pos++] = n->val.val.str[pos];
                                }
@@ -5430,16 +5437,22 @@ static Node *makeIndexable(char *opname, Node *lexpr, Node *rexpr)
        
                        for (pos = 0; n->val.val.str[pos]; pos++)
                        {
-                               if (n->val.val.str[pos] == '%' &&
-                                        n->val.val.str[pos+1] != '%')
-                                       break;
-                               if(n->val.val.str[pos] == '_')
+                               /* % and _ are wildcard characters in LIKE */
+                               if (n->val.val.str[pos] == '%' ||
+                                       n->val.val.str[pos] == '_')
                                        break;
-                       if (n->val.val.str[pos] == '\\' ||
-                               n->val.val.str[pos+1] == '%')
+                               /* Backslash quotes the next character */
+                               if (n->val.val.str[pos] == '\\')
+                               {
                                        pos++;
-                               if (n->val.val.str[pos] == '\0')
-                                       break;
+                                       if (n->val.val.str[pos] == '\0')
+                                               break;
+                               }
+                               /*
+                                * NOTE: this code used to think that %% meant a literal %,
+                                * but textlike() itself does not think that, and the SQL92
+                                * spec doesn't say any such thing either.
+                                */
                                match_least[match_pos] = n->val.val.str[pos];
                                match_most[match_pos++] = n->val.val.str[pos];
                        }

diff --git a/src/backend/utils/adt/like.c b/src/backend/utils/adt/like.c
index 14fc531..2d35bfa 100644 (file)
--- a/src/backend/utils/adt/like.c
+++ b/src/backend/utils/adt/like.c
@@ -111,7 +111,7 @@ textnlike(struct varlena * s, struct varlena * p)
 }
 
 
-/*     $Revision: 1.24 $
+/*     $Revision: 1.25 $
 **     "like.c" A first attempt at a LIKE operator for Postgres95.
 **
 **     Originally written by Rich $alz, mirror!rs, Wed Nov 26 19:03:17 EST 1986.
@@ -191,7 +191,9 @@ DoMatch(pg_wchar * text, pg_wchar * p)
        else
        {
                /* End of input string.  Do we have matching string remaining? */
-               if (p[0] == '\0' || (p[0] == '%' && p[1] == '\0'))
+               while (*p == '%')               /* allow multiple %'s at end of pattern */
+                       p++;
+               if (*p == '\0')
                        return LIKE_TRUE;
                else
                        return LIKE_ABORT;