From 2de946be6a4de5986ba7b1528aeb7eec0dec194f Mon Sep 17 00:00:00 2001 From: Tom Lane Date: Wed, 7 Nov 2007 22:37:24 +0000 Subject: [PATCH] Improve the performance of LIKE/regex estimation in non-C locales, by making make_greater_string() try harder to generate a string that's actually greater than its input string. Before we just assumed that making a string that was memcmp-greater was enough, but it is easy to generate examples where this is not so when the locale is not C. Instead, loop until the relevant comparison function agrees that the generated string is greater than the input. Unfortunately this is probably not enough to guarantee that the generated string is greater than all extensions of the input, so we cannot relax the restriction to C locale for the LIKE/regex index optimization. But it should at least improve the odds of getting a useful selectivity estimate in prefix_selectivity(). Per example from Guillaume Smet. Backpatch to 8.1, mainly because that's what the complainant is using... --- src/backend/optimizer/path/indxpath.c | 14 ++++---- src/backend/utils/adt/selfuncs.c | 52 +++++++++++++++++---------- src/include/utils/selfuncs.h | 4 +-- 3 files changed, 43 insertions(+), 27 deletions(-) diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/path/indxpath.c index b9439e60a8..47dd3ec55b 100644 --- a/src/backend/optimizer/path/indxpath.c +++ b/src/backend/optimizer/path/indxpath.c @@ -9,7 +9,7 @@ * * * IDENTIFICATION - * $PostgreSQL: pgsql/src/backend/optimizer/path/indxpath.c,v 1.222 2007/05/22 01:40:33 tgl Exp $ + * $PostgreSQL: pgsql/src/backend/optimizer/path/indxpath.c,v 1.223 2007/11/07 22:37:24 tgl Exp $ * *------------------------------------------------------------------------- */ @@ -2668,6 +2668,7 @@ prefix_quals(Node *leftop, Oid opfamily, Oid datatype; Oid oproid; Expr *expr; + FmgrInfo ltproc; Const *greaterstr; Assert(pstatus != Pattern_Prefix_None); @@ -2759,13 +2760,14 @@ prefix_quals(Node *leftop, Oid opfamily, * "x < greaterstr". *------- */ - greaterstr = make_greater_string(prefix_const); + oproid = get_opfamily_member(opfamily, datatype, datatype, + BTLessStrategyNumber); + if (oproid == InvalidOid) + elog(ERROR, "no < operator for opfamily %u", opfamily); + fmgr_info(get_opcode(oproid), <proc); + greaterstr = make_greater_string(prefix_const, <proc); if (greaterstr) { - oproid = get_opfamily_member(opfamily, datatype, datatype, - BTLessStrategyNumber); - if (oproid == InvalidOid) - elog(ERROR, "no < operator for opfamily %u", opfamily); expr = make_opclause(oproid, BOOLOID, false, (Expr *) leftop, (Expr *) greaterstr); result = lappend(result, diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 5bf04d052a..095b389eb1 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -15,7 +15,7 @@ * * * IDENTIFICATION - * $PostgreSQL: pgsql/src/backend/utils/adt/selfuncs.c,v 1.237 2007/11/07 21:00:37 tgl Exp $ + * $PostgreSQL: pgsql/src/backend/utils/adt/selfuncs.c,v 1.238 2007/11/07 22:37:24 tgl Exp $ * *------------------------------------------------------------------------- */ @@ -4302,17 +4302,17 @@ prefix_selectivity(VariableStatData *vardata, * "x < greaterstr". *------- */ - greaterstrcon = make_greater_string(prefixcon); + cmpopr = get_opfamily_member(opfamily, vartype, vartype, + BTLessStrategyNumber); + if (cmpopr == InvalidOid) + elog(ERROR, "no < operator for opfamily %u", opfamily); + fmgr_info(get_opcode(cmpopr), &opproc); + + greaterstrcon = make_greater_string(prefixcon, &opproc); if (greaterstrcon) { Selectivity topsel; - cmpopr = get_opfamily_member(opfamily, vartype, vartype, - BTLessStrategyNumber); - if (cmpopr == InvalidOid) - elog(ERROR, "no < operator for opfamily %u", opfamily); - fmgr_info(get_opcode(cmpopr), &opproc); - topsel = ineq_histogram_selectivity(vardata, &opproc, false, greaterstrcon->constvalue, greaterstrcon->consttype); @@ -4589,8 +4589,17 @@ pattern_selectivity(Const *patt, Pattern_Type ptype) * in the form of a Const pointer; else return NULL. * * The key requirement here is that given a prefix string, say "foo", - * we must be able to generate another string "fop" that is greater - * than all strings "foobar" starting with "foo". + * we must be able to generate another string "fop" that is greater than + * all strings "foobar" starting with "foo". We can test that we have + * generated a string greater than the prefix string, but in non-C locales + * that is not a bulletproof guarantee that an extension of the string might + * not sort after it; an example is that "foo " is less than "foo!", but it + * is not clear that a "dictionary" sort ordering will consider "foo!" less + * than "foo bar". Therefore, this function should be used only for + * estimation purposes when working in a non-C locale. + * + * The caller must provide the appropriate "less than" comparison function + * for testing the strings. * * If we max out the righthand byte, truncate off the last character * and start incrementing the next. For example, if "z" were the last @@ -4599,20 +4608,15 @@ pattern_selectivity(Const *patt, Pattern_Type ptype) * * This could be rather slow in the worst case, but in most cases we * won't have to try more than one or two strings before succeeding. - * - * NOTE: at present this assumes we are in the C locale, so that simple - * bytewise comparison applies. However, we might be in a multibyte - * encoding such as UTF8, so we do have to watch out for generating - * invalid encoding sequences. */ Const * -make_greater_string(const Const *str_const) +make_greater_string(const Const *str_const, FmgrInfo *ltproc) { Oid datatype = str_const->consttype; char *workstr; int len; - /* Get the string and a modifiable copy */ + /* Get a modifiable copy of the string in C-string format */ if (datatype == NAMEOID) { workstr = DatumGetCString(DirectFunctionCall1(nameout, @@ -4660,8 +4664,18 @@ make_greater_string(const Const *str_const) else workstr_const = string_to_bytea_const(workstr, len); - pfree(workstr); - return workstr_const; + if (DatumGetBool(FunctionCall2(ltproc, + str_const->constvalue, + workstr_const->constvalue))) + { + /* Successfully made a string larger than the input */ + pfree(workstr); + return workstr_const; + } + + /* No good, release unusable value and try again */ + pfree(DatumGetPointer(workstr_const->constvalue)); + pfree(workstr_const); } /* restore last byte so we don't confuse pg_mbcliplen */ diff --git a/src/include/utils/selfuncs.h b/src/include/utils/selfuncs.h index fc409d9bdd..f92bb16d07 100644 --- a/src/include/utils/selfuncs.h +++ b/src/include/utils/selfuncs.h @@ -8,7 +8,7 @@ * Portions Copyright (c) 1996-2007, PostgreSQL Global Development Group * Portions Copyright (c) 1994, Regents of the University of California * - * $PostgreSQL: pgsql/src/include/utils/selfuncs.h,v 1.40 2007/08/31 23:35:22 tgl Exp $ + * $PostgreSQL: pgsql/src/include/utils/selfuncs.h,v 1.41 2007/11/07 22:37:24 tgl Exp $ * *------------------------------------------------------------------------- */ @@ -118,7 +118,7 @@ extern Pattern_Prefix_Status pattern_fixed_prefix(Const *patt, Pattern_Type ptype, Const **prefix, Const **rest); -extern Const *make_greater_string(const Const *str_const); +extern Const *make_greater_string(const Const *str_const, FmgrInfo *ltproc); extern Datum eqsel(PG_FUNCTION_ARGS); extern Datum neqsel(PG_FUNCTION_ARGS); -- 2.40.0