of an index can now be a computed expression instead of a simple variable.
Restrictions on expressions are the same as for predicates (only immutable
functions, no sub-selects). This fixes problems recently introduced with
inlining SQL functions, because the inlining transformation is applied to
both expression trees so the planner can still match them up. Along the
way, improve efficiency of handling index predicates (both predicates and
index expressions are now cached by the relcache) and fix 7.3 oversight
that didn't record dependencies of predicate expressions.
/* we're only interested if it is the primary key */
if (index->indisprimary == TRUE)
{
- i = 0;
- while (index->indkey[i++] != 0)
- (*numatts)++;
-
+ *numatts = index->indnatts;
if (*numatts > 0)
{
result = (char **) palloc(*numatts * sizeof(char *));
<!--
Documentation of the system catalogs, directed toward PostgreSQL developers
- $Header: /cvsroot/pgsql/doc/src/sgml/catalogs.sgml,v 2.70 2003/05/08 22:19:55 tgl Exp $
+ $Header: /cvsroot/pgsql/doc/src/sgml/catalogs.sgml,v 2.71 2003/05/28 16:03:55 tgl Exp $
-->
<chapter id="catalogs">
<entry>The OID of the <structname>pg_class</> entry for the table this index is for</entry>
</row>
- <row>
- <entry><structfield>indproc</structfield></entry>
- <entry><type>regproc</type></entry>
- <entry><literal>pg_proc.oid</literal></entry>
- <entry>The function's OID if this is a functional index,
- else zero</entry>
- </row>
-
<row>
<entry><structfield>indkey</structfield></entry>
<entry><type>int2vector</type></entry>
<entry>pg_attribute.attnum</entry>
<entry>
- This is an array of up to
- <symbol>INDEX_MAX_KEYS</symbol> values that indicate which
- table columns this index pertains to. For example a value of
- <literal>1 3</literal> would mean that the first and the third
- column make up the index key. For a functional index, these
- columns are the inputs to the function, and the function's return
- value is the index key.
+ This is an array of <structfield>indnatts</structfield> (up to
+ <symbol>INDEX_MAX_KEYS</symbol>) values that indicate which
+ table columns this index indexes. For example a value of
+ <literal>1 3</literal> would mean that the first and the third table
+ columns make up the index key. A zero in this array indicates that the
+ corresponding index attribute is an expression over the table columns,
+ rather than a simple column reference.
</entry>
</row>
<entry><type>oidvector</type></entry>
<entry>pg_opclass.oid</entry>
<entry>
- For each column in the index key this contains a reference to
+ For each column in the index key this contains the OID of
the operator class to use. See
<structname>pg_opclass</structname> for details.
</entry>
</row>
<row>
- <entry><structfield>indisclustered</structfield></entry>
- <entry><type>bool</type></entry>
+ <entry><structfield>indnatts</structfield></entry>
+ <entry><type>int2</type></entry>
<entry></entry>
- <entry>If true, the table was last clustered on this index.</entry>
+ <entry>The number of columns in the index (duplicates
+ <literal>pg_class.relnatts</literal>)</entry>
</row>
<row>
</row>
<row>
- <entry><structfield>indreference</structfield></entry>
- <entry><type>oid</type></entry>
+ <entry><structfield>indisclustered</structfield></entry>
+ <entry><type>bool</type></entry>
<entry></entry>
- <entry>unused</entry>
+ <entry>If true, the table was last clustered on this index.</entry>
+ </row>
+
+ <row>
+ <entry><structfield>indexprs</structfield></entry>
+ <entry><type>text</type></entry>
+ <entry></entry>
+ <entry>Expression trees (in <function>nodeToString()</function> representation)
+ for index attributes that are not simple column references. This is a
+ list with one element for each zero entry in <structfield>indkey</>.
+ Null if all index attributes are simple references.</entry>
</row>
<row>
<entry><structfield>indpred</structfield></entry>
<entry><type>text</type></entry>
<entry></entry>
- <entry>Expression tree (in the form of a <function>nodeToString()</function> representation)
- for partial index predicate. Empty string if not a partial
- index.</entry>
+ <entry>Expression tree (in <function>nodeToString()</function> representation)
+ for partial index predicate. Null if not a partial index.</entry>
</row>
</tbody>
</tgroup>
-<!-- $Header: /cvsroot/pgsql/doc/src/sgml/indices.sgml,v 1.41 2003/05/15 15:50:18 petere Exp $ -->
+<!-- $Header: /cvsroot/pgsql/doc/src/sgml/indices.sgml,v 1.42 2003/05/28 16:03:55 tgl Exp $ -->
<chapter id="indexes">
<title id="indexes-title">Indexes</title>
<title>Introduction</title>
<para>
- The classical example for the need of an index is if there is a
- table similar to this:
+ Suppose we have a table similar to this:
<programlisting>
CREATE TABLE test1 (
id integer,
<programlisting>
SELECT content FROM test1 WHERE id = <replaceable>constant</replaceable>;
</programlisting>
- Ordinarily, the system would have to scan the entire
- <structname>test1</structname> table row by row to find all
+ With no advance preparation, the system would have to scan the entire
+ <structname>test1</structname> table, row by row, to find all
matching entries. If there are a lot of rows in
- <structname>test1</structname> and only a few rows (possibly zero
- or one) returned by the query, then this is clearly an inefficient
- method. If the system were instructed to maintain an index on the
- <structfield>id</structfield> column, then it could use a more
+ <structname>test1</structname> and only a few rows (perhaps only zero
+ or one) that would be returned by such a query, then this is clearly an
+ inefficient method. But if the system has been instructed to maintain an
+ index on the <structfield>id</structfield> column, then it can use a more
efficient method for locating matching rows. For instance, it
might only have to walk a few levels deep into a search tree.
</para>
<para>
- A similar approach is used in most books of non-fiction: Terms and
+ A similar approach is used in most books of non-fiction: terms and
concepts that are frequently looked up by readers are collected in
an alphabetic index at the end of the book. The interested reader
can scan the index relatively quickly and flip to the appropriate
- page, and would not have to read the entire book to find the
- interesting location. As it is the task of the author to
+ page(s), rather than having to read the entire book to find the
+ material of interest. Just as it is the task of the author to
anticipate the items that the readers are most likely to look up,
it is the task of the database programmer to foresee which indexes
would be of advantage.
<para>
Once the index is created, no further intervention is required: the
- system will use the index when it thinks it would be more efficient
+ system will update the index when the table is modified, and it will
+ use the index in queries when it thinks this would be more efficient
than a sequential table scan. But you may have to run the
<command>ANALYZE</command> command regularly to update
statistics to allow the query planner to make educated decisions.
Also read <xref linkend="performance-tips"> for information about
how to find out whether an index is used and when and why the
- planner may choose to <emphasis>not</emphasis> use an index.
+ planner may choose <emphasis>not</emphasis> to use an index.
</para>
<para>
than B-tree indexes, and the index size and build time for hash
indexes is much worse. Hash indexes also suffer poor performance
under high concurrency. For these reasons, hash index use is
- discouraged.
+ presently discouraged.
</para>
</note>
</para>
Currently, only the B-tree and GiST implementations support multicolumn
indexes. Up to 32 columns may be specified. (This limit can be
altered when building <productname>PostgreSQL</productname>; see the
- file <filename>pg_config.h</filename>.)
+ file <filename>pg_config_manual.h</filename>.)
</para>
<para>
The query planner can use a multicolumn index for queries that
- involve the leftmost column in the index definition and any number
- of columns listed to the right of it without a gap (when
- used with appropriate operators). For example,
+ involve the leftmost column in the index definition plus any number
+ of columns listed to the right of it, without a gap. For example,
an index on <literal>(a, b, c)</literal> can be used in queries
involving all of <literal>a</literal>, <literal>b</literal>, and
<literal>c</literal>, or in queries involving both
(In a query involving <literal>a</literal> and <literal>c</literal>
the planner might choose to use the index for
<literal>a</literal> only and treat <literal>c</literal> like an
- ordinary unindexed column.)
+ ordinary unindexed column.) Of course, each column must be used with
+ operators appropriate to the index type; clauses that involve other
+ operators will not be considered.
</para>
<para>
<para>
Multicolumn indexes should be used sparingly. Most of the time,
an index on a single column is sufficient and saves space and time.
- Indexes with more than three columns are almost certainly
- inappropriate.
+ Indexes with more than three columns are unlikely to be helpful
+ unless the usage of the table is extremely stylized.
</para>
</sect1>
</sect1>
- <sect1 id="indexes-functional">
- <title>Functional Indexes</title>
+ <sect1 id="indexes-expressional">
+ <title>Indexes on Expressions</title>
- <indexterm zone="indexes-functional">
+ <indexterm zone="indexes-expressional">
<primary>indexes</primary>
- <secondary>on functions</secondary>
+ <secondary>on expressions</secondary>
</indexterm>
<para>
- For a <firstterm>functional index</firstterm>, an index is defined
- on the result of a function applied to one or more columns of a
- single table. Functional indexes can be used to obtain fast access
- to data based on the result of function calls.
+ An index column need not be just a column of the underlying table,
+ but can be a function or scalar expression computed from one or
+ more columns of the table. This feature is useful to obtain fast
+ access to tables based on the results of computations.
</para>
<para>
</para>
<para>
- The function in the index definition can take more than one
- argument, but they must be table columns, not constants.
- Functional indexes are always single-column (namely, the function
- result) even if the function uses more than one input column; there
- cannot be multicolumn indexes that contain function calls.
+ As another example, if one often does queries like this:
+<programlisting>
+SELECT * FROM people WHERE (first_name || ' ' || last_name) = 'John Smith';
+</programlisting>
+ then it might be worth creating an index like this:
+<programlisting>
+CREATE INDEX people_names ON people ((first_name || ' ' || last_name));
+</programlisting>
</para>
- <tip>
- <para>
- The restrictions mentioned in the previous paragraph can easily be
- worked around by defining a custom function to use in the index
- definition that computes any desired result internally.
- </para>
- </tip>
+ <para>
+ The syntax of the <command>CREATE INDEX</> command normally requires
+ writing parentheses around index expressions, as shown in the second
+ example. The parentheses may be omitted when the expression is just
+ a function call, as in the first example.
+ </para>
+
+ <para>
+ Index expressions are relatively expensive to maintain, since the
+ derived expression(s) must be computed for each row upon insertion
+ or whenever it is updated. Therefore they should be used only when
+ queries that can use the index are very frequent.
+ </para>
</sect1>
The operator class identifies the operators to be used by the index
for that column. For example, a B-tree index on the type <type>int4</type>
would use the <literal>int4_ops</literal> class; this operator
- class includes comparison functions for values of type <type>int4</type>. In
- practice the default operator class for the column's data type is
+ class includes comparison functions for values of type <type>int4</type>.
+ In practice the default operator class for the column's data type is
usually sufficient. The main point of having operator classes is
that for some data types, there could be more than one meaningful
ordering. For example, we might want to sort a complex-number data
<literal>name_pattern_ops</literal> support B-tree indexes on
the types <type>text</type>, <type>varchar</type>,
<type>char</type>, and <type>name</type>, respectively. The
- difference to the ordinary operator classes is that the values
+ difference from the ordinary operator classes is that the values
are compared strictly character by character rather than
according to the locale-specific collation rules. This makes
these operator classes suitable for use by queries involving
pattern matching expressions (<literal>LIKE</literal> or POSIX
regular expressions) if the server does not use the standard
- <quote>C</quote> locale. As an example, to index a
+ <quote>C</quote> locale. As an example, you might index a
<type>varchar</type> column like this:
<programlisting>
CREATE INDEX test_index ON test_table (col varchar_pattern_ops);
</programlisting>
- If you do use the C locale, you should instead create an index
- with the default operator class. Also note that you should
+ If you do use the C locale, you may instead create an index
+ with the default operator class, and it will still be useful
+ for pattern-matching queries. Also note that you should
create an index with the default operator class if you want
queries involving ordinary comparisons to use an index. Such
queries cannot use the
<literal><replaceable>xxx</replaceable>_pattern_ops</literal>
- operator classes. It is possible, however, to create multiple
+ operator classes. It is allowed to create multiple
indexes on the same column with different operator classes.
</para>
</listitem>
<!--
-$Header: /cvsroot/pgsql/doc/src/sgml/plpgsql.sgml,v 1.18 2003/04/27 22:21:22 tgl Exp $
+$Header: /cvsroot/pgsql/doc/src/sgml/plpgsql.sgml,v 1.19 2003/05/28 16:03:55 tgl Exp $
-->
<chapter id="plpgsql">
<para>
Except for input/output conversion and calculation functions
for user-defined types, anything that can be defined in C language
- functions can also be done with <application>PL/pgSQL</application>. For example, it is possible to
+ functions can also be done with <application>PL/pgSQL</application>.
+ For example, it is possible to
create complex conditional computation functions and later use
- them to define operators or use them in functional indexes.
+ them to define operators or use them in index expressions.
</para>
<sect2 id="plpgsql-advantages">
<!--
-$Header: /cvsroot/pgsql/doc/src/sgml/ref/create_index.sgml,v 1.38 2003/04/22 10:08:08 petere Exp $
+$Header: /cvsroot/pgsql/doc/src/sgml/ref/create_index.sgml,v 1.39 2003/05/28 16:03:55 tgl Exp $
PostgreSQL documentation
-->
<refsynopsisdiv>
<synopsis>
-CREATE [ UNIQUE ] INDEX <replaceable class="parameter">index_name</replaceable> ON <replaceable class="parameter">table</replaceable>
- [ USING <replaceable class="parameter">method</replaceable> ] ( <replaceable class="parameter">column</replaceable> [ <replaceable class="parameter">ops_name</replaceable> ] [, ...] )
- [ WHERE <replaceable class="parameter">predicate</replaceable> ]
-
-CREATE [ UNIQUE ] INDEX <replaceable class="parameter">index_name</replaceable> ON <replaceable class="parameter">table</replaceable>
- [ USING <replaceable class="parameter">method</replaceable> ] ( <replaceable class="parameter">func_name</replaceable>( <replaceable class="parameter">column</replaceable> [, ... ]) [ <replaceable class="parameter">ops_name</replaceable> ] )
+CREATE [ UNIQUE ] INDEX <replaceable class="parameter">index_name</replaceable> ON <replaceable class="parameter">table</replaceable> [ USING <replaceable class="parameter">method</replaceable> ]
+ ( { <replaceable class="parameter">column</replaceable> | ( <replaceable class="parameter">expression</replaceable> ) } [ <replaceable class="parameter">opclass</replaceable> ] [, ...] )
[ WHERE <replaceable class="parameter">predicate</replaceable> ]
</synopsis>
</refsynopsisdiv>
<para>
<command>CREATE INDEX</command> constructs an index <replaceable
class="parameter">index_name</replaceable> on the specified table.
- Indexes are primarily used to enhance database performance. But
- inappropriate use will result in slower performance.
+ Indexes are primarily used to enhance database performance (though
+ inappropriate use will result in slower performance).
</para>
<para>
- In the first syntax shown above, the key field(s) for the
- index are specified as column names.
+ The key field(s) for the index are specified as column names,
+ or alternatively as expressions written in parentheses.
Multiple fields can be specified if the index method supports
multicolumn indexes.
</para>
<para>
- In the second syntax shown above, an index is defined on the result
- of a user-specified function <replaceable
- class="parameter">func_name</replaceable> applied to one or more
- columns of a single table. These <firstterm>functional
- indexes</firstterm> can be used to obtain fast access to data based
- on operators that would normally require some transformation to apply
- them to the base data. For example, a functional index on
+ An index field can be an expression computed from the values of
+ one or more columns of the table row. This feature can be used
+ to obtain fast access to data based on some transformation of
+ the basic data. For example, an index computed on
<literal>upper(col)</> would allow the clause
<literal>WHERE upper(col) = 'JIM'</> to use an index.
</para>
only to columns of the underlying table (but it can use all columns,
not only the one(s) being indexed). Presently, subqueries and
aggregate expressions are also forbidden in <literal>WHERE</literal>.
+ The same restrictions apply to index fields that are expressions.
</para>
<para>
their arguments and never on any outside influence (such as
the contents of another table or the current time). This restriction
ensures that the behavior of the index is well-defined. To use a
- user-defined function in an index, remember to mark the function immutable
- when you create it.
+ user-defined function in an index expression or <literal>WHERE</literal>
+ clause, remember to mark the function immutable when you create it.
</para>
</refsect1>
</varlistentry>
<varlistentry>
- <term><replaceable class="parameter">ops_name</replaceable></term>
+ <term><replaceable class="parameter">expression</replaceable></term>
<listitem>
<para>
- An associated operator class. See below for details.
+ An expression based on one or more columns of the table. The
+ expression usually must be written with surrounding parentheses,
+ as shown in the syntax. However, the parentheses may be omitted
+ if the expression has the form of a function call.
</para>
</listitem>
</varlistentry>
<varlistentry>
- <term><replaceable class="parameter">func_name</replaceable></term>
+ <term><replaceable class="parameter">opclass</replaceable></term>
<listitem>
<para>
- A function, which returns a value that can be indexed.
+ The name of an operator class. See below for details.
</para>
</listitem>
</varlistentry>
<term><replaceable class="parameter">predicate</replaceable></term>
<listitem>
<para>
- Defines the constraint expression for a partial index.
+ The constraint expression for a partial index.
</para>
</listitem>
</varlistentry>
<!--
-$Header: /cvsroot/pgsql/doc/src/sgml/release.sgml,v 1.191 2003/05/26 18:58:26 tgl Exp $
+$Header: /cvsroot/pgsql/doc/src/sgml/release.sgml,v 1.192 2003/05/28 16:03:55 tgl Exp $
-->
<appendix id="release">
worries about funny characters.
-->
<literallayout><![CDATA[
+Functional indexes have been generalized into expressional indexes
CHAR(n) to TEXT conversion automatically strips trailing blanks
Pattern matching operations can use indexes regardless of locale
New frontend/backend protocol supports many long-requested features
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/bootstrap/bootparse.y,v 1.57 2003/05/27 17:49:45 momjian Exp $
+ * $Header: /cvsroot/pgsql/src/backend/bootstrap/bootparse.y,v 1.58 2003/05/28 16:03:55 tgl Exp $
*
*-------------------------------------------------------------------------
*/
{
IndexElem *n = makeNode(IndexElem);
n->name = LexIDStr($1);
- n->funcname = n->args = NIL; /* no func indexes */
+ n->expr = NULL;
n->opclass = makeList1(makeString(LexIDStr($2)));
$$ = n;
}
* Portions Copyright (c) 1994, Regents of the University of California
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/bootstrap/bootstrap.c,v 1.158 2003/05/27 17:49:45 momjian Exp $
+ * $Header: /cvsroot/pgsql/src/backend/bootstrap/bootstrap.c,v 1.159 2003/05/28 16:03:55 tgl Exp $
*
*-------------------------------------------------------------------------
*/
newind->il_info = (IndexInfo *) palloc(sizeof(IndexInfo));
memcpy(newind->il_info, indexInfo, sizeof(IndexInfo));
+ /* expressions will likely be null, but may as well copy it */
+ newind->il_info->ii_Expressions = (List *)
+ copyObject(indexInfo->ii_Expressions);
+ newind->il_info->ii_ExpressionsState = NIL;
/* predicate will likely be null, but may as well copy it */
newind->il_info->ii_Predicate = (List *)
copyObject(indexInfo->ii_Predicate);
* Portions Copyright (c) 1994, Regents of the University of California
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/catalog/dependency.c,v 1.24 2003/05/27 17:49:45 momjian Exp $
+ * $Header: /cvsroot/pgsql/src/backend/catalog/dependency.c,v 1.25 2003/05/28 16:03:55 tgl Exp $
*
*-------------------------------------------------------------------------
*/
term_object_addresses(&context.addrs);
}
+/*
+ * recordDependencyOnSingleRelExpr - find expression dependencies
+ *
+ * As above, but only one relation is expected to be referenced (with
+ * varno = 1 and varlevelsup = 0). Pass the relation OID instead of a
+ * range table. An additional frammish is that dependencies on that
+ * relation (or its component columns) will be marked with 'self_behavior',
+ * whereas 'behavior' is used for everything else.
+ */
+void
+recordDependencyOnSingleRelExpr(const ObjectAddress *depender,
+ Node *expr, Oid relId,
+ DependencyType behavior,
+ DependencyType self_behavior)
+{
+ find_expr_references_context context;
+ RangeTblEntry rte;
+
+ init_object_addresses(&context.addrs);
+
+ /* We gin up a rather bogus rangetable list to handle Vars */
+ MemSet(&rte, 0, sizeof(rte));
+ rte.type = T_RangeTblEntry;
+ rte.rtekind = RTE_RELATION;
+ rte.relid = relId;
+
+ context.rtables = makeList1(makeList1(&rte));
+
+ /* Scan the expression tree for referenceable objects */
+ find_expr_references_walker(expr, &context);
+
+ /* Remove any duplicates */
+ eliminate_duplicate_dependencies(&context.addrs);
+
+ /* Separate self-dependencies if necessary */
+ if (behavior != self_behavior && context.addrs.numrefs > 0)
+ {
+ ObjectAddresses self_addrs;
+ ObjectAddress *outobj;
+ int oldref,
+ outrefs;
+
+ init_object_addresses(&self_addrs);
+
+ outobj = context.addrs.refs;
+ outrefs = 0;
+ for (oldref = 0; oldref < context.addrs.numrefs; oldref++)
+ {
+ ObjectAddress *thisobj = context.addrs.refs + oldref;
+
+ if (thisobj->classId == RelOid_pg_class &&
+ thisobj->objectId == relId)
+ {
+ /* Move this ref into self_addrs */
+ add_object_address(OCLASS_CLASS, relId, thisobj->objectSubId,
+ &self_addrs);
+ }
+ else
+ {
+ /* Keep it in context.addrs */
+ outobj->classId = thisobj->classId;
+ outobj->objectId = thisobj->objectId;
+ outobj->objectSubId = thisobj->objectSubId;
+ outobj++;
+ outrefs++;
+ }
+ }
+ context.addrs.numrefs = outrefs;
+
+ /* Record the self-dependencies */
+ recordMultipleDependencies(depender,
+ self_addrs.refs, self_addrs.numrefs,
+ self_behavior);
+
+ term_object_addresses(&self_addrs);
+ }
+
+ /* Record the external dependencies */
+ recordMultipleDependencies(depender,
+ context.addrs.refs, context.addrs.numrefs,
+ behavior);
+
+ term_object_addresses(&context.addrs);
+}
+
/*
* Recursively search an expression tree for object references.
*
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/catalog/heap.c,v 1.244 2003/05/12 00:17:02 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/catalog/heap.c,v 1.245 2003/05/28 16:03:55 tgl Exp $
*
*
* INTERFACE ROUTINES
static void
RelationTruncateIndexes(Oid heapId)
{
- Relation indexRelation;
- ScanKeyData entry;
- SysScanDesc scan;
- HeapTuple indexTuple;
-
- /* Scan pg_index to find indexes on specified heap */
- indexRelation = heap_openr(IndexRelationName, AccessShareLock);
- ScanKeyEntryInitialize(&entry, 0,
- Anum_pg_index_indrelid,
- F_OIDEQ,
- ObjectIdGetDatum(heapId));
- scan = systable_beginscan(indexRelation, IndexIndrelidIndex, true,
- SnapshotNow, 1, &entry);
-
- while (HeapTupleIsValid(indexTuple = systable_getnext(scan)))
- {
- Form_pg_index indexform = (Form_pg_index) GETSTRUCT(indexTuple);
- Oid indexId;
- IndexInfo *indexInfo;
- Relation heapRelation,
- currentIndex;
+ Relation heapRelation;
+ List *indlist;
- /*
- * For each index, fetch info needed for index_build
- */
- indexId = indexform->indexrelid;
- indexInfo = BuildIndexInfo(indexform);
+ /*
+ * Open the heap rel. We need grab no lock because we assume
+ * heap_truncate is holding an exclusive lock on the heap rel.
+ */
+ heapRelation = heap_open(heapId, NoLock);
- /*
- * We have to re-open the heap rel each time through this loop
- * because index_build will close it again. We need grab no lock,
- * however, because we assume heap_truncate is holding an
- * exclusive lock on the heap rel.
- */
- heapRelation = heap_open(heapId, NoLock);
+ /* Ask the relcache to produce a list of the indexes of the rel */
+ foreach(indlist, RelationGetIndexList(heapRelation))
+ {
+ Oid indexId = lfirsto(indlist);
+ Relation currentIndex;
+ IndexInfo *indexInfo;
/* Open the index relation */
currentIndex = index_open(indexId);
/* Obtain exclusive lock on it, just to be sure */
LockRelation(currentIndex, AccessExclusiveLock);
+ /* Fetch info needed for index_build */
+ indexInfo = BuildIndexInfo(currentIndex);
+
/*
* Drop any buffers associated with this index. If they're dirty,
* they're just dropped without bothering to flush to disk.
/*
* index_build will close both the heap and index relations (but
- * not give up the locks we hold on them).
+ * not give up the locks we hold on them). We're done with this
+ * index, but we must re-open the heap rel.
*/
+ heapRelation = heap_open(heapId, NoLock);
}
- /* Complete the scan and close pg_index */
- systable_endscan(scan);
- heap_close(indexRelation, AccessShareLock);
+ /* Finish by closing the heap rel again */
+ heap_close(heapRelation, NoLock);
}
/*
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/catalog/index.c,v 1.209 2003/02/19 04:06:28 momjian Exp $
+ * $Header: /cvsroot/pgsql/src/backend/catalog/index.c,v 1.210 2003/05/28 16:03:56 tgl Exp $
*
*
* INTERFACE ROUTINES
#include "miscadmin.h"
#include "optimizer/clauses.h"
#include "optimizer/prep.h"
+#include "parser/parse_expr.h"
#include "parser/parse_func.h"
#include "storage/sinval.h"
#include "storage/smgr.h"
((natts) * AVG_ATTR_SIZE + MAXALIGN(sizeof(HeapTupleHeaderData))))
/* non-export function prototypes */
-static TupleDesc BuildFuncTupleDesc(Oid funcOid,
- Oid *classObjectId);
static TupleDesc ConstructTupleDescriptor(Relation heapRelation,
- int numatts, AttrNumber *attNums,
- Oid *classObjectId);
+ IndexInfo *indexInfo,
+ Oid *classObjectId);
static void UpdateRelationRelation(Relation indexRelation);
static void InitializeAttributeOids(Relation indexRelation,
int numatts, Oid indexoid);
return reindexing;
}
-static TupleDesc
-BuildFuncTupleDesc(Oid funcOid,
- Oid *classObjectId)
-{
- TupleDesc funcTupDesc;
- HeapTuple tuple;
- Oid keyType;
- Oid retType;
- Form_pg_type typeTup;
-
- /*
- * Allocate and zero a tuple descriptor for a one-column tuple.
- */
- funcTupDesc = CreateTemplateTupleDesc(1, false);
- funcTupDesc->attrs[0] = (Form_pg_attribute) palloc0(ATTRIBUTE_TUPLE_SIZE);
-
- /*
- * Lookup the function to get its name and return type.
- */
- tuple = SearchSysCache(PROCOID,
- ObjectIdGetDatum(funcOid),
- 0, 0, 0);
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "Function %u does not exist", funcOid);
- retType = ((Form_pg_proc) GETSTRUCT(tuple))->prorettype;
-
- /*
- * make the attributes name the same as the functions
- */
- namestrcpy(&funcTupDesc->attrs[0]->attname,
- NameStr(((Form_pg_proc) GETSTRUCT(tuple))->proname));
-
- ReleaseSysCache(tuple);
-
- /*
- * Check the opclass to see if it provides a keytype (overriding the
- * function result type).
- */
- tuple = SearchSysCache(CLAOID,
- ObjectIdGetDatum(classObjectId[0]),
- 0, 0, 0);
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "Opclass %u does not exist", classObjectId[0]);
- keyType = ((Form_pg_opclass) GETSTRUCT(tuple))->opckeytype;
- ReleaseSysCache(tuple);
-
- if (!OidIsValid(keyType))
- keyType = retType;
-
- /*
- * Lookup the key type in pg_type for the type length etc.
- */
- tuple = SearchSysCache(TYPEOID,
- ObjectIdGetDatum(keyType),
- 0, 0, 0);
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "Type %u does not exist", keyType);
- typeTup = (Form_pg_type) GETSTRUCT(tuple);
-
- /*
- * Assign some of the attributes values. Leave the rest as 0.
- */
- funcTupDesc->attrs[0]->attnum = 1;
- funcTupDesc->attrs[0]->atttypid = keyType;
- funcTupDesc->attrs[0]->attlen = typeTup->typlen;
- funcTupDesc->attrs[0]->attbyval = typeTup->typbyval;
- funcTupDesc->attrs[0]->attstorage = typeTup->typstorage;
- funcTupDesc->attrs[0]->attalign = typeTup->typalign;
- funcTupDesc->attrs[0]->attcacheoff = -1;
- funcTupDesc->attrs[0]->atttypmod = -1;
- funcTupDesc->attrs[0]->attislocal = true;
-
- ReleaseSysCache(tuple);
-
- return funcTupDesc;
-}
-
-/* ----------------------------------------------------------------
+/*
* ConstructTupleDescriptor
*
- * Build an index tuple descriptor for a new index (plain not functional)
- * ----------------------------------------------------------------
+ * Build an index tuple descriptor for a new index
*/
static TupleDesc
ConstructTupleDescriptor(Relation heapRelation,
- int numatts,
- AttrNumber *attNums,
+ IndexInfo *indexInfo,
Oid *classObjectId)
{
+ int numatts = indexInfo->ii_NumIndexAttrs;
+ List *indexprs = indexInfo->ii_Expressions;
TupleDesc heapTupDesc;
TupleDesc indexTupDesc;
int natts; /* #atts in heap rel --- for error checks */
/*
* allocate the new tuple descriptor
*/
-
indexTupDesc = CreateTemplateTupleDesc(numatts, false);
- /* ----------------
- * for each attribute we are indexing, obtain its attribute
- * tuple form from either the static table of system attribute
- * tuple forms or the relation tuple descriptor
- * ----------------
+ /*
+ * For simple index columns, we copy the pg_attribute row from the
+ * parent relation and modify it as necessary. For expressions we
+ * have to cons up a pg_attribute row the hard way.
*/
for (i = 0; i < numatts; i++)
{
- AttrNumber atnum; /* attributeNumber[attributeOffset] */
- Form_pg_attribute from;
+ AttrNumber atnum = indexInfo->ii_KeyAttrNumbers[i];
Form_pg_attribute to;
HeapTuple tuple;
+ Form_pg_type typeTup;
Oid keyType;
- /*
- * get the attribute number and make sure it's valid; determine
- * which attribute descriptor to copy
- */
- atnum = attNums[i];
+ indexTupDesc->attrs[i] = to =
+ (Form_pg_attribute) palloc0(ATTRIBUTE_TUPLE_SIZE);
- if (!AttrNumberIsForUserDefinedAttr(atnum))
+ if (atnum != 0)
{
+ /* Simple index column */
+ Form_pg_attribute from;
+
+ if (atnum < 0)
+ {
+ /*
+ * here we are indexing on a system attribute (-1...-n)
+ */
+ from = SystemAttributeDefinition(atnum,
+ heapRelation->rd_rel->relhasoids);
+ }
+ else
+ {
+ /*
+ * here we are indexing on a normal attribute (1...n)
+ */
+ if (atnum > natts)
+ elog(ERROR, "cannot create index: column %d does not exist",
+ atnum);
+ from = heapTupDesc->attrs[AttrNumberGetAttrOffset(atnum)];
+ }
+
+ /*
+ * now that we've determined the "from", let's copy the tuple desc
+ * data...
+ */
+ memcpy(to, from, ATTRIBUTE_TUPLE_SIZE);
+
/*
- * here we are indexing on a system attribute (-1...-n)
+ * Fix the stuff that should not be the same as the underlying
+ * attr
*/
- from = SystemAttributeDefinition(atnum,
- heapRelation->rd_rel->relhasoids);
+ to->attnum = i + 1;
+
+ to->attstattarget = 0;
+ to->attcacheoff = -1;
+ to->attnotnull = false;
+ to->atthasdef = false;
+ to->attislocal = true;
+ to->attinhcount = 0;
}
else
{
+ /* Expressional index */
+ Node *indexkey;
+
+ if (indexprs == NIL)
+ elog(ERROR, "too few entries in indexprs list");
+ indexkey = (Node *) lfirst(indexprs);
+ indexprs = lnext(indexprs);
+
/*
- * here we are indexing on a normal attribute (1...n)
+ * Make the attribute's name "pg_expresssion_nnn" (maybe think
+ * of something better later)
*/
- if (atnum > natts)
- elog(ERROR, "cannot create index: column %d does not exist",
- atnum);
+ sprintf(NameStr(to->attname), "pg_expression_%d", i + 1);
- from = heapTupDesc->attrs[AttrNumberGetAttrOffset(atnum)];
- }
-
- /*
- * now that we've determined the "from", let's copy the tuple desc
- * data...
- */
- indexTupDesc->attrs[i] = to =
- (Form_pg_attribute) palloc(ATTRIBUTE_TUPLE_SIZE);
- memcpy(to, from, ATTRIBUTE_TUPLE_SIZE);
+ /*
+ * Lookup the expression type in pg_type for the type length etc.
+ */
+ keyType = exprType(indexkey);
+ tuple = SearchSysCache(TYPEOID,
+ ObjectIdGetDatum(keyType),
+ 0, 0, 0);
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "Type %u does not exist", keyType);
+ typeTup = (Form_pg_type) GETSTRUCT(tuple);
- /*
- * Fix the stuff that should not be the same as the underlying
- * attr
- */
- to->attnum = i + 1;
+ /*
+ * Assign some of the attributes values. Leave the rest as 0.
+ */
+ to->attnum = i + 1;
+ to->atttypid = keyType;
+ to->attlen = typeTup->typlen;
+ to->attbyval = typeTup->typbyval;
+ to->attstorage = typeTup->typstorage;
+ to->attalign = typeTup->typalign;
+ to->attcacheoff = -1;
+ to->atttypmod = -1;
+ to->attislocal = true;
- to->attstattarget = 0;
- to->attcacheoff = -1;
- to->attnotnull = false;
- to->atthasdef = false;
- to->attislocal = true;
- to->attinhcount = 0;
+ ReleaseSysCache(tuple);
+ }
/*
* We do not yet have the correct relation OID for the index, so
if (OidIsValid(keyType) && keyType != to->atttypid)
{
/* index value and heap value have different types */
- Form_pg_type typeTup;
-
tuple = SearchSysCache(TYPEOID,
ObjectIdGetDatum(keyType),
0, 0, 0);
{
int16 indkey[INDEX_MAX_KEYS];
Oid indclass[INDEX_MAX_KEYS];
+ Datum exprsDatum;
Datum predDatum;
Datum values[Natts_pg_index];
char nulls[Natts_pg_index];
/*
* Copy the index key and opclass info into zero-filled vectors
- *
- * (zero filling is essential 'cause we don't store the number of
- * index columns explicitly in pg_index, which is pretty grotty...)
*/
MemSet(indkey, 0, sizeof(indkey));
- for (i = 0; i < indexInfo->ii_NumKeyAttrs; i++)
- indkey[i] = indexInfo->ii_KeyAttrNumbers[i];
-
MemSet(indclass, 0, sizeof(indclass));
for (i = 0; i < indexInfo->ii_NumIndexAttrs; i++)
+ {
+ indkey[i] = indexInfo->ii_KeyAttrNumbers[i];
indclass[i] = classOids[i];
+ }
+
+ /*
+ * Convert the index expressions (if any) to a text datum
+ */
+ if (indexInfo->ii_Expressions != NIL)
+ {
+ char *exprsString;
+
+ exprsString = nodeToString(indexInfo->ii_Expressions);
+ exprsDatum = DirectFunctionCall1(textin,
+ CStringGetDatum(exprsString));
+ pfree(exprsString);
+ }
+ else
+ exprsDatum = (Datum) 0;
/*
- * Convert the index-predicate (if any) to a text datum
+ * Convert the index predicate (if any) to a text datum
*/
if (indexInfo->ii_Predicate != NIL)
{
pfree(predString);
}
else
- predDatum = DirectFunctionCall1(textin,
- CStringGetDatum(""));
+ predDatum = (Datum) 0;
/*
* open the system catalog index relation
values[Anum_pg_index_indexrelid - 1] = ObjectIdGetDatum(indexoid);
values[Anum_pg_index_indrelid - 1] = ObjectIdGetDatum(heapoid);
- values[Anum_pg_index_indproc - 1] = ObjectIdGetDatum(indexInfo->ii_FuncOid);
values[Anum_pg_index_indkey - 1] = PointerGetDatum(indkey);
values[Anum_pg_index_indclass - 1] = PointerGetDatum(indclass);
- values[Anum_pg_index_indisclustered - 1] = BoolGetDatum(false);
+ values[Anum_pg_index_indnatts - 1] = Int16GetDatum(indexInfo->ii_NumIndexAttrs);
values[Anum_pg_index_indisunique - 1] = BoolGetDatum(indexInfo->ii_Unique);
values[Anum_pg_index_indisprimary - 1] = BoolGetDatum(primary);
- values[Anum_pg_index_indreference - 1] = ObjectIdGetDatum(InvalidOid);
+ values[Anum_pg_index_indisclustered - 1] = BoolGetDatum(false);
+ values[Anum_pg_index_indexprs - 1] = exprsDatum;
+ if (exprsDatum == (Datum) 0)
+ nulls[Anum_pg_index_indexprs - 1] = 'n';
values[Anum_pg_index_indpred - 1] = predDatum;
+ if (predDatum == (Datum) 0)
+ nulls[Anum_pg_index_indpred - 1] = 'n';
tuple = heap_formtuple(RelationGetDescr(pg_index), values, nulls);
/*
* check parameters
*/
- if (indexInfo->ii_NumIndexAttrs < 1 ||
- indexInfo->ii_NumKeyAttrs < 1)
+ if (indexInfo->ii_NumIndexAttrs < 1)
elog(ERROR, "must index at least one column");
if (!allow_system_table_mods &&
/*
* construct tuple descriptor for index tuples
*/
- if (OidIsValid(indexInfo->ii_FuncOid))
- indexTupDesc = BuildFuncTupleDesc(indexInfo->ii_FuncOid,
- classObjectId);
- else
- indexTupDesc = ConstructTupleDescriptor(heapRelation,
- indexInfo->ii_NumKeyAttrs,
- indexInfo->ii_KeyAttrNumbers,
- classObjectId);
-
- indexTupDesc->tdhasoid = false;
+ indexTupDesc = ConstructTupleDescriptor(heapRelation,
+ indexInfo,
+ classObjectId);
/*
* create the index relation's relcache entry and physical disk file.
constraintType = 0; /* keep compiler quiet */
}
+ /* Shouldn't have any expressions */
+ if (indexInfo->ii_Expressions)
+ elog(ERROR, "constraints can't have index expressions");
+
conOid = CreateConstraintEntry(indexRelationName,
namespaceId,
constraintType,
false, /* isDeferred */
heapRelationId,
indexInfo->ii_KeyAttrNumbers,
- indexInfo->ii_NumKeyAttrs,
+ indexInfo->ii_NumIndexAttrs,
InvalidOid, /* no domain */
InvalidOid, /* no foreign key */
NULL,
}
else
{
- for (i = 0; i < indexInfo->ii_NumKeyAttrs; i++)
+ /* Create auto dependencies on simply-referenced columns */
+ for (i = 0; i < indexInfo->ii_NumIndexAttrs; i++)
{
- referenced.classId = RelOid_pg_class;
- referenced.objectId = heapRelationId;
- referenced.objectSubId = indexInfo->ii_KeyAttrNumbers[i];
-
- recordDependencyOn(&myself, &referenced, DEPENDENCY_AUTO);
+ if (indexInfo->ii_KeyAttrNumbers[i] != 0)
+ {
+ referenced.classId = RelOid_pg_class;
+ referenced.objectId = heapRelationId;
+ referenced.objectSubId = indexInfo->ii_KeyAttrNumbers[i];
+
+ recordDependencyOn(&myself, &referenced, DEPENDENCY_AUTO);
+ }
}
}
recordDependencyOn(&myself, &referenced, DEPENDENCY_NORMAL);
}
- /* Store the dependency on the function (if appropriate) */
- if (OidIsValid(indexInfo->ii_FuncOid))
+ /* Store dependencies on anything mentioned in index expressions */
+ if (indexInfo->ii_Expressions)
{
- referenced.classId = RelOid_pg_proc;
- referenced.objectId = indexInfo->ii_FuncOid;
- referenced.objectSubId = 0;
+ recordDependencyOnSingleRelExpr(&myself,
+ (Node *) indexInfo->ii_Expressions,
+ heapRelationId,
+ DEPENDENCY_NORMAL,
+ DEPENDENCY_AUTO);
+ }
- recordDependencyOn(&myself, &referenced, DEPENDENCY_NORMAL);
+ /* Store dependencies on anything mentioned in predicate */
+ if (indexInfo->ii_Predicate)
+ {
+ recordDependencyOnSingleRelExpr(&myself,
+ (Node *) indexInfo->ii_Predicate,
+ heapRelationId,
+ DEPENDENCY_NORMAL,
+ DEPENDENCY_AUTO);
}
}
/* ----------------
* BuildIndexInfo
- * Construct an IndexInfo record given the index's pg_index tuple
+ * Construct an IndexInfo record for an open index
*
* IndexInfo stores the information about the index that's needed by
* FormIndexDatum, which is used for both index_build() and later insertion
* ----------------
*/
IndexInfo *
-BuildIndexInfo(Form_pg_index indexStruct)
+BuildIndexInfo(Relation index)
{
IndexInfo *ii = makeNode(IndexInfo);
+ Form_pg_index indexStruct = index->rd_index;
int i;
int numKeys;
- /*
- * count the number of keys, and copy them into the IndexInfo
- */
- numKeys = 0;
- for (i = 0; i < INDEX_MAX_KEYS &&
- indexStruct->indkey[i] != InvalidAttrNumber; i++)
- {
+ /* check the number of keys, and copy attr numbers into the IndexInfo */
+ numKeys = indexStruct->indnatts;
+ if (numKeys < 1 || numKeys > INDEX_MAX_KEYS)
+ elog(ERROR, "invalid indnatts %d for index %u",
+ numKeys, RelationGetRelid(index));
+ ii->ii_NumIndexAttrs = numKeys;
+ for (i = 0; i < numKeys; i++)
ii->ii_KeyAttrNumbers[i] = indexStruct->indkey[i];
- numKeys++;
- }
- ii->ii_NumKeyAttrs = numKeys;
- /*
- * Handle functional index.
- *
- * If we have a functional index then the number of attributes defined in
- * the index must be 1 (the function's single return value). Otherwise
- * it's same as number of keys.
- */
- ii->ii_FuncOid = indexStruct->indproc;
+ /* fetch any expressions needed for expressional indexes */
+ ii->ii_Expressions = RelationGetIndexExpressions(index);
+ ii->ii_ExpressionsState = NIL;
- if (OidIsValid(indexStruct->indproc))
- {
- ii->ii_NumIndexAttrs = 1;
- /* Do a lookup on the function, too */
- fmgr_info(indexStruct->indproc, &ii->ii_FuncInfo);
- }
- else
- ii->ii_NumIndexAttrs = numKeys;
-
- /*
- * If partial index, convert predicate into expression nodetree
- */
- if (VARSIZE(&indexStruct->indpred) > VARHDRSZ)
- {
- char *predString;
-
- predString = DatumGetCString(DirectFunctionCall1(textout,
- PointerGetDatum(&indexStruct->indpred)));
- ii->ii_Predicate = stringToNode(predString);
- ii->ii_PredicateState = NIL;
- pfree(predString);
- }
- else
- {
- ii->ii_Predicate = NIL;
- ii->ii_PredicateState = NIL;
- }
+ /* fetch index predicate if any */
+ ii->ii_Predicate = RelationGetIndexPredicate(index);
+ ii->ii_PredicateState = NIL;
- /* Other info */
+ /* other info */
ii->ii_Unique = indexStruct->indisunique;
return ii;
* indexInfo Info about the index
* heapTuple Heap tuple for which we must prepare an index entry
* heapDescriptor tupledesc for heap tuple
- * resultCxt Temporary memory context for any palloc'd datums created
+ * estate executor state for evaluating any index expressions
* datum Array of index Datums (output area)
* nullv Array of is-null indicators (output area)
*
+ * When there are no index expressions, estate may be NULL. Otherwise it
+ * must be supplied, *and* the ecxt_scantuple slot of its per-tuple expr
+ * context must point to the heap tuple passed in.
+ *
* For largely historical reasons, we don't actually call index_formtuple()
* here, we just prepare its input arrays datum[] and nullv[].
* ----------------
FormIndexDatum(IndexInfo *indexInfo,
HeapTuple heapTuple,
TupleDesc heapDescriptor,
- MemoryContext resultCxt,
+ EState *estate,
Datum *datum,
char *nullv)
{
- MemoryContext oldContext;
+ List *indexprs;
int i;
- Datum iDatum;
- bool isNull;
-
- oldContext = MemoryContextSwitchTo(resultCxt);
- if (OidIsValid(indexInfo->ii_FuncOid))
+ if (indexInfo->ii_Expressions != NIL &&
+ indexInfo->ii_ExpressionsState == NIL)
{
- /*
- * Functional index --- compute the single index attribute
- */
- FunctionCallInfoData fcinfo;
- bool anynull = false;
+ /* First time through, set up expression evaluation state */
+ indexInfo->ii_ExpressionsState = (List *)
+ ExecPrepareExpr((Expr *) indexInfo->ii_Expressions,
+ estate);
+ /* Check caller has set up context correctly */
+ Assert(GetPerTupleExprContext(estate)->ecxt_scantuple->val == heapTuple);
+ }
+ indexprs = indexInfo->ii_ExpressionsState;
- MemSet(&fcinfo, 0, sizeof(fcinfo));
- fcinfo.flinfo = &indexInfo->ii_FuncInfo;
- fcinfo.nargs = indexInfo->ii_NumKeyAttrs;
+ for (i = 0; i < indexInfo->ii_NumIndexAttrs; i++)
+ {
+ int keycol = indexInfo->ii_KeyAttrNumbers[i];
+ Datum iDatum;
+ bool isNull;
- for (i = 0; i < indexInfo->ii_NumKeyAttrs; i++)
- {
- fcinfo.arg[i] = heap_getattr(heapTuple,
- indexInfo->ii_KeyAttrNumbers[i],
- heapDescriptor,
- &fcinfo.argnull[i]);
- anynull |= fcinfo.argnull[i];
- }
- if (indexInfo->ii_FuncInfo.fn_strict && anynull)
+ if (keycol != 0)
{
- /* force a null result for strict function */
- iDatum = (Datum) 0;
- isNull = true;
+ /*
+ * Plain index column; get the value we need directly from the
+ * heap tuple.
+ */
+ iDatum = heap_getattr(heapTuple, keycol, heapDescriptor, &isNull);
}
else
{
- iDatum = FunctionCallInvoke(&fcinfo);
- isNull = fcinfo.isnull;
- }
- datum[0] = iDatum;
- nullv[0] = (isNull) ? 'n' : ' ';
- }
- else
- {
- /*
- * Plain index --- for each attribute we need from the heap tuple,
- * get the attribute and stick it into the datum and nullv arrays.
- */
- for (i = 0; i < indexInfo->ii_NumIndexAttrs; i++)
- {
- iDatum = heap_getattr(heapTuple,
- indexInfo->ii_KeyAttrNumbers[i],
- heapDescriptor,
- &isNull);
- datum[i] = iDatum;
- nullv[i] = (isNull) ? 'n' : ' ';
+ /*
+ * Index expression --- need to evaluate it.
+ */
+ if (indexprs == NIL)
+ elog(ERROR, "wrong number of index expressions");
+ iDatum = ExecEvalExprSwitchContext((ExprState *) lfirst(indexprs),
+ GetPerTupleExprContext(estate),
+ &isNull,
+ NULL);
+ indexprs = lnext(indexprs);
}
+ datum[i] = iDatum;
+ nullv[i] = (isNull) ? 'n' : ' ';
}
- MemoryContextSwitchTo(oldContext);
+ if (indexprs != NIL)
+ elog(ERROR, "wrong number of index expressions");
}
heapDescriptor = RelationGetDescr(heapRelation);
/*
- * Need an EState for evaluation of functional-index functions
+ * Need an EState for evaluation of index expressions
* and partial-index predicates.
*/
estate = CreateExecutorState();
/*
* If this is a predicate (partial) index, we will need to evaluate
* the predicate using ExecQual, which requires the current tuple to
- * be in a slot of a TupleTable.
+ * be in a slot of a TupleTable. Likewise if there are any expressions.
*/
- if (indexInfo->ii_Predicate != NIL)
+ if (indexInfo->ii_Predicate != NIL || indexInfo->ii_Expressions != NIL)
{
tupleTable = ExecCreateTupleTable(1);
slot = ExecAllocTableSlot(tupleTable);
/* Arrange for econtext's scan tuple to be the tuple under test */
econtext->ecxt_scantuple = slot;
- /*
- * Set up execution state for predicate. Note: we mustn't attempt to
- * cache this in the indexInfo, since we're building it in a transient
- * EState.
- */
+ /* Set up execution state for predicate. */
predicate = (List *)
ExecPrepareExpr((Expr *) indexInfo->ii_Predicate,
estate);
/*
* For the current heap tuple, extract all the attributes we use
* in this index, and note which are null. This also performs
- * evaluation of the function, if this is a functional index.
+ * evaluation of any expressions needed.
*/
FormIndexDatum(indexInfo,
heapTuple,
heapDescriptor,
- econtext->ecxt_per_tuple_memory,
+ estate,
attdata,
nulls);
FreeExecutorState(estate);
+ /* These may have been pointing to the now-gone estate */
+ indexInfo->ii_ExpressionsState = NIL;
+ indexInfo->ii_PredicateState = NIL;
+
return reltuples;
}
}
/* Fetch info needed for index_build */
- indexInfo = BuildIndexInfo(iRel->rd_index);
+ indexInfo = BuildIndexInfo(iRel);
if (inplace)
{
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/catalog/indexing.c,v 1.102 2002/09/04 20:31:14 momjian Exp $
+ * $Header: /cvsroot/pgsql/src/backend/catalog/indexing.c,v 1.103 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
* In the current implementation, we share code for opening/closing the
* indexes with execUtils.c. But we do not use ExecInsertIndexTuples,
* because we don't want to create an EState. This implies that we
- * do not support partial indexes on system catalogs. Nor do we handle
- * functional indexes very well (the code will work, but will leak memory
- * intraquery, because the index function is called in the per-query context
- * that we are invoked in). This could be fixed with localized changes here
- * if we wanted to pay the extra overhead of building an EState.
+ * do not support partial or expressional indexes on system catalogs.
+ * This could be fixed with localized changes here if we wanted to pay
+ * the extra overhead of building an EState.
*/
CatalogIndexState
CatalogOpenIndexes(Relation heapRel)
indexInfo = indexInfoArray[i];
- /* Partial indexes on system catalogs are not supported */
+ /*
+ * Expressional and partial indexes on system catalogs are not
+ * supported
+ */
+ Assert(indexInfo->ii_Expressions == NIL);
Assert(indexInfo->ii_Predicate == NIL);
/*
FormIndexDatum(indexInfo,
heapTuple,
heapDescriptor,
- CurrentMemoryContext,
+ NULL, /* no expression eval to do */
datum,
nullv);
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/catalog/pg_constraint.c,v 1.12 2002/12/12 20:35:11 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/catalog/pg_constraint.c,v 1.13 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
{
/*
* Register dependencies from constraint to objects mentioned in
- * CHECK expression. We gin up a rather bogus rangetable list to
- * handle any Vars in the constraint.
+ * CHECK expression.
*/
- RangeTblEntry rte;
-
- MemSet(&rte, 0, sizeof(rte));
- rte.type = T_RangeTblEntry;
- rte.rtekind = RTE_RELATION;
- rte.relid = relId;
-
- recordDependencyOnExpr(&conobject, conExpr, makeList1(&rte),
- DEPENDENCY_NORMAL);
+ recordDependencyOnSingleRelExpr(&conobject, conExpr, relId,
+ DEPENDENCY_NORMAL,
+ DEPENDENCY_NORMAL);
}
return conOid;
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/commands/cluster.c,v 1.109 2003/05/14 03:26:01 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/commands/cluster.c,v 1.110 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
/* Start a new transaction for each relation. */
StartTransactionCommand();
- SetQuerySnapshot(); /* might be needed for functional index */
+ SetQuerySnapshot(); /* might be needed for functions in indexes */
cluster_rel(rvtc, true);
CommitTransactionCommand();
}
* seqscan pass over the table to copy the missing rows, but that seems
* expensive and tedious.
*/
- if (VARSIZE(&OldIndex->rd_index->indpred) > VARHDRSZ) /* partial? */
+ if (!heap_attisnull(OldIndex->rd_indextuple, Anum_pg_index_indpred))
elog(ERROR, "CLUSTER: cannot cluster on partial index");
if (!OldIndex->rd_am->amindexnulls)
{
* at the first column; multicolumn-capable AMs are *required* to
* index nulls in columns after the first.
*/
- if (OidIsValid(OldIndex->rd_index->indproc))
- elog(ERROR, "CLUSTER: cannot cluster on functional index when index access method does not handle nulls");
colno = OldIndex->rd_index->indkey[0];
- if (colno > 0) /* system columns are non-null */
+ if (colno > 0)
+ {
+ /* ordinary user attribute */
if (!OldHeap->rd_att->attrs[colno - 1]->attnotnull)
elog(ERROR, "CLUSTER: cannot cluster when index access method does not handle nulls"
"\n\tYou may be able to work around this by marking column \"%s\" NOT NULL",
NameStr(OldHeap->rd_att->attrs[colno - 1]->attname));
+ }
+ else if (colno < 0)
+ {
+ /* system column --- okay, always non-null */
+ }
+ else
+ {
+ /* index expression, lose... */
+ elog(ERROR, "CLUSTER: cannot cluster on expressional index when index access method does not handle nulls");
+ }
}
/*
foreach(indlist, RelationGetIndexList(OldHeap))
{
Oid indexOID = lfirsto(indlist);
- HeapTuple indexTuple;
- HeapTuple classTuple;
- Form_pg_index indexForm;
- Form_pg_class classForm;
+ Relation oldIndex;
IndexAttrs *attrs;
- indexTuple = SearchSysCache(INDEXRELID,
- ObjectIdGetDatum(indexOID),
- 0, 0, 0);
- if (!HeapTupleIsValid(indexTuple))
- elog(ERROR, "Cache lookup failed for index %u", indexOID);
- indexForm = (Form_pg_index) GETSTRUCT(indexTuple);
- Assert(indexForm->indexrelid == indexOID);
+ oldIndex = index_open(indexOID);
attrs = (IndexAttrs *) palloc(sizeof(IndexAttrs));
attrs->indexOID = indexOID;
- attrs->indexInfo = BuildIndexInfo(indexForm);
+ attrs->indexName = pstrdup(NameStr(oldIndex->rd_rel->relname));
+ attrs->accessMethodOID = oldIndex->rd_rel->relam;
+ attrs->indexInfo = BuildIndexInfo(oldIndex);
attrs->classOID = (Oid *)
palloc(sizeof(Oid) * attrs->indexInfo->ii_NumIndexAttrs);
- memcpy(attrs->classOID, indexForm->indclass,
+ memcpy(attrs->classOID, oldIndex->rd_index->indclass,
sizeof(Oid) * attrs->indexInfo->ii_NumIndexAttrs);
/* We adjust the isclustered attribute to correct new state */
attrs->isclustered = (indexOID == OldIndex);
- /* Name and access method of each index come from pg_class */
- classTuple = SearchSysCache(RELOID,
- ObjectIdGetDatum(indexOID),
- 0, 0, 0);
- if (!HeapTupleIsValid(classTuple))
- elog(ERROR, "Cache lookup failed for index %u", indexOID);
- classForm = (Form_pg_class) GETSTRUCT(classTuple);
-
- attrs->indexName = pstrdup(NameStr(classForm->relname));
- attrs->accessMethodOID = classForm->relam;
-
- ReleaseSysCache(classTuple);
- ReleaseSysCache(indexTuple);
+ index_close(oldIndex);
/*
* Cons the gathered data into the list. We do not care about
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/commands/indexcmds.c,v 1.99 2003/05/14 03:26:01 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/commands/indexcmds.c,v 1.100 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
#include "optimizer/prep.h"
#include "parser/parsetree.h"
#include "parser/parse_coerce.h"
+#include "parser/parse_expr.h"
#include "parser/parse_func.h"
#include "utils/acl.h"
#include "utils/builtins.h"
#include "utils/syscache.h"
-#define IsFuncIndex(ATTR_LIST) (((IndexElem*)lfirst(ATTR_LIST))->funcname != NIL)
-
/* non-export function prototypes */
-static void CheckPredicate(List *predList, List *rangeTable, Oid baseRelOid);
-static void FuncIndexArgs(IndexInfo *indexInfo, Oid *classOidP,
- IndexElem *funcIndex,
- Oid relId,
- char *accessMethodName, Oid accessMethodId);
-static void NormIndexAttrs(IndexInfo *indexInfo, Oid *classOidP,
+static void CheckPredicate(List *predList);
+static void ComputeIndexAttrs(IndexInfo *indexInfo, Oid *classOidP,
List *attList,
Oid relId,
char *accessMethodName, Oid accessMethodId);
-static Oid GetAttrOpClass(IndexElem *attribute, Oid attrType,
+static Oid GetIndexOpClass(List *opclass, Oid attrType,
char *accessMethodName, Oid accessMethodId);
static Oid GetDefaultOpClass(Oid attrType, Oid accessMethodId);
* DefineIndex
* Creates a new index.
*
- * 'attributeList' is a list of IndexElem specifying either a functional
- * index or a list of attributes to index on.
+ * 'attributeList' is a list of IndexElem specifying columns and expressions
+ * to index on.
* 'predicate' is the qual specified in the where clause.
* 'rangetable' is needed to interpret the predicate.
*/
ReleaseSysCache(tuple);
+ /*
+ * If a range table was created then check that only the base rel is
+ * mentioned.
+ */
+ if (rangetable != NIL)
+ {
+ if (length(rangetable) != 1 || getrelid(1, rangetable) != relationId)
+ elog(ERROR, "index expressions and predicates may refer only to the base relation");
+ }
+
/*
* Convert the partial-index predicate from parsetree form to an
* implicit-AND qual expression, for easier evaluation at runtime.
if (predicate)
{
cnfPred = canonicalize_qual((Expr *) copyObject(predicate), true);
- CheckPredicate(cnfPred, rangetable, relationId);
+ CheckPredicate(cnfPred);
}
/*
* Check that all of the attributes in a primary key are marked
* as not null, otherwise attempt to ALTER TABLE .. SET NOT NULL
*/
- if (primary && !IsFuncIndex(attributeList))
+ if (primary)
{
List *keys;
IndexElem *key = (IndexElem *) lfirst(keys);
HeapTuple atttuple;
+ if (!key->name)
+ elog(ERROR, "primary keys cannot be expressions");
+
/* System attributes are never null, so no problem */
if (SystemAttributeByName(key->name, rel->rd_rel->relhasoids))
continue;
* structure
*/
indexInfo = makeNode(IndexInfo);
+ indexInfo->ii_NumIndexAttrs = numberOfAttributes;
+ indexInfo->ii_Expressions = NIL; /* for now */
+ indexInfo->ii_ExpressionsState = NIL;
indexInfo->ii_Predicate = cnfPred;
indexInfo->ii_PredicateState = NIL;
- indexInfo->ii_FuncOid = InvalidOid;
indexInfo->ii_Unique = unique;
- if (IsFuncIndex(attributeList))
- {
- IndexElem *funcIndex = (IndexElem *) lfirst(attributeList);
- int nargs;
-
- /* Parser should have given us only one list item, but check */
- if (numberOfAttributes != 1)
- elog(ERROR, "Functional index can only have one attribute");
-
- nargs = length(funcIndex->args);
- if (nargs > INDEX_MAX_KEYS)
- elog(ERROR, "Index function can take at most %d arguments",
- INDEX_MAX_KEYS);
-
- indexInfo->ii_NumIndexAttrs = 1;
- indexInfo->ii_NumKeyAttrs = nargs;
-
- classObjectId = (Oid *) palloc(sizeof(Oid));
-
- FuncIndexArgs(indexInfo, classObjectId, funcIndex,
+ classObjectId = (Oid *) palloc(numberOfAttributes * sizeof(Oid));
+ ComputeIndexAttrs(indexInfo, classObjectId, attributeList,
relationId, accessMethodName, accessMethodId);
- }
- else
- {
- indexInfo->ii_NumIndexAttrs = numberOfAttributes;
- indexInfo->ii_NumKeyAttrs = numberOfAttributes;
-
- classObjectId = (Oid *) palloc(numberOfAttributes * sizeof(Oid));
-
- NormIndexAttrs(indexInfo, classObjectId, attributeList,
- relationId, accessMethodName, accessMethodId);
- }
index_create(relationId, indexRelationName,
indexInfo, accessMethodId, classObjectId,
/*
* CheckPredicate
- * Checks that the given list of partial-index predicates refer
- * (via the given range table) only to the given base relation oid.
+ * Checks that the given list of partial-index predicates is valid.
*
* This used to also constrain the form of the predicate to forms that
* indxpath.c could do something with. However, that seems overly
* any evaluatable predicate will work. So accept any predicate here
* (except ones requiring a plan), and let indxpath.c fend for itself.
*/
-
static void
-CheckPredicate(List *predList, List *rangeTable, Oid baseRelOid)
+CheckPredicate(List *predList)
{
- if (length(rangeTable) != 1 || getrelid(1, rangeTable) != baseRelOid)
- elog(ERROR,
- "Partial-index predicates may refer only to the base relation");
-
/*
* We don't currently support generation of an actual query plan for a
* predicate, only simple scalar expressions; hence these
/*
* A predicate using mutable functions is probably wrong, for the same
- * reasons that we don't allow a functional index to use one.
+ * reasons that we don't allow an index expression to use one.
*/
if (contain_mutable_functions((Node *) predList))
elog(ERROR, "Functions in index predicate must be marked IMMUTABLE");
}
-
static void
-FuncIndexArgs(IndexInfo *indexInfo,
- Oid *classOidP,
- IndexElem *funcIndex,
- Oid relId,
- char *accessMethodName,
- Oid accessMethodId)
-{
- Oid argTypes[FUNC_MAX_ARGS];
- List *arglist;
- int nargs = 0;
- int i;
- FuncDetailCode fdresult;
- Oid funcid;
- Oid rettype;
- bool retset;
- Oid *true_typeids;
-
- /*
- * process the function arguments, which are a list of T_String
- * (someday ought to allow more general expressions?)
- *
- * Note caller already checked that list is not too long.
- */
- MemSet(argTypes, 0, sizeof(argTypes));
-
- foreach(arglist, funcIndex->args)
- {
- char *arg = strVal(lfirst(arglist));
- HeapTuple tuple;
- Form_pg_attribute att;
-
- tuple = SearchSysCacheAttName(relId, arg);
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "DefineIndex: column \"%s\" named in key does not exist", arg);
- att = (Form_pg_attribute) GETSTRUCT(tuple);
- indexInfo->ii_KeyAttrNumbers[nargs] = att->attnum;
- argTypes[nargs] = att->atttypid;
- ReleaseSysCache(tuple);
- nargs++;
- }
-
- /*
- * Lookup the function procedure to get its OID and result type.
- *
- * We rely on parse_func.c to find the correct function in the possible
- * presence of binary-compatible types. However, parse_func may do
- * too much: it will accept a function that requires run-time coercion
- * of input types, and the executor is not currently set up to support
- * that. So, check to make sure that the selected function has
- * exact-match or binary-compatible input types.
- */
- fdresult = func_get_detail(funcIndex->funcname, funcIndex->args,
- nargs, argTypes,
- &funcid, &rettype, &retset,
- &true_typeids);
- if (fdresult != FUNCDETAIL_NORMAL)
- {
- if (fdresult == FUNCDETAIL_AGGREGATE)
- elog(ERROR, "DefineIndex: functional index may not use an aggregate function");
- else if (fdresult == FUNCDETAIL_COERCION)
- elog(ERROR, "DefineIndex: functional index must use a real function, not a type coercion"
- "\n\tTry specifying the index opclass you want to use, instead");
- else
- func_error("DefineIndex", funcIndex->funcname, nargs, argTypes,
- NULL);
- }
-
- if (retset)
- elog(ERROR, "DefineIndex: cannot index on a function returning a set");
-
- for (i = 0; i < nargs; i++)
- {
- if (!IsBinaryCoercible(argTypes[i], true_typeids[i]))
- func_error("DefineIndex", funcIndex->funcname, nargs, true_typeids,
- "Index function must be binary-compatible with table datatype");
- }
-
- /*
- * Require that the function be marked immutable. Using a mutable
- * function for a functional index is highly questionable, since if
- * you aren't going to get the same result for the same data every
- * time, it's not clear what the index entries mean at all.
- */
- if (func_volatile(funcid) != PROVOLATILE_IMMUTABLE)
- elog(ERROR, "DefineIndex: index function must be marked IMMUTABLE");
-
- /* Process opclass, using func return type as default type */
-
- classOidP[0] = GetAttrOpClass(funcIndex, rettype,
- accessMethodName, accessMethodId);
-
- /* OK, return results */
-
- indexInfo->ii_FuncOid = funcid;
- /* Need to do the fmgr function lookup now, too */
- fmgr_info(funcid, &indexInfo->ii_FuncInfo);
-}
-
-static void
-NormIndexAttrs(IndexInfo *indexInfo,
- Oid *classOidP,
- List *attList, /* list of IndexElem's */
- Oid relId,
- char *accessMethodName,
- Oid accessMethodId)
+ComputeIndexAttrs(IndexInfo *indexInfo,
+ Oid *classOidP,
+ List *attList, /* list of IndexElem's */
+ Oid relId,
+ char *accessMethodName,
+ Oid accessMethodId)
{
List *rest;
int attn = 0;
foreach(rest, attList)
{
IndexElem *attribute = (IndexElem *) lfirst(rest);
- HeapTuple atttuple;
- Form_pg_attribute attform;
-
- if (attribute->name == NULL)
- elog(ERROR, "missing attribute for define index");
-
- atttuple = SearchSysCacheAttName(relId, attribute->name);
- if (!HeapTupleIsValid(atttuple))
- elog(ERROR, "DefineIndex: attribute \"%s\" not found",
- attribute->name);
- attform = (Form_pg_attribute) GETSTRUCT(atttuple);
+ Oid atttype;
- indexInfo->ii_KeyAttrNumbers[attn] = attform->attnum;
+ if (attribute->name != NULL)
+ {
+ /* Simple index attribute */
+ HeapTuple atttuple;
+ Form_pg_attribute attform;
+
+ Assert(attribute->expr == NULL);
+ atttuple = SearchSysCacheAttName(relId, attribute->name);
+ if (!HeapTupleIsValid(atttuple))
+ elog(ERROR, "DefineIndex: attribute \"%s\" not found",
+ attribute->name);
+ attform = (Form_pg_attribute) GETSTRUCT(atttuple);
+ indexInfo->ii_KeyAttrNumbers[attn] = attform->attnum;
+ atttype = attform->atttypid;
+ ReleaseSysCache(atttuple);
+ }
+ else if (attribute->expr && IsA(attribute->expr, Var))
+ {
+ /* Tricky tricky, he wrote (column) ... treat as simple attr */
+ Var *var = (Var *) attribute->expr;
- classOidP[attn] = GetAttrOpClass(attribute, attform->atttypid,
- accessMethodName, accessMethodId);
+ indexInfo->ii_KeyAttrNumbers[attn] = var->varattno;
+ atttype = get_atttype(relId, var->varattno);
+ }
+ else
+ {
+ /* Index expression */
+ Assert(attribute->expr != NULL);
+ indexInfo->ii_KeyAttrNumbers[attn] = 0; /* marks expression */
+ indexInfo->ii_Expressions = lappend(indexInfo->ii_Expressions,
+ attribute->expr);
+ atttype = exprType(attribute->expr);
+
+ /*
+ * We don't currently support generation of an actual query plan
+ * for an index expression, only simple scalar expressions;
+ * hence these restrictions.
+ */
+ if (contain_subplans(attribute->expr))
+ elog(ERROR, "Cannot use subselect in index expression");
+ if (contain_agg_clause(attribute->expr))
+ elog(ERROR, "Cannot use aggregate in index expression");
+
+ /*
+ * A expression using mutable functions is probably wrong,
+ * since if you aren't going to get the same result for the same
+ * data every time, it's not clear what the index entries mean at
+ * all.
+ */
+ if (contain_mutable_functions(attribute->expr))
+ elog(ERROR, "Functions in index expression must be marked IMMUTABLE");
+ }
- ReleaseSysCache(atttuple);
+ classOidP[attn] = GetIndexOpClass(attribute->opclass,
+ atttype,
+ accessMethodName,
+ accessMethodId);
attn++;
}
}
+/*
+ * Resolve possibly-defaulted operator class specification
+ */
static Oid
-GetAttrOpClass(IndexElem *attribute, Oid attrType,
- char *accessMethodName, Oid accessMethodId)
+GetIndexOpClass(List *opclass, Oid attrType,
+ char *accessMethodName, Oid accessMethodId)
{
char *schemaname;
char *opcname;
Oid opClassId,
opInputType;
- if (attribute->opclass == NIL)
+ /*
+ * Release 7.0 removed network_ops, timespan_ops, and
+ * datetime_ops, so we ignore those opclass names
+ * so the default *_ops is used. This can be
+ * removed in some later release. bjm 2000/02/07
+ *
+ * Release 7.1 removes lztext_ops, so suppress that too
+ * for a while. tgl 2000/07/30
+ *
+ * Release 7.2 renames timestamp_ops to timestamptz_ops,
+ * so suppress that too for awhile. I'm starting to
+ * think we need a better approach. tgl 2000/10/01
+ */
+ if (length(opclass) == 1)
+ {
+ char *claname = strVal(lfirst(opclass));
+
+ if (strcmp(claname, "network_ops") == 0 ||
+ strcmp(claname, "timespan_ops") == 0 ||
+ strcmp(claname, "datetime_ops") == 0 ||
+ strcmp(claname, "lztext_ops") == 0 ||
+ strcmp(claname, "timestamp_ops") == 0)
+ opclass = NIL;
+ }
+
+ if (opclass == NIL)
{
/* no operator class specified, so find the default */
opClassId = GetDefaultOpClass(attrType, accessMethodId);
*/
/* deconstruct the name list */
- DeconstructQualifiedName(attribute->opclass, &schemaname, &opcname);
+ DeconstructQualifiedName(opclass, &schemaname, &opcname);
if (schemaname)
{
if (!HeapTupleIsValid(tuple))
elog(ERROR, "DefineIndex: operator class \"%s\" not supported by access method \"%s\"",
- NameListToString(attribute->opclass), accessMethodName);
+ NameListToString(opclass), accessMethodName);
/*
* Verify that the index operator class accepts this datatype. Note
if (!IsBinaryCoercible(attrType, opInputType))
elog(ERROR, "operator class \"%s\" does not accept data type %s",
- NameListToString(attribute->opclass), format_type_be(attrType));
+ NameListToString(opclass), format_type_be(attrType));
ReleaseSysCache(tuple);
for (i = 0; i < relcnt; i++)
{
StartTransactionCommand();
- SetQuerySnapshot(); /* might be needed for functional index */
+ SetQuerySnapshot(); /* might be needed for functions in indexes */
if (reindex_relation(relids[i], force))
elog(NOTICE, "relation %u was reindexed", relids[i]);
CommitTransactionCommand();
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/commands/tablecmds.c,v 1.72 2003/04/29 22:13:08 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/commands/tablecmds.c,v 1.73 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
Relation attrelation;
HeapTuple atttup;
Form_pg_attribute attform;
+ int attnum;
List *indexoidlist;
List *indexoidscan;
oldattname);
attform = (Form_pg_attribute) GETSTRUCT(atttup);
- if (attform->attnum < 0)
+ attnum = attform->attnum;
+ if (attnum < 0)
elog(ERROR, "renameatt: system attribute \"%s\" may not be renamed",
oldattname);
{
Oid indexoid = lfirsto(indexoidscan);
HeapTuple indextup;
+ Form_pg_index indexform;
+ int i;
/*
- * First check to see if index is a functional index. If so, its
- * column name is a function name and shouldn't be renamed here.
+ * Scan through index columns to see if there's any simple index
+ * entries for this attribute. We ignore expressional entries.
*/
indextup = SearchSysCache(INDEXRELID,
ObjectIdGetDatum(indexoid),
0, 0, 0);
if (!HeapTupleIsValid(indextup))
elog(ERROR, "renameatt: can't find index id %u", indexoid);
- if (OidIsValid(((Form_pg_index) GETSTRUCT(indextup))->indproc))
- {
- ReleaseSysCache(indextup);
- continue;
- }
- ReleaseSysCache(indextup);
+ indexform = (Form_pg_index) GETSTRUCT(indextup);
- /*
- * Okay, look to see if any column name of the index matches the
- * old attribute name.
- */
- atttup = SearchSysCacheCopyAttName(indexoid, oldattname);
- if (!HeapTupleIsValid(atttup))
- continue; /* Nope, so ignore it */
+ for (i = 0; i < indexform->indnatts; i++)
+ {
+ if (attnum != indexform->indkey[i])
+ continue;
+ /*
+ * Found one, rename it.
+ */
+ atttup = SearchSysCacheCopy(ATTNUM,
+ ObjectIdGetDatum(indexoid),
+ Int16GetDatum(i + 1),
+ 0, 0);
+ if (!HeapTupleIsValid(atttup))
+ continue; /* should we raise an error? */
+ /*
+ * Update the (copied) attribute tuple.
+ */
+ namestrcpy(&(((Form_pg_attribute) GETSTRUCT(atttup))->attname),
+ newattname);
- /*
- * Update the (copied) attribute tuple.
- */
- namestrcpy(&(((Form_pg_attribute) GETSTRUCT(atttup))->attname),
- newattname);
+ simple_heap_update(attrelation, &atttup->t_self, atttup);
- simple_heap_update(attrelation, &atttup->t_self, atttup);
+ /* keep system catalog indexes current */
+ CatalogUpdateIndexes(attrelation, atttup);
- /* keep system catalog indexes current */
- CatalogUpdateIndexes(attrelation, atttup);
+ heap_freetuple(atttup);
+ }
- heap_freetuple(atttup);
+ ReleaseSysCache(indextup);
}
freeList(indexoidlist);
* Loop over each attribute in the primary key and see if it
* matches the to-be-altered attribute
*/
- for (i = 0; i < INDEX_MAX_KEYS &&
- indexStruct->indkey[i] != InvalidAttrNumber; i++)
+ for (i = 0; i < indexStruct->indnatts; i++)
{
if (indexStruct->indkey[i] == attnum)
elog(ERROR, "ALTER TABLE: Attribute \"%s\" is in a primary key", colName);
/*
* Now build the list of PK attributes from the indkey definition
+ * (we assume a primary key cannot have expressional elements)
*/
*attnamelist = NIL;
- for (i = 0; i < INDEX_MAX_KEYS && indexStruct->indkey[i] != 0; i++)
+ for (i = 0; i < indexStruct->indnatts; i++)
{
int pkattno = indexStruct->indkey[i];
indexStruct = (Form_pg_index) GETSTRUCT(indexTuple);
/*
- * Must be unique, not a functional index, and not a partial index
+ * Must have the right number of columns; must be unique and not a
+ * partial index; forget it if there are any expressions, too
*/
- if (indexStruct->indisunique &&
- indexStruct->indproc == InvalidOid &&
- VARSIZE(&indexStruct->indpred) <= VARHDRSZ)
+ if (indexStruct->indnatts == numattrs &&
+ indexStruct->indisunique &&
+ heap_attisnull(indexTuple, Anum_pg_index_indpred) &&
+ heap_attisnull(indexTuple, Anum_pg_index_indexprs))
{
- for (i = 0; i < INDEX_MAX_KEYS && indexStruct->indkey[i] != 0; i++)
- ;
- if (i == numattrs)
+ /*
+ * The given attnum list may match the index columns in any
+ * order. Check that each list is a subset of the other.
+ */
+ for (i = 0; i < numattrs; i++)
+ {
+ found = false;
+ for (j = 0; j < numattrs; j++)
+ {
+ if (attnums[i] == indexStruct->indkey[j])
+ {
+ found = true;
+ break;
+ }
+ }
+ if (!found)
+ break;
+ }
+ if (found)
{
- /*
- * The given attnum list may match the index columns in any
- * order. Check that each list is a subset of the other.
- */
for (i = 0; i < numattrs; i++)
{
found = false;
for (j = 0; j < numattrs; j++)
{
- if (attnums[i] == indexStruct->indkey[j])
+ if (attnums[j] == indexStruct->indkey[i])
{
found = true;
break;
if (!found)
break;
}
- if (found)
- {
- for (i = 0; i < numattrs; i++)
- {
- found = false;
- for (j = 0; j < numattrs; j++)
- {
- if (attnums[j] == indexStruct->indkey[i])
- {
- found = true;
- break;
- }
- }
- if (!found)
- break;
- }
- }
}
}
ReleaseSysCache(indexTuple);
indexInfo = makeNode(IndexInfo);
indexInfo->ii_NumIndexAttrs = 2;
- indexInfo->ii_NumKeyAttrs = 2;
indexInfo->ii_KeyAttrNumbers[0] = 1;
indexInfo->ii_KeyAttrNumbers[1] = 2;
+ indexInfo->ii_Expressions = NIL;
+ indexInfo->ii_ExpressionsState = NIL;
indexInfo->ii_Predicate = NIL;
indexInfo->ii_PredicateState = NIL;
- indexInfo->ii_FuncOid = InvalidOid;
indexInfo->ii_Unique = true;
classObjectId[0] = OID_BTREE_OPS_OID;
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/commands/vacuum.c,v 1.254 2003/05/27 17:49:45 momjian Exp $
+ * $Header: /cvsroot/pgsql/src/backend/commands/vacuum.c,v 1.255 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
if (vacstmt->vacuum)
{
StartTransactionCommand();
- SetQuerySnapshot(); /* might be needed for functional index */
+ SetQuerySnapshot(); /* might be needed for functions in indexes */
}
else
old_context = MemoryContextSwitchTo(anl_context);
/* Begin a transaction for vacuuming this relation */
StartTransactionCommand();
- SetQuerySnapshot(); /* might be needed for functional index */
+ SetQuerySnapshot(); /* might be needed for functions in indexes */
/*
* Check for user-requested abort. Note we want this to be inside a
return true;
/* Otherwise, look to see if there's a partial-index predicate */
- return (VARSIZE(&indrel->rd_index->indpred) > VARHDRSZ);
+ if (!heap_attisnull(indrel->rd_indextuple, Anum_pg_index_indpred))
+ return true;
+
+ return false;
}
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/executor/execUtils.c,v 1.99 2003/05/05 17:57:47 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/executor/execUtils.c,v 1.100 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
if (!indexDesc->rd_am->amconcurrent)
LockRelation(indexDesc, AccessExclusiveLock);
- /*
- * extract index key information from the index's pg_index tuple
- */
- ii = BuildIndexInfo(indexDesc->rd_index);
+ /* extract index key information from the index's pg_index info */
+ ii = BuildIndexInfo(indexDesc);
relationDescs[i] = indexDesc;
indexInfoArray[i] = ii;
/*
* We will use the EState's per-tuple context for evaluating
- * predicates and functional-index functions (creating it if it's not
+ * predicates and index expressions (creating it if it's not
* already there).
*/
econtext = GetPerTupleExprContext(estate);
/*
* FormIndexDatum fills in its datum and null parameters with
* attribute information taken from the given heap tuple.
+ * It also computes any expressions needed.
*/
FormIndexDatum(indexInfo,
heapTuple,
heapDescriptor,
- econtext->ecxt_per_tuple_memory,
+ estate,
datum,
nullv);
* Portions Copyright (c) 1994, Regents of the University of California
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/nodes/copyfuncs.c,v 1.250 2003/05/06 00:20:32 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/nodes/copyfuncs.c,v 1.251 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
IndexElem *newnode = makeNode(IndexElem);
COPY_STRING_FIELD(name);
- COPY_NODE_FIELD(funcname);
- COPY_NODE_FIELD(args);
+ COPY_NODE_FIELD(expr);
COPY_NODE_FIELD(opclass);
return newnode;
* Portions Copyright (c) 1994, Regents of the University of California
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/nodes/equalfuncs.c,v 1.193 2003/05/06 00:20:32 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/nodes/equalfuncs.c,v 1.194 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
_equalIndexElem(IndexElem *a, IndexElem *b)
{
COMPARE_STRING_FIELD(name);
- COMPARE_NODE_FIELD(funcname);
- COMPARE_NODE_FIELD(args);
+ COMPARE_NODE_FIELD(expr);
COMPARE_NODE_FIELD(opclass);
return true;
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/nodes/outfuncs.c,v 1.205 2003/05/06 00:20:32 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/nodes/outfuncs.c,v 1.206 2003/05/28 16:03:56 tgl Exp $
*
* NOTES
* Every node type that can appear in stored rules' parsetrees *must*
WRITE_NODE_TYPE("INDEXELEM");
WRITE_STRING_FIELD(name);
- WRITE_NODE_FIELD(funcname);
- WRITE_NODE_FIELD(args);
+ WRITE_NODE_FIELD(expr);
WRITE_NODE_FIELD(opclass);
}
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/optimizer/path/indxpath.c,v 1.141 2003/05/27 17:49:46 momjian Exp $
+ * $Header: /cvsroot/pgsql/src/backend/optimizer/path/indxpath.c,v 1.142 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
/*
* DoneMatchingIndexKeys() - MACRO
- *
- * Formerly this looked at indexkeys, but that's the wrong thing for a
- * functional index.
*/
-#define DoneMatchingIndexKeys(indexkeys, classes) \
- (classes[0] == InvalidOid)
+#define DoneMatchingIndexKeys(classes) (classes[0] == InvalidOid)
#define is_indexable_operator(clause,opclass,indexkey_on_left) \
(indexable_operator(clause,opclass,indexkey_on_left) != InvalidOid)
IndexOptInfo *index,
Relids outer_relids,
bool isouterjoin);
-static bool match_clause_to_indexkey(RelOptInfo *rel, IndexOptInfo *index,
- int indexkey, Oid opclass, Expr *clause);
-static bool match_join_clause_to_indexkey(RelOptInfo *rel, IndexOptInfo *index,
- int indexkey, Oid opclass, Expr *clause);
+static bool match_clause_to_indexcol(RelOptInfo *rel, IndexOptInfo *index,
+ int indexcol, Oid opclass, Expr *clause);
+static bool match_join_clause_to_indexcol(RelOptInfo *rel, IndexOptInfo *index,
+ int indexcol, Oid opclass, Expr *clause);
static Oid indexable_operator(Expr *clause, Oid opclass,
bool indexkey_on_left);
static bool pred_test(List *predicate_list, List *restrictinfo_list,
- List *joininfo_list, int relvarno);
+ List *joininfo_list);
static bool pred_test_restrict_list(Expr *predicate, List *restrictinfo_list);
static bool pred_test_recurse_clause(Expr *predicate, Node *clause);
static bool pred_test_recurse_pred(Expr *predicate, Node *clause);
static Path *make_innerjoin_index_path(Query *root,
RelOptInfo *rel, IndexOptInfo *index,
List *clausegroups);
-static bool match_index_to_operand(int indexkey, Node *operand,
+static bool match_index_to_operand(Node *operand, int indexcol,
RelOptInfo *rel, IndexOptInfo *index);
-static bool function_index_operand(Expr *funcOpnd, RelOptInfo *rel,
- IndexOptInfo *index);
static bool match_special_index_operator(Expr *clause, Oid opclass,
bool indexkey_on_left);
static List *expand_indexqual_condition(Expr *clause, Oid opclass);
* predicate test.
*/
if (index->indpred != NIL)
- if (!pred_test(index->indpred, restrictinfo_list, joininfo_list,
- rel->relid))
+ if (!pred_test(index->indpred, restrictinfo_list, joininfo_list))
continue;
/*
IndexOptInfo *index,
Expr *clause)
{
- int indexkey = index->indexkeys[0];
Oid opclass = index->classlist[0];
if (and_clause((Node *) clause))
foreach(item, ((BoolExpr *) clause)->args)
{
- if (match_clause_to_indexkey(rel, index, indexkey, opclass,
+ if (match_clause_to_indexcol(rel, index, 0, opclass,
lfirst(item)))
return true;
}
return false;
}
else
- return match_clause_to_indexkey(rel, index, indexkey, opclass,
+ return match_clause_to_indexcol(rel, index, 0, opclass,
clause);
}
Expr *orsubclause)
{
List *quals = NIL;
- int *indexkeys = index->indexkeys;
+ int indexcol = 0;
Oid *classes = index->classlist;
/*
*/
do
{
- int curIndxKey = indexkeys[0];
Oid curClass = classes[0];
List *clausegroup = NIL;
List *item;
{
Expr *subsubclause = (Expr *) lfirst(item);
- if (match_clause_to_indexkey(rel, index,
- curIndxKey, curClass,
+ if (match_clause_to_indexcol(rel, index,
+ indexcol, curClass,
subsubclause))
clausegroup = nconc(clausegroup,
expand_indexqual_condition(subsubclause,
curClass));
}
}
- else if (match_clause_to_indexkey(rel, index,
- curIndxKey, curClass,
+ else if (match_clause_to_indexcol(rel, index,
+ indexcol, curClass,
orsubclause))
clausegroup = expand_indexqual_condition(orsubclause, curClass);
{
RestrictInfo *rinfo = (RestrictInfo *) lfirst(item);
- if (match_clause_to_indexkey(rel, index,
- curIndxKey, curClass,
+ if (match_clause_to_indexcol(rel, index,
+ indexcol, curClass,
rinfo->clause))
clausegroup = nconc(clausegroup,
expand_indexqual_condition(rinfo->clause,
quals = nconc(quals, clausegroup);
- indexkeys++;
+ indexcol++;
classes++;
- } while (!DoneMatchingIndexKeys(indexkeys, classes));
+ } while (!DoneMatchingIndexKeys(classes));
if (quals == NIL)
elog(ERROR, "extract_or_indexqual_conditions: no matching clause");
{
List *clausegroup_list = NIL;
List *restrictinfo_list = rel->baserestrictinfo;
- int *indexkeys = index->indexkeys;
+ int indexcol = 0;
Oid *classes = index->classlist;
if (restrictinfo_list == NIL)
do
{
- int curIndxKey = indexkeys[0];
Oid curClass = classes[0];
List *clausegroup = NIL;
List *i;
{
RestrictInfo *rinfo = (RestrictInfo *) lfirst(i);
- if (match_clause_to_indexkey(rel,
+ if (match_clause_to_indexcol(rel,
index,
- curIndxKey,
+ indexcol,
curClass,
rinfo->clause))
clausegroup = lappend(clausegroup, rinfo);
clausegroup_list = lappend(clausegroup_list, clausegroup);
- indexkeys++;
+ indexcol++;
classes++;
- } while (!DoneMatchingIndexKeys(indexkeys, classes));
+ } while (!DoneMatchingIndexKeys(classes));
return clausegroup_list;
}
{
List *clausegroup_list = NIL;
bool jfound = false;
- int *indexkeys = index->indexkeys;
+ int indexcol = 0;
Oid *classes = index->classlist;
do
{
- int curIndxKey = indexkeys[0];
Oid curClass = classes[0];
List *clausegroup = NIL;
List *i;
if (isouterjoin && rinfo->ispusheddown)
continue;
- if (match_join_clause_to_indexkey(rel,
+ if (match_join_clause_to_indexcol(rel,
index,
- curIndxKey,
+ indexcol,
curClass,
rinfo->clause))
{
if (isouterjoin && rinfo->ispusheddown)
continue;
- if (match_clause_to_indexkey(rel,
+ if (match_clause_to_indexcol(rel,
index,
- curIndxKey,
+ indexcol,
curClass,
rinfo->clause))
clausegroup = lappend(clausegroup, rinfo);
clausegroup_list = lappend(clausegroup_list, clausegroup);
- indexkeys++;
+ indexcol++;
classes++;
- } while (!DoneMatchingIndexKeys(indexkeys, classes));
+ } while (!DoneMatchingIndexKeys(classes));
/* if no join clause was matched then forget it, per comments above */
if (!jfound)
/*
- * match_clause_to_indexkey()
- * Determines whether a restriction clause matches a key of an index.
+ * match_clause_to_indexcol()
+ * Determines whether a restriction clause matches a column of an index.
*
* To match, the clause:
*
* (1) must be in the form (indexkey op const) or (const op indexkey);
* and
* (2) must contain an operator which is in the same class as the index
- * operator for this key, or is a "special" operator as recognized
+ * operator for this column, or is a "special" operator as recognized
* by match_special_index_operator().
*
* Presently, the executor can only deal with indexquals that have the
*
* 'rel' is the relation of interest.
* 'index' is an index on 'rel'.
- * 'indexkey' is a key of 'index'.
+ * 'indexcol' is a column number of 'index' (counting from 0).
* 'opclass' is the corresponding operator class.
* 'clause' is the clause to be tested.
*
* responsibility of higher-level routines to cope with those.
*/
static bool
-match_clause_to_indexkey(RelOptInfo *rel,
+match_clause_to_indexcol(RelOptInfo *rel,
IndexOptInfo *index,
- int indexkey,
+ int indexcol,
Oid opclass,
Expr *clause)
{
* (indexkey operator constant) or (constant operator indexkey).
* Anything that is a "pseudo constant" expression will do.
*/
- if (match_index_to_operand(indexkey, leftop, rel, index) &&
+ if (match_index_to_operand(leftop, indexcol, rel, index) &&
is_pseudo_constant_clause(rightop))
{
if (is_indexable_operator(clause, opclass, true))
return false;
}
- if (match_index_to_operand(indexkey, rightop, rel, index) &&
+ if (match_index_to_operand(rightop, indexcol, rel, index) &&
is_pseudo_constant_clause(leftop))
{
if (is_indexable_operator(clause, opclass, false))
}
/*
- * match_join_clause_to_indexkey()
- * Determines whether a join clause matches a key of an index.
+ * match_join_clause_to_indexcol()
+ * Determines whether a join clause matches a column of an index.
*
* To match, the clause:
*
* where others is an expression involving only vars of the other
* relation(s); and
* (2) must contain an operator which is in the same class as the index
- * operator for this key, or is a "special" operator as recognized
+ * operator for this column, or is a "special" operator as recognized
* by match_special_index_operator().
*
* As above, we must be able to commute the clause to put the indexkey
*
* 'rel' is the relation of interest.
* 'index' is an index on 'rel'.
- * 'indexkey' is a key of 'index'.
+ * 'indexcol' is a column number of 'index' (counting from 0).
* 'opclass' is the corresponding operator class.
* 'clause' is the clause to be tested.
*
* responsibility of higher-level routines to cope with those.
*/
static bool
-match_join_clause_to_indexkey(RelOptInfo *rel,
+match_join_clause_to_indexcol(RelOptInfo *rel,
IndexOptInfo *index,
- int indexkey,
+ int indexcol,
Oid opclass,
Expr *clause)
{
* expression that uses none of the indexed relation's vars and
* contains no volatile functions.
*/
- if (match_index_to_operand(indexkey, leftop, rel, index))
+ if (match_index_to_operand(leftop, indexcol, rel, index))
{
Relids othervarnos = pull_varnos(rightop);
bool isIndexable;
return isIndexable;
}
- if (match_index_to_operand(indexkey, rightop, rel, index))
+ if (match_index_to_operand(rightop, indexcol, rel, index))
{
Relids othervarnos = pull_varnos(leftop);
bool isIndexable;
* to CNF format). --Nels, Jan '93
*/
static bool
-pred_test(List *predicate_list, List *restrictinfo_list, List *joininfo_list,
- int relvarno)
+pred_test(List *predicate_list, List *restrictinfo_list, List *joininfo_list)
{
List *pred;
return false; /* no restriction clauses: the test must
* fail */
- /*
- * The predicate as stored in the index definition will use varno 1
- * for its Vars referencing the indexed relation. If the indexed
- * relation isn't varno 1 in the query, we must adjust the predicate
- * to make the Vars match, else equal() won't work.
- */
- if (relvarno != 1)
- {
- predicate_list = copyObject(predicate_list);
- ChangeVarNodes((Node *) predicate_list, 1, relvarno, 0);
- }
-
foreach(pred, predicate_list)
{
/*
{
RestrictInfo *rinfo = (RestrictInfo *) lfirst(j);
Expr *clause = rinfo->clause;
- int *indexkeys = index->indexkeys;
+ int indexcol = 0;
Oid *classes = index->classlist;
do
{
- int curIndxKey = indexkeys[0];
Oid curClass = classes[0];
- if (match_join_clause_to_indexkey(rel,
+ if (match_join_clause_to_indexcol(rel,
index,
- curIndxKey,
+ indexcol,
curClass,
clause))
{
break;
}
- indexkeys++;
+ indexcol++;
classes++;
- } while (!DoneMatchingIndexKeys(indexkeys, classes));
+ } while (!DoneMatchingIndexKeys(classes));
if (match_found)
break;
* match_index_to_operand()
* Generalized test for a match between an index's key
* and the operand on one side of a restriction or join clause.
- * Now check for functional indices as well.
+ *
+ * operand: the nodetree to be compared to the index
+ * indexcol: the column number of the index (counting from 0)
+ * rel: the parent relation
+ * index: the index of interest
*/
static bool
-match_index_to_operand(int indexkey,
- Node *operand,
+match_index_to_operand(Node *operand,
+ int indexcol,
RelOptInfo *rel,
IndexOptInfo *index)
{
+ int indkey;
+
/*
- * Ignore any RelabelType node above the indexkey. This is needed to
+ * Ignore any RelabelType node above the operand. This is needed to
* be able to apply indexscanning in binary-compatible-operator cases.
* Note: we can assume there is at most one RelabelType node;
* eval_const_expressions() will have simplified if more than one.
if (operand && IsA(operand, RelabelType))
operand = (Node *) ((RelabelType *) operand)->arg;
- if (index->indproc == InvalidOid)
+ indkey = index->indexkeys[indexcol];
+ if (indkey != 0)
{
/*
- * Simple index.
+ * Simple index column; operand must be a matching Var.
*/
if (operand && IsA(operand, Var) &&
rel->relid == ((Var *) operand)->varno &&
- indexkey == ((Var *) operand)->varattno)
+ indkey == ((Var *) operand)->varattno)
return true;
- else
- return false;
}
-
- /*
- * Functional index.
- */
- return function_index_operand((Expr *) operand, rel, index);
-}
-
-static bool
-function_index_operand(Expr *funcOpnd, RelOptInfo *rel, IndexOptInfo *index)
-{
- FuncExpr *function;
- List *funcargs;
- int *indexKeys = index->indexkeys;
- List *arg;
- int i;
-
- /*
- * sanity check, make sure we know what we're dealing with here.
- */
- if (funcOpnd == NULL || !IsA(funcOpnd, FuncExpr) ||
- indexKeys == NULL)
- return false;
-
- function = (FuncExpr *) funcOpnd;
- funcargs = function->args;
-
- if (function->funcid != index->indproc)
- return false;
-
- /*----------
- * Check that the arguments correspond to the same arguments used to
- * create the functional index. To do this we must check that
- * 1. they refer to the right relation.
- * 2. the args have the right attr. numbers in the right order.
- * We must ignore RelabelType nodes above the argument Vars in order
- * to recognize binary-compatible-function cases correctly.
- *----------
- */
- i = 0;
- foreach(arg, funcargs)
+ else
{
- Var *var = (Var *) lfirst(arg);
+ /*
+ * Index expression; find the correct expression. (This search could
+ * be avoided, at the cost of complicating all the callers of this
+ * routine; doesn't seem worth it.)
+ */
+ List *indexprs;
+ int i;
+ Node *indexkey;
- if (var && IsA(var, RelabelType))
- var = (Var *) ((RelabelType *) var)->arg;
- if (var == NULL || !IsA(var, Var))
- return false;
- if (indexKeys[i] == 0)
- return false;
- if (var->varno != rel->relid || var->varattno != indexKeys[i])
- return false;
+ indexprs = index->indexprs;
+ for (i = 0; i < indexcol; i++)
+ {
+ if (index->indexkeys[i] == 0)
+ {
+ if (indexprs == NIL)
+ elog(ERROR, "wrong number of index expressions");
+ indexprs = lnext(indexprs);
+ }
+ }
+ if (indexprs == NIL)
+ elog(ERROR, "wrong number of index expressions");
+ indexkey = (Node *) lfirst(indexprs);
+ /*
+ * Does it match the operand? Again, strip any relabeling.
+ */
+ if (indexkey && IsA(indexkey, RelabelType))
+ indexkey = (Node *) ((RelabelType *) indexkey)->arg;
- i++;
+ if (equal(indexkey, operand))
+ return true;
}
- if (indexKeys[i] != 0)
- return false; /* not enough arguments */
-
- return true;
+ return false;
}
/****************************************************************************
*
* Two routines are provided here, match_special_index_operator() and
* expand_indexqual_conditions(). match_special_index_operator() is
- * just an auxiliary function for match_clause_to_indexkey(); after
+ * just an auxiliary function for match_clause_to_indexcol(); after
* the latter fails to recognize a restriction opclause's operator
* as a member of an index's opclass, it asks match_special_index_operator()
* whether the clause should be considered an indexqual anyway.
expand_indexqual_conditions(IndexOptInfo *index, List *clausegroups)
{
List *resultquals = NIL;
- int *indexkeys = index->indexkeys;
Oid *classes = index->classlist;
if (clausegroups == NIL)
clausegroups = lnext(clausegroups);
- indexkeys++;
classes++;
- } while (clausegroups != NIL &&
- !DoneMatchingIndexKeys(indexkeys, classes));
+ } while (clausegroups != NIL && !DoneMatchingIndexKeys(classes));
Assert(clausegroups == NIL); /* else more groups than indexkeys... */
* Portions Copyright (c) 1994, Regents of the University of California
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/optimizer/path/pathkeys.c,v 1.48 2003/05/02 19:48:53 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/optimizer/path/pathkeys.c,v 1.49 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
List *retval = NIL;
int *indexkeys = index->indexkeys;
Oid *ordering = index->ordering;
- PathKeyItem *item;
- Oid sortop;
-
- if (!indexkeys || indexkeys[0] == 0 ||
- !ordering || ordering[0] == InvalidOid)
- return NIL; /* unordered index? */
+ List *indexprs = index->indexprs;
- if (index->indproc)
+ while (*ordering != InvalidOid)
{
- /* Functional index: build a representation of the function call */
- Expr *funcnode;
- List *funcargs = NIL;
+ PathKeyItem *item;
+ Oid sortop;
+ Node *indexkey;
+ List *cpathkey;
sortop = *ordering;
if (ScanDirectionIsBackward(scandir))
{
sortop = get_commutator(sortop);
if (sortop == InvalidOid)
- return NIL; /* oops, no reverse sort operator? */
+ break; /* oops, no reverse sort operator? */
}
- while (*indexkeys != 0)
+ if (*indexkeys != 0)
{
- funcargs = lappend(funcargs,
- find_indexkey_var(root, rel, *indexkeys));
- indexkeys++;
+ /* simple index column */
+ indexkey = (Node *) find_indexkey_var(root, rel, *indexkeys);
}
-
- funcnode = make_funcclause(index->indproc,
- get_func_rettype(index->indproc),
- false, /* cannot be a set */
- COERCE_DONTCARE, /* to match any user expr */
- funcargs);
-
- /* Make a one-sublist pathkeys list for the function expression */
- item = makePathKeyItem((Node *) funcnode, sortop);
- retval = makeList1(make_canonical_pathkey(root, item));
- }
- else
- {
- /* Normal non-functional index */
- while (*indexkeys != 0 && *ordering != InvalidOid)
+ else
{
- Var *relvar = find_indexkey_var(root, rel, *indexkeys);
- List *cpathkey;
+ /* expression --- assume we need not copy it */
+ if (indexprs == NIL)
+ elog(ERROR, "wrong number of index expressions");
+ indexkey = (Node *) lfirst(indexprs);
+ indexprs = lnext(indexprs);
+ }
- sortop = *ordering;
- if (ScanDirectionIsBackward(scandir))
- {
- sortop = get_commutator(sortop);
- if (sortop == InvalidOid)
- break; /* oops, no reverse sort operator? */
- }
+ /* OK, make a sublist for this sort key */
+ item = makePathKeyItem(indexkey, sortop);
+ cpathkey = make_canonical_pathkey(root, item);
- /* OK, make a sublist for this sort key */
- item = makePathKeyItem((Node *) relvar, sortop);
- cpathkey = make_canonical_pathkey(root, item);
+ /*
+ * Eliminate redundant ordering info; could happen if query is
+ * such that index keys are equijoined...
+ */
+ if (!ptrMember(cpathkey, retval))
+ retval = lappend(retval, cpathkey);
- /*
- * Eliminate redundant ordering info; could happen if query is
- * such that index keys are equijoined...
- */
- if (!ptrMember(cpathkey, retval))
- retval = lappend(retval, cpathkey);
- indexkeys++;
- ordering++;
- }
+ indexkeys++;
+ ordering++;
}
return retval;
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/optimizer/plan/createplan.c,v 1.142 2003/05/12 00:17:03 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/optimizer/plan/createplan.c,v 1.143 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
fix_indxqual_operand(Node *node, int baserelid, IndexOptInfo *index,
Oid *opclass)
{
- /*
- * Remove any binary-compatible relabeling of the indexkey
- */
- if (IsA(node, RelabelType))
- node = (Node *) ((RelabelType *) node)->arg;
-
/*
* We represent index keys by Var nodes having the varno of the base
* table but varattno equal to the index's attribute number (index
* a special-purpose node type that could not be mistaken for a
* regular Var. But it will do for now.
*/
- if (IsA(node, Var))
+ Var *result;
+ int pos;
+ List *indexprs;
+
+ /*
+ * Remove any binary-compatible relabeling of the indexkey
+ */
+ if (IsA(node, RelabelType))
+ node = (Node *) ((RelabelType *) node)->arg;
+
+ if (IsA(node, Var) &&
+ ((Var *) node)->varno == baserelid)
{
- /* If it's a var, find which index key position it occupies */
- Assert(index->indproc == InvalidOid);
+ /* Try to match against simple index columns */
+ int varatt = ((Var *) node)->varattno;
- if (((Var *) node)->varno == baserelid)
+ if (varatt != 0)
{
- int varatt = ((Var *) node)->varattno;
- int pos;
-
- for (pos = 0; pos < index->nkeys; pos++)
+ for (pos = 0; pos < index->ncolumns; pos++)
{
if (index->indexkeys[pos] == varatt)
{
- Node *newnode = copyObject(node);
-
- ((Var *) newnode)->varattno = pos + 1;
+ result = (Var *) copyObject(node);
+ result->varattno = pos + 1;
/* return the correct opclass, too */
*opclass = index->classlist[pos];
- return newnode;
+ return (Node *) result;
}
}
}
-
- /*
- * Oops, this Var isn't an indexkey!
- */
- elog(ERROR, "fix_indxqual_operand: var is not index attribute");
}
- /*
- * Else, it must be a func expression matching a functional index.
- * Since we currently only support single-column functional indexes,
- * the returned varattno must be 1.
- */
- Assert(index->indproc != InvalidOid);
- Assert(is_funcclause(node)); /* not a very thorough check, but easy */
-
- /* classlist[0] is the only class of a functional index */
- *opclass = index->classlist[0];
+ /* Try to match against index expressions */
+ indexprs = index->indexprs;
+ for (pos = 0; pos < index->ncolumns; pos++)
+ {
+ if (index->indexkeys[pos] == 0)
+ {
+ Node *indexkey;
+
+ if (indexprs == NIL)
+ elog(ERROR, "too few entries in indexprs list");
+ indexkey = (Node *) lfirst(indexprs);
+ if (indexkey && IsA(indexkey, RelabelType))
+ indexkey = (Node *) ((RelabelType *) indexkey)->arg;
+ if (equal(node, indexkey))
+ {
+ /* Found a match */
+ result = makeVar(baserelid, pos + 1,
+ exprType(lfirst(indexprs)), -1,
+ 0);
+ /* return the correct opclass, too */
+ *opclass = index->classlist[pos];
+ return (Node *) result;
+ }
+ indexprs = lnext(indexprs);
+ }
+ }
- return (Node *) makeVar(baserelid, 1, exprType(node), -1, 0);
+ /* Ooops... */
+ elog(ERROR, "fix_indxqual_operand: node is not index attribute");
+ return NULL; /* keep compiler quiet */
}
/*
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/optimizer/util/clauses.c,v 1.136 2003/04/29 22:13:09 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/optimizer/util/clauses.c,v 1.137 2003/05/28 16:03:56 tgl Exp $
*
* HISTORY
* AUTHOR DATE MAJOR EVENT
return NULL;
}
-/*****************************************************************************
- * FUNCTION clause functions
- *****************************************************************************/
-
-/*
- * make_funcclause
- * Creates a function clause given its function info and argument list.
- */
-Expr *
-make_funcclause(Oid funcid, Oid funcresulttype, bool funcretset,
- CoercionForm funcformat, List *funcargs)
-{
- FuncExpr *expr = makeNode(FuncExpr);
-
- expr->funcid = funcid;
- expr->funcresulttype = funcresulttype;
- expr->funcretset = funcretset;
- expr->funcformat = funcformat;
- expr->args = funcargs;
- return (Expr *) expr;
-}
-
/*****************************************************************************
* NOT clause functions
*****************************************************************************/
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/optimizer/util/plancat.c,v 1.82 2003/05/12 00:17:03 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/optimizer/util/plancat.c,v 1.83 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
#include "optimizer/clauses.h"
#include "optimizer/plancat.h"
#include "parser/parsetree.h"
+#include "rewrite/rewriteManip.h"
#include "utils/builtins.h"
#include "utils/fmgroids.h"
#include "utils/lsyscache.h"
Relation indexRelation;
Form_pg_index index;
IndexOptInfo *info;
+ int ncolumns;
int i;
int16 amorderstrategy;
/* Extract info from the relation descriptor for the index */
indexRelation = index_open(indexoid);
+ index = indexRelation->rd_index;
info = makeNode(IndexOptInfo);
- /*
- * Need to make these arrays large enough to be sure there is room
- * for a terminating 0 at the end of each one.
- */
- info->classlist = (Oid *) palloc(sizeof(Oid) * (INDEX_MAX_KEYS + 1));
- info->indexkeys = (int *) palloc(sizeof(int) * (INDEX_MAX_KEYS + 1));
- info->ordering = (Oid *) palloc(sizeof(Oid) * (INDEX_MAX_KEYS + 1));
-
- /* Extract info from the pg_index tuple */
- index = indexRelation->rd_index;
info->indexoid = index->indexrelid;
- info->indproc = index->indproc; /* functional index ?? */
- if (VARSIZE(&index->indpred) > VARHDRSZ) /* partial index ?? */
- {
- char *predString;
+ info->ncolumns = ncolumns = index->indnatts;
- predString = DatumGetCString(DirectFunctionCall1(textout,
- PointerGetDatum(&index->indpred)));
- info->indpred = (List *) stringToNode(predString);
- pfree(predString);
- }
- else
- info->indpred = NIL;
- info->unique = index->indisunique;
+ /*
+ * Need to make classlist and ordering arrays large enough to put
+ * a terminating 0 at the end of each one.
+ */
+ info->indexkeys = (int *) palloc(sizeof(int) * ncolumns);
+ info->classlist = (Oid *) palloc0(sizeof(Oid) * (ncolumns + 1));
+ info->ordering = (Oid *) palloc0(sizeof(Oid) * (ncolumns + 1));
- for (i = 0; i < INDEX_MAX_KEYS; i++)
+ for (i = 0; i < ncolumns; i++)
{
- if (index->indclass[i] == (Oid) 0)
- break;
info->classlist[i] = index->indclass[i];
- }
- info->classlist[i] = (Oid) 0;
- info->ncolumns = i;
-
- for (i = 0; i < INDEX_MAX_KEYS; i++)
- {
- if (index->indkey[i] == 0)
- break;
info->indexkeys[i] = index->indkey[i];
}
- info->indexkeys[i] = 0;
- info->nkeys = i;
info->relam = indexRelation->rd_rel->relam;
info->pages = indexRelation->rd_rel->relpages;
info->tuples = indexRelation->rd_rel->reltuples;
info->amcostestimate = index_cost_estimator(indexRelation);
- amorderstrategy = indexRelation->rd_am->amorderstrategy;
/*
* Fetch the ordering operators associated with the index, if any.
*/
- MemSet(info->ordering, 0, sizeof(Oid) * (INDEX_MAX_KEYS + 1));
+ amorderstrategy = indexRelation->rd_am->amorderstrategy;
if (amorderstrategy != 0)
{
int oprindex = amorderstrategy - 1;
- for (i = 0; i < info->ncolumns; i++)
+ for (i = 0; i < ncolumns; i++)
{
info->ordering[i] = indexRelation->rd_operator[oprindex];
oprindex += indexRelation->rd_am->amstrategies;
}
}
+ /*
+ * Fetch the index expressions and predicate, if any. We must
+ * modify the copies we obtain from the relcache to have the
+ * correct varno for the parent relation, so that they match up
+ * correctly against qual clauses.
+ */
+ info->indexprs = RelationGetIndexExpressions(indexRelation);
+ info->indpred = RelationGetIndexPredicate(indexRelation);
+ if (info->indexprs && varno != 1)
+ ChangeVarNodes((Node *) info->indexprs, 1, varno, 0);
+ if (info->indpred && varno != 1)
+ ChangeVarNodes((Node *) info->indpred, 1, varno, 0);
+ info->unique = index->indisunique;
+
/* initialize cached join info to empty */
info->outer_relids = NULL;
info->inner_paths = NIL;
IndexOptInfo *index = (IndexOptInfo *) lfirst(ilist);
/*
- * Note: ignore functional and partial indexes, since they don't
- * allow us to conclude that all attr values are distinct. Also, a
- * multicolumn unique index doesn't allow us to conclude that just
- * the specified attr is unique.
+ * Note: ignore partial indexes, since they don't allow us to conclude
+ * that all attr values are distinct. We don't take any interest in
+ * expressional indexes either. Also, a multicolumn unique index
+ * doesn't allow us to conclude that just the specified attr is
+ * unique.
*/
if (index->unique &&
- index->nkeys == 1 &&
+ index->ncolumns == 1 &&
index->indexkeys[0] == attno &&
- index->indproc == InvalidOid &&
index->indpred == NIL)
return true;
}
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $Header: /cvsroot/pgsql/src/backend/parser/analyze.c,v 1.271 2003/05/06 00:20:32 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/parser/analyze.c,v 1.272 2003/05/28 16:03:56 tgl Exp $
*
*-------------------------------------------------------------------------
*/
/* OK, add it to the index definition */
iparam = makeNode(IndexElem);
iparam->name = pstrdup(key);
- iparam->funcname = NIL;
- iparam->args = NIL;
+ iparam->expr = NULL;
iparam->opclass = NIL;
index->indexParams = lappend(index->indexParams, iparam);
}
if (index->idxname == NULL && index->indexParams != NIL)
{
- iparam = lfirst(index->indexParams);
+ iparam = (IndexElem *) lfirst(index->indexParams);
+ /* we should never see an expression item here */
+ Assert(iparam->expr == NULL);
index->idxname = CreateIndexName(cxt->relation->relname,
- iparam->name ? iparam->name :
- strVal(llast(iparam->funcname)),
- "key", cxt->alist);
+ iparam->name,
+ "key",
+ cxt->alist);
}
if (index->idxname == NULL) /* should not happen */
elog(ERROR, "%s: failed to make implicit index name",
transformIndexStmt(ParseState *pstate, IndexStmt *stmt)
{
Query *qry;
- RangeTblEntry *rte;
+ RangeTblEntry *rte = NULL;
+ List *l;
qry = makeNode(Query);
qry->commandType = CMD_UTILITY;
stmt->whereClause = transformWhereClause(pstate, stmt->whereClause);
}
+ /* take care of any index expressions */
+ foreach(l, stmt->indexParams)
+ {
+ IndexElem *ielem = (IndexElem *) lfirst(l);
+
+ if (ielem->expr)
+ {
+ /* Set up rtable as for predicate, see notes above */
+ if (rte == NULL)
+ {
+ rte = addRangeTableEntry(pstate, stmt->relation, NULL,
+ false, true);
+ /* no to join list, yes to namespace */
+ addRTEtoQuery(pstate, rte, false, true);
+ }
+ ielem->expr = transformExpr(pstate, ielem->expr);
+ /*
+ * We check only that the result type is legitimate; this is
+ * for consistency with what transformWhereClause() checks for
+ * the predicate. DefineIndex() will make more checks.
+ */
+ if (expression_returns_set(ielem->expr))
+ elog(ERROR, "index expression may not return a set");
+ }
+ }
+
qry->hasSubLinks = pstate->p_hasSubLinks;
stmt->rangetable = pstate->p_rtable;
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/parser/gram.y,v 2.414 2003/05/15 16:35:28 momjian Exp $
+ * $Header: /cvsroot/pgsql/src/backend/parser/gram.y,v 2.415 2003/05/28 16:03:57 tgl Exp $
*
* HISTORY
* AUTHOR DATE MAJOR EVENT
oper_argtypes RuleActionList RuleActionMulti
opt_column_list columnList opt_name_list
sort_clause opt_sort_clause sortby_list index_params
- index_list name_list from_clause from_list opt_array_bounds
+ name_list from_clause from_list opt_array_bounds
qualified_name_list any_name any_name_list
any_operator expr_list dotted_name attrs
target_list update_target_list insert_column_list
%type <columnref> columnref
%type <alias> alias_clause
%type <sortgroupby> sortby
-%type <ielem> index_elem func_index
+%type <ielem> index_elem
%type <node> table_ref
%type <jexpr> joined_table
%type <range> relation_expr
%token UNIONJOIN
/* Special keywords, not in the query language - see the "lex" file */
-%token <str> IDENT FCONST SCONST NCONST BCONST XCONST Op
+%token <str> IDENT FCONST SCONST BCONST XCONST Op
%token <ival> ICONST PARAM
/* precedence: lowest to highest */
*
* QUERY:
* create index <indexname> on <relname>
- * [ using <access> ] "(" (<col> with <op>)+ ")"
+ * [ using <access> ] "(" ( <col> [ using <opclass> ] )+ ")"
* [ where <predicate> ]
*
*****************************************************************************/
access_method_clause:
USING access_method { $$ = $2; }
- /* If btree changes as our default, update pg_get_indexdef() */
| /*EMPTY*/ { $$ = DEFAULT_INDEX_TYPE; }
;
-index_params:
- index_list { $$ = $1; }
- | func_index { $$ = makeList1($1); }
+index_params: index_elem { $$ = makeList1($1); }
+ | index_params ',' index_elem { $$ = lappend($1, $3); }
;
-index_list: index_elem { $$ = makeList1($1); }
- | index_list ',' index_elem { $$ = lappend($1, $3); }
- ;
-
-func_index: func_name '(' name_list ')' opt_class
+/*
+ * Index attributes can be either simple column references, or arbitrary
+ * expressions in parens. For backwards-compatibility reasons, we allow
+ * an expression that's just a function call to be written without parens.
+ */
+index_elem: attr_name opt_class
+ {
+ $$ = makeNode(IndexElem);
+ $$->name = $1;
+ $$->expr = NULL;
+ $$->opclass = $2;
+ }
+ | func_name '(' expr_list ')' opt_class
{
+ FuncCall *n = makeNode(FuncCall);
+ n->funcname = $1;
+ n->args = $3;
+ n->agg_star = FALSE;
+ n->agg_distinct = FALSE;
+
$$ = makeNode(IndexElem);
$$->name = NULL;
- $$->funcname = $1;
- $$->args = $3;
+ $$->expr = (Node *)n;
$$->opclass = $5;
}
- ;
-
-index_elem: attr_name opt_class
+ | '(' a_expr ')' opt_class
{
$$ = makeNode(IndexElem);
- $$->name = $1;
- $$->funcname = NIL;
- $$->args = NIL;
- $$->opclass = $2;
+ $$->name = NULL;
+ $$->expr = $2;
+ $$->opclass = $4;
}
;
-opt_class: any_name
- {
- /*
- * Release 7.0 removed network_ops, timespan_ops, and
- * datetime_ops, so we suppress it from being passed to
- * the parser so the default *_ops is used. This can be
- * removed in some later release. bjm 2000/02/07
- *
- * Release 7.1 removes lztext_ops, so suppress that too
- * for a while. tgl 2000/07/30
- *
- * Release 7.2 renames timestamp_ops to timestamptz_ops,
- * so suppress that too for awhile. I'm starting to
- * think we need a better approach. tgl 2000/10/01
- */
- if (length($1) == 1)
- {
- char *claname = strVal(lfirst($1));
-
- if (strcmp(claname, "network_ops") != 0 &&
- strcmp(claname, "timespan_ops") != 0 &&
- strcmp(claname, "datetime_ops") != 0 &&
- strcmp(claname, "lztext_ops") != 0 &&
- strcmp(claname, "timestamp_ops") != 0)
- $$ = $1;
- else
- $$ = NIL;
- }
- else
- $$ = $1;
- }
+opt_class: any_name { $$ = $1; }
| USING any_name { $$ = $2; }
| /*EMPTY*/ { $$ = NIL; }
;
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/tcop/pquery.c,v 1.65 2003/05/27 17:49:46 momjian Exp $
+ * $Header: /cvsroot/pgsql/src/backend/tcop/pquery.c,v 1.66 2003/05/28 16:03:58 tgl Exp $
*
*-------------------------------------------------------------------------
*/
* without freezing a snapshot. By extension we allow SHOW
* not to set a snapshot. The other stmts listed are just
* efficiency hacks. Beware of listing anything that can
- * modify the database --- if, say, it has to update a
- * functional index, then it had better have a snapshot.
+ * modify the database --- if, say, it has to update an
+ * index with expressions that invoke user-defined functions,
+ * then it had better have a snapshot.
*/
if (! (IsA(utilityStmt, TransactionStmt) ||
IsA(utilityStmt, LockStmt) ||
* back to source text
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/utils/adt/ruleutils.c,v 1.140 2003/05/20 20:35:10 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/utils/adt/ruleutils.c,v 1.141 2003/05/28 16:03:59 tgl Exp $
*
* This software is copyrighted by Jan Wieck - Hamburg.
*
Form_pg_index idxrec;
Form_pg_class idxrelrec;
Form_pg_am amrec;
+ List *indexprs;
+ List *context;
Oid indrelid;
int len;
int keyno;
- Oid keycoltypes[INDEX_MAX_KEYS];
+ Oid keycoltype;
StringInfoData buf;
- StringInfoData keybuf;
+ char *str;
char *sep;
/*
elog(ERROR, "syscache lookup for AM %u failed", idxrelrec->relam);
amrec = (Form_pg_am) GETSTRUCT(ht_am);
+ /*
+ * Get the index expressions, if any. (NOTE: we do not use the relcache
+ * versions of the expressions and predicate, because we want to display
+ * non-const-folded expressions.)
+ */
+ if (!heap_attisnull(ht_idx, Anum_pg_index_indexprs))
+ {
+ Datum exprsDatum;
+ bool isnull;
+ char *exprsString;
+
+ exprsDatum = SysCacheGetAttr(INDEXRELID, ht_idx,
+ Anum_pg_index_indexprs, &isnull);
+ Assert(!isnull);
+ exprsString = DatumGetCString(DirectFunctionCall1(textout,
+ exprsDatum));
+ indexprs = (List *) stringToNode(exprsString);
+ pfree(exprsString);
+ }
+ else
+ indexprs = NIL;
+
+ context = deparse_context_for(get_rel_name(indrelid), indrelid);
+
/*
* Start the index definition. Note that the index's name should
* never be schema-qualified, but the indexed rel's name may be.
quote_identifier(NameStr(amrec->amname)));
/*
- * Collect the indexed attributes in keybuf
+ * Report the indexed attributes
*/
- initStringInfo(&keybuf);
sep = "";
- for (keyno = 0; keyno < INDEX_MAX_KEYS; keyno++)
+ for (keyno = 0; keyno < idxrec->indnatts; keyno++)
{
AttrNumber attnum = idxrec->indkey[keyno];
- char *attname;
-
- if (attnum == InvalidAttrNumber)
- break;
-
- attname = get_relid_attribute_name(indrelid, attnum);
- keycoltypes[keyno] = get_atttype(indrelid, attnum);
- appendStringInfo(&keybuf, sep);
+ appendStringInfo(&buf, sep);
sep = ", ";
- /*
- * Add the indexed field name
- */
- appendStringInfo(&keybuf, "%s", quote_identifier(attname));
+ if (attnum != 0)
+ {
+ /* Simple index column */
+ char *attname;
- /*
- * If not a functional index, add the operator class name
- */
- if (idxrec->indproc == InvalidOid)
- get_opclass_name(idxrec->indclass[keyno],
- keycoltypes[keyno],
- &keybuf);
- }
+ attname = get_relid_attribute_name(indrelid, attnum);
+ appendStringInfo(&buf, "%s", quote_identifier(attname));
+ keycoltype = get_atttype(indrelid, attnum);
+ }
+ else
+ {
+ /* expressional index */
+ Node *indexkey;
+
+ if (indexprs == NIL)
+ elog(ERROR, "too few entries in indexprs list");
+ indexkey = (Node *) lfirst(indexprs);
+ indexprs = lnext(indexprs);
+ /* Deparse */
+ str = deparse_expression(indexkey, context, false, false);
+ /* Need parens if it's not a bare function call */
+ if (indexkey && IsA(indexkey, FuncExpr) &&
+ ((FuncExpr *) indexkey)->funcformat == COERCE_EXPLICIT_CALL)
+ appendStringInfo(&buf, "%s", str);
+ else
+ appendStringInfo(&buf, "(%s)", str);
+ keycoltype = exprType(indexkey);
+ }
- if (idxrec->indproc != InvalidOid)
- {
/*
- * For functional index say 'func (attrs) opclass'
+ * Add the operator class name
*/
- appendStringInfo(&buf, "%s(%s)",
- generate_function_name(idxrec->indproc,
- keyno, keycoltypes),
- keybuf.data);
- get_opclass_name(idxrec->indclass[0],
- get_func_rettype(idxrec->indproc),
+ get_opclass_name(idxrec->indclass[keyno], keycoltype,
&buf);
}
- else
- {
- /*
- * Otherwise say 'attr opclass [, ...]'
- */
- appendStringInfo(&buf, "%s", keybuf.data);
- }
appendStringInfoChar(&buf, ')');
/*
* If it's a partial index, decompile and append the predicate
*/
- if (VARSIZE(&idxrec->indpred) > VARHDRSZ)
+ if (!heap_attisnull(ht_idx, Anum_pg_index_indpred))
{
Node *node;
- List *context;
- char *exprstr;
- char *str;
+ Datum predDatum;
+ bool isnull;
+ char *predString;
- /* Convert TEXT object to C string */
- exprstr = DatumGetCString(DirectFunctionCall1(textout,
- PointerGetDatum(&idxrec->indpred)));
- /* Convert expression to node tree */
- node = (Node *) stringToNode(exprstr);
+ /* Convert text string to node tree */
+ predDatum = SysCacheGetAttr(INDEXRELID, ht_idx,
+ Anum_pg_index_indpred, &isnull);
+ Assert(!isnull);
+ predString = DatumGetCString(DirectFunctionCall1(textout,
+ predDatum));
+ node = (Node *) stringToNode(predString);
+ pfree(predString);
/*
* If top level is a List, assume it is an implicit-AND structure,
if (node && IsA(node, List))
node = (Node *) make_ands_explicit((List *) node);
/* Deparse */
- context = deparse_context_for(get_rel_name(indrelid), indrelid);
str = deparse_expression(node, context, false, false);
appendStringInfo(&buf, " WHERE %s", str);
}
memcpy(VARDATA(indexdef), buf.data, buf.len);
pfree(buf.data);
- pfree(keybuf.data);
ReleaseSysCache(ht_idx);
ReleaseSysCache(ht_idxrel);
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/utils/adt/selfuncs.c,v 1.138 2003/05/26 00:11:27 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/utils/adt/selfuncs.c,v 1.139 2003/05/28 16:03:59 tgl Exp $
*
*-------------------------------------------------------------------------
*/
indexSelectivity, indexCorrelation);
/*
- * If it's a functional index, leave the default zero-correlation
- * estimate in place. If not, and if we can get an estimate for the
- * first variable's ordering correlation C from pg_statistic, estimate
+ * If the first column is a simple variable, and we can get an estimate
+ * for its ordering correlation C from pg_statistic, estimate
* the index correlation as C / number-of-columns. (The idea here is
* that multiple columns dilute the importance of the first column's
* ordering, but don't negate it entirely.)
*/
- if (index->indproc == InvalidOid)
+ if (index->indexkeys[0] != 0)
{
Oid relid;
HeapTuple tuple;
Assert(nnumbers == 1);
varCorrelation = numbers[0];
- for (nKeys = 1; index->indexkeys[nKeys] != 0; nKeys++)
- /* skip */ ;
+ nKeys = index->ncolumns;
*indexCorrelation = varCorrelation / nKeys;
*
*
* IDENTIFICATION
- * $Header: /cvsroot/pgsql/src/backend/utils/cache/relcache.c,v 1.184 2003/02/09 06:56:28 tgl Exp $
+ * $Header: /cvsroot/pgsql/src/backend/utils/cache/relcache.c,v 1.185 2003/05/28 16:03:59 tgl Exp $
*
*-------------------------------------------------------------------------
*/
#include "catalog/pg_type.h"
#include "commands/trigger.h"
#include "miscadmin.h"
+#include "optimizer/clauses.h"
+#include "optimizer/planmain.h"
#include "storage/smgr.h"
#include "utils/builtins.h"
#include "utils/catcache.h"
RelationInitIndexAccessInfo(Relation relation)
{
HeapTuple tuple;
- Size iformsize;
- Form_pg_index iform;
Form_pg_am aform;
MemoryContext indexcxt;
+ MemoryContext oldcontext;
IndexStrategy strategy;
Oid *operator;
RegProcedure *support;
uint16 amsupport;
/*
- * Make a copy of the pg_index entry for the index. Note that this is
- * a variable-length tuple.
+ * Make a copy of the pg_index entry for the index. Since pg_index
+ * contains variable-length and possibly-null fields, we have to do
+ * this honestly rather than just treating it as a Form_pg_index struct.
*/
tuple = SearchSysCache(INDEXRELID,
ObjectIdGetDatum(RelationGetRelid(relation)),
if (!HeapTupleIsValid(tuple))
elog(ERROR, "RelationInitIndexAccessInfo: no pg_index entry for index %u",
RelationGetRelid(relation));
- iformsize = tuple->t_len - tuple->t_data->t_hoff;
- iform = (Form_pg_index) MemoryContextAlloc(CacheMemoryContext, iformsize);
- memcpy(iform, GETSTRUCT(tuple), iformsize);
+ oldcontext = MemoryContextSwitchTo(CacheMemoryContext);
+ relation->rd_indextuple = heap_copytuple(tuple);
+ relation->rd_index = (Form_pg_index) GETSTRUCT(relation->rd_indextuple);
+ MemoryContextSwitchTo(oldcontext);
ReleaseSysCache(tuple);
- relation->rd_index = iform;
/*
* Make a copy of the pg_am entry for the index's access method
relation->rd_am = aform;
natts = relation->rd_rel->relnatts;
+ if (natts != relation->rd_index->indnatts)
+ elog(ERROR, "RelationInitIndexAccessInfo: relnatts disagrees with indnatts for index %u",
+ RelationGetRelid(relation));
amstrategies = aform->amstrategies;
amsupport = aform->amsupport;
* Fill the strategy map and the support RegProcedure arrays.
* (supportinfo is left as zeroes, and is filled on-the-fly when used)
*/
- IndexSupportInitialize(iform,
+ IndexSupportInitialize(relation->rd_index,
strategy, operator, support,
amstrategies, amsupport, natts);
+
+ /*
+ * expressions and predicate cache will be filled later
+ */
+ relation->rd_indexprs = NIL;
+ relation->rd_indpred = NIL;
}
/*
{
OpClassCacheEnt *opcentry;
- if (iform->indkey[attIndex] == InvalidAttrNumber ||
- !OidIsValid(iform->indclass[attIndex]))
+ if (!OidIsValid(iform->indclass[attIndex]))
elog(ERROR, "IndexSupportInitialize: bogus pg_index tuple");
/* look up the info for this opclass, using a cache */
* with, we can only get rid of these fields:
*/
FreeTriggerDesc(relation->trigdesc);
- if (relation->rd_index)
- pfree(relation->rd_index);
+ if (relation->rd_indextuple)
+ pfree(relation->rd_indextuple);
if (relation->rd_am)
pfree(relation->rd_am);
if (relation->rd_rel)
return list;
}
+/*
+ * RelationGetIndexExpressions -- get the index expressions for an index
+ *
+ * We cache the result of transforming pg_index.indexprs into a node tree.
+ * If the rel is not an index or has no expressional columns, we return NIL.
+ * Otherwise, the returned tree is copied into the caller's memory context.
+ * (We don't want to return a pointer to the relcache copy, since it could
+ * disappear due to relcache invalidation.)
+ */
+List *
+RelationGetIndexExpressions(Relation relation)
+{
+ List *result;
+ Datum exprsDatum;
+ bool isnull;
+ char *exprsString;
+ MemoryContext oldcxt;
+
+ /* Quick exit if we already computed the result. */
+ if (relation->rd_indexprs)
+ return (List *) copyObject(relation->rd_indexprs);
+
+ /* Quick exit if there is nothing to do. */
+ if (relation->rd_indextuple == NULL ||
+ heap_attisnull(relation->rd_indextuple, Anum_pg_index_indexprs))
+ return NIL;
+
+ /*
+ * We build the tree we intend to return in the caller's context.
+ * After successfully completing the work, we copy it into the relcache
+ * entry. This avoids problems if we get some sort of
+ * error partway through.
+ *
+ * We make use of the syscache's copy of pg_index's tupledesc
+ * to access the non-fixed fields of the tuple. We assume that
+ * the syscache will be initialized before any access of a
+ * partial index could occur. (This would probably fail if we
+ * were to allow partial indexes on system catalogs.)
+ */
+ exprsDatum = SysCacheGetAttr(INDEXRELID, relation->rd_indextuple,
+ Anum_pg_index_indexprs, &isnull);
+ Assert(!isnull);
+ exprsString = DatumGetCString(DirectFunctionCall1(textout, exprsDatum));
+ result = (List *) stringToNode(exprsString);
+ pfree(exprsString);
+
+ /*
+ * Run the expressions through eval_const_expressions. This is not just
+ * an optimization, but is necessary, because the planner will be
+ * comparing them to const-folded qual clauses, and may fail to detect
+ * valid matches without this.
+ */
+ result = (List *) eval_const_expressions((Node *) result);
+
+ /* May as well fix opfuncids too */
+ fix_opfuncids((Node *) result);
+
+ /* Now save a copy of the completed tree in the relcache entry. */
+ oldcxt = MemoryContextSwitchTo(CacheMemoryContext);
+ relation->rd_indexprs = (List *) copyObject(result);
+ MemoryContextSwitchTo(oldcxt);
+
+ return result;
+}
+
+/*
+ * RelationGetIndexPredicate -- get the index predicate for an index
+ *
+ * We cache the result of transforming pg_index.indpred into a node tree.
+ * If the rel is not an index or has no predicate, we return NIL.
+ * Otherwise, the returned tree is copied into the caller's memory context.
+ * (We don't want to return a pointer to the relcache copy, since it could
+ * disappear due to relcache invalidation.)
+ */
+List *
+RelationGetIndexPredicate(Relation relation)
+{
+ List *result;
+ Datum predDatum;
+ bool isnull;
+ char *predString;
+ MemoryContext oldcxt;
+
+ /* Quick exit if we already computed the result. */
+ if (relation->rd_indpred)
+ return (List *) copyObject(relation->rd_indpred);
+
+ /* Quick exit if there is nothing to do. */
+ if (relation->rd_indextuple == NULL ||
+ heap_attisnull(relation->rd_indextuple, Anum_pg_index_indpred))
+ return NIL;
+
+ /*
+ * We build the tree we intend to return in the caller's context.
+ * After successfully completing the work, we copy it into the relcache
+ * entry. This avoids problems if we get some sort of
+ * error partway through.
+ *
+ * We make use of the syscache's copy of pg_index's tupledesc
+ * to access the non-fixed fields of the tuple. We assume that
+ * the syscache will be initialized before any access of a
+ * partial index could occur. (This would probably fail if we
+ * were to allow partial indexes on system catalogs.)
+ */
+ predDatum = SysCacheGetAttr(INDEXRELID, relation->rd_indextuple,
+ Anum_pg_index_indpred, &isnull);
+ Assert(!isnull);
+ predString = DatumGetCString(DirectFunctionCall1(textout, predDatum));
+ result = (List *) stringToNode(predString);
+ pfree(predString);
+
+ /*
+ * Run the expression through eval_const_expressions. This is not just
+ * an optimization, but is necessary, because the planner will be
+ * comparing it to const-folded qual clauses, and may fail to detect
+ * valid matches without this.
+ */
+ result = (List *) eval_const_expressions((Node *) result);
+
+ /* May as well fix opfuncids too */
+ fix_opfuncids((Node *) result);
+
+ /* Now save a copy of the completed tree in the relcache entry. */
+ oldcxt = MemoryContextSwitchTo(CacheMemoryContext);
+ relation->rd_indpred = (List *) copyObject(result);
+ MemoryContextSwitchTo(oldcxt);
+
+ return result;
+}
+
/*
* load_relcache_init_file, write_relcache_init_file
if (rel->rd_isnailed)
nailed_indexes++;
- /* next, read the pg_index tuple form */
+ /* next, read the pg_index tuple */
if ((nread = fread(&len, 1, sizeof(len), fp)) != sizeof(len))
goto read_failed;
- rel->rd_index = (Form_pg_index) palloc(len);
- if ((nread = fread(rel->rd_index, 1, len, fp)) != len)
+ rel->rd_indextuple = (HeapTuple) palloc(len);
+ if ((nread = fread(rel->rd_indextuple, 1, len, fp)) != len)
goto read_failed;
+ /* Fix up internal pointers in the tuple -- see heap_copytuple */
+ rel->rd_indextuple->t_datamcxt = CurrentMemoryContext;
+ rel->rd_indextuple->t_data = (HeapTupleHeader) ((char *) rel->rd_indextuple + HEAPTUPLESIZE);
+ rel->rd_index = (Form_pg_index) GETSTRUCT(rel->rd_indextuple);
+
/* next, read the access method tuple form */
if ((nread = fread(&len, 1, sizeof(len), fp)) != sizeof(len))
goto read_failed;
nailed_rels++;
Assert(rel->rd_index == NULL);
+ Assert(rel->rd_indextuple == NULL);
Assert(rel->rd_am == NULL);
Assert(rel->rd_indexcxt == NULL);
Assert(rel->rd_istrat == NULL);
* format is complex and subject to change). They must be rebuilt
* if needed by RelationCacheInitializePhase2. This is not
* expected to be a big performance hit since few system catalogs
- * have such.
+ * have such. Ditto for index expressions and predicates.
*/
rel->rd_rules = NULL;
rel->rd_rulescxt = NULL;
rel->trigdesc = NULL;
+ rel->rd_indexprs = NIL;
+ rel->rd_indpred = NIL;
/*
* Reset transient-state fields in the relcache entry
if (rel->rd_rel->relkind == RELKIND_INDEX)
{
Form_pg_am am = rel->rd_am;
- HeapTuple tuple;
- /*
- * We need to write the index tuple form, but this is a bit
- * tricky since it's a variable-length struct. Rather than
- * hoping to intuit the length, fetch the pg_index tuple
- * afresh using the syscache, and write that.
- */
- tuple = SearchSysCache(INDEXRELID,
- ObjectIdGetDatum(RelationGetRelid(rel)),
- 0, 0, 0);
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "write_relcache_init_file: no pg_index entry for index %u",
- RelationGetRelid(rel));
- len = tuple->t_len - tuple->t_data->t_hoff;
+ /* write the pg_index tuple */
+ /* we assume this was created by heap_copytuple! */
+ len = HEAPTUPLESIZE + rel->rd_indextuple->t_len;
if (fwrite(&len, 1, sizeof(len), fp) != sizeof(len))
- elog(FATAL, "cannot write init file -- index tuple form length");
- if (fwrite(GETSTRUCT(tuple), 1, len, fp) != len)
- elog(FATAL, "cannot write init file -- index tuple form");
- ReleaseSysCache(tuple);
+ elog(FATAL, "cannot write init file -- index tuple length");
+
+ if (fwrite(rel->rd_indextuple, 1, len, fp) != len)
+ elog(FATAL, "cannot write init file -- index tuple");
/* next, write the access method tuple form */
len = sizeof(FormData_pg_am);
*
* Copyright 2000-2002 by PostgreSQL Global Development Group
*
- * $Header: /cvsroot/pgsql/src/bin/psql/describe.c,v 1.76 2003/03/27 16:57:39 momjian Exp $
+ * $Header: /cvsroot/pgsql/src/bin/psql/describe.c,v 1.77 2003/05/28 16:03:59 tgl Exp $
*/
#include "postgres_fe.h"
#include "describe.h"
headers[cols] = NULL;
/* Get column info (index requires additional checks) */
- if (tableinfo.relkind == 'i')
- printfPQExpBuffer(&buf, "SELECT\n CASE i.indproc WHEN ('-'::pg_catalog.regproc) THEN a.attname\n ELSE SUBSTR(pg_catalog.pg_get_indexdef(attrelid),\n POSITION('(' in pg_catalog.pg_get_indexdef(attrelid)))\n END,");
- else
- printfPQExpBuffer(&buf, "SELECT a.attname,");
+ printfPQExpBuffer(&buf, "SELECT a.attname,");
appendPQExpBuffer(&buf, "\n pg_catalog.format_type(a.atttypid, a.atttypmod),"
"\n (SELECT substring(d.adsrc for 128) FROM pg_catalog.pg_attrdef d"
"\n WHERE d.adrelid = a.attrelid AND d.adnum = a.attnum AND a.atthasdef),"
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $Id: catversion.h,v 1.196 2003/05/26 00:11:27 tgl Exp $
+ * $Id: catversion.h,v 1.197 2003/05/28 16:03:59 tgl Exp $
*
*-------------------------------------------------------------------------
*/
*/
/* yyyymmddN */
-#define CATALOG_VERSION_NO 200305241
+#define CATALOG_VERSION_NO 200305271
#endif
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $Id: dependency.h,v 1.7 2003/03/06 22:54:49 tgl Exp $
+ * $Id: dependency.h,v 1.8 2003/05/28 16:03:59 tgl Exp $
*
*-------------------------------------------------------------------------
*/
Node *expr, List *rtable,
DependencyType behavior);
+extern void recordDependencyOnSingleRelExpr(const ObjectAddress *depender,
+ Node *expr, Oid relId,
+ DependencyType behavior,
+ DependencyType self_behavior);
+
/* in pg_depend.c */
extern void recordDependencyOn(const ObjectAddress *depender,
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $Id: index.h,v 1.50 2002/09/23 00:42:48 tgl Exp $
+ * $Id: index.h,v 1.51 2003/05/28 16:04:00 tgl Exp $
*
*-------------------------------------------------------------------------
*/
#include "catalog/pg_index.h"
#include "nodes/execnodes.h"
+
#define DEFAULT_INDEX_TYPE "btree"
/* Typedef for callback function for IndexBuildHeapScan */
extern void index_drop(Oid indexId);
-extern IndexInfo *BuildIndexInfo(Form_pg_index indexStruct);
+extern IndexInfo *BuildIndexInfo(Relation index);
extern void FormIndexDatum(IndexInfo *indexInfo,
HeapTuple heapTuple,
TupleDesc heapDescriptor,
- MemoryContext resultCxt,
+ EState *estate,
Datum *datum,
char *nullv);
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $Id: pg_index.h,v 1.30 2003/03/10 22:28:19 tgl Exp $
+ * $Id: pg_index.h,v 1.31 2003/05/28 16:04:00 tgl Exp $
*
* NOTES
* the genbki.sh script reads this file and generates .bki
{
Oid indexrelid; /* OID of the index */
Oid indrelid; /* OID of the relation it indexes */
- regproc indproc; /* OID of function for functional index */
- int2vector indkey; /* column numbers of indexed attributes */
+ int2vector indkey; /* column numbers of indexed cols, or 0 */
oidvector indclass; /* opclass identifiers */
- bool indisclustered; /* is this the index last clustered by? */
+ int2 indnatts; /* number of columns in index */
bool indisunique; /* is this a unique index? */
bool indisprimary; /* is this index for primary key? */
- Oid indreference; /* oid of index of referenced relation (ie
- * - this index for foreign key) */
+ bool indisclustered; /* is this the index last clustered by? */
- /* VARIABLE LENGTH FIELD: */
+ /* VARIABLE LENGTH FIELDS: */
+ text indexprs; /* expression trees for index attributes
+ * that are not simple column references;
+ * one for each zero entry in indkey[] */
text indpred; /* expression tree for predicate, if a
- * partial index */
+ * partial index; else NULL */
} FormData_pg_index;
/* ----------------
#define Natts_pg_index 10
#define Anum_pg_index_indexrelid 1
#define Anum_pg_index_indrelid 2
-#define Anum_pg_index_indproc 3
-#define Anum_pg_index_indkey 4
-#define Anum_pg_index_indclass 5
-#define Anum_pg_index_indisclustered 6
-#define Anum_pg_index_indisunique 7
-#define Anum_pg_index_indisprimary 8
-#define Anum_pg_index_indreference 9
+#define Anum_pg_index_indkey 3
+#define Anum_pg_index_indclass 4
+#define Anum_pg_index_indnatts 5
+#define Anum_pg_index_indisunique 6
+#define Anum_pg_index_indisprimary 7
+#define Anum_pg_index_indisclustered 8
+#define Anum_pg_index_indexprs 9
#define Anum_pg_index_indpred 10
#endif /* PG_INDEX_H */
* <opcamid, opcname> --- that is, there is a row for each valid combination
* of opclass name and index access method type. This row specifies the
* expected input data type for the opclass (the type of the heap column,
- * or the function output type in the case of a functional index). Note
+ * or the expression output type in the case of an index expression). Note
* that types binary-coercible to the specified type will be accepted too.
*
* For a given <opcamid, opcintype> pair, there can be at most one row that
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $Id: pg_opclass.h,v 1.49 2003/05/26 00:11:27 tgl Exp $
+ * $Id: pg_opclass.h,v 1.50 2003/05/28 16:04:00 tgl Exp $
*
* NOTES
* the genbki.sh script reads this file and generates .bki
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $Id: execnodes.h,v 1.97 2003/04/08 23:20:04 tgl Exp $
+ * $Id: execnodes.h,v 1.98 2003/05/28 16:04:00 tgl Exp $
*
*-------------------------------------------------------------------------
*/
/* ----------------
* IndexInfo information
*
- * this class holds the information needed to construct new index
+ * this struct holds the information needed to construct new index
* entries for a particular index. Used for both index_build and
* retail creation of index entries.
*
* NumIndexAttrs number of columns in this index
- * (1 if a func. index, else same as NumKeyAttrs)
- * NumKeyAttrs number of key attributes for this index
- * (ie, number of attrs from underlying relation)
* KeyAttrNumbers underlying-rel attribute numbers used as keys
+ * (zeroes indicate expressions)
+ * Expressions expr trees for expression entries, or NIL if none
+ * ExpressionsState exec state for expressions, or NIL if none
* Predicate partial-index predicate, or NIL if none
* PredicateState exec state for predicate, or NIL if none
- * FuncOid OID of function, or InvalidOid if not f. index
- * FuncInfo fmgr lookup data for function, if FuncOid valid
* Unique is it a unique index?
* ----------------
*/
{
NodeTag type;
int ii_NumIndexAttrs;
- int ii_NumKeyAttrs;
AttrNumber ii_KeyAttrNumbers[INDEX_MAX_KEYS];
+ List *ii_Expressions; /* list of Expr */
+ List *ii_ExpressionsState; /* list of ExprState */
List *ii_Predicate; /* list of Expr */
List *ii_PredicateState; /* list of ExprState */
- Oid ii_FuncOid;
- FmgrInfo ii_FuncInfo;
bool ii_Unique;
} IndexInfo;
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $Id: parsenodes.h,v 1.237 2003/05/02 20:54:36 tgl Exp $
+ * $Id: parsenodes.h,v 1.238 2003/05/28 16:04:02 tgl Exp $
*
*-------------------------------------------------------------------------
*/
/*
* IndexElem - index parameters (used in CREATE INDEX)
*
- * For a plain index, each 'name' is an attribute name in the heap relation;
- * 'funcname' and 'args' are NIL. For a functional index, only one IndexElem
- * is allowed. It has name = NULL, funcname = name of function and args =
- * list of attribute names that are the function's arguments.
+ * For a plain index attribute, 'name' is the name of the table column to
+ * index, and 'expr' is NULL. For an index expression, 'name' is NULL and
+ * 'expr' is the expression tree.
*/
typedef struct IndexElem
{
NodeTag type;
char *name; /* name of attribute to index, or NULL */
- List *funcname; /* qualified name of function */
- List *args; /* list of names of function arguments */
+ Node *expr; /* expression to index, or NULL */
List *opclass; /* name of desired opclass; NIL = default */
} IndexElem;
char *accessMethod; /* name of access method (eg. btree) */
List *indexParams; /* a list of IndexElem */
Node *whereClause; /* qualification (partial-index predicate) */
- List *rangetable; /* range table for qual, filled in by
- * transformStmt() */
+ List *rangetable; /* range table for qual and/or expressions,
+ * filled in by transformStmt() */
bool unique; /* is index unique? */
bool primary; /* is index on primary key? */
bool isconstraint; /* is it from a CONSTRAINT clause? */
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $Id: relation.h,v 1.79 2003/02/08 20:20:55 tgl Exp $
+ * $Id: relation.h,v 1.80 2003/05/28 16:04:02 tgl Exp $
*
*-------------------------------------------------------------------------
*/
* and indexes, but that created confusion without actually doing anything
* useful. So now we have a separate IndexOptInfo struct for indexes.
*
- * ncolumns and nkeys are the same except for a functional index,
- * wherein ncolumns is 1 (the single function output) while nkeys
- * is the number of table columns passed to the function. classlist[]
- * and ordering[] have ncolumns entries, while indexkeys[] has nkeys
- * entries.
+ * classlist[], indexkeys[], and ordering[] have ncolumns entries.
+ * Zeroes in the indexkeys[] array indicate index columns that are
+ * expressions; there is one element in indexprs for each such column.
*
- * Note: for historical reasons, the arrays classlist, indexkeys and
- * ordering have an extra entry that is always zero. Some code scans
- * until it sees a zero rather than looking at ncolumns or nkeys.
+ * Note: for historical reasons, the classlist and ordering arrays have
+ * an extra entry that is always zero. Some code scans until it sees a
+ * zero entry, rather than looking at ncolumns.
+ *
+ * The indexprs and indpred expressions have been run through
+ * eval_const_expressions() for ease of matching to WHERE clauses.
*/
typedef struct IndexOptInfo
/* index descriptor information */
int ncolumns; /* number of columns in index */
- int nkeys; /* number of keys used by index */
Oid *classlist; /* OIDs of operator classes for columns */
- int *indexkeys; /* column numbers of index's keys */
+ int *indexkeys; /* column numbers of index's keys, or 0 */
Oid *ordering; /* OIDs of sort operators for each column */
Oid relam; /* OID of the access method (in pg_am) */
RegProcedure amcostestimate; /* OID of the access method's cost fcn */
- Oid indproc; /* OID of func if functional index, else 0 */
+ List *indexprs; /* expressions for non-simple index columns */
List *indpred; /* predicate if a partial index, else NIL */
bool unique; /* true if a unique index */
/*
* key typically points to a Var node, ie a relation attribute, but it
- * can also point to a FuncExpr clause representing the value indexed by a
- * functional index. Someday we might allow arbitrary expressions as
- * path keys, so don't assume more than you must.
+ * can also point to an arbitrary expression representing the value
+ * indexed by an index expression.
*/
} PathKeyItem;
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $Id: clauses.h,v 1.62 2003/02/04 00:50:01 tgl Exp $
+ * $Id: clauses.h,v 1.63 2003/05/28 16:04:02 tgl Exp $
*
*-------------------------------------------------------------------------
*/
extern Node *get_leftop(Expr *clause);
extern Node *get_rightop(Expr *clause);
-extern Expr *make_funcclause(Oid funcid, Oid funcresulttype, bool funcretset,
- CoercionForm funcformat, List *funcargs);
-
extern bool not_clause(Node *clause);
extern Expr *make_notclause(Expr *notclause);
extern Expr *get_notclausearg(Expr *notclause);
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $Id: rel.h,v 1.64 2002/11/23 03:59:09 momjian Exp $
+ * $Id: rel.h,v 1.65 2003/05/28 16:04:02 tgl Exp $
*
*-------------------------------------------------------------------------
*/
/* These are non-NULL only for an index relation: */
Form_pg_index rd_index; /* pg_index tuple describing this index */
+ struct HeapTupleData *rd_indextuple; /* all of pg_index tuple */
+ /* "struct HeapTupleData *" avoids need to include htup.h here */
Form_pg_am rd_am; /* pg_am tuple for index's AM */
/* index access support info (used only for an index relation) */
struct FmgrInfo *rd_supportinfo; /* lookup info for support
* procedures */
/* "struct FmgrInfo" avoids need to include fmgr.h here */
+ List *rd_indexprs; /* index expression trees, if any */
+ List *rd_indpred; /* index predicate tree, if any */
/* statistics collection area */
PgStat_Info pgstat_info;
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $Id: relcache.h,v 1.34 2002/08/06 02:36:35 tgl Exp $
+ * $Id: relcache.h,v 1.35 2003/05/28 16:04:02 tgl Exp $
*
*-------------------------------------------------------------------------
*/
* Routines to compute/retrieve additional cached information
*/
extern List *RelationGetIndexList(Relation relation);
+extern List *RelationGetIndexExpressions(Relation relation);
+extern List *RelationGetIndexPredicate(Relation relation);
extern void RelationInitIndexAccessInfo(Relation relation);
ic.relname AS index_name, a.attname
FROM pg_class bc, pg_class ic, pg_index i, pg_attribute a
WHERE i.indrelid = bc.oid AND i.indexrelid = bc.oid
- AND i.indkey[0] = a.attnum AND a.attrelid = bc.oid
- AND i.indproc = '0'::oid AND a.attisdropped = 'f'
+ AND i.indkey[0] = a.attnum AND i.indnatts = 1
+ AND a.attrelid = bc.oid AND a.attisdropped = 'f'
ORDER BY class_name, index_name, attname""")
return result
ERROR: Cannot insert a duplicate key into unique index func_index_index
-- but this shouldn't:
INSERT INTO func_index_heap VALUES('QWERTY');
+--
+-- Same test, expressional index
+--
+DROP TABLE func_index_heap;
+CREATE TABLE func_index_heap (f1 text, f2 text);
+CREATE UNIQUE INDEX func_index_index on func_index_heap ((f1 || f2) text_ops);
+INSERT INTO func_index_heap VALUES('ABC','DEF');
+INSERT INTO func_index_heap VALUES('AB','CDEFG');
+INSERT INTO func_index_heap VALUES('QWE','RTY');
+-- this should fail because of unique index:
+INSERT INTO func_index_heap VALUES('ABCD', 'EF');
+ERROR: Cannot insert a duplicate key into unique index func_index_index
+-- but this shouldn't:
+INSERT INTO func_index_heap VALUES('QWERTY');
WHERE relhasoids
AND ((nspname ~ '^pg_') IS NOT FALSE)
AND NOT EXISTS (SELECT 1 FROM pg_index i WHERE indrelid = c.oid
- AND indkey[0] = -2 AND indkey[1] = 0 AND indisunique);
+ AND indkey[0] = -2 AND indnatts = 1 AND indisunique);
relname | nspname
---------+---------
(0 rows)
INSERT INTO func_index_heap VALUES('ABCD', 'EF');
-- but this shouldn't:
INSERT INTO func_index_heap VALUES('QWERTY');
+
+
+--
+-- Same test, expressional index
+--
+DROP TABLE func_index_heap;
+CREATE TABLE func_index_heap (f1 text, f2 text);
+CREATE UNIQUE INDEX func_index_index on func_index_heap ((f1 || f2) text_ops);
+
+INSERT INTO func_index_heap VALUES('ABC','DEF');
+INSERT INTO func_index_heap VALUES('AB','CDEFG');
+INSERT INTO func_index_heap VALUES('QWE','RTY');
+-- this should fail because of unique index:
+INSERT INTO func_index_heap VALUES('ABCD', 'EF');
+-- but this shouldn't:
+INSERT INTO func_index_heap VALUES('QWERTY');
WHERE relhasoids
AND ((nspname ~ '^pg_') IS NOT FALSE)
AND NOT EXISTS (SELECT 1 FROM pg_index i WHERE indrelid = c.oid
- AND indkey[0] = -2 AND indkey[1] = 0 AND indisunique);
+ AND indkey[0] = -2 AND indnatts = 1 AND indisunique);
-- Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
-- Portions Copyright (c) 1994, Regents of the University of California
--
--- $Id: syscat.source,v 1.7 2002/06/20 20:29:54 momjian Exp $
+-- $Id: syscat.source,v 1.8 2003/05/28 16:04:02 tgl Exp $
--
---------------------------------------------------------------------------
--
--- lists all simple indices (ie. those that are not defined over a function
--- of several attributes)
+-- lists all simple indices (ie. those that are defined over one simple
+-- column reference)
--
SELECT bc.relname AS class_name,
ic.relname AS index_name,
WHERE i.indrelid = bc.oid
and i.indexrelid = ic.oid
and i.indkey[0] = a.attnum
+ and i.indnatts = 1
and a.attrelid = bc.oid
- and i.indproc = '0'::oid -- no functional indices
ORDER BY class_name, index_name, attname;
projects / postgresql.git / commitdiff