September 26, 2024: PostgreSQL 17 Released!
Supported Versions: Current (17) / 16 / 15 / 14 / 13 / 12
Development Versions: devel
Unsupported versions: 11 / 10 / 9.6 / 9.5 / 9.4 / 9.3 / 9.2 / 9.1 / 9.0 / 8.4 / 8.3 / 8.2 / 8.1

62.2. Index Access Method Functions

The index construction and maintenance functions that an index access method must provide in IndexAmRoutine are:

IndexBuildResult *
ambuild (Relation heapRelation,
         Relation indexRelation,
         IndexInfo *indexInfo);

Build a new index. The index relation has been physically created, but is empty. It must be filled in with whatever fixed data the access method requires, plus entries for all tuples already existing in the table. Ordinarily the ambuild function will call table_index_build_scan() to scan the table for existing tuples and compute the keys that need to be inserted into the index. The function must return a palloc'd struct containing statistics about the new index.

void
ambuildempty (Relation indexRelation);

Build an empty index, and write it to the initialization fork (INIT_FORKNUM) of the given relation. This method is called only for unlogged indexes; the empty index written to the initialization fork will be copied over the main relation fork on each server restart.

bool
aminsert (Relation indexRelation,
          Datum *values,
          bool *isnull,
          ItemPointer heap_tid,
          Relation heapRelation,
          IndexUniqueCheck checkUnique,
          bool indexUnchanged,
          IndexInfo *indexInfo);

Insert a new tuple into an existing index. The values and isnull arrays give the key values to be indexed, and heap_tid is the TID to be indexed. If the access method supports unique indexes (its amcanunique flag is true) then checkUnique indicates the type of uniqueness check to perform. This varies depending on whether the unique constraint is deferrable; see Section 62.5 for details. Normally the access method only needs the heapRelation parameter when performing uniqueness checking (since then it will have to look into the heap to verify tuple liveness).

The indexUnchanged Boolean value gives a hint about the nature of the tuple to be indexed. When it is true, the tuple is a duplicate of some existing tuple in the index. The new tuple is a logically unchanged successor MVCC tuple version. This happens when an UPDATE takes place that does not modify any columns covered by the index, but nevertheless requires a new version in the index. The index AM may use this hint to decide to apply bottom-up index deletion in parts of the index where many versions of the same logical row accumulate. Note that updating a non-key column or a column that only appears in a partial index predicate does not affect the value of indexUnchanged. The core code determines each tuple's indexUnchanged value using a low overhead approach that allows both false positives and false negatives. Index AMs must not treat indexUnchanged as an authoritative source of information about tuple visibility or versioning.

The function's Boolean result value is significant only when checkUnique is UNIQUE_CHECK_PARTIAL. In this case a true result means the new entry is known unique, whereas false means it might be non-unique (and a deferred uniqueness check must be scheduled). For other cases a constant false result is recommended.

Some indexes might not index all tuples. If the tuple is not to be indexed, aminsert should just return without doing anything.

If the index AM wishes to cache data across successive index insertions within an SQL statement, it can allocate space in indexInfo->ii_Context and store a pointer to the data in indexInfo->ii_AmCache (which will be NULL initially).

IndexBulkDeleteResult *
ambulkdelete (IndexVacuumInfo *info,
              IndexBulkDeleteResult *stats,
              IndexBulkDeleteCallback callback,
              void *callback_state);

Delete tuple(s) from the index. This is a bulk delete operation that is intended to be implemented by scanning the whole index and checking each entry to see if it should be deleted. The passed-in callback function must be called, in the style callback(TID, callback_state) returns bool, to determine whether any particular index entry, as identified by its referenced TID, is to be deleted. Must return either NULL or a palloc'd struct containing statistics about the effects of the deletion operation. It is OK to return NULL if no information needs to be passed on to amvacuumcleanup.

Because of limited maintenance_work_mem, ambulkdelete might need to be called more than once when many tuples are to be deleted. The stats argument is the result of the previous call for this index (it is NULL for the first call within a VACUUM operation). This allows the AM to accumulate statistics across the whole operation. Typically, ambulkdelete will modify and return the same struct if the passed stats is not null.

IndexBulkDeleteResult *
amvacuumcleanup (IndexVacuumInfo *info,
                 IndexBulkDeleteResult *stats);

Clean up after a VACUUM operation (zero or more ambulkdelete calls). This does not have to do anything beyond returning index statistics, but it might perform bulk cleanup such as reclaiming empty index pages. stats is whatever the last ambulkdelete call returned, or NULL if ambulkdelete was not called because no tuples needed to be deleted. If the result is not NULL it must be a palloc'd struct. The statistics it contains will be used to update pg_class, and will be reported by VACUUM if VERBOSE is given. It is OK to return NULL if the index was not changed at all during the VACUUM operation, but otherwise correct stats should be returned.

amvacuumcleanup will also be called at completion of an ANALYZE operation. In this case stats is always NULL and any return value will be ignored. This case can be distinguished by checking info->analyze_only. It is recommended that the access method do nothing except post-insert cleanup in such a call, and that only in an autovacuum worker process.

bool
amcanreturn (Relation indexRelation, int attno);

Check whether the index can support index-only scans on the given column, by returning the column's original indexed value. The attribute number is 1-based, i.e., the first column's attno is 1. Returns true if supported, else false. This function should always return true for included columns (if those are supported), since there's little point in an included column that can't be retrieved. If the access method does not support index-only scans at all, the amcanreturn field in its IndexAmRoutine struct can be set to NULL.

void
amcostestimate (PlannerInfo *root,
                IndexPath *path,
                double loop_count,
                Cost *indexStartupCost,
                Cost *indexTotalCost,
                Selectivity *indexSelectivity,
                double *indexCorrelation,
                double *indexPages);

Estimate the costs of an index scan. This function is described fully in Section 62.6, below.

bytea *
amoptions (ArrayType *reloptions,
           bool validate);

Parse and validate the reloptions array for an index. This is called only when a non-null reloptions array exists for the index. reloptions is a text array containing entries of the form name=value. The function should construct a bytea value, which will be copied into the rd_options field of the index's relcache entry. The data contents of the bytea value are open for the access method to define; most of the standard access methods use struct StdRdOptions. When validate is true, the function should report a suitable error message if any of the options are unrecognized or have invalid values; when validate is false, invalid entries should be silently ignored. (validate is false when loading options already stored in pg_catalog; an invalid entry could only be found if the access method has changed its rules for options, and in that case ignoring obsolete entries is appropriate.) It is OK to return NULL if default behavior is wanted.

bool
amproperty (Oid index_oid, int attno,
            IndexAMProperty prop, const char *propname,
            bool *res, bool *isnull);

The amproperty method allows index access methods to override the default behavior of pg_index_column_has_property and related functions. If the access method does not have any special behavior for index property inquiries, the amproperty field in its IndexAmRoutine struct can be set to NULL. Otherwise, the amproperty method will be called with index_oid and attno both zero for pg_indexam_has_property calls, or with index_oid valid and attno zero for pg_index_has_property calls, or with index_oid valid and attno greater than zero for pg_index_column_has_property calls. prop is an enum value identifying the property being tested, while propname is the original property name string. If the core code does not recognize the property name then prop is AMPROP_UNKNOWN. Access methods can define custom property names by checking propname for a match (use pg_strcasecmp to match, for consistency with the core code); for names known to the core code, it's better to inspect prop. If the amproperty method returns true then it has determined the property test result: it must set *res to the Boolean value to return, or set *isnull to true to return a NULL. (Both of the referenced variables are initialized to false before the call.) If the amproperty method returns false then the core code will proceed with its normal logic for determining the property test result.

Access methods that support ordering operators should implement AMPROP_DISTANCE_ORDERABLE property testing, as the core code does not know how to do that and will return NULL. It may also be advantageous to implement AMPROP_RETURNABLE testing, if that can be done more cheaply than by opening the index and calling amcanreturn, which is the core code's default behavior. The default behavior should be satisfactory for all other standard properties.

char *
ambuildphasename (int64 phasenum);

Return the textual name of the given build phase number. The phase numbers are those reported during an index build via the pgstat_progress_update_param interface. The phase names are then exposed in the pg_stat_progress_create_index view.

bool
amvalidate (Oid opclassoid);

Validate the catalog entries for the specified operator class, so far as the access method can reasonably do that. For example, this might include testing that all required support functions are provided. The amvalidate function must return false if the opclass is invalid. Problems should be reported with ereport messages, typically at INFO level.

void
amadjustmembers (Oid opfamilyoid,
                 Oid opclassoid,
                 List *operators,
                 List *functions);

Validate proposed new operator and function members of an operator family, so far as the access method can reasonably do that, and set their dependency types if the default is not satisfactory. This is called during CREATE OPERATOR CLASS and during ALTER OPERATOR FAMILY ADD; in the latter case opclassoid is InvalidOid. The List arguments are lists of OpFamilyMember structs, as defined in amapi.h. Tests done by this function will typically be a subset of those performed by amvalidate, since amadjustmembers cannot assume that it is seeing a complete set of members. For example, it would be reasonable to check the signature of a support function, but not to check whether all required support functions are provided. Any problems can be reported by throwing an error. The dependency-related fields of the OpFamilyMember structs are initialized by the core code to create hard dependencies on the opclass if this is CREATE OPERATOR CLASS, or soft dependencies on the opfamily if this is ALTER OPERATOR FAMILY ADD. amadjustmembers can adjust these fields if some other behavior is more appropriate. For example, GIN, GiST, and SP-GiST always set operator members to have soft dependencies on the opfamily, since the connection between an operator and an opclass is relatively weak in these index types; so it is reasonable to allow operator members to be added and removed freely. Optional support functions are typically also given soft dependencies, so that they can be removed if necessary.

The purpose of an index, of course, is to support scans for tuples matching an indexable WHERE condition, often called a qualifier or scan key. The semantics of index scanning are described more fully in Section 62.3, below. An index access method can support plain index scans, bitmap index scans, or both. The scan-related functions that an index access method must or may provide are:

IndexScanDesc
ambeginscan (Relation indexRelation,
             int nkeys,
             int norderbys);

Prepare for an index scan. The nkeys and norderbys parameters indicate the number of quals and ordering operators that will be used in the scan; these may be useful for space allocation purposes. Note that the actual values of the scan keys aren't provided yet. The result must be a palloc'd struct. For implementation reasons the index access method must create this struct by calling RelationGetIndexScan(). In most cases ambeginscan does little beyond making that call and perhaps acquiring locks; the interesting parts of index-scan startup are in amrescan.

void
amrescan (IndexScanDesc scan,
          ScanKey keys,
          int nkeys,
          ScanKey orderbys,
          int norderbys);

Start or restart an index scan, possibly with new scan keys. (To restart using previously-passed keys, NULL is passed for keys and/or orderbys.) Note that it is not allowed for the number of keys or order-by operators to be larger than what was passed to ambeginscan. In practice the restart feature is used when a new outer tuple is selected by a nested-loop join and so a new key comparison value is needed, but the scan key structure remains the same.

bool
amgettuple (IndexScanDesc scan,
            ScanDirection direction);

Fetch the next tuple in the given scan, moving in the given direction (forward or backward in the index). Returns true if a tuple was obtained, false if no matching tuples remain. In the true case the tuple TID is stored into the scan structure. Note that success means only that the index contains an entry that matches the scan keys, not that the tuple necessarily still exists in the heap or will pass the caller's snapshot test. On success, amgettuple must also set scan->xs_recheck to true or false. False means it is certain that the index entry matches the scan keys. True means this is not certain, and the conditions represented by the scan keys must be rechecked against the heap tuple after fetching it. This provision supports lossy index operators. Note that rechecking will extend only to the scan conditions; a partial index predicate (if any) is never rechecked by amgettuple callers.

If the index supports index-only scans (i.e., amcanreturn returns true for any of its columns), then on success the AM must also check scan->xs_want_itup, and if that is true it must return the originally indexed data for the index entry. Columns for which amcanreturn returns false can be returned as nulls. The data can be returned in the form of an IndexTuple pointer stored at scan->xs_itup, with tuple descriptor scan->xs_itupdesc; or in the form of a HeapTuple pointer stored at scan->xs_hitup, with tuple descriptor scan->xs_hitupdesc. (The latter format should be used when reconstructing data that might possibly not fit into an IndexTuple.) In either case, management of the data referenced by the pointer is the access method's responsibility. The data must remain good at least until the next amgettuple, amrescan, or amendscan call for the scan.

The amgettuple function need only be provided if the access method supports plain index scans. If it doesn't, the amgettuple field in its IndexAmRoutine struct must be set to NULL.

int64
amgetbitmap (IndexScanDesc scan,
             TIDBitmap *tbm);

Fetch all tuples in the given scan and add them to the caller-supplied TIDBitmap (that is, OR the set of tuple IDs into whatever set is already in the bitmap). The number of tuples fetched is returned (this might be just an approximate count, for instance some AMs do not detect duplicates). While inserting tuple IDs into the bitmap, amgetbitmap can indicate that rechecking of the scan conditions is required for specific tuple IDs. This is analogous to the xs_recheck output parameter of amgettuple. Note: in the current implementation, support for this feature is conflated with support for lossy storage of the bitmap itself, and therefore callers recheck both the scan conditions and the partial index predicate (if any) for recheckable tuples. That might not always be true, however. amgetbitmap and amgettuple cannot be used in the same index scan; there are other restrictions too when using amgetbitmap, as explained in Section 62.3.

The amgetbitmap function need only be provided if the access method supports bitmap index scans. If it doesn't, the amgetbitmap field in its IndexAmRoutine struct must be set to NULL.

void
amendscan (IndexScanDesc scan);

End a scan and release resources. The scan struct itself should not be freed, but any locks or pins taken internally by the access method must be released, as well as any other memory allocated by ambeginscan and other scan-related functions.

void
ammarkpos (IndexScanDesc scan);

Mark current scan position. The access method need only support one remembered scan position per scan.

The ammarkpos function need only be provided if the access method supports ordered scans. If it doesn't, the ammarkpos field in its IndexAmRoutine struct may be set to NULL.

void
amrestrpos (IndexScanDesc scan);

Restore the scan to the most recently marked position.

The amrestrpos function need only be provided if the access method supports ordered scans. If it doesn't, the amrestrpos field in its IndexAmRoutine struct may be set to NULL.

In addition to supporting ordinary index scans, some types of index may wish to support parallel index scans, which allow multiple backends to cooperate in performing an index scan. The index access method should arrange things so that each cooperating process returns a subset of the tuples that would be performed by an ordinary, non-parallel index scan, but in such a way that the union of those subsets is equal to the set of tuples that would be returned by an ordinary, non-parallel index scan. Furthermore, while there need not be any global ordering of tuples returned by a parallel scan, the ordering of that subset of tuples returned within each cooperating backend must match the requested ordering. The following functions may be implemented to support parallel index scans:

Size
amestimateparallelscan (void);

Estimate and return the number of bytes of dynamic shared memory which the access method will be needed to perform a parallel scan. (This number is in addition to, not in lieu of, the amount of space needed for AM-independent data in ParallelIndexScanDescData.)

It is not necessary to implement this function for access methods which do not support parallel scans or for which the number of additional bytes of storage required is zero.

void
aminitparallelscan (void *target);

This function will be called to initialize dynamic shared memory at the beginning of a parallel scan. target will point to at least the number of bytes previously returned by amestimateparallelscan, and this function may use that amount of space to store whatever data it wishes.

It is not necessary to implement this function for access methods which do not support parallel scans or in cases where the shared memory space required needs no initialization.

void
amparallelrescan (IndexScanDesc scan);

This function, if implemented, will be called when a parallel index scan must be restarted. It should reset any shared state set up by aminitparallelscan such that the scan will be restarted from the beginning.

Submit correction

If you see anything in the documentation that is not correct, does not match your experience with the particular feature or requires further clarification, please use this form to report a documentation issue.