postgresql/doc/src/sgml/trigger.sgml

<!--
$Header: /cvsroot/pgsql/doc/src/sgml/trigger.sgml,v 1.25 2002/09/21 18:32:54 petere Exp $
-->

 <chapter id="triggers">
  <title>Triggers</title>

  <para>
   <productname>PostgreSQL</productname> has various server-side
   function interfaces. Server-side functions can be written in SQL,
   C, or any defined procedural language. Trigger functions can be
   written in C and most procedural languages, but not in SQL. Note that
   statement-level trigger events are not supported in the current
   version. You can currently specify BEFORE or AFTER on INSERT,
   DELETE or UPDATE of a tuple as a trigger event.
  </para>

  <sect1 id="trigger-definition">
   <title>Trigger Definition</title>

   <para>
    If a trigger event occurs, the trigger manager (called by the Executor)
    sets up a <structname>TriggerData</> information structure (described below) and calls
    the trigger function to handle the event.
   </para>

   <para>
    The trigger function must be defined before the trigger itself can be
    created.  The trigger function must be declared as a 
    function taking no arguments and returning type <literal>trigger</>.
    (The trigger function receives its input through a <structname>TriggerData</>
    structure, not in the form of ordinary function arguments.)
    If the function is written in C, it must use the <quote>version 1</>
    function manager interface.
   </para>

   <para>
    The syntax for creating triggers is:

<programlisting>
CREATE TRIGGER <replaceable>trigger</replaceable> [ BEFORE | AFTER ] [ INSERT | DELETE | UPDATE [ OR ... ] ]
    ON <replaceable>relation</replaceable> FOR EACH [ ROW | STATEMENT ]
    EXECUTE PROCEDURE <replaceable>procedure</replaceable>
     (<replaceable>args</replaceable>);
</programlisting>

    where the arguments are:

    <variablelist>
     <varlistentry>
      <term>
       <replaceable>trigger</replaceable>
      </term>
      <listitem>
       <para>
        The trigger must have a name distinct from all other triggers on
	the same table.  The name is needed
	if you ever have to delete the trigger.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term>BEFORE</term>
      <term>AFTER</term>
      <listitem>
       <para>
	Determines whether the function is called before or after
	the event.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term>INSERT</term>
      <term>DELETE</term>
      <term>UPDATE</term>
      <listitem>
       <para>
	The next element of the command determines what event(s) will trigger
	the function.  Multiple events can be specified separated by OR.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term><replaceable>relation</replaceable></term>
      <listitem>
       <para>
	The relation name indicates which table the event applies to.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term>ROW</term>
      <term>STATEMENT</term>
      <listitem>
       <para>
	The FOR EACH clause determines whether the trigger is fired for each
	affected row or before (or after) the entire statement has completed.
	Currently only the ROW case is supported.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term><replaceable>procedure</replaceable></term>
      <listitem>
       <para>
	The procedure name is the function to be called.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term><replaceable>args</replaceable></term>
      <listitem>
       <para>
	The arguments passed to the function in the <structname>TriggerData</> structure.
	This is either empty or a list of one or more simple literal
	constants (which will be passed to the function as strings).
       </para>

       <para>
	The purpose of including arguments in the trigger definition
	is to allow different
	triggers with similar requirements to call the same function.
	As an example, there could be a generalized trigger
	function that takes as its arguments two field names and puts the
	current user in one and the current time stamp in the other.
	Properly written, this trigger function would be independent of
	the specific table it is triggering on.  So the same function
	could be used for INSERT events on any table with suitable fields,
	to automatically track creation of records in a transaction table for
	example. It could also be used to track last-update events if
	defined as an UPDATE trigger.
       </para>
      </listitem>
     </varlistentry>
    </variablelist>
   </para>

   <para>
    Trigger functions return a <structname>HeapTuple</> to the calling executor.  The return
    value is ignored for triggers fired AFTER an operation,
    but it allows BEFORE triggers to:

    <itemizedlist>
     <listitem>
      <para>
       Return a <symbol>NULL</> pointer to skip the operation for the
       current tuple (and so the tuple will not be
       inserted/updated/deleted).
      </para>
     </listitem>

     <listitem>
      <para>
       For INSERT and UPDATE triggers only, the returned tuple becomes the
       tuple which will be inserted or will replace the tuple being updated.
       This allows the trigger function to modify the row being inserted or
       updated.
      </para>
     </listitem>
    </itemizedlist>

    A BEFORE trigger that does not intend to cause either of these behaviors
    must be careful to return the same NEW tuple it is passed.
   </para>

   <para>
    Note that there is no initialization performed by the CREATE TRIGGER
    handler.  This may be changed in the future.
   </para>

   <para>
    If more than one trigger is defined for the same event on the same
    relation, the triggers will be fired in alphabetical order by
    name.  In the case of BEFORE triggers, the possibly-modified tuple
    returned by each trigger becomes the input to the next trigger.
    If any BEFORE trigger returns <symbol>NULL</>, the operation is
    abandoned and subsequent triggers are not fired.
   </para>

   <para>
    If a trigger function executes SQL-queries (using SPI) then these queries
    may fire triggers again. This is known as cascading triggers.  There is no
    direct limitation on the number of cascade levels.  It is possible for
    cascades to cause recursive invocation of the same trigger --- for
    example, an INSERT trigger might execute a query that inserts an
    additional tuple into the same table, causing the INSERT trigger to be
    fired again.  It is the trigger programmer's
    responsibility to avoid infinite recursion in such scenarios.
   </para>
  </sect1>

  <sect1 id="trigger-manager">
   <title>Interaction with the Trigger Manager</title>

   <para>
    This section describes the low-level details of the interface to a
    trigger function.  This information is only needed when writing a
    trigger function in C.  If you are using a higher-level function
    language then these details are handled for you.
   </para>

    <note>
     <para>
      The interface described here applies for
      <productname>PostgreSQL</productname> 7.1 and later.
      Earlier versions passed the <structname>TriggerData</> pointer in a global
      variable <varname>CurrentTriggerData</>.
     </para>
    </note>

   <para>
    When a function is called by the trigger manager, it is not passed any
    normal parameters, but it is passed a <quote>context</> pointer pointing to a
    <structname>TriggerData</> structure.  C functions can check whether they were called
    from the trigger manager or not by executing the macro
    <literal>CALLED_AS_TRIGGER(fcinfo)</literal>, which expands to
<programlisting>
((fcinfo)->context != NULL && IsA((fcinfo)->context, TriggerData))
</programlisting>
    If this returns true, then it is safe to cast <literal>fcinfo->context</> to type
    <literal>TriggerData *</literal> and make use of the pointed-to
    <structname>TriggerData</> structure.
    The function must <emphasis>not</emphasis> alter the <structname>TriggerData</>
    structure or any of the data it points to.
   </para>

   <para>
    <structname>struct TriggerData</structname> is defined in
    <filename>commands/trigger.h</filename>:

<programlisting>
typedef struct TriggerData
{
    NodeTag       type;
    TriggerEvent  tg_event;
    Relation      tg_relation;
    HeapTuple     tg_trigtuple;
    HeapTuple     tg_newtuple;
    Trigger      *tg_trigger;
} TriggerData;
</programlisting>

    where the members are defined as follows:

    <variablelist>
     <varlistentry>
      <term><structfield>type</></term>
      <listitem>
       <para>
        Always <literal>T_TriggerData</literal> if this is a trigger event.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term><structfield>tg_event</></term>
      <listitem>
       <para>
	describes the event for which the function is called. You may use the
	following macros to examine <literal>tg_event</literal>:

	<variablelist>
	 <varlistentry>
	  <term>TRIGGER_FIRED_BEFORE(tg_event)</term>
	  <listitem>
	   <para>
	    returns TRUE if trigger fired BEFORE.
	   </para>
	  </listitem>
	 </varlistentry>

	 <varlistentry>
	  <term>TRIGGER_FIRED_AFTER(tg_event)</term>
	  <listitem>
	   <para>
	    Returns TRUE if trigger fired AFTER.
	   </para>
	  </listitem>
	 </varlistentry>

	 <varlistentry>
	  <term>TRIGGER_FIRED_FOR_ROW(event)</term>
	  <listitem>
	   <para>
	    Returns TRUE if trigger fired for
	    a ROW-level event.
	   </para>
	  </listitem>
	 </varlistentry>

	 <varlistentry>
	  <term>TRIGGER_FIRED_FOR_STATEMENT(event)</term>
	  <listitem>
	   <para>
	    Returns TRUE if trigger fired for
	    STATEMENT-level event.
	   </para>
	  </listitem>
	 </varlistentry>

	 <varlistentry>
	  <term>TRIGGER_FIRED_BY_INSERT(event)</term>
	  <listitem>
	   <para>
	    Returns TRUE if trigger fired by INSERT.
	   </para>
	  </listitem>
	 </varlistentry>

	 <varlistentry>
	  <term>TRIGGER_FIRED_BY_DELETE(event)</term>
	  <listitem>
	   <para>
	    Returns TRUE if trigger fired by DELETE.
	   </para>
	  </listitem>
	 </varlistentry>

	 <varlistentry>
	  <term>TRIGGER_FIRED_BY_UPDATE(event)</term>
	  <listitem>
	   <para>
	    Returns TRUE if trigger fired by UPDATE.
	   </para>
	  </listitem>
	 </varlistentry>
	</variablelist>
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term><structfield>tg_relation</></term>
      <listitem>
       <para>
	is a pointer to structure describing the triggered
	relation. Look at <filename>utils/rel.h</> for details about
	this structure.  The most interesting things are
	<literal>tg_relation->rd_att</> (descriptor of the relation
	tuples) and <literal>tg_relation->rd_rel->relname</>
	(relation's name. This is not <type>char*</>, but
	<type>NameData</>.  Use
	<literal>SPI_getrelname(tg_relation)</> to get <type>char*</> if you
	need a copy of the name).
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term><structfield>tg_trigtuple</></term>
      <listitem>
       <para>
	is a pointer to the tuple for which the trigger is fired. This is the tuple
	being inserted (if INSERT), deleted (if DELETE) or updated (if UPDATE).
	If INSERT/DELETE then this is what you are to return to Executor if 
	you don't want to replace tuple with another one (INSERT) or skip the
	operation.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term><structfield>tg_newtuple</></term>
      <listitem>
       <para>
	is a pointer to the new version of tuple if UPDATE and <symbol>NULL</> if this is
	for an INSERT or a DELETE. This is what you are to return to Executor if
	UPDATE and you don't want to replace this tuple with another one or skip
	the operation.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term><structfield>tg_trigger</></term>
      <listitem>
       <para>
	is pointer to structure <structname>Trigger</> defined in <filename>utils/rel.h</>:

<programlisting>
typedef struct Trigger
{
    Oid         tgoid;
    char       *tgname;
    Oid         tgfoid;
    int16       tgtype;
    bool        tgenabled;
    bool        tgisconstraint;
    Oid         tgconstrrelid;
    bool        tgdeferrable;
    bool        tginitdeferred;
    int16       tgnargs;
    int16       tgattr[FUNC_MAX_ARGS];
    char      **tgargs;
} Trigger;
</programlisting>

       where <structfield>tgname</> is the trigger's name,
       <structfield>tgnargs</> is number of arguments in
       <structfield>tgargs</>, <structfield>tgargs</> is an array of
       pointers to the arguments specified in the CREATE TRIGGER
       statement. Other members are for internal use only.
       </para>
      </listitem>
     </varlistentry>
    </variablelist>
   </para>
  </sect1>

  <sect1 id="trigger-datachanges">
   <title>Visibility of Data Changes</title>

   <para>
    <productname>PostgreSQL</productname> data changes visibility rule: during a query execution, data
    changes made by the query itself (via SQL-function, SPI-function, triggers)
    are invisible to the query scan.  For example, in query

<programlisting>
INSERT INTO a SELECT * FROM a;
</programlisting>

    tuples inserted are invisible for SELECT scan.  In effect, this
    duplicates the database table within itself (subject to unique index
    rules, of course) without recursing.
   </para>

   <para>
    But keep in mind this notice about visibility in the SPI documentation:

    <blockquote>
     <para>
Changes made by query Q are visible by queries that are started after
query Q, no matter whether they are started inside Q (during the
execution of Q) or after Q is done.
     </para>
    </blockquote>
   </para>

   <para>
    This is true for triggers as well so, though a tuple being inserted
    (<structfield>tg_trigtuple</>) is not visible to queries in a BEFORE trigger, this tuple
    (just inserted) is visible to queries in an AFTER trigger, and to queries
    in BEFORE/AFTER triggers fired after this!
   </para>
  </sect1>

  <sect1 id="trigger-examples">
   <title>Examples</title>

   <para>
    There are more complex examples in
    <filename>src/test/regress/regress.c</filename> and
    in <filename>contrib/spi</filename>.
   </para>

   <para>
    Here is a very simple example of trigger usage.  Function <function>trigf</> reports
    the number of tuples in the triggered relation <literal>ttest</> and skips the
    operation if the query attempts to insert a null value into x (i.e - it acts as a
    not-null constraint but doesn't abort the transaction).

<programlisting>
#include "executor/spi.h"       /* this is what you need to work with SPI */
#include "commands/trigger.h"   /* -"- and triggers */

extern Datum trigf(PG_FUNCTION_ARGS);

PG_FUNCTION_INFO_V1(trigf);

Datum
trigf(PG_FUNCTION_ARGS)
{
    TriggerData *trigdata = (TriggerData *) fcinfo->context;
    TupleDesc   tupdesc;
    HeapTuple   rettuple;
    char       *when;
    bool        checknull = false;
    bool        isnull;
    int         ret, i;

    /* Make sure trigdata is pointing at what I expect */
    if (!CALLED_AS_TRIGGER(fcinfo))
        elog(ERROR, "trigf: not fired by trigger manager");

    /* tuple to return to Executor */
    if (TRIGGER_FIRED_BY_UPDATE(trigdata->tg_event))
        rettuple = trigdata->tg_newtuple;
    else
        rettuple = trigdata->tg_trigtuple;

    /* check for null values */
    if (!TRIGGER_FIRED_BY_DELETE(trigdata->tg_event)
        && TRIGGER_FIRED_BEFORE(trigdata->tg_event))
        checknull = true;

    if (TRIGGER_FIRED_BEFORE(trigdata->tg_event))
        when = "before";
    else
        when = "after ";

    tupdesc = trigdata->tg_relation->rd_att;

    /* Connect to SPI manager */
    if ((ret = SPI_connect()) < 0)
        elog(INFO, "trigf (fired %s): SPI_connect returned %d", when, ret);

    /* Get number of tuples in relation */
    ret = SPI_exec("SELECT count(*) FROM ttest", 0);

    if (ret < 0)
        elog(NOTICE, "trigf (fired %s): SPI_exec returned %d", when, ret);

    /* count(*) returns int8 as of PG 7.2, so be careful to convert */
    i = (int) DatumGetInt64(SPI_getbinval(SPI_tuptable->vals[0],
                                          SPI_tuptable->tupdesc,
                                          1,
                                          &amp;isnull));

    elog (NOTICE, "trigf (fired %s): there are %d tuples in ttest", when, i);

    SPI_finish();

    if (checknull)
    {
        (void) SPI_getbinval(rettuple, tupdesc, 1, &amp;isnull);
        if (isnull)
            rettuple = NULL;
    }

    return PointerGetDatum(rettuple);
}
</programlisting>
   </para>

   <para>
    Now, compile and create the trigger function:

<programlisting>
CREATE FUNCTION trigf () RETURNS TRIGGER AS 
'...path_to_so' LANGUAGE C;

CREATE TABLE ttest (x int4);
</programlisting>

<programlisting>
vac=> CREATE TRIGGER tbefore BEFORE INSERT OR UPDATE OR DELETE ON ttest 
FOR EACH ROW EXECUTE PROCEDURE trigf();
CREATE
vac=> CREATE TRIGGER tafter AFTER INSERT OR UPDATE OR DELETE ON ttest 
FOR EACH ROW EXECUTE PROCEDURE trigf();
CREATE
vac=> INSERT INTO ttest VALUES (NULL);
WARNING:  trigf (fired before): there are 0 tuples in ttest
INSERT 0 0

-- Insertion skipped and AFTER trigger is not fired

vac=> SELECT * FROM ttest;
 x
---
(0 rows)

vac=> INSERT INTO ttest VALUES (1);
INFO:  trigf (fired before): there are 0 tuples in ttest
INFO:  trigf (fired after ): there are 1 tuples in ttest
                                       ^^^^^^^^
                             remember what we said about visibility.
INSERT 167793 1
vac=> SELECT * FROM ttest;
 x
---
 1
(1 row)

vac=> INSERT INTO ttest SELECT x * 2 FROM ttest;
INFO:  trigf (fired before): there are 1 tuples in ttest
INFO:  trigf (fired after ): there are 2 tuples in ttest
                                       ^^^^^^^^
                             remember what we said about visibility.
INSERT 167794 1
vac=> SELECT * FROM ttest;
 x
---
 1
 2
(2 rows)

vac=> UPDATE ttest SET x = NULL WHERE x = 2;
INFO:  trigf (fired before): there are 2 tuples in ttest
UPDATE 0
vac=> UPDATE ttest SET x = 4 WHERE x = 2;
INFO:  trigf (fired before): there are 2 tuples in ttest
INFO:  trigf (fired after ): there are 2 tuples in ttest
UPDATE 1
vac=> SELECT * FROM ttest;
 x
---
 1
 4
(2 rows)

vac=> DELETE FROM ttest;
INFO:  trigf (fired before): there are 2 tuples in ttest
INFO:  trigf (fired after ): there are 1 tuples in ttest
INFO:  trigf (fired before): there are 1 tuples in ttest
INFO:  trigf (fired after ): there are 0 tuples in ttest
                                       ^^^^^^^^
                             remember what we said about visibility.
DELETE 2
vac=> SELECT * FROM ttest;
 x
---
(0 rows)
</programlisting>

   </para>
  </sect1>
 </chapter>

<!-- Keep this comment at the end of the file
Local variables:
mode:sgml
sgml-omittag:nil
sgml-shorttag:t
sgml-minimize-attributes:nil
sgml-always-quote-attributes:t
sgml-indent-step:1
sgml-indent-data:t
sgml-parent-document:nil
sgml-default-dtd-file:"./reference.ced"
sgml-exposed-tags:nil
sgml-local-catalogs:("/usr/lib/sgml/catalog")
sgml-local-ecat-files:nil
End:
-->