path: root/doc/src/sgml/pltcl.sgml

<!-- doc/src/sgml/pltcl.sgml -->

 <chapter id="pltcl">
  <title>PL/Tcl &mdash; Tcl Procedural Language</title>

  <indexterm zone="pltcl">
   <primary>PL/Tcl</primary>
  </indexterm>

  <indexterm zone="pltcl">
   <primary>Tcl</primary>
  </indexterm>

  <para>
   PL/Tcl is a loadable procedural language for the
   <productname>PostgreSQL</productname> database system
   that enables the <ulink url="https://www.tcl.tk/">
   Tcl language</ulink> to be used to write
   <productname>PostgreSQL</productname> functions and procedures.
  </para>

  <!-- **** PL/Tcl overview **** -->

  <sect1 id="pltcl-overview">
   <title>Overview</title>

   <para>
    PL/Tcl offers most of the capabilities a function writer has in
    the C language, with a few restrictions, and with the addition of
    the powerful string processing libraries that are available for
    Tcl.
   </para>
   <para>
    One compelling <emphasis>good</emphasis> restriction is that
    everything is executed from within the safety of the context of a
    Tcl interpreter.  In addition to the limited command set of safe
    Tcl, only a few commands are available to access the database via
    SPI and to raise messages via <function>elog()</function>.  PL/Tcl
    provides no way to access internals of the database server or to
    gain OS-level access under the permissions of the
    <productname>PostgreSQL</productname> server process, as a C
    function can do.  Thus, unprivileged database users can be trusted
    to use this language; it does not give them unlimited authority.
   </para>
   <para>
    The other notable implementation restriction is that Tcl functions
    cannot be used to create input/output functions for new data
    types.
   </para>
   <para>
    Sometimes it is desirable to write Tcl functions that are not restricted
    to safe Tcl.  For example, one might want a Tcl function that sends
    email.  To handle these cases, there is a variant of <application>PL/Tcl</application> called <literal>PL/TclU</literal>
    (for untrusted Tcl).  This is exactly the same language except that a full
    Tcl interpreter is used.  <emphasis>If <application>PL/TclU</application> is used, it must be
    installed as an untrusted procedural language</emphasis> so that only
    database superusers can create functions in it.  The writer of a <application>PL/TclU</application>
    function must take care that the function cannot be used to do anything
    unwanted, since it will be able to do anything that could be done by
    a user logged in as the database administrator.
   </para>
   <para>
    The shared object code for the <application>PL/Tcl</application> and
    <application>PL/TclU</application> call handlers is automatically built and
    installed in the <productname>PostgreSQL</productname> library
    directory if Tcl support is specified in the configuration step of
    the installation procedure.  To install <application>PL/Tcl</application>
    and/or <application>PL/TclU</application> in a particular database, use the
    <command>CREATE EXTENSION</command> command, for example
    <literal>CREATE EXTENSION pltcl</literal> or
    <literal>CREATE EXTENSION pltclu</literal>.
   </para>
  </sect1>

  <!-- **** PL/Tcl description **** -->

   <sect1 id="pltcl-functions">
    <title>PL/Tcl Functions and Arguments</title>

    <para>
     To create a function in the <application>PL/Tcl</application> language, use
     the standard <xref linkend="sql-createfunction"/> syntax:

<programlisting>
CREATE FUNCTION <replaceable>funcname</replaceable> (<replaceable>argument-types</replaceable>) RETURNS <replaceable>return-type</replaceable> AS $$
    # PL/Tcl function body
$$ LANGUAGE pltcl;
</programlisting>

     <application>PL/TclU</application> is the same, except that the language has to be specified as
     <literal>pltclu</literal>.
    </para>

    <para>
     The body of the function is simply a piece of Tcl script.
     When the function is called, the argument values are passed to the
     Tcl script as variables named <literal>1</literal>
     ... <literal><replaceable>n</replaceable></literal>.  The result is
     returned from the Tcl code in the usual way, with
     a <literal>return</literal> statement.  In a procedure, the return value
     from the Tcl code is ignored.
    </para>

    <para>
     For example, a function
     returning the greater of two integer values could be defined as:

<programlisting>
CREATE FUNCTION tcl_max(integer, integer) RETURNS integer AS $$
    if {$1 &gt; $2} {return $1}
    return $2
$$ LANGUAGE pltcl STRICT;
</programlisting>

     Note the clause <literal>STRICT</literal>, which saves us from
     having to think about null input values: if a null value is passed, the
     function will not be called at all, but will just return a null
     result automatically.
    </para>

    <para>
     In a nonstrict function,
     if the actual value of an argument is null, the corresponding
     <literal>$<replaceable>n</replaceable></literal> variable will be set to an empty string.
     To detect whether a particular argument is null, use the function
     <literal>argisnull</literal>.  For example, suppose that we wanted <function>tcl_max</function>
     with one null and one nonnull argument to return the nonnull
     argument, rather than null:

<programlisting>
CREATE FUNCTION tcl_max(integer, integer) RETURNS integer AS $$
    if {[argisnull 1]} {
        if {[argisnull 2]} { return_null }
        return $2
    }
    if {[argisnull 2]} { return $1 }
    if {$1 &gt; $2} {return $1}
    return $2
$$ LANGUAGE pltcl;
</programlisting>
    </para>

    <para>
     As shown above,
     to return a null value from a PL/Tcl function, execute
     <literal>return_null</literal>.  This can be done whether the
     function is strict or not.
    </para>

    <para>
     Composite-type arguments are passed to the function as Tcl
     arrays.  The element names of the array are the attribute names
     of the composite type. If an attribute in the passed row has the
     null value, it will not appear in the array. Here is an example:

<programlisting>
CREATE TABLE employee (
    name text,
    salary integer,
    age integer
);

CREATE FUNCTION overpaid(employee) RETURNS boolean AS $$
    if {200000.0 &lt; $1(salary)} {
        return "t"
    }
    if {$1(age) &lt; 30 &amp;&amp; 100000.0 &lt; $1(salary)} {
        return "t"
    }
    return "f"
$$ LANGUAGE pltcl;
</programlisting>
    </para>

    <para>
     PL/Tcl functions can return composite-type results, too.  To do this,
     the Tcl code must return a list of column name/value pairs matching
     the expected result type.  Any column names omitted from the list
     are returned as nulls, and an error is raised if there are unexpected
     column names.  Here is an example:

<programlisting>
CREATE FUNCTION square_cube(in int, out squared int, out cubed int) AS $$
    return [list squared [expr {$1 * $1}] cubed [expr {$1 * $1 * $1}]]
$$ LANGUAGE pltcl;
</programlisting>
    </para>

    <para>
     Output arguments of procedures are returned in the same way, for example:

<programlisting>
CREATE PROCEDURE tcl_triple(INOUT a integer, INOUT b integer) AS $$
    return [list a [expr {$1 * 3}] b [expr {$2 * 3}]]
$$ LANGUAGE pltcl;

CALL tcl_triple(5, 10);
</programlisting>
    </para>

    <tip>
     <para>
      The result list can be made from an array representation of the
      desired tuple with the <literal>array get</literal> Tcl command.  For example:

<programlisting>
CREATE FUNCTION raise_pay(employee, delta int) RETURNS employee AS $$
    set 1(salary) [expr {$1(salary) + $2}]
    return [array get 1]
$$ LANGUAGE pltcl;
</programlisting>
     </para>
    </tip>

    <para>
     PL/Tcl functions can return sets.  To do this, the Tcl code should
     call <function>return_next</function> once per row to be returned,
     passing either the appropriate value when returning a scalar type,
     or a list of column name/value pairs when returning a composite type.
     Here is an example returning a scalar type:

<programlisting>
CREATE FUNCTION sequence(int, int) RETURNS SETOF int AS $$
    for {set i $1} {$i &lt; $2} {incr i} {
        return_next $i
    }
$$ LANGUAGE pltcl;
</programlisting>

     and here is one returning a composite type:

<programlisting>
CREATE FUNCTION table_of_squares(int, int) RETURNS TABLE (x int, x2 int) AS $$
    for {set i $1} {$i &lt; $2} {incr i} {
        return_next [list x $i x2 [expr {$i * $i}]]
    }
$$ LANGUAGE pltcl;
</programlisting>
    </para>

   </sect1>

   <sect1 id="pltcl-data">
    <title>Data Values in PL/Tcl</title>

    <para>
     The argument values supplied to a PL/Tcl function's code are simply
     the input arguments converted to text form (just as if they had been
     displayed by a <command>SELECT</command> statement).  Conversely, the
     <literal>return</literal> and <literal>return_next</literal> commands will accept
     any string that is acceptable input format for the function's declared
     result type, or for the specified column of a composite result type.
    </para>

   </sect1>

   <sect1 id="pltcl-global">
    <title>Global Data in PL/Tcl</title>

    <indexterm zone="pltcl-global">
     <primary>global data</primary>
     <secondary>in PL/Tcl</secondary>
    </indexterm>

    <para>
     Sometimes it
     is useful to have some global data that is held between two
     calls to a function or is shared between different functions.
     This is easily done in PL/Tcl, but there are some restrictions that
     must be understood.
    </para>

    <para>
     For security reasons, PL/Tcl executes functions called by any one SQL
     role in a separate Tcl interpreter for that role.  This prevents
     accidental or malicious interference by one user with the behavior of
     another user's PL/Tcl functions.  Each such interpreter will have its own
     values for any <quote>global</quote> Tcl variables.  Thus, two PL/Tcl
     functions will share the same global variables if and only if they are
     executed by the same SQL role.  In an application wherein a single
     session executes code under multiple SQL roles (via <literal>SECURITY
     DEFINER</literal> functions, use of <command>SET ROLE</command>, etc.) you may need to
     take explicit steps to ensure that PL/Tcl functions can share data.  To
     do that, make sure that functions that should communicate are owned by
     the same user, and mark them <literal>SECURITY DEFINER</literal>.  You must of
     course take care that such functions can't be used to do anything
     unintended.
    </para>

    <para>
     All PL/TclU functions used in a session execute in the same Tcl
     interpreter, which of course is distinct from the interpreter(s)
     used for PL/Tcl functions.  So global data is automatically shared
     between PL/TclU functions.  This is not considered a security risk
     because all PL/TclU functions execute at the same trust level,
     namely that of a database superuser.
    </para>

    <para>
     To help protect PL/Tcl functions from unintentionally interfering
     with each other, a global
     array is made available to each function via the <function>upvar</function>
     command. The global name of this variable is the function's internal
     name, and the local name is <literal>GD</literal>.  It is recommended that
     <literal>GD</literal> be used
     for persistent private data of a function.  Use regular Tcl global
     variables only for values that you specifically intend to be shared among
     multiple functions.  (Note that the <literal>GD</literal> arrays are only
     global within a particular interpreter, so they do not bypass the
     security restrictions mentioned above.)
    </para>

    <para>
     An example of using <literal>GD</literal> appears in the
     <function>spi_execp</function> example below.
    </para>
   </sect1>

   <sect1 id="pltcl-dbaccess">
    <title>Database Access from PL/Tcl</title>

    <para>
     In this section, we follow the usual Tcl convention of using question
     marks, rather than brackets, to indicate an optional element in a
     syntax synopsis.  The following commands are available to access
     the database from the body of a PL/Tcl function:

    <variablelist>

     <varlistentry>
      <term><literal><function>spi_exec</function> <optional role="tcl">-count <replaceable>n</replaceable></optional> <optional role="tcl">-array <replaceable>name</replaceable></optional> <replaceable>command</replaceable> <optional role="tcl"><replaceable>loop-body</replaceable></optional></literal></term>
      <listitem>
       <para>
        Executes an SQL command given as a string.  An error in the command
        causes an error to be raised.  Otherwise, the return value of <function>spi_exec</function>
        is the number of rows processed (selected, inserted, updated, or
        deleted) by the command, or zero if the command is a utility
        statement.  In addition, if the command is a <command>SELECT</command> statement, the
        values of the selected columns are placed in Tcl variables as
        described below.
       </para>
       <para>
        The optional <literal>-count</literal> value tells
        <function>spi_exec</function> to stop
        once <replaceable>n</replaceable> rows have been retrieved,
        much as if the query included a <literal>LIMIT</literal> clause.
        If <replaceable>n</replaceable> is zero, the query is run to
        completion, the same as when <literal>-count</literal> is omitted.
       </para>
       <para>
        If the command is a <command>SELECT</command> statement, the values of the
        result columns are placed into Tcl variables named after the columns.
        If the <literal>-array</literal> option is given, the column values are
        instead stored into elements of the named associative array, with the
        column names used as array indexes.  In addition, the current row
        number within the result (counting from zero) is stored into the array
        element named <quote><literal>.tupno</literal></quote>, unless that name is
        in use as a column name in the result.
       </para>
       <para>
        If the command is a <command>SELECT</command> statement and no <replaceable>loop-body</replaceable>
        script is given, then only the first row of results are stored into
        Tcl variables or array elements; remaining rows, if any, are ignored.
        No storing occurs if the query returns no rows.  (This case can be
        detected by checking the result of <function>spi_exec</function>.)
        For example:
<programlisting>
spi_exec "SELECT count(*) AS cnt FROM pg_proc"
</programlisting>
        will set the Tcl variable <literal>$cnt</literal> to the number of rows in
        the <structname>pg_proc</structname> system catalog.
       </para>
       <para>
        If the optional <replaceable>loop-body</replaceable> argument is given, it is
        a piece of Tcl script that is executed once for each row in the
        query result.  (<replaceable>loop-body</replaceable> is ignored if the given
        command is not a <command>SELECT</command>.)
        The values of the current row's columns
        are stored into Tcl variables or array elements before each iteration.
        For example:
<programlisting>
spi_exec -array C "SELECT * FROM pg_class" {
    elog DEBUG "have table $C(relname)"
}
</programlisting>
        will print a log message for every row of <literal>pg_class</literal>.  This
        feature works similarly to other Tcl looping constructs; in
        particular <literal>continue</literal> and <literal>break</literal> work in the
        usual way inside the loop body.
       </para>
       <para>
        If a column of a query result is null, the target
        variable for it is <quote>unset</quote> rather than being set.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term><function>spi_prepare</function> <replaceable>query</replaceable> <replaceable>typelist</replaceable></term>
      <listitem>
       <para>
        Prepares and saves a query plan for later execution.  The
        saved plan will be retained for the life of the current
        session.<indexterm><primary>preparing a query</primary>
        <secondary>in PL/Tcl</secondary></indexterm>
       </para>
       <para>
        The query can use parameters, that is, placeholders for
        values to be supplied whenever the plan is actually executed.
        In the query string, refer to parameters
        by the symbols <literal>$1</literal> ... <literal>$<replaceable>n</replaceable></literal>.
        If the query uses parameters, the names of the parameter types
        must be given as a Tcl list.  (Write an empty list for
        <replaceable>typelist</replaceable> if no parameters are used.)
       </para>
       <para>
        The return value from <function>spi_prepare</function> is a query ID
        to be used in subsequent calls to <function>spi_execp</function>. See
        <function>spi_execp</function> for an example.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term><literal><function>spi_execp</function> <optional role="tcl">-count <replaceable>n</replaceable></optional> <optional role="tcl">-array <replaceable>name</replaceable></optional> <optional role="tcl">-nulls <replaceable>string</replaceable></optional> <replaceable>queryid</replaceable> <optional role="tcl"><replaceable>value-list</replaceable></optional> <optional role="tcl"><replaceable>loop-body</replaceable></optional></literal></term>
      <listitem>
       <para>
        Executes a query previously prepared with <function>spi_prepare</function>.
        <replaceable>queryid</replaceable> is the ID returned by
        <function>spi_prepare</function>.  If the query references parameters,
        a <replaceable>value-list</replaceable> must be supplied.  This
        is a Tcl list of actual values for the parameters.  The list must be
        the same length as the parameter type list previously given to
        <function>spi_prepare</function>.  Omit <replaceable>value-list</replaceable>
        if the query has no parameters.
       </para>
       <para>
        The optional value for <literal>-nulls</literal> is a string of spaces and
        <literal>'n'</literal> characters telling <function>spi_execp</function>
        which of the parameters are null values. If given, it must have exactly the
        same length as the <replaceable>value-list</replaceable>.  If it
        is not given, all the parameter values are nonnull.
       </para>
       <para>
        Except for the way in which the query and its parameters are specified,
        <function>spi_execp</function> works just like <function>spi_exec</function>.
        The <literal>-count</literal>, <literal>-array</literal>, and
        <replaceable>loop-body</replaceable> options are the same,
        and so is the result value.
       </para>
       <para>
        Here's an example of a PL/Tcl function using a prepared plan:

<programlisting>
CREATE FUNCTION t1_count(integer, integer) RETURNS integer AS $$
    if {![ info exists GD(plan) ]} {
        # prepare the saved plan on the first call
        set GD(plan) [ spi_prepare \
                "SELECT count(*) AS cnt FROM t1 WHERE num &gt;= \$1 AND num &lt;= \$2" \
                [ list int4 int4 ] ]
    }
    spi_execp -count 1 $GD(plan) [ list $1 $2 ]
    return $cnt
$$ LANGUAGE pltcl;
</programlisting>

        We need backslashes inside the query string given to
        <function>spi_prepare</function> to ensure that the
        <literal>$<replaceable>n</replaceable></literal> markers will be passed
        through to <function>spi_prepare</function> as-is, and not replaced by Tcl
        variable substitution.

       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term><function>subtransaction</function> <replaceable>command</replaceable></term>
      <listitem>
       <para>
        The Tcl script contained in <replaceable>command</replaceable> is
        executed within an SQL subtransaction.  If the script returns an
        error, that entire subtransaction is rolled back before returning the
        error out to the surrounding Tcl code.
        See <xref linkend="pltcl-subtransactions"/> for more details and an
        example.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term><function>quote</function> <replaceable>string</replaceable></term>
      <listitem>
       <para>
        Doubles all occurrences of single quote and backslash characters
        in the given string.  This can be used to safely quote strings
        that are to be inserted into SQL commands given
        to <function>spi_exec</function> or
        <function>spi_prepare</function>.
        For example, think about an SQL command string like:

<programlisting>
"SELECT '$val' AS ret"
</programlisting>

        where the Tcl variable <literal>val</literal> actually contains
        <literal>doesn't</literal>. This would result
        in the final command string:

<programlisting>
SELECT 'doesn't' AS ret
</programlisting>

        which would cause a parse error during
        <function>spi_exec</function> or
        <function>spi_prepare</function>.
        To work properly, the submitted command should contain:

<programlisting>
SELECT 'doesn''t' AS ret
</programlisting>

        which can be formed in PL/Tcl using:

<programlisting>
"SELECT '[ quote $val ]' AS ret"
</programlisting>

        One advantage of <function>spi_execp</function> is that you don't
        have to quote parameter values like this, since the parameters are never
        parsed as part of an SQL command string.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
      <term>
       <function>elog</function> <replaceable>level</replaceable> <replaceable>msg</replaceable>
       <indexterm>
        <primary>elog</primary>
        <secondary>in PL/Tcl</secondary>
       </indexterm>
      </term>
      <listitem>
       <para>
        Emits a log or error message. Possible levels are
        <literal>DEBUG</literal>, <literal>LOG</literal>, <literal>INFO</literal>,
        <literal>NOTICE</literal>, <literal>WARNING</literal>, <literal>ERROR</literal>, and
        <literal>FATAL</literal>. <literal>ERROR</literal>
        raises an error condition; if this is not trapped by the surrounding
        Tcl code, the error propagates out to the calling query, causing
        the current transaction or subtransaction to be aborted.  This
        is effectively the same as the Tcl <literal>error</literal> command.
        <literal>FATAL</literal> aborts the transaction and causes the current
        session to shut down.  (There is probably no good reason to use
        this error level in PL/Tcl functions, but it's provided for
        completeness.)  The other levels only generate messages of different
        priority levels.
        Whether messages of a particular priority are reported to the client,
        written to the server log, or both is controlled by the
        <xref linkend="guc-log-min-messages"/> and
        <xref linkend="guc-client-min-messages"/> configuration
        variables. See <xref linkend="runtime-config"/>
        and <xref linkend="pltcl-error-handling"/>
        for more information.
       </para>
      </listitem>
     </varlistentry>

    </variablelist>
    </para>

   </sect1>

   <sect1 id="pltcl-trigger">
    <title>Trigger Functions in PL/Tcl</title>

    <indexterm>
     <primary>trigger</primary>
     <secondary>in PL/Tcl</secondary>
    </indexterm>

    <para>
     Trigger functions can be written in PL/Tcl.
     <productname>PostgreSQL</productname> requires that a function that is to be called
     as a trigger must be declared as a function with no arguments
     and a return type of <literal>trigger</literal>.
    </para>
    <para>
     The information from the trigger manager is passed to the function body
     in the following variables:

     <variablelist>

      <varlistentry>
       <term><varname>$TG_name</varname></term>
       <listitem>
        <para>
         The name of the trigger from the <command>CREATE TRIGGER</command> statement.
        </para>
       </listitem>
      </varlistentry>

      <varlistentry>
       <term><varname>$TG_relid</varname></term>
       <listitem>
        <para>
         The object ID of the table that caused the trigger function
         to be invoked.
        </para>
       </listitem>
      </varlistentry>

      <varlistentry>
       <term><varname>$TG_table_name</varname></term>
       <listitem>
        <para>
         The name of the table that caused the trigger function
         to be invoked.
        </para>
       </listitem>
      </varlistentry>

      <varlistentry>
       <term><varname>$TG_table_schema</varname></term>
       <listitem>
        <para>
         The schema of the table that caused the trigger function
         to be invoked.
        </para>
       </listitem>
      </varlistentry>

      <varlistentry>
       <term><varname>$TG_relatts</varname></term>
       <listitem>
        <para>
         A Tcl list of the table column names, prefixed with an empty list
         element. So looking up a column name in the list with <application>Tcl</application>'s
         <function>lsearch</function> command returns the element's number starting
         with 1 for the first column, the same way the columns are customarily
         numbered in <productname>PostgreSQL</productname>.  (Empty list
         elements also appear in the positions of columns that have been
         dropped, so that the attribute numbering is correct for columns
         to their right.)
        </para>
       </listitem>
      </varlistentry>

      <varlistentry>
       <term><varname>$TG_when</varname></term>
       <listitem>
        <para>
         The string <literal>BEFORE</literal>, <literal>AFTER</literal>, or
         <literal>INSTEAD OF</literal>, depending on the type of trigger event.
        </para>
       </listitem>
      </varlistentry>

      <varlistentry>
       <term><varname>$TG_level</varname></term>
       <listitem>
        <para>
         The string <literal>ROW</literal> or <literal>STATEMENT</literal> depending on the
         type of trigger event.
        </para>
       </listitem>
      </varlistentry>

      <varlistentry>
       <term><varname>$TG_op</varname></term>
       <listitem>
        <para>
         The string <literal>INSERT</literal>, <literal>UPDATE</literal>,
         <literal>DELETE</literal>, or <literal>TRUNCATE</literal> depending on the type of
         trigger event.
        </para>
       </listitem>
      </varlistentry>

      <varlistentry>
       <term><varname>$NEW</varname></term>
       <listitem>
        <para>
         An associative array containing the values of the new table
         row for <command>INSERT</command> or <command>UPDATE</command> actions, or
         empty for <command>DELETE</command>.  The array is indexed by column
         name.  Columns that are null will not appear in the array.
         This is not set for statement-level triggers.
        </para>
       </listitem>
      </varlistentry>

      <varlistentry>
       <term><varname>$OLD</varname></term>
       <listitem>
        <para>
         An associative array containing the values of the old table
         row for <command>UPDATE</command> or <command>DELETE</command> actions, or
         empty for <command>INSERT</command>.  The array is indexed by column
         name.  Columns that are null will not appear in the array.
         This is not set for statement-level triggers.
        </para>
       </listitem>
      </varlistentry>

      <varlistentry>
       <term><varname>$args</varname></term>
       <listitem>
        <para>
         A Tcl list of the arguments to the function as given in the
         <command>CREATE TRIGGER</command> statement. These arguments are also accessible as
         <literal>$1</literal> ... <literal>$<replaceable>n</replaceable></literal> in the function body.
        </para>
       </listitem>
      </varlistentry>

     </variablelist>
    </para>

    <para>
     The return value from a trigger function can be one of the strings
     <literal>OK</literal> or <literal>SKIP</literal>, or a list of column name/value pairs.
     If the return value is <literal>OK</literal>,
     the operation (<command>INSERT</command>/<command>UPDATE</command>/<command>DELETE</command>)
     that fired the trigger will proceed
     normally. <literal>SKIP</literal> tells the trigger manager to silently suppress
     the operation for this row. If a list is returned, it tells PL/Tcl to
     return a modified row to the trigger manager; the contents of the
     modified row are specified by the column names and values in the list.
     Any columns not mentioned in the list are set to null.
     Returning a modified row is only meaningful
     for row-level <literal>BEFORE</literal> <command>INSERT</command> or <command>UPDATE</command>
     triggers, for which the modified row will be inserted instead of the one
     given in <varname>$NEW</varname>; or for row-level <literal>INSTEAD OF</literal>
     <command>INSERT</command> or <command>UPDATE</command> triggers where the returned row
     is used as the source data for <command>INSERT RETURNING</command> or
     <command>UPDATE RETURNING</command> clauses.
     In row-level <literal>BEFORE</literal> <command>DELETE</command> or <literal>INSTEAD
     OF</literal> <command>DELETE</command> triggers, returning a modified row has the same
     effect as returning <literal>OK</literal>, that is the operation proceeds.
     The trigger return value is ignored for all other types of triggers.
    </para>

    <tip>
     <para>
      The result list can be made from an array representation of the
      modified tuple with the <literal>array get</literal> Tcl command.
     </para>
    </tip>

    <para>
     Here's a little example trigger function that forces an integer value
     in a table to keep track of the number of updates that are performed on the
     row. For new rows inserted, the value is initialized to 0 and then
     incremented on every update operation.

<programlisting>
CREATE FUNCTION trigfunc_modcount() RETURNS trigger AS $$
    switch $TG_op {
        INSERT {
            set NEW($1) 0
        }
        UPDATE {
            set NEW($1) $OLD($1)
            incr NEW($1)
        }
        default {
            return OK
        }
    }
    return [array get NEW]
$$ LANGUAGE pltcl;

CREATE TABLE mytab (num integer, description text, modcnt integer);

CREATE TRIGGER trig_mytab_modcount BEFORE INSERT OR UPDATE ON mytab
    FOR EACH ROW EXECUTE FUNCTION trigfunc_modcount('modcnt');
</programlisting>

     Notice that the trigger function itself does not know the column
     name; that's supplied from the trigger arguments.  This lets the
     trigger function be reused with different tables.
    </para>
   </sect1>

   <sect1 id="pltcl-event-trigger">
    <title>Event Trigger Functions in PL/Tcl</title>

    <indexterm>
     <primary>event trigger</primary>
     <secondary>in PL/Tcl</secondary>
    </indexterm>

    <para>
     Event trigger functions can be written in PL/Tcl.
     <productname>PostgreSQL</productname> requires that a function that is
     to be called as an event trigger must be declared as a function with no
     arguments and a return type of <literal>event_trigger</literal>.
    </para>
    <para>
     The information from the trigger manager is passed to the function body
     in the following variables:

     <variablelist>

      <varlistentry>
       <term><varname>$TG_event</varname></term>
       <listitem>
        <para>
         The name of the event the trigger is fired for.
        </para>
       </listitem>
      </varlistentry>

      <varlistentry>
       <term><varname>$TG_tag</varname></term>
       <listitem>
        <para>
         The command tag for which the trigger is fired.
        </para>
       </listitem>
      </varlistentry>
     </variablelist>
    </para>

    <para>
     The return value of the trigger function is ignored.
    </para>

    <para>
     Here's a little example event trigger function that simply raises
     a <literal>NOTICE</literal> message each time a supported command is
     executed:

<programlisting>
CREATE OR REPLACE FUNCTION tclsnitch() RETURNS event_trigger AS $$
  elog NOTICE "tclsnitch: $TG_event $TG_tag"
$$ LANGUAGE pltcl;

CREATE EVENT TRIGGER tcl_a_snitch ON ddl_command_start EXECUTE FUNCTION tclsnitch();
</programlisting>
    </para>
   </sect1>

   <sect1 id="pltcl-error-handling">
    <title>Error Handling in PL/Tcl</title>

    <indexterm>
     <primary>exceptions</primary>
     <secondary>in PL/Tcl</secondary>
    </indexterm>

    <para>
     Tcl code within or called from a PL/Tcl function can raise an error,
     either by executing some invalid operation or by generating an error
     using the Tcl <function>error</function> command or
     PL/Tcl's <function>elog</function> command.  Such errors can be caught
     within Tcl using the Tcl <function>catch</function> command.  If an
     error is not caught but is allowed to propagate out to the top level of
     execution of the PL/Tcl function, it is reported as an SQL error in the
     function's calling query.
    </para>

    <para>
     Conversely, SQL errors that occur within PL/Tcl's
     <function>spi_exec</function>, <function>spi_prepare</function>,
     and <function>spi_execp</function> commands are reported as Tcl errors,
     so they are catchable by Tcl's <function>catch</function> command.
     (Each of these PL/Tcl commands runs its SQL operation in a
     subtransaction, which is rolled back on error, so that any
     partially-completed operation is automatically cleaned up.)
     Again, if an error propagates out to the top level without being caught,
     it turns back into an SQL error.
    </para>

    <para>
     Tcl provides an <varname>errorCode</varname> variable that can represent
     additional information about an error in a form that is easy for Tcl
     programs to interpret.  The contents are in Tcl list format, and the
     first word identifies the subsystem or library reporting the error;
     beyond that the contents are left to the individual subsystem or
     library.  For database errors reported by PL/Tcl commands, the first
     word is <literal>POSTGRES</literal>, the second word is the PostgreSQL
     version number, and additional words are field name/value pairs
     providing detailed information about the error.
     Fields <varname>SQLSTATE</varname>, <varname>condition</varname>,
     and <varname>message</varname> are always supplied
     (the first two represent the error code and condition name as shown
     in <xref linkend="errcodes-appendix"/>).
     Fields that may be present include
     <varname>detail</varname>, <varname>hint</varname>, <varname>context</varname>,
     <varname>schema</varname>, <varname>table</varname>, <varname>column</varname>,
     <varname>datatype</varname>, <varname>constraint</varname>,
     <varname>statement</varname>, <varname>cursor_position</varname>,
     <varname>filename</varname>, <varname>lineno</varname>, and
     <varname>funcname</varname>.
    </para>

    <para>
     A convenient way to work with PL/Tcl's <varname>errorCode</varname>
     information is to load it into an array, so that the field names become
     array subscripts.  Code for doing that might look like
<programlisting>
if {[catch { spi_exec $sql_command }]} {
    if {[lindex $::errorCode 0] == "POSTGRES"} {
        array set errorArray $::errorCode
        if {$errorArray(condition) == "undefined_table"} {
            # deal with missing table
        } else {
            # deal with some other type of SQL error
        }
    }
}
</programlisting>
     (The double colons explicitly specify that <varname>errorCode</varname>
     is a global variable.)
    </para>
   </sect1>

   <sect1 id="pltcl-subtransactions">
    <title>Explicit Subtransactions in PL/Tcl</title>

    <indexterm>
     <primary>subtransactions</primary>
     <secondary>in PL/Tcl</secondary>
    </indexterm>

    <para>
     Recovering from errors caused by database access as described in
     <xref linkend="pltcl-error-handling"/> can lead to an undesirable
     situation where some operations succeed before one of them fails,
     and after recovering from that error the data is left in an
     inconsistent state.  PL/Tcl offers a solution to this problem in
     the form of explicit subtransactions.
    </para>

    <para>
     Consider a function that implements a transfer between two accounts:
<programlisting>
CREATE FUNCTION transfer_funds() RETURNS void AS $$
    if [catch {
        spi_exec "UPDATE accounts SET balance = balance - 100 WHERE account_name = 'joe'"
        spi_exec "UPDATE accounts SET balance = balance + 100 WHERE account_name = 'mary'"
    } errormsg] {
        set result [format "error transferring funds: %s" $errormsg]
    } else {
        set result "funds transferred successfully"
    }
    spi_exec "INSERT INTO operations (result) VALUES ('[quote $result]')"
$$ LANGUAGE pltcl;
</programlisting>
     If the second <command>UPDATE</command> statement results in an
     exception being raised, this function will log the failure, but
     the result of the first <command>UPDATE</command> will
     nevertheless be committed.  In other words, the funds will be
     withdrawn from Joe's account, but will not be transferred to
     Mary's account.  This happens because each <function>spi_exec</function>
     is a separate subtransaction, and only one of those subtransactions
     got rolled back.
    </para>

    <para>
     To handle such cases, you can wrap multiple database operations in an
     explicit subtransaction, which will succeed or roll back as a whole.
     PL/Tcl provides a <function>subtransaction</function> command to manage
     this.  We can rewrite our function as:
<programlisting>
CREATE FUNCTION transfer_funds2() RETURNS void AS $$
    if [catch {
        subtransaction {
            spi_exec "UPDATE accounts SET balance = balance - 100 WHERE account_name = 'joe'"
            spi_exec "UPDATE accounts SET balance = balance + 100 WHERE account_name = 'mary'"
        }
    } errormsg] {
        set result [format "error transferring funds: %s" $errormsg]
    } else {
        set result "funds transferred successfully"
    }
    spi_exec "INSERT INTO operations (result) VALUES ('[quote $result]')"
$$ LANGUAGE pltcl;
</programlisting>
     Note that use of <function>catch</function> is still required for this
     purpose.  Otherwise the error would propagate to the top level of the
     function, preventing the desired insertion into
     the <structname>operations</structname> table.
     The <function>subtransaction</function> command does not trap errors, it
     only assures that all database operations executed inside its scope will
     be rolled back together when an error is reported.
    </para>

    <para>
     A rollback of an explicit subtransaction occurs on any error reported
     by the contained Tcl code, not only errors originating from database
     access.  Thus a regular Tcl exception raised inside
     a <function>subtransaction</function> command will also cause the
     subtransaction to be rolled back.  However, non-error exits out of the
     contained Tcl code (for instance, due to <function>return</function>) do
     not cause a rollback.
    </para>
   </sect1>

   <sect1 id="pltcl-transactions">
    <title>Transaction Management</title>

    <para>
     In a procedure called from the top level or an anonymous code block
     (<command>DO</command> command) called from the top level it is possible
     to control transactions.  To commit the current transaction, call the
     <literal>commit</literal> command.  To roll back the current transaction,
     call the <literal>rollback</literal> command.  (Note that it is not
     possible to run the SQL commands <command>COMMIT</command> or
     <command>ROLLBACK</command> via <function>spi_exec</function> or similar.
     It has to be done using these functions.)  After a transaction is ended,
     a new transaction is automatically started, so there is no separate
     command for that.
    </para>

    <para>
     Here is an example:
<programlisting>
CREATE PROCEDURE transaction_test1()
LANGUAGE pltcl
AS $$
for {set i 0} {$i &lt; 10} {incr i} {
    spi_exec "INSERT INTO test1 (a) VALUES ($i)"
    if {$i % 2 == 0} {
        commit
    } else {
        rollback
    }
}
$$;

CALL transaction_test1();
</programlisting>
    </para>

    <para>
     Transactions cannot be ended when an explicit subtransaction is active.
    </para>
   </sect1>

   <sect1 id="pltcl-config">
    <title>PL/Tcl Configuration</title>

    <para>
     This section lists configuration parameters that
     affect <application>PL/Tcl</application>.
    </para>

    <variablelist>

     <varlistentry id="guc-pltcl-start-proc" xreflabel="pltcl.start_proc">
      <term>
       <varname>pltcl.start_proc</varname> (<type>string</type>)
       <indexterm>
        <primary><varname>pltcl.start_proc</varname> configuration parameter</primary>
       </indexterm>
      </term>
      <listitem>
       <para>
        This parameter, if set to a nonempty string, specifies the name
        (possibly schema-qualified) of a parameterless PL/Tcl function that
        is to be executed whenever a new Tcl interpreter is created for
        PL/Tcl.  Such a function can perform per-session initialization, such
        as loading additional Tcl code.  A new Tcl interpreter is created
        when a PL/Tcl function is first executed in a database session, or
        when an additional interpreter has to be created because a PL/Tcl
        function is called by a new SQL role.
       </para>

       <para>
        The referenced function must be written in the <literal>pltcl</literal>
        language, and must not be marked <literal>SECURITY DEFINER</literal>.
        (These restrictions ensure that it runs in the interpreter it's
        supposed to initialize.)  The current user must have permission to
        call it, too.
       </para>

       <para>
        If the function fails with an error it will abort the function call
        that caused the new interpreter to be created and propagate out to
        the calling query, causing the current transaction or subtransaction
        to be aborted.  Any actions already done within Tcl won't be undone;
        however, that interpreter won't be used again.  If the language is
        used again the initialization will be attempted again within a fresh
        Tcl interpreter.
       </para>

       <para>
        Only superusers can change this setting.  Although this setting
        can be changed within a session, such changes will not affect Tcl
        interpreters that have already been created.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry id="guc-pltclu-start-proc" xreflabel="pltclu.start_proc">
      <term>
       <varname>pltclu.start_proc</varname> (<type>string</type>)
       <indexterm>
        <primary><varname>pltclu.start_proc</varname> configuration parameter</primary>
       </indexterm>
      </term>
      <listitem>
       <para>
        This parameter is exactly like <varname>pltcl.start_proc</varname>,
        except that it applies to PL/TclU.  The referenced function must
        be written in the <literal>pltclu</literal> language.
       </para>
      </listitem>
     </varlistentry>

    </variablelist>
   </sect1>

   <sect1 id="pltcl-procnames">
    <title>Tcl Procedure Names</title>

    <para>
     In <productname>PostgreSQL</productname>, the same function name can be used for
     different function definitions as long as the number of arguments or their types
     differ. Tcl, however, requires all procedure names to be distinct.
     PL/Tcl deals with this by making the internal Tcl procedure names contain
     the object
     ID of the function from the system table <structname>pg_proc</structname> as part of their name. Thus,
     <productname>PostgreSQL</productname> functions with the same name
     and different argument types will be different Tcl procedures, too.  This
     is not normally a concern for a PL/Tcl programmer, but it might be visible
     when debugging.
    </para>

   </sect1>
 </chapter>