PL/Python patch for Universal Newline Support - Mailing list pgsql-patches

From Michael Fuhr
Subject PL/Python patch for Universal Newline Support
Date
Msg-id 20050319050847.GA90344@winnie.fuhr.org
Whole thread Raw
Responses Re: PL/Python patch for Universal Newline Support  (Neil Conway <neilc@samurai.com>)
List pgsql-patches
The attached patch mimics Universal Newline Support for PL/Python
functions.  For background, see PEP 278 and the recent "plpython
function problem workaround" thread in pgsql-general:

http://www.python.org/peps/pep-0278.html
http://archives.postgresql.org/pgsql-general/2005-03/msg00823.php

In short, embedded Python code must have lines that end in \n, not
\r\n or \r.  This patch adds logic to PLy_procedure_munge_source()
to skip \r followed by \n, otherwise to convert \r to \n.  I tested
it in HEAD but REL8_0_STABLE has the same version of plpython.c (1.58),
so it should work there as well.

Should the PL/Python documentation mention this behavior?  If so
then I can resubmit with a documentation patch.

How should I submit regression tests?  Here's what I did:

CREATE FUNCTION test_lf() RETURNS integer AS
'x=100\ny=23\nreturn x+y\n'
LANGUAGE plpythonu;

CREATE FUNCTION test_crlf() RETURNS integer AS
'x=100\r\ny=23\r\nreturn x+y\r\n'
LANGUAGE plpythonu;

CREATE FUNCTION test_cr() RETURNS integer AS
'x=100\ry=23\rreturn x+y\r'
LANGUAGE plpythonu;

Here's what an unpatched system does:

SELECT test_lf();
 test_lf
---------
     123
(1 row)

SELECT test_cr();
ERROR:  plpython: could not compile function "test_cr"
DETAIL:  exceptions.SyntaxError: invalid syntax (line 2)

SELECT test_crlf();
ERROR:  plpython: could not compile function "test_crlf"
DETAIL:  exceptions.SyntaxError: invalid syntax (line 2)

Here's a patched system:

SELECT test_lf();
 test_lf
---------
     123
(1 row)

SELECT test_cr();
 test_cr
---------
     123
(1 row)

SELECT test_crlf();
 test_crlf
-----------
       123
(1 row)

This patch doesn't address the indentation problem mentioned here:

http://archives.postgresql.org/pgsql-general/2005-03/msg00762.php

The solution to that problem is still being discussed.

--
Michael Fuhr
http://www.fuhr.org/~mfuhr/

Attachment

pgsql-patches by date:

Previous
From: Alvaro Herrera
Date:
Subject: Re: Shared dependency patch
Next
From: Alvaro Herrera
Date:
Subject: HeapTupleSatisfiesUpdate result as enum