psycopg/psycopg2
1
1
Fork 0
mirror of https://github.com/psycopg/psycopg2.git synced 2025-02-08 05:10:32 +03:00
Code Issues Packages Projects Releases Wiki Activity
2,337 Commits 12 Branches 87 Tags 12 MiB
8decf34ad7
Commit Graph

942 Commits

Author SHA1 Message Date
Daniele Varrazzo
a79a5292e7 Added support with cursors without scroll clause 
Using nothing is different from NO SCROLL, see DECLARE notes
in PG docs.
 2012-08-15 09:44:44 +01:00
Daniele Varrazzo
fa4994e471 Added scrollable cursor implementation 
Patch provided by Jon Nelson (ticket #108).
 2012-08-15 00:30:54 +01:00
Daniele Varrazzo
91c2ff9296 Fixed tpc_recover() with RealDictStuff 
Same problem and correction of ticket #114.
 2012-08-14 23:29:19 +01:00
Daniele Varrazzo
f9a13eb563 connection.reset() implemented using DISCARD ALL 2012-06-25 16:37:11 +01:00
Daniele Varrazzo
e8fb84306d Fixed cursor name as unicode string on Python 3 2012-05-27 08:58:55 +01:00
Daniele Varrazzo
7982a6ac0b Dropped GIL release around function calling PyMem_Malloc 
Closes ticket #110.
 2012-05-22 17:22:57 +01:00
Daniele Varrazzo
27421f1e41 Name can be passed as None to cursor() 
Makes invocation from subclasses and generic code easier.

Code simplified by using default values for keyword arguments
and avoiding needless conversions back and forth between Python and C
strings. Also added connection type check to cursor's init.
 2012-04-11 17:32:10 +01:00
Daniele Varrazzo
c63891af8d Fixed bad error return code from cursor's init 2012-04-11 17:26:11 +01:00
Federico Di Gregorio
550130b19e Merge remote-tracking branch 'piro/devel' into devel 
Conflicts:
	psycopg/lobject_int.c
 2012-03-08 12:28:52 +01:00
Daniele Varrazzo
b8c75d9de0 Merge branch 'gcc-python-plugin' into devel 2012-03-05 02:48:11 +00:00
Daniele Varrazzo
37aa62ca52 Merge branch 'close-idempotent' into devel 2012-03-05 02:47:52 +00:00
Daniele Varrazzo
8707d8c399 Fixed iterator refcount in case of memory error during COPY 2012-03-05 02:09:20 +00:00
Daniele Varrazzo
735d50c782 Check if the object wrapped in binary is not None before trying the other types 
Otherwise it seems we clobber some result with NULL.
 2012-03-05 02:08:45 +00:00
Daniele Varrazzo
a9dc1b83ad Methods callbacks signatures match the flags they are exported with 2012-03-05 01:26:28 +00:00
Daniele Varrazzo
531084d561 Stricter types usage in several PyArg_ParseTuple calls 2012-03-04 18:01:08 +00:00
Daniele Varrazzo
0e832b97ea Proper type check in prepare() methods for list, binary, qstring 2012-03-04 17:59:51 +00:00
Daniele Varrazzo
84f2a370f6 close() methods don't raise errors if called on closed objects 2012-03-04 05:10:07 +00:00
Daniele Varrazzo
76cc838a93 Expressions rewritten in a more normal way 
(double)'0'? :)
 2012-03-04 04:47:21 +00:00
Daniele Varrazzo
5bbfd38dfb Check for errors in float adaptation 2012-03-04 04:43:14 +00:00
Daniele Varrazzo
6cece00958 Check failed list creation in array adaptation 2012-03-04 04:42:44 +00:00
Daniele Varrazzo
be35df3818 Fixed typecasters refcount 2012-03-04 04:41:36 +00:00
Daniele Varrazzo
dc4c3d3143 Guard from failed keys creation during adaptation 2012-03-04 04:40:44 +00:00
Daniele Varrazzo
18085201c8 Guard from NULL dereference if Xid allocation fails 2012-03-04 04:39:41 +00:00
Daniele Varrazzo
31812c01e6 Further modeling of exception raising 2012-03-04 04:38:44 +00:00
Daniele Varrazzo
d93732558d Raise an exception in case of failed localtime_r call 2012-03-04 04:17:03 +00:00
Daniele Varrazzo
4d15b973b0 Attempt to enforce signature for the "O&" converter functions 
It seems causing a traceback in the static checker. Enforcing it simplifies
the code, but doesn't help the checker.
 2012-03-01 02:53:29 +00:00
Daniele Varrazzo
e1266d52cd More functions annotated for static analysis 
Also more return values checked for values < 0 for errors, instead of
checking == 0 and leaving the positive side unchecked
 2012-03-01 02:53:28 +00:00
Daniele Varrazzo
5bfb6cdefe Use more compact macros to annotate functions for the static checker 2012-03-01 02:53:28 +00:00
Daniele Varrazzo
f2e4a8ed78 Functions setting exception return a negative value on error 
This works around another shortcoming of the static checker; also to be
discussed with the author.
 2012-03-01 02:53:28 +00:00
Daniele Varrazzo
9432787279 Work around a false positive returned by the static checker 
To be submitted to the author.
 2012-03-01 02:53:28 +00:00
Daniele Varrazzo
a167822e26 Use the newly provided attributes to validate exceptions raising 
Be more consistent in using 0 for success, <0 for failure, and to check
for values < 0 instead of specific -1.
 2012-03-01 02:53:28 +00:00
Marti Raudsepp
b97599166e Update all links to PostgreSQL docs to the current version. 
I also checked all links and anchors to make sure they're still valid.
 2012-02-28 18:28:07 +02:00
Daniele Varrazzo
67712e4226 Fixed possible NULL dereferencing in fetchmany()/fetchall() 2012-02-24 22:26:02 +00:00
Daniele Varrazzo
6d76e81166 Fixed possible NULL dereferencing in callproc() 2012-02-24 22:26:02 +00:00
Daniele Varrazzo
94a53b48df Building rows simplified 
Dropped repeated checks for tuple_factory.
Internal functions refactored a bit.
 2012-02-24 22:26:02 +00:00
Daniele Varrazzo
efee049338 Added error check in _mogrify for failed tuple creation 2012-02-24 22:26:02 +00:00
Daniele Varrazzo
0ee641361b Flag a few other functions returning borrowed refs 2012-02-24 22:26:01 +00:00
Daniele Varrazzo
4ecfd48671 Fixed possible NULL dereferencing in notice process 2012-02-24 22:25:56 +00:00
Daniele Varrazzo
a6df55f4e3 Flag the psycopg_ensure_*() functions as stealing a ref 2012-02-24 10:41:02 +00:00
Daniele Varrazzo
5f3f35a2c2 Mark getnextarg function as returning a borrowed reference 2012-02-24 04:25:08 +00:00
Daniele Varrazzo
fc78fb09c0 Dropped unused pq_resolve_critical() return value 2012-02-24 04:00:12 +00:00
Daniele Varrazzo
4eea8bc912 Dropped redundant check on the status before rollback 
The check is better done inside the critical section.
 2012-02-24 03:28:20 +00:00
Daniele Varrazzo
5fcbe7bd0f Check/set connection status at commit inside the critical section 
Failing to do so was causing the issue reported in ticket #103. The issue
as reported was fixed when SET ISOLATION LEVEL was dropped, but the real
problem wasn't fixed.
 2012-02-24 03:28:20 +00:00
Daniele Varrazzo
3b36100ec1 Dropped hardcoded list of exceptions in init functions 
Use the already available exctable array.

This stops the gcc-python-plugin complaining about access to potentially
uninitialized memory.
 2012-02-23 20:09:28 +00:00
Daniele Varrazzo
09be4dc5d1 Fixed potential failures while setting exceptions attributes 2012-02-23 19:48:46 +00:00
Daniele Varrazzo
7d67ecbed3 Fixed potential NULL incref 2012-02-23 19:47:36 +00:00
Daniele Varrazzo
1b27820389 Fixed refcount of exceptions dicts 2012-02-23 19:36:30 +00:00
Daniele Varrazzo
08fbd86495 Check errors in module typecasters init 2012-02-23 19:20:51 +00:00
Daniele Varrazzo
ff61cf25b6 Fixed refcount of None if namedtuples are not available 2012-02-23 18:50:06 +00:00
Daniele Varrazzo
026899e0c1 Check errors when populating encodings map 2012-02-23 18:50:06 +00:00
Daniele Varrazzo
db987250c8 Avoid installing Error.__reduce_ex__ on Python 2.4 
It is not used by the pickle protocol, and if called manually fails in an
unsettling way, probably because the exceptions were old-style classes.
 2012-01-14 18:01:15 +00:00
Daniele Varrazzo
43daba38e7 Make Error and subclasses picklable 
Useful for multiprocessing interaction.
Closes ticket #90.
 2012-01-14 17:34:09 +00:00
Daniele Varrazzo
6f21111a92 Docs/cleanup for the lo_creat patch 2012-01-10 21:51:34 +00:00
Giovanni Mascellani
e1700fe28b Use lo_creat() instead of lo_create() when creating large objects without new_oid. 
This enhances the compatibility with the pgpool-II replication middleware,
that doesn't support lo_create() calls.
 2012-01-10 20:49:58 +00:00
Giovanni Mascellani
8ec9b0f3d9 Use lo_creat() instead of lo_create() when creating large objects without new_oid. 
This enhances the compatibility with the pgpool-II replication middleware,
that doesn't support lo_create() calls.
 2012-01-10 11:12:06 +01:00
Daniele Varrazzo
dcc60131a9 fetchmany accepts None as size, meaning the default arraysize 
without this care, extending fetchmany in subclasses becomes tricky.

Closes ticket #84.
 2012-01-10 01:32:45 +00:00
Daniele Varrazzo
09a8e7bf1f Fixed cursor.fetchmany docstring 2012-01-10 01:28:14 +00:00
Daniele Varrazzo
d6e0b284e7 Map error classes 20 and HV to more specific exceptions 2011-12-16 12:26:27 +00:00
Daniele Varrazzo
ba7a0a3008 Raise DatabaseError instead of error with bad exception informations 
We can actually raise these exceptions in weird situations,
e.g. see ticket #82.
 2011-12-15 17:50:31 +00:00
Daniele Varrazzo
b5de04d2ff Put back a distinct ISOLATION_LEVEL_READ_UNCOMMITTED value 2011-12-15 12:53:48 +00:00
Federico Di Gregorio
d2d94e203f Reverted isolation level values to backward compatible values 
This basically removes the READ UNCOMMITED level (that internally
PostgreSQL maps to READ COMMITED anyway) to keep the numeric values
compattible with old psycopg versions. For full details and discussion
see this thread:
http://archives.postgresql.org/psycopg/2011-12/msg00008.php
 2011-12-15 12:25:19 +01:00
Daniele Varrazzo
088978e7b1 Typo fixed writeable -> writable 
Closes ticket #79.
 2011-12-11 02:55:29 +00:00
Daniele Varrazzo
ad3a198919 Fixed reference leak with arguments referenced more than once in queries 
Plus, some more care in objects life cycle, mostly in exceptions handling.

Closes ticket #81.
 2011-12-11 02:52:06 +00:00
Daniele Varrazzo
d2b67364fd connect() supports generic keyword arguments passed to the dsn 2011-11-17 01:51:25 +00:00
Daniele Varrazzo
dc94a3cb2d Check for connection closed before getting the isolation level 
Closes ticket #74

Also added test to check regressions in isolation_level,
set_isolation_level, set_session, autocommit.
 2011-11-16 23:51:05 +00:00
Daniele Varrazzo
a76d0525d1 Dropped unused variable 2011-11-01 07:42:22 +00:00
Daniele Varrazzo
2671472de8 Dropped leftover extra char, already accounted for before 2011-10-20 11:11:57 +01:00
Daniele Varrazzo
60b49f5c45 Avoid PyOS_snprintf to calculate the copy command buffer size 
On windows it returns -1 instead of sometihing portable. So just ditch
the static buffer and just use a dynamic one to compose the command.

Also squashed a couple of buglets in copy_to: copyfile was decremented
before being set to null, size_t was used instead of Py_ssize_t.
 2011-10-19 21:01:53 +01:00
Daniele Varrazzo
ff8158d7c0 Simplification in the COPY command composition 
Dropped the branch if NULL is specified or not: just use the default \N.

Also fixed copy_from/copy_to docstrings.
 2011-10-19 20:31:09 +01:00
Daniele Varrazzo
dde4c0de3d Decimal adapter registration moved from C to Python 
Fixes Decimal adaptation in sub-interpreter, where the Decimal class has
a different identity from the one in the main interpreter.

Closes ticket #52.
 2011-10-14 22:35:56 +01:00
Daniele Varrazzo
c4e6d7d982 Fixed typecasting of arrays containing consecutive backslashes 2011-09-22 18:14:16 +01:00
Daniele Varrazzo
e3054ac9f3 Added new_array_type() function 
Allows the creation of a generic array typecaster from Python.
 2011-09-22 15:51:21 +01:00
Daniele Varrazzo
8fb08efae7 Allocate dynamically memory for the list of columns in COPY 
Some bloke finds the limit of 8K too restrictive... ticket #68.
 2011-09-12 02:21:59 +01:00
Federico Di Gregorio
880aa07a58 WITH HOLD documentation a argument parsing changes 
Now any true value will do for the withhold parameter.
 2011-08-10 19:21:12 +02:00
Federico Di Gregorio
a59d88c703 Merge remote-tracking branch 'piro/devel' into devel 2011-08-10 18:36:24 +02:00
Federico Di Gregorio
479bdf7458 New 'withhold' parameter for connection.cursor() 2011-08-10 18:25:46 +02:00
Daniele Varrazzo
de6f2ac387 Grab the GIL when checking for errors occurred 
The problem was causing a segfault on BEGIN if the server is disconnected
after the connection is created.
 2011-08-09 11:44:30 +01:00
Federico Di Gregorio
2f6336ea78 First try at curs.withhold implementation 2011-07-05 10:28:34 +02:00
Daniele Varrazzo
d2b28abced Method set_transaction() renamed to set_session() 
In fact it doesn't change "the transaction", as there has to be no
transaction when invoked. The effect instead is to execute SET SESSION
CHARACTERISTICS.
 2011-06-08 14:22:11 +01:00
Daniele Varrazzo
1a51cfe274 Better error message if deferrable is used in PG < 9.1 2011-06-08 10:59:27 +01:00
Daniele Varrazzo
0a1bbb56cd Dropped redundant semicolons at the end of internal queries 
For consistency with other queries, and probably we give less work to do
to the server parser (a ridiculously tiny amount).
 2011-06-08 09:22:35 +01:00
Daniele Varrazzo
6d907df14d Fixed documentation for COPY methods 
The size parameter in copy_from was undocumented (ticket #59).
 2011-06-07 11:16:10 +01:00
Daniele Varrazzo
9b5ac79513 Fixed default size for read copy buffer 
The original commit stated it should have been 8192.
 2011-06-07 11:16:06 +01:00
Daniele Varrazzo
679af4a975 Fixed copyfile refcount in copy_expert 
In case of early error, jumping to exit would have decref'd the borrowed
reference to file.

Issue spotted by Dave Malcolm, thanks!
 2011-06-07 01:20:25 +01:00
Daniele Varrazzo
b6e710b0fc Fixed refcount bug in copy_to() and copy_expert() methods too 2011-06-07 00:08:29 +01:00
Daniele Varrazzo
1888bf41c0 Added patch for refcount bug in copy_from 
By Dave Malcolm. https://bugzilla.redhat.com/show_bug.cgi?id=711095
(slightly edited to increment the refcount before storing the pointer
in the cursor).
 2011-06-07 00:08:29 +01:00
Jason Erickson
dd7ee7093a No strcasecmp function with MSVC 
The MSVC compiler does not have the strcasecmp(x, y) function, which is a
case insensitve string compare function.  Instead, MSVC has a similar function,
lstrcmpi(x, y).  Modified config.h to use this function when building with
MSVC.
 2011-06-05 23:33:28 +01:00
Daniele Varrazzo
442a0606fe Merge branch 'guc-cleanup' into devel 2011-06-05 16:31:25 +01:00
Daniele Varrazzo
709df38d79 Don't clobber an eventual Python exception set by a green thread 2011-06-05 16:30:37 +01:00
Daniele Varrazzo
869d48b6f0 Use the pqpath functions to get/set GUC parameters 
Functions conn_setup(), conn_get_isolation_level(), conn_set_transaction(),
conn_switch_isolation_level(), conn_set_client_encoding() reimplemented
using the pqpath funtitons.

Dropped analogous function in the connection, as it had to take the lock,
thus it was hard to build consistent pieces of functionality with it.
 2011-06-05 16:26:01 +01:00
Daniele Varrazzo
8f876d4b5d Avoid a deadlock using concurrent green threads on the same connection 
Use the async_cursor property to store an indication that something is
running (even if it is not necessarily a cursor running the query).
 2011-06-05 16:22:54 +01:00
Daniele Varrazzo
cf6a4ec062 Added pqpath functions to get/set the value for GUC parameters 
The aim of these function is to allow the connection to make a better use
of the pqpath functions instead of using PQexec for these small things.
Also, the functions are to be called with the connection lock: this makes
composing higher level functions using them easier.
 2011-06-05 15:36:02 +01:00
Daniele Varrazzo
a0d16fcfb2 Avoid a ton of warnings when building on mingw 
mingw doesn't support visibility hidden even if gcc can.
 2011-06-04 22:19:56 +01:00
Daniele Varrazzo
c8ec747903 Don't fail import if mx.DateTime module is not found at import time 
A better fix for ticket #53.
 2011-06-04 14:16:24 +01:00
Daniele Varrazzo
05659c0d16 Cleanup of notice processing 
The function is always called in the context of functions grabbing the
connection lock, so just use the same critical section instead of releasing
and re-acquiring it. It is not a problem as serious as the notifies process
(ticket #55) as the notices are a psycopg structure, not libpq. However the
change allows again processing notices/notifies in the same place,
which makes sense conceptually, plus we save some lock dance.
 2011-06-04 01:49:03 +01:00
Daniele Varrazzo
d9c0b8166f Process notifies when data is received, not when the result is parsed 
Notifies process access the connection, is not limited to the result, so
There is the possibility of loss of protocol sync in multithread programs.

Closes ticket #55.
 2011-06-04 01:31:36 +01:00
Daniele Varrazzo
0eb5e0430e Merge branch 'transaction-control' into devel 2011-06-03 01:48:24 +01:00
Daniele Varrazzo
4d3c6865ee Use only the isolation levels available on old PG versions 2011-06-03 00:40:54 +01:00
Daniele Varrazzo
c2d1f1f2e6 Dropped isolation level from the connection object 
Don't issue a SET TRANSACTION ISOLATION LEVEL at every begin: use PG's
GUC default, eventually set by set_transaction.

Dropped the last query at connection, yay!

Method set_isolation_level() and property isolation_level refactored using
the new structures, keeping the previous semantic.
 2011-06-03 00:10:24 +01:00
Daniele Varrazzo
389f2cf1d0 Added autocommit property on connection 2011-06-02 01:16:22 +01:00
Daniele Varrazzo
ea03ffbf76 Added partial implementation for set_transaction 
autocommit to be implemented yet.
 2011-06-01 09:07:02 +01:00
Daniele Varrazzo
531292bca5 Merge branch 'neg-escape' into devel 2011-05-30 22:09:23 +01:00
Daniele Varrazzo
281427f450 Fixed escape for negative numbers prefixed by minus operator 
Closes ticket #57.
 2011-05-30 22:00:20 +01:00
Daniele Varrazzo
19ec8809fd Use all the isolation levels accepted by PostgreSQL 
In PG 9.1 repeatable read and serializable are distinct levels.
 2011-05-11 13:01:20 +01:00
Daniele Varrazzo
834c7d1288 Fixed a few docstrings mixed up 2011-05-11 12:59:31 +01:00
Daniele Varrazzo
c08799b0b0 Fixed SystemError clobbering libpq errors raised without SQLSTATE 
Bug vivisectioned by Eric Snow
<http://archives.postgresql.org/psycopg/2011-04/msg00019.php>.
 2011-04-24 02:59:28 +01:00
Daniele Varrazzo
19653a88ec Store a normalized version of the PG encoding in the connection 
This way looking up into extensions.encodings will not break.
 2011-04-08 13:48:11 +01:00
Daniele Varrazzo
88803695ac Normalize the encoding name at connection 
The encoding can be set by PGCLIENTENCODING, which may be an alternative
spelling. Bug reported by Peter Eisentraut.

At this point the idea of considering one of the random spellings such as
EUC_CN as somewhat "blessed" is debunked. So just store the cleaned-up
version of the encoding in the mapping table. Note that the cleaned-up
version was needed by the unicode adapter: this requirement has been
surpassed as the connection now contains a copy of the Python codec name
set whenever the client encoding is set.
 2011-04-08 13:48:11 +01:00
Daniele Varrazzo
e0cd6f0f00 Added tests for our own bytea parser 
Because the parse function is not supposed to be exposed in Python,
use ctypes to directly inspect the C function.
 2011-03-26 12:59:14 +00:00
Daniele Varrazzo
66c543b16c Parse bytea output format ourselves instead of using the libpq 
PG 9.0 uses the hex format by default, and clients < 9.0 can't parse that
format, requiring client update and great care in what is linked at runtime,
and generally giving headache to users and transitively us.
 2011-03-26 12:59:14 +00:00
Daniele Varrazzo
fcbe0466a6 Correctly detect an empty query sent to the backend 
Closes ticket #46.
 2011-03-04 20:30:43 +00:00
Daniele Varrazzo
f1d69f6dec Fixed detection of empty error from pq_raise 
Avoid a system error in case err is set to an empty string.
 2011-03-04 20:20:56 +00:00
Daniele Varrazzo
194447fbbf Merge branch 'msvc-cleanup' into devel 
Merged Jason Erickson devel branch after collapsing a few commits together
where it made sense.
 2011-02-25 10:29:19 +00:00
Jason Erickson
7c2fa77c4b Windows MSVC: Fixed warning of incompatible types 
Fixed incompatible type warning from XidObject * to PyObject * by
casting.
 2011-02-25 10:25:12 +00:00
Jason Erickson
961e855bbd Windows MSVC: Fix Compiler Warning: getpid 
Fix a compiler warning when using PSYCOPG_DEBUG on MSVC where getpid is
undefined.
 2011-02-25 10:25:12 +00:00
Jason Erickson
631883f62f Windows MSVC: Fix data loss compiler warnings 
Fixed MSVC compiler warnings where it was indicating a conversion
from a larger data type to smaller data type might have data loss.
 2011-02-25 10:25:12 +00:00
Daniele Varrazzo
29ac03ef04 Check for failed allocation in the notice callback 2011-02-24 10:14:06 +00:00
Daniele Varrazzo
894d3f653c Raise an exception if the libpq fails to decode bytea in hex format 2011-02-23 14:04:27 +00:00
Daniele Varrazzo
f96b68d8c6 Fixed docstring for connection.lobject() 2011-02-23 01:53:56 +00:00
Daniele Varrazzo
1db9c9b8ce The cursor name can be a non-valid PostgreSQL identifier 2011-02-23 01:53:25 +00:00
Daniele Varrazzo
66555c5f11 Fixed call of memory functions without the GIL 2011-02-23 00:48:10 +00:00
Daniele Varrazzo
20f714f17c Added error checking to _pq_fetch_tuples() 2011-02-23 00:32:23 +00:00
Daniele Varrazzo
c1715f66fe More careful memory management 
- Check return value of PyErr_Malloc and set an exception in case of error
- Avoid exposing variables with refcount 0 as connection attributes.
- PyErr_Free guards itself for NULL input
 2011-02-23 00:32:23 +00:00
Daniele Varrazzo
beffb02d56 Stricter declaration and correct use of psyco_set_error 
It has long been used in wrong ways, with the function receiving a
connection or lobject instead of a cursor. It has always been unnoticed
(nobody has noticed the wrong object attached to the exception in the
wrong cases) but it started crashing the interpreter with Python 3.2 on
Windows.

Thanks to Jason Erickson for finding the problem and helping fixing it.
 2011-02-21 00:31:09 +00:00
Daniele Varrazzo
d74f777339 Check for memory errors in the connection init 2011-02-20 18:57:04 +00:00
Benjamin Poulain
5ee60571a5 Add a type converter to handle untyped empty arrays. 
Empty array can be returned untyped by postgres. To handle
this case, a special handler is added for the type UNKNOWNOID.
If the value return by the database is strictly equal to "{}",
the value is converted. Otherwise, the conversion fallback on
the default handler.
 2011-02-20 12:28:10 +00:00
Daniele Varrazzo
4bc4f85229 Merge remote branch 'jason/devel' into devel 
Conflicts:
	psycopg/notify_type.c
 2011-02-19 16:25:31 +00:00
Daniele Varrazzo
556b4d461e Documentation cleanup 
Added several links to the Python documentation using the 'intersphinx'
extension.
 2011-02-19 16:16:28 +00:00
Daniele Varrazzo
6098ced761 Assume there may be files returning Unicode in Python 2 too 
This is the case in Python 2.7 with files implementing io.TextIOBase.
 2011-02-19 14:24:15 +00:00
Daniele Varrazzo
8a08114314 Correctly handle exceptions with non-ascii chars in the message 
Previous implementation would have barfed in canse of non-utf-8 data in
the message.
 2011-02-19 14:16:53 +00:00
Jason Erickson
4bc47d47a5 Python 3.2 hash() return value is arch dependant 
Python 3.2 hash() function will now return a 64bit value when run on a 64bit
architecture, where as previously, it would always return a 32bit value.
Modified the code to use the now Py_hash_t typedef and for Python versions
less than 3.2, hard code Py_hash_t to long and Py_uhash_t to unsigned long.
 2011-02-18 23:57:25 -07:00
Daniele Varrazzo
837f1e5e4f Set hidden visibility to a few functions not public 2011-02-19 00:25:23 +00:00
Daniele Varrazzo
c620f18be1 Provide cursor.description as named tuple if possible 
If namedtuple() is not available, use regular tuples.
 2011-02-19 00:05:43 +00:00
Daniele Varrazzo
836f8a1aa1 Make Binary(None) work as expected, adapting to NULL 
Issue reported by Stefano Dal Pra.
 2011-02-18 14:19:57 +00:00
Daniele Varrazzo
99b3c72312 Some cleanup in mogrify 
- Raise an exception on incomplete placeholders.
- Minor speedups.
- Don't change the string in place (??!!) if the placeholder is not s
  and the value is null.

The latter point can be done because downstream we don't accept anything
different from s anyway (in the Bytes_Format function).

Notice that now the format string is constant whatever the arguments.
This means that executemany is still more inefficient than it should be
as mogrify may work only on the parameters. However this is an
implementation only worthwhile if we start supporting real parameters.

Let's talk about that for the next release.
 2011-02-18 02:33:42 +00:00
Daniele Varrazzo
b6d6fbbe8c Use a global object for NULL 
Small optimization as NULL is a frequent value to build.
 2011-02-17 20:09:52 +00:00
Daniele Varrazzo
c51165e2aa Indentation fixed 2011-02-17 19:53:04 +00:00
Daniele Varrazzo
63ac6cdde5 Added cursor.itersize 
The value is used to control the number of records to fetch per network
roundtrip in named cursors iteration. Used to avoid the inefficient
arraysize default of 1 without giving this value the magic meaning of
2000.
 2011-02-17 12:36:02 +00:00
Daniele Varrazzo
bccbcf42d0 Added adaptation for objects supporting the new-style buffer interface 
Supporting this interface is required to adapt memoryview on Python 2.7 as they
don't support the old style. But because the old style is long deprecated it
makes sense to start supporting the new one.
 2011-02-16 01:30:25 +00:00
Daniele Varrazzo
3ae2f221b3 Adapt bytearray and memoryview to bytes if available 2011-02-15 17:30:43 +00:00
Daniele Varrazzo
e4a84b9ce9 Fixed error message on Binary(str) in Python 3 2011-02-15 15:53:07 +00:00
Daniele Varrazzo
c1fe0b675a Dropped correct roundtrip of empty array 
The feature in itself is not extremely useful and instead PostgreSQL is
not always able to cast away from text[], which is a regression see
(ticket #42).
 2011-02-15 10:27:47 +00:00
Daniele Varrazzo
8ab7fa596c Merge remote branch 'jason/devel' into devel 2011-02-12 20:19:07 +00:00
Marti Raudsepp
9cb72a38da Add negative infinity support for Python->pg conversion 2011-02-12 20:19:02 +00:00
Jason Erickson
560f52106c Windows manifest changes based upon architecture 
Apparently, using * for the architecture has the potential to not work on
on some amd64 systems.  Added checks and split the manifest based upon
architecture.
 2011-02-11 10:01:38 -07:00
Jason Erickson
b075017ad9 Pulled down changes from dvarrazzo branch on gh 
Pulled the master branch from of Daniele's psycopg branch on github and
merged the changes.
 2011-02-10 15:59:31 -07:00
Jason Erickson
8d28509f49 Change win32 build to reinsert VC Library Manifest 
Added a change at the end of the build process that would reinsert the VC library manifest.  This patch will fix issues when an embedded program does not have a manifest pointing to the VC 2008 runtime library, such as in an apache/mod_python situation.

Signed-off-by: Jason Erickson <jerickso@stickpeople.com>
 2011-02-10 13:17:14 -07:00
Federico Di Gregorio
da27142882 Merge branch 'python3' into python2 2011-02-06 16:47:05 +01:00
Daniele Varrazzo
fab31e9441 Fetch 'arraysize' records per roundtrip in named cursors iteration 
Closes ticket #33.
 2011-02-05 15:24:00 +01:00
Daniele Varrazzo
b544354db2 COPY sends unicode to a file if it derives from io.TextIoBase 
Fixes ticket #36.
 2011-02-05 15:12:37 +01:00
Daniele Varrazzo
b358c54f02 More efficient cursor.iter: fetch many records at time. 2011-02-04 12:22:07 +00:00
Daniele Varrazzo
d40b394c50 Merge branch 'python2' into python3 2011-02-01 02:27:45 +00:00
Daniele Varrazzo
a2dcf504b5 Hopefully really fixed crash in datetime adapter. 
Verified with gcc 4.3. Notice that on gcc 4.4 the bug was not present.
 2011-02-01 02:00:47 +00:00
Daniele Varrazzo
789dda1173 lobject read and write can deal with both bytes and unicode 
On write, unicode is encoded in connection encoding.
On read, respect the lobject mode 't' or 'b'.
 2011-01-10 00:46:51 +00:00
Daniele Varrazzo
ba1d77a297 Large object mode parsing refactored 
Added parsing of text/binary mode.
 2011-01-10 00:46:51 +00:00
Daniele Varrazzo
79048ff19a Merge branch 'python2' into python3 
Conflicts:
	NEWS-2.3
	tests/__init__.py
	tests/test_lobject.py
	tests/test_quote.py
	tests/testutils.py
 2011-01-10 00:46:02 +00:00
Daniele Varrazzo
48588e5f69 Invalidate large objects after a two-phase commit operation 2011-01-10 00:20:55 +00:00
Daniele Varrazzo
935c25730a Fixed segfault in large object close. 
Check that the connection is not closed/faulty before attempting lo_close.
 2011-01-10 00:11:14 +00:00
Daniele Varrazzo
9deb16484d Don't define a CObject API in Python 3.2 
The API is not available: a PyCapsule should be used. Nobody seems
needing it for now.
 2011-01-08 01:24:16 +00:00
Daniele Varrazzo
b8c8cddc2d Fixed argument parsing in lobject.read 
Using an int instead of a Py_ssize_t randomly crashed Python 3.1 64 bit.
 2011-01-04 02:27:02 +01:00
Daniele Varrazzo
80bd6e2794 Merge branch 'python2' into python3 
Conflicts:
	NEWS-2.3
	psycopg/connection_type.c
	tests/test_connection.py
	tests/types_basic.py
 2011-01-03 21:43:02 +01:00
Daniele Varrazzo
abd7991968 Fixed refcount bug with connection destroyed in a 2PC transaction 2011-01-03 21:34:49 +01:00
Daniele Varrazzo
5f6e773575 Broken circular reference in async execution 
If a connection is destroyed before an async operation is completed, the
`async_cursor` member creates a reference loop, leaving the connection and
the cursor alive. `async_cursor` is now a weak reference.
 2011-01-03 21:34:49 +01:00
Daniele Varrazzo
19ff51ae75 The cursor is weakly referenceable 2011-01-03 21:34:49 +01:00
Daniele Varrazzo
04cf90cc21 The connection is weakly referenceable 2011-01-03 21:34:49 +01:00
Daniele Varrazzo
5888b03608 Fixed "historical" reference leak in TimestampFromTicks 
Added an internal function with C signature to avoid the creation of a
tuple to be later unpacked. When the tuple was decref'd, Python 2.4 64
bits regularly segfaulted; Python 2.5 less regularly; don't know about
other versions.
 2011-01-03 21:34:49 +01:00
Daniele Varrazzo
39dd577c90 Use the proper printf placeholders to avoid warnings on 64 bit builds 2011-01-03 20:44:57 +01:00
Daniele Varrazzo
b276e3b05d Fixed compiling on Python versions before 2.6 
Added a few macros not defined in Py 2.4. Don't know about 2.5.
 2011-01-03 19:18:50 +01:00
Daniele Varrazzo
9eae66e8cf Added Py3 compatibility macro for Py_TPFLAGS_HAVE_WEAKREFS 2011-01-03 16:56:26 +01:00
Daniele Varrazzo
929d62053a Merge branch 'python2' into python3 
Conflicts:
	NEWS-2.3
	setup.py
 2011-01-02 03:28:00 +01:00
Daniele Varrazzo
159cda3688 Added cursor.cast() method 
The method exposes the typecasters lookup algorithm. Useful to create
recursive typecasters.
 2011-01-01 22:55:10 +01:00
Daniele Varrazzo
3e94375cf7 Merge branch 'python2' into python3 
Conflicts:
	ChangeLog
	NEWS-2.3
	lib/extensions.py
	psycopg/microprotocols.c
	setup.py
 2011-01-01 17:14:54 +01:00
Daniele Varrazzo
30921b58a8 Dropped conversion from None to NULL in adapter. 
It was an error as the string is not conform to the protocol. The error
is masked by the None fast path in _mogrify, but surfaces when adapting
a tuple contains a None.
 2010-12-31 18:55:33 +01:00
Daniele Varrazzo
ac25816962 Use the same Bytes_Format function for both Python 2 and 3. 
This makes the behaviour between the two versions similar. It also have
the effect of a more specific error message in case an user specifies a
placeholder different from 's'.
 2010-12-31 03:22:19 +01:00
Daniele Varrazzo
2930ed3d59 Dropped support for all format specifiers except s in PyBytes_Format. 2010-12-31 03:18:27 +01:00
Daniele Varrazzo
6882ac31d4 Dropped warnings in PyBytes_Format function. 2010-12-31 03:18:27 +01:00
Daniele Varrazzo
96a950d3eb Fixed 2-phase commit support in Python 3. 2010-12-31 03:18:27 +01:00
Daniele Varrazzo
89fb60de4b Column names in copy methods can be unicode. 2010-12-31 03:18:27 +01:00
Daniele Varrazzo
73917c15e1 Fixed COPY FROM to deal with decoded files. 2010-12-31 03:18:27 +01:00
Daniele Varrazzo
bc28cc8b00 Use unicode keys as strings in Py3. 
This fixes pyformat style argument passing. Unicode and bytes don't
compare equal (even if they hash the same).
 2010-12-31 03:18:27 +01:00
Daniele Varrazzo
061079c918 In Py3, decode the tuple values before passing to the typecaster. 
Not sure this is the best way to go: it is now impossible to write a
binary typecaster in Python; furthermore it is the opposite approach of
the codecs, which should return bytes.
 2010-12-31 03:18:27 +01:00
Daniele Varrazzo
f6fefbea64 Function psycopg_ensure_bytes converted in a "filter" stealing a ref. 2010-12-31 03:18:27 +01:00
Daniele Varrazzo
3214c23f51 Fixed adaptation in several adapters. 
The getquoted methods always return bytes. The str() convert this
representation to string on the fly.
 2010-12-31 03:18:27 +01:00
Daniele Varrazzo
2e22eef727 Added utility function to convert bytes to string in Python 3. 2010-12-31 03:18:27 +01:00
Daniele Varrazzo
d3f3f1caf0 Added utility method to return a string in the connection encoding. 
In Py2 the result is plain string, in Py3 an unicode decoded in the
connection encoding.
 2010-12-31 03:18:27 +01:00
Daniele Varrazzo
56e4c2bd55 Redefining the microprotocol on Py3 as returning bytes. 2010-12-31 03:18:26 +01:00
Daniele Varrazzo
014b6a6d5b Use psycopg_ensure_bytes() to unify Py2/3 code paths. 2010-12-31 03:18:26 +01:00
Daniele Varrazzo
89e4d4c7bb Empty lists correctly roundtrip. 2010-12-31 03:18:26 +01:00
Daniele Varrazzo
b5ef5ef21d Added typecasters repr(). 2010-12-31 03:18:26 +01:00
Daniele Varrazzo
b4685bba4a Added utility function to get bytes from a str/unicode. 2010-12-31 03:18:26 +01:00
Daniele Varrazzo
03dde732f6 Datetime adaptation in bytes. 2010-12-31 03:18:26 +01:00
Daniele Varrazzo
87a7ebac10 Query mogrification in bytes. 2010-12-31 03:18:26 +01:00
Daniele Varrazzo
c3196ebd9d Added PyBytes_Format function. 2010-12-31 03:18:26 +01:00
Daniele Varrazzo
cb6b52945b The library can be compiled with Python 3. 
Just compiled! No test run yet and many points to review, marked in the
code.

The patch is largely Martin von Löwis work, simplified after refactoring
in the previous commits and adapted to the new code (as the patch was
originally for Psycopg 2.0.9)
 2010-12-21 04:24:36 +00:00
Daniele Varrazzo
e182201e6e Added Python codec name to the connection. 
This allows dropping repeated dictionary lookups with unicode
query/parameters.
 2010-12-21 04:02:14 +00:00
Daniele Varrazzo
ae06fb03e7 Added psycopg_strdup utility function. 2010-12-21 04:02:14 +00:00
Daniele Varrazzo
a50a91fc7b No need to put connection fields to zero: tp_alloc already did. 2010-12-21 04:02:14 +00:00
Daniele Varrazzo
7b5d80d36d Added a few missing encodings. 
EUC_CN, EUC_JIS_2004, ISO885910, ISO885916, LATIN10, SHIFT_JIS_2004.
 2010-12-21 04:02:14 +00:00
Daniele Varrazzo
657bcb4828 Encodings mapping reordered in a more maintainable order. 2010-12-21 04:02:14 +00:00
Daniele Varrazzo
4635c2aa4f Import structmember/stringobject headers from python.h. 
stringobject is not to be imported with Python 3.
 2010-12-21 04:02:13 +00:00
Daniele Varrazzo
9b30147341 Using PyVarObject_HEAD_INIT macro. 2010-12-21 04:02:13 +00:00
Daniele Varrazzo
8dfa9915eb Using Py_TYPE and Py_REFCNT macros. 2010-12-21 04:02:13 +00:00
Daniele Varrazzo
2196ff5488 Added a few compatibility macros defined in Py 2.6. 2010-12-21 04:02:13 +00:00
Marti Raudsepp
a08419406c Don't segfault when PQparameterStatus(DateStyle) returns NULL 
pgbouncer for example does not pass on DateStyle.
 2010-12-18 15:27:15 +00:00
Daniele Varrazzo
89f70bdb3c Changed Python const RO -> READONLY. 2010-12-12 13:39:32 +00:00
Daniele Varrazzo
6d7916cfe1 Internal imports simplified. 
.c files only need to import psycopg.h: it will in turn import
dependencies from Python and libpq and configure.h. psycopg.h should be
the first to be imported, so the basic imports are not required in
the .h's

As a guideline I'm trying to import from the most specific to the most
generic to detect missing imports in the .h's.
 2010-12-12 13:39:32 +00:00
Daniele Varrazzo
9fa1eac2b4 Dropped unused include file. 2010-12-12 13:39:32 +00:00
Daniele Varrazzo
557e28d744 Dropped interface for two private functions. 
Note: the functions are private because typecast.c imports the .c's of
typecast_[mx]datetime, not the .h's.

Work around the warning for 'skip_until_space' not used with an #ifdef.
Furthermore, those functions are now static.
 2010-12-04 13:51:21 +00:00
Daniele Varrazzo
288f9ee809 Work around CentOS 5.5 x86_64 buld problem. 
Closes ticket #23
 2010-12-04 13:51:21 +00:00
Daniele Varrazzo
0c7b0a943b A prepared connection can't be canceled. 2010-11-28 12:15:26 +00:00
Jan Urbański
751bfa1ea6 Support query cancellation. 
Add a cancel() method do the connection object that will interrupt
the current query using the libpq PQcancel() function.
 2010-11-28 11:50:02 +00:00
Daniele Varrazzo
fdf1ba1aac Dropped notices hack to get COPY errors from V2 protocol. 2010-11-22 00:49:07 +00:00
Daniele Varrazzo
163cf5bfb4 mx.DateTime module initialized as it is supposed to be. 
No need to pass the api pointer around. Dropped compiler warnings.
 2010-11-19 00:28:49 +00:00
Daniele Varrazzo
3e3aa676a9 datetime module initialized as it is supposed to be. 
Dropped compiler warnings. It wouldn't blow up if any api was called.
No need to pass type pointers around.
 2010-11-19 00:28:39 +00:00
Daniele Varrazzo
576b01f6a3 Functions exported to drop warnings. 2010-11-19 00:20:49 +00:00
Daniele Varrazzo
67793ed989 Compiler warning dropped. 2010-11-18 23:13:16 +00:00
Daniele Varrazzo
bcacdc8461 Added enum with possilbe isolation level states. 
Also, general isolation levels cleanup and tests added.
 2010-11-18 00:31:51 +00:00
Daniele Varrazzo
c3c54aab38 Don't clobber exception if conn_switch_isolation_level fails. 
Which shouldn't.
 2010-11-17 01:43:50 +00:00
Daniele Varrazzo
e1a04bec0e Don't inconditionately run a query at connection to set datestyle to ISO. 
Run it only if PQparameterStatus() reports the datestyle is set to
something different.
 2010-11-16 18:15:00 +00:00
Daniele Varrazzo
15bba2966f Don't run a query at connection to detect client encoding. 
Use PQparameterStatus() instead. This is only guaranteed in protocol 3.
 2010-11-16 10:16:52 +00:00
Daniele Varrazzo
a88d7ab424 Dropped support for protocol 2. 
Dropped both the setup constant allowing conditional compiling and the
code specific to V2 protocol (mostly COPY and error handling).
 2010-11-16 01:08:50 +00:00
Daniele Varrazzo
73265e7ece Refuse connection with server with protocol version 2. 
This cuts off server whose version is older than 7.4. But enables us to
remove large portions of code rarely used and tested (e.g. p2 copy) and
will allow us to drop the query we do at each connection to establish
the client encoding and the datestyle.
 2010-11-16 01:01:32 +00:00
Jean-Baptiste Quenot
b8ce51d9c2 Mention connection object in debug logs 2010-11-13 03:56:23 +00:00
Daniele Varrazzo
cd19f9115d Fixed notices order (ticket #9). 2010-11-11 11:54:43 +00:00
Daniele Varrazzo
92ee893f0f Functions unused outside the module marked static. 2010-11-10 15:50:05 +00:00
Daniele Varrazzo
a39e98f9e5 Check the presence of a mro. 
The mere presence of the Py_TPFLAGS_HAVE_CLASS doesn't ensure it is
there.
 2010-11-09 14:28:00 +00:00
Daniele Varrazzo
ed6a4c8b1a Dropped PyArg_ParseTuple() calls in functions taking no arguments. 2010-11-09 03:18:54 +00:00
Daniele Varrazzo
2dc28ee7d8 Less lookups and more efficient calls in microprotocols_adapt(). 
Also more explicit handling of the exception indicator.
 2010-11-09 02:41:43 +00:00
Daniele Varrazzo
753b580d72 Less lookups and more efficient calls in microprotocols_getquoted(). 2010-11-09 02:21:21 +00:00
Daniele Varrazzo
422fede38e Replaced PyObject_CallFunction() with *ObjArgs() where more efficient. 2010-11-09 01:49:22 +00:00
Jason Erickson
c79d20855a Win32 *time_r fix 
Fixed the thread safetyness of the Windows *time_r functions.
 2010-11-09 00:20:58 +00:00
Daniele Varrazzo
d601ab8239 fixed crash in pdecimal_str with a few Python 2.5.x releases. 
is_finite() is not available in 2.5.1, it is in 2.5.5 but is officially
supported only since 2.6.
 2010-11-09 00:14:22 +00:00
Daniele Varrazzo
af3681cc1c Fixed repr for Decimal wrapper. 2010-11-09 00:09:19 +00:00
Daniele Varrazzo
916c172cf7 Dropped file imported by mistake. 2010-11-09 00:09:09 +00:00
Daniele Varrazzo
62d3a1533b Use the adapter of an object superclass if available. 2010-11-08 01:35:06 +00:00
Daniele Varrazzo
225b276de5 Use faster function to build tuples in adaptation. 
Fixed a refcount bug too.
 2010-11-08 01:28:01 +00:00
Daniele Varrazzo
df05ea7a33 Py_TYPE defined as it is in Python 2.6. 2010-11-08 01:28:01 +00:00
Daniele Varrazzo
4125b3fce0 Added compatibility macro for pre 2.6 2010-11-05 12:59:32 +00:00
Daniele Varrazzo
2480f587e1 Added warning note about equivalence inconsistence between Notify and tuples. 2010-11-05 09:34:52 +00:00
Daniele Varrazzo
77c0ab02d8 The Notify type is hashable. 
If there is no payload, hash the same way the equivalent 2-tuple does.
Otherwise hash on the payload too.
 2010-11-05 09:34:52 +00:00
Daniele Varrazzo
56ae1fe4bf Payload default is the empty string. 2010-11-05 09:34:52 +00:00
Daniele Varrazzo
9c58c01e0f Added comparison between Notify objects and Notify or tuple. 
Explicit comparison with the tuple is required if we want to make
Notify() == (pid, channel) work: item access is not enough (and a test
in the suite fails if we get this wrong).
 2010-11-05 09:34:52 +00:00
Daniele Varrazzo
90e0e2f47d Added documentation for the Xid object. 2010-11-05 09:34:50 +00:00
Daniele Varrazzo
3e658c33b5 Ensure unicode is accepted as type for transaction ids. 
We don't do somersaults to ensure people can use snowmen as transaction
ids anyway: it would require passing the connection to xid_ensure and
down below to use the correct encoding.
 2010-11-05 09:34:50 +00:00
Daniele Varrazzo
4f3976681a Added Xid.from_string() constructor. 
It exposes the parsing rules used in restore.
 2010-11-05 09:34:49 +00:00
Daniele Varrazzo
2f29429e88 Expose the Xid object in the extension module. 2010-11-05 09:34:49 +00:00
Daniele Varrazzo
f0f5f095c7 Fixed handling of commit/rollback prepared in green mode 2010-11-05 09:34:49 +00:00
Daniele Varrazzo
5dbeeba0f2 Raise NotSuppoertdError if tpc is used with PostgreSQL < 8.1 2010-11-05 09:34:49 +00:00
Daniele Varrazzo
ab8e145063 Encoding/decoding in base64 refactored. 2010-11-05 09:34:49 +00:00
Daniele Varrazzo
3312897e5d Added str() and repr() for Xid objects. 2010-11-05 09:34:49 +00:00
Daniele Varrazzo
4fdcfe365c xid_get_tid returns a Python string, not a buffer. 2010-11-05 09:34:49 +00:00
Daniele Varrazzo
774be1d616 Dropped XID_UNPARSED: we use format_id = None for PG xact ids. 2010-11-05 09:34:49 +00:00
Daniele Varrazzo
978cac3a1b XA transaction ids can be decoded from PostgreSQL transaction ids. 2010-11-05 09:34:48 +00:00
Daniele Varrazzo
6309e038e5 Dropped pg_xact_id from XidObject 2010-11-05 09:34:48 +00:00
Daniele Varrazzo
c0c116dcc4 Use pgjdbc algorithm to convert XA xids into strings. 2010-11-05 09:34:48 +00:00
Daniele Varrazzo
17d70babb1 Keep the GIL while converting the xid into the PostgreSQL transaction id. 2010-11-05 09:34:48 +00:00
Daniele Varrazzo
56c02b0f94 Added tpc_recover method. 2010-11-05 09:34:48 +00:00
Daniele Varrazzo
09983db6ed Added tpc_commit and tpc_rollback methods. 2010-11-05 09:34:48 +00:00
Daniele Varrazzo
4f66de494b Added tpc_prepare and CONN_STATUS_PREPARED. 2010-11-05 09:34:48 +00:00
Daniele Varrazzo
4588fa50f2 begin and commit can't be called during a two-phase transaction. 2010-11-05 09:34:48 +00:00
Daniele Varrazzo
98c5b1d374 Added tpc_begin() and current xid on connection. 2010-11-05 09:34:48 +00:00
Daniele Varrazzo
60ee39fa3d Copyright of TPC files moved to LGPL V3. 
With the permission of James Henstridge.
 2010-11-05 09:34:47 +00:00
Daniele Varrazzo
8bfd34faf2 hook up two phase commit tests. 
By James Henstridge on 2008-07-24.

Merged from lp:~jamesh/psycopg/two-phase-commit/revision/359
 2010-11-05 09:34:47 +00:00
Daniele Varrazzo
7a9d678050 There is no point in allowing subclasses of Xid. 
By James Henstridge on 2008-07-24.

Merged from lp:~jamesh/psycopg/two-phase-commit/revision/358
 2010-11-05 09:34:47 +00:00
Daniele Varrazzo
22aea9114b implement sequence behaviour, as required for transaction IDs. 
By James Henstridge on 2008-07-24.

Merged from lp:~jamesh/psycopg/two-phase-commit/revision/357
 2010-11-05 09:34:47 +00:00
Daniele Varrazzo
0021e56d80 Added connection.xid() and related objects. 
By James Henstridge on 2008-07-23.

Merged from lp:~jamesh/psycopg/two-phase-commit/revision/356

* psycopg/connection_type.c (psyco_conn_xid): add a
Connection.xid() method that instantiates Xid objects.

* psycopg/psycopgmodule.c (init_psycopg): initialise the Xid
object type.

* psycopg/xid.h:
* psycopg/xid_type.c: Implement a basic transaction ID object for
use in two phase commit.
 2010-11-05 09:34:47 +00:00
Daniele Varrazzo
1a0fca09d9 Added documentation for the Notify object. 2010-11-05 09:34:47 +00:00
Daniele Varrazzo
4ec298e112 Notify object exposed in the extensions module. 2010-11-05 09:34:47 +00:00
Daniele Varrazzo
8db9c7085d Added repr method for Notify object. 2010-11-05 09:34:47 +00:00
Daniele Varrazzo
f435d15c95 Adding Notify object with payload. 2010-11-05 09:34:46 +00:00
Daniele Varrazzo
7e482756c5 Use PQfreemem to free memory allocated by the libpq. 
Bug reported by Anton Kovalev.
 2010-10-08 14:26:40 +01:00
Daniele Varrazzo
6e71b3db05 Dropped PSYCOPG_OWN_QUOTING. 
It was deprecated for version 2.1. There are bugs to be fixed made more
complex by its presence and it doesn't keep into account PostgreSQL 9.0
new binary format.

Time to go!
 2010-10-08 12:27:47 +01:00
Daniele Varrazzo
baf65a0dda Fixed access to freed memory in conn_get_isolation_level(). 
Bug reported by Anton Kovalev.
 2010-10-08 12:02:53 +01:00
Daniele Varrazzo
bc2aefeacf cursor.mogrify() accepts unicode queries. 2010-10-05 03:13:44 +01:00
Daniele Varrazzo
5b4d366f4e Common code in execute() and mogrify() merged. 2010-10-05 01:43:23 +01:00
Daniele Varrazzo
2081ceffde Don't execute a ROLLBACK on close()/GC. 
The command wasn't sent since 2.2.0 due to a bug, but after a ML
discussion this behaviour proved more correct so the bug has become a
feature.
 2010-09-23 23:28:30 +01:00
Federico Di Gregorio
557608f715 Applied spelling fixes patch from Peter Eisentraut 2010-07-18 12:14:46 +02:00
Daniele Varrazzo
deb2d9050f Exceptions raised by the columns iterator propagated. 2010-07-10 00:12:50 +01:00
Daniele Varrazzo
3773b50bb3 Propagate iterable exceptions to the executemany caller. 2010-07-09 23:22:35 +01:00
Daniele Varrazzo
305d86f38a Handle time zones with seconds in the UTC offset. 
The offset is rounded to the nearest minute.
 2010-05-20 02:16:04 +01:00
Jason Erickson
1f1133b871 Micrsoft Visual C Compile Fix 
Moved the declarations of variables/pointers to the beginning of the block, where MSVC wants them.
 2010-05-17 09:06:14 +02:00
Federico Di Gregorio
6521fb5a44 Fixed error related to calling C typecasters from Python ones 2010-05-15 19:34:56 +02:00
Daniele Varrazzo
c46a1dad63 Fixed TimeFromTicks for second values > 59.5. 2010-05-15 14:27:24 +01:00
Daniele Varrazzo
afea19651c Fixed TimestampFromTicks for second values > 59.5. 
Bug reported and fixed by Jozsef Szalay on 2010-05-06 at 14:11:59.999920.
 2010-05-15 14:27:24 +01:00
Daniele Varrazzo
5e3f240ac9 Cast datetime into timestamptz when the tzinfo field is set. 2010-05-15 14:27:23 +01:00
Daniele Varrazzo
bf1c76b792 Explicit cast of the SQL representation of time-related objects. 
Allow the objects to be recognized as the proper type by Postgres in not
strong contexts: problem reported by Peter Eisentraut.

Added tests to check the types are respected in a complete Py -> PG ->
Py roundtrip without context.
 2010-05-15 14:27:23 +01:00
Daniele Varrazzo
e8c2a14362 Added typecasters for arrays of specific MX/Py time-related types. 2010-05-15 14:27:23 +01:00
Daniele Varrazzo
c8aef26e15 Adapt buffer objects using an explicit cast on the string literal. 
Don't rely on Postgres casting the literal according to the context:
this doesn't work e.g. passing the object as function argument where a
function with the same name but taking a text exists. It doesn't work
either when the object is in an ARRAY construct.

Added test to check the type is respected in a complete Py -> PG -> Py
roundtrip without context.

Bug and solution reported by Peter Eisentraut.
 2010-05-15 14:27:23 +01:00
Daniele Varrazzo
73db6bee01 Keep the connection in blocking mode unless it is not async. 
Dropped set/unset nonblocking mode for copy and lobject operations:
lobjects don't work in nonblocking mode so they will hardly be supported
in green/async branches.  Support for copy is still feasible, but it
will be done in other code paths (called by poll).
 2010-05-09 20:37:48 +01:00
Daniele Varrazzo
e29424a230 Dropped large object support when psycopg is in green mode. 
Async mode and large object are not compatible, albeit I haven't found
an authoritative source yet.
 2010-05-09 20:34:02 +01:00
Daniele Varrazzo
2596cf7125 Simplified psyco_wait() function interface. 2010-04-25 23:08:08 +01:00
Daniele Varrazzo
8efa1fa6af Dropped compiling warning. 2010-04-25 23:08:08 +01:00
Daniele Varrazzo
cb40342afa poll implementation for async, sync and green connection unified. 2010-04-23 13:28:49 +01:00
Daniele Varrazzo
0da4befe78 Added function to advance the state after flush attempt. 2010-04-23 13:28:38 +01:00
Daniele Varrazzo
f3b3483df3 Added a function to advance the state after a read attempt. 
Dropped calls to PQisBusy/PQconsumeInput in the connection code.
 2010-04-23 13:24:36 +01:00
Daniele Varrazzo
1edbd16577 Dropped curs_get_last_result function. 
The result is read from a pqpath function and put on the async_cursor by
a function that also checks for is presence and dismiss it after its
usage.
 2010-04-23 13:24:35 +01:00
Daniele Varrazzo
a66de9808f Added 'pq_get_last_result()' function. 
The function reads the last result after an asynchronous query.
 2010-04-23 13:24:35 +01:00
Daniele Varrazzo
5be0fc52ca Reusable parts of the green polling refactored out. 
The functions _conn_poll_connecting() and _conn_poll_query() will be
usable by the async connections too.
 2010-04-23 13:24:35 +01:00
Daniele Varrazzo
df959c20be Making sync and async connection setup somewhat more consistent. 2010-04-23 13:24:35 +01:00
Daniele Varrazzo
d915cb12a8 Functions to poll in ready state moved on the connection. 2010-04-23 13:24:35 +01:00
Daniele Varrazzo
9f5d835de6 Functions conn_poll_* renamed to conn_poll_connect_*. 
These functions are used only during async connections, not for polling
afterwards.
 2010-04-23 13:24:35 +01:00
Daniele Varrazzo
7af0bf0b54 Set the async_status to ASYNC_DONE after a query with wait callback. 
Failing in doing that broke notifications reception.

The responsibility for changing the async_status has been moved to the
poll function: this is consistent with how the async branch is
implemented.

With this commit all the test suite passes in "green" mode.
 2010-04-21 16:20:16 +01:00
Daniele Varrazzo
d71520db9a Read the server version only after connection established. 2010-04-21 15:31:07 +01:00
Daniele Varrazzo
8fed0aa57d Forbid COPY-related methods in green mode. 
With the current implementation, at best they would silently block. They
actually hang everything.

Implementation posponed after some refactoring of the polling system,
because it will be probably possible to provide an implementation for
'poll()' during COPY which is good for both async and green modes.
 2010-04-21 15:21:33 +01:00
Daniele Varrazzo
c1f0d4d46c Read notifications when polling an idle sync connection. 2010-04-21 15:21:33 +01:00
Daniele Varrazzo
4bb0c3bdf6 Use the wait callback in the connection setup queries. 2010-04-21 15:21:33 +01:00
Daniele Varrazzo
0ec73a18b4 'pq_execute_command_locked()' calls the wait callback if set. 
The function is called without holding the GIL. Because it is necessary
to execute the Python callback if set, we need to re-acquire the GIL and
tnen release it again. In order to correctly bookkeep the thread state,
the pointer of the _save variable is passed to the function.
 2010-04-21 15:21:33 +01:00
Daniele Varrazzo
7a06c0455b Try to restore the connection state after a wait callback error. 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
a54932ee9c Added documentation for the green features. 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
127f92f9db Changed 'psyco_wait()' to only take the connection. 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
0dd5d3f1d9 Correctly flush async queries in 'green' mode. 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
8ba0f00d21 Added 'get_wait_callback()' function. 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
a7c41a0388 'pq_execute()' uses the user wait callback if provided. 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
55270cab9f 'poll()' can raise a Python exception in case of error. 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
442b3b2f6c Added 'psyco_exec_green()' to replace 'PQexec' using the user-provided block function. 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
6dd6bee217 Added async read support to 'connection.poll()'. 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
1607f89bb1 Avoid the locks dance when reading the connection socket number. 
It is a fast and non critical operation, so the lock is not required.
Furthermore, called from the wait callback creates a deadlock.
 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
389ad08965 Use the wait callback during connection if set. 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
04d66a6c82 Added poll implementation for sync connection with external wait. 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
1446f046e9 Added wait callback and functions to deal with it. 2010-04-21 15:21:32 +01:00
Daniele Varrazzo
02a28ff028 Notifications are automatically read after each query. 
Added tests for basic notifications process.
 2010-04-21 15:09:14 +01:00
Daniele Varrazzo
12ef826d50 Dropped redundant call to conn_notice_process(). 
If the connection is sync, notices will be processed by pq_fetch()
downstream.

If the connection is async, here we have only sent the query: no result
is ready yet, and neither notices have had a chance to arrive: they will
be retrieved later by pq_is_busy().

Added tests to check the above statement don't break.
 2010-04-21 15:09:14 +01:00
Daniele Varrazzo
bcfcea4b49 Fixed reference leak in notify reception. 2010-04-21 15:09:13 +01:00
Daniele Varrazzo
6fecc36b7f Connection method 'executing()' renamed to 'isexecuting()'. 2010-04-21 15:09:13 +01:00
Daniele Varrazzo
7ee09353ec Connection method 'issync()' changed into the attribute 'async'. 2010-04-21 15:09:13 +01:00
Jan Urbański
067161d5f3 Remove fileno() and poll() from cursors 
Instead, the code should be using the fileno() and poll() methods of
the cursor's connection. Handle the case when poll() is called on an
already built connection as a request to poll the asynchronous query
(if there is one) and get NOTIFY events.
Update the tests to reflect that change, add a test for NOTIFY.
 2010-04-21 15:09:13 +01:00
Jan Urbański
c4ebc0f702 Handle errors in asynchronous queries. 
Do it by keeping the reference to the last PGresult in the cursor and
calling pq_fetch() before ending the asynchronous execution. This
takes care of handling the possible error state of the PGresult and
also allows the removal of the needsfetch flag, since now after
execution ends the results are already fetched and parsed.
 2010-04-20 23:01:01 +01:00
Jan Urbański
5f4ef5da13 Make connection.executing() be True during an async connection attempt. 2010-04-14 10:00:37 +02:00
Jan Urbański
463724690c Set the execution status to ASYNC_READ after flushing all output. 
Without this a query that did not get flushed completely to the server
would cause cursor.poll() to always go into the curs_poll_send()
branch even if it was retuning ASYNC_READ.

Bug report by Daniele Varrazzo.
 2010-04-14 09:56:44 +02:00
Jan Urbański
4574bde3a2 Check for possible errors from PQflush. 
Noticed by Daniele Varrazzo.
 2010-04-14 09:56:44 +02:00
Jan Urbański
61f890c317 Set the initial connection status to CONN_STATUS_SETUP. 
Without it the changes in e79528f9aa1f7457573720fce5548302fefffd37 are
never used.
 2010-04-14 09:56:44 +02:00
Jan Urbański
6108e4dc92 Make the first poll() of an asynchronous connection return POLL_WRITE. 
This hides from the user the libpq's implementation detail of
requiring the first select() to wait for the connection socket to
become writable and makes it possible to have a uniform select loop
for both cursors and connections, in which you always start by polling
the object and then acting according to the result from poll().

Idea and implementation by Daniele Varrazzo.
 2010-04-14 09:56:44 +02:00
Jan Urbański
4afc1baf35 Make polling a cursor that's not in an async query raise an exception. 
If there is an asynchronous query, polling a cursor that did not
initiate it will raise an exception. Polling while there is no
asynchronous query underway still works, because the user needs to
have a way to get asynchronous NOTIFYs.
 2010-04-14 09:56:44 +02:00
Federico Di Gregorio
a90935930b Merge remote branch 'piro/fix22' into python2 2010-04-14 08:57:30 +02:00
Federico Di Gregorio
eaa97def73 Applied GC patch from Michael Tharp 2010-04-14 08:50:16 +02:00
Daniele Varrazzo
88fe5f91d2 Fixed parameters parsing in the connection constructor. 
Keyword arguments were silently discarded.
 2010-04-11 00:30:54 +01:00
Federico Di Gregorio
b8953cb345 Fixed Python 2.4 compatibility breakage 2010-04-09 12:53:38 +02:00
Daniele Varrazzo
e8286084e9 Free the GIL in blocking operations in V2 COPY FROM. 2010-04-07 23:52:53 +01:00
Daniele Varrazzo
aeec583ff1 Ensure running COPY in blocking mode. 2010-04-07 23:41:32 +01:00
Daniele Varrazzo
7e0dcfdda5 Added pq_set_non_blocking utility function. 2010-04-07 23:41:18 +01:00
Daniele Varrazzo
d339466c6a Speedup of the copy functions. 
Python objects are evaluated only once outside the I/O loops.
 2010-04-07 18:22:31 +01:00
Daniele Varrazzo
5a75d693a3 Dropped warnings when compiling on Python 2.5 2010-04-07 18:12:24 +01:00
Daniele Varrazzo
c75ddea895 Pass the async parameter to the connection only if nonzero. 
This avoids breaking connection subclasses redefining '__init__' (at
least when used sync, which is good for most of the people).
 2010-04-05 18:31:09 +01:00
Daniele Varrazzo
c486136df7 Delegate the destruction of the PGconn to the connection finalization. 
This way we can have an usable connection before complete
initialization: this will be required to allow a Python callback to
interact with connect().
 2010-04-05 18:19:37 +01:00
Federico Di Gregorio
c1a24f4ca2 NOTIFYs fix and poll status changes 
POLL_OK has been changed from 3 to 0 to let the user specify a short loop
just as "if not curs.poll()" instead of having to check for write and read
separately. For an example of this, see examples/notify.py.
 2010-04-05 16:52:25 +02:00
Jan Urbański
31f60be000 Be a bit more explicit 2010-04-05 16:27:59 +02:00
Jan Urbański
58eb868db6 Avoid the possibility when curs_get_last_result would block 
It was trying to get all pending results from the connection and if
the client sent many and anyone except the first one would not be
immediately available the loop in curs_get_last_result would call
PQgetResult blockingly.
Avoid that by calling PQisBusy every time and telling the client to
wait for more data if it returns 1.
 2010-04-05 16:27:39 +02:00
Jan Urbański
25a609c9a7 Handle initial connection queries being sent partly 
The CONN_STATUS_SENT_* statuses were not being handled at all, and
they indicate that a query has been sent, but not fully, so the client
should wait for the socket to become writable again and flush the output.
 2010-04-05 16:25:17 +02:00
Jan Urbański
b99eac18f8 Make cursor.scroll() raise an exception in asynchronous mode 
The exception is raised if an asynchronous query is in progress,
otherwise it will behave normally (since the data is already on the
client side).
 2010-04-05 16:24:45 +02:00
Jan Urbański
727670c269 Make some methods raise an exception in asynchronous mode 
The methods changed are connection.commit(), rollback(), reset(),
set_isolation_level(), set_client_encoding(), lobject(), cursor(str)
as well as cursor.execute() and cursor.callproc() if another query is
in progress and cursor.executemany(), cursor.copy_{from,to,expert)().
 2010-04-05 16:23:44 +02:00