Mike Bayer [Sat, 26 Jan 2019 19:53:45 +0000 (14:53 -0500)]
Ensure of_type subclass taken into account with wildcards
Fixed a regression in 1.2 where a wildcard/load_only loader option would
not work correctly against a loader path where of_type() were used to limit
to a particular subclass. The fix only works for of_type() of a simple
subclass so far, not a with_polymorphic entity which will be addressed in a
separate issue; it is unlikely this latter case was working previously.
Since we ensure that the entity is broken out into its superclasses
when a wilcard is encountered, we can limit the entity path to the
specific entity given in this case.
Within this issue some additional issues with with_polymorphic()
loaders were found which will be addressed in #4469.
Mike Bayer [Wed, 9 Jan 2019 07:01:16 +0000 (02:01 -0500)]
Improve error messages in the area of loader options
Improved error messages emitted by the ORM in the area of loader option
traversal. This includes early detection of mis-matched loader strategies
along with a clearer explanation why these strategies don't match.
Lele Gaifax [Mon, 14 Jan 2019 16:26:33 +0000 (11:26 -0500)]
Fix many spell glitches
This affects mostly docstrings, except in orm/events.py::dispose_collection()
where one parameter gets renamed: given that the method is
empty, it seemed reasonable to me to fix that too.
Mike Bayer [Fri, 25 Jan 2019 16:36:54 +0000 (11:36 -0500)]
Fix mssql quote schema warning
The deprecations review didn't include tests of identifier_preparer.quote.force
for backends, so MSSQL slipped through. We have to fully reimplement
the deprecation warning here so that it passes tests which are now
enabled for all backends.
Daniel Lister [Thu, 24 Jan 2019 21:35:16 +0000 (16:35 -0500)]
Add getters for all execution_options
Added accessors for execution options to Core and ORM, via
:meth:`.Query.get_execution_options`,
:meth:`.Connection.get_execution_options`,
:meth:`.Engine.get_execution_options`, and
:meth:`.Executable.get_execution_options`. PR courtesy Daniel Lister.
Mike Bayer [Sun, 30 Dec 2018 01:54:29 +0000 (20:54 -0500)]
Implement relationship to AliasedClass; deprecate non primary mappers
Implemented a new feature whereby the :class:`.AliasedClass` construct can
now be used as the target of a :func:`.relationship`. This allows the
concept of "non primary mappers" to no longer be necessary, as the
:class:`.AliasedClass` is much easier to configure and automatically inherits
all the relationships of the mapped class, as well as preserves the
ability for loader options to work normally.
- introduce new name for mapped_table, "persist_selectable". this is
the selectable that selects against the local mapper and its superclasses,
but does not include columns local only to subclasses.
- relationship gains "entity" which is the mapper or aliasedinsp.
- clarfiy name "entity" vs. "query_entity" in loader strategies.
Mike Bayer [Thu, 24 Jan 2019 21:56:44 +0000 (16:56 -0500)]
Use pg_get_constraintdef for CHECK constraint reflection
Revised the query used when reflecting CHECK constraints to make use of the
``pg_get_constraintdef`` function, as the ``consrc`` column is being
deprecated in PG 12. Thanks to John A Stevenson for the tip.
Mike Bayer [Fri, 21 Dec 2018 03:05:36 +0000 (22:05 -0500)]
Add deprecation warnings to all deprecated APIs
A large change throughout the library has ensured that all objects, parameters,
and behaviors which have been noted as deprecated or legacy now emit
``DeprecationWarning`` warnings when invoked. As the Python 3 interpreter now
defaults to displaying deprecation warnings, as well as that modern test suites
based on tools like tox and pytest tend to display deprecation warnings,
this change should make it easier to note what API features are obsolete.
See the notes added to the changelog and migration notes for further
details.
Added new event hooks :meth:`.QueryEvents.before_compile_update` and
:meth:`.QueryEvents.before_compile_delete` which complement
:meth:`.QueryEvents.before_compile` in the case of the :meth:`.Query.update`
and :meth:`.Query.delete` methods.
Mike Bayer [Sat, 19 Jan 2019 02:30:21 +0000 (21:30 -0500)]
Don't use cx_Oracle.NATIVE_INT in output type handlers
Fixed regression in integer precision logic due to the refactor of the
cx_Oracle dialect in 1.2. We now no longer apply the cx_Oracle.NATIVE_INT
type to result columns sending integer values (detected as positive
precision with scale ==0) which encounters integer overflow issues with
values that go beyond the 32 bit boundary. Instead, the output variable
is left untyped so that cx_Oracle can choose the best option.
Mike Bayer [Fri, 18 Jan 2019 04:38:40 +0000 (23:38 -0500)]
Cleanup with query aliasing
Try to simplify some of the "adapter" stuff in query:
1. identify that join(.., aliased=True) doesn't work if the
right side has no mapper. The adaption of the right side is
done via the mapper with aliased(), so that doesn't effect
a selectable only. raise an error, so we can simplify
the code.
2. build fewer adapter objects. these are confusing to follow
and we should try to figure out exactly what purpose which
one serves where and make that clear.
Mike Bayer [Fri, 18 Jan 2019 06:02:07 +0000 (01:02 -0500)]
Feature mysqlclient
MySQL-Python hasn't been released in about five years and there
is no reason that it should be listed as a viable DBAPI;
all MySQL / MariaDB users should be using mysqlclient or PyMySQL today.
Mike Bayer [Fri, 18 Jan 2019 01:08:10 +0000 (20:08 -0500)]
Adapt single inh criteria more specifically
Fixed issue where when using single-table inheritance in conjunction with a
joined inheritance hierarchy that uses "with polymorphic" loading, the
"single table criteria" for that single-table entity could get confused for
that of other entities from the same hierarchy used in the same query.The
adaption of the "single table criteria" is made more specific to the target
entity to avoid it accidentally getting adapted to other tables in the
query.
Mike Bayer [Thu, 17 Jan 2019 18:16:04 +0000 (13:16 -0500)]
Repair use of deprecated text() typemap, bindparams parameters
These will emit a deprecation warning once
If0ea11a1fc24f9a8029352eeadfc49a7a54c0a1b is merged, modernize these
ahead of time as this should likely be backported to 1.2 as well.
Mike Bayer [Sun, 13 Jan 2019 23:15:52 +0000 (18:15 -0500)]
Render N'' for SQL Server unicode literals
The ``literal_processor`` for the :class:`.Unicode` and
:class:`.UnicodeText` datatypes now render an ``N`` character in front of
the literal string expression as required by SQL Server for Unicode string
values rendered in SQL expressions.
Note that this adds full unicode characters to the standard test suite,
which means we also need to bump MySQL provisioning up to utf8mb4.
Modern installs do not seem to be reproducing the 1271 issue locally,
if it reproduces in CI it would be better for us to skip those ORM-centric
tests for MySQL.
Also remove unused _StringType from SQL Server dialect
Mike Bayer [Mon, 14 Jan 2019 23:53:58 +0000 (18:53 -0500)]
Relax "ambiguous" association proxy restrictions, support Proxy
Fixed issue in association proxy due to :ticket:`3423` which caused the use
of custom :class:`.PropComparator` objects with hybrid attribites, such as
the one demonstrated in the ``dictlike-polymorphic`` example to not
function within an association proxy. The strictness that was added in
:ticket:`3423` has been relaxed, and additional logic to accomodate for
an association proxy that links to a custom hybrid have been added.
Add standalone orm.close_all method and deprecate SessionMaker.close_all
Added a new function :func:`.close_all_sessions` which takes
over the task of the :meth:`.Session.close_all` method, which
is now deprecated as this is confusing as a classmethod.
Pull request courtesy Augustin Trancart.
Mike Bayer [Tue, 8 Jan 2019 22:46:55 +0000 (17:46 -0500)]
use ..deprecated directive w/ version in all cases
These changes should be ported from 1.3 back to 1.0 or
possibly 0.9 to the extent they are relevant in each
version. In 1.3 we hope to turn all deprecation documentation
into warnings.
Mike Bayer [Thu, 10 Jan 2019 17:03:40 +0000 (12:03 -0500)]
Leave bytestring exception messages as bytestrings
Fixed a regression introduced in version 1.2 where a refactor
of the :class:`.SQLAlchemyError` base exception class introduced an
inappropriate coercion of a plain string message into Unicode under
python 2k, which is not handled by the Python interpreter for characters
outside of the platform's encoding (typically ascii). The
:class:`.SQLAlchemyError` class now passes a bytestring through under
Py2K for ``__str__()`` as is the behavior of exception objects in general
under Py2K, does a safe coercion to unicode utf-8 with
backslash fallback for ``__unicode__()``. For Py3K the message is
typically unicode already, but if not is again safe-coerced with utf-8
with backslash fallback for the ``__str__()`` method.
Mike Bayer [Wed, 9 Jan 2019 16:42:02 +0000 (11:42 -0500)]
Skip expression-based index reflection for SQLite
Reflection of an index based on SQL expressions are now skipped with a
warning, in the same way as that of the Postgresql dialect, where we currently
do not support reflecting indexes that have SQL expressions within them.
Previously, an index with columns of None were produced which would break
tools like Alembic.
Mike Bayer [Wed, 9 Jan 2019 16:18:02 +0000 (11:18 -0500)]
Render correct DDL for unsetting table comments
Fixed issue where the DDL emitted for :class:`.DropTableComment`, which
will be used by an upcoming version of Alembic, was incorrect for the MySQL
and Oracle databases.
Mike Bayer [Sun, 6 Jan 2019 06:19:47 +0000 (01:19 -0500)]
Post black reformatting
Applied on top of a pure run of black -l 79 in
I7eda77fed3d8e73df84b3651fd6cfcfe858d4dc9, this set of changes
resolves all remaining flake8 conditions for those codes
we have enabled in setup.cfg.
Included are resolutions for all remaining flake8 issues
including shadowed builtins, long lines, import order, unused
imports, duplicate imports, and docstring issues.
Mike Bayer [Sun, 6 Jan 2019 06:14:26 +0000 (01:14 -0500)]
Run black -l 79 against all source files
This is a straight reformat run using black as is, with no edits
applied at all.
The black run will format code consistently, however in
some cases that are prevalent in SQLAlchemy code it produces
too-long lines. The too-long lines will be resolved in the
following commit that will resolve all remaining flake8 issues
including shadowed builtins, long lines, import order, unused
imports, duplicate imports, and docstring issues.
Mike Bayer [Sun, 6 Jan 2019 03:42:08 +0000 (22:42 -0500)]
Assorted pre-Black fixes
Fixes to the test suite, a few errant imports, and setup.py:
- mysql and postgresql have unused 'json' imports; remove
- postgresql is exporting the 'json' symbol, remove
- make sure setup.py can find __version__ using " or '
- retry logic in provision create database for postgresql fixed
- refactor test_magazine to use cls.tables rather than globals
- remove unused class in test_scoping
- add a comment to test_deprecations that this test suite itself
is deprecated
- don't use mapper() and orm_mapper() in test_unitofwork, just
use mapper()
- remove dupe test_scalar_set_None test in test_attributes
- Python 2.7 and above includes unittest.SkipTest, remove pre-2.7
fallback
- use imported SkipTest in profiling
- declarative test_reflection tests with "reflectable_autoincrement"
already don't run on oracle or firebird; remove conditional logic
for these, which also removes an "id" symbol
- clean up test in test_functions, remove print statement
- remove dupe test_literal_processor_coercion_native_int_out_of_range
in test/sql/test_types.py
- fix psycopg2_hstore ref
Mike Bayer [Fri, 4 Jan 2019 03:28:09 +0000 (22:28 -0500)]
Prep for flake8 refactoring
a few code changes ahead of time to handle some __all__
issues better. also include new flake8 rules, since the
existing flake8 doesn't pass in any case.
Mike Bayer [Fri, 28 Dec 2018 14:33:07 +0000 (09:33 -0500)]
Add missing disambiguation to non-primary mapper example
The mapping to B over a join defines an alternate primary key
based on all the primary key columns in the join unless we
re-define it explicitly. Similarly, people expect that
``.id`` looks the same. make sure these line up with the
old mapping.
Mike Bayer [Sun, 23 Dec 2018 00:16:50 +0000 (19:16 -0500)]
Check collection less than two items remaining before firing scalar backref remove
Fixed long-standing issue where duplicate collection members would cause a
backref to delete the association between the member and its parent object
when one of the duplicates were removed, as occurs as a side effect of
swapping two objects in one statement.
Mike Bayer [Fri, 28 Dec 2018 03:10:45 +0000 (22:10 -0500)]
Call __del() before remove()
The "remove" event for collections is now called before the item is removed
in the case of the ``collection.remove()`` method, as is consistent with the
behavior for most other forms of collection item removal (such as
``__delitem__``, replacement under ``__setitem__``). The ``pop()`` methods
are now the only exception as the target item is not available until after
the pop operation proceeds.
This allows ``remove()`` to be consistent in its behavior with all
the other collection operations, allows the "before_delete" hook
to be local to "pop()" operations only, and removes some method overhead.
We are also looking here to gain some more predictability in terms
of the fix for #1103.
Mike Bayer [Thu, 27 Dec 2018 17:39:15 +0000 (12:39 -0500)]
Implement SynonymProperty.get_history()
Implemented the ``.get_history()`` method, which also implies availability
of :attr:`.AttributeState.history`, for :func:`.synonym` attributes.
Previously, trying to access attribute history via a synonym would raise an
``AttributeError``.
Mike Bayer [Fri, 21 Dec 2018 22:35:12 +0000 (17:35 -0500)]
Maintain compiled_params / replacement_expressions within expanding IN
Fixed issue in "expanding IN" feature where using the same bound parameter
name more than once in a query would lead to a KeyError within the process
of rewriting the parameters in the query.
Mike Bayer [Fri, 21 Dec 2018 16:04:14 +0000 (11:04 -0500)]
Handle PostgreSQL enums in remote schemas
Fixed issue where a :class:`.postgresql.ENUM` or a custom domain present
in a remote schema would not be recognized within column reflection if
the name of the enum/domain or the name of the schema required quoting.
A new parsing scheme now fully parses out quoted or non-quoted tokens
including support for SQL-escaped quotes.
Fixed issue where multiple :class:`.postgresql.ENUM` objects referred to
by the same :class:`.MetaData` object would fail to be created if
multiple objects had the same name under different schema names. The
internal memoization the Postgresql dialect uses to track if it has
created a particular :class:`.postgresql.ENUM` in the database during
a DDL creation sequence now takes schema name into account.
Michael Howitz [Mon, 17 Dec 2018 16:04:44 +0000 (11:04 -0500)]
Omit include of no longer existing files
Warnings are generated during setup.py install such as:
```
Getting distribution for 'SQLAlchemy==1.2.15'.
warning: no files found matching '*.jpg' under directory 'doc'
warning: no files found matching '*.mako' under directory 'doc'
warning: no files found matching 'distribute_setup.py'
warning: no files found matching 'sa2to3.py'
warning: no files found matching 'ez_setup.py'
```
This change removes those expressions from MANIFEST.in.
The current example code does not pass `**kw` down to the `compiler.process` calls, thus the example does not work when invoked with, `literal_binds=True`.
Besides, the calls to `process` each argument twice are wasteful, and reusing the built-in `case` expression instead of hard-coding the SQL statements is slightly nicer overall, isn't it?
Agam Rafaeli [Mon, 17 Dec 2018 16:20:09 +0000 (11:20 -0500)]
Set a reading order to the compatibility import file
The order is as follows:
1. Imports, in the same format of "import X"
2. Members denoting versions of python
3. Members of imports that are shadowed for usage across the system
4. Global variables
5. If conditions of imports. (internally organized by this order)
6. Function definitions
Mike Bayer [Wed, 12 Dec 2018 17:51:20 +0000 (12:51 -0500)]
Document and support nested composites
Composites can behave in a "nested" fashion by defining the
class in that way. To make the constructor more convenient,
a callable can be passed to :func:`.composite` instead of the
class itself. This works now, so add a test to ensure this
pattern remains available.
Chris Wilson [Mon, 10 Dec 2018 21:37:51 +0000 (16:37 -0500)]
Add before_mapper_configured event
This event is intended to allow a specific mapper to be skipped during the
configure step, by returning a value of `.orm.interfaces.EXT_SKIP` which means
the mapper will be skipped within this configure run. The "new mappers" flag
will remain set in this case and the configure operation will occur again.
This event, and its return value, make it possible to query one base while a
different one still needs configuration, which cannot be completed at this
time.
Mike Bayer [Fri, 7 Dec 2018 18:08:29 +0000 (13:08 -0500)]
Take instance into account when determining AssociationProxyInstance
Fixed a regression in 1.3.0b1 caused by :ticket:`3423` where association
proxy objects that access an attribute that's only present on a polymorphic
subclass would raise an ``AttributeError`` even though the actual instance
being accessed was an instance of that subclass.
Mike Bayer [Fri, 7 Dec 2018 21:01:04 +0000 (16:01 -0500)]
Refer to existing of_type when resolving string attribute name
Fixed bug where chaining of mapper options using
:meth:`.RelationshipProperty.of_type` in conjunction with a chained option
that refers to an attribute name by string only would fail to locate the
attribute.
Mike Bayer [Fri, 7 Dec 2018 19:01:23 +0000 (14:01 -0500)]
Add versioned_update_old_row example
A modification to the "turn UPDATE into INSERT" recipe that
also UPDATEs the previous row. The example is using timestamps
to maintain a relationship between two objects.
projects / thirdparty / sqlalchemy / sqlalchemy.git / log
