python/python-textX
SHA256
14
0
Fork 0
forked from pool/python-textX
Code Pull Requests Activity
Files
14523e7c46b7e74c1948b6010d4fd747e86716798e8dee2b43725c5594eabd67
python-textX/python-textX.changes

385 lines
20 KiB
Plaintext
Raw Blame History

-------------------------------------------------------------------
Mon Feb 5 13:29:19 UTC 2024 - Sebastian Wagner <sebix@sebix.at>
- Require Arpeggio >= 2.0.0 fixing build fails
-------------------------------------------------------------------
Mon Sep 11 05:42:48 UTC 2023 - Steve Kowalik <steven.kowalik@suse.com>
- Update to 3.1.1:
* Source distribution packaging issue ([392]). Thanks yurivict@GitHub.
* Fixed RREL lookup in case of multi-meta models (some special cases were not
handled correctly; [#379]).
* Fixed test suite invocation to use `pytest` over `py.test` that stopped
working in pytest-7.2.0. ([#389]). Thanks mgorny@GitHub.
* Changed separator in obj. rule refs from `|` to `:`. Old separator
will still be allowed until version 4.0. ([#385], [#384])
* Removed the dependency on `future` package ([#388]). Thanks mgorny@GitHub.
* Removed vendored `six` library. We don't need 2.x support anymore. Thanks
davidchall@GitHub for reporting the issue ([#390]).
- Switch to pyproject macros.
- Stop using greedy globs in %files.
- Drop a Python 2 requires.
-------------------------------------------------------------------
Tue Mar 22 20:51:24 UTC 2022 - Sebastian Wagner <sebix+novell.com@sebix.at>
- update to version 3.0.0:
- Added:
- Added RREL-'fixed name'-extension, allowing to follow model elements
given a fixed name (e.g. an object defined in a builtin model).
Details described in rrel.md ([#364]).
- Added ability to access the full path of named objects traversed while
resolving a RREL expression ([#304]).
- Added decorator `textx.textxerror_wrap` for object processors to automatically
transform non-TextXErrors to TextXErrors in order to indicate the filename and
position of the element being processed ([#306]).
- Fixed:
- `model_param_defs` on `TextXMetaMetaModel` ([#360]).
- Interpreting of backslash special chars (e.g. `\n`, `\t`) in grammar string
matches ([#323]). Possible **(BIC)** - backslash chars were not interpreted in
grammar files and raw Python strings prior to this fix.
- Exception/error messages ([#320])
- Relaxed assert in model creation enabling some model changes in user classes
([#311])
- Model export to dot in cases where textX object is replaced in the
processor([#301])
- Do not allow "empty" RREL expressions (compare unittests in `test_rrel.py`; [#355])
- Changed:
- Inheritance chain calculation. Possible **(BIC)** ([#369]).
- Added `def_file_name` attribute to `RefRulePosition` for storing the definition's
model file name in case of cross-references between models. ([#313],[#277])
- Migrated from Travis CI to GitHub Actions ([#307])
- Dropped support for deprecated Python versions. The lowest supported version
is 3.6. **(BIC)**
-------------------------------------------------------------------
Sun Jul 4 09:56:22 UTC 2021 - Sebastian Wagner <sebix+novell.com@sebix.at>
- Fix package builds for Leap 15.2 and 15.3, requires python-xml
-------------------------------------------------------------------
Sun Nov 1 17:56:21 UTC 2020 - Sebastian Wagner <sebix+novell.com@sebix.at>
- remove fix_tests_setupcfg.patch, included upstream
- update to version 2.3.0:
- Added:
- `textx generate`. Documented passing in arbitrary parameters which can be
used in the generator function. Also, implemented passing of model
parameters defined in the meta-model (`model_param_defs` and
`_tx_model_params`) ([#299])
- `get_location` function for producing `line/col/filename` from any textX
object. ([#294])
- `builtin_models` of type `ModelRepository` to meta-model constructor. Used
to supply pre-loaded models for `ImportURI` based scoping providers as a
fallback to search into. ([#284])
- Initial implementation of TEP-001 ([#111]) allowing to specify scope
provider behavior within the grammar itself. [#274] and [#287] introduce
the RREL (reference resolving expression language) to define how
references are resolved. Details see `rrel.md`.
- Parameter `should_follow` of callable type to `get_children` and
`get_children_of_type` model API calls to decide if the element should be
traversed. ([#281])
- Fixed:
- Fixed bug with Falsy user classes in `get_children` ([#288])
- Fixed bug with unhashable objects during dot export ([#283])
- Fixed bug where (Ext)RelativeName scope providers accepted any referenced
object that contained the lookup name in its name. Thanks ipa-mdl@GitHub
([#267])
- Fixed bug in `flow_dsl` test project causing static files not being included
in package build/installation. Thanks sebix@GitHub ([#272]).
- Fixed bug, where user classes not used in the grammar caused exceptions
([#270]): now, when passing a list of user classes, you need to use them in
your grammar. You can alternatively also pass a callable (see metamodel.md;
[#273]). Also, using base classes for rules from imported grammars in
conjunction with user classes is not allowed and results in an exception.
- Fixed bug in `export.py` concerning html escaping in the dot export of a
textx meta-model ([#276]).
- Changed:
- `_tx_model_param_definitions` deprecated in favor of `model_param_defs` ([#298]).
- `click` is now an optional dependency, only when CLI is needed ([#292])
- Make warning about not overwriting generated file more visible
([01341ec3](https://github.com/textX/textX/commit/01341ec381bfb4c8c27bcec5d2998a34d207f430))
- Truncate long strings during dot export for better diagram readability ([#282]).
- Changed `unhashable type` exception when a list is used for `name` attributes by
raising a more informative exception and extending docs to document the issue
and a proper way to solve it ([#40], [#266]).
-------------------------------------------------------------------
Thu Aug 6 13:16:43 UTC 2020 - Sebastian Wagner <sebix+novell.com@sebix.at>
- update to version 2.2.0:
- Added:
- Initial docs for Jinja code generator support (from
[textX-jinja](https://github.com/textX/textX-jinja)) ([#264]).
- Analyzing grammars programmatically as plain textX models
(`grammar_model_from_str/file`) ([#235])
- Initial `startproject` scaffolding (from
[textX-dev](https://github.com/textX/textX-dev)) docs ([#234])
- Generator helper functions `get_output_filename` and `gen_file` ([#233])
- `textx version` command ([#219])
- Versions for languages/packages in `list-languages` and `list-generators`
commands ([#228])
- Added the ability to specify extra parameters during `model_from_file` or
`model_from_str` and to define which extra parameters exist in the
meta-model ([#243]).
- Fixed:
- Fixed several instances of invalid truthiness checking. Thanks
markusschmaus@GitHub ([#250])
- Fixed applying multiple grammar rule modifiers ([#246])
- Fixed exception on calling `check` CLI command with relative path name.
- Fixed return value of textx generate and check commands: we return a failure
on error now ([#222])
- Fixed type checking for references to builtin elements ([#218])
- Changed:
- User classes can now be immutable (e.g. `attr.frozen`) or can use
`__slots__`. Thanks markusschmaus@GitHub ([#256, #260, #261])
- Cleanup of setup configuration and install scripts [#231]
- Dot/PlantUML rendering of meta-models: remove rendering of base types,
improve rendering of required/optional, render match rules as a single
table. ([#225])
- Allow passing kwargs in `metamodel_for_file/language` registration API
calls. ([#224])
- Add patch to fix tests:
* fix_tests_setupcfg.patch
-------------------------------------------------------------------
Mon May 18 09:10:29 UTC 2020 - Petr Gajdos <pgajdos@suse.com>
- %python3_only -> %python_alternative
-------------------------------------------------------------------
Wed Mar 11 11:27:12 UTC 2020 - pgajdos@suse.com
- version update to 2.1.0
### Added
- Added new function `textx.scoping.is_file_included` ([#197]).
### Changed
- Allow passing kwargs (specially - file_name) argument when loading metamodel
from string (needed for `textX-LS v0.1.0`) ([#211]).
- Changed the parser rule for regex matches. Spaces are not stripped any more
from the beginning and the end of the regexp-pattern. This could be possible
**BIC** for some special cases [#208].
- Changed function name `textx.scoping.get_all_models_including_attached_models`
to `textx.scoping.get_included_models` (marked old function
as deprecated) ([#197]).
- Delete all models touched while loading a model, when an error occurs
while loading in all repositories (strong exception safety guarantee) ([#200]).
-------------------------------------------------------------------
Tue Jul 9 11:53:21 UTC 2019 - Sebastian Wagner <sebix+novell.com@sebix.at>
- Fix Tests for Leap 15.1: click requires UTF-8 locale.
-------------------------------------------------------------------
Sat Jun 8 14:09:51 UTC 2019 - Sebastian Wagner <sebix+novell.com@sebix.at>
- Use documentation page as homepage instead of repository.
-------------------------------------------------------------------
Thu Jun 6 18:56:06 UTC 2019 - Sebastian Wagner <sebix+novell.com@sebix.at>
- Fix the tests
-------------------------------------------------------------------
Tue Jun 4 14:23:23 UTC 2019 - Tomáš Chvátal <tchvatal@suse.com>
- Make sure the package actually again gets to the test phase
* the tests are actually failing now
-------------------------------------------------------------------
Sun Jun 2 15:22:59 UTC 2019 - Sebastian Wagner <sebix+novell.com@sebix.at>
- update to version 2.0.1:
- Added:
- [Registration and discovery] of languages and generators ([#187])
- New textx CLI commands for listing generators and languages
(`list-generators`, `list-languages`) and calling a generator (`generate`) ([#187])
- Meta-models may now [reference other registered meta-models] using the
`reference` statement ([#187])
- Adding examples and documentation related to scope providers (related to
model modification through scope providers) ([#168])
- metamodel export feature for [PlantUML] ([#165])
- `textx_isinstance` from `textx.scoping.tools` made available in `textx`
([#164], [#157])
- CLI extensibility ([#162], [#161])
- An initial version of FAQ page ([#138]). Thanks Aluriak@GitHub
- A version of `calc.py` exercises usage of
`text.scoping.tools.textx_isinstance()` to inspect model objects types
during traversal. ([#136], [#123]). Thanks dkrikun@GitHub
- A version of `calc.py` in expression example that exercises dynamically adding
properties to object classes ([#126]). Thanks dkrikun@GitHub
- python like imports (named import rules, scope providers affected) ([#114])
- Added `STRICTFLOAT` as buildin type to allow to distinguish ints from floats
in `NUMBER`. Fixed docu link ([#98]). Possible **(BIC)**
- Added [flake8] and [coverage] checking ([#92])
- Changed:
- All textX commands implemented using textX CLI extensibility. `check`
command reworked to support the new registration feature ([#187]) **(BIC)**
- (Meta-)model visualization reworked as a set of textX generators ([#187]).
**(BIC)**
- Made scope provider implementation of `RelativeName` and `ExtRelativeName`
more readable ([#186]). Minor functional changes, not very probable to have
any impact (only affects model-paths containing a list not at the end of the
path; see [#186]). Possible **(BIC)**.
- Improved handling of abstract rules references. Improved the definition of
rules for various cases. Docs + tests ([#185], [#166]) **(BIC)**
- Changed the time of call of match rule object processors to be during the
model construction. This enable proper override of base types processors and
calls on nested match rules ([#183], [#182], [#96]). Possible **(BIC)**
- CLI migrated to the [click] library ([#162])
- Docs improvements ([#146], [#153], [#151]). Thanks simkimsia@GitHub.
- `FQN` constuctor param `follow_loaded_models` removed and introduced
callback `scope_rediction_logic` to implement arbitrary logic on FQN
resolution ([#133], [#114], [#103]) **(BIC)**
- Changed attribute name for the meta-model object (from `metamodel._parser`
to `metamodel._parser_blueprint`). ([#93]) **(BIC)**
- Started using _[Keep a Changelog][keepachangelog]_ ([#174])
- Started using _[Semantic Versioning][semver]_ ([#174])
- Dropped support for Python 3.3
- Fixed:
- White-spaces in string matches were erroneously stripped ([#188])
- Calling of match rule object processors ([#183], [#182], [#96])
- Circular rule references in grammars ([#173], [#159], [#155])
- Assertion error while calling object processors with multi meta models
(extended grammars) and custom types ([#141], [#140])
- Using a sequence of matches and rule reference in an abstract rule choice
alternative, and explicitly disallowing referencing more than one rule in an
abstract rule choice alternative ([#134])
- Unicode requirement for (meta)-model strings API parameters made strict.
This should prevent common errors with Python 2.x. ([#120]) (related: [#99],
[#105], [#117]). Possible **(BIC)**
- OS incompatibility fixes (path separator). ([#114])
- Object processors called twice for imported models ([#108], [#118])
- Documentation and examples regarding `NUMBER` base type ([#97], [#100]).
Thanks approxit@GitHub
-------------------------------------------------------------------
Fri Apr 5 08:46:58 UTC 2019 - pgajdos@suse.com
- version update to 1.8.0
- https://github.com/igordejanovic/textX/pull/71
- Regular expression with group support
- See [the
docs](http://www.igordejanovic.net/textX/development/grammar/#matches)
for usage.
- https://github.com/igordejanovic/textX/pull/69
- Added search path feature (issue #66) - search path support for model
files (importURI scope providers; see docs/scoping.md).
- https://github.com/igordejanovic/textX/pull/77
- New multi meta model support for references-only for better meta model
modularity (referencing models without having access to the grammar,
see docs/multimetamodel.md).
- https://github.com/igordejanovic/textX/pull/79
- Fixing obj_processors calling.
- https://github.com/igordejanovic/textX/pull/84
- New contribution guide.
- https://github.com/igordejanovic/textX/pull/81
- Bugfix: lost encoding for multi meta-model.
- https://github.com/igordejanovic/textX/pull/68
- changed parser access in metamodel (private attribute "_parser")
- mkdocs documentation now uses [mike](https://github.com/jimporter/mike) for
multiversion support.
-------------------------------------------------------------------
Tue Dec 4 12:55:15 UTC 2018 - Matej Cepl <mcepl@suse.com>
- Remove superfluous devel dependency for noarch package
-------------------------------------------------------------------
Thu Nov 22 14:31:16 UTC 2018 - Bernhard Wiedemann <bwiedemann@suse.com>
- drop indeterministic .dot files to make package build reproducible
(boo#1114571)
-------------------------------------------------------------------
Fri Jul 6 20:48:57 UTC 2018 - sebix+novell.com@sebix.at
- update to version 1.7.0:
- A major feature of this release is multi-(meta-)model support with
configurable resolving techniques. Thanks Pierre Bayerl (goto40@GitHub)!
The docs sections are [here](http://www.igordejanovic.net/textX/scoping/)
and [here](http://www.igordejanovic.net/textX/multimetamodel/).
Details follow:
- added new function textx.get_model.children to search arbritrary children
using a lambda predicate.
- remapped textx.model.get_children_of_type to the new children function
(changed the logic, such that the root node is also checked to be model
object).
- added new metamodel function to register scope providers. Scope providers
are callables, which return the referenced object.
- added optional attribute "_tx_model_repository", see metamodel.py
documentation
- added attribute "scope_provider" like "obj_processors" to organize scope
providers
- added an optional argument to model_from_str and model_from_file:
"pre_ref_resolution_callback": this is required internally to prepare the
loading of other model files.
- changed reference resolution in model.py
- moved default resolution to textx.scoping.py
- select the scope provider based on rule and rule-attribute (see
scoping.py documentation)
- added a Postponed type to postpone the resolution
- introduced a multi-pass resolution (implemented at the end of
parse_tree_to_objgraph; introduced new helper argument, e.g., a new
optional argument "is_this_the_main_model" and
"pre_ref_resolution_callback" (see metamodel.py above) to support
reference resolution in the presence of different model files.
- added a new module textx.scoping, to provide some scope providers (e.g. a
fully qualified name provider) - see scoping.py:
- full qualified names for reference names (e.g. package.package.class)
- global scope (model distributed over different files - loaded globally)
- import scope (model distributed over different files - loaded when
imported)
- relative scopes (e.g. instance.method where method is defined for the
class of the instance in a model of classes, methods and instances)
- selecting the metamodel based on a file pattern when loading models
- added tests (mostly scope related - some of them also test other stuff,
like buildins)
- exceptions where adapted to include a file name (makes errors more
visible)
- The metamodel now allows to specify a global model repository. With this
you can now share models across metamodels (before you could only do this
within one metamodel or language).
- The metamodel clones the parser when parsing a model file. The meta model
holds one parser, which is clone for every model to be parsed.
- TextXModelParser now has a clone method.
(TBC: is the clone ok: see responsibility of the method)
- model.py: the resolution loop logic now mostly moved to a separate object
ReferenceResolver, which holds the parser.
- The reference resolver are build from all model files affected (loaded).
This may involve multiple meta models.
- Then all references are resolved in one loop.
- Finally the helper objects (ReferenceResolver) are purged.
- The MetaModelProvider has a clear-method now (useful for tests).
- Backward incompatible change: match filters and object processors unified.
Now there are only object processors but they are called for any type of
rule and the returned value if exists is used instead of the original
object. See [the
docs](http://www.igordejanovic.net/textX/metamodel/#object-processors). In
order to migrate your match filters just register them as object processors.
- Fixing FLOAT regex. Thanks Boris Marin (borismarin@GitHub)!
- Fixing position info on obj cross ref. Thanks Daniel Elero (danixeee@GitHub)!
-------------------------------------------------------------------
Sun Feb 18 18:52:02 UTC 2018 - jengelh@inai.de
- Compact description
-------------------------------------------------------------------
Wed Jan 31 13:41:03 UTC 2018 - sebix+novell.com@sebix.at
- update to version 1.6.1. No notable changes.
- update to version 1.6.0. Shortened changelog:
* Introduced _tx_fqn class attribute.
* textX api functions and classes are now available directly from textx module.
* bugfixes
- update to version 1.5.2. Bugfixes.
-------------------------------------------------------------------
Mon Aug 28 12:37:08 UTC 2017 - sebix+novell.com@sebix.at
- initial package
View Git Blame Copy Permalink