Commit Graph

23 Commits

Author SHA1 Message Date
Ines Montani
43b960c01b
Refactor pipeline components, config and language data (#5759)
* Update with WIP

* Update with WIP

* Update with pipeline serialization

* Update types and pipe factories

* Add deep merge, tidy up and add tests

* Fix pipe creation from config

* Don't validate default configs on load

* Update spacy/language.py

Co-authored-by: Ines Montani <ines@ines.io>

* Adjust factory/component meta error

* Clean up factory args and remove defaults

* Add test for failing empty dict defaults

* Update pipeline handling and methods

* provide KB as registry function instead of as object

* small change in test to make functionality more clear

* update example script for EL configuration

* Fix typo

* Simplify test

* Simplify test

* splitting pipes.pyx into separate files

* moving default configs to each component file

* fix batch_size type

* removing default values from component constructors where possible (TODO: test 4725)

* skip instead of xfail

* Add test for config -> nlp with multiple instances

* pipeline.pipes -> pipeline.pipe

* Tidy up, document, remove kwargs

* small cleanup/generalization for Tok2VecListener

* use DEFAULT_UPSTREAM field

* revert to avoid circular imports

* Fix tests

* Replace deprecated arg

* Make model dirs require config

* fix pickling of keyword-only arguments in constructor

* WIP: clean up and integrate full config

* Add helper to handle function args more reliably

Now also includes keyword-only args

* Fix config composition and serialization

* Improve config debugging and add visual diff

* Remove unused defaults and fix type

* Remove pipeline and factories from meta

* Update spacy/default_config.cfg

Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>

* Update spacy/default_config.cfg

* small UX edits

* avoid printing stack trace for debug CLI commands

* Add support for language-specific factories

* specify the section of the config which holds the model to debug

* WIP: add Language.from_config

* Update with language data refactor WIP

* Auto-format

* Add backwards-compat handling for Language.factories

* Update morphologizer.pyx

* Fix morphologizer

* Update and simplify lemmatizers

* Fix Japanese tests

* Port over tagger changes

* Fix Chinese and tests

* Update to latest Thinc

* WIP: xfail first Russian lemmatizer test

* Fix component-specific overrides

* fix nO for output layers in debug_model

* Fix default value

* Fix tests and don't pass objects in config

* Fix deep merging

* Fix lemma lookup data registry

Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed)

* Add types

* Add Vocab.from_config

* Fix typo

* Fix tests

* Make config copying more elegant

* Fix pipe analysis

* Fix lemmatizers and is_base_form

* WIP: move language defaults to config

* Fix morphology type

* Fix vocab

* Remove comment

* Update to latest Thinc

* Add morph rules to config

* Tidy up

* Remove set_morphology option from tagger factory

* Hack use_gpu

* Move [pipeline] to top-level block and make [nlp.pipeline] list

Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them

* Fix use_gpu and resume in CLI

* Auto-format

* Remove resume from config

* Fix formatting and error

* [pipeline] -> [components]

* Fix types

* Fix tagger test: requires set_morphology?

Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com>
Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 13:42:59 +02:00
Ines Montani
644074b954 Merge branch 'develop' into master-tmp 2020-07-20 14:58:04 +02:00
Matthew Honnibal
2d715451a2
Revert "Convert custom user_data to token extension format for Japanese tokenizer (#5652)" (#5665)
This reverts commit 1dd38191ec.
2020-06-29 14:34:15 +02:00
Adriane Boyd
1dd38191ec
Convert custom user_data to token extension format for Japanese tokenizer (#5652)
* Convert custom user_data to token extension format

Convert the user_data values so that they can be loaded as custom token
extensions for `inflection`, `reading_form`, `sub_tokens`, and `lemma`.

* Reset Underscore state in ja tokenizer tests
2020-06-29 14:20:26 +02:00
Hiroshi Matsuda
150a39ccca
Japanese model: add user_dict entries and small refactor (#5573)
* user_dict fields: adding inflections, reading_forms, sub_tokens
deleting: unidic_tags
improve code readability around the token alignment procedure

* add test cases, replace fugashi with sudachipy in conftest

* move bunsetu.py to spaCy Universe as a pipeline component BunsetuRecognizer

* tag is space -> both surface and tag are spaces

* consider len(text)==0
2020-06-22 14:32:25 +02:00
Ines Montani
ef5f548fb0 Tidy up and auto-format 2020-06-21 22:38:04 +02:00
Ines Montani
40bb918a4c Remove unicode declarations and tidy up 2020-06-21 22:34:10 +02:00
Ines Montani
52728d8fa3 Merge branch 'develop' into master-tmp 2020-06-20 15:52:00 +02:00
adrianeboyd
b7e6e1b9a7
Disable sentence segmentation in ja tokenizer (#5566) 2020-06-09 12:00:59 +02:00
adrianeboyd
f162815f45
Handle empty and whitespace-only docs for Japanese (#5564)
Handle empty and whitespace-only docs in the custom alignment method
used by the Japanese tokenizer.
2020-06-08 21:09:23 +02:00
adrianeboyd
3bf111585d
Update Japanese tokenizer config and add serialization (#5562)
* Use `config` dict for tokenizer settings
* Add serialization of split mode setting
* Add tests for tokenizer split modes and serialization of split mode
setting

Based on #5561
2020-06-08 16:29:05 +02:00
Paul O'Leary McCann
410fb7ee43
Add Japanese Model (#5544)
* Add more rules to deal with Japanese UD mappings

Japanese UD rules sometimes give different UD tags to tokens with the
same underlying POS tag. The UD spec indicates these cases should be
disambiguated using the output of a tool called "comainu", but rules are
enough to get the right result.

These rules are taken from Ginza at time of writing, see #3756.

* Add new tags from GSD

This is a few rare tags that aren't in Unidic but are in the GSD data.

* Add basic Japanese sentencization

This code is taken from Ginza again.

* Add sentenceizer quote handling

Could probably add more paired characters but this will do for now. Also
includes some tests.

* Replace fugashi with SudachiPy

* Modify tag format to match GSD annotations

Some of the tests still need to be updated, but I want to get this up
for testing training.

* Deal with case with closing punct without opening

* refactor resolve_pos()

* change tag field separator from "," to "-"

* add TAG_ORTH_MAP

* add TAG_BIGRAM_MAP

* revise rules for 連体詞

* revise rules for 連体詞

* improve POS about 2%

* add syntax_iterator.py (not mature yet)

* improve syntax_iterators.py

* improve syntax_iterators.py

* add phrases including nouns and drop NPs consist of STOP_WORDS

* First take at noun chunks

This works in many situations but still has issues in others.

If the start of a subtree has no noun, then nested phrases can be
generated.

    また行きたい、そんな気持ちにさせてくれるお店です。
    [そんな気持ち, また行きたい、そんな気持ちにさせてくれるお店]

For some reason て gets included sometimes. Not sure why.

    ゲンに連れ添って円盤生物を調査するパートナーとなる。
    [て円盤生物, ...]

Some phrases that look like they should be split are grouped together;
not entirely sure that's wrong. This whole thing becomes one chunk:

    道の駅遠山郷北側からかぐら大橋南詰現道交点までの1.060kmのみ開通済み

* Use new generic get_words_and_spaces

The new get_words_and_spaces function is simpler than what was used in
Japanese, so it's good to be able to switch to it. However, there was an
issue. The new function works just on text, so POS info could get out of
sync. Fixing this required a small change to the way dtokens (tokens
with POS and lemma info) were generated.

Specifically, multiple extraneous spaces now become a single token, so
when generating dtokens multiple space tokens should be created in a
row.

* Fix noun_chunks, should be working now

* Fix some tests, add naughty strings tests

Some of the existing tests changed because the tokenization mode of
Sudachi changed to the more fine-grained A mode.

Sudachi also has issues with some strings, so this adds a test against
the naughty strings.

* Remove empty Sudachi tokens

Not doing this creates zero-length tokens and causes errors in the
internal spaCy processing.

* Add yield_bunsetu back in as a separate piece of code

Co-authored-by: Hiroshi Matsuda <40782025+hiroshi-matsuda-rit@users.noreply.github.com>
Co-authored-by: hiroshi <hiroshi_matsuda@megagon.ai>
2020-06-04 19:15:43 +02:00
Ines Montani
db55577c45
Drop Python 2.7 and 3.5 (#4828)
* Remove unicode declarations

* Remove Python 3.5 and 2.7 from CI

* Don't require pathlib

* Replace compat helpers

* Remove OrderedDict

* Use f-strings

* Set Cython compiler language level

* Fix typo

* Re-add OrderedDict for Table

* Update setup.cfg

* Revert CONTRIBUTING.md

* Revert lookups.md

* Revert top-level.md

* Small adjustments and docs [ci skip]
2019-12-22 01:53:56 +01:00
Ines Montani
3d8fd4b461 Revert #4334 2019-09-29 17:32:12 +02:00
Ines Montani
c9cd516d96 Move tests out of package (#4334)
* Move tests out of package

* Fix typo
2019-09-28 18:05:00 +02:00
Ines Montani
3126dd0904 Tidy up and auto-format [ci skip] 2019-09-14 12:58:06 +02:00
Paul O'Leary McCann
29a9e636eb Fix half-width space handling in JA (#4284) (closes #4262)
Before this patch, half-width spaces between words were simply lost in
Japanese text. This wasn't immediately noticeable because much Japanese
text never uses spaces at all.
2019-09-13 16:28:12 +02:00
Hiromu Hota
914b9ff3d2 Tags are joined with a comma and padded with asterisks (#3491)
<!--- Provide a general summary of your changes in the title. -->

## Description
<!--- Use this section to describe your changes. If your changes required
testing, include information about the testing environment and the tests you
ran. If your test fixes a bug reported in an issue, don't forget to include the
issue number. If your PR is still a work in progress, that's totally fine – just
include a note to let us know. -->

Fix a bug in the test of JapaneseTokenizer.
This PR may require @polm's review.

### Types of change
<!-- What type of change does your PR cover? Is it a bug fix, an enhancement
or new feature, or a change to the documentation? -->

Bug fix

## Checklist
<!--- Before you submit the PR, go over this checklist and make sure you can
tick off all the boxes. [] -> [x] -->
- [x] I have submitted the spaCy Contributor Agreement.
- [x] I ran the tests, and all new and existing tests passed.
- [x] My changes don't require a change to the documentation, or if they do, I've added all required information.
2019-03-28 16:17:31 +01:00
Ines Montani
b6e991440c 💫 Tidy up and auto-format tests (#2967)
* Auto-format tests with black

* Add flake8 config

* Tidy up and remove unused imports

* Fix redefinitions of test functions

* Replace orths_and_spaces with words and spaces

* Fix compatibility with pytest 4.0

* xfail test for now

Test was previously overwritten by following test due to naming conflict, so failure wasn't reported

* Unfail passing test

* Only use fixture via arguments

Fixes pytest 4.0 compatibility
2018-11-27 01:09:36 +01:00
Ines Montani
75f3234404
💫 Refactor test suite (#2568)
## Description

Related issues: #2379 (should be fixed by separating model tests)

* **total execution time down from > 300 seconds to under 60 seconds** 🎉
* removed all model-specific tests that could only really be run manually anyway – those will now live in a separate test suite in the [`spacy-models`](https://github.com/explosion/spacy-models) repository and are already integrated into our new model training infrastructure
* changed all relative imports to absolute imports to prepare for moving the test suite from `/spacy/tests` to `/tests` (it'll now always test against the installed version)
* merged old regression tests into collections, e.g. `test_issue1001-1500.py` (about 90% of the regression tests are very short anyways)
* tidied up and rewrote existing tests wherever possible

### Todo

- [ ] move tests to `/tests` and adjust CI commands accordingly
- [x] move model test suite from internal repo to `spacy-models`
- [x] ~~investigate why `pipeline/test_textcat.py` is flakey~~
- [x] review old regression tests (leftover files) and see if they can be merged, simplified or deleted
- [ ] update documentation on how to run tests


### Types of change
enhancement, tests

## Checklist
<!--- Before you submit the PR, go over this checklist and make sure you can
tick off all the boxes. [] -> [x] -->
- [x] I have submitted the spaCy Contributor Agreement.
- [x] I ran the tests, and all new and existing tests passed.
- [ ] My changes don't require a change to the documentation, or if they do, I've added all required information.
2018-07-24 23:38:44 +02:00
Paul O'Leary McCann
1987f3f784 Add Japanese lemmas (#2543)
This info was already available from Mecab, forgot to add it before.
2018-07-13 10:55:14 +02:00
Paul O'Leary McCann
bd72fbf09c Port Japanese mecab tokenizer from v1 (#2036)
* Port Japanese mecab tokenizer from v1

This brings the Mecab-based Japanese tokenization introduced in #1246 to
spaCy v2. There isn't a JapaneseTagger implementation yet, but POS tag
information from Mecab is stored in a token extension. A tag map is also
included.

As a reminder, Mecab is required because Universal Dependencies are
based on Unidic tags, and Janome doesn't support Unidic.

Things to check:

1. Is this the right way to use a token extension?

2. What's the right way to implement a JapaneseTagger? The approach in
 #1246 relied on `tag_from_strings` which is just gone now. I guess the
best thing is to just try training spaCy's default Tagger?

-POLM

* Add tagging/make_doc and tests
2018-05-03 18:38:26 +02:00
ines
612224c10d Port over changes from #1157 2017-10-14 13:11:39 +02:00