Preparation for release

Author: richardpaulhudson

Diff for: .github/workflows/test-coreferee.yml

@@ -8,6 +8,17 @@ jobs:
       matrix:
         os: [macos-latest, windows-latest, ubuntu-latest]
         python_version: ['3.6', '3.7', '3.8', '3.9', '3.10']
+        spacy_version: ['3.3.0']
+      include:
+        - os: 'ubuntu-latest'
+          python-version: '3.10'
+          spacy_version: '3.2.0'
+        - os: 'ubuntu-latest'
+          python-version: '3.10'
+          spacy_version: '3.1.0'
+        - os: 'ubuntu-latest'
+          python-version: '3.10'
+          spacy_version: '3.0.0'
     runs-on: ${{ matrix.os }}
     steps:
     - name: Checkout repository code
@@ -23,7 +34,7 @@ jobs:
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
-        pip install spacy pytest
+        pip install spacy==${{ matrix.spacy_version }} pytest
         
         # see https://github.com/explosion/spaCy/issues/10564
         pip uninstall click -y

Diff for: coreferee/lang/de/config.cfg

@@ -1,17 +1,17 @@
 [sm_3_2_0]
 model:         core_news_sm
 from_version:  3.0.0
-to_version:    3.2.0
+to_version:    3.3.0
 train_version: 3.2.0
 
 [md_3_2_0]
 model:         core_news_md
 from_version:  3.0.0
-to_version:    3.2.0
+to_version:    3.3.0
 train_version: 3.2.0
 
 [lg_3_2_0]
 model:         core_news_lg
 from_version:  3.0.0
-to_version:    3.2.0
+to_version:    3.3.0
 train_version: 3.2.0

Diff for: coreferee/lang/en/config.cfg

@@ -1,27 +1,25 @@
 [lg_3_2_0]
 model:                  core_web_lg
 from_version:           3.0.0
-to_version:             3.2.0
+to_version:             3.3.0
 train_version:          3.2.0
 
 [md_3_2_0]
 model:                  core_web_md
 from_version:           3.0.0
-to_version:             3.2.0
+to_version:             3.3.0
 train_version:          3.2.0
 
 [sm_3_2_0]
 model:                  core_web_sm
 from_version:           3.0.0
-to_version:             3.2.0
+to_version:             3.3.0
 train_version:          3.2.0
 
 [trf_3_2_0]
 model:                  core_web_trf
 from_version:           3.0.0
-to_version:             3.2.0
+to_version:             3.3.0
 train_version:          3.2.0
 vectors_model:          core_web_lg
-vectors_from_version:   3.0.0
-vectors_to_version:     3.2.0
-vectors_train_version:  3.2.0
+

Diff for: coreferee/lang/pl/config.cfg

@@ -1,12 +1,11 @@
-[md_3_2_0]
-model:                  core_news_md
+[lg_3_2_0]
+model:                  core_news_lg
 from_version:           3.0.0
-to_version:             3.2.0
+to_version:             3.3.0
 train_version:          3.2.0
 
-
-[lg_3_2_0]
-model:                  core_news_lg
+[md_3_2_0]
+model:                  core_news_md
 from_version:           3.0.0
-to_version:             3.2.0
+to_version:             3.3.0
 train_version:          3.2.0

Diff for: coreferee/manager.py

@@ -86,9 +86,9 @@ def get_annotator(nlp: Language) -> Annotator:
                         msg.fail(error_msg)
                         raise VectorsModelNotInstalledError(error_msg)
                     if version.parse(vectors_nlp.meta["version"]) < version.parse(
-                        config_entry["vectors_from_version"]
+                        config_entry["from_version"]
                     ) or version.parse(vectors_nlp.meta["version"]) > version.parse(
-                        config_entry["vectors_to_version"]
+                        config_entry["to_version"]
                     ):
                         msg = Printer()
                         error_msg = "".join(
@@ -100,9 +100,9 @@ def get_annotator(nlp: Language) -> Annotator:
                                 "_",
                                 config_entry["vectors_model"],
                                 " between versions ",
-                                config_entry["vectors_from_version"],
+                                config_entry["from_version"],
                                 " and ",
-                                config_entry["vectors_to_version"],
+                                config_entry["to_version"],
                                 " inclusive.",
                             )
                         )

Diff for: coreferee/training/train.py

@@ -73,23 +73,25 @@ def __init__(
                 )
             )
             if not os.path.isdir(this_model_dir) or not train_not_check:
-                self.relevant_config_entry_names.append(config_entry_name)
                 model_name = "_".join((lang, config_entry["model"]))
-                self.load_model(
+                if not self.load_model(
                     model_name,
                     config_entry_name,
                     config_entry["from_version"],
                     config_entry["to_version"],
-                )
+                ):
+                    print("Skipping config entry", config_entry_name, "as specified version range does not match.")
+                    continue
                 if "vectors_model" in config_entry:
                     vectors_model_name = "_".join((lang, config_entry["vectors_model"]))
                     self.load_model(
                         vectors_model_name,
                         config_entry_name,
-                        config_entry["vectors_from_version"],
-                        config_entry["vectors_to_version"],
+                        config_entry["from_version"],
+                        config_entry["to_version"],
                         is_vector_model=True,
                     )
+                self.relevant_config_entry_names.append(config_entry_name)
             else:
                 print("Skipping config entry", config_entry_name, "as model exists")
 
@@ -119,7 +121,7 @@ def load_model(
         to_version,
         *,
         is_vector_model=False
-    ):
+    ) -> bool:
         if name not in self.nlp_dict:
             print("Loading model", name, "...")
             try:
@@ -147,24 +149,9 @@ def load_model(
         if version.parse(nlp.meta["version"]) < version.parse(
             from_version
         ) or version.parse(nlp.meta["version"]) > version.parse(to_version):
-            if is_vector_model:
-                print(
-                    "Config entry",
-                    config_entry_name,
-                    "specifies a version range for vectors model",
-                    name,
-                    "that does not include the loaded version.",
-                )
-            else:
-                print(
-                    "Config entry",
-                    config_entry_name,
-                    "specifies a version range for model",
-                    name,
-                    "that does not include the loaded version.",
-                )
-            sys.exit(1)
+            return False
         self.nlp_dict[name] = nlp
+        return True
 
     def set_up_models_dir(self):
         os.mkdir(self.models_dirname)

Diff for: setup.cfg

@@ -35,6 +35,6 @@ include_package_data = True
 packages = find:
 python_requires = >=3.6,<3.11
 install_requires =
-  spacy>=3.0.0,<3.3.0
+  spacy>=3.0.0,<3.4.0
 [options.package_data]
 * = *.dat, *.cfg

Diff for: tests/de/test_rules_de.py

@@ -204,7 +204,7 @@ def test_substituting_indefinite_pronoun(self):
     def test_pronoun_noun(self):
         self.compare_independent_noun(
             "Diejenigen der Jungen, die heimgekommen sind, waren müde",
-            [0, 2],
+            [2],
             excluded_nlps=["core_news_md", "core_news_sm"],
         )
 
@@ -694,7 +694,7 @@ def test_potential_pair_person_neut_2(self):
     @unittest.skipIf(train_version_mismatch, train_version_mismatch_message)
     def test_potential_pair_person_neut_3(self):
         self.compare_potential_pair(
-            "Ich sah ein Kind. Dann lächelte dieses und weinte", 3, False, 7, 2
+            "Ich sah ein Kind. Dann lächelte es und weinte", 3, False, 7, 2
         )
 
     def test_potential_pair_person_neut_control(self):

Diff for: tests/de/test_smoke_tests_de.py

@@ -56,7 +56,7 @@ def test_simple_conjunction_different_words(self):
     @unittest.skipIf(train_version_mismatch, train_version_mismatch_message)
     def test_conjunction_different_pronouns(self):
         self.compare_annotations(
-            'Peter und das Mädchen haben gesprochen, und es und er jagten ein Katze', '[0: [0], [10], 1: [3], [8]]')
+            'Peter und das Mädchen haben gesprochen, und dieses und er jagten ein Katze', '[0: [0], [10], 1: [3], [8]]')
 
     def test_conjunction_involving_pronoun(self):
         self.compare_annotations(

Diff for: tests/fr/test_rules_fr.py

@@ -1,8 +1,13 @@
 import unittest
+from coreferee.errors import ModelNotSupportedError
 from coreferee.rules import RulesAnalyzerFactory
 from coreferee.test_utils import get_nlps
 from coreferee.data_model import Mention
 
+try:
+    nlps = get_nlps("fr")
+except ModelNotSupportedError:
+    raise unittest.SkipTest("Model version not supported.")
 
 class FrenchRulesTest(unittest.TestCase):
     def setUp(self):

Diff for: tests/fr/test_smoke_tests_fr.py

@@ -1,18 +1,23 @@
 import unittest
+from coreferee.errors import ModelNotSupportedError
 from coreferee.test_utils import get_nlps
 
-nlps = get_nlps('fr')
+try:
+    nlps = get_nlps("fr")
+except ModelNotSupportedError:
+    raise unittest.SkipTest("Model version not supported.")
+
 train_version_mismatch = False
+train_version_mismatch_message = "Loaded model version does not match train model version"
 for nlp in nlps:
     if not nlp.meta["matches_train_version"]:
         train_version_mismatch = True
-train_version_mismatch_message = "Loaded model version does not match train model version"
+
 
 class FrenchSmokeTest(unittest.TestCase):
 
     def setUp(self):
-
-        self.nlps = get_nlps('fr')
+        self.nlps = get_nlps("fr")
 
     def all_nlps(self, func):
         for nlp in self.nlps:

Diff for: tests/pl/test_rules_pl.py

@@ -1059,7 +1059,7 @@ def test_potential_pair_non_personal_subject_personal_verb_control_conjunction(
         self,
     ):
         self.compare_potential_pair(
-            "Dom i dom stoją. One powiedziały, wszystko dobrze", 0, True, 5, 2, 
+            "Dom i dom stoją. One powiedziały, wszystko dobrze", 0, True, 5, 1, 
             excluded_nlps=['core_news_md', 'core_news_sm']
         )
 
@@ -1068,7 +1068,7 @@ def test_potential_pair_non_personal_subject_personal_verb_control_z_conjunction
         self,
     ):
         self.compare_potential_pair(
-            "Dom z domem stoją. One powiedziały, wszystko dobrze", 0, True, 5, 2, 
+            "Dom z domem stoją. One powiedziały, wszystko dobrze", 0, True, 5, 1, 
             excluded_nlps=['core_news_md', 'core_news_sm']
         )
 
@@ -1468,12 +1468,12 @@ def test_potentially_introducing_with_preposition(self):
     @unittest.skipIf(train_version_mismatch, train_version_mismatch_message)
     def test_potentially_introducing_with_ten_control(self):
         self.compare_potentially_introducing(
-            "On mieszka również z tym facetem", 5, False, excluded_nlps=['core_news_md', 'core_news_sm']
+            "On mieszka z tym kolegą", 4, False, excluded_nlps=['core_news_md', 'core_news_sm']
         )
 
     def test_potentially_introducing_with_ten_and_relative_clause(self):
         self.compare_potentially_introducing(
-            "On mieszka również z tym facetem, którego znasz", 5, True,
+            "On mieszka z tym kolegą, którego znasz", 4, True,
             excluded_nlps=["core_news_sm"]
         )