Make gensim optional

.github/workflows/pytest.yml

@@ -28,6 +28,6 @@ jobs:
 
       - name: Install package and run tests
         run: |
-          pip install .
+          pip install ".[doc2vec]"
           pip install pytest
           pytest tests -v

asreviewcontrib/dory/feature_extractors/doc2vec.py

@@ -2,12 +2,18 @@
 
 import numpy as np
 from asreview.models.feature_extractors import TextMerger
-from gensim.models.doc2vec import Doc2Vec as GenSimDoc2Vec
-from gensim.models.doc2vec import TaggedDocument
-from gensim.utils import simple_preprocess
 from sklearn.base import BaseEstimator, TransformerMixin
 from sklearn.pipeline import Pipeline
 
+try:
+    from gensim.models.doc2vec import Doc2Vec as GenSimDoc2Vec
+    from gensim.models.doc2vec import TaggedDocument
+    from gensim.utils import simple_preprocess
+
+    GENSIM_AVAILABLE = True
+except ImportError:
+    GENSIM_AVAILABLE = False
+
 
 class Doc2Vec(Pipeline):
     name = "doc2vec"
@@ -72,6 +78,11 @@ def __init__(
         dbow_words=False,
         verbose=True,
     ):
+        if not GENSIM_AVAILABLE:
+            raise ImportError(
+                "The 'gensim' package is required for Doc2Vec but is not installed. "
+                "Install it with: pip install gensim"
+            )
         self.vector_size = int(vector_size)
         self.epochs = int(epochs)
         self.min_count = int(min_count)

pyproject.toml

@@ -28,10 +28,12 @@ dependencies = [
     "sentence-transformers>=3.4.1",
     "accelerate>=1.0.0",
     "torch>=2.1.0",
-    "xgboost>=2.1.4",
-    "gensim>=4.4.0"
+    "xgboost>=2.1.4"
 ]
 
+[project.optional-dependencies]
+doc2vec = ["gensim>=4.4.0"]
+
 [project.urls]
 Homepage = "https://asreview.ai"
 Repository = "https://github.com/asreview/asreview-dory/"