Merge pull request #75 from aCLImatise/spacy-3

Spacy 3

Author: multimeric

aclimatise/nlp.py

@@ -2,17 +2,18 @@
 import wordsegment
 
 
+@spacy.language.Language.component("prevent_sentence_boundary")
 def prevent_sentence_boundary_detection(doc):
     for token in doc:
         token.is_sent_start = False
     return doc
 
 
 try:
-    nlp = spacy.load("en")
-    no_sentences = spacy.load("en")
+    nlp = spacy.load("en_core_web_sm")
+    no_sentences = spacy.load("en_core_web_sm")
     no_sentences.add_pipe(
-        prevent_sentence_boundary_detection, name="prevent-sbd", before="parser"
+        "prevent_sentence_boundary", name="prevent-sbd", before="parser"
     )
 except IOError:
     raise Exception(

docs/changes.rst

@@ -1,8 +1,12 @@
 Changelog
 =========
+3.0.1 (2021-02-10)
+------------------
+* Pin spacy dependency above 3.x.x
+* Remove support for older spacy versions
 
 3.0.0 (2021-01-27)
-----------------
+------------------
 Breaking Changes
 ****************
 * Rework of executors

setup.py

@@ -6,7 +6,7 @@
     install_requires=[
         "pyparsing",
         "jinja2",
-        "spacy",
+        "spacy~=3.0",
         "miniwdl",
         "wordsegment",
         "inflection",
@@ -21,7 +21,7 @@
         "psutil",
         "deprecated",
         "attrs",
-        "janis-pipelines.core >= 0.11.2",
+        "janis-pipelines.core>=0.11.2",
     ],
     python_requires=">=3.6",
     entry_points={"console_scripts": ["aclimatise = aclimatise.cli:main"]},