Merge pull request #9 from jenojp/develop

Spacy 3.3 support

Author: jenojp

azure-pipelines.yml

@@ -12,34 +12,34 @@ trigger:
 
 strategy:
   matrix:
-    Python36Linux:
-      imageName: 'ubuntu-18.04'
-      python.version: '3.6'
     Python37Linux:
-      imageName: 'ubuntu-18.04'
+      imageName: 'ubuntu-20.04'
       python.version: '3.7'
     Python38Linux:
-      imageName: 'ubuntu-18.04'
+      imageName: 'ubuntu-20.04'
       python.version: '3.8'
-    Python36Mac:
-      imageName: 'macos-10.15'
-      python.version: '3.6'
+    Python39Linux:
+      imageName: 'ubuntu-20.04'
+      python.version: '3.9'
     Python37Mac:
-      imageName: 'macos-10.15'
+      imageName: 'macos-11'
       python.version: '3.7'
     Python38Mac:
-      imageName: 'macos-10.15'
+      imageName: 'macos-11'
       python.version: '3.8'
-    Python36Windows:
-      imageName: 'vs2017-win2016'
-      python.version: '3.6'
+    Python39Mac:
+      imageName: 'macos-11'
+      python.version: '3.9'
     Python37Windows:
-      imageName: 'vs2017-win2016'
+      imageName: 'windows-2019'
       python.version: '3.7'
     Python38Windows:
-      imageName: 'vs2017-win2016'
+      imageName: 'windows-2019'
       python.version: '3.8'
-  maxParallel: 4
+    Python39Windows:
+      imageName: 'windows-2019'
+      python.version: '3.9'
+  maxParallel: 9
 
 pool:
   vmImage: $(imageName)

extractacy/extract.py

@@ -23,6 +23,8 @@ def __call__(self, doc):
         in the pipeline, if available.
         """
         matches = self.matcher(doc)
+        for match_id, start, end in matches:
+            print(self.nlp.vocab.strings[match_id], start, end)
         for e in doc.ents:
             if e.label_ not in self.ent_patterns.keys():
                 e._.value_extract = []

extractacy/test.py

@@ -11,12 +11,12 @@ def build_docs():
             "Discharge Date: 11/15/2008. Patient had temp reading of 102.6 degrees. Insurance claim sent to patient's account on file: 1112223. 12/31/2008: Payment received.",
             [
                 ("Discharge Date", ["11/15/2008"]),
-                ("11/15/2008", []),
+                # ("11/15/2008", []),
                 ("temp", ["102.6 degrees"]),
                 ("102.6 degrees", []),
                 ("account", ["1112223"]),
                 ("1112223", []),
-                ("12/31/2008", []),
+                # ("12/31/2008", []),
                 ("Payment received", ["12/31/2008"]),
             ],
         )
@@ -60,6 +60,7 @@ def build_docs():
         (
             "We believe 01/01/1980 is his date of birth but it could also be 01/02/1980",
             [
+                ("01/01/1980",[]),
                 ("date of birth", ["01/01/1980", "01/02/1980"]),
                 ("01/02/1980",[])
             ],
@@ -78,14 +79,17 @@ def build_docs():
     docs.append(
         (
             "We believe 01/01/1980 is his date of birth",
-            [("date of birth", ["01/01/1980"])],
+            [
+                ("01/01/1980", []),
+                ("date of birth", ["01/01/1980"])
+                ],
         )
     )
     # test outside boundary
     docs.append(
         (
             "Discharge date unknown. 12/12/1999 date of confirmation.",
-            [("Discharge date", []), ("12/12/1999 date", [])],
+            [("Discharge date", []), ("12/12/1999", [])],
         )
     )
 
@@ -164,9 +168,20 @@ def test():
     docs = build_docs()
     for d in docs:
         doc = nlp(d[0])
+        print()
+        print()
+        print("sentences...")
+        for s in doc.sents:
+            print(s)
+        print("tokens....")
+        print([t.text for t in doc])
+        print("entities....")
+        for e in doc.ents:
+            print(e.text, e.start, e.end)
+        print("value extracts....")
         for i, e in enumerate(doc.ents):
             print(e.text, e._.value_extract)
-            print([t.text for t in doc])
+        for i, e in enumerate(doc.ents):
             assert (e.text, e._.value_extract) == d[1][i]
 
 

setup.py

@@ -3,7 +3,7 @@
 
 setup(
     name = 'extractacy',
-    version = 'v1.0.1',
+    version = 'v1.0.2',
     url = 'https://github.com/jenojp/extractacy',
     author = 'Jeno Pizarro',
     author_email = '[email protected]',
@@ -24,7 +24,7 @@
     packages = find_packages(exclude=["*.tests", "*.tests.*", "tests.*", "tests"]),
     license="MIT",
     install_requires=[
-        "spacy>=3.0.1,<3.2.0",
+        "spacy>=3.0.1,<4.0.0",
         ],
     tests_require=[
         "pytest",