Merge pull request #393 from DigitalSlideArchive/match-methods

Make it clearer to adjust API match methods.
DigitalSlideArchive · Nov 10, 2023 · a164923 · a164923
2 parents a0910bf + 1e534c0
commit a164923
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 13 deletions.
diff --git a/tests/test_seer_api.py b/tests/test_seer_api.py
@@ -30,7 +30,7 @@ def test_api_search(api_search):
     queries = api_search.getOCRQueryList(ocr_record)
 
     # Ensure each query has enough data
-    assert all([len(query) >= 3 for query in queries])
+    assert all([len(query) >= 2 for query in queries])
 
     # Ensure tokens with confidence < .5 aren't used for queries
     query_tokens_list = [list(query.values()) for query in queries]

diff --git a/wsi_deid/matching_api.py b/wsi_deid/matching_api.py
@@ -57,6 +57,17 @@ class APISearch:
         },
     }
 
+    apiMatchMethods = [
+        {'path_case_num', 'name_last', 'name_first', 'date_of_birth'},
+        {'path_case_num', 'name_last', 'name_first'},
+        {'path_case_num', 'name_last', 'date_of_birth'},
+        {'path_case_num', 'name_first', 'date_of_birth'},
+        {'path_case_num', 'name_last', 'name_first', 'date_of_service'},
+        {'name_last', 'name_first', 'date_of_birth'},
+        {'path_case_num', 'name_last'},
+        {'path_case_num', 'date_of_birth'},
+    ]
+
     confidenceThreshold = 0.50
 
     def __init__(self, url=None, apikey=None, logger=None):
@@ -115,9 +126,7 @@ def getQueryList(self, words):
                 if v is not None and (v['word'], v['value']) not in used:
                     params[k] = v['value']
                     used.add((v['word'], v['value']))
-            if (len(params) >= 3 and (
-                    params.get('name_first') or params.get('name_last')) and (
-                    params.get('date_of_birth') or params.get('path_case_num'))):
+            if any(not len(apiset - set(params)) for apiset in self.apiMatchMethods):
                 if params not in [entry[-1] for entry in queries]:
                     queries.append((-len(params), len(queries), params))
         queries = [entry[-1] for entry in sorted(queries)]

diff --git a/wsi_deid/rest.py b/wsi_deid/rest.py
@@ -865,6 +865,8 @@ def callMatchingAPI(self, match):
     )
     @access.public
     def fakeMatchingAPI(self, match):  # noqa
+        from .matching_api import APISearch
+
         fakeData = []
         try:
             filepath = '/conf/fake_matches.json'
@@ -873,16 +875,8 @@ def fakeMatchingAPI(self, match):  # noqa
             fakeData = json.load(open(filepath))
         except Exception:
             pass
-        matchMethods = [
-            {'date_of_birth', 'name_last', 'name_first', 'path_case_num'},
-            {'name_last', 'name_first', 'path_case_num'},
-            {'date_of_birth', 'name_last', 'path_case_num'},
-            {'date_of_birth', 'name_first', 'path_case_num'},
-            {'date_of_birth', 'name_last', 'name_first', 'date_of_service'},
-            {'date_of_birth', 'name_last', 'name_first'},
-        ]
         results = []
-        for matchMethod in matchMethods:
+        for matchMethod in APISearch.apiMatchMethods:
             for entry in fakeData:
                 matched = True
                 try: