Merge pull request #62 from QuanMPhm/59/institute_domains

Allow a PI's institution name to be search with subdomains

process_report/institute_map.json

@@ -10,15 +10,6 @@
     "dfci.harvard.edu"      : "Dana-Farber Cancer Institute",
     "bwh.harvard.edu"       : "Brigham and Women's Hospital",
     "bidmc.harvard.edu"     : "Beth Israel Deaconess Medical Center",
-    "fas.harvard.edu"       : "Harvard University",
-    "cga.harvard.edu"       : "Harvard University",
-    "iq.harvard.edu"        : "Harvard University",
-    "hks.harvard.edu"       : "Harvard University",
-    "hsph.harvard.edu"      : "Harvard University",
-    "seas.harvard.edu"      : "Harvard University",
-    "gse.harvard.edu"       : "Harvard University",
-    "gov.harvard.edu"       : "Harvard University",
-    "oeb.harvard.edu"       : "Harvard University",
     "harvard.edu"           : "Harvard University",
     "wpi.edu"               : "Worcester Polytechnic Institute",
     "mit.edu"               : "Massachusetts Institute of Technology",

process_report/process_report.py

@@ -45,8 +45,11 @@
 
 
 def get_institution_from_pi(institute_map, pi_uname):
-    institution_key = pi_uname.split("@")[-1]
-    institution_name = institute_map.get(institution_key, "")
+    institution_domain = pi_uname.split("@")[-1]
+    for i in range(institution_domain.count(".") + 1):
+        if institution_name := institute_map.get(institution_domain, ""):
+            break
+        institution_domain = institution_domain[institution_domain.find(".") + 1 :]
 
     if institution_name == "":
         print(f"Warning: PI name {pi_uname} does not match any institution!")

process_report/tests/unit_tests.py

@@ -229,34 +229,32 @@ def test_get_pi_institution(self):
             "bu.edu": "Boston University",
             "bentley.edu": "Bentley",
             "mclean.harvard.edu": "McLean Hospital",
+            "northeastern.edu": "Northeastern University",
+            "childrens.harvard.edu": "Boston Children's Hospital",
             "meei.harvard.edu": "Massachusetts Eye & Ear",
             "dfci.harvard.edu": "Dana-Farber Cancer Institute",
-            "northeastern.edu": "Northeastern University",
+            "bwh.harvard.edu": "Brigham and Women's Hospital",
+            "bidmc.harvard.edu": "Beth Israel Deaconess Medical Center",
         }
 
-        self.assertEqual(
-            process_report.get_institution_from_pi(institute_map, "[email protected]"),
-            "Boston University",
-        )
-        self.assertEqual(
-            process_report.get_institution_from_pi(
-                institute_map, "[email protected]"
-            ),
-            "McLean Hospital",
-        )
-        self.assertEqual(
-            process_report.get_institution_from_pi(institute_map, "[email protected]"),
-            "Harvard University",
-        )
-        self.assertEqual(
-            process_report.get_institution_from_pi(institute_map, "fake"), ""
-        )
-        self.assertEqual(
-            process_report.get_institution_from_pi(
-                institute_map, "[email protected]"
-            ),
-            "Northeastern University",
-        )
+        answers = {
+            "[email protected]": "Boston University",
+            "[email protected]": "McLean Hospital",
+            "[email protected]": "Harvard University",
+            "e@edu": "",
+            "[email protected]": "Northeastern University",
+            "[email protected]": "Harvard University",
+            "[email protected]": "Boston Children's Hospital",
+            "[email protected]": "Massachusetts Eye & Ear",
+            "[email protected]": "",
+            "[email protected]": "Brigham and Women's Hospital",
+            "[email protected]": "Beth Israel Deaconess Medical Center",
+        }
+
+        for pi_email, answer in answers.items():
+            self.assertEqual(
+                process_report.get_institution_from_pi(institute_map, pi_email), answer
+            )
 
 
 class TestAlias(TestCase):