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

Now, for PIs whose emails contain multiple subdomains (i.e `[email protected]`), each of their subdomains will be matched against institute_map.json, from the most qualified to least qualified subdomain. In the example of `[email protected]`, this means `b.c.edu` would be matched first, then `c.edu`, then `edu`
QuanMPhm · Jun 11, 2024 · 5e86c9e · 5e86c9e
1 parent 0400557
commit 5e86c9e
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 34 deletions.
diff --git a/process_report/institute_map.json b/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",

diff --git a/process_report/process_report.py b/process_report/process_report.py
@@ -44,8 +44,11 @@
 
 
 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!")

diff --git a/process_report/tests/unit_tests.py b/process_report/tests/unit_tests.py
@@ -234,29 +234,19 @@ def test_get_pi_institution(self):
             "northeastern.edu": "Northeastern University",
         }
 
-        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",
+        }
+
+        for pi_email, answer in answers.items():
+            self.assertEqual(
+                process_report.get_institution_from_pi(institute_map, pi_email), answer
+            )
 
 
 class TestAlias(TestCase):