Phase 1 project prototype

tests/test_scraper_classify.py

@@ -0,0 +1,76 @@
+from admin_analytics.scraper.classify import classify_title, is_overhead
+
+
+def test_grants_analyst():
+    assert classify_title("Grants Analyst II") == "GRANTS_ADMIN"
+    assert classify_title("Senior Grants Analyst") == "GRANTS_ADMIN"
+    assert classify_title("Manager, Grant Administration, Pre-Award") == "GRANTS_ADMIN"
+    assert classify_title("Closeout Coordinator, Research") == "GRANTS_ADMIN"
+
+
+def test_research_staff():
+    assert classify_title("Research Associate") == "RESEARCH"
+    assert classify_title("Associate Scientist") == "RESEARCH"
+    assert classify_title("Computer Scientist NIST") == "RESEARCH"
+    assert classify_title("Postdoctoral Researcher") == "RESEARCH"
+
+
+def test_academic_support():
+    assert classify_title("Undergraduate Academic Advisor") == "ACADEMIC_SUPPORT"
+    assert classify_title("Graduate Services Coordinator") == "ACADEMIC_SUPPORT"
+    assert classify_title("Academic Program Manager") == "ACADEMIC_SUPPORT"
+
+
+def test_admin_support():
+    assert classify_title("Administrative Assistant IV") == "ADMIN_SUPPORT"
+    assert classify_title("Administrative Specialist") == "ADMIN_SUPPORT"
+
+
+def test_it():
+    assert classify_title("Computing Support Specialist II") == "IT"
+    assert classify_title("Systems Programmer IV") == "IT"
+    assert classify_title("Director of Computing Operations") == "IT"
+
+
+def test_finance():
+    assert classify_title("Financial Specialist") == "FINANCE"
+    assert classify_title("Director, Procurement & Financial Processing") == "FINANCE"
+    assert classify_title("Sr. Business Officer") == "FINANCE"
+
+
+def test_leadership():
+    assert classify_title("Dean") == "LEADERSHIP"
+    assert classify_title("Associate Dean for Academic Affairs") == "LEADERSHIP"
+    assert classify_title("Chief of Staff") == "LEADERSHIP"
+
+
+def test_communications():
+    assert classify_title("Communications Director") == "COMMUNICATIONS"
+    assert classify_title("Digital Communications Specialist") == "COMMUNICATIONS"
+
+
+def test_technical():
+    assert classify_title("Master Machinist") == "TECHNICAL"
+    assert classify_title("Lab Manager") == "TECHNICAL"
+    assert classify_title("Lab Coordinator II") == "TECHNICAL"
+
+
+def test_faculty_not_admin():
+    assert classify_title("Adjunct Professor NIST") == "FACULTY"
+    assert classify_title("Affiliated Associate Professor") == "FACULTY"
+
+
+def test_overhead_classification():
+    assert is_overhead("LEADERSHIP") is True
+    assert is_overhead("FINANCE") is True
+    assert is_overhead("IT") is True
+    assert is_overhead("RESEARCH") is False
+    assert is_overhead("ACADEMIC_SUPPORT") is False
+    assert is_overhead("TECHNICAL") is False
+    assert is_overhead("GRANTS_ADMIN") is None  # debatable
+
+
+def test_unknown():
+    assert classify_title("Football Coach") == "UNKNOWN"
+    assert classify_title(None) == "UNKNOWN"
+    assert classify_title("") == "UNKNOWN"