Fix report heuristic

olms/discover.py

@@ -96,16 +96,14 @@ def fetch(session, rpt_id, form):
 
 def fetch_assigned(session, rpt_id, form):
     """True if `rpt_id` is assigned to `form`.
-
-    Real form pages embed an Angular app (`ng-app="LM20App"` etc.) and
-    fetch their data asynchronously. The OLMS "not found" page is a
-    plain HTML stub without ng-app. Body size is unreliable as a
-    discriminator (form templates are ~8.5K, the stub is ~8.2K).
+
+    Body size is unreliable as a discriminator (form templates are ~8.5K, 
+    the stub is ~8.2K). Error pages are labeled as such.
     """
     content_type, body = fetch(session, rpt_id, form)
     if content_type == "application/pdf":
         return True
-    return content_type == "text/html" and b"ng-app=" in body
+    return content_type == "text/html" and b"Error Page" not in body
 
 
 def is_assigned(session, rpt_id):
@@ -182,7 +180,7 @@ def discover(config, db):
     if max_known and not is_assigned(sess, max_known):
         raise RuntimeError(
             f"rptId {max_known} is in {db} but the OLMS probe reports"
-            " it unassigned; the ng-app/content-type heuristics are broken"
+            " it unassigned; the error label/content-type heuristics are broken"
         )
 
     max_assigned = bisect_max_assigned(sess, max_known)

tests/test_discover.py

@@ -15,7 +15,7 @@
 )
 
 
-def response(status=200, content_type="text/html", body=b"ng-app="):
+def response(status=200, content_type="text/html", body=b"success body"):
     r = Mock()
     r.status_code = status
     r.headers = {"Content-Type": content_type}
@@ -31,7 +31,7 @@ def test_fetch_backs_off_then_succeeds():
     with patch.object(discover.time, "sleep") as sleep:
         content_type, body = fetch(session, 1, "LM10Form")
     assert content_type == "text/html"
-    assert body == b"ng-app="
+    assert body == b"success body"
     assert [c.args[0] for c in sleep.call_args_list] == [5, 10]
 
 
@@ -55,8 +55,8 @@ def test_fetch_assigned():
     session = Mock()
     session.get.side_effect = [
         response(content_type="application/pdf"),
-        response(body=b"<html>not found stub</html>"),
-        response(body=b'<html ng-app="LM20App">'),
+        response(body=b"Error Page"),
+        response(),
     ]
     assert fetch_assigned(session, 1, "LM2Form") is True
     assert fetch_assigned(session, 2, "LM2Form") is False
@@ -125,7 +125,7 @@ def test_fetch_hit_html_falls_through_mismatched_forms():
 
     session = Mock()
     session.get.side_effect = [
-        response(body=b'<html ng-app="LM20App">bare shell</html>'),
+        response(body=b'<html>bare shell</html>'),
         response(body=b"<html>Signature ... rendered LM-21 ...</html>"),
     ]
     body = fetch_hit_html(session, 941283, ("LM20Form", "LM21Form"))