fix(compliance-check): respect auto-discovery 'not found' verdict; DSB not canonical

Two related bugs in the BMW test result:

1. AGB rendered as 'MANGELHAFT 0/13' even though BMW has no public AGB:
   - Auto-discovery correctly returned 'not found' for AGB (no link on
     bmw.de matches AGB keywords).
   - But auto_fill_from_dsi then found the substring 'AGB' in a section
     of the DSI and pseudo-filled the AGB entry with a 264-word DSI
     fragment.
   - cross_search_documents would have done the same.
   - Both now skip entries where discovery_attempted=True AND
     auto_discovered=False — the 'not found' verdict stands.

2. DSB-Kontakt rendered as a separate 100% OK document with 7566 words
   = the entire DSI text:
   - GDPR practice: the DSB is named *inside* the DSI as an email or
     contact block (Art. 13(1)(b)), not as a stand-alone page.
   - cross_search_documents had been assigning the full DSI to the DSB
     row because it matched 'datenschutzbeauftragte' keywords.
   - DSB removed from _ALL_DOC_TYPES — no longer canonical, no longer
     padded as missing, no longer auto-discovered. The frontend row
     remains so a tenant with a separate DSB page can still submit one.

After this fix BMW should render:
- DSE: OK
- Impressum: LUECKENHAFT (unchanged — regex gaps to fix separately)
- Cookie-Richtlinie: OK
- Social Media: NICHT GEFUNDEN (bmw.de does not link to it)
- AGB: NICHT GEFUNDEN (correct — BMW has no public AGB)
- Nutzungsbedingungen: NICHT GEFUNDEN
- Widerruf: NICHT GEFUNDEN

backend-compliance/compliance/api/agent_compliance_check_routes.py

@@ -890,13 +890,19 @@ _DOC_TYPE_LABELS = {
     "dsb": "DSB-Kontakt",
 }
 
-# Canonical 8 doc types in the same order as the frontend ComplianceCheckTab.
+# Canonical doc types in the same order as the frontend ComplianceCheckTab.
 # The route pads `results` to always contain an entry for each — even if
 # the user did not submit a URL — so the email + frontend always show
 # the complete checklist (missing rows marked as 'Nicht eingereicht').
+#
+# DSB-Kontakt is intentionally NOT canonical: per GDPR practice the DSB is
+# named *inside* the DSI/datenschutz document (email or contact block), not
+# as a separate page. We check 'DSB benannt' as a sub-check of the DSE
+# instead. If a tenant insists on a separate DSB document, they can still
+# submit one — it just won't appear as a missing checklist row.
 _ALL_DOC_TYPES = [
     "dse", "impressum", "social_media", "cookie",
-    "agb", "nutzungsbedingungen", "widerruf", "dsb",
+    "agb", "nutzungsbedingungen", "widerruf",
 ]
 
 

backend-compliance/compliance/services/section_splitter.py

@@ -199,6 +199,12 @@ def auto_fill_from_dsi(doc_entries: list[dict]) -> None:
     for entry in doc_entries:
         if entry.get("text") or entry.get("url"):
             continue  # Already has content
+        # Auto-discovery already tried + decided: skip. Don't override its
+        # 'NICHT GEFUNDEN' verdict with a pseudo-match from DSI sections
+        # (which produces false MANGELHAFT findings for genuinely missing
+        # docs like BMW's AGB).
+        if entry.get("discovery_attempted") and not entry.get("auto_discovered"):
+            continue
 
         doc_type = entry["doc_type"]
         section_text = _find_section_for_type(sections, doc_type)
@@ -267,8 +273,10 @@ def cross_search_documents(doc_entries: list[dict]) -> list[dict]:
         return findings
 
     # For each entry, check if:
-    # a) It's empty → search other texts
-    # b) It has text but the text doesn't match the doc_type → search other texts
+    # a) It has text but the text doesn't match the doc_type → search other texts
+    # (Empty entries from auto-discovery 'not found' are NOT pseudo-filled
+    # from other docs — that would silently revive a 'NICHT GEFUNDEN' verdict
+    # as a misleading MANGELHAFT row.)
     for entry in doc_entries:
         target_type = entry["doc_type"]
         keywords = _DOC_TYPE_KEYWORDS.get(target_type, [])
@@ -278,13 +286,15 @@ def cross_search_documents(doc_entries: list[dict]) -> list[dict]:
         has_text = entry.get("text") and len(entry["text"].split()) > 50
         text_matches = False
         if has_text:
-            # Check if the current text actually contains this doc_type's content
             entry_lower = entry["text"].lower()
             match_score = sum(1 for kw in keywords if kw in entry_lower)
             text_matches = match_score >= 2
 
         if has_text and text_matches:
             continue  # Text present AND matches doc_type → skip
+        # Skip empty entries the auto-discovery has already ruled on.
+        if not has_text and entry.get("discovery_attempted") and not entry.get("auto_discovered"):
+            continue
 
         # Search all other texts for this doc_type's keywords
         best_match: dict | None = None