From 566202dcad02b0a32d2fcabab32d107a3e7e7027 Mon Sep 17 00:00:00 2001
From: pewdiepie-archdaemon <pewdiepie-archdaemon@users.noreply.github.com>
Date: Thu, 17 Sep 2026 20:07:51 +0000
Subject: [PATCH] require domain evidence for official source shortcut

---
 src/clean_agent_preview.py        | 20 +++++++-------------
 tests/test_clean_agent_preview.py | 19 ++++++-------------
 2 files changed, 13 insertions(+), 26 deletions(-)

diff --git a/src/clean_agent_preview.py b/src/clean_agent_preview.py
index 1c14b61de..d44f0f385 100644
--- a/src/clean_agent_preview.py
+++ b/src/clean_agent_preview.py
@@ -3251,17 +3251,7 @@ def web_source_links(raw, *, max_items=1, prefer_official=False, query=''):
                 host == domain or host.endswith('.' + domain)
                 for domain in official_domains
             )
-            direct_document = bool(re.search(
-                r'\.(?:pdf|docx?|xlsx?|pptx?)(?:$|[?#])', row[1], re.I,
-            ))
-            query_tokens_in_url = any(
-                token in row[1].casefold()
-                for token in query_tokens
-                if len(token) >= 4
-            )
-            if query_host_match or (
-                not official_domains and direct_document and query_tokens_in_url
-            ):
+            if query_host_match:
                 primary.append(row)
         rows = primary
     links = []
@@ -5185,10 +5175,14 @@ async def stream_preview(*, endpoint_url, model, messages, headers, turn_contrac
                         and requested_web_source_links(direct_user_text)
                     ):
                         requested_links = requested_web_link_limit(direct_user_text)
+                        official_requested = bool(re.search(r'\bofficial\b', direct_user_text, re.I))
+                        known_official_domains = official_domains_for_text(
+                            direct_user_text + ' ' + str(args.get('query', ''))
+                        )
                         source_links = web_source_links(
                             output,
                             max_items=requested_links or 1,
-                            prefer_official=bool(re.search(r'\bofficial\b', direct_user_text, re.I)),
+                            prefer_official=official_requested,
                             query=args.get('query', ''),
                         )
                         source_only_request = not re.search(
@@ -5203,7 +5197,7 @@ async def stream_preview(*, endpoint_url, model, messages, headers, turn_contrac
                                 link for _, link in source_links[:requested_links]
                             )
                         elif (requested_links and not source_links
-                              and re.search(r'\bofficial\b', direct_user_text, re.I)):
+                              and official_requested and known_official_domains):
                             if not official_source_retry_attempted and round_number < round_limit:
                                 official_source_retry_attempted = True
                                 force_web_search_next_round = True
diff --git a/tests/test_clean_agent_preview.py b/tests/test_clean_agent_preview.py
index 717f78d34..51727750a 100644
--- a/tests/test_clean_agent_preview.py
+++ b/tests/test_clean_agent_preview.py
@@ -2012,7 +2012,7 @@ def test_official_manual_does_not_invent_pdf_requirement():
     assert explicit['query'].endswith('filetype:pdf')
 
 
-def test_unknown_official_domain_rejects_reseller_but_accepts_direct_document():
+def test_unknown_official_domain_cannot_be_proven_by_pdf_suffix():
     raw = '''
 [1] WIKING Miro 4 Wood Burning Stove
     https://scottishstovecentre.co.uk/product/wiking-miro-4/
@@ -2025,11 +2025,9 @@ def test_unknown_official_domain_rejects_reseller_but_accepts_direct_document():
         query='WIKING Miro stove manual official source filetype:pdf',
     )
 
-    assert links == [(
-        'https://www.hwam.com/pub/media/wiking/53-0756_Miro_EN.pdf',
-        '[Source: WIKING Miro Installation and User Manual]'
-        '(https://www.hwam.com/pub/media/wiking/53-0756_Miro_EN.pdf)',
-    )]
+    assert links == []
+    # Candidates remain in the full tool output for the model to inspect.
+    assert len(web_source_links(raw, max_items=2, query='WIKING Miro')) == 2
 
 
 def test_web_fetch_collapses_single_and_batch_url_fields_without_losing_targets():
@@ -2044,7 +2042,7 @@ def test_web_fetch_collapses_single_and_batch_url_fields_without_losing_targets(
     }
 
 
-def test_manual_locator_prefers_manufacturer_link_and_requests_one_result():
+def test_manual_locator_requests_one_result_but_needs_manufacturer_evidence():
     from src.clean_agent_preview import (
         requested_web_link_limit, requested_web_source_links, web_source_links,
     )
@@ -2061,12 +2059,7 @@ def test_manual_locator_prefers_manufacturer_link_and_requests_one_result():
     assert requested_web_link_limit(prompt) == 1
     assert web_source_links(
         raw, max_items=1, prefer_official=True, query=prompt,
-    ) == [
-        (
-            'https://www.hwam.com/media/wiking/miro-en.pdf',
-            '[Source: WIKING Miro English manual PDF](https://www.hwam.com/media/wiking/miro-en.pdf)',
-        )
-    ]
+    ) == []
 
 
 def test_preview_allows_only_reversible_client_local_ui_control():