debug: instrument platform presence scan for missing listings regression

app/services/isbn_catalog.py

@@ -487,9 +487,33 @@ async def verify_isbn_anchor(
             snippet = (resp.text or "")[:120_000].lower()
             tokens = [t for t in re.sub(r"[^\w\s]", " ", core).split() if len(t) > 3]
             if tokens and sum(1 for t in tokens[:3] if t.lower() in snippet) < 2:
+                # #region agent log
+                try:
+                    import time as _tv, json as _jv
+                    with open("debug-7c8974.log", "a") as _fv:
+                        _fv.write(_jv.dumps({"sessionId":"7c8974","hypothesisId":"A","location":"isbn_catalog.py:verify_isbn_anchor","message":"anchor_title_fail","data":{"platform":platform_id,"anchor":anchor,"final":final,"clean":clean,"found_tokens":sum(1 for t in tokens[:3] if t.lower() in snippet)},"timestamp":int(_tv.time()*1000)}) + "\n")
+                except Exception:
+                    pass
+                # #endregion
                 return None
+        # #region agent log
+        try:
+            import time as _tv2, json as _jv2
+            with open("debug-7c8974.log", "a") as _fv2:
+                _fv2.write(_jv2.dumps({"sessionId":"7c8974","hypothesisId":"A","location":"isbn_catalog.py:verify_isbn_anchor","message":"anchor_success","data":{"platform":platform_id,"anchor":anchor,"final":final,"clean":clean,"status":resp.status_code},"timestamp":int(_tv2.time()*1000)}) + "\n")
+        except Exception:
+            pass
+        # #endregion
         return clean
     except Exception as exc:
+        # #region agent log
+        try:
+            import time as _tve, json as _jve
+            with open("debug-7c8974.log", "a") as _fve:
+                _fve.write(_jve.dumps({"sessionId":"7c8974","hypothesisId":"A","location":"isbn_catalog.py:verify_isbn_anchor","message":"anchor_exception","data":{"platform":platform_id,"anchor":anchor,"error":str(exc)[:200]},"timestamp":int(_tve.time()*1000)}) + "\n")
+        except Exception:
+            pass
+        # #endregion
         logger.debug(
             "ISBN anchor verify failed",
             platform=platform_id,

app/services/platform_presence.py

@@ -754,6 +754,15 @@ def _finalize_listing_url(listing: PlatformListing) -> None:
                 clean = rewritten
         except Exception:
             pass
+    # #region agent log
+    _cls = classify_retailer_url(clean, pid) if clean else "none"
+    try:
+        import time as _t2, json as _j2
+        with open("debug-7c8974.log", "a") as _f2:
+            _f2.write(_j2.dumps({"sessionId":"7c8974","hypothesisId":"B","location":"platform_presence.py:_finalize","message":"finalize_url","data":{"pid":pid,"raw":listing.listing_url,"clean":clean,"cls":_cls},"timestamp":int(_t2.time()*1000)}) + "\n")
+    except Exception:
+        pass
+    # #endregion
     if not clean or classify_retailer_url(clean, pid) != "product":
         listing.listing_url = None
         listing.status = "not_found"
@@ -952,6 +961,16 @@ async def scan_platform_presence(
             author=snap.author or author,
         )
 
+        # #region agent log
+        import time as _time
+        try:
+            with open("debug-7c8974.log", "a") as _f:
+                import json as _j
+                _f.write(_j.dumps({"sessionId":"7c8974","hypothesisId":"A-D","location":"platform_presence.py:after_enrich","message":"retailer_links_after_enrich","data":{"links":{k:v for k,v in (snap.retailer_links or {}).items()}},"timestamp":int(_time.time()*1000)}) + "\n")
+        except Exception:
+            pass
+        # #endregion
+
         if not snap.asin:
             resolved_asin = _extract_asin(scan_isbn or isbn)
             if not resolved_asin and gr_links.get("amazon"):