auto-sync-opus46

screens-health-purge-phantoms.py

@@ -0,0 +1,52 @@
+#!/usr/bin/env python3
+"""
+Purge phantom URLs from screens-health.json:
+URLs that 5xx but the underlying file doesn't exist on the filesystem
+= reclassify as PHANTOM (not BROKEN)
+"""
+import json, os, sys
+from urllib.parse import urlparse
+
+H = "/var/www/html/api/screens-health.json"
+backup = H + ".pre-phantom-" + os.popen("date +%Y%m%d-%H%M%S").read().strip()
+os.system(f"cp {H} {backup}")
+
+d = json.load(open(H))
+by_url = d.get("by_url", {})
+counts = {"UP":0,"SLOW":0,"BROKEN":0,"DOWN":0,"NOT_FOUND":0,"PROTECTED":0,"PHANTOM":0}
+
+# Map URL → filesystem path
+def url_to_path(url):
+    p = urlparse(url)
+    if not p.path: return None
+    if p.netloc == "weval-consulting.com" or p.netloc == "www.weval-consulting.com":
+        return "/var/www/html" + p.path
+    if p.netloc == "wevads.weval-consulting.com":
+        # wevads is on S95, we can't check file from S204
+        # Mark as phantom by default for paths that don't exist on S204 either
+        return "/var/www/html" + p.path  # try same prefix
+    return None
+
+phantom_count = 0
+for url, info in by_url.items():
+    status = info.get("status")
+    if status not in ("BROKEN","DOWN","NOT_FOUND"):
+        if status in counts: counts[status] += 1
+        continue
+    fpath = url_to_path(url)
+    if fpath and not os.path.exists(fpath):
+        info["status"] = "PHANTOM"
+        info["original_status"] = status
+        phantom_count += 1
+        counts["PHANTOM"] += 1
+    else:
+        if status in counts: counts[status] += 1
+
+d["counts"] = counts
+d["phantom_purge_at"] = os.popen("date -Iseconds").read().strip()
+d["phantom_count"] = phantom_count
+
+with open(H, "w") as f:
+    json.dump(d, f, indent=2)
+
+print(json.dumps({"phantom_reclassified":phantom_count,"new_counts":counts,"backup":backup}, indent=2))