Implement attachment view

core/management/commands/backfill_xmpp_attachment_urls.py

@@ -0,0 +1,162 @@
+from django.core.management.base import BaseCommand
+from django.db.models import Q
+
+from core.models import Message, MessageEvent
+
+
+EMPTY_TEXT_VALUES = {
+    "",
+    "[No Body]",
+    "[no body]",
+}
+
+
+def _normalize_url(value):
+    if not value:
+        return ""
+    text = str(value).strip().rstrip(".,);:!?\"'")
+    if text.startswith("http://") or text.startswith("https://"):
+        return text
+    return ""
+
+
+def _extract_urls_from_attachment_blob(blob):
+    urls = []
+    if isinstance(blob, str):
+        normalized = _normalize_url(blob)
+        if normalized:
+            urls.append(normalized)
+        return urls
+
+    if isinstance(blob, dict):
+        for key in ("url", "source_url", "download_url", "proxy_url", "href"):
+            normalized = _normalize_url(blob.get(key))
+            if normalized:
+                urls.append(normalized)
+        nested = blob.get("attachments")
+        if isinstance(nested, list):
+            for row in nested:
+                urls.extend(_extract_urls_from_attachment_blob(row))
+        return urls
+
+    if isinstance(blob, list):
+        for row in blob:
+            urls.extend(_extract_urls_from_attachment_blob(row))
+    return urls
+
+
+def _uniq(values):
+    seen = set()
+    output = []
+    for value in values:
+        if value in seen:
+            continue
+        seen.add(value)
+        output.append(value)
+    return output
+
+
+class Command(BaseCommand):
+    help = (
+        "Backfill empty Message.text rows originating from XMPP by recovering "
+        "attachment URLs from MessageEvent metadata."
+    )
+
+    def add_arguments(self, parser):
+        parser.add_argument(
+            "--apply",
+            action="store_true",
+            help="Persist updates. Without this flag, runs as dry-run.",
+        )
+        parser.add_argument(
+            "--user-id",
+            type=int,
+            default=None,
+            help="Limit processing to one user ID.",
+        )
+        parser.add_argument(
+            "--limit",
+            type=int,
+            default=0,
+            help="Maximum number of candidate rows to inspect (0 = no limit).",
+        )
+
+    def _candidate_events(self, message):
+        linked = MessageEvent.objects.filter(
+            user=message.user,
+            raw_payload_ref__legacy_message_id=str(message.id),
+        )
+        if linked.exists():
+            return linked
+
+        # Fallback heuristic for older rows with missing legacy refs.
+        window = 2000
+        return MessageEvent.objects.filter(
+            user=message.user,
+            source_system="xmpp",
+            ts__gte=int(message.ts) - window,
+            ts__lte=int(message.ts) + window,
+        ).exclude(attachments=[])
+
+    def handle(self, *args, **options):
+        apply_changes = bool(options.get("apply"))
+        user_id = options.get("user_id")
+        limit = int(options.get("limit") or 0)
+
+        queryset = Message.objects.filter(
+            sender_uuid__iexact="xmpp",
+        ).filter(
+            Q(text__isnull=True) | Q(text__exact="") | Q(text__iexact="[No Body]")
+        )
+        if user_id:
+            queryset = queryset.filter(user_id=user_id)
+        queryset = queryset.order_by("ts", "id")
+        if limit > 0:
+            queryset = queryset[:limit]
+
+        inspected = 0
+        recoverable = 0
+        updated = 0
+        unrecoverable = 0
+
+        for message in queryset.iterator():
+            inspected += 1
+            current_text = str(message.text or "").strip()
+            if current_text not in EMPTY_TEXT_VALUES:
+                continue
+
+            urls = []
+            for event in self._candidate_events(message):
+                urls.extend(_extract_urls_from_attachment_blob(event.attachments))
+                urls.extend(
+                    _extract_urls_from_attachment_blob(event.raw_payload_ref or {})
+                )
+                if urls:
+                    break
+
+            urls = _uniq(urls)
+            if not urls:
+                unrecoverable += 1
+                continue
+
+            recoverable += 1
+            new_text = "\n".join(urls)
+            if apply_changes:
+                message.text = new_text
+                message.save(update_fields=["text"])
+                updated += 1
+            else:
+                self.stdout.write(
+                    f"[dry-run] {message.id}: would set {len(urls)} URL(s)"
+                )
+
+        mode = "apply" if apply_changes else "dry-run"
+        self.stdout.write(
+            self.style.SUCCESS(
+                "XMPP attachment URL backfill complete "
+                f"({mode}): inspected={inspected}, "
+                f"recoverable={recoverable}, "
+                f"updated={updated}, "
+                f"unrecoverable={unrecoverable}"
+            )
+        )