thread fix

backend/app/services/external_reply_service.py

@@ -17,6 +17,7 @@ from app.services.message_ids import (
     message_id_fields,
 )
 from app.services.mime_parser import parse_mime
+from app.services.reply_extractor import build_reply_display
 from app.services.raw_mime_storage import get_raw_mime_storage
 from app.services.thread_service import threading_fields_from_parent
 
@@ -410,6 +411,10 @@ async def send_external_reply(
     )
 
     now = datetime.now(timezone.utc)
+    reply_display = build_reply_display(
+        body_text=plan.body_text,
+        body_html=plan.body_html,
+    )
     document = {
         "direction": "outbound",
         "message_scope": MESSAGE_SCOPE_EXTERNAL,
@@ -417,7 +422,10 @@ async def send_external_reply(
         "sender": {"name": sender_name, "address": sender},
         "to": [{"name": value.split("@")[0], "address": value} for value in plan.recipients],
         "subject": plan.subject,
-        "snippet": re.sub(r"\s+", " ", plan.body_text).strip()[:240],
+        "snippet": reply_display["snippet"],
+        "new_reply_content": reply_display["new_reply_content"],
+        "quoted_content": reply_display["quoted_content"],
+        "has_quoted_text": reply_display["has_quoted_text"],
         "body_text": plan.body_text,
         "body_html": plan.body_html,
         "attachments": [],

backend/app/services/inbound_smtp.py

@@ -13,6 +13,7 @@ from app.services.external_reply_service import log_original_external_message
 from app.services.human_handoff_service import MESSAGE_SCOPE_EXTERNAL
 from app.services.message_ids import canonical_message_id, message_id_fields
 from app.services.mime_parser import parse_mime
+from app.services.reply_extractor import build_reply_display
 from app.services.raw_mime_storage import delivery_digest, get_raw_mime_storage
 from app.services.thread_resolver import participants_from_parsed, resolve_thread
 from app.services.thread_service import ensure_thread_for_message, message_event_at
@@ -155,12 +156,26 @@ class InboundHandler:
                 raise RuntimeError("Inbound message was not persisted.")
 
             if body_text or body_html:
+                reply_display = build_reply_display(body_text=body_text, body_html=body_html)
                 await mongo_db.messages.update_one(
                     {"_id": persisted["_id"]},
-                    {"$set": {"body_text": body_text, "body_html": body_html}},
+                    {
+                        "$set": {
+                            "body_text": body_text,
+                            "body_html": body_html,
+                            "new_reply_content": reply_display["new_reply_content"],
+                            "quoted_content": reply_display["quoted_content"],
+                            "has_quoted_text": reply_display["has_quoted_text"],
+                            "snippet": reply_display["snippet"],
+                        }
+                    },
                 )
                 persisted["body_text"] = body_text
                 persisted["body_html"] = body_html
+                persisted["snippet"] = reply_display["snippet"]
+                persisted["new_reply_content"] = reply_display["new_reply_content"]
+                persisted["quoted_content"] = reply_display["quoted_content"]
+                persisted["has_quoted_text"] = reply_display["has_quoted_text"]
 
             participants = participants_from_parsed(parsed)
             for recipient in recipients:
@@ -214,7 +229,7 @@ class InboundHandler:
                             "sender": parsed["sender"],
                             "recipients": parsed["to"],
                             "subject": parsed["subject"],
-                            "snippet": parsed["snippet"],
+                            "snippet": persisted.get("snippet", parsed["snippet"]),
                             "attachments": parsed["attachments"],
                             "status": "received",
                             "is_read": False,

backend/app/services/mime_parser.py

@@ -6,6 +6,7 @@ from email.utils import getaddresses, parsedate_to_datetime
 from html import unescape
 
 from app.services.message_ids import first_message_id, parse_message_id_list
+from app.services.reply_extractor import build_reply_display
 
 
 def _addresses(values: list[str]) -> list[dict[str, str]]:
@@ -36,11 +37,7 @@ def _header_str(value) -> str | None:
 
 
 def _snippet(text: str, html: str | None, limit: int = 240) -> str:
-    source = text
-    if not source and html:
-        source = re.sub(r"<[^>]+>", " ", html)
-        source = unescape(source)
-    return re.sub(r"\s+", " ", source).strip()[:limit]
+    return build_reply_display(body_text=text, body_html=html, limit=limit)["snippet"]
 
 
 def parse_mime(raw: bytes) -> dict:

backend/app/services/reply_extractor.py

@@ -4,19 +4,13 @@ from html import unescape
 # Patterns that mark the start of quoted previous-message content.
 TEXT_QUOTE_MARKERS = [
     re.compile(r"\n-{2,}\s*Original Message\s*-{2,}", re.IGNORECASE),
-    re.compile(
-        r"\nOn .+? wrote:\s*\n",
-        re.IGNORECASE,
-    ),
-    re.compile(
-        r"\nOn .+?, .+? at .+? .+? wrote:\s*\n",
-        re.IGNORECASE,
-    ),
+    re.compile(r"(?:^|\n)\s*On .+? wrote:\s*", re.IGNORECASE),
     re.compile(
         r"\nFrom:\s*.+\nSent:\s*.+\nTo:\s*.+\nSubject:\s*.+",
         re.IGNORECASE,
     ),
     re.compile(r"\n_{5,}\n"),
+    re.compile(r"\n\s*> ?"),
 ]
 
 HTML_QUOTE_PATTERNS = [
@@ -110,6 +104,46 @@ def extract_new_reply_content(
     }
 
 
+def reply_fields_for_message(
+    message: dict,
+    *,
+    body_text: str | None,
+    body_html: str | None = None,
+) -> dict[str, str | bool]:
+    if "new_reply_content" in message:
+        return {
+            "new_reply_content": message.get("new_reply_content", ""),
+            "quoted_content": message.get("quoted_content", ""),
+            "has_quoted_text": bool(message.get("has_quoted_text") or message.get("quoted_content")),
+        }
+    built = build_reply_display(body_text=body_text, body_html=body_html)
+    return {
+        "new_reply_content": built["new_reply_content"],
+        "quoted_content": built["quoted_content"],
+        "has_quoted_text": built["has_quoted_text"],
+    }
+
+
+def build_reply_display(
+    *,
+    body_text: str | None,
+    body_html: str | None = None,
+    limit: int = 240,
+) -> dict[str, str | bool]:
+    extracted = extract_new_reply_content(body_text=body_text, body_html=body_html)
+    new_content = extracted["new_reply_content"]
+    quoted = extracted["quoted_content"]
+    fallback = re.sub(r"\s+", " ", (body_text or "")).strip()
+    snippet_source = new_content or fallback
+    return {
+        "new_reply_content": new_content,
+        "quoted_content": quoted,
+        "has_quoted_text": bool(quoted),
+        "raw_body": extracted["raw_body"],
+        "snippet": re.sub(r"\s+", " ", snippet_source).strip()[:limit],
+    }
+
+
 def validate_external_reply_body(body: str) -> bool:
     normalized = (body or "").strip()
     if not normalized: