feat: Increase max email fetch results and enhance email body extraction

src/app.py

@@ -364,7 +364,7 @@ def _sync_emails_once(cfg: dict) -> int:
             days_back = max(1, delta_days)
         except Exception:
             pass
-    max_results = 5
+    max_results = 100
     client = ZohoClient(
         email=cfg.get("zoho_email") or account_email,
         app_password=cfg.get("zoho_app_password"),

src/zoho_client.py

@@ -132,21 +132,28 @@ class ZohoClient:
                         email_message = email.message_from_bytes(raw_email)
                         date_header = email_message.get("Date", "")
                         email_date = parse_email_date_safely(date_header)
-                       
-                        
+
                         # Ensure both dates are timezone-aware for comparison
                         if email_date and latest_date:
                             print(f"📅 Email date: {email_date} Latest: {latest_date}")
                             # If latest_date is timezone-naive, make it timezone-aware (assume UTC)
                             if latest_date.tzinfo is None:
                                 latest_date = latest_date.replace(tzinfo=timezone.utc)
-                            
+
                             if (email_date > latest_date) or first_time:
-                            # Extract headers
-                                print(f"📅 Email date: {email_date} Latest: {latest_date}")
-                                subject = self._decode_header(email_message.get("Subject", ""))
-                                from_header = self._decode_header(email_message.get("From", ""))
-                                to_header = self._decode_header(email_message.get("To", ""))
+                                # Extract headers
+                                print(
+                                    f"📅 Email date: {email_date} Latest: {latest_date}"
+                                )
+                                subject = self._decode_header(
+                                    email_message.get("Subject", "")
+                                )
+                                from_header = self._decode_header(
+                                    email_message.get("From", "")
+                                )
+                                to_header = self._decode_header(
+                                    email_message.get("To", "")
+                                )
 
                                 message_id = email_message.get("Message-ID", "")
                                 in_reply_to = email_message.get("In-Reply-To", "")
@@ -156,7 +163,6 @@ class ZohoClient:
 
                                 # Get email body snippet
                                 body = self._get_email_body(email_message)
-                                
 
                                 email_data = {
                                     "id": num.decode(),
@@ -170,7 +176,7 @@ class ZohoClient:
                                     "folder": folder,
                                     "snippet": body,
                                 }
-                                
+
                                 emails.append(email_data)
                                 logging.info(f"Long body: {body}")
                 except Exception as e:
@@ -227,18 +233,22 @@ class ZohoClient:
             return str(header_value)
 
     def _get_email_body(self, email_message) -> str:
-        """Extract email body text"""
+        """Extract email body text - get only the main content, not quoted replies"""
         body = ""
 
         if email_message.is_multipart():
+            # Get only the first text/plain part (main content)
             for part in email_message.walk():
-                if part.get_content_type() == "text/plain":
+                if part.get_content_type() == "text/plain" and not part.get_filename():
                     try:
-                        body += part.get_payload(decode=True).decode(
+                        content = part.get_payload(decode=True).decode(
                             "utf-8", errors="ignore"
                         )
+                        # Take only the first text part we find
+                        body = content
+                        break  # Stop after first text/plain part
                     except Exception:
-                        pass
+                        continue
         else:
             try:
                 body = email_message.get_payload(decode=True).decode(
@@ -247,8 +257,41 @@ class ZohoClient:
             except Exception:
                 pass
 
+        # Optional: Clean up the body by removing quoted content
+        body = self._clean_email_body(body)
         return body
 
+    def _clean_email_body(self, body: str) -> str:
+        """Clean email body by removing quoted content and signatures"""
+        if not body:
+            return ""
+
+        lines = body.split("\n")
+        cleaned_lines = []
+
+        for line in lines:
+            line = line.strip()
+
+            # Stop at common quote indicators
+            if (
+                line.startswith("---- On ")
+                or line.startswith("On ")
+                and "wrote:" in line
+                or line.startswith("From:")
+                or line.startswith("> ")
+                or line.startswith("-----Original Message-----")
+                or line.startswith("---------- Forwarded message ---------")
+            ):
+                break
+
+            cleaned_lines.append(line)
+
+        # Remove trailing empty lines
+        while cleaned_lines and not cleaned_lines[-1]:
+            cleaned_lines.pop()
+
+        return "\n".join(cleaned_lines)
+
     def get_thread_messages(self, thread_id: str) -> List[Dict[str, Any]]:
         """Get all messages in a thread (simplified for IMAP)"""
         # For IMAP, we'll return a single message since thread grouping is more complex