Spaces:
Running
Running
File size: 3,394 Bytes
7fee926 e00943e 2eba2a0 5de8b59 2eba2a0 5de8b59 7fee926 7a4edb9 7fee926 7a4edb9 5de8b59 c860ff8 7a4edb9 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 | import re
import json
from collections import defaultdict
from datetime import datetime, timezone, timedelta
_VN_TZ = timezone(timedelta(hours=7))
def format_created_at(raw: str) -> str:
if not raw:
return "Unknown time"
try:
dt = datetime.fromisoformat(raw).astimezone(_VN_TZ)
return dt.strftime("%d/%m/%Y %H:%M:%S")
except (ValueError, TypeError):
return raw
def extract_metadata_from_messages(messages: list[dict]) -> dict:
"""
Trích xuất danh sách Link và File từ danh sách tin nhắn.
"""
links = []
files = []
# Regex đơn giản cho URL
url_pattern = r'https?://[^\s]+'
for msg in messages:
content = msg.get("content", "")
# Tìm links
found_links = re.findall(url_pattern, content)
links.extend(found_links)
# Tìm files từ trường attachment
attachment_raw = msg.get("attachment")
if attachment_raw and attachment_raw != "null":
try:
# Trường attachment có thể là string JSON hoặc dict tùy vào RedisClient
if isinstance(attachment_raw, str):
att = json.loads(attachment_raw)
else:
att = attachment_raw
files.append({
"id": att.get("id"),
"type": att.get("type"),
"name": att.get("name"),
"sender": msg.get("senderName")
})
except:
pass
return {
"links": list(set(links)), # Loại bỏ trùng
"files": files
}
def group_messages_by_thread(messages: list[dict]) -> dict[str, list[dict]]:
# ... (rest of the file)
"""
Gom nhóm các tin nhắn có cùng roomId lại với nhau.
"""
threads = defaultdict(list)
for msg in messages:
# Use roomId as defined in docs/redis-data-structure.md
room_id = msg.get("roomId", "unknown")
threads[room_id].append(msg)
# Sort each thread by timestamp to maintain chronological order
for room_id in threads:
threads[room_id].sort(key=lambda m: m.get("timestamp", ""))
return dict(threads)
def format_thread_for_prompt(room_id: str, messages: list[dict]) -> str:
"""
Chuyển một thread thành đoạn văn bản rõ ràng để đưa vào prompt.
"""
lines = []
for msg in messages:
raw_ts = msg.get("created_at") or msg.get("timestamp", "")
timestamp = format_created_at(raw_ts) if raw_ts else "Unknown time"
user = msg.get("sender_username") or msg.get("name") or msg.get("senderName") or msg.get("sender_id", "Unknown")
content = msg.get("content", "")
lines.append(f"[{timestamp}] {user}: {content}")
thread_text = "\n".join(lines)
return f"=== ROOM/THREAD: {room_id} ===\n{thread_text}\n=== END ===\n"
def preprocess_messages(messages: list[dict]) -> str:
"""
Tiền xử lý toàn bộ: gom nhóm theo room → format thành prompt-ready string.
"""
threads = group_messages_by_thread(messages)
formatted_parts = []
for room_id, thread_msgs in threads.items():
formatted_parts.append(format_thread_for_prompt(room_id, thread_msgs))
return "\n".join(formatted_parts)
|