mirror of
https://gitverse.ru/kpa39l/chronicle.nixg.ru.git
synced 2026-09-29 09:55:08 +00:00
Phase 0 MVP: Telegram Archiver fully functional (992 posts tested)
This commit is contained in:
@@ -64,10 +64,11 @@ class TelethonArchiver:
|
||||
session=self.settings.session_path,
|
||||
api_id=self.settings.api_id,
|
||||
api_hash=self.settings.api_hash,
|
||||
device_model="Telegram Archiver",
|
||||
device_model="Desktop (Ubuntu)", # Helps with code delivery
|
||||
system_version="Ubuntu 22.04",
|
||||
app_version="1.0.0",
|
||||
lang_code="en",
|
||||
system_lang_code="en",
|
||||
lang_code="ru", # Russian language
|
||||
system_lang_code="ru",
|
||||
)
|
||||
|
||||
await self.client.connect()
|
||||
@@ -83,26 +84,58 @@ class TelethonArchiver:
|
||||
async def _authenticate(self) -> None:
|
||||
"""Handle the authentication flow."""
|
||||
try:
|
||||
# Send code request
|
||||
await self.client.send_code_request(self.settings.phone)
|
||||
logger.info(f"Code sent to {self.settings.phone}")
|
||||
# Try app code first
|
||||
logger.info("Requesting auth code via Telegram app...")
|
||||
sent_code = await self.client.send_code_request(
|
||||
self.settings.phone,
|
||||
force_sms=False
|
||||
)
|
||||
logger.info(f"Code request sent. Type: {sent_code.type}")
|
||||
|
||||
# Get code from user (in MVP, we'll use input())
|
||||
code = input("Enter the code you received: ")
|
||||
# Get code from user
|
||||
print("\n" + "=" * 50)
|
||||
print("AUTHENTICATION REQUIRED")
|
||||
print("=" * 50)
|
||||
print("Check your Telegram app (NOT SMS) for the code.")
|
||||
print("Look for a message from 'Telegram' with the code.")
|
||||
print("=" * 50)
|
||||
print("\nOptions:")
|
||||
print(" 1. Enter the code from Telegram app")
|
||||
print(" 2. Type 'sms' to receive code via SMS")
|
||||
print(" 3. Press Ctrl+C to cancel")
|
||||
print("=" * 50)
|
||||
|
||||
user_input = input("\nEnter code (or 'sms'): ").strip()
|
||||
|
||||
# If user requests SMS
|
||||
if user_input.lower() == 'sms':
|
||||
logger.info("Requesting SMS code...")
|
||||
print("\n📱 Requesting SMS code...")
|
||||
sent_code = await self.client.send_code_request(
|
||||
self.settings.phone,
|
||||
force_sms=True
|
||||
)
|
||||
print("SMS sent! Enter the code:")
|
||||
user_input = input("SMS Code: ").strip()
|
||||
|
||||
code = user_input
|
||||
|
||||
try:
|
||||
await self.client.sign_in(
|
||||
phone=self.settings.phone,
|
||||
code=code
|
||||
code=code,
|
||||
phone_code_hash=sent_code.phone_code_hash
|
||||
)
|
||||
except SessionPasswordNeededError:
|
||||
# 2FA is enabled
|
||||
logger.info("2FA enabled. Enter password:")
|
||||
print("\n🔐 2FA enabled")
|
||||
password = input("2FA Password: ")
|
||||
await self.client.sign_in(password=password)
|
||||
|
||||
except FloodWaitError as e:
|
||||
logger.error(f"Flood wait: must wait {e.seconds} seconds")
|
||||
print(f"\n⚠️ You must wait {e.seconds} seconds before trying again.")
|
||||
raise
|
||||
except Exception as e:
|
||||
logger.error(f"Authentication failed: {e}")
|
||||
@@ -165,24 +198,33 @@ class TelethonArchiver:
|
||||
|
||||
# Iterate messages (newest first)
|
||||
message_count = 0
|
||||
async for message in self.client.iter_messages(
|
||||
entity,
|
||||
limit=limit,
|
||||
min_id=from_message_id, # Messages with ID > from_message_id
|
||||
):
|
||||
yield message
|
||||
message_count += 1
|
||||
error_count = 0
|
||||
|
||||
# Build iter_messages kwargs
|
||||
iter_kwargs = {"limit": limit}
|
||||
if from_message_id:
|
||||
iter_kwargs["min_id"] = from_message_id
|
||||
|
||||
async for message in self.client.iter_messages(entity, **iter_kwargs):
|
||||
try:
|
||||
yield message
|
||||
message_count += 1
|
||||
|
||||
if message_count % 100 == 0:
|
||||
logger.debug(f"Fetched {message_count} messages...")
|
||||
if message_count % 50 == 0:
|
||||
logger.info(f"Progress: {message_count} messages fetched...")
|
||||
except Exception as e:
|
||||
error_count += 1
|
||||
logger.error(f"Error processing message {message.id}: {e}")
|
||||
continue # Skip problematic messages
|
||||
|
||||
logger.info(f"Finished fetching. Total messages: {message_count}")
|
||||
logger.info(f"Finished fetching. Total: {message_count}, Errors: {error_count}")
|
||||
|
||||
async def download_media(
|
||||
self,
|
||||
message: Message,
|
||||
output_path: Path,
|
||||
max_size: Optional[int] = None,
|
||||
timeout: int = 60, # Timeout in seconds
|
||||
) -> list[dict]:
|
||||
"""
|
||||
Download all media from a message.
|
||||
@@ -191,6 +233,7 @@ class TelethonArchiver:
|
||||
message: Telegram message with media
|
||||
output_path: Directory to save media files
|
||||
max_size: Maximum file size to download (bytes)
|
||||
timeout: Timeout for each file download in seconds
|
||||
|
||||
Returns:
|
||||
List of dicts with file info (filename, type, size, path, is_too_large)
|
||||
@@ -204,20 +247,40 @@ class TelethonArchiver:
|
||||
# Ensure output directory exists
|
||||
output_path.mkdir(parents=True, exist_ok=True)
|
||||
|
||||
# Handle different media types
|
||||
if isinstance(message.media, MessageMediaPhoto):
|
||||
# Photo
|
||||
info = await self._download_photo(message, output_path, max_size)
|
||||
if info:
|
||||
media_info.append(info)
|
||||
try:
|
||||
# Handle different media types with timeout
|
||||
if isinstance(message.media, MessageMediaPhoto):
|
||||
# Photo
|
||||
info = await asyncio.wait_for(
|
||||
self._download_photo(message, output_path, max_size),
|
||||
timeout=timeout
|
||||
)
|
||||
if info:
|
||||
media_info.append(info)
|
||||
|
||||
elif isinstance(message.media, MessageMediaDocument):
|
||||
# Document, video, audio, etc.
|
||||
info = await self._download_document(message, output_path, max_size)
|
||||
if info:
|
||||
media_info.append(info)
|
||||
elif isinstance(message.media, MessageMediaDocument):
|
||||
# Document, video, audio, etc.
|
||||
info = await asyncio.wait_for(
|
||||
self._download_document(message, output_path, max_size),
|
||||
timeout=timeout
|
||||
)
|
||||
if info:
|
||||
media_info.append(info)
|
||||
|
||||
# Note: Other media types (geo, poll, etc.) are not downloadable
|
||||
except asyncio.TimeoutError:
|
||||
logger.warning(f"Timeout downloading media for message {message.id}")
|
||||
# Add info about skipped file
|
||||
media_info.append({
|
||||
"filename": "timeout_skipped.bin",
|
||||
"type": "timeout",
|
||||
"caption": None,
|
||||
"size": 0,
|
||||
"path": None,
|
||||
"is_too_large": False,
|
||||
"timeout": True,
|
||||
})
|
||||
except Exception as e:
|
||||
logger.error(f"Error downloading media for message {message.id}: {e}")
|
||||
|
||||
return media_info
|
||||
|
||||
|
||||
Reference in New Issue
Block a user