Skip to content

Fix TikTok extraction with proxies #60

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
karilaa-dev merged 7 commits into from
Jan 16, 2026
Merged

Fix TikTok extraction with proxies #60

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
e195d38
Remove performance config, hardcode maximum throughput values
karilaa-dev Jan 15, 2026
7295e49
Change positions of values in .env.example
karilaa-dev Jan 15, 2026
8d109a8
Add Telegram API credentials to .env.example
karilaa-dev Jan 15, 2026
acff6be
Update CODEBASE_MAP.md
karilaa-dev Jan 16, 2026
8745d5a
Fix TikTok extraction with proxies by using direct connection for met…
karilaa-dev Jan 16, 2026
248c050
Improve error handling in ydl recreation for proxy extraction
karilaa-dev Jan 16, 2026
14c1abe
Add video_data validation after TikTok extraction
karilaa-dev Jan 16, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
31 changes: 20 additions & 11 deletions docs/CODEBASE_MAP.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,12 +1,12 @@
---
last_mapped: 2026-01-14T22:45:00Z
last_mapped: 2026-01-15T12:00:00Z
total_files: 61
total_tokens: 71358
total_tokens: 69443
---

# Codebase Map

> Auto-generated by Cartographer. Last mapped: 2026-01-14
> Auto-generated by Cartographer. Last mapped: 2026-01-15

## System Overview

Expand Down Expand Up @@ -168,11 +168,13 @@ tt-bot/

| File | Purpose | Tokens |
|------|---------|--------|
| client.py | Main TikTokClient + ProxySession + 3-part retry | 18,676 |
| client.py | Main TikTokClient + ProxySession + 3-part retry | 16,499 |
| proxy_manager.py | Thread-safe round-robin proxy rotation | 1,303 |
| models.py | VideoInfo, MusicInfo dataclasses | 1,091 |
| models.py | VideoInfo, MusicInfo dataclasses | 1,079 |
| exceptions.py | Exception hierarchy (9 error types) | 233 |

**Note:** `VideoInfo.author` field was removed (unused in codebase).

**Key Classes:**
- `TikTokClient`: Main extraction client with integrated retry
- `ProxySession`: Manages proxy state per request flow (sticky until retry)
Expand Down Expand Up @@ -263,8 +265,8 @@ tt-bot/

| File | Purpose | Tokens |
|------|---------|--------|
| video_types.py | Video/image sending, slideshow retry, HEIC conversion | 6,374 |
| queue_manager.py | Per-user concurrency limits | 1,021 |
| video_types.py | Video/image sending, slideshow retry, HEIC conversion | 6,322 |
| queue_manager.py | Per-user concurrency limits | 1,029 |
| utils.py | Helpers (lang resolution, user registration) | 692 |

**Key Functions:**
Expand All @@ -273,6 +275,10 @@ tt-bot/
- `send_music_result()`: Send audio with cover
- `QueueManager.info_queue()`: Acquire/release queue slot

**Note:** Thumbnail download thresholds:
- Inline messages: >30s (lowered from 60s)
- Regular messages: >60s

---

### Stats Module (`stats/`)
Expand Down Expand Up @@ -516,6 +522,8 @@ sequenceDiagram
| `BOT_TOKEN` | Main bot token |
| `DB_URL` | PostgreSQL connection string |
| `TG_SERVER` | Telegram API server URL |
| `TELEGRAM_API_ID` | Telegram API ID (for custom Bot API server) |
| `TELEGRAM_API_HASH` | Telegram API hash (for custom Bot API server) |

### Retry Configuration (NEW)
| Variable | Default | Description |
Expand All @@ -527,10 +535,11 @@ sequenceDiagram
### Performance
| Variable | Default | Description |
|----------|---------|-------------|
| `THREAD_POOL_SIZE` | 128 | ThreadPoolExecutor workers |
| `MAX_USER_QUEUE_SIZE` | 3 | Max concurrent per user |
| `MAX_CONCURRENT_IMAGES` | 20 | Max parallel image downloads |
| `MAX_VIDEO_DURATION` | 1800 | Max video duration (seconds, 0=unlimited) |
| `MAX_USER_QUEUE_SIZE` | 0 | Max concurrent per user (0=unlimited) |
| `MAX_VIDEO_DURATION` | 0 | Max video duration (seconds, 0=unlimited) |
| `STREAMING_DURATION_THRESHOLD` | 300 | Stream videos longer than this (seconds) |
| `LOG_LEVEL` | INFO | Logging level |

**Note:** Thread pool (500 workers) and curl_cffi connections (10,000) are hardcoded for maximum throughput.

See `.env.example` for complete list.
108 changes: 36 additions & 72 deletions tiktok_api/client.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -848,76 +848,28 @@ def _extract_with_context_sync(
try:
# Use yt-dlp's internal method to get raw webpage data
# This also sets up all necessary cookies
# NOTE: When using a proxy, yt-dlp's impersonate=True feature
# doesn't work correctly. We need to download without impersonate.
# NOTE: TikTok's impersonate feature doesn't work through HTTP proxies.
# Always use direct connection for extraction, proxy is used for downloads.
saved_proxy = None # Will store proxy for download context
if self.proxy_manager and self.proxy_manager.has_proxies():
# Download webpage without impersonate to avoid proxy issues
res = ie._download_webpage_handle(
normalized_url, video_id, fatal=False, impersonate=False
# Download webpage without proxy but with impersonate
# Save current proxy setting and temporarily disable it
saved_proxy = ydl_opts.get("proxy")
if "proxy" in ydl_opts:
del ydl_opts["proxy"]
# Recreate YDL without proxy for extraction
# Create new instance first to ensure we have a valid ydl
# even if something goes wrong during recreation
old_ydl = ydl
ydl = yt_dlp.YoutubeDL(ydl_opts)
old_ydl.close() # Close old instance after new one is ready
ie = ydl.get_info_extractor("TikTok")
ie.set_downloader(ydl)

# Use standard extraction with impersonate (no proxy)
video_data, status = ie._extract_web_data_and_status(
normalized_url, video_id
)
if res is False:
raise TikTokExtractionError(
f"Failed to download webpage for video {video_id}"
)

webpage, urlh = res

# Check for login redirect
import urllib.parse

if urllib.parse.urlparse(urlh.url).path == "/login":
raise TikTokExtractionError(
"TikTok is requiring login for access to this content"
)

# Extract data manually using yt-dlp's helper methods
video_data = None
status = -1

# Try universal data first
if universal_data := ie._get_universal_data(webpage, video_id):
from yt_dlp.utils import traverse_obj

status = (
traverse_obj(
universal_data,
("webapp.video-detail", "statusCode", {int}),
)
or 0
)
video_data = traverse_obj(
universal_data,
("webapp.video-detail", "itemInfo", "itemStruct", {dict}),
)

# Try sigi state data
elif sigi_data := ie._get_sigi_state(webpage, video_id):
from yt_dlp.utils import traverse_obj

status = (
traverse_obj(sigi_data, ("VideoPage", "statusCode", {int}))
or 0
)
video_data = traverse_obj(
sigi_data, ("ItemModule", video_id, {dict})
)

# Try next.js data
elif next_data := ie._search_nextjs_data(
webpage, video_id, default={}
):
from yt_dlp.utils import traverse_obj

status = (
traverse_obj(
next_data, ("props", "pageProps", "statusCode", {int})
)
or 0
)
video_data = traverse_obj(
next_data,
("props", "pageProps", "itemInfo", "itemStruct", {dict}),
)

# Check TikTok status codes for errors
# 10204 = Video not found / deleted
Expand All @@ -930,10 +882,10 @@ def _extract_with_context_sync(
elif status == 10216:
return None, "deleted", None # Treat under review as deleted

# Validate that we got video data
if not video_data:
raise TikTokExtractionError(
f"Unable to extract webpage video data (status: {status})"
)
logger.error(f"No video data returned for {video_id} (status={status})")
return None, "extraction", None
else:
# No proxy, use the standard method with impersonate
video_data, status = ie._extract_web_data_and_status(
Expand All @@ -947,6 +899,11 @@ def _extract_with_context_sync(
return None, "private", None
elif status == 10216:
return None, "deleted", None # Treat under review as deleted

# Validate that we got video data
if not video_data:
logger.error(f"No video data returned for {video_id} (status={status})")
return None, "extraction", None
except AttributeError as e:
logger.error(
f"Failed to call yt-dlp internal method: {e}. "
Expand All @@ -958,11 +915,18 @@ def _extract_with_context_sync(
) from e

# Create download context with the live instances
# For proxy path, use the saved_proxy (extraction was without proxy, downloads use proxy)
# For non-proxy path, use request_proxy as before
context_proxy = (
saved_proxy
if self.proxy_manager and self.proxy_manager.has_proxies()
else request_proxy
)
download_context = {
"ydl": ydl,
"ie": ie,
"referer_url": url,
"proxy": request_proxy, # Store proxy for per-request assignment
"proxy": context_proxy, # Store proxy for per-request assignment
}

# Success - transfer ownership of ydl to caller via download_context
Expand Down