Fix TikTok extraction with proxies

[karilaa-dev]

User description

Summary

• Fix TikTok video extraction failing when proxies are configured
• Use direct connection for metadata extraction (with browser impersonation)
• Use proxies for media downloads to hide server IP

Test plan

• Tested video extraction with proxies enabled
• Verified proxy is used for downloads
• Confirmed extraction bypasses proxy to use impersonate feature

---

PR Type

Bug fix, Documentation

---

Description

• Fix TikTok extraction with proxies using direct connection for metadata

• Simplify proxy handling by temporarily disabling it for extraction

• Update CODEBASE_MAP.md with new configuration details

• Document hardcoded performance values and removed env vars

---

Diagram Walkthrough

flowchart LR
  A[Proxy Configured] --> B[Disable Proxy Temporarily]
  B --> C[Extract Metadata with Impersonate]
  C --> D[Restore Proxy for Downloads]
  D --> E[Complete Extraction]

Loading

File Walkthrough

	Relevant files
Bug fix	client.py Fix TikTok extraction with proxies                                              tiktok_api/client.py Simplify proxy handling by temporarily disabling it for metadata extraction Use ie._extract_web_data_and_status() for both proxy and non-proxy cases Remove complex manual extraction logic for proxy scenario Update download context to use saved proxy for downloads +25/-74
Documentation	CODEBASE_MAP.md Update codebase documentation                                                        docs/CODEBASE_MAP.md Update last_mapped date and token counts Add Telegram API credentials documentation Update performance configuration section Add notes about removed fields and hardcoded values +20/-11

---

[zam-review]

PR Description updated to latest commit (8745d5a)

[karilaa-dev]

/review

[zam-review]

PR Reviewer Guide 🔍

(Review updated until commit 14c1abe)

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 3 🔵🔵🔵⚪⚪

🧪 No relevant tests

🔒 No security concerns identified

⚡ Recommended focus areas for review Resource Leak Potential If yt_dlp.YoutubeDL(ydl_opts) fails on line 864, the old_ydl instance will never be closed, potentially leaking resources. Consider wrapping this in a try-finally block or using a context manager pattern to ensure cleanup even if initialization fails. old_ydl = ydl ydl = yt_dlp.YoutubeDL(ydl_opts) old_ydl.close() # Close old instance after new one is ready Error Handling Change The code now returns None, "extraction", None instead of raising TikTokExtractionError when video data extraction fails (lines 888, 906). This changes the error handling contract and may confuse callers expecting exceptions. Verify that all calling code handles this return value correctly. # 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 Hardcoded Configuration Values The documentation indicates that thread pool (500 workers) and curl_cffi connections (10,000) are now hardcoded for maximum throughput, removing previous environment variables (THREAD_POOL_SIZE, MAX_USER_QUEUE_SIZE, MAX_CONCURRENT_IMAGES). This reduces deployment flexibility and should be validated against different deployment scenarios. | `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.

[karilaa-dev]

/review

[zam-review]

Persistent review updated to latest commit 248c050

[karilaa-dev]

/review

[zam-review]

Persistent review updated to latest commit 14c1abe