[BUG] All searches return None - Library non-functional due to HLTB API changes and bot protection (probably)

[lihuelworks]

1st of all sorry for the trouble but I AI assisted this issue report to have as much data as possible for the author. I'm not a bot but I don't want YOU 🫵🏻 the author to have to ask menial questions like "have you updated the requirements", "do you have X library" and such.

Library completely broken - All searches return None due to HLTB API changes

Summary

The HowLongToBeat Python API library is completely non-functional as of November 2024. All search operations return None regardless of game name, similarity settings, or search parameters. The issue affects both synchronous and asynchronous search methods.

Environment

• Library version: 1.0.18 (latest from PyPI and master branch)
• Python version: 3.13.0 (also tested on 3.11)
• Operating System: Windows 10/11 (likely affects all platforms)
• Installation method: Both pip install howlongtobeatpy and from source
• fake_useragent version: 2.2.0

Problem Description

Observed Behavior

from howlongtobeatpy import HowLongToBeat

# ALL of these return None
results = HowLongToBeat().search("Celeste") # None
results = HowLongToBeat(0.0).search("Celeste") # None (no similarity filter)
results = HowLongToBeat().search("The Witcher 3") # None
results = HowLongToBeat().search("Halo Combat Evolved") # None

Expected Behavior

Should return a list of HowLongToBeatEntry objects with game data.

Root Cause Analysis

After extensive debugging, we've identified two critical issues:

Issue 1: Empty API Key Extraction

The library attempts to extract an API key from HLTB's JavaScript files, but the extraction now returns an empty string instead of None or a valid key.

Evidence:

from howlongtobeatpy.HTMLRequests import HTMLRequests

search_info = HTMLRequests.send_website_request_getcode(False)
print(f"API Key: '{search_info.api_key}'") # Prints: API Key: ''
print(f"Search URL: {search_info.search_url}") # Prints: Search URL: None

The regex patterns in __extract_api_from_script() no longer match HLTB's current JavaScript structure.

Issue 2: Bot Protection / Firewall Blocking

HLTB has implemented anti-bot measures (likely Cloudflare or similar) that block programmatic API requests with 403 Forbidden errors.

Direct API Test:

import requests
from fake_useragent import UserAgent

ua = UserAgent()
headers = {
 'content-type': 'application/json',
 'accept': '*/*',
 'User-Agent': ua.random.strip(),
 'referer': 'https://howlongtobeat.com/'
}

payload = {
 'searchType': "games",
 'searchTerms': ["Celeste"],
 'searchPage': 1,
 'size': 20,
 'searchOptions': {
 'games': {
 'userId': 0,
 'platform': "",
 'sortCategory': "popular",
 'rangeCategory': "main",
 'rangeTime': {'min': 0, 'max': 0},
 'gameplay': {'perspective': "", 'flow': "", 'genre': ""},
 'rangeYear': {'max': "", 'min': ""},
 'modifier': ""
 },
 'users': {'sortCategory': "postcount"},
 'lists': {'sortCategory': "follows"},
 'filter': "",
 'sort': 0,
 'randomizer': 0
 },
 'useCache': True
}

resp = requests.post('https://howlongtobeat.com/api/search', 
 headers=headers, 
 json=payload, 
 timeout=10)
print(resp.status_code) # 403
print(resp.text) # {"error":"Session expired"}

Different header combinations all result in 403:

• Minimal headers: 403 with HTML firewall page
• With Referer/Origin: 403 with HTML firewall page
• Full browser-like headers: 403 with JSON {"error":"Session expired"}

Investigation Details

JavaScript Analysis

We analyzed HLTB's current JavaScript bundles and found:

• The search endpoint is now simply /api/search (previously /api/s/)
• No API key suffix is required in the URL path
• The API key extraction patterns no longer match the JavaScript structure

Current JavaScript patterns found:

/api/search
/api/user
/api/logout
/api/search/init
/api/user/game_detail/
/api/game/
/api/error

Comparison with Other Libraries

We checked the Swift implementation (https://github.com/holgerkrupp/HowLongToBeat-PythonAPI) which was updated November 1, 2025:

• Uses the same API key extraction approach
• Likely has the same issues
• Their .httprequest test file shows outdated endpoint patterns

Test Results

All official library tests fail:

$ pytest howlongtobeatpy/tests/test_normal_request.py::TestNormalRequest::test_simple_game_name -v
FAILED - AssertionError: None == None : Search Results are None

Attempted Fixes

We attempted several fixes:

1. ✅ Updated SEARCH_URL from /api/s/ to /api/search
2. ✅ Removed API key dependency - made requests without API key suffix
3. ❌ Comprehensive browser headers - still blocked by firewall
4. ❌ Static user agents - still blocked
5. ❌ Session management - cookies not helping
6. ❌ Authorization headers - still blocked

Modified code (still fails due to bot protection):

# In HTMLRequests.py
SEARCH_URL = BASE_URL + "api/search" # Changed from "api/s/"

@staticmethod
def send_web_request(game_name: str, search_modifiers: SearchModifiers = SearchModifiers.NONE,
 page: int = 1):
 headers = HTMLRequests.get_search_request_headers()
 search_url = HTMLRequests.SEARCH_URL # No API key suffix needed
 payload = HTMLRequests.get_search_request_data(game_name, search_modifiers, page, None)
 resp = requests.post(search_url, headers=headers, data=payload, timeout=60)
 if resp.status_code == 200:
 return resp.text
 return None

Additional Context

Network/IP Considerations

The bot protection may be:

• IP-based (certain regions/IPs blocked)
• Rate-limit based
• Fingerprint-based (detecting automation)

Timeline

• Library worked fine until approximately October/November 2024
• HLTB website structure changed (new JavaScript bundles)
• Bot protection appears to have been added/strengthened

Reproduction Steps

1. Install the library: pip install howlongtobeatpy
2. Run any search:

   from howlongtobeatpy import HowLongToBeat
   results = HowLongToBeat().search("Celeste")
   print(results) # None

3. Observe that None is returned instead of game results

System Information

Python: 3.13.0
howlongtobeatpy: 1.0.18
fake-useragent: 2.2.0
requests: 2.32.3
aiohttp: 3.13.2
beautifulsoup4: 4.14.2

[GamerKingFaiz]

I believe something is just broken on HLTB's side. If you go to the actual website and try to do a search, it's not working on their end either.

The response is just {"error":"Session expired"}.

[GamerKingFaiz]

Website's back up. Now the endpoint requires an x-auth-token request header.

[ScrappyCocco]

It looks like for the x-auth-token it's calling
let e = await fetch("/api/search/init?t=".concat(Date.now()));
I'll try to implement that as soon as I have the time

[ScrappyCocco]

You can have a look at #53

[ScrappyCocco]

This release should work https://github.com/ScrappyCocco/HowLongToBeat-PythonAPI/releases/tag/1.0.19

[lihuelworks]

It does work! Thank you so much Cocco