fix(http): sanitize non-ASCII in URL before urllib.request to prevent latin-1 encode crash (fixes #817)

[23241a6749]

Fixes #817

Problem

Non-Latin-script topics (Arabic, CJK, Cyrillic, etc.) cause TikTok and Instagram ScrapeCreators requests to crash with:

UnicodeEncodeError: 'latin-1' codec can't encode character '\u0627' in position 28: ordinal not in range(256)

Fix

Added a urllib.parse.quote pass on the URL string after urlencode construction in http.request(), using a safe set that preserves all valid URL structural characters. This guarantees the URL is always 100% ASCII-safe before being passed to urllib.request.Request.

safe='/:@!$&\'()*+,;=-._~%?#[]=+'

Testing

• test_http_v3.py: 14/14 passed
• test_grounding_v3.py: 28/28 passed
• test_env_v3.py: 13/13 passed

[greptile-apps]

Greptile Summary

This PR makes HTTP request URLs ASCII-safe before they reach urllib.request. The main changes are:

• Adds urllib.parse.quote to the HTTP helper imports.
• Applies a final whole-URL quoting pass after query parameter construction.
• Preserves URL structural characters while encoding non-ASCII text.

Confidence Score: 5/5

This looks safe to merge.

• No blocking issues found in the changed code.

Important Files Changed

Filename	Overview
skills/last30days/scripts/lib/http.py	Adds final URL quoting in request() to prevent non-ASCII URLs from crashing urllib.request.

_{Reviews (1): Last reviewed commit: "fix(http): sanitize non-ASCII in URL bef..." | Re-trigger Greptile}