fix: parse_resource still corrupts every https:// resource - off-by-one from #1325 never fixed

[ionfwsrijan]

Summary

parse_resource in src/utils/url_validator.py still corrupts every labeled resource URL — the exact bug reported in #1325 (closed) was never actually fixed. The split marker ": http" is a prefix of ": https://", so splitting on it chops http off every https:// resource, leaving a broken s://... URL.

Evidence

src/utils/url_validator.py:69-75:

split_marker = ": http"
idx = raw.find(split_marker)
if idx != -1:
    label = raw[:idx].strip()
    url   = raw[idx + len(split_marker):].strip()

Verified by execution:

parse_resource("Python official docs: https://docs.python.org")
  -> {"label": "Python official docs", "url": "s://docs.python.org"}   # corrupt
parse_resource("MDN Docs: https://developer.mozilla.org/")
  -> url = "s://developer.mozilla.org/"                                 # corrupt

data/projects.json contains ~100 labeled resources in the "Label: https://..." form, so all of them are corrupted:

• /api/project/<id>/resources (main_routes.py:315 via validate_resources) flags them all as invalid.
• data_loader.validate_projects logs spurious warnings for every one.
• tests/test_url_validator.py:123-127, 138-143, 202-206 assert the correct output and therefore fail against the current code.

Impact

• Every learning resource link rendered by the project detail page is a broken s://... URL.
• The shipped test suite for url_validator is failing.
• fix : add off-by-one fix in url_validator.py parse_resource #1325 ("add off-by-one fix in url_validator.py parse_resource") was closed, but the fix is not present — this is a regression/open-bug against that fix.

Suggested Fix

Split on the scheme token properly, e.g. find ": " and slice the URL starting at "http" (url = raw[idx+2:].lstrip() or regex r"^(.+?):\s*(https?://.*)$"), and add a regression test using a https:// label.

Related minor flaw in the same function: the marker is case-sensitive, so "Docs: HTTPS://example.com" isn't split at all and is treated as a bare URL.