mirror of
https://github.com/assafelovic/gpt-researcher.git
synced 2026-09-14 20:17:32 +08:00
Assaf Elovic
f29f3e182e
Consolidates 10 single-file PRs from @Bartok9. Scrapers assumed a live session and well-formed responses: a None session raised instead of falling back, None content hit len() before any check, TavilyExtract indexed keys the API omits on partial results, and image/dimension parsing assumed str where BeautifulSoup can return a list or None. Squashed from #1915 #1916 #1939 #1948 #1949 #1965 #1990 #1991 #2004 #2005. Two of the tests needed repair before they could be kept. Both stubbed bs4 via setdefault() and then unconditionally assigned sys.modules['bs4'].BeautifulSoup = MagicMock -- setdefault returns the real module when bs4 is already imported, so the real BeautifulSoup became a MagicMock process-wide and test_scraper_extract_title failed once both were collected together. They now stub only when bs4 is genuinely absent. Several also left gpt_researcher stubs in sys.modules permanently, breaking collection for every module after them; those now snapshot and restore. Co-Authored-By: Bartok9 <noreply@github.com> Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
26 lines
710 B
Python
26 lines
710 B
Python
"""parse_dimension must tolerate None/non-str width/height attrs."""
|
|
from __future__ import annotations
|
|
|
|
import importlib.util
|
|
from pathlib import Path
|
|
|
|
|
|
def _load():
|
|
path = Path(__file__).resolve().parents[1] / "gpt_researcher" / "scraper" / "utils.py"
|
|
spec = importlib.util.spec_from_file_location("scraper_utils_ut", path)
|
|
mod = importlib.util.module_from_spec(spec)
|
|
spec.loader.exec_module(mod)
|
|
return mod
|
|
|
|
|
|
def test_parse_dimension_none():
|
|
u = _load()
|
|
assert u.parse_dimension(None) is None
|
|
|
|
|
|
def test_parse_dimension_px_and_int():
|
|
u = _load()
|
|
assert u.parse_dimension("100px") == 100
|
|
assert u.parse_dimension(80) == 80
|
|
assert u.parse_dimension("auto") is None
|