def parse_page(page, url, dump=False):
"""
Parse the page using all parsing methods available.
:param str page: the content of the fatality page
:param str url: detail page URL
:return: a dictionary representing a fatality.
:rtype: dict
"""
report = model.Report(case='19-123456')
# Parse the twitter fields.
twitter_report, twitter_err = twitter.parse(page)
report.update(twitter_report)
# Parse the page.
article_report, artricle_err = article.parse_content(page)
report.update(article_report)
if twitter_err or artricle_err: # pragma: no cover
twitter_err_str = f'\nTwitter fields:\n\t * ' + "\n\t * ".join(
twitter_err) if twitter_err else ''
article_err_str = f'\nArticle fields:\n\t * ' + "\n\t * ".join(
artricle_err) if artricle_err else ''
logger.debug(
f'Errors while parsing {url}:{twitter_err_str}{article_err_str}')
# Dump the file.
if dump:
dumpr_dir = Path(constant.DUMP_DIR)
dumpr_dir.mkdir(parents=True, exist_ok=True)
dump_file_name = url.split('/')[-1]
dump_file = dumpr_dir / dump_file_name
dump_file.write_text(page)
return report
def test_parse_page_content_02(self, page, expected, errors):
"""Ensure special cases are handled."""
actual, err = article.parse_content(page)
if errors or err:
assert errors == len(err)
else:
assert actual == expected
def test_parse_page_content_00(self, page, expected, errors):
"""Ensure location information is properly extracted from the page."""
p = load_test_page(page)
actual, err = article.parse_content(p)
if errors or err:
assert errors == len(err)
else:
assert actual == expected
def test_dumped_page(page_dump):
"""
Helper test to allow debugging offline.
Run the following command: `pytest -s -n0 -x -vvv -m dump`
"""
try:
page = load_dumped_page(page_dump)
except FileNotFoundError:
raise FileNotFoundError(
f'Dump file "{page_dump}" not found: run "scrapd --dump" first.')
else:
twitter_report, twitter_err = twitter.parse(page)
assert not twitter_err
article_report, artricle_err = article.parse_content(page)
assert not artricle_err
def test_parse_page_content_03(self, page, expected, errors):
"""Ensure special cases are handled."""
with pytest.raises(ValueError, match='a date is mandatory'):
actual, err = article.parse_content(page)
def test_parse_page_content_01(self):
"""Ensure a missing case number raises an exception."""
with pytest.raises(ValueError):
article.parse_content('There is no case number here.')