def test_oth3(self):
"""4 authors."""
i = (
'https://arstechnica.com/science/2007/09/'
'the-pseudoscience-behind-homeopathy/'
)
o = urls_scr(i)
ct = (
'* {{cite web '
'| last=Timmer '
'| first=John '
'| last2=Ford '
'| first2=Matt '
'| last3=Lee '
'| first3=Chris '
'| last4=Gitlin '
'| first4=Jonathan '
'| title=Diluting the scientific method: Ars looks at homeopathy '
'| website=Ars Technica '
'| date=2007-09-12 '
'| url=https://arstechnica.com/science/2007/09/'
'the-pseudoscience-behind-homeopathy/ '
'| access-date='
)
self.assertIn(ct, o[1])
def test_dm1(self):
"""4 authors"""
o = urls_scr(
'http://www.dailymail.co.uk/news/article-2633025/'
'London-cleric-convicted-NYC-terrorism-trial.html'
)
self.assertEqual(
'{{sfn | Malm | Witheridge | Drury | Bates | 2014}}', o[0]
)
self.assertIn(
'* {{cite web '
'| last=Malm '
'| first=Sara '
'| last2=Witheridge '
'| first2=Annette '
'| last3=Drury '
'| first3=Ian '
'| last4=Bates '
'| first4=Daniel '
'| title=Abu Hamza found guilty in US court of helping'
' Al-Qaeda terrorists '
'| website=Daily Mail Online '
'| date=2014-05-19 '
'| url=http://www.dailymail.co.uk/news/article-2633025/'
'London-cleric-convicted-NYC-terrorism-trial.html '
'| access-date=',
o[1],
)
def test_nyt2(self):
"""newstylct, 2 authors"""
ct = (
'* {{cite web '
'| last=Belson '
'| first=Ken '
'| last2=Sandomir '
'| first2=Richard '
'| title=$2 Billion for Clippers? In Time, '
'It May Be a Steal for Steve Ballmer '
'| website=The New York Times '
'| date=2014-05-30 '
'| url=https://www.nytimes.com/2014/05/31/sports/basketball/'
'steven-a-ballmers-2-billion-play-for-clippers-is-a-big-bet-on-'
'the-nba.html '
'| access-date='
)
self.assertIn(
ct,
urls_scr(
'https://www.nytimes.com/2014/05/31/sports/basketball/'
'steven-a-ballmers-2-billion-play-for-clippers-is-a-big-bet-'
'on-the-nba.html?hp'
)[1],
)
def test_citation_author_reverse_order(self):
"""Test correct detection of citation_author.
first name and last name are in reverse order.
"""
self.assertIn(
'* {{cite web '
'| last=Hartman '
'| first=JudithAnn R. '
'| last2=Nelson '
'| first2=Eric A. '
'| title=Automaticity in Computation and Student Success in '
'Introductory Physical Science Courses '
'| website=arXiv.org e-Print archive '
'| date=2016-08-17 '
'| url=https://arxiv.org/abs/1608.05006?utm_medium=email&'
'utm_source=other&utm_campaign=opencourse.GdeNrll1EeSROyIACtiVvg.'
'announcements%257Eopencourse.GdeNrll1EeSROyIACtiVvg.'
'4xDVKzx5EeeJjRJrkGD1dA '
'| access-date=',
urls_scr(
'https://arxiv.org/abs/1608.05006?utm_medium=email&utm_source='
'other&utm_campaign=opencourse.GdeNrll1EeSROyIACtiVvg.'
'announcements%257Eopencourse.GdeNrll1EeSROyIACtiVvg.'
'4xDVKzx5EeeJjRJrkGD1dA'
)[1],
)
def test_oth12(self):
"""Times of India, author could not be detected."""
i = ('http://timesofindia.indiatimes.com/city/pune/'
'UK-allows-working-visas-for-Indian-students/'
'articleshow/1163528927.cms?')
o = urls_scr(i)
sfn = "{{sfn | Kashyap | 2001}}"
self.assertIn(sfn, o[0])
def google_encrypted_scr(url, parsed_url, date_format):
if parsed_url[2][:7] in {'/books', '/books/'}:
# sample urls:
# https://encrypted.google.com/books?id=6upvonUt0O8C
# https://www.google.com/books?id=bwfoCAAAQBAJ&pg=PA32
# https://www.google.com/books/edition/_/bwfoCAAAQBAJ?gbpv=1&pg=PA32
return googlebooks_scr(parsed_url, date_format)
return urls_scr(url, date_format)
def test_single_line_meta_tags():
"""Issue #9."""
assert ("* {{cite web | last=Shoichet | first=Catherine E. "
"| title=Spill spews tons of coal ash into North Carolina's "
"Dan River | website=CNN | date=2014-02-09 "
"| url=http://www.cnn.com/2014/02/09/us/north-carolina-coal-ash"
"-spill/index.html | access-date=") in urls_scr(
'https://edition.cnn.com/'
'2014/02/09/us/north-carolina-coal-ash-spill/')[1]
def test_indaily(self):
self.assertIn(
"* {{cite web | last=Siebert | first=Bension "
"| title=Epidemics expert questions Marshall's schools advice "
"| website=InDaily | date=2020-03-19 "
"| url=https://indaily.com.au/news/2020/03/19/epidemics-expert-contradicts-marshalls-schools-advice/ "
"| access-date=",
urls_scr(
'https://indaily.com.au/news/2020/03/19/epidemics-expert-contradicts-marshalls-schools-advice/'
)[1])
def test_reverse_name():
"""Author is `Martin, Tracy`. Tracy should be the first name."""
assert ('* {{cite web '
'| last=Martin '
'| first=Tracy '
'| title=Dynamometers Explained '
'| website=HighBeam Research '
'| date=2014-07-01 '
'| url=http://www.highbeam.com/doc/1P3-3372742961.html '
'| access-date='
) in urls_scr('http://www.highbeam.com/doc/1P3-3372742961.html')[1]
def test_bbc5():
"""news.bbc.co.uk, 1 author"""
assert ("* {{cite web "
"| last=Madslien "
"| first=Jorn "
"| title=Inside the Bentley factory "
"| website=BBC NEWS "
"| date=2002-12-24 "
"| url=http://news.bbc.co.uk/2/hi/business/2570109.stm "
"| access-date="
) in urls_scr('http://news.bbc.co.uk/2/hi/business/2570109.stm')[1]
def test_abc_author():
assert ('* {{cite web | last=Ferguson | first=Kathleen '
'| title=Glow worms in Wollemi National Park survived Gospers '
'Mountain bushfire - ABC News '
'| website=ABC (Australian Broadcasting Corporation) '
'| date=2020-09-06 | url=https://www.abc.net.au/news/2020-09-06/'
'glow-worms-in-wollemi-national-park-survived-summer-bushfire/'
'12634762 | access-date=') in urls_scr(
'https://www.abc.net.au/news/2020-09-06/'
'glow-worms-in-wollemi-national-park-survived-summer-bushfire/'
'12634762')[1]
def test_bbc2():
"""1 author"""
assert ('* {{cite web '
'| last=Gage '
'| first=Suzi '
'| title=Sea otter return boosts ailing seagrass in California '
'| website=BBC News '
'| date=2013-08-26 '
'| url=http://www.bbc.com/news/science-environment-23814524 '
'| access-date=') in urls_scr(
'http://www.bbc.com/news/science-environment-23814524')[1]
def test_bbc1(self):
"""no authors"""
i = 'https://www.bbc.com/news/world-asia-27653361'
o = urls_scr(i)
ct = ("* {{cite web "
"| title=US 'received Qatar assurances' on Afghan prisoner deal "
"| website=BBC News "
"| date=2014-06-01 "
"| url=http://www.bbc.com/news/world-asia-27653361 "
"| ref={{sfnref | BBC News | 2014}} "
"| access-date=")
self.assertIn(ct, o[1])
def test_oth7():
"""Contains a By Topic line and also the byline contains ' | '."""
assert ('* {{cite web '
'| last=Chandler '
'| first=David L. '
'| title=Traffic lights: There’s a better way '
'| website=MIT News '
'| date=2014-07-07 '
'| url=http://news.mit.edu/2014/'
'traffic-lights-theres-a-better-way-0707 '
'| access-date=') in urls_scr(
'http://news.mit.edu/2014/'
'traffic-lights-theres-a-better-way-0707')[1]
def test_bbc4():
"""news.bbc.co.uk, 1 author"""
assert (
"* {{cite web "
"| last=Jones "
"| first=Meirion "
"| title=Malaria advice 'risks lives' "
"| website=BBC NEWS "
"| date=2006-07-13 "
"| url="
"http://news.bbc.co.uk/2/hi/programmes/newsnight/5178122.stm "
"| access-date=") in urls_scr(
'http://news.bbc.co.uk/2/hi/programmes/newsnight/5178122.stm')[1]
def test_empty_meta_author_content():
"""Test that the output will not be malformed because empty meta."""
assert (
"* {{cite web "
"| title=UAE's Enoc pays Iran $4 billion in oil dues "
"| website=Al Jazeera "
"| date=2017-05-29 "
"| url=http://www.aljazeera.com/news/2017/05/uae-enoc-pays-iran-4-"
"billion-oil-dues-170529171315570.html "
"| ref={{sfnref | Al Jazeera | 2017}} "
"| access-date=") in urls_scr(
'http://www.aljazeera.com/news/2017/05/'
'uae-enoc-pays-iran-4-billion-oil-dues-170529171315570.html')[1]
def test_bbc3(self):
"""https version of bbc2 (differs a lot!)"""
i = 'https://www.bbc.com/news/science-environment-23814524'
o = urls_scr(i)
ct = ('* {{cite web '
'| last=Gage '
'| first=Suzi '
'| title=Sea otter return boosts ailing seagrass in California '
'| website=BBC News '
'| date=2013-08-26 '
'| url=http://www.bbc.com/news/science-environment-23814524 '
'| access-date=')
self.assertIn(ct, o[1])
def test_bbc6():
"""bbc.com, 1 author"""
i = 'http://www.bbc.com/news/science-environment-26267918'
o = urls_scr(i)
ct = ("* {{cite web "
"| last=Amos "
"| first=Jonathan "
"| title=European Space Agency picks Plato planet-hunting mission "
"| website=BBC News "
"| date=2014-02-20 "
"| url=http://www.bbc.com/news/science-environment-26267918 "
"| access-date=")
assert ct in o[1]
def test_language_not_de_csbc():
assert (
"{{cite web "
"| last=Martin "
"| first=Emmie "
"| title=In San Francisco, households earning $117,000 qualify as ‘low income’ "
"| website=CNBC "
"| date=2018-06-28 "
"| url=https://www.cnbc.com/2018/06/28/families-earning-117000-qualify-as-low-income-in-san-francisco.html "
"| access-date="
) == urls_scr(
'https://www.cnbc.com/2018/06/28/families-earning-117000-qualify-as-low-income-in-san-francisco.html'
)[1][2:-12]
def test_oth5():
"""Getting the date is tricky here."""
o = urls_scr('http://www.magiran.com/npview.asp?ID=1410487')
assert '{{sfn | نوري | 2007}}' in o[0]
assert ('* {{cite web '
'| last=نوري '
'| first=آزاده شهمير '
'| title=روزنامه سرمايه86/3/1: دكتر طاهر صباحي، محقق و مجموعه دار'
' فرش: بازار جهاني با توليد فرش هنري نصيب ايران مي شود '
'| website=magiran.com '
'| date=2007-05-22 '
'| url=http://www.magiran.com/npview.asp?ID=1410487 '
'| language=fa '
'| access-date=') in o[1]
def test_oth6():
"""Detection of website name."""
o = urls_scr('http://www.farsnews.com/newstext.php?nn=13930418000036')
assert "{{sfn | ''خبرگزاری فارس'' | 2014}}" in o[0]
# Fars news is using 'خبرگزاری فارس' as og:author which is wrong
# and thats why its name is not italicized in sfn.
assert ('* {{cite web '
'| title=آیت\u200cالله محمدی گیلانی دارفانی را وداع گفت '
'| website=خبرگزاری فارس '
'| date=2014-07-09 '
'| url=http://www.farsnews.com/newstext.php?nn=13930418000036 '
'| language=fa '
'| ref={{sfnref | خبرگزاری فارس | 2014}} '
'| access-date=') in o[1]
def test_invalid_name():
"""Test that URL does not fail with InvalidNameError."""
assert ('* {{cite web | title=انتخابات 96 به روایت آمار '
'| website=پایگاه اطلاع رسانی شبکه خبر صدا'
' و سیمای جمهوری اسلامی ایران |'
' date=2017-05-24 | url=http://www.irinn.ir/fa/news/499654 '
'| language=fa | ref={{sfnref |'
' پایگاه اطلاع رسانی شبکه خبر'
' صدا و سیمای جمهوری اسلامی ایران | 2017}} |'
' access-date=') in urls_scr(
'http://www.irinn.ir/fa/news/499654/'
'%D8%A7%D9%86%D8%AA%D8%AE%D8%A7%D8%A8%D8%A7%D8%AA-96-'
'%D8%A8%D9%87-%D8%B1%D9%88%D8%A7%DB%8C%D8%AA-'
'%D8%A2%D9%85%D8%A7%D8%B1')[1]
def test_oth14(self):
"""thebulletin.org"""
i = ('http://www.independent.co.uk/news/business/'
'the-investment-column-tt-group-1103208.html')
o = urls_scr(i)
ct = ('* {{cite web '
'| title=The Investment column: TT Group '
'| website=The Independent '
'| date=1999-06-29 '
'| url=http://www.independent.co.uk/news/business/'
'the-investment-column-tt-group-1103208.html '
'| ref={{sfnref | The Independent | 1999}} '
'| access-date=')
self.assertIn(ct, o[1])
def test_oth12(self):
"""thebulletin.org"""
i = 'http://thebulletin.org/evidence-shows-iron-dome-not-working7318'
o = urls_scr(i)
ct = ('* {{cite web '
'| last=Postol '
'| first=Theodore A. '
'| title=The evidence that shows Iron Dome is not working '
'| website=Bulletin of the Atomic Scientists '
'| date=2014-07-19 '
'| url=http://thebulletin.org/'
'evidence-shows-iron-dome-not-working7318 '
'| access-date=')
self.assertIn(ct, o[1])
def test_tgd1(self):
"""ABCNews. Wrong author: | last=News | first=ABC."""
i = 'http://abcnews.go.com/blogs/headlines/2006/12/saddam_executed/'
o = urls_scr(i)
ct = ('* {{cite web '
'| last=Ross '
'| first=Brian '
'| title=Saddam Executed; An Era Comes to an End '
'| website=ABC News Blogs '
'| date=2006-12-30 '
'| url=http://abcnews.go.com/blogs/headlines/2006/12/'
'saddam_executed/ '
'| access-date=')
self.assertIn(ct, o[1])
def test_tgd3(self):
""""Staff" in author name."""
i = ('http://www.tgdaily.com/space-features/'
'82906-sma-reveals-giant-star-cluster-in-the-making')
o = urls_scr(i)
ct = ('* {{cite web '
'| title=SMA reveals giant star cluster in the making '
'| website=TG Daily '
'| date=2013-12-17 '
'| url=http://www.tgdaily.com/space-features/'
'82906-sma-reveals-giant-star-cluster-in-the-making '
'| ref={{sfnref | TG Daily | 2013}} '
'| access-date=')
self.assertIn(ct, o[1])
def test_tgd2(self):
"""Hard to find author and date."""
i = ('http://www.tgdaily.com/web/'
'100381-apple-might-buy-beats-for-32-billion')
o = urls_scr(i)
ct = ('* {{cite web '
'| title=Apple might buy Beats for $3.2 billion '
'| website=TG Daily '
'| date=2014-05-09 '
'| url=http://www.tgdaily.com/web/'
'100381-apple-might-buy-beats-for-32-billion '
'| ref={{sfnref | TG Daily | 2014}} '
'| access-date=')
self.assertIn(ct, o[1])
def test_nyt5(self):
"""special case for date format (not in usual meta tags)"""
i = ('https://www.nytimes.com/2007/06/13/world/americas/'
'13iht-whale.1.6123654.html')
o = urls_scr(i)
ct = ('* {{cite web '
'| title=19th-century harpoon gives clue on whales '
'| website=The New York Times '
'| date=2007-06-13 '
'| url=https://www.nytimes.com/2007/06/13/world/americas/'
'13iht-whale.1.6123654.html '
'| ref={{sfnref | The New York Times | 2007}} '
'| access-date=')
self.assertIn(ct, o[1])
def test_nyt3(self):
"""oldstylct, 1 author"""
i = 'http://www.nytimes.com/2007/12/25/world/africa/25kenya.html'
o = urls_scr(i)
ct = ('* {{cite web '
'| last=Gettleman '
'| first=Jeffrey '
'| title=Election Rules Complicate Kenya Race '
'| website=The New York Times '
'| date=2007-12-25 '
'| url=https://www.nytimes.com/2007/12/25/world/africa/'
'25kenya.html '
'| access-date=')
self.assertIn(ct, o[1])
def test_oth11(self):
"""Business News Daily."""
i = ('http://www.businessnewsdaily.com/6762-male-female-entrepreneurs'
'.html?cmpid=514642_20140715_27858876')
o = urls_scr(i)
ct = ('* {{cite web '
'| last=Helmrich '
'| first=Brittney '
'| title=Male vs. Female Entrepreneurs: How Are They Different? '
'| website=Business News Daily '
'| date=2014-07-10 '
'| url=http://www.businessnewsdaily.com/6762-male-female-'
'entrepreneurs.html '
'| access-date=')
self.assertIn(ct, o[1])