def test_make_doc1_url(self):
"""Can we make good doc1 urls?"""
qa_pairs = (
(('cand', '01712427473', False),
'https://ecf.cand.uscourts.gov/doc1/01712427473'),
(('cand', '01702427473', False),
'https://ecf.cand.uscourts.gov/doc1/01702427473'),
(('cand', '01712427473', True),
'https://ecf.cand.uscourts.gov/doc1/01712427473'),
(('cand', '01702427473', True),
'https://ecf.cand.uscourts.gov/doc1/01712427473'),
)
for q, a in qa_pairs:
self.assertEqual(make_doc1_url(*q), a)
def test_make_doc1_url(self):
"""Can we make good doc1 urls?"""
qa_pairs = (
(('cand', '01712427473', False),
'https://ecf.cand.uscourts.gov/doc1/01712427473'),
(('cand', '01702427473', False),
'https://ecf.cand.uscourts.gov/doc1/01702427473'),
(('cand', '01712427473', True),
'https://ecf.cand.uscourts.gov/doc1/01712427473'),
(('cand', '01702427473', True),
'https://ecf.cand.uscourts.gov/doc1/01712427473'),
)
for q, a in qa_pairs:
self.assertEqual(make_doc1_url(*q), a)
def test_auto_login(self, mock_login):
"""Do we automatically log in if needed?"""
court_id = 'ksd'
pacer_doc_id = '07902639735'
url = make_doc1_url(court_id, pacer_doc_id, True)
pacer_case_id = '81531'
# This triggers and auto-login because we aren't logged in yet.
self.session.username = PACER_USERNAME
self.session.password = PACER_PASSWORD
_ = self.session.get(url, params={
'case_id': pacer_case_id,
'got_receipt': '1',
}, allow_redirects=True)
self.assertTrue(mock_login.called,
'PacerSession.login() should be called.')
def test_auto_login(self, mock_login):
"""Do we automatically log in if needed?"""
court_id = 'ksd'
pacer_doc_id = '07902639735'
url = make_doc1_url(court_id, pacer_doc_id, True)
pacer_case_id = '81531'
# This triggers and auto-login because we aren't logged in yet.
self.session.username = PACER_USERNAME
self.session.password = PACER_PASSWORD
_ = self.session.get(url, params={
'case_id': pacer_case_id,
'got_receipt': '1',
}, allow_redirects=True)
self.assertTrue(mock_login.called,
'PacerSession.login() should be called.')
def download_pdf(self, pacer_case_id, pacer_document_number):
"""Download a PDF from PACER.
Note that this doesn't support attachments yet.
"""
timeout = (60, 300)
url = make_doc1_url(self.court_id, pacer_document_number, True)
data = {
'caseid': pacer_case_id,
'got_receipt': '1',
}
logger.info("GETting PDF at URL: %s with params: %s" % (url, data))
r = self.session.get(url, params=data, timeout=timeout)
# The request above sometimes generates an HTML page with an iframe
# containing the PDF, and other times returns the PDF. Our task is thus
# to either get the src of the iframe and download the PDF or just
# return the pdf.
r.raise_for_status()
if is_pdf(r):
logger.info('Got PDF binary data for case %s at: %s' % (url, data))
return r
text = clean_html(r.text)
tree = get_html_parsed_text(text)
tree.rewrite_links(fix_links_in_lxml_tree, base_href=r.url)
try:
iframe_src = tree.xpath('//iframe/@src')[0]
except IndexError:
if 'pdf:Producer' in text:
logger.error("Unable to download PDF. PDF content was placed "
"directly in HTML. URL: %s, caseid: %s" %
(url, pacer_case_id))
else:
logger.error(
"Unable to download PDF. PDF not served as binary "
"data and unable to find iframe src attribute. "
"URL: %s, caseid: %s" % (url, pacer_case_id))
return None
r = self.session.get(iframe_src, timeout=timeout)
if is_pdf(r):
logger.info('Got iframed PDF data for case %s at: %s' %
(url, iframe_src))
return r
def test_auto_login(self, mock_login):
"""Do we automatically log in if needed?"""
court_id = "ksd"
pacer_doc_id = "07902639735"
url = make_doc1_url(court_id, pacer_doc_id, True)
pacer_case_id = "81531"
# This triggers and auto-login because we aren't logged in yet.
self.session.username = PACER_USERNAME
self.session.password = PACER_PASSWORD
_ = self.session.get(
url,
params={
"case_id": pacer_case_id,
"got_receipt": "1",
},
allow_redirects=True,
)
self.assertTrue(mock_login.called,
"PacerSession.login() should be called.")
