def test_pagerank_calculation(self):
"""Create a few Documents and fake citation relation among them, then run the pagerank
algorithm. Check whether this simple case can get the correct result.
"""
# Set up some handy variables
self.court = Court.objects.get(pk='test')
#create 3 documents with their citations
c1, c2, c3 = Citation(case_name=u"c1"), Citation(case_name=u"c2"), Citation(case_name=u"c3")
c1.save(index=False)
c2.save(index=False)
c3.save(index=False)
d1, d2, d3 = Document(date_filed=date.today()), Document(date_filed=date.today()), Document(date_filed=date.today())
d1.citation, d2.citation, d3.citation = c1, c2, c3
doc_list = [d1, d2, d3]
for d in doc_list:
d.court = self.court
d.citation.save(index=False)
d.save(index=False)
#create simple citing relation: 1 cites 2 and 3; 2 cites 3; 3 cites 1;
d1.cases_cited.add(d2.citation)
d2.citation_count += 1
d2.cases_cited.add(d3.citation)
d3.citation_count += 1
d3.cases_cited.add(d1.citation)
d1.citation_count += 1
d1.cases_cited.add(d3.citation)
d3.citation_count += 1
d1.save(index=False)
d2.save(index=False)
d3.save(index=False)
#calculate pagerank of these 3 document
comm = Command()
self.verbosity = 1
comm.do_pagerank(chown=False)
# read in the pagerank file, converting to a dict
pr_values_from_file = {}
with open(get_data_dir_location() + "external_pagerank") as f:
for line in f:
pk, value = line.split('=')
pr_values_from_file[pk] = float(value.strip())
# Verify that whether the answer is correct, based on calculations in Gephi
answers = {
'1': 0.387790,
'2': 0.214811,
'3': 0.397400,
}
for key, value in answers.iteritems():
self.assertTrue(
(abs(pr_values_from_file[key]) - value) < 0.0001,
msg="The answer for item %s was %s when it should have been %s" % (key, answers['1'], pr_values_from_file['1'])
)
def import_law_box_case(case_path):
"""Open the file, get its contents, convert to XML and extract the meta data.
Return a document object for saving in the database
"""
raw_text = open(case_path).read()
clean_html_tree, complete_html_tree, clean_html_str, body_text = get_html_from_raw_text(
raw_text)
sha1 = hashlib.sha1(clean_html_str).hexdigest()
citations = get_citations_from_tree(complete_html_tree, case_path)
judges = get_judge(clean_html_tree, case_path)
court = get_court_object(clean_html_tree, citations, case_path, judges)
doc = Document(
source='L',
sha1=sha1,
html=
clean_html_str, # we clear this field later, putting the value into html_lawbox.
date_filed=get_date_filed(clean_html_tree,
citations=citations,
case_path=case_path,
court=court),
precedential_status=get_precedential_status(),
judges=judges,
download_url=case_path,
)
cite = Citation(docket_number=get_docket_number(
clean_html_tree, case_path=case_path, court=court))
docket = Docket(
case_name=get_case_name(complete_html_tree, case_path),
court=court,
)
# Necessary for dup_finder.
path = '//p/text()'
doc.body_text = ' '.join(clean_html_tree.xpath(path))
# Add the dict of citations to the object as its attributes.
citations_as_dict = map_citations_to_models(citations)
for k, v in citations_as_dict.iteritems():
setattr(cite, k, v)
doc.citation = cite
doc.docket = docket
return doc
def import_mayer(case_path):
"""Open the file, get its contents, convert to XML and extract the meta data.
Return a document object for saving in the database
"""
#raw_text = open(case_path).read()
#clean_html_tree, complete_html_tree, clean_html_str, body_text = get_html_from_raw_text(raw_text)
tree = html.parse(case_path)
sha1 = hashlib.sha1(clean_html_str).hexdigest()
citations = get_citations_from_tree(complete_html_tree, case_path)
judges = get_judge(clean_html_tree, case_path)
court = get_court_object(clean_html_tree, citations, case_path, judges)
doc = Document(
source='L',
sha1=sha1,
html=clean_html_str, # we clear this field later, putting the value into html_lawbox.
date_filed=get_date_filed(clean_html_tree, citations=citations, case_path=case_path, court=court),
precedential_status=get_precedential_status(),
judges=judges,
download_url=case_path,
)
cite = Citation()
docket = Docket(
docket_number=get_docket_number(
clean_html_tree,
case_path=case_path,
court=court
),
case_name=get_case_name(complete_html_tree, case_path),
court=court,
)
# Necessary for dup_finder.
path = '//p/text()'
doc.body_text = ' '.join(clean_html_tree.xpath(path))
# Add the dict of citations to the object as its attributes.
citations_as_dict = map_citations_to_models(citations)
for k, v in citations_as_dict.iteritems():
setattr(cite, k, v)
doc.citation = cite
doc.docket = docket
return doc
def setUp(self):
self.court = Court.objects.get(pk='test')
# create 3 documents with their citations and dockets
c1, c2, c3 = Citation(case_name=u"c1"), Citation(
case_name=u"c2"), Citation(case_name=u"c3")
c1.save(index=False)
c2.save(index=False)
c3.save(index=False)
docket1 = Docket(
case_name=u"c1",
court=self.court,
)
docket2 = Docket(
case_name=u"c2",
court=self.court,
)
docket3 = Docket(
case_name=u"c3",
court=self.court,
)
docket1.save()
docket2.save()
docket3.save()
d1, d2, d3 = Document(date_filed=date.today()), Document(
date_filed=date.today()), Document(date_filed=date.today())
d1.citation, d2.citation, d3.citation = c1, c2, c3
d1.docket, d2.docket, d3.docket = docket1, docket2, docket3
doc_list = [d1, d2, d3]
for d in doc_list:
d.citation.save(index=False)
d.save(index=False)
# create simple citing relation: 1 cites 2 and 3; 2 cites 3; 3 cites 1;
d1.cases_cited.add(d2.citation)
d2.citation_count += 1
d2.cases_cited.add(d3.citation)
d3.citation_count += 1
d3.cases_cited.add(d1.citation)
d1.citation_count += 1
d1.cases_cited.add(d3.citation)
d3.citation_count += 1
d1.save(index=False)
d2.save(index=False)
d3.save(index=False)
def setUp(self):
self.court = Court.objects.get(pk='test')
# create 3 documents with their citations and dockets
c1, c2, c3 = Citation(case_name=u"c1"), Citation(
case_name=u"c2"), Citation(case_name=u"c3")
c1.save(index=False)
c2.save(index=False)
c3.save(index=False)
docket1 = Docket(
case_name=u"c1",
court=self.court,
)
docket2 = Docket(
case_name=u"c2",
court=self.court,
)
docket3 = Docket(
case_name=u"c3",
court=self.court,
)
docket1.save()
docket2.save()
docket3.save()
d1, d2, d3 = Document(date_filed=date.today()), Document(
date_filed=date.today()), Document(date_filed=date.today())
d1.citation, d2.citation, d3.citation = c1, c2, c3
d1.docket, d2.docket, d3.docket = docket1, docket2, docket3
doc_list = [d1, d2, d3]
for d in doc_list:
d.citation.save(index=False)
d.save(index=False)
# create simple citing relation: 1 cites 2 and 3; 2 cites 3; 3 cites 1;
d1.cases_cited.add(d2.citation)
d2.citation_count += 1
d2.cases_cited.add(d3.citation)
d3.citation_count += 1
d3.cases_cited.add(d1.citation)
d1.citation_count += 1
d1.cases_cited.add(d3.citation)
d3.citation_count += 1
d1.save(index=False)
d2.save(index=False)
d3.save(index=False)
def import_resource_org_item(case_location):
"""Using the path to a case, import it, gathering all needed meta data.
Path is any valid URI that the requests library can handle.
"""
def get_file(location):
if location.startswith('/'):
with open(location) as f:
r = requests.Session()
r.content = f.read()
else:
r = requests.get(location)
return fromstring(r.content), get_clean_body_content(r.content)
# Get trees and text for the opinion itself and for the index page
# that links to it. Each has useful data.
case_tree, case_text = get_file(case_location)
vol_location = case_location.rsplit('/', 1)[-2] + '/index.html'
vol_tree, vol_text = get_file(vol_location)
html, blocked = anonymize(get_case_body(case_tree))
case_location_relative = case_location.rsplit('/', 1)[1]
case_name, status = get_case_name_and_status(
vol_tree, case_location_relative)
cite = Citation(
case_name=case_name,
docket_number=get_docket_number(case_location),
federal_cite_one=get_west_cite(vol_tree, case_location_relative),
)
docket = Docket(
court=Court.objects.get(pk=get_court_id(case_tree)),
case_name=case_name,
)
doc = Document(
date_filed=get_date_filed(vol_tree, case_location_relative),
source='R',
sha1=hashlib.sha1(case_text).hexdigest(),
citation=cite,
docket=docket,
download_url=case_location,
html=html,
precedential_status=status,
)
if blocked:
doc.blocked = True
docket.blocked = True
doc.date_blocked = datetime.date.today()
docket.date_blocked = datetime.date.today()
cite.save()
docket.save()
doc.docket = docket
doc.citation = cite
doc.save()
# Update the citation graph
from alert.citations.tasks import update_document_by_id
update_document_by_id(doc.pk)
return doc
def import_resource_org_item(case_location):
"""Using the path to a case, import it, gathering all needed meta data.
Path is any valid URI that the requests library can handle.
"""
def get_file(location):
if location.startswith('/'):
with open(location) as f:
r = requests.Session()
r.content = f.read()
else:
r = requests.get(location)
return fromstring(r.content), get_clean_body_content(r.content)
# Get trees and text for the opinion itself and for the index page
# that links to it. Each has useful data.
case_tree, case_text = get_file(case_location)
vol_location = case_location.rsplit('/', 1)[-2] + '/index.html'
vol_tree, vol_text = get_file(vol_location)
html, blocked = anonymize(get_case_body(case_tree))
case_location_relative = case_location.rsplit('/', 1)[1]
case_name, status = get_case_name_and_status(vol_tree,
case_location_relative)
cite = Citation(
case_name=case_name,
docket_number=get_docket_number(case_location),
federal_cite_one=get_west_cite(vol_tree, case_location_relative),
)
docket = Docket(
court=Court.objects.get(pk=get_court_id(case_tree)),
case_name=case_name,
)
doc = Document(
date_filed=get_date_filed(vol_tree, case_location_relative),
source='R',
sha1=hashlib.sha1(case_text).hexdigest(),
citation=cite,
docket=docket,
download_url=case_location,
html=html,
precedential_status=status,
)
if blocked:
doc.blocked = True
docket.blocked = True
doc.date_blocked = datetime.date.today()
docket.date_blocked = datetime.date.today()
cite.save()
docket.save()
doc.docket = docket
doc.citation = cite
doc.save()
# Update the citation graph
from alert.citations.tasks import update_document_by_id
update_document_by_id(doc.pk)
return doc
def scrape_court(site, full_crawl=False):
download_error = False
# Get the court object early for logging
# opinions.united_states.federal.ca9_u --> ca9
court_str = site.court_id.split('.')[-1].split('_')[0]
court = Court.objects.get(pk=court_str)
dup_checker = DupChecker(site.court_id, full_crawl=full_crawl)
abort = dup_checker.abort_by_hash(site.hash)
if not abort:
for i in range(0, len(site.case_names)):
msg, r = get_binary_content(site.download_urls[i], site._get_cookies())
clean_content = site._cleanup_content(r.content)
if msg:
logger.warn(msg)
ErrorLog(log_level='WARNING',
court=court,
message=msg).save()
continue
current_date = site.case_dates[i]
try:
next_date = site.case_dates[i + 1]
except IndexError:
next_date = None
# Make a hash of the data. Need to convert unicode to binary before hashing.
if type(clean_content) == unicode:
hash_content = clean_content.encode('utf-8')
else:
hash_content = clean_content
sha1_hash = hashlib.sha1(hash_content).hexdigest()
if court_str == 'nev' and site.precedential_statuses[i] == 'Unpublished':
# Nevada's non-precedential cases have different SHA1 sums every time.
onwards = dup_checker.should_we_continue_break_or_carry_on(
current_date,
next_date,
lookup_value=site.download_urls[i],
lookup_by='download_url'
)
else:
onwards = dup_checker.should_we_continue_break_or_carry_on(
current_date,
next_date,
lookup_value=sha1_hash,
lookup_by='sha1'
)
if onwards == 'CONTINUE':
# It's a duplicate, but we haven't hit any thresholds yet.
continue
elif onwards == 'BREAK':
# It's a duplicate, and we hit a date or dup_count threshold.
dup_checker.update_site_hash(sha1_hash)
break
elif onwards == 'CARRY_ON':
# Not a duplicate, carry on
logger.info('Adding new document found at: %s' % site.download_urls[i])
dup_checker.reset()
# Make a citation
cite = Citation(case_name=site.case_names[i])
if site.docket_numbers:
cite.docket_number = site.docket_numbers[i]
if site.neutral_citations:
cite.neutral_cite = site.neutral_citations[i]
if site.west_citations:
cite.federal_cite_one = site.west_citations[i]
if site.west_state_citations:
cite.west_state_cite = site.west_state_citations[i]
# Make the document object
doc = Document(source='C',
sha1=sha1_hash,
date_filed=site.case_dates[i],
court=court,
download_url=site.download_urls[i],
precedential_status=site.precedential_statuses[i])
# Make and associate the file object
try:
cf = ContentFile(clean_content)
extension = get_extension(r.content)
# See issue #215 for why this must be lower-cased.
file_name = trunc(site.case_names[i].lower(), 75) + extension
doc.local_path.save(file_name, cf, save=False)
except:
msg = 'Unable to save binary to disk. Deleted document: % s.\n % s' % \
(cite.case_name, traceback.format_exc())
logger.critical(msg)
ErrorLog(log_level='CRITICAL', court=court, message=msg).save()
download_error = True
continue
if site.judges:
doc.judges = site.judges[i]
if site.nature_of_suit:
doc.nature_of_suit = site.nature_of_suit[i]
# Save everything, but don't update Solr index yet
cite.save(index=False)
doc.citation = cite
doc.save(index=False)
# Extract the contents asynchronously.
extract_doc_content(doc.pk, callback=subtask(extract_by_ocr))
logger.info("Successfully added doc %s: %s" % (doc.pk, site.case_names[i]))
# Update the hash if everything finishes properly.
logger.info("%s: Successfully crawled." % site.court_id)
if not download_error and not full_crawl:
# Only update the hash if no errors occurred.
dup_checker.update_site_hash(site.hash)
def test_pagerank_calculation(self):
"""Create a few Documents and fake citation relation among them, then
run the pagerank algorithm. Check whether this simple case can get the
correct result.
"""
# Set up some handy variables
self.court = Court.objects.get(pk='test')
# create 3 documents with their citations and dockets
c1, c2, c3 = Citation(case_name=u"c1"), Citation(
case_name=u"c2"), Citation(case_name=u"c3")
c1.save(index=False)
c2.save(index=False)
c3.save(index=False)
docket1 = Docket(
case_name=u"c1",
court=self.court,
)
docket2 = Docket(
case_name=u"c2",
court=self.court,
)
docket3 = Docket(
case_name=u"c3",
court=self.court,
)
d1, d2, d3 = Document(date_filed=date.today()), Document(
date_filed=date.today()), Document(date_filed=date.today())
d1.citation, d2.citation, d3.citation = c1, c2, c3
d1.docket, d2.docket, d3.docket = docket1, docket2, docket3
doc_list = [d1, d2, d3]
for d in doc_list:
d.citation.save(index=False)
d.save(index=False)
#create simple citing relation: 1 cites 2 and 3; 2 cites 3; 3 cites 1;
d1.cases_cited.add(d2.citation)
d2.citation_count += 1
d2.cases_cited.add(d3.citation)
d3.citation_count += 1
d3.cases_cited.add(d1.citation)
d1.citation_count += 1
d1.cases_cited.add(d3.citation)
d3.citation_count += 1
d1.save(index=False)
d2.save(index=False)
d3.save(index=False)
#calculate pagerank of these 3 document
comm = Command()
self.verbosity = 1
comm.do_pagerank(chown=False)
# read in the pagerank file, converting to a dict
pr_values_from_file = {}
with open(get_data_dir_location() + "external_pagerank") as f:
for line in f:
pk, value = line.split('=')
pr_values_from_file[pk] = float(value.strip())
# Verify that whether the answer is correct, based on calculations in
# Gephi
answers = {
'1': 0.387790,
'2': 0.214811,
'3': 0.397400,
}
for key, value in answers.iteritems():
self.assertTrue(
(abs(pr_values_from_file[key]) - value) < 0.0001,
msg="The answer for item %s was %s when it should have been "
"%s" % (
key,
pr_values_from_file[key],
answers[key],
))
