def test_parse_000038(self):
"""
parse 000038.html"""
p = parsepolicelog.PoliceLogParser(TEST_DATA_IN)
L = p.parse_log("000038.html", 2009)
#
# should be 70 crime reports in 000038
self.assert_(len(L) == 70)
#
# check distribution of parsed crimes
distrib = {}
for crime in L:
category = crime['category']
if distrib.has_key(category):
distrib[category] = distrib[category] + 1
else:
distrib[category] = 1
self.assert_(distrib['auto burglary'] == 12)
self.assert_(distrib['grand theft'] == 5)
self.assert_(distrib['suspicious circumstances'] == 10)
self.assert_(distrib['domestic disturbance'] == 4)
self.assert_(distrib['battery'] == 5)
self.assert_(distrib['commercial burglary'] == 2)
self.assert_(distrib['residential burglary'] == 5)
self.assert_(distrib['vandalism'] == 13)
self.assert_(distrib['robbery'] == 4)
self.assert_(distrib['assault'] == 1)
self.assert_(distrib['stolen vehicle'] == 9)
def test_parse_short_log(self):
"""
parse short_log.html"""
p = parsepolicelog.PoliceLogParser(source_dir=TEST_DATA_IN)
L = p.parse_log("short_log.html", 2009)
#
# two reports in short_log.html
self.assert_(len(L) == 2)
#
# verify first report
#</table>Auto Burglary <P></p>
#<p class=story_text>2100 block Creeden Way, 1/23<P></p>
R = L[0]
self.assert_(R['category'] == 'auto burglary')
self.assert_(R['date_day'] == 23)
self.assert_(R['date_month'] == 1)
self.assert_(R['date_year'] == 2009)
self.assert_(R['address'].lower() == '2100 block creeden way')
self.assert_(R['map_scale'] == mapscale.BLOCK)
# verify report
#<p class=story_text>Commercial Burglary<P></p>
#<p class=story_text>California St./S. Rengstorff Ave., 2/26<P></p>
R = L[1]
self.assert_(R['category'] == 'commercial burglary')
self.assert_(R['date_day'] == 26)
self.assert_(R['date_month'] == 2)
self.assert_(R['date_year'] == 2009)
self.assert_(
R['address'].lower() == 'california st./s. rengstorff ave.')
self.assert_(R['map_scale'] == mapscale.INTERSECTION)
def test_map_scale_id(self):
p = parsepolicelog.PoliceLogParser()
line = "<p class=story_text>200 block Rock St., 1/29<P></p>"
report = p.parse_report_line(line)
self.assert_(report['map_scale'] == mapscale.BLOCK)
self.assert_(report['date_month'] == 1)
self.assert_(report['date_day'] == 29)
line = "<p class=story_text>2500 West Middlefield Rd., 12/30<P></p>"
report = p.parse_report_line(line)
self.assert_(report['map_scale'] == mapscale.EXACT)
self.assert_(report['date_month'] == 12)
self.assert_(report['date_day'] == 30)
line = "<p class=story_text>Crittenden Lane/N. Shoreline Blvd., 1/30<P></p>"
report = p.parse_report_line(line)
self.assert_(report['map_scale'] == mapscale.INTERSECTION)
self.assert_(report['date_month'] == 1)
self.assert_(report['date_day'] == 30)
line = "<p class=story_text>Safeway, North Shoreline Blvd., 1/25 <P></p>"
report = p.parse_report_line(line)
self.assert_(report['map_scale'] == mapscale.OTHER)
self.assert_(report['date_month'] == 1)
self.assert_(report['date_day'] == 25)
def setUp(self):
print 'setUp: database_populated=%s' % self.database_populated
if TestParsePoliceLog.database_populated == False:
#
# set up database
issue_num = 39
police_log = PoliceLog(issue_number=issue_num,
issue_exists=True,
pub_date=datetime.date(2006, 3, 10))
police_log.save()
#
# associate db entry with
# log file in test directory
test_dir = TEST_DATA_IN
findpolicelog.sync_download_issue_with_db(issue_num, test_dir)
#
# parse crime log and load crimes into db
p = parsepolicelog.PoliceLogParser(test_dir)
p.parse_log_and_populate_db(issue_num, issue_num)
#
# remember that we've performed setup
TestParsePoliceLog.database_populated = True
else:
pass
def test_crimereport_unique_hash(self):
"""
verify that crimereport hash prevents duplicate crime report entries in db"""
#
# set up database
issue_num = 39
# police_log=PoliceLog(issue_number=issue_num,
# issue_exists=True,
# pub_date=datetime.date(2006,3,10))
# police_log.save()
# #
# # associate db entry with
# # log file in test directory
test_dir = TEST_DATA_IN
# findpolicelog.sync_download_issue_with_db(issue_num,test_dir)
# #
# # parse crime log and load crimes into db
# p=parsepolicelog.PoliceLogParser(test_dir)
# p.parse_log_and_populate_db(issue_num,issue_num)
#
# count entries in db
crime_report_list = CrimeReport.objects.filter(
policelog__issue_number__exact=issue_num)
num_reports1 = len(crime_report_list)
#
# parse crime log and load crimes into db AGAIN
p = parsepolicelog.PoliceLogParser(test_dir)
p.parse_log_and_populate_db(issue_num, issue_num)
crime_report_list = CrimeReport.objects.filter(
policelog__issue_number__exact=issue_num)
num_reports2 = len(crime_report_list)
#
# verify that second parse_log_and_populate_db didn't result
# in additional db entries
self.assert_(num_reports1 == num_reports2)
def test_parse_000039_01(self):
line = '''<p class=story_text>Moffett Blvd./Stevens Creek, 1/30<P></p>'''
p = parsepolicelog.PoliceLogParser()
crime_report = p.parse_report_line(line)
self.assert_(crime_report['map_scale'] == mapscale.INTERSECTION)
def test_parse_000039_00(self):
line = '''<p class=story_text>700 block San Pablo, 1/30<P></p>'''
p = parsepolicelog.PoliceLogParser()
crime_report = p.parse_report_line(line)
self.assert_(crime_report['map_scale'] == mapscale.BLOCK)
def test_populate_db_issue_000038(self):
"""
parse issue #38 police log, stuff into database"""
#
# set up database
issue_num = 38
police_log = PoliceLog(issue_number=issue_num,
issue_exists=True,
pub_date=datetime.date(2006, 3, 10))
police_log.save()
#
# associate db entry with
# log file in test directory
test_dir = TEST_DATA_IN
findpolicelog.sync_download_issue_with_db(issue_num, test_dir)
#
# parse crime log and load crimes into db
p = parsepolicelog.PoliceLogParser(test_dir)
p.parse_log_and_populate_db(issue_num, issue_num)
#
# should be 70 crime reports associated with 000038
# the following query set is equivalent to:
# police_log=PoliceLog.objects.get(issue_number__exact=issue_num)
# crime_report_list=police_log.crimereport_set.all()
crime_report_list = CrimeReport.objects.filter(
policelog__issue_number__exact=issue_num)
self.assert_(len(crime_report_list) == 70)
if False:
for crime_report in crime_report_list:
print "%d %d %s (%s)" % (
crime_report.category, crime_report.line_num,
crime_report.address,
mapscale.dict[crime_report.map_scale])
#
# parse_and_populate_db shouldn't have created an issue 39
try:
police_log = PoliceLog.objects.get(issue_number__exact=issue_num +
1)
except PoliceLog.DoesNotExist:
pass
#
# check distribution of parsed crimes
category_distrib = get_crime_type_distribution(crime_report_list)
scale_distrib = get_mapscale_distribution(crime_report_list)
self.assert_(category_distrib['auto burglary'] == 12)
self.assert_(category_distrib['grand theft'] == 5)
self.assert_(category_distrib['suspicious circumstances'] == 10)
self.assert_(category_distrib['domestic disturbance'] == 4)
self.assert_(category_distrib['battery'] == 5)
self.assert_(category_distrib['commercial burglary'] == 2)
self.assert_(category_distrib['residential burglary'] == 5)
self.assert_(category_distrib['vandalism'] == 13)
self.assert_(category_distrib['robbery'] == 4)
self.assert_(category_distrib['stolen vehicle'] == 9)
self.assert_(category_distrib['assault'] == 1)
self.assert_(scale_distrib[mapscale.BLOCK] == 56)
self.assert_(scale_distrib[mapscale.INTERSECTION] == 1)
self.assert_(scale_distrib[mapscale.EXACT] == 2)
self.assert_(scale_distrib[mapscale.OTHER] == 11)