def test_converting_pdf_to_html(self):
"""Test that the pdf becomes the html that we expect"""
pdf_file = open( self.sample_pdf, 'r' )
html = KenyaParser.convert_pdf_to_html( pdf_file )
expected_html = open( self.sample_html, 'r' ).read()
self.assertEqual( html, expected_html )
def test_converting_pdf_to_html(self):
"""Test that the pdf becomes the html that we expect"""
pdf_file = open(self.sample_pdf, 'r')
html = KenyaParser.convert_pdf_to_html(pdf_file)
expected_html = open(self.sample_html, 'r').read()
self.assertEqual(html, expected_html)
def handle_noargs(self, **options):
for source in Source.objects.all().requires_processing():
if int(options.get('verbosity')) >= 2:
print "Looking at %s" % source
source.last_processing_attempt = datetime.datetime.now()
source.save()
pdf = source.file()
html = KenyaParser.convert_pdf_to_html(pdf)
data = KenyaParser.convert_html_to_data(html)
KenyaParser.create_entries_from_data_and_source(data, source)
def handle_noargs(self, **options):
for source in Source.objects.all().requires_processing():
if int(options.get('verbosity')) >= 2:
print "Looking at %s" % source
source.last_processing_attempt = datetime.datetime.now()
source.save()
pdf = source.file()
html = KenyaParser.convert_pdf_to_html( pdf )
data = KenyaParser.convert_html_to_data( html )
KenyaParser.create_entries_from_data_and_source( data, source )
def handle_noargs(self, **options):
verbose = int(options.get('verbosity')) >= 2
for source in Source.objects.all().requires_processing():
if verbose:
message = "{0}: Looking at {1}"
print message.format(source.list_page, source)
source.last_processing_attempt = datetime.datetime.now()
source.save()
pdf = source.file()
try:
html = KenyaParser.convert_pdf_to_html( pdf )
data = KenyaParser.convert_html_to_data( html )
KenyaParser.create_entries_from_data_and_source( data, source )
except Exception as e:
print "There was an exception when parsing {0}".format(pdf)
raise
