def testPdfParserPhysical(self):
"""Get Pdf physical structure."""
url = "file://%s" % pdf_file_name
pdf_parser = PdfParser(pdf_file_name, url, pdf_file_name)
phys = pdf_parser.get_physical_structure()
self.assertEqual(phys[0]['label'], pdf_file_name, "Physical Structure "\
"missmatch: %s != %s" % (phys[0]['label'], pdf_file_name))
def testPdfParserLogical(self):
"""Get Pdf logical structure."""
pdf_parser = PdfParser(pdf_file_name, "file://%s" % pdf_file_name, pdf_file_name)
logic = pdf_parser.get_logical_structure()
first_section = logic[0]["label"]
self.assertEqual(
first_section, "Introduction", "TOC is not well " "detected: %s != %s" % (first_section, "Introduction")
)
def testPdfParserLogical(self):
"""Get Pdf logical structure."""
pdf_parser = PdfParser(pdf_file_name, "file://%s" % pdf_file_name,
pdf_file_name)
logic = pdf_parser.get_logical_structure()
first_section = logic[0]['label']
self.assertEqual (first_section, 'Introduction', "TOC is not well "\
"detected: %s != %s" %(first_section, 'Introduction'))
def testPdfParserMeta(self):
"""Get Pdf Metadata."""
pdf_parser = PdfParser(pdf_file_name, "file://%s" % pdf_file_name,
pdf_file_name)
meta = pdf_parser.get_metadata()
title = meta['title']
self.assertEqual(title, u'Multivio: Project description', "Metadata has "\
"not been correctly detected %s != %s" %
(title, u'Multivio: Project description'))
def testPdfParserMetaFileSize(self):
"""Get Pdf Metadata and test file size."""
pdf_parser = PdfParser(pdf_file_name, "file://%s" % pdf_file_name, pdf_file_name)
meta = pdf_parser.get_metadata()
file_size = meta["fileSize"]
ref_size = os.path.getsize(pdf_file_name)
self.assertEqual(
file_size, ref_size, "File size has " "not been correctly detected %s != %s" % (file_size, ref_size)
)
def testPdfParserMetaFileSize(self):
"""Get Pdf Metadata and test file size."""
pdf_parser = PdfParser(pdf_file_name, "file://%s" % pdf_file_name,
pdf_file_name)
meta = pdf_parser.get_metadata()
file_size = meta['fileSize']
ref_size = os.path.getsize(pdf_file_name)
self.assertEqual(file_size, ref_size, "File size has "\
"not been correctly detected %s != %s" %
(file_size, ref_size))
def testPdfParserPhysical(self):
"""Get Pdf physical structure."""
url = "file://%s" % pdf_file_name
pdf_parser = PdfParser(pdf_file_name, url, pdf_file_name)
phys = pdf_parser.get_physical_structure()
self.assertEqual(
phys[0]["label"],
pdf_file_name,
"Physical Structure " "missmatch: %s != %s" % (phys[0]["label"], pdf_file_name),
)
def testPdfParserMeta(self):
"""Get Pdf Metadata."""
pdf_parser = PdfParser(pdf_file_name, "file://%s" % pdf_file_name, pdf_file_name)
meta = pdf_parser.get_metadata()
title = meta["title"]
self.assertEqual(
title,
u"Multivio: Project description",
"Metadata has " "not been correctly detected %s != %s" % (title, u"Multivio: Project description"),
)
def testPdfParserLogicalSegfault(self):
"""Get Pdf logical structure for a malformed pdf:
http://www.tendancesit.com/pdf/19.pdf."""
file_name = "examples/toc_segfault.pdf"
try:
file(file_name)
except IOError:
raise Exception(
"Please go the examples directory and run: get_examples.sh")
pdf_parser = PdfParser(file_name, "file://%s" % file_name, file_name)
logic = pdf_parser.get_logical_structure()
first_section = logic[0]['label']
first_section_ref = 'Tendances IT 19 WEB.pdf'
self.assertEqual (first_section, first_section_ref, "TOC is not well "\
"detected: %s != %s" %(first_section, first_section_ref))
def testPdfParserLogicalSegfault(self):
"""Get Pdf logical structure for a malformed pdf:
http://www.tendancesit.com/pdf/19.pdf."""
file_name = "examples/toc_segfault.pdf"
try:
file(file_name)
except IOError:
raise Exception("Please go the examples directory and run: get_examples.sh")
pdf_parser = PdfParser(file_name, "file://%s" % file_name, file_name)
logic = pdf_parser.get_logical_structure()
first_section = logic[0]["label"]
first_section_ref = "Tendances IT 19 WEB.pdf"
self.assertEqual(
first_section,
first_section_ref,
"TOC is not well " "detected: %s != %s" % (first_section, first_section_ref),
)
def testPdfParser(self):
"""Check PdfParser instance."""
pdf_parser = PdfParser(pdf_file_name, "file://%s" % pdf_file_name,
pdf_file_name)
self.assert_(pdf_parser, "Can not create simple Parser Object")
