def test_for_loop(self):
pdf = pdftotext.PDF(get_file("two_page.pdf"))
result = ""
for page in pdf:
result = result + page
self.assertIn("one", result)
self.assertIn("two", result)
def test_read_landscape_0(self):
pdf = pdftotext.PDF(get_file("landscape_0.pdf"))
result = pdf[0]
self.assertIn("a", result)
self.assertIn("b", result)
self.assertIn("c", result)
self.assertIn("d", result)
def test_read_portrait(self):
pdf = pdftotext.PDF(get_file("portrait.pdf"))
result = pdf[0]
self.assertIn("a", result)
self.assertIn("b", result)
self.assertIn("c", result)
self.assertIn("d", result)
def test_read_landscape_90(self):
pdf = pdftotext.PDF(get_file("landscape_90.pdf"),
layout=pdftotext.TextLayout.raw,
encoding='latin1')
result = pdf[0]
self.assertIn("a", result)
self.assertIn("b", result)
self.assertIn("c", result)
self.assertIn("d", result)
def test_raw_vs_not(self):
filename = "table.pdf"
pdf = pdftotext.PDF(get_file(filename))
raw_pdf = pdftotext.PDF(get_file(filename),
layout=pdftotext.TextLayout.raw)
self.assertNotEqual(pdf[0], raw_pdf[0])
def test_iter(self):
pdf = pdftotext.PDF(get_file('two_page.pdf'))
self.assertEqual([x.strip() for x in list(pdf)], ['one.', 'two.'])
def test_read(self):
pdf = pdftotext.PDF(get_file("abcde.pdf"))
result = pdf[0]
self.assertIn("abcde", result)
def test_init_invalid_pdf_file(self):
pdf_file = io.BytesIO(b"wrong")
with self.assertRaises(TypeError):
pdftotext.PDF(pdf_file)
def test_init_corrupt_pdf_file(self):
with self.assertRaises(ValueError):
pdftotext.PDF(get_file("corrupt.pdf"))
def test_double_init_success(self):
pdf = pdftotext.PDF(get_file("abcde.pdf"))
pdf.__init__(get_file("blank.pdf"))
self.assertEqual(len(pdf), 1)
def test_init_file_in_text_mode(self):
text_file = io.StringIO(u"wrong")
with self.assertRaises(TypeError):
pdftotext.PDF(text_file)
def test_read_corrupt_page(self):
with self.assertRaises(IndexError):
pdf = pdftotext.PDF(get_file("corrupt_page.pdf"))
pdf[0]
def test_pdf_read_wrong_arg_type(self):
pdf = pdftotext.PDF(get_file("blank.pdf"))
with self.assertRaises(TypeError):
pdf["wrong"]
def test_read_page_two(self):
pdf = pdftotext.PDF(get_file("two_page.pdf"))
result = pdf[1]
self.assertIn("two", result)
def test_list_last_element(self):
pdf = pdftotext.PDF(get_file("two_page.pdf"))
self.assertIn("two", pdf[-1])
def test_list_invalid_element(self):
pdf = pdftotext.PDF(get_file("two_page.pdf"))
with self.assertRaises(IndexError):
pdf[2]
def test_list_length(self):
pdf = pdftotext.PDF(get_file("two_page.pdf"))
self.assertEqual(len(pdf), 2)
def test_length_one(self):
pdf = pdftotext.PDF(get_file("blank.pdf"))
self.assertEqual(len(pdf), 1)
def test_raw_is_not_default(self):
filename = "table.pdf"
pdf_default = pdftotext.PDF(get_file(filename))
pdf_raw_false = pdftotext.PDF(get_file(filename),
layout=pdftotext.TextLayout.physical)
self.assertEqual(pdf_default[0], pdf_raw_false[0])
def test_pdf_read_invalid_page_number(self):
pdf = pdftotext.PDF(get_file("blank.pdf"))
with self.assertRaises(IndexError):
pdf[100]