def test_get_output(self):
boxes = [
builders.Box("word1", ((10, 11), (12, 13)), 95),
builders.Box("word2", ((11, 12), (13, 14))),
builders.Box("word3", ((12, 13), (14, 15))),
builders.Box("word4", ((13, 14), (15, 16)), 87),
]
for box in boxes:
self.builder.add_word(box.content, box.position, box.confidence)
output = self.builder.get_output()
for box, box_expected in zip(output, boxes):
self.assertIsInstance(box, builders.Box)
self.assertEqual(box, box_expected)
self.assertEqual(box.content, box_expected.content)
def setUp(self):
box1 = builders.Box("word1", ((15, 22), (23, 30)))
box2 = builders.Box("word2", ((25, 23), (30, 32)))
box3 = builders.Box("word3", ((32, 25), (40, 32)), 95)
box4 = builders.Box("word4", ((41, 18), (44, 33)), 98)
box_unicode = builders.Box("\xe9", ((1, 2), (3, 4)), 98)
self.line1 = builders.LineBox([box1, box2, box3, box4],
((14, 15), (45, 33)))
self.line1_bis = builders.LineBox([box1, box2], ((14, 15), (45, 33)))
self.line2 = builders.LineBox([box3, box4], ((30, 5), (53, 20)))
self.line1_dupl = builders.LineBox([box1, box2, box3, box4],
((14, 15), (45, 33)))
self.line_unicode = builders.LineBox([box1, box_unicode],
((1, 2), (3, 4)))
def test_write_file(self):
output_fh = StringIO()
lines = []
for l in range(4):
boxes = []
for b in range(4):
word = "word" + str(4 * l + b)
position = ((4 * l + b, 4 * l + b + 1), (4 * l + b + 2,
4 * l + b + 3))
boxes.append(builders.Box(word, position, randint(0, 100)))
line_position = ((4 * l, 4 * (l + 1)), (4 * l + 2,
4 * (l + 1) + 2))
lines.append(builders.LineBox(boxes, line_position))
self.builder.write_file(output_fh, lines)
output_fh.seek(0)
output = output_fh.read()
for line in lines:
for box in line.word_boxes:
self.assertIn(box.content, output)
self.assertIn(
"{} {} {} {}".format(
box.position[0][0],
box.position[0][1],
box.position[1][0],
box.position[1][1],
), output)
self.assertIn(str(box.confidence), output)
def test_add_word(self):
box = builders.Box("word", ((10, 11), (12, 13)))
self.builder.start_line(box)
self.builder.add_word(box.content, box)
self.assertEqual(self.builder.built_text[0], box.content)
self.builder.add_word(box.content, box)
self.assertEqual(self.builder.built_text[0],
box.content + " " + box.content)
def test_get_output(self):
box = builders.Box("word", ((10, 11), (12, 13)))
self.builder.start_line(box)
self.builder.add_word("word1", box)
self.builder.add_word("word2", box)
self.builder.start_line(box)
self.builder.add_word("word3", box)
self.builder.add_word("word4", box)
self.assertEqual(self.builder.get_output(), "word1 word2\nword3 word4")
def test_write_file(self):
output = StringIO()
boxes = [
builders.Box("word1", ((10, 11), (12, 13)), 95),
builders.Box("word2", ((11, 12), (13, 14))),
builders.Box("word3", ((12, 13), (14, 15))),
builders.Box("word4", ((13, 14), (15, 16)), 87),
]
self.builder.write_file(output, boxes)
output.seek(0)
output = output.read()
for box in boxes:
self.assertIn(box.content, output)
self.assertIn(
"{} {} {} {}".format(
box.position[0][0],
box.position[0][1],
box.position[1][0],
box.position[1][1],
), output)
self.assertIn(str(box.confidence), output)
def test_write_file(self):
builder = tesseract.CharBoxBuilder()
output = StringIO()
boxes = [
builders.Box("a", ((10, 11), (12, 13)), 95),
builders.Box("b", ((11, 12), (13, 14))),
builders.Box("c", ((12, 13), (14, 15))),
builders.Box("d", ((13, 14), (15, 16)), 87),
builders.Box(u"\xe9", ((14, 15), (16, 17)), 88),
]
builder.write_file(output, boxes)
output.seek(0)
output = output.read()
for box in boxes:
self.assertIn(box.content, output)
self.assertIn(
u"{} {} {} {}".format(
box.position[0][0],
box.position[0][1],
box.position[1][0],
box.position[1][1],
), output)
def test_get_output(self):
lines = []
for l in range(4):
boxes = []
for b in range(4):
word = "word" + str(4 * l + b)
position = ((4 * l + b, 0), (0, 0))
boxes.append(builders.Box(word, position, randint(0, 100)))
line_position = ((4 * l, 4 * (l + 1)), (4 * l + 2,
4 * (l + 1) + 2))
lines.append(builders.LineBox(boxes, line_position))
for line in lines:
self.builder.start_line(line.position)
for word in line.word_boxes:
self.builder.add_word(word.content, word.position,
word.confidence)
self.builder.end_line() # could be useful in future
output = self.builder.get_output()
for line, line_expected in zip(output, lines):
self.assertIsInstance(line, builders.LineBox)
self.assertEqual(line, line_expected)
def setUp(self):
self.box1 = builders.Box("word1", ((15, 22), (23, 42)))
self.box1_bis = builders.Box("word1_bis", ((15, 22), (23, 42)))
self.box2 = builders.Box("word2", ((30, 5), (40, 15)), 95)
self.box_unicode = builders.Box("\xe9", ((1, 2), (3, 4)))
def test_add_word_no_line(self):
box = builders.Box("word", ((10, 11), (12, 13)))
with self.assertRaises(IndexError):
self.builder.add_word(box.content, box)
def test_start_line(self):
box = builders.Box("word", ((10, 11), (12, 13)))
self.builder.start_line(box)
self.assertListEqual(self.builder.built_text, [""])
def test_add_word_no_line(self):
box = builders.Box("word", ((1, 2), (3, 4)), 42)
with self.assertRaises(IndexError):
self.builder.add_word(box.content, box.position, box.confidence)
self.assertListEqual(self.builder.lines, [])
def test_add_word(self):
box = builders.Box("word", ((1, 2), (3, 4)), 42)
self.builder.add_word(box.content, box.position, box.confidence)
for box in self.builder.word_boxes:
self.assertIsInstance(box, builders.Box)
self.assertEqual(self.builder.word_boxes[0], box)
def test_start_line(self):
box = builders.Box("word", ((1, 2), (3, 4)))
before = list(self.builder.word_boxes)
self.builder.start_line(box)
self.assertEqual(self.builder.word_boxes, before)