def test_starts_and_width_height(self):
b = BoundingBox(start_x=start.x,
start_y=start.y,
width=width,
height=height)
self.compare_points(b.start, start)
self.compare_points(b.end, end)
def get_bounding_box_and_baseline(line_tag):
bbox = None
baseline = None
title = line_tag.attrib["title"]
title_parts = title.split(";")
bbox_part = get_part("bbox", title_parts)
if bbox_part is not None:
bbox_parts = bbox_part.split()
bbox = BoundingBox(start_x=int(bbox_parts[1]),
start_y=int(bbox_parts[2]),
end_x=int(bbox_parts[3]),
end_y=int(bbox_parts[4]))
baseline_part = get_part("baseline", title_parts)
if baseline_part is not None:
baseline_parts = baseline_part.split()
baseline_slope = float(baseline_parts[1])
baseline_offset = int(float(baseline_parts[2]))
bbox_width = bbox.end.x - bbox.start.x
baseline_y_start = bbox.end.y + baseline_offset
baseline_y_end = int(bbox.end.y + baseline_offset +
bbox_width * baseline_slope)
baseline = Baseline(start_x=bbox.start.x,
start_y=baseline_y_start,
end_x=bbox.end.x,
end_y=baseline_y_end)
return bbox, baseline
def test_starts_and_ends(self):
b = BoundingBox(start_x=start.x,
start_y=start.y,
end_x=end.x,
end_y=end.y)
self.compare_points(b.start, start)
self.compare_points(b.end, end)
def process_dataset(dataset, log_path):
new_dataset = Dataset()
total_pages = len(dataset.pages)
for index, page_id in enumerate(dataset.pages):
print("\r{current}/{total} ({percentage:.2f} %)".format(
current=index + 1,
total=total_pages,
percentage=100 * float(index + 1) / total_pages),
end="")
transformation = read_log(
os.path.join(log_path,
page_id.split(".")[0] + ".jpg.log"))
try:
inversion = np.linalg.inv(transformation)
except:
print("\nPage ID: {id}".format(id=page_id))
print("Transformation matrix: {matrix}".format(
matrix=transformation))
continue
page = dataset.get_page(page_id)
lines = []
if page.lines is not None:
for line in page.lines:
new_bb_start = transform_point(line.bounding_box.start,
inversion)
new_bb_end = transform_point(line.bounding_box.end, inversion)
new_bb_inner_points = []
if line.bounding_box.inner_points is not None:
for point in line.bounding_box.inner_points:
new_bb_inner_points.append(
transform_point(point, inversion))
new_baseline_start = transform_point(line.baseline.start,
inversion)
new_baseline_end = transform_point(line.baseline.end,
inversion)
lines.append(
Line(
line.text,
BoundingBox(new_bb_start, new_bb_end,
new_bb_inner_points),
Baseline(new_baseline_start, new_baseline_end), None))
new_dataset.add_page(Page(page_id + ".jpg_rec", lines))
print()
return new_dataset
def test_empty_raises(self):
with self.assertRaises(ValueError):
b = BoundingBox()
def test_starts_and_height_raises(self):
with self.assertRaises(ValueError):
b = BoundingBox(start_x=start.x, start_y=start.y, height=height)
def test_start_and_height_raises(self):
with self.assertRaises(ValueError):
b = BoundingBox(start=start, height=height)
def test_starts_and_width_raises(self):
with self.assertRaises(ValueError):
b = BoundingBox(start_x=start.x, start_y=start.y, width=width)
def test_start_and_width_raises(self):
with self.assertRaises(ValueError):
b = BoundingBox(start=start, width=width)
def test_starts_and_end_y_raises(self):
with self.assertRaises(ValueError):
b = BoundingBox(start_x=start.x, start_y=start.y, end_y=end.y)
def test_start_and_end_x_raises(self):
with self.assertRaises(ValueError):
b = BoundingBox(start=start, end_x=end.x)
def test_start_and_end(self):
b = BoundingBox(start=start, end=end)
self.compare_points(b.start, start)
self.compare_points(b.end, end)
def test_ends_raises(self):
with self.assertRaises(ValueError):
b = BoundingBox(end_x=end.x, end_y=end.y)
def test_end_raises(self):
with self.assertRaises(ValueError):
b = BoundingBox(end=end)
def test_start_raises(self):
with self.assertRaises(ValueError):
b = BoundingBox(start=start)