def report_extracted(self, input_filename, output_dir, project):
input_filename_base = os.path.basename(input_filename)
input_md5 = md5sum(input_filename)
pathname = output_prefix(input_filename, output_dir) + "*.pgm"
for fname in glob.glob(pathname):
output_filename = os.path.basename(fname)
output_md5 = md5sum(fname)
page_num = self.page_num(output_filename)
extract_image.send(self, input_filename=input_filename_base,
input_md5=input_md5, filename=output_filename,
md5=output_md5, page=page_num, project=project)
def split(self, output_dir, **kwargs):
"""Split an image that contains a table into per-cell images"""
for box in self.get_boxes(**kwargs):
col, row, left, upper, right, lower = box
region = self.im.crop((left, upper, right, lower))
region = self.transform(region, col, row)
# Increase contrast. Whiteish regions become absolute white
region = threshold(region, 200)
output_filename = cell_basename(self.filename, col, row) + ".tiff"
output_path = os.path.join(output_dir, output_filename)
region.save(output_path)
md5 = md5sum(output_path)
split_image.send(self, input_filename=self.filename,
input_md5=self.md5, filename=output_filename, md5=md5,
column=col, row=row, left=left, upper=upper, right=right,
lower=lower)
def ocr_image(self, input_filename, num_cols, num_rows, split_dir, ocr_dir,
user):
for col in range(num_cols):
for row in range(num_rows):
basename = cell_basename(input_filename, col, row)
img_filename = basename + '.tiff'
path = os.path.join(split_dir, img_filename)
outbase = os.path.join(ocr_dir, basename)
if os.path.exists(path):
img_md5 = md5sum(path)
cmd = "tesseract {} {}".format(path, outbase)
cmd_args = shlex.split(cmd)
subprocess.check_call(cmd_args)
output_filename = outbase + '.txt'
with open(output_filename, 'r') as f:
text = f.read()
image_text.send(self, source_filename=img_filename,
source_md5=img_md5, method='ocr',
text=text, user=user)
def __init__(self, filename):
self.md5 = md5sum(filename)
self.filename = os.path.basename(filename)
self._im = Image.open(filename)
