def optimized_image2text(filename, lang='eng', verbose=False):
ocr_txt = False
ocr_temp_dirname = tempfile.mkdtemp(prefix="opensemanticetl_ocr_descew_")
if verbose:
print("Optimizing image {}".format(filename))
# start external OCR Program
result = subprocess.call(['scantailor-cli', filename, ocr_temp_dirname])
if result == 0:
images = os.listdir(ocr_temp_dirname)
images.sort()
for image in images:
try:
result = False
imagefilename = ocr_temp_dirname + os.path.sep + image
# ignore the cache directory of scantailor, only files in directory
if os.path.isfile(imagefilename):
result = enhance_ocr.image2text(imagefilename,
lang,
verbose=verbose)
os.remove(imagefilename)
if result:
if ocr_txt:
ocr_txt = ocr_txt + '\n' + result
else:
ocr_txt = result
except KeyboardInterrupt:
raise KeyboardInterrupt
except BaseException as e:
sys.stderr.write(
"Exception while OCR descewed image of: {} - Maybe descewed image {} corrupt? Exception: {}\n"
.format(filename, imagefilename, e))
else:
sys.stderr.write(
"Error: Descewing images for OCR failed for {} with return code {}"
.format(filename, result))
shutil.rmtree(ocr_temp_dirname)
return ocr_txt
def optimized_image2text(filename, lang='eng', verbose=False):
ocr_txt = False
ocr_temp_dirname = tempfile.mkdtemp(prefix="opensemanticetl_ocr_descew_")
if verbose:
print ( "Optimizing image {}".format(filename) )
# start external OCR Program
result = subprocess.call(['scantailor-cli', filename, ocr_temp_dirname])
if result == 0:
images = os.listdir(ocr_temp_dirname)
images.sort()
for image in images:
try:
result = False
imagefilename = ocr_temp_dirname + os.path.sep + image
# ignore the cache directory of scantailor, only files in directory
if os.path.isfile(imagefilename):
result = enhance_ocr.image2text(imagefilename, lang, verbose=verbose)
os.remove(imagefilename)
if result:
if ocr_txt:
ocr_txt = ocr_txt + '\n' + result
else:
ocr_txt = result
except KeyboardInterrupt:
raise KeyboardInterrupt
except BaseException as e:
sys.stderr.write( "Exception while OCR descewed image of: {} - Maybe descewed image {} corrupt? Exception: {}\n" .format(filename, imagefilename, e) )
else:
sys.stderr.write ( "Error: Descewing images for OCR failed for {} with return code {}".format(filename, result) )
shutil.rmtree(ocr_temp_dirname)
return ocr_txt
def pdfimages2text(filename,
lang='eng',
verbose=False,
pdf_ocr=True,
pdf_ocr_descew=False):
ocr_txt = {}
ocr_descew_txt = {}
ocr_temp_dirname = tempfile.mkdtemp(prefix="opensemanticetl_pdf_ocr_")
# Extract all images of the pdf to tempdir with commandline tool "pdfimages" from poppler pdf toolbox
# -j = export as JPEG
# -p = write page name in image filename
result = subprocess.call([
'pdfimages', '-p', '-j', filename,
ocr_temp_dirname + os.path.sep + 'image'
])
if result == 0:
images = os.listdir(ocr_temp_dirname)
images.sort()
for image in images:
imagefilename = ocr_temp_dirname + os.path.sep + image
if pdf_ocr:
try:
# extract page number from extracted image filename (image-pagenumber-imagenumber.jpg)
pagenumber = int(image.split('-')[1])
result = enhance_ocr.image2text(filename=imagefilename,
lang=lang,
verbose=verbose)
if result:
if pagenumber in ocr_txt:
ocr_txt[pagenumber] += '\n' + result
else:
ocr_txt[pagenumber] = result
except KeyboardInterrupt:
raise KeyboardInterrupt
except BaseException as e:
sys.stderr.write(
"Exception while OCR of PDF: {} - maybe corrupt image: {} - exception: {}"
.format(filename, imagefilename, e))
if pdf_ocr_descew:
try:
# extract page number from extracted image filename (image-pagenumber-imagenumber.jpg)
pagenumber = int(image.split('-')[1])
result = enhance_ocr_descew.optimized_image2text(
imagefilename, lang, verbose=verbose)
if result:
if pagenumber in ocr_descew_txt:
ocr_descew_txt[pagenumber] += '\n\n' + result
else:
ocr_descew_txt[pagenumber] = result
except KeyboardInterrupt:
raise KeyboardInterrupt
except BaseException as e:
sys.stderr.write(
"Exception while optimized ocr pdf: {} - maybe corrupt image: {} - exception: {}"
.format(filename, imagefilename, e))
os.remove(imagefilename)
os.rmdir(ocr_temp_dirname)
else:
sys.stderr.write(
"Error: Extracting images from PDF failed for {} {}".format(
filename, result))
return ocr_txt, ocr_descew_txt
def pdfimages2text(filename, lang='eng', verbose=False, pdf_ocr=True, pdf_ocr_descew=False):
ocr_txt = {}
ocr_descew_txt = {}
ocr_temp_dirname = tempfile.mkdtemp(prefix="opensemanticetl_pdf_ocr_")
# Extract all images of the pdf to tempdir with commandline tool "pdfimages" from poppler pdf toolbox
# -j = export as JPEG
# -p = write page name in image filename
result = subprocess.call(['pdfimages', '-p' ,'-j', filename, ocr_temp_dirname + os.path.sep + 'image'])
if result == 0:
images = os.listdir(ocr_temp_dirname)
images.sort()
for image in images:
imagefilename = ocr_temp_dirname + os.path.sep + image
if pdf_ocr:
try:
# extract page number from extracted image filename (image-pagenumber-imagenumber.jpg)
pagenumber = int( image.split('-')[1] )
result = enhance_ocr.image2text(filename=imagefilename, lang=lang, verbose=verbose)
if result:
if pagenumber in ocr_txt:
ocr_txt[pagenumber] += '\n' + result
else:
ocr_txt[pagenumber] = result
except KeyboardInterrupt:
raise KeyboardInterrupt
except BaseException as e:
sys.stderr.write( "Exception while OCR of PDF: {} - maybe corrupt image: {} - exception: {}".format(filename, imagefilename, e) )
if pdf_ocr_descew:
try:
# extract page number from extracted image filename (image-pagenumber-imagenumber.jpg)
pagenumber = int( image.split('-')[1] )
result = enhance_ocr_descew.optimized_image2text(imagefilename, lang, verbose=verbose)
if result:
if pagenumber in ocr_descew_txt:
ocr_descew_txt[pagenumber] += '\n\n' + result
else:
ocr_descew_txt[pagenumber] = result
except KeyboardInterrupt:
raise KeyboardInterrupt
except BaseException as e:
sys.stderr.write( "Exception while optimized ocr pdf: {} - maybe corrupt image: {} - exception: {}".format(filename, imagefilename, e) )
os.remove(imagefilename)
os.rmdir(ocr_temp_dirname)
else:
sys.stderr.write( "Error: Extracting images from PDF failed for {} {}".format(filename, result) )
return ocr_txt, ocr_descew_txt
def pdfimages2text(filename,
lang='eng',
verbose=False,
pdf_ocr=True,
pdf_ocr_descew=False,
cache=None):
ocr_txt = {}
ocr_descew_txt = {}
if cache is not None:
try:
return load_cache(filename, cache, lang, pdf_ocr, pdf_ocr_descew)
except (FileNotFoundError, KeyError):
if verbose:
print('Not in OCR cache, starting OCR for {}'.format(filename))
ocr_temp_dirname = tempfile.mkdtemp(prefix="opensemanticetl_pdf_ocr_")
# Extract all images of the pdf to tempdir with commandline tool
# "pdfimages" from poppler pdf toolbox
# -j = export as JPEG
# -p = write page name in image filename
result = subprocess.call([
'pdfimages', '-p', '-j', filename,
ocr_temp_dirname + os.path.sep + 'image'
])
if result != 0:
sys.stderr.write(
"Error: Extracting images from PDF failed for {} {}".format(
filename, result))
return {}, {}
images = os.listdir(ocr_temp_dirname)
images.sort()
for image in images:
imagefilename = ocr_temp_dirname + os.path.sep + image
if pdf_ocr:
try:
result = enhance_ocr.image2text(filename=imagefilename,
lang=lang,
verbose=verbose)
if result:
# extract page number from extracted image
# filename (image-pagenumber-imagenumber.jpg)
pagenumber = int(image.split('-')[1])
append_page(ocr_txt, pagenumber, result)
except BaseException as e:
sys.stderr.write(
"Exception while OCR of PDF: {} - "
"maybe corrupt image: {} - exception: {}\n".format(
filename, imagefilename, e))
if pdf_ocr_descew:
try:
result = enhance_ocr_descew.optimized_image2text(
imagefilename, lang, verbose=verbose)
if result:
# extract page number from extracted image
# filename (image-pagenumber-imagenumber.jpg)
pagenumber = int(image.split('-')[1])
append_page(enhance_ocr_descew, pagenumber, result)
except BaseException as e:
sys.stderr.write(
"Exception while optimized ocr pdf: {} - "
"maybe corrupt image: {} - exception: {}\n".format(
filename, imagefilename, e))
os.remove(imagefilename)
os.rmdir(ocr_temp_dirname)
return ocr_txt, ocr_descew_txt
