def ocr(args):
"""Performs optical character recognition (OCR) on given image. The image
can be defined either as path to a file or as the primary key of a Page
object. The latter case also enables saving the resulting text to the
object. By default the result is simply printed.
"""
try:
assert args.pk or args.path, '--path or --pk must be provided'
assert not (args.pk and args.path), 'specify either --path or --pk'
assert not (args.path and args.save), 'use --save only with --pk'
except AssertionError as e:
raise CommandError(e)
if args.pk:
db = app.get_feature('document_storage').default_db
page = Page.object(db, args.pk)
path = page.image.full_path
else:
path = args.path
text = image_to_text(path, language=args.language)
if args.save:
page.details = text
page.save()
yield 'Page saved with the new text.'
else:
yield text
def page_index(request):
db = app.get_feature('document_storage').default_db
pages = Page.objects(db).order_by('date_time')
for p in pages:
print p.pk, p, p.summary, p.image
return {'pages': pages,
'thumbnail': _get_thumbnail}
def add_pages(args):
_args_to_unicode(args, ['language', 'summary', 'summary_prefix'])
db = app.get_feature('document_storage').default_db
# check if the files exist
for path in args.paths:
assert os.path.exists(path)
# import
for path in args.paths:
yield '* importing {0} (language {1})'.format(path, args.language)
fingerprint = get_file_hash(open(path, 'rb'))
# check file hash uniqueness
if Page.objects(db).where(source_fingerprint=fingerprint):
yield '...already in the database.'
continue
p = Page()
p.summary = args.summary or get_summary_from_path(path)
if args.summary_prefix:
p.summary = u'{0} {1}'.format(args.summary_prefix, p.summary)
p.language = args.language or None
p.source_fingerprint = fingerprint
if not args.no_ocr:
try:
p.details = image_to_text(path=path, language=p.language)
except RuntimeError as e:
if not args.skip_ocr_errors:
raise CommandError(e)
yield '(OCR failed, saving only image itself)'
# usually we don't need heavy formats like ppm or tiff even for OCR
img = Image.open(path)
if args.format:
fmt = args.format
elif img.format not in IMAGE_FORMATS:
fmt = IMAGE_FORMATS[0]
else:
fmt = img.format
img.save(TMP_FILENAME, fmt)
p['image'] = open(TMP_FILENAME, 'rb')
# provide original path so that the resulting filename is inherited
p['image'].path = path
p.save(db)