qurator-spk
/
modstool
mirror of https://github.com/qurator-spk/modstool.git
1
0
Fork 0
Code Issues Releases Wiki Activity

✨ ALTO: preProcessingStep/processingAgency/sourceImageInformation etc.

Browse Source
master
Gerber, Mike 3 years ago
parent 01326050d3
commit c91c9b1714
1 changed files with 16 additions and 0 deletions
Show Stats Download Patch File Download Diff File

16
qurator/modstool/altotool.py
Unescape Escape View File

@ -42,16 +42,32 @@ def alto_to_dict(alto, raise_errors=True):
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}ocrProcessingStep': elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}ocrProcessingStep':
for n, e in enumerate(group): for n, e in enumerate(group):
value['ocrProcessingStep{}'.format(n)] = alto_to_dict(e, raise_errors) value['ocrProcessingStep{}'.format(n)] = alto_to_dict(e, raise_errors)
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}preProcessingStep':
# TODO This enumerated descent is used more than once, DRY!
for n, e in enumerate(group):
value['preProcessingStep{}'.format(n)] = alto_to_dict(e, raise_errors)
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}processingDateTime': elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}processingDateTime':
value['processingDateTime'] = TagGroup(tag, group).is_singleton().has_no_attributes().text() value['processingDateTime'] = TagGroup(tag, group).is_singleton().has_no_attributes().text()
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}processingSoftware': elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}processingSoftware':
value['processingSoftware'] = TagGroup(tag, group).is_singleton().descend(raise_errors) value['processingSoftware'] = TagGroup(tag, group).is_singleton().descend(raise_errors)
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}processingAgency':
value['processingAgency'] = TagGroup(tag, group).is_singleton().has_no_attributes().text()
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}processingStepDescription':
value['processingStepDescription'] = TagGroup(tag, group).is_singleton().has_no_attributes().text()
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}processingStepSettings':
value['processingStepSettings'] = TagGroup(tag, group).is_singleton().has_no_attributes().text()
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}softwareCreator': elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}softwareCreator':
value['softwareCreator'] = TagGroup(tag, group).is_singleton().has_no_attributes().text() value['softwareCreator'] = TagGroup(tag, group).is_singleton().has_no_attributes().text()
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}softwareName': elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}softwareName':
value['softwareName'] = TagGroup(tag, group).is_singleton().has_no_attributes().text() value['softwareName'] = TagGroup(tag, group).is_singleton().has_no_attributes().text()
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}softwareVersion': elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}softwareVersion':
value['softwareVersion'] = TagGroup(tag, group).is_singleton().has_no_attributes().text() value['softwareVersion'] = TagGroup(tag, group).is_singleton().has_no_attributes().text()
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}sourceImageInformation':
value['sourceImageInformation'] = TagGroup(tag, group).is_singleton().has_no_attributes().descend(raise_errors)
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}fileName':
value['fileName'] = TagGroup(tag, group).is_singleton().has_no_attributes().text()
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}Layout': elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}Layout':
value['Layout'] = TagGroup(tag, group).is_singleton().has_no_attributes().descend(raise_errors) value['Layout'] = TagGroup(tag, group).is_singleton().has_no_attributes().descend(raise_errors)
elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}Page': elif tag == '{http://www.loc.gov/standards/alto/ns-v2#}Page':

Write Preview
Loading…
Cancel
Save