You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
ocrd-galley/extra/ppn2ocr

218 lines
6.1 KiB
Python

#!/usr/bin/env python3
"""Get OCR results as a OCR-D workspace for a given PPN"""
import os
import requests
import sys
import lxml.etree as ET
import re
import subprocess
import click
from copy import deepcopy
XMLNS = {
'mets': 'http://www.loc.gov/METS/',
'xlink': 'http://www.w3.org/1999/xlink'
}
API_URL = 'https://oai.sbb.berlin'
IDENTIFIER_TEMPLATE = 'oai:digital.staatsbibliothek-berlin.de:%s'
for prefix, uri in XMLNS.items():
ET.register_namespace(prefix, uri)
def oai_mets(ppn):
"""Retrieve METS metadata for a given PPN."""
params = {
'verb': 'GetRecord',
'metadataPrefix': 'mets',
'identifier': IDENTIFIER_TEMPLATE % ppn
}
s = requests.Session()
r = s.get(API_URL, params=params)
mets = ET.XML(r.content).find(f".//{{{XMLNS['mets']}}}mets")
mets = ET.ElementTree(mets)
return mets
def iiif_url_for_sbb_url(sbb_url, ppn, size, format):
"""
Construct an IIIF URL from a dms or an IIIF URL.
This function exists as long as dms URL exist (or as long as we may need to
rewrite IIIF URLs for a different size)
"""
if "/dms/" in sbb_url:
return iiif_url_for_dms_url(sbb_url, ppn, size, format)
else:
return iiif_url_for_iiif_url(sbb_url, ppn, size, format)
def iiif_url_for_dms_url(dms_url, ppn, size, format):
"""
Construct an IIIF URL from a dms URL.
This function exists to encapsulate the hack of rewriting the URL to get IIIF.
"""
if ppn not in dms_url:
raise ValueError(f"Unexpected URL {dms_url}")
m = re.search(r'/dms/.*/([0-9]+)\.jpg$', dms_url)
if m:
page_num = m.group(1)
else:
raise ValueError(f"Unexpected URL {dms_url}")
iiif_identifier = f'{ppn}-{page_num}'
iiif_quality = 'default'
iiif_url = f'https://content.staatsbibliothek-berlin.de/dc/{iiif_identifier}/full/{size}/0/{iiif_quality}.{format}'
return iiif_url
def iiif_url_for_iiif_url(iiif_url, ppn, size, format):
"""
Construct an IIIF URL from an already existing IIIF URL.
"""
if ppn not in iiif_url:
raise ValueError(f"Unexpected URL {iiif_url}")
m = re.search(rf'/dc/{ppn}-([0-9]+)/', iiif_url)
if m:
page_num = m.group(1)
else:
raise ValueError(f"Unexpected URL {iiif_url}")
iiif_identifier = f'{ppn}-{page_num}'
iiif_quality = 'default'
iiif_url = f'https://content.staatsbibliothek-berlin.de/dc/{iiif_identifier}/full/{size}/0/{iiif_quality}.{format}'
return iiif_url
def remove_file_grp(mets, use):
for bad_fileid in mets.xpath(f'//mets:fileGrp[@USE="{use}"]/mets:file/@ID', namespaces=XMLNS):
for bad in mets.xpath(f'//mets:fptr[@FILEID="{bad_fileid}"]', namespaces=XMLNS):
bad.getparent().remove(bad)
for bad in mets.xpath(f'//mets:fileGrp[@USE="{use}"]', namespaces=XMLNS):
bad.getparent().remove(bad)
def mime_type_for_format(format_):
if format_ == 'tif':
mime_type = 'image/tiff'
elif format_ == 'jpg':
mime_type = 'image/jpg'
else:
raise ValueError()
return mime_type
def prune_file_grps(mets):
"""
Prune unwanted file groups
We only want to keep the MAX file group (we created it ourselves) and
possibly ABBYY full texts in FULLTEXT.
For the PRESENTATION + LOCAL file groups we definitely want to delete
because they contain local file:/// or file:/ links, which are not handled
well by "ocrd workspace". They are not explicitly mentioned, as we
only keep a whitelist.
"""
wanted_file_grps = ["MAX", "FULLTEXT"]
for u in mets.xpath('//mets:fileGrp/@USE', namespaces=XMLNS):
if u not in wanted_file_grps:
remove_file_grp(mets, u)
def make_workspace(ppn, workspace):
# Make workspace directory
os.mkdir(workspace)
os.chdir(workspace)
mets = oai_mets(ppn)
# Delete MAX file group - we assume that, if it exists, it is not as
# we expect it, e.g. IIIF full URLs
remove_file_grp(mets, 'MAX')
# Duplicate DEFAULT file group into a new file group MAX
format_ = 'tif'
file_grp_default = mets.find('//mets:fileGrp[@USE="DEFAULT"]', namespaces=XMLNS)
if file_grp_default is None:
raise ValueError("This document has no DEFAULT file group, could be a multi-volume work")
file_grp_best = deepcopy(file_grp_default)
file_grp_best.attrib['USE'] = 'MAX'
for f in file_grp_best.findall('./mets:file', namespaces=XMLNS):
old_id = f.attrib['ID']
new_id = re.sub('DEFAULT', 'MAX', old_id)
f.attrib['ID'] = new_id
f.attrib['MIMETYPE'] = mime_type_for_format(format_)
for fptr in mets.findall(f'//mets:fptr[@FILEID="{old_id}"]', namespaces=XMLNS):
new_fptr = deepcopy(fptr)
new_fptr.attrib['FILEID'] = new_id
fptr.getparent().append(new_fptr)
# XXX Need to fumble around with the URL for now
flocat = f.find(f".//{{{XMLNS['mets']}}}FLocat")
old_url = flocat.attrib[f"{{{XMLNS['xlink']}}}href"]
url_iiif_full = iiif_url_for_sbb_url(old_url, ppn, 'full', format_)
flocat.attrib[f"{{{XMLNS['xlink']}}}href"] = url_iiif_full
mets.find('//mets:fileSec', namespaces=XMLNS).append(file_grp_best)
prune_file_grps(mets)
# Write mets.xml
mets.write('mets.xml', pretty_print=True)
# TODO
# Validate workspace
#ocrd workspace validate mets.xml | grep -v "<notice>Won't download remote image"
def validate_ppn(ctx, param, value):
"""Validate a PPN argument"""
if not value.startswith('PPN'):
raise click.BadParameter('PPN must be in format PPNxxxxxxxx')
else:
return value
@click.command()
@click.argument('ppn', callback=validate_ppn)
def ppn2ocr(ppn):
"""
Get METS with best images for a document PPN
For example, to get the document "PROPOSITIONES PHILOSOPHICAE: [...]" use this:
\b
ppn2ocr PPN699887615
ls PPN699887615
"""
self_dir = os.path.realpath(os.path.dirname(sys.argv[0]))
make_workspace(ppn, ppn)
# XXX
# subprocess.run([
# os.path.join(self_dir, 'run-docker-hub'),
# '-I', 'MAX',
# '--skip-validation'
# ])
if __name__ == '__main__':
ppn2ocr()