mirror of
https://github.com/qurator-spk/ocrd-galley.git
synced 2025-06-09 14:49:53 +02:00
🚧 ppn2ocr: Properly remove the PRESENTATION file group
This commit is contained in:
parent
014e70fe35
commit
f893b339c5
1 changed files with 9 additions and 2 deletions
11
ppn2ocr
11
ppn2ocr
|
@ -70,6 +70,14 @@ def iiif_url_for_dms_url(dms_url, ppn, size):
|
|||
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 make_workspace(ppn, workspace):
|
||||
# Make workspace directory
|
||||
os.mkdir(workspace)
|
||||
|
@ -80,8 +88,7 @@ def make_workspace(ppn, workspace):
|
|||
# XXX
|
||||
# Delete PRESENTATION file group
|
||||
# (local file:/// links, not handled well by "ocrd workspace")
|
||||
for bad in mets.xpath('//mets:fileGrp[@USE="PRESENTATION"]', namespaces=XMLNS):
|
||||
bad.getparent().remove(bad)
|
||||
remove_file_grp(mets, 'PRESENTATION')
|
||||
|
||||
|
||||
# Duplicate DEFAULT file group into a new file group BEST
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue