qurator-spk/modstool
1
0
Fork 0
mirror of https://github.com/qurator-spk/modstool.git synced 2025-06-10 04:09:56 +02:00
Code Issues Releases Wiki Activity

⚡ Make gettstruct_log faster by using precise predicates

Browse source
This commit is contained in:
Gerber, Mike 2023-12-09 11:35:24 +01:00
parent 1dac77a2f5
commit 448639b05b
1 changed files with 9 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

12
qurator/mods4pandas/mods4pandas.py
View file

@ -266,9 +266,12 @@ def pages_to_dict(mets, raise_errors=True) -> List[Dict]:
ppn = get_mets_recordIdentifier() ppn = get_mets_recordIdentifier()
# Getting per-page/structure information is a bit different # Getting per-page/structure information is a bit different
structMap_PHYSICAL = (mets.xpath('//mets:structMap[@TYPE="PHYSICAL"]', namespaces=ns) or [None])[0] structMap_PHYSICAL = mets.find('./mets:structMap[@TYPE="PHYSICAL"]', ns)
if not structMap_PHYSICAL: structMap_LOGICAL = mets.find('./mets:structMap[@TYPE="LOGICAL"]', ns)
if structMap_PHYSICAL is None:
raise ValueError("No structMap[@TYPE='PHYSICAL'] found") raise ValueError("No structMap[@TYPE='PHYSICAL'] found")
if structMap_LOGICAL is None:
raise ValueError("No structMap[@TYPE='LOGICAL'] found")
div_physSequence = structMap_PHYSICAL[0] div_physSequence = structMap_PHYSICAL[0]
assert div_physSequence.attrib.get("TYPE") == "physSequence" assert div_physSequence.attrib.get("TYPE") == "physSequence"
@ -278,6 +281,9 @@ def pages_to_dict(mets, raise_errors=True) -> List[Dict]:
file_ = mets.find(f'.//{{{ns["mets"]}}}file[@ID="{ID}"]') file_ = mets.find(f'.//{{{ns["mets"]}}}file[@ID="{ID}"]')
return file_ return file_
def get_mets_div(*, ID):
if ID:
return structMap_LOGICAL.findall(f'.//mets:div[@ID="{ID}"]', ns)
for page in div_physSequence: for page in div_physSequence:
@ -315,7 +321,7 @@ def pages_to_dict(mets, raise_errors=True) -> List[Dict]:
targets = [] targets = []
for sm_link in sm_links: for sm_link in sm_links:
xlink_from = sm_link.attrib.get(f"{{{ns['xlink']}}}from") xlink_from = sm_link.attrib.get(f"{{{ns['xlink']}}}from")
targets.extend(mets.findall(f'//mets:div[@ID="{xlink_from}"]', ns)) targets.extend(get_mets_div(ID=xlink_from))
return targets return targets
struct_divs = set(get_struct_log(to_phys=page_dict["ID"])) struct_divs = set(get_struct_log(to_phys=page_dict["ID"]))