def docTypeExists(dict_tree,doc_type):
log_struct_map = log_struct_map_tools.getLogicalStructMap(dict_tree)
ns = '{http://www.loc.gov/METS/}'
elem_name = 'div'
elem_attrib_key = 'TYPE'
elem_attrib_val = doc_type
dmd_sec = xml_tools.getAllSubTrees(log_struct_map,
ns,
elem_name,
elem_attrib_key,
elem_attrib_val)
#pprint.pprint(dmd_sec)
return len(dmd_sec) > 0
def getDmdsWithContent(dict_tree, doc_type, metadata_key, content):
'''
Get a list of logical IDs (LOG_NNNN) for dmd_secs with TYPE = doc_type and
metadata_key = content.
E.g. a list of all articles where 'TitleDocMain' = content
:param dict_tree:
:param doc_type: type of dmd_sec to look in, e.g. 'Article'
:param metadata_key: metadata key to look in, e.g. 'TitleDocMain'
:param content: metadata content to check for, e.g. the name of the article
'''
ns = '{http://www.loc.gov/METS/}'
mets = dict_tree[ns+'mets'] if ns+'mets' in dict_tree else {}
elem_name = 'div'
elem_attrib_key = 'TYPE'
elem_attrib_val = doc_type
log_struct_maps = xml_tools.getAllSubTrees(mets,
ns,
elem_name,
elem_attrib_key,
elem_attrib_val)
dmd_sec_ids = dict([(x['@DMDID'],x['@ID']) for x in log_struct_maps if '@DMDID' in x])
goobi_ns = '{http://meta.goobi.org/v1.5.1/}'
# Take list of dmd_secs
dmd_secs = mets[ns+'dmdSec'] if ns+'dmdSec' in mets else []
named_dmd_secs = []
for dmd_sec in dmd_secs:
if '@ID' in dmd_sec and dmd_sec['@ID'] in dmd_sec_ids:
# This is an article
dmd_sec_metadata = xml_tools.getAllSubTrees(dmd_sec,goobi_ns,
'dmdSec','name',
metadata_key)
if [a for a in dmd_sec_metadata
if '#text' in a and a['#text'] == content]:
# This dmd_sec is a article and has the name "metadata_key"
# Get LOG-id for dmd, as this is used outsude
named_dmd_secs.append(dmd_sec_ids[dmd_sec['@ID']])
#return True # The article has the title 'article_title'
return named_dmd_secs
def getMetadata(dmd_secs):
'''
Returns all the metadata from dmd_secs from a list of dmd_secs. If a
metadata field exist more than once, only one will be outputted
:param dmd_secs: a list of dmd_secs
'''
metadata = []
goobi_ns = '{http://meta.goobi.org/v1.5.1/}'
metadata_key = 'metadata'
for dmd_sec in dmd_secs:
temp_metadata = xml_tools.getAllSubTrees(dmd_sec,
goobi_ns,
metadata_key)
if temp_metadata and isinstance(temp_metadata[0],list):
if len(temp_metadata) > 1: raise ValueError('To many metadata lists')
metadata.extend(temp_metadata[0])
else: metadata.extend(temp_metadata)
return metadata