def get_record_provenance(recid):
"""
Return the provenance XML representation of a record, suitable to be put
in the about tag.
"""
record = get_record(recid)
provenances = record_get_field_instances(
record, CFG_BIBUPLOAD_EXTERNAL_OAIID_TAG[:3],
CFG_BIBUPLOAD_EXTERNAL_OAIID_TAG[3],
CFG_BIBUPLOAD_EXTERNAL_OAIID_TAG[4])
out = ""
for provenance in provenances:
base_url = identifier = datestamp = metadata_namespace = origin_description = harvest_date = altered = ""
for (code, value) in provenance[0]:
if code == CFG_OAI_PROVENANCE_BASEURL_SUBFIELD:
base_url = value
elif code == CFG_BIBUPLOAD_EXTERNAL_OAIID_TAG[5]:
identifier = value
elif code == CFG_OAI_PROVENANCE_DATESTAMP_SUBFIELD:
datestamp = value
elif code == CFG_OAI_PROVENANCE_METADATANAMESPACE_SUBFIELD:
metadata_namespace = value
elif code == CFG_OAI_PROVENANCE_ORIGINDESCRIPTION_SUBFIELD:
origin_description = value
elif code == CFG_OAI_PROVENANCE_HARVESTDATE_SUBFIELD:
harvest_date = value
elif code == CFG_OAI_PROVENANCE_ALTERED_SUBFIELD:
altered = value
if base_url:
out += """<provenance xmlns="http://www.openarchives.org/OAI/2.0/provenance" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/provenance http://www.openarchives.org/OAI/2.0/provenance.xsd">"""
out += X.originDescription(
harvestDate=harvest_date, altered=altered)(
X.baseURL()(base_url),
X.identifier()(identifier),
X.datestamp()(datestamp),
X.metadataNamespace()(metadata_namespace),
origin_description
and X.originDescription(origin_description)
or '' ## This is already XML
)
out += """</provenance>"""
return out
def get_record_provenance(recid):
"""
Return the provenance XML representation of a record, suitable to be put
in the about tag.
"""
record = get_record(recid)
provenances = record_get_field_instances(record, CFG_BIBUPLOAD_EXTERNAL_OAIID_TAG[:3], CFG_BIBUPLOAD_EXTERNAL_OAIID_TAG[3], CFG_BIBUPLOAD_EXTERNAL_OAIID_TAG[4])
out = ""
for provenance in provenances:
base_url = identifier = datestamp = metadata_namespace = origin_description = harvest_date = altered = ""
for (code, value) in provenance[0]:
if code == CFG_OAI_PROVENANCE_BASEURL_SUBFIELD:
base_url = value
elif code == CFG_BIBUPLOAD_EXTERNAL_OAIID_TAG[5]:
identifier = value
elif code == CFG_OAI_PROVENANCE_DATESTAMP_SUBFIELD:
datestamp = value
elif code == CFG_OAI_PROVENANCE_METADATANAMESPACE_SUBFIELD:
metadata_namespace = value
elif code == CFG_OAI_PROVENANCE_ORIGINDESCRIPTION_SUBFIELD:
origin_description = value
elif code == CFG_OAI_PROVENANCE_HARVESTDATE_SUBFIELD:
harvest_date = value
elif code == CFG_OAI_PROVENANCE_ALTERED_SUBFIELD:
altered = value
if base_url:
out += """<provenance xmlns="http://www.openarchives.org/OAI/2.0/provenance" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/provenance http://www.openarchives.org/OAI/2.0/provenance.xsd">"""
out += X.originDescription(harvestDate=harvest_date, altered=altered)(
X.baseURL()(base_url),
X.identifier()(identifier),
X.datestamp()(datestamp),
X.metadataNamespace()(metadata_namespace),
origin_description and X.originDescription(origin_description) or '' ## This is already XML
)
out += """</provenance>"""
return out
def print_record(recid, prefix='marcxml', verb='ListRecords', set_spec=None, set_last_updated=None):
"""Prints record 'recid' formatted according to 'prefix'.
- if record does not exist, return nothing.
- if record has been deleted and CFG_OAI_DELETED_POLICY is
'transient' or 'deleted', then return only header, with status
'deleted'.
- if record has been deleted and CFG_OAI_DELETED_POLICY is 'no',
then return nothing.
"""
record_exists_result = record_exists(recid) == 1
if record_exists_result:
sets = get_field(recid, CFG_OAI_SET_FIELD)
if set_spec is not None and not set_spec in sets and not [set_ for set_ in sets if set_.startswith("%s:" % set_spec)]:
## the record is not in the requested set, and is not
## in any subset
record_exists_result = False
if record_exists_result:
status = None
else:
status = 'deleted'
if not record_exists_result and CFG_OAI_DELETED_POLICY not in ('persistent', 'transient'):
return ""
idents = get_field(recid, CFG_OAI_ID_FIELD)
if not idents:
return ""
## FIXME: Move these checks in a bibtask
#try:
#assert idents, "No OAI ID for record %s, please do your checks!" % recid
#except AssertionError, err:
#register_exception(alert_admin=True)
#return ""
#try:
#assert len(idents) == 1, "More than OAI ID found for recid %s. Considering only the first one, but please do your checks: %s" % (recid, idents)
#except AssertionError, err:
#register_exception(alert_admin=True)
ident = idents[0]
header_body = EscapedXMLString('')
header_body += X.identifier()(ident)
if set_last_updated:
header_body += X.datestamp()(max(get_modification_date(recid), set_last_updated))
else:
header_body += X.datestamp()(get_modification_date(recid))
for set_spec in get_field(recid, CFG_OAI_SET_FIELD):
if set_spec and set_spec != CFG_OAI_REPOSITORY_GLOBAL_SET_SPEC:
# Print only if field not empty
header_body += X.setSpec()(set_spec)
header = X.header(status=status)(header_body)
if verb == 'ListIdentifiers':
return header
else:
if record_exists_result:
metadata_body = format_record(recid, CFG_OAI_METADATA_FORMATS[prefix][0])
metadata = X.metadata(body=metadata_body)
provenance_body = get_record_provenance(recid)
if provenance_body:
provenance = X.about(body=provenance_body)
else:
provenance = ''
rights_body = get_record_rights(recid)
if rights_body:
rights = X.about(body=rights_body)
else:
rights = ''
else:
metadata = ''
provenance = ''
rights = ''
return X.record()(header, metadata, provenance, rights)
try:
assert idents, "No OAI ID for record %s, please do your checks!" % recid
except AssertionError, err:
register_exception(alert_admin=True)
return
try:
assert len(
idents
) == 1, "More than OAI ID found for recid %s. Considering only the first one, but please do your checks: %s" % (
recid, idents)
except AssertionError, err:
register_exception(alert_admin=True)
ident = idents[0]
header_body = EscapedXMLString('')
header_body += X.identifier()(ident)
header_body += X.datestamp()(get_modification_date(recid))
for set_spec in get_field(recid, CFG_OAI_SET_FIELD):
if set_spec and set_spec != CFG_OAI_REPOSITORY_GLOBAL_SET_SPEC:
# Print only if field not empty
header_body += X.setSpec()(set_spec)
header = X.header(status=status)(header_body)
if verb == 'ListIdentifiers':
return header
else:
if record_exists_result:
metadata_body = format_record(recid,
CFG_OAI_METADATA_FORMATS[prefix][0])
metadata = X.metadata(body=metadata_body)
def print_record(recid,
prefix='marcxml',
verb='ListRecords',
set_spec=None,
set_last_updated=None):
"""Prints record 'recid' formatted according to 'prefix'.
- if record does not exist, return nothing.
- if record has been deleted and CFG_OAI_DELETED_POLICY is
'transient' or 'deleted', then return only header, with status
'deleted'.
- if record has been deleted and CFG_OAI_DELETED_POLICY is 'no',
then return nothing.
"""
record_exists_result = record_exists(recid) == 1
if record_exists_result:
sets = get_field(recid, CFG_OAI_SET_FIELD)
if set_spec is not None and not set_spec in sets and not [
set_ for set_ in sets if set_.startswith("%s:" % set_spec)
]:
## the record is not in the requested set, and is not
## in any subset
record_exists_result = False
if record_exists_result:
status = None
else:
status = 'deleted'
if not record_exists_result and CFG_OAI_DELETED_POLICY not in (
'persistent', 'transient'):
return ""
idents = get_field(recid, CFG_OAI_ID_FIELD)
if not idents:
return ""
## FIXME: Move these checks in a bibtask
#try:
#assert idents, "No OAI ID for record %s, please do your checks!" % recid
#except AssertionError, err:
#register_exception(alert_admin=True)
#return ""
#try:
#assert len(idents) == 1, "More than OAI ID found for recid %s. Considering only the first one, but please do your checks: %s" % (recid, idents)
#except AssertionError, err:
#register_exception(alert_admin=True)
ident = idents[0]
header_body = EscapedXMLString('')
header_body += X.identifier()(ident)
if set_last_updated:
header_body += X.datestamp()(max(get_modification_date(recid),
set_last_updated))
else:
header_body += X.datestamp()(get_modification_date(recid))
for set_spec in get_field(recid, CFG_OAI_SET_FIELD):
if set_spec and set_spec != CFG_OAI_REPOSITORY_GLOBAL_SET_SPEC:
# Print only if field not empty
header_body += X.setSpec()(set_spec)
header = X.header(status=status)(header_body)
if verb == 'ListIdentifiers':
return header
else:
if record_exists_result:
metadata_body = format_record(recid,
CFG_OAI_METADATA_FORMATS[prefix][0])
metadata = X.metadata(body=metadata_body)
provenance_body = get_record_provenance(recid)
if provenance_body:
provenance = X.about(body=provenance_body)
else:
provenance = ''
rights_body = get_record_rights(recid)
if rights_body:
rights = X.about(body=rights_body)
else:
rights = ''
else:
metadata = ''
provenance = ''
rights = ''
return X.record()(header, metadata, provenance, rights)
return
idents = get_field(recid, CFG_OAI_ID_FIELD)
try:
assert idents, "No OAI ID for record %s, please do your checks!" % recid
except AssertionError, err:
register_exception(alert_admin=True)
return
try:
assert len(idents) == 1, "More than OAI ID found for recid %s. Considering only the first one, but please do your checks: %s" % (recid, idents)
except AssertionError, err:
register_exception(alert_admin=True)
ident = idents[0]
header_body = EscapedXMLString('')
header_body += X.identifier()(ident)
header_body += X.datestamp()(get_modification_date(recid))
for set_spec in get_field(recid, CFG_OAI_SET_FIELD):
if set_spec and set_spec != CFG_OAI_REPOSITORY_GLOBAL_SET_SPEC:
# Print only if field not empty
header_body += X.setSpec()(set_spec)
header = X.header(status=status)(header_body)
if verb == 'ListIdentifiers':
return header
else:
if record_exists_result:
metadata_body = format_record(recid, CFG_OAI_METADATA_FORMATS[prefix][0])
metadata = X.metadata(body=metadata_body)
provenance_body = get_record_provenance(recid)