def format_element(bfo,
name,
tag_name='',
tag='',
kb='',
kb_default_output='',
var='',
protocol='googlescholar'):
"""Prints a custom field in a way suitable to be used in HTML META
tags. In particular conforms to Google Scholar harvesting protocol as
defined http://scholar.google.com/intl/en/scholar/inclusion.html and
Open Graph http://ogp.me/
@param tag_name: the name, from tag table, of the field to be exported
looks initially for names prefixed by "meta-"<tag_name>
then looks for exact name, then falls through to "tag"
@param tag: the MARC tag to be exported (only if not defined by tag_name)
@param name: name to be displayed in the meta headers, labelling this value.
@param kb: a knowledge base through which to process the retrieved value if necessary.
@param kb: when a '<code>kb</code>' is specified and no match for value is found, what shall we
return? Either return the given parameter or specify "{value}" to return the retrieved
value before processing though kb.
@param var: the name of a variable to output instead of field from metadata.
Allowed values are those supported by bfe_server_info and
bfe_client_info. Overrides <code>name</code> and <code>tag_name</code>
@param protocol: the protocol this tag is aimed at. Can be used to switch on/off support for a given "protocol". Can take values among 'googlescholar', 'opengraph'
@see: bfe_server_info.py, bfe_client_info.py
"""
if protocol == 'googlescholar' and not CFG_WEBSEARCH_ENABLE_GOOGLESCHOLAR:
return ""
elif protocol == 'opengraph' and not CFG_WEBSEARCH_ENABLE_OPENGRAPH:
return ""
tags = []
if var:
# delegate to bfe_server_info or bfe_client_info:
value = server_info(bfo, var)
if value.startswith("Unknown variable: "):
# Oops variable was not defined there
value = client_info(bfo, var)
return not value.startswith("Unknown variable: ") and \
create_metatag(name=name, content=cgi.escape(value, True)) \
or ""
elif tag_name:
# First check for special meta named tags
tags = get_field_tags("meta-" + tag_name)
if not tags:
# then check for regular tags
tags = get_field_tags(tag_name)
if not tags and tag:
# fall back to explicit marc tag
tags = [tag]
if not tags:
return ''
out = []
values = [bfo.fields(marctag, escape=9) for marctag in tags]
for value in values:
if isinstance(value, list):
for val in value:
if isinstance(val, dict):
out.extend(val.values())
else:
out.append(val)
elif isinstance(value, dict):
out.extend(value.values())
else:
out.append(value)
out = dict(zip(out, len(out) * [''])).keys() # Remove duplicates
if name == 'citation_date':
for idx in range(len(out)):
out[idx] = out[idx].replace('-', '/')
if kb:
if kb_default_output == "{value}":
out = [bfo.kb(kb, value, value) for value in out]
else:
out = [bfo.kb(kb, value, kb_default_output) for value in out]
return '\n'.join(
[create_metatag(name=name, content=value) for value in out])
def format_element(bfo,
name,
tag_name='',
tag='',
kb='',
kb_default_output='',
var='',
protocol='googlescholar'):
"""Prints a custom field in a way suitable to be used in HTML META
tags. In particular conforms to Google Scholar harvesting protocol as
defined http://scholar.google.com/intl/en/scholar/inclusion.html and
Open Graph http://ogp.me/
@param tag_name: the name, from tag table, of the field to be exported
looks initially for names prefixed by "meta-"<tag_name>
then looks for exact name, then falls through to "tag"
@param tag: the MARC tag to be exported (only if not defined by tag_name). Comma-separated list of tags.
@param name: name to be displayed in the meta headers, labelling this value.
@param kb: a knowledge base through which to process the retrieved value if necessary.
@param kb: when a '<code>kb</code>' is specified and no match for value is found, what shall we
return? Either return the given parameter or specify "{value}" to return the retrieved
value before processing though kb.
@param var: the name of a variable to output instead of field from metadata.
Allowed values are those supported by bfe_server_info and
bfe_client_info. Overrides <code>name</code> and <code>tag_name</code>
@param protocol: the protocol this tag is aimed at. Can be used to switch on/off support for a given "protocol". Can take values among 'googlescholar', 'opengraph'
@see: bfe_server_info.py, bfe_client_info.py
"""
if protocol == 'googlescholar' and not CFG_WEBSEARCH_ENABLE_GOOGLESCHOLAR:
return ""
elif protocol == 'opengraph' and not CFG_WEBSEARCH_ENABLE_OPENGRAPH:
return ""
matched_by_tag_name_p = False
tags = []
if var:
# delegate to bfe_server_info or bfe_client_info:
value = server_info(bfo, var)
if value.startswith("Unknown variable: "):
# Oops variable was not defined there
value = client_info(bfo, var)
return not value.startswith("Unknown variable: ") and \
create_metatag(name=name, content=cgi.escape(value, True)) \
or ""
elif tag_name:
# First check for special meta named tags
tags = get_field_tags("meta-" + tag_name)
if not tags:
# then check for regular tags
tags = get_field_tags(tag_name)
matched_by_tag_name_p = tags and True or False
if not tags and tag:
# fall back to explicit marc tag
if ',' in tag:
tags = tag.split(',')
else:
tags = [tag]
if not tags:
return ''
out = []
if protocol == 'googlescholar' and \
(tags == ['100__a'] or tags == ['700__a']):
# Authors for Google Scholar: remove names that are not purely
# author (thesis director, coordinator, etc). Assume that
# existence of $e subfield is a sign. Since this assumption
# might be wrong, put some strong conditions in order to get
# into this branch, with easy way to bypass.
values = [
field_instance[tags[0][-1]]
for field_instance in bfo.fields(tags[0][:-1], escape=9)
if 'e' not in field_instance and tags[0][-1] in field_instance
]
else:
# Standard fetching of values
values = [bfo.fields(marctag, escape=9) for marctag in tags]
if name == 'citation_dissertation_institution':
if CFG_CERN_SITE and \
'THESIS' in bfo.fields('980__a'):
authors = bfo.fields('100__', escape=9)
authors.extend(bfo.fields('700__', escape=9))
values = [field_instance['u'] for field_instance in authors \
if not field_instance.has_key('e') and field_instance.has_key('u')]
elif tag == '100__u' and not matched_by_tag_name_p:
# TODO: find way to map correctly this tag
values = []
for value in values:
if isinstance(value, list):
for val in value:
if isinstance(val, dict):
out.extend(val.values())
else:
out.append(val)
elif isinstance(value, dict):
out.extend(value.values())
else:
out.append(value)
if name == 'citation_title':
out = [' : '.join(out)]
if name == 'citation_date':
for idx in range(len(out)):
out[idx] = out[idx].replace('-', '/')
elif name == 'citation_publication_date':
for idx in range(len(out)):
# Stop at first match
parsed_date = parse_date_for_googlescholar(out[idx])
if parsed_date:
out = [parsed_date]
break
out = dict(zip(out, len(out) * [''])).keys() # Remove duplicates
if kb:
if kb_default_output == "{value}":
out = [bfo.kb(kb, value, value) for value in out]
else:
out = [bfo.kb(kb, value, kb_default_output) for value in out]
return '\n'.join(
[create_metatag(name=name, content=value) for value in out])
def format_element(bfo, name, tag_name='', tag='', kb='', kb_default_output='', var='', protocol='googlescholar'):
"""Prints a custom field in a way suitable to be used in HTML META
tags. In particular conforms to Google Scholar harvesting protocol as
defined http://scholar.google.com/intl/en/scholar/inclusion.html and
Open Graph http://ogp.me/
@param tag_name: the name, from tag table, of the field to be exported
looks initially for names prefixed by "meta-"<tag_name>
then looks for exact name, then falls through to "tag"
@param tag: the MARC tag to be exported (only if not defined by tag_name). Comma-separated list of tags.
@param name: name to be displayed in the meta headers, labelling this value.
@param kb: a knowledge base through which to process the retrieved value if necessary.
@param kb: when a '<code>kb</code>' is specified and no match for value is found, what shall we
return? Either return the given parameter or specify "{value}" to return the retrieved
value before processing though kb.
@param var: the name of a variable to output instead of field from metadata.
Allowed values are those supported by bfe_server_info and
bfe_client_info. Overrides <code>name</code> and <code>tag_name</code>
@param protocol: the protocol this tag is aimed at. Can be used to switch on/off support for a given "protocol". Can take values among 'googlescholar', 'opengraph'
@see: bfe_server_info.py, bfe_client_info.py
"""
if protocol == 'googlescholar' and not CFG_WEBSEARCH_ENABLE_GOOGLESCHOLAR:
return ""
elif protocol == 'opengraph' and not CFG_WEBSEARCH_ENABLE_OPENGRAPH:
return ""
matched_by_tag_name_p = False
tags = []
if var:
# delegate to bfe_server_info or bfe_client_info:
value = server_info(bfo, var)
if value.startswith("Unknown variable: "):
# Oops variable was not defined there
value = client_info(bfo, var)
return not value.startswith("Unknown variable: ") and \
create_metatag(name=name, content=cgi.escape(value, True)) \
or ""
elif tag_name:
# First check for special meta named tags
tags = get_field_tags("meta-" + tag_name)
if not tags:
# then check for regular tags
tags = get_field_tags(tag_name)
matched_by_tag_name_p = tags and True or False
if not tags and tag:
# fall back to explicit marc tag
if ',' in tag:
tags = tag.split(',')
else:
tags = [tag]
if not tags:
return ''
out = []
if protocol == 'googlescholar' and \
(tags == ['100__a'] or tags == ['700__a']):
# Authors for Google Scholar: remove names that are not purely
# author (thesis director, coordinator, etc). Assume that
# existence of $e subfield is a sign. Since this assumption
# might be wrong, put some strong conditions in order to get
# into this branch, with easy way to bypass.
values = [field_instance[tags[0][-1]] for field_instance in bfo.fields(tags[0][:-1], escape=9) \
if not field_instance.has_key('e')]
else:
# Standard fetching of values
values = [bfo.fields(marctag, escape=9) for marctag in tags]
if name == 'citation_dissertation_institution':
if CFG_CERN_SITE and \
'THESIS' in bfo.fields('980__a'):
authors = bfo.fields('100__', escape=9)
authors.extend(bfo.fields('700__', escape=9))
values = [field_instance['u'] for field_instance in authors \
if not field_instance.has_key('e') and field_instance.has_key('u')]
elif tag == '100__u' and not matched_by_tag_name_p:
# TODO: find way to map correctly this tag
values = []
for value in values:
if isinstance(value, list):
for val in value:
if isinstance(val, dict):
out.extend(val.values())
else:
out.append(val)
elif isinstance(value, dict):
out.extend(value.values())
else:
out.append(value)
if name == 'citation_date':
for idx in range(len(out)):
out[idx] = out[idx].replace('-', '/')
elif name == 'citation_publication_date':
for idx in range(len(out)):
# Stop at first match
parsed_date = parse_date_for_googlescholar(out[idx])
if parsed_date:
out = [parsed_date]
break
out = dict(zip(out, len(out)*[''])).keys() # Remove duplicates
if kb:
if kb_default_output == "{value}":
out = [bfo.kb(kb, value, value) for value in out]
else:
out = [bfo.kb(kb, value, kb_default_output) for value in out]
return '\n'.join([create_metatag(name=name, content=value) for value in out])
def format_element(bfo, name, tag_name='', tag='', kb='', kb_default_output='', var='', protocol='googlescholar'):
"""Prints a custom field in a way suitable to be used in HTML META
tags. In particular conforms to Google Scholar harvesting protocol as
defined http://scholar.google.com/intl/en/scholar/inclusion.html and
Open Graph http://ogp.me/
@param tag_name: the name, from tag table, of the field to be exported
looks initially for names prefixed by "meta-"<tag_name>
then looks for exact name, then falls through to "tag"
@param tag: the MARC tag to be exported (only if not defined by tag_name)
@param name: name to be displayed in the meta headers, labelling this value.
@param kb: a knowledge base through which to process the retrieved value if necessary.
@param kb: when a '<code>kb</code>' is specified and no match for value is found, what shall we
return? Either return the given parameter or specify "{value}" to return the retrieved
value before processing though kb.
@param var: the name of a variable to output instead of field from metadata.
Allowed values are those supported by bfe_server_info and
bfe_client_info. Overrides <code>name</code> and <code>tag_name</code>
@param protocol: the protocol this tag is aimed at. Can be used to switch on/off support for a given "protocol". Can take values among 'googlescholar', 'opengraph'
@see: bfe_server_info.py, bfe_client_info.py
"""
if protocol == 'googlescholar' and not CFG_WEBSEARCH_ENABLE_GOOGLESCHOLAR:
return ""
elif protocol == 'opengraph' and not CFG_WEBSEARCH_ENABLE_OPENGRAPH:
return ""
tags = []
if var:
# delegate to bfe_server_info or bfe_client_info:
value = server_info(bfo, var)
if value.startswith("Unknown variable: "):
# Oops variable was not defined there
value = client_info(bfo, var)
return not value.startswith("Unknown variable: ") and \
create_metatag(name=name, content=cgi.escape(value, True)) \
or ""
elif tag_name:
# First check for special meta named tags
tags = get_field_tags("meta-" + tag_name)
if not tags:
# then check for regular tags
tags = get_field_tags(tag_name)
if not tags and tag:
# fall back to explicit marc tag
tags = [tag]
if not tags:
return ''
out = []
values = [bfo.fields(marctag, escape=9) for marctag in tags]
for value in values:
if isinstance(value, list):
for val in value:
if isinstance(val, dict):
out += val.values()
else:
out.append(val)
out += [isinstance(val, dict) and val.values() or val for val in value]
elif isinstance(value, dict):
out += value.values()
else:
out.append(value)
out = dict(zip(out, len(out)*[''])).keys() # Remove duplicates
if name == 'citation_date':
for idx in range(len(out)):
out[idx] = out[idx].replace('-', '/')
if kb:
if kb_default_output == "{value}":
out = [bfo.kb(kb, value, value) for value in out]
else:
out = [bfo.kb(kb, value, kb_default_output) for value in out]
return '\n'.join([create_metatag(name=name, content=value) for value in out])