def before_index(self, pkg_dict):
"""
Dynamically creates a license_id-is-ogl field to index on, and clean
up resource formats prior to indexing.
"""
SearchIndexing.clean_title_string(pkg_dict)
SearchIndexing.add_field__is_ogl(pkg_dict)
SearchIndexing.resource_format_cleanup(pkg_dict)
SearchIndexing.add_field__group_titles(pkg_dict)
SearchIndexing.add_field__publisher(pkg_dict)
if is_plugin_enabled('harvest'):
SearchIndexing.add_field__harvest_document(pkg_dict)
SearchIndexing.add_field__openness(pkg_dict)
SearchIndexing.add_popularity(pkg_dict)
SearchIndexing.add_field__group_abbreviation(pkg_dict)
SearchIndexing.add_inventory(pkg_dict)
# Extract multiple theme values (concatted with ' ') into one multi-value schema field
all_themes = set()
for value in (pkg_dict.get('theme-primary', ''), pkg_dict.get('theme-secondary', '')):
for theme in value.split(' '):
if theme:
all_themes.add(theme)
pkg_dict['all_themes'] = list(all_themes)
return pkg_dict
def before_index(self, pkg_dict):
"""
Dynamically creates a license_id-is-ogl field to index on, and clean
up resource formats prior to indexing.
"""
SearchIndexing.clean_title_string(pkg_dict)
SearchIndexing.add_field__is_ogl(pkg_dict)
SearchIndexing.resource_format_cleanup(pkg_dict)
SearchIndexing.add_field__group_titles(pkg_dict)
SearchIndexing.add_field__publisher(pkg_dict)
if is_plugin_enabled('harvest'):
SearchIndexing.add_field__harvest_document(pkg_dict)
SearchIndexing.add_field__openness(pkg_dict)
SearchIndexing.add_popularity(pkg_dict)
SearchIndexing.add_field__group_abbreviation(pkg_dict)
SearchIndexing.add_inventory(pkg_dict)
# Extract multiple theme values (concatted with ' ') into one multi-value schema field
all_themes = set()
for value in (pkg_dict.get('theme-primary',
''), pkg_dict.get('theme-secondary', '')):
for theme in value.split(' '):
if theme:
all_themes.add(theme)
pkg_dict['all_themes'] = list(all_themes)
return pkg_dict
def before_index(self, pkg_dict):
"""
DGU-specific changes to the pkg_dict before it is indexed. The main
reason is so that we can add search facets.
"""
log.info('Indexing: %s', pkg_dict['name'])
SearchIndexing.clean_title_string(pkg_dict)
SearchIndexing.add_field__is_ogl(pkg_dict)
SearchIndexing.resource_format_cleanup(pkg_dict)
SearchIndexing.add_field__publisher(pkg_dict)
SearchIndexing.add_field__organization_title_and_abbreviation(pkg_dict)
if is_plugin_enabled('harvest'):
SearchIndexing.add_field__harvest_document(pkg_dict)
SearchIndexing.add_field__openness(pkg_dict)
SearchIndexing.add_popularity(pkg_dict)
SearchIndexing.add_inventory(pkg_dict)
SearchIndexing.add_its(pkg_dict)
SearchIndexing.add_register(pkg_dict)
SearchIndexing.add_api_flag(pkg_dict)
SearchIndexing.add_theme(pkg_dict)
if is_plugin_enabled('dgu_schema'):
SearchIndexing.add_schema(pkg_dict)
SearchIndexing.add_collections(pkg_dict)
return pkg_dict
def before_index(self, pkg_dict):
"""
Dynamically creates a license_id-is-ogl field to index on, and clean
up resource formats prior to indexing.
"""
SearchIndexing.add_field__is_ogl(pkg_dict)
SearchIndexing.resource_format_cleanup(pkg_dict)
SearchIndexing.add_field__group_titles(pkg_dict)
SearchIndexing.add_field__publisher(pkg_dict)
SearchIndexing.add_field__harvest_document(pkg_dict)
SearchIndexing.add_field__openness(pkg_dict)
SearchIndexing.add_popularity(pkg_dict)
return pkg_dict
def before_index(self, pkg_dict):
"""
Dynamically creates a license_id-is-ogl field to index on, and clean
up resource formats prior to indexing.
"""
SearchIndexing.add_field__is_ogl(pkg_dict)
SearchIndexing.resource_format_cleanup(pkg_dict)
SearchIndexing.add_field__group_titles(pkg_dict)
SearchIndexing.add_field__publisher(pkg_dict)
SearchIndexing.add_field__harvest_document(pkg_dict)
SearchIndexing.add_field__openness(pkg_dict)
SearchIndexing.add_popularity(pkg_dict)
return pkg_dict
def before_index(self, pkg_dict):
"""
Dynamically creates a license_id-is-ogl field to index on, and clean
up resource formats prior to indexing.
"""
log.info('Indexing: %s', pkg_dict['name'])
SearchIndexing.clean_title_string(pkg_dict)
SearchIndexing.add_field__is_ogl(pkg_dict)
SearchIndexing.resource_format_cleanup(pkg_dict)
SearchIndexing.add_field__group_titles(pkg_dict)
SearchIndexing.add_field__publisher(pkg_dict)
if is_plugin_enabled('harvest'):
SearchIndexing.add_field__harvest_document(pkg_dict)
SearchIndexing.add_field__openness(pkg_dict)
SearchIndexing.add_popularity(pkg_dict)
SearchIndexing.add_field__group_abbreviation(pkg_dict)
SearchIndexing.add_inventory(pkg_dict)
SearchIndexing.add_theme(pkg_dict)
if is_plugin_enabled('dgu_schema'):
SearchIndexing.add_schema(pkg_dict)
return pkg_dict
def before_index(self, pkg_dict):
"""
Dynamically creates a license_id-is-ogl field to index on, and clean
up resource formats prior to indexing.
"""
log.info('Indexing: %s', pkg_dict['name'])
SearchIndexing.clean_title_string(pkg_dict)
SearchIndexing.add_field__is_ogl(pkg_dict)
SearchIndexing.resource_format_cleanup(pkg_dict)
SearchIndexing.add_field__group_titles(pkg_dict)
SearchIndexing.add_field__publisher(pkg_dict)
if is_plugin_enabled('harvest'):
SearchIndexing.add_field__harvest_document(pkg_dict)
SearchIndexing.add_field__openness(pkg_dict)
SearchIndexing.add_popularity(pkg_dict)
SearchIndexing.add_field__group_abbreviation(pkg_dict)
SearchIndexing.add_inventory(pkg_dict)
SearchIndexing.add_theme(pkg_dict)
if is_plugin_enabled('dgu_schema'):
SearchIndexing.add_schema(pkg_dict)
return pkg_dict