def expected_fields(cls, include_nullable=True):
"""
Returns a list of fields we expect to be present in the mapping and
in the extraction method.
Should be updated whenever you change the mapping to add/remove fields.
"""
# Fields that can not be directly compared with the property of the
# same name on the Addon instance, either because the property doesn't
# exist on the model, or it has a different name, or the value we need
# to store in ES differs from the one in the db.
complex_fields = [
'app', 'boost', 'category', 'colors', 'current_version',
'description', 'has_eula', 'has_privacy_policy', 'listed_authors',
'name', 'platforms', 'previews', 'ratings', 'summary', 'tags',
]
# Fields that need to be present in the mapping, but might be skipped
# for extraction because they can be null.
nullable_fields = []
# For each translated field that needs to be indexed, we store one
# version for each language we have an analyzer for.
_indexed_translated_fields = ('name', 'description', 'summary')
analyzer_fields = list(chain.from_iterable(
[['%s_l10n_%s' % (field, lang) for lang, analyzer
in SEARCH_LANGUAGE_TO_ANALYZER.items()]
for field in _indexed_translated_fields]))
# It'd be annoying to hardcode `analyzer_fields`, so we generate it,
# but to make sure the test is correct we still do a simple check of
# the length to make sure we properly flattened the list.
assert len(analyzer_fields) == (len(SEARCH_LANGUAGE_TO_ANALYZER) *
len(_indexed_translated_fields))
# Each translated field that we want to return to the API.
raw_translated_fields = [
'%s_translations' % field for field in
['name', 'description', 'developer_comments', 'homepage',
'summary', 'support_email', 'support_url']]
# Return a list with the base fields and the dynamic ones added.
fields = (cls.simple_fields + complex_fields + analyzer_fields +
raw_translated_fields)
if include_nullable:
fields += nullable_fields
return fields
def attach_language_specific_analyzers(cls, mapping, field_names):
"""
For each field in field_names, attach language-specific mappings that
will use specific analyzers for these fields in every language that we
support.
These mappings are used by the search filtering code if they exist.
"""
doc_name = cls.get_doctype_name()
for lang, analyzer in SEARCH_LANGUAGE_TO_ANALYZER.items():
for field in field_names:
property_name = '%s_l10n_%s' % (field, lang)
mapping[doc_name]['properties'][property_name] = {
'type': 'text',
'analyzer': analyzer,
}
def attach_language_specific_analyzers_with_raw_variant(
cls, mapping, field_names):
"""
Like attach_language_specific_analyzers() but with an extra field to
storethe "raw" variant of the value, for exact matches.
"""
doc_name = cls.get_doctype_name()
for lang, analyzer in SEARCH_LANGUAGE_TO_ANALYZER.items():
for field in field_names:
property_name = '%s_l10n_%s' % (field, lang)
mapping[doc_name]['properties'][property_name] = {
'type': 'text',
'analyzer': analyzer,
'fields': {
'raw': cls.get_raw_field_definition(),
}
}