def add_faceted_document(index):
document = search.Document(
doc_id='doc1',
fields=[search.AtomField(name='name', value='x86')],
facets=[
search.AtomFacet(name='type', value='computer'),
search.NumberFacet(name='ram_size_gb', value=8)
])
index.put(document)
def add_to_search_index(object_id,
labels,
metadata,
mapped_category=None,
most_similar_category=None,
thumb_url=None,
preview_url=None):
"""Adds document to the search index."""
try:
# Define document search fields - these can be queried using keyword search
fields = [
search.TextField(name='image_id', value=object_id),
]
# Define document facet fields
facets = []
# Add label descriptions into search and facet fields. Search API allows
# multiple values for the same field.
for label in labels:
fields.append(
search.TextField(name='label', value=label['description']))
facets.append(
search.AtomFacet(name='label_facet',
value=label['description']))
# Add mapped category and most similar category as facets
if mapped_category:
fields.append(
search.TextField(name='mapped_category',
value=mapped_category))
facets.append(
search.AtomFacet(name='mapped_category_facet',
value=mapped_category))
if most_similar_category:
fields.append(
search.TextField(name='most_similar_category',
value=most_similar_category))
facets.append(
search.AtomFacet(name='most_similar_category_facet',
value=most_similar_category))
# We're not using a database, so store the image URLs to the index.
# We don't need to search on the image URL, but we will need them to display
# images in the user interface.
# Add thumbnail url
if thumb_url:
fields.append(search.TextField(name='thumb_url', value=thumb_url))
# Add preview url
if thumb_url:
fields.append(
search.TextField(name='preview_url', value=preview_url))
# Add any other object metadata as document search fields
for k, v in metadata.iteritems():
fields.append(search.TextField(name=k, value=v))
# Add the document to the search index
d = search.Document(doc_id=hashing.hash_value(object_id),
fields=fields,
facets=facets)
add_result = search.Index(name='imagesearch').put(d)
doc_id = add_result[0].id
return doc_id
except search.Error:
logging.exception('Something went wrong in add_to_search_index()')
def item_data(item):
fields = [
search.AtomField('store',
item.key.parent().id()),
search.AtomField('sku', item.key.id()),
search.TextField('title', item.title),
# "Unsupported field type TOKENIZED_PREFIX"
# search.TokenizedPrefixField('title_prefix', item.title),
search.AtomField('url', item.url),
search.AtomField('image', item.image),
# DateField supports only date accuracy (ie. not second)
search.NumberField('added', to_unix(item.added)),
search.NumberField('checked', to_unix(item.checked))
]
if item.custom:
custom = set()
for val in item.custom.itervalues():
if isinstance(val, basestring):
custom.add(val)
elif isinstance(val, (int, long)):
custom.add(str(val))
if custom:
fields.append(
search.TextField('custom', " ".join(sorted(custom))))
facets = []
if item.category:
id_path = [
"%d" % ck.id() for ck in cat_path(item.key, item.category)
]
if id_path:
fields.append(search.TextField('categories',
" ".join(id_path)))
# NumberFacet is 30 bit
facets += [
search.AtomFacet('category', cat_id) for cat_id in id_path
]
prices = Price.query(ancestor=item.key) \
.order(-Price.timestamp) \
.fetch()
if prices:
us_cents = map(to_us_cents, prices)
fields += [
search.NumberField('us_cents', us_cents[0]),
search.TextField('price_history',
" ".join(map(format_history_price, prices))),
]
if len(us_cents) > 1:
discount_usc = max(us_cents) - us_cents[0]
if discount_usc:
fields.append(
search.NumberField('discount_us_cents', discount_usc))
discount_pc = discount_usc * 100 / max(us_cents)
if discount_pc:
fields.append(
search.NumberField('discount_pc', discount_pc))
tags = []
if item.removed:
fields.append(search.NumberField('removed', to_unix(item.removed)))
tags.append('removed')
else:
# NOT queries are expensive, thus providing information in both forms
tags.append('active')
fields += [search.AtomField('tags', "#%s" % t) for t in tags]
return fields, facets