def update(self, **caches):
"""Refresh index version.
**POST** /update
Reopen searcher, optionally reloading caches, and return document count.
{"spellcheckers": true,... }
.. versionchanged:: 1.2 request body is an object instead of an array
:return: *int*
"""
names = ()
while self.urls:
url = self.urls[0]
try:
names = self.sync(url)
break
except IOError:
with contextlib.suppress(ValueError):
self.urls.remove(url)
self.searcher = self.searcher.reopen(**caches)
self.updated = time.time()
if names:
engine.IndexWriter(self.searcher.directory).close()
if not self.urls and hasattr(self, 'fields'):
other = WebIndexer(self.searcher.directory,
analyzer=self.searcher.analyzer)
other.indexer.shared, other.indexer.fields = self.searcher.shared, self.fields
app, = (app for app in cherrypy.tree.apps.values()
if app.root is self)
mount(other, app=app, autoupdate=getattr(self, 'autoupdate', 0))
return len(self.searcher)
def test_facets(tempdir, servers, zipcodes):
writer = engine.IndexWriter(tempdir)
writer.commit()
resource = servers.start(servers.ports[0], '-r', tempdir)
writer.set('zipcode', dimensions=1, stored=True)
writer.fields['location'] = engine.NestedField('county.city', docValuesType='sorted')
for doc in zipcodes:
if doc['state'] == 'CA':
writer.add(zipcode=int(doc['zipcode']), location='{}.{}'.format(doc['county'], doc['city']))
writer.commit()
assert resource.post('update') == resource().popitem()[1] == len(writer)
result = resource.search(count=0, facets='county', **{'count.min': 10000})
facets = result['facets']['county']
assert result['count'] == sum(facets.values()) and 'Los Angeles' in facets
result = resource.search(q='Los Angeles', count=0, facets='county.city', **{'q.type': 'term', 'q.field': 'county'})
facets = result['facets']['county.city']
assert result['count'] == sum(facets.values()) and all(location.startswith('Los Angeles.') for location in facets)
result = resource.search(count=0, facets='county', **{'facets.count': 3})
assert sorted(result['facets']['county']) == ['Los Angeles', 'Orange', 'San Diego']
result = resource.search(count=0, facets='county', **{'facets.min': 140})
assert sorted(result['facets']['county']) == ['Los Angeles', 'Orange', 'San Diego']
result = resource.search(
q='Los Angeles', group='county.city', **{'group.count': 2, 'q.field': 'county', 'q.type': 'prefix'}
)
assert all(group['value'].startswith('Los Angeles') for group in result['groups'])
assert sum(map(operator.itemgetter('count'), result['groups'])) == sum(facets.values()) == result['count']
def __init__(self, *directories, **kwargs):
self.urls = collections.deque(kwargs.pop('urls', ()))
if self.urls:
engine.IndexWriter(*directories).close()
self.searcher = engine.MultiSearcher(
directories, **
kwargs) if len(directories) > 1 else engine.IndexSearcher(
*directories, **kwargs)
self.updated = time.time()
self.query_map = {}
def test_config(tempdir, servers):
assert subprocess.call((sys.executable, '-m', 'lupyne.server', '-c', __file__), stderr=subprocess.PIPE)
engine.IndexWriter(tempdir).close()
config = {'tools.validate.last_modified': True, 'tools.validate.expires': 0, 'tools.validate.max_age': 0}
client = servers.start(servers.ports[0], tempdir, tempdir, '--autoreload=0.1', **config).client
response = client.get()
assert response.ok
((directory, size),) = response.json().items()
assert 'Directory@' in directory and size == 0
assert int(response.headers['age']) >= 0
assert response.headers['cache-control'] == 'max-age=0'
assert float(response.headers['x-response-time']) > 0.0
dates = [parsedate(response.headers[header]) for header in ('last-modified', 'expires', 'date')]
assert all(dates) and sorted(dates) == dates
w, verion, _ = response.headers['etag'].split('"')
assert w == 'W/' and verion.isdigit()
def test_replication(tempdir, servers):
primary = servers.start(servers.ports[0], tempdir, '--autoupdate=1')
directory = os.path.join(tempdir, 'backup')
engine.IndexWriter(directory).close()
secondary = servers.start(servers.ports[1], '-r', directory, '--autoupdate=1')
primary.post('docs', [{}])
assert primary.client.get('update/0').status_code == http.client.NOT_FOUND
assert primary.client.get('update/x').status_code == http.client.NOT_FOUND
response = primary.client.post('update', {'snapshot': True})
assert response.status_code == http.client.CREATED
location, filenames = response.headers['location'], response.json()
assert primary.get(location) == filenames
for filename in filenames:
shutil.copy(os.path.join(tempdir, filename), directory)
assert primary.delete(location) == filenames
time.sleep(1.1)
assert secondary.get('docs') == [0]
def index(tempdir, fields, constitution):
with engine.IndexWriter(tempdir) as writer:
writer.fields.update({field.name: field for field in fields})
for doc in constitution:
writer.add(doc)
return tempdir