def test_facet():
config = {'formatter': 'csv'}
f = facet.FacetFactory(config).create()
f.install('data/install/american-english', nrows=50000)
matches = f.match('beautiful window in Apollo spacecraft')
print(matches)
f.close()
def server(
config,
host,
port,
query,
alpha,
similarity,
ngram,
formatter,
tokenizer,
database,
install,
dump_config,
):
# Resolve settings for dumping configuration
full_config = copy.deepcopy(config)
dump_config = config.pop('dump_config', dump_config)
if dump_config:
dump_output, dump_format = parse_dump_configuration(dump_config)
# Support CLI shortcut options from configuration files
host = config.pop('host', host)
port = config.pop('port', port)
query = config.pop('query', query)
install = config.pop('install', install)
if isinstance(install, str):
install = {'filename': install}
# Prepare factory options from configuration
factory_config = copy.deepcopy(config)
factory_config['class'] = config.get('class', 'facet')
factory_config['tokenizer'] = config.get('tokenizer', tokenizer)
factory_config['formatter'] = config.get('formatter', formatter)
factory_config['matcher'] = config.get(
'matcher', {
'class': 'simstring',
'db': database,
'alpha': config.get('alpha', alpha),
'similarity': config.get('similarity', similarity),
'ngram': config.get('ngram', ngram),
})
if dump_config:
full_config.update(factory_config)
dump_configuration({'SERVER': full_config}, dump_output, dump_format)
return
f = facet.FacetFactory(factory_config).create()
if install:
f.install(**install)
with facet.network.SocketServer(
(host, port),
facet.network.SocketServerHandler,
served_object=f,
) as server:
server.serve_forever()
f.close()
def run(
config,
query,
alpha,
similarity,
ngram,
formatter,
output,
tokenizer,
database,
install,
dump_config,
):
# Resolve settings for dumping configuration
full_config = copy.deepcopy(config)
dump_config = config.pop('dump_config', dump_config)
if dump_config:
dump_output, dump_format = parse_dump_configuration(dump_config)
# Support CLI shortcut options from configuration files
query = config.pop('query', query)
output = config.pop('output', output)
install = config.pop('install', install)
if isinstance(install, str):
install = {'filename': install}
# Prepare factory options from configuration
factory_config = copy.deepcopy(config)
factory_config['class'] = config.get('class', 'facet')
factory_config['tokenizer'] = config.get('tokenizer', tokenizer)
factory_config['formatter'] = config.get('formatter', formatter)
factory_config['matcher'] = config.get(
'matcher', {
'class': 'simstring',
'db': database,
'alpha': config.get('alpha', alpha),
'similarity': config.get('similarity', similarity),
'ngram': config.get('ngram', ngram),
})
if dump_config:
full_config.update(factory_config)
dump_configuration({'FACET': full_config}, dump_output, dump_format)
return
f = facet.FacetFactory(factory_config).create()
if install:
f.install(**install)
if query:
matches = f.match(query, output=output)
if matches is not None:
print(matches)
else:
repl_loop(f)
f.close()
def __init__(
self,
config: Union[str, Dict[str, Any]],
num_procs: int = multiprocessing.cpu_count(),
):
self.num_procs = num_procs
self._factory = facet.FacetFactory(config)
self._formatter = get_formatter(
self._factory.get_config().pop('formatter', None)
)
import facet
def generate_sources(n):
source = 'acetate'
return [source] * n
factory = facet.FacetFactory('config/umls_sqlite.yaml:Search')
host = 'localhost'
port = 4444
with facet.network.SocketClient(
(host, port),
target_class=factory.get_class(),
) as f:
matches = f.match(generate_sources(8))
print(matches)
import facet
factory = facet.FacetFactory('config/umls_sqlite.yaml:Install')
f = factory.create()
f.install('data/umls', nrows=40000)
host = 'localhost'
port = 4444
with facet.network.SocketServer(
(host, port),
facet.network.SocketServerHandler,
served_object=f,
) as server:
server.serve_forever()
f.close()
import facet
def generate_sources(n):
source = 'acetate'
return [source] * n
factory = facet.FacetFactory('config/umls_redis.yaml:Search')
host = 'localhost'
port = 4444
with facet.network.SocketClient(
(host, port),
target_class=factory.get_class(),
) as f:
matches = f.match(generate_sources(8))
print(matches)
import facet
# Install
factory = facet.FacetFactory('config/umls_redis.yaml:Install')
tmp_f = factory.create()
tmp_f.install('data/umls', nrows=40000)
# Parallel matcher
f = facet.ParallelFacet('config/umls_redis.yaml:Search')
host = 'localhost'
port = 4444
with facet.network.SocketServer(
(host, port),
facet.network.SocketServerHandler,
served_object=f,
) as server:
server.serve_forever()
f.close()