}
mq_client = message_queue_client_from_config(mq_config)
mq_codec = JSONCodec()
processed_resource_queue = 'processed_resources'
# ElasticSearch
es = ES('localhost:9200', timeout=60)
es_index = 'topic_tracking'
# dequeue one resource
mq_client.connect()
message = mq_client.get_message(processed_resource_queue)
resource = mq_codec.decode(message.body, Resource)
mq_client.delete_message(processed_resource_queue, message.id)
mq_client.disconnect()
# save the resource to mongo
resource._id = makeIdFromURI(resource.uri)
resource_collection.insert_model(resource)
# index the resource
for boost in [1, 1000]:
es_doc = {}
es_doc['content'] = resource.content
es_doc['title'] = resource.title
es_doc['entities'] = build_payload_string(resource.entities, boost)
es_doc['terms'] = build_payload_string(resource.terms, boost)
id = '%s_%d' % (resource._id, boost)
r = es.index(es_doc, es_index, 'resource', id)
pprint(r)
class ESDataTarget(base.DataTarget):
"""docstring for ClassName
"""
def __init__(self, document_type, database="test", host="127.0.0.1", port="9200",
truncate=False, expand=False, **elasticsearch_args):
"""Creates a ElasticSearch data target stream.
:Attributes:
* document_ElasticSearch elasticsearch document_type name
* database: database name
* host: ElasticSearch database server host, default is ``localhost``
* port: ElasticSearch port, default is ``9200``
* expand: expand dictionary values and treat children as top-level keys with dot '.'
separated key path to the child..
* truncate: delete existing data in the document_type. Default: False
"""
self.document_type = document_type
self.database_name = database
self.host = host
self.port = port
self.elasticsearch_args = elasticsearch_args
self.expand = expand
self.truncate = truncate
self._fields = None
def initialize(self):
"""Initialize ElasticSearch source stream:
"""
from pyes.es import ES
from pyes.exceptions import IndexAlreadyExistsException
args = self.elasticsearch_args.copy()
server = ""
if self.host:
server = self.host
if self.port:
server += ":" + self.port
create = args.pop("create", False)
replace = args.pop("replace", False)
self.connection = ES(server, **args)
self.connection.default_indices = self.database_name
self.connection.default_types = self.document_type
created = False
if create:
try:
self.connection.create_index(self.database_name)
self.connection.refresh(self.database_name)
created = True
except IndexAlreadyExistsException:
pass
if replace and not created:
self.connection.delete_index_if_exists(self.database_name)
time.sleep(2)
self.connection.create_index(self.database_name)
self.connection.refresh(self.database_name)
if self.truncate:
self.connection.delete_mapping(self.database_name, self.document_type)
self.connection.refresh(self.database_name)
def append(self, obj):
record = obj
if not isinstance(obj, dict):
record = dict(zip(self.fields.names(), obj))
if self.expand:
record = expand_record(record)
id = record.get('id') or record.get('_id')
self.connection.index(record, self.database_name, self.document_type, id, bulk=True)
def finalize(self):
self.connection.flush_bulk(forced=True)
class FullTextSearch(object):
def __init__(self, server, settings=None):
# These timeout and bulk_size parameters were determined through
# trial and error to be necessary to avoid timeout errors when
# generating indices on Sandbox. They should not be taken as gospel.
self.conn = ES(server, timeout=120.0) # Default timeout: 30.0
self.conn.bulker.bulk_size = 25 # Default: 400
if settings:
self.settings = settings
else:
self.settings = {
'index': {
'analysis': {
'analyzer': {
'ngram_analyzer': {
'tokenizer': 'keyword',
'filter': ['lowercase', 'filter_ngram'],
'type': 'custom'
}
},
'filter': {
'filter_ngram': {
'type': 'nGram',
'max_gram': 30,
'min_gram': 1
}
}
}
}
}
self.refresh_index_cache()
def search_index_text(self, query_string, fields="_all", **args):
q = query.MatchQuery(fields, query_string)
return self.search_index(q, **args)
def search_index(self, query, indices=None, num_results=None, node_type=None):
results = self.conn.search(
query=query, indices=indices, doc_types=node_type)
meta_list = [r.get_meta() for r in results[0:num_results]]
node_dict = {}
# fetch nodes grouped by type to reduce number of db calls
key = itemgetter('type')
for t, grouped_list in groupby(sorted(meta_list, key=key), key=key):
ids = [meta['id'] for meta in grouped_list]
for node in self.datastore.get_nodes(t, ids):
node_dict[(node.type, node.key)] = node
# return nodes in original order
nodelist = [node_dict[(meta['type'], meta['id'])]
for meta in meta_list]
return nodelist
def create_index(self, type, indexed_variables, index_name):
self.conn.indices.create_index_if_missing(index_name, self.settings)
mapping = {}
for arg in indexed_variables:
mapping[arg] = {'boost': 1.0,
'analyzer': 'ngram_analyzer',
'type': 'string',
'term_vector': 'with_positions_offsets'}
index_settings = {'index_analyzer': 'ngram_analyzer',
'search_analyzer': 'standard',
'properties': mapping}
self.conn.indices.put_mapping(str(type), index_settings, [index_name])
self.refresh_index_cache()
self.populate_index(type, index_name)
def refresh_index_cache(self):
try:
indices = self.conn.indices.get_mapping(raw=True)
except exceptions.IndexMissingException:
indices = {}
else:
indices = dict((k, v.get('mappings', {})) for k, v in indices.items())
self.indices = indices
def delete_index(self, index_name):
self.conn.indices.delete_index_if_exists(index_name)
self.refresh_index_cache()
def populate_index(self, type, index_name):
#add all the currently existing nodes into the index
ref_node = self.datastore.get_reference_node(type)
node_list = [rel.target_node for rel in ref_node.instance.outgoing]
for node in node_list:
key = node.key
index_dict = self.populate_index_document(node, index_name)
try:
self.conn.delete(index_name, type, key)
except ELASTIC_SEARCH_EXCEPTIONS as err:
log.exception(err)
pass
self.conn.index(index_dict, index_name, type, key, bulk=True)
self.conn.indices.refresh([index_name])
def on_create(self, node):
type_indices = self.get_indices_of_type(node.type)
for index_name in type_indices:
index_dict = self.populate_index_document(node, index_name)
try:
self.conn.index(index_dict, index_name, node.type, node.key, bulk=True)
self.conn.indices.refresh([index_name])
except ELASTIC_SEARCH_EXCEPTIONS as err:
log.exception(err)
pass
def on_delete(self, node):
type_indices = self.get_indices_of_type(node.type)
for index_name in type_indices:
try:
self.conn.delete(index_name, node.type, node.key, bulk=True)
self.conn.indices.refresh([index_name])
except ELASTIC_SEARCH_EXCEPTIONS as err:
log.exception(err)
pass
def on_modify(self, node):
type_indices = self.get_indices_of_type(node.type)
for index_name in type_indices:
index_dict = self.populate_index_document(node, index_name)
try:
self.conn.delete(index_name, node.type, node.key)
self.conn.index(index_dict, index_name, node.type, node.key, bulk=True)
self.conn.indices.refresh([index_name])
except ELASTIC_SEARCH_EXCEPTIONS as err:
log.exception(err)
pass
def get_indices_of_type(self, type):
type_indices = [
key for key, value in self.indices.items()
if type in value
]
return type_indices
def populate_index_document(self, node, index_name):
indexed_variables = self.indices[index_name][node.type]['properties'].keys()
index_dict = {
field: node[field] for field in indexed_variables
}
return index_dict
class FullTextSearch(object):
def __init__(self,server,settings = None ):
self.conn = ES(server)
self.indices = {}
if settings:
self.settings = settings
else:
self.settings = {
'index': {
'analysis' : {
'analyzer' : {
'ngram_analyzer' : {
'tokenizer' : 'keyword',
'filter' : ['lowercase', 'filter_ngram'],
'type' : 'custom'
}
},
'filter' : {
'filter_ngram' : {
'type' : 'nGram',
'max_gram' : 30,
'min_gram' : 1
}
}
}
}
}
def search_index(self, type, index_names, query_string, num_results=-1):
ns_index_names= [str(type) + "-_-" + index_name for index_name in index_names]
q = WildcardQuery('_all',lower(query_string))
results = self.conn.search(query=q, indices=ns_index_names, doc_types=type)
num_found = len(results)
if(num_results > num_found):
num_results = num_found
nodelist = [self.datastore.get_node(type,r['_id']) for r in results['hits']['hits']]
if(num_results!=-1):
return nodelist[0:num_results]
else:
return nodelist
def create_index(self, type, indexed_variables, index_name):
ns_index_name = str(type) + "-_-" + index_name
self.conn.delete_index_if_exists(ns_index_name)
self.conn.create_index(ns_index_name,self.settings)
mapping = {}
for arg in indexed_variables:
mapping[arg] = {'boost':1.0,
'analyzer' : 'ngram_analyzer',
'type': u'string',
'term_vector': 'with_positions_offsets'}
index_settings = {'index_analyzer':'ngram_analyzer',
'search_analyzer':'standard',
'properties':mapping}
self.conn.put_mapping(str(type),index_settings,[ns_index_name])
self.refresh_index_cache()
self.populate_index(type, index_name)
def refresh_index_cache(self):
self.indices = self.conn.get_indices()
def delete_index(self,type,index_name):
ns_index_name = str(type) + "-_-" + index_name
self.conn.delete_index_if_exists(ns_index_name)
self.refresh_index_cache()
def populate_index(self, type, index_name):
#add all the currently existing nodes into the index
ns_index_name = str(type) + "-_-" + index_name
ref_node = self.datastore.get_reference_node(type)
node_list = [rel.target_node for rel in ref_node.instance.outgoing]
mapping = self.conn.get_mapping(type,ns_index_name)
for node in node_list:
key = node.key
index_dict = self.populate_index_document(type,ns_index_name,node.attributes,mapping)
try:
self.conn.delete(ns_index_name,type,key)
except exceptions.NotFoundException:
pass
try:
self.conn.index(index_dict,ns_index_name,type,key)
except exceptions.ElasticSearchParseException:
pass
self.conn.refresh([ns_index_name])
def on_create(self,node):
type_indices = self.get_indices_of_type(node.type)
for ns_index_name in type_indices:
mapping = self.conn.get_mapping(node.type,ns_index_name)
index_dict = self.populate_index_document(node.type,ns_index_name,node.attributes,mapping)
self.conn.index(index_dict,ns_index_name,node.type,node.key)
self.conn.refresh([ns_index_name])
def on_delete(self, node):
type_indices = self.get_indices_of_type(node.type)
for ns_index_name in type_indices:
try:
self.conn.delete(ns_index_name,node.type,node.key)
self.conn.refresh([ns_index_name])
except exceptions.NotFoundException:
pass
def on_modify(self, node):
type_indices = self.get_indices_of_type(node.type)
for ns_index_name in type_indices:
mapping = self.conn.get_mapping(node.type,ns_index_name)
index_dict = self.populate_index_document(node.type,ns_index_name,node.attributes,mapping)
try:
self.conn.delete(ns_index_name,node.type,node.key)
self.conn.index(index_dict,ns_index_name,node.type,node.key)
self.conn.refresh([ns_index_name])
except exceptions.NotFoundException:
pass
def get_indices_of_type(self,type):
type_indices = []
for index in self.indices.keys():
if index.startswith(type+"-_-"):
type_indices.append(index)
return type_indices
def populate_index_document(self,type,ns_index_name,attributes,mapping):
indexed_variables = mapping[type]['properties'].keys()
index_dict = {}
for arg in indexed_variables:
try:
index_dict[arg] = attributes[arg]
except KeyError:
#if this attribute doesn't exist for this node, just pass
pass
return index_dict
'boost': 1.0,
'index': 'analyzed',
'store': 'true',
'type': u'string',
"term_vector": "with_positions_offsets"
},
u'name': {
'boost': 1.0,
'index': 'analyzed',
'store': 'true',
'type': u'string',
"term_vector": "with_positions_offsets"
},
u'age': {
'store': 'true',
'type': u'integer'
},
}
conn.create_index("test-index")
conn.put_mapping("test-type", {'properties': mapping}, ["test-index"])
start = datetime.now()
for k, userdata in dataset.items():
# conn.index(userdata, "test-index", "test-type", k)
conn.index(userdata, "test-index", "test-type", k, bulk=True)
conn.force_bulk()
end = datetime.now()
print "time:", end - start
dataset.close()
class ESDataTarget(DataTarget):
"""docstring for ClassName
"""
def __init__(self, document_type, index="test", host="127.0.0.1", port="9200", truncate=False, expand=False,
**elasticsearch_args):
"""Creates a ElasticSearch data target stream.
:Attributes:
* document_ElasticSearch elasticsearch document_type name
* index: database name
* host: ElasticSearch database server host, default is ``localhost``
* port: ElasticSearch port, default is ``9200``
* expand: expand dictionary values and treat children as top-level keys with dot '.'
separated key path to the child..
* truncate: delete existing data in the document_type. Default: False
"""
super(ESDataTarget, self).__init__()
self.document_type = document_type
self.index = index
self.host = host
self.port = port
self.elasticsearch_args = elasticsearch_args
self.expand = expand
self.truncate = truncate
self._fields = None
def initialize(self):
"""
Initialize ElasticSearch source stream:
"""
from pyes.es import ES
from pyes.exceptions import IndexAlreadyExistsException
args = self.elasticsearch_args.copy()
server = ""
if self.host:
server = self.host
if self.port:
server += ":" + self.port
create = args.pop("create", False)
replace = args.pop("replace", False)
self.connection = ES(server, **args)
self.connection.default_indices = self.index
self.connection.default_types = self.document_type
created = False
if create:
try:
self.connection.create_index(self.index)
self.connection.refresh(self.index)
created = True
except IndexAlreadyExistsException:
pass
if replace and not created:
self.connection.delete_index_if_exists(self.index)
self.connection.refresh(self.index)
self.connection.create_index(self.index)
self.connection.refresh(self.index)
if self.truncate:
self.connection.delete_mapping(self.index, self.document_type)
self.connection.refresh(self.index)
#check mapping
try:
self.connection.get_mapping(self.document_type, self.index)
except TypeMissingException:
self.connection.put_mapping(self.document_type, self._get_mapping(), self.index)
def _get_mapping(self):
"""Build an ES optimized mapping for the given fields"""
from pyes.mappings import DocumentObjectField, IntegerField, StringField, BooleanField, FloatField, DateField
document = DocumentObjectField(name=self.document_type)
for field in self.fields:
st = field.storage_type
if st == "unknown":
#lets es detect the type
continue
elif st in ["string", "text"]:
document.add_property(StringField(name=field.name))
elif st == "integer":
document.add_property(IntegerField(name=field.name))
elif st == "boolean":
document.add_property(BooleanField(name=field.name))
elif st == "date":
document.add_property(DateField(name=field.name))
elif st == "float":
document.add_property(FloatField(name=field.name))
return document
def append(self, obj):
record = obj
if not isinstance(obj, dict):
record = dict(zip(self.field_names, obj))
if self.expand:
record = expand_record(record)
id = record.get('id') or record.get('_id')
self.connection.index(record, self.index, self.document_type, id, bulk=True)
def finalize(self):
self.connection.flush_bulk(forced=True)
dataset = shelve.open("samples.shelve")
mapping = {u'description': {'boost': 1.0,
'index': 'analyzed',
'store': 'yes',
'type': u'string',
"term_vector": "with_positions_offsets"
},
u'name': {'boost': 1.0,
'index': 'analyzed',
'store': 'yes',
'type': u'string',
"term_vector": "with_positions_offsets"
},
u'age': {'store': 'yes',
'type': u'integer'},
}
conn.create_index("test-index")
conn.put_mapping("test-type", {'properties': mapping}, ["test-index"])
start = datetime.now()
for k, userdata in dataset.items():
# conn.index(userdata, "test-index", "test-type", k)
conn.index(userdata, "test-index", "test-type", k, bulk=True)
conn.force_bulk()
end = datetime.now()
print "time:", end - start
dataset.close()
class ESDataTarget(DataTarget):
"""docstring for ClassName
"""
def __init__(self,
document_type,
index="test",
host="127.0.0.1",
port="9200",
truncate=False,
expand=False,
**elasticsearch_args):
"""Creates a ElasticSearch data target stream.
:Attributes:
* document_ElasticSearch elasticsearch document_type name
* index: database name
* host: ElasticSearch database server host, default is ``localhost``
* port: ElasticSearch port, default is ``9200``
* expand: expand dictionary values and treat children as top-level keys with dot '.'
separated key path to the child..
* truncate: delete existing data in the document_type. Default: False
"""
super(ESDataTarget, self).__init__()
self.document_type = document_type
self.index = index
self.host = host
self.port = port
self.elasticsearch_args = elasticsearch_args
self.expand = expand
self.truncate = truncate
self._fields = None
def initialize(self):
"""
Initialize ElasticSearch source stream:
"""
from pyes.es import ES
from pyes.exceptions import IndexAlreadyExistsException
args = self.elasticsearch_args.copy()
server = ""
if self.host:
server = self.host
if self.port:
server += ":" + self.port
create = args.pop("create", False)
replace = args.pop("replace", False)
self.connection = ES(server, **args)
self.connection.default_indices = self.index
self.connection.default_types = self.document_type
created = False
if create:
try:
self.connection.create_index(self.index)
self.connection.refresh(self.index)
created = True
except IndexAlreadyExistsException:
pass
if replace and not created:
self.connection.delete_index_if_exists(self.index)
self.connection.refresh(self.index)
self.connection.create_index(self.index)
self.connection.refresh(self.index)
if self.truncate:
self.connection.delete_mapping(self.index, self.document_type)
self.connection.refresh(self.index)
#check mapping
try:
self.connection.get_mapping(self.document_type, self.index)
except TypeMissingException:
self.connection.put_mapping(self.document_type,
self._get_mapping(), self.index)
def _get_mapping(self):
"""Build an ES optimized mapping for the given fields"""
from pyes.mappings import DocumentObjectField, IntegerField, StringField, BooleanField, FloatField, DateField
document = DocumentObjectField(name=self.document_type)
for field in self.fields:
st = field.storage_type
if st == "unknown":
#lets es detect the type
continue
elif st in ["string", "text"]:
document.add_property(StringField(name=field.name))
elif st == "integer":
document.add_property(IntegerField(name=field.name))
elif st == "boolean":
document.add_property(BooleanField(name=field.name))
elif st == "date":
document.add_property(DateField(name=field.name))
elif st == "float":
document.add_property(FloatField(name=field.name))
return document
def append(self, obj):
record = obj
if not isinstance(obj, dict):
record = dict(zip(self.field_names, obj))
if self.expand:
record = expand_record(record)
id = record.get('id') or record.get('_id')
self.connection.index(record,
self.index,
self.document_type,
id,
bulk=True)
def finalize(self):
self.connection.flush_bulk(forced=True)
class FullTextSearch(object):
def __init__(self, server, settings=None):
self.conn = ES(server)
self.indices = {}
if settings:
self.settings = settings
else:
self.settings = {
'index': {
'analysis': {
'analyzer': {
'ngram_analyzer': {
'tokenizer': 'keyword',
'filter': ['lowercase', 'filter_ngram'],
'type': 'custom'
}
},
'filter': {
'filter_ngram': {
'type': 'nGram',
'max_gram': 30,
'min_gram': 1
}
}
}
}
}
def search_index(self, type, index_names, query_string, num_results=-1):
ns_index_names = [
str(type) + "-_-" + index_name for index_name in index_names
]
q = WildcardQuery('_all', lower(query_string))
results = self.conn.search(query=q,
indices=ns_index_names,
doc_types=type)
num_found = len(results)
if (num_results > num_found):
num_results = num_found
nodelist = [
self.datastore.get_node(type, r['_id'])
for r in results['hits']['hits']
]
if (num_results != -1):
return nodelist[0:num_results]
else:
return nodelist
def create_index(self, type, indexed_variables, index_name):
ns_index_name = str(type) + "-_-" + index_name
self.conn.delete_index_if_exists(ns_index_name)
self.conn.create_index(ns_index_name, self.settings)
mapping = {}
for arg in indexed_variables:
mapping[arg] = {
'boost': 1.0,
'analyzer': 'ngram_analyzer',
'type': u'string',
'term_vector': 'with_positions_offsets'
}
index_settings = {
'index_analyzer': 'ngram_analyzer',
'search_analyzer': 'standard',
'properties': mapping
}
self.conn.put_mapping(str(type), index_settings, [ns_index_name])
self.refresh_index_cache()
self.populate_index(type, index_name)
def refresh_index_cache(self):
self.indices = self.conn.get_indices()
def delete_index(self, type, index_name):
ns_index_name = str(type) + "-_-" + index_name
self.conn.delete_index_if_exists(ns_index_name)
self.refresh_index_cache()
def populate_index(self, type, index_name):
#add all the currently existing nodes into the index
ns_index_name = str(type) + "-_-" + index_name
ref_node = self.datastore.get_reference_node(type)
node_list = [rel.target_node for rel in ref_node.instance.outgoing]
mapping = self.conn.get_mapping(type, ns_index_name)
for node in node_list:
key = node.key
index_dict = self.populate_index_document(type, ns_index_name,
node.attributes, mapping)
try:
self.conn.delete(ns_index_name, type, key)
except exceptions.NotFoundException:
pass
try:
self.conn.index(index_dict, ns_index_name, type, key)
except exceptions.ElasticSearchParseException:
pass
self.conn.refresh([ns_index_name])
def on_create(self, node):
type_indices = self.get_indices_of_type(node.type)
for ns_index_name in type_indices:
mapping = self.conn.get_mapping(node.type, ns_index_name)
index_dict = self.populate_index_document(node.type, ns_index_name,
node.attributes, mapping)
self.conn.index(index_dict, ns_index_name, node.type, node.key)
self.conn.refresh([ns_index_name])
def on_delete(self, node):
type_indices = self.get_indices_of_type(node.type)
for ns_index_name in type_indices:
try:
self.conn.delete(ns_index_name, node.type, node.key)
self.conn.refresh([ns_index_name])
except exceptions.NotFoundException:
pass
def on_modify(self, node):
type_indices = self.get_indices_of_type(node.type)
for ns_index_name in type_indices:
mapping = self.conn.get_mapping(node.type, ns_index_name)
index_dict = self.populate_index_document(node.type, ns_index_name,
node.attributes, mapping)
try:
self.conn.delete(ns_index_name, node.type, node.key)
self.conn.index(index_dict, ns_index_name, node.type, node.key)
self.conn.refresh([ns_index_name])
except exceptions.NotFoundException:
pass
def get_indices_of_type(self, type):
type_indices = []
for index in self.indices.keys():
if index.startswith(type + "-_-"):
type_indices.append(index)
return type_indices
def populate_index_document(self, type, ns_index_name, attributes,
mapping):
indexed_variables = mapping[type]['properties'].keys()
index_dict = {}
for arg in indexed_variables:
try:
index_dict[arg] = attributes[arg]
except KeyError:
#if this attribute doesn't exist for this node, just pass
pass
return index_dict
from pyes.es import ES
import random
import datetime
now = datetime.datetime.now()
es = ES()
for i in range(1000):
d = {'created_at': now - datetime.timedelta(seconds=i),
'level': 'ERROR' if random.random() < 0.1 else 'WARN',
'message': "Test message"}
es.index(d, 'my_index', 'my_type')
class FullTextSearch(object):
def __init__(self, server, settings=None):
self.conn = ES(server)
if settings:
self.settings = settings
else:
self.settings = {
'index': {
'analysis': {
'analyzer': {
'ngram_analyzer': {
'tokenizer': 'keyword',
'filter': ['lowercase', 'filter_ngram'],
'type': 'custom'
}
},
'filter': {
'filter_ngram': {
'type': 'nGram',
'max_gram': 30,
'min_gram': 1
}
}
}
}
}
self.refresh_index_cache()
def search_index_text(self, query_string, fields="_all", **args):
q = query.TextQuery(fields, query_string)
return self.search_index(q, **args)
def search_index(self, query, indices=None, num_results=None, node_type=None):
results = self.conn.search(
query=query, indices=indices, doc_types=node_type)
meta_list = [r.get_meta() for r in results[0:num_results]]
node_dict = {}
# fetch nodes grouped by type to reduce number of db calls
key = itemgetter('type')
for t, grouped_list in groupby(sorted(meta_list, key=key), key=key):
ids = [meta['id'] for meta in grouped_list]
for node in self.datastore.get_nodes(t, ids):
node_dict[(node.type, node.key)] = node
# return nodes in original order
nodelist = [node_dict[(meta['type'], meta['id'])]
for meta in meta_list]
return nodelist
def create_index(self, type, indexed_variables, index_name):
self.conn.create_index_if_missing(index_name, self.settings)
mapping = {}
for arg in indexed_variables:
mapping[arg] = {'boost': 1.0,
'analyzer': 'ngram_analyzer',
'type': 'string',
'term_vector': 'with_positions_offsets'}
index_settings = {'index_analyzer': 'ngram_analyzer',
'search_analyzer': 'standard',
'properties': mapping}
self.conn.put_mapping(str(type), index_settings, [index_name])
self.refresh_index_cache()
self.populate_index(type, index_name)
def refresh_index_cache(self):
try:
self.indices = self.conn.get_mapping()
except exceptions.IndexMissingException:
self.indices = {}
def delete_index(self, index_name):
self.conn.delete_index_if_exists(index_name)
self.refresh_index_cache()
def populate_index(self, type, index_name):
#add all the currently existing nodes into the index
ref_node = self.datastore.get_reference_node(type)
node_list = [rel.target_node for rel in ref_node.instance.outgoing]
for node in node_list:
key = node.key
index_dict = self.populate_index_document(node, index_name)
try:
self.conn.delete(index_name, type, key)
except exceptions.NotFoundException:
pass
self.conn.index(index_dict, index_name, type, key)
self.conn.refresh([index_name])
def on_create(self, node):
type_indices = self.get_indices_of_type(node.type)
for index_name in type_indices:
index_dict = self.populate_index_document(node, index_name)
self.conn.index(index_dict, index_name, node.type, node.key)
self.conn.refresh([index_name])
def on_delete(self, node):
type_indices = self.get_indices_of_type(node.type)
for index_name in type_indices:
try:
self.conn.delete(index_name, node.type, node.key)
self.conn.refresh([index_name])
except exceptions.NotFoundException:
pass
def on_modify(self, node):
type_indices = self.get_indices_of_type(node.type)
for index_name in type_indices:
index_dict = self.populate_index_document(node, index_name)
try:
self.conn.delete(index_name, node.type, node.key)
self.conn.index(index_dict, index_name, node.type, node.key)
self.conn.refresh([index_name])
except exceptions.NotFoundException:
pass
def get_indices_of_type(self, type):
type_indices = [
key for key, value in self.indices.items()
if type in value
]
return type_indices
def populate_index_document(self, node, index_name):
indexed_variables = self.indices[index_name][node.type]['properties'].keys()
index_dict = {
field: node[field] for field in indexed_variables
}
return index_dict
class ESDataTarget(base.DataTarget):
"""docstring for ClassName
"""
def __init__(self,
document_type,
database="test",
host="127.0.0.1",
port="9200",
truncate=False,
expand=False,
**elasticsearch_args):
"""Creates a ElasticSearch data target stream.
:Attributes:
* document_ElasticSearch elasticsearch document_type name
* database: database name
* host: ElasticSearch database server host, default is ``localhost``
* port: ElasticSearch port, default is ``9200``
* expand: expand dictionary values and treat children as top-level keys with dot '.'
separated key path to the child..
* truncate: delete existing data in the document_type. Default: False
"""
self.document_type = document_type
self.database_name = database
self.host = host
self.port = port
self.elasticsearch_args = elasticsearch_args
self.expand = expand
self.truncate = truncate
self._fields = None
def initialize(self):
"""Initialize ElasticSearch source stream:
"""
from pyes.es import ES
from pyes.exceptions import IndexAlreadyExistsException
args = self.elasticsearch_args.copy()
server = ""
if self.host:
server = self.host
if self.port:
server += ":" + self.port
create = args.pop("create", False)
replace = args.pop("replace", False)
self.connection = ES(server, **args)
self.connection.default_indices = self.database_name
self.connection.default_types = self.document_type
created = False
if create:
try:
self.connection.create_index(self.database_name)
self.connection.refresh(self.database_name)
created = True
except IndexAlreadyExistsException:
pass
if replace and not created:
self.connection.delete_index_if_exists(self.database_name)
time.sleep(2)
self.connection.create_index(self.database_name)
self.connection.refresh(self.database_name)
if self.truncate:
self.connection.delete_mapping(self.database_name,
self.document_type)
self.connection.refresh(self.database_name)
def append(self, obj):
record = obj
if not isinstance(obj, dict):
record = dict(zip(self.fields.names(), obj))
if self.expand:
record = expand_record(record)
id = record.get('id') or record.get('_id')
self.connection.index(record,
self.database_name,
self.document_type,
id,
bulk=True)
def finalize(self):
self.connection.flush_bulk(forced=True)