Switch Milvus for Qdrant in YAMLs (#43)

* Qdrant working

* - Fix missing prompt templates
- Bump version
- Add Qdrant to packages

* Switch Milvus for Qdrant in config files

trustgraph/base/consumer.py

@@ -10,7 +10,6 @@ class Consumer(BaseProcessor):
 
     def __init__(self, **params):
 
-        print("HERE2")
         super(Consumer, self).__init__(**params)
 
         input_queue = params.get("input_queue")
@@ -30,7 +29,6 @@ class Consumer(BaseProcessor):
                 'pubsub', 'Pub/sub configuration'
             )
 
-        print("HERE")
         if not hasattr(__class__, "processing_metric"):
             __class__.processing_metric = Counter(
                 'processing_count', 'Processing count', ["status"]

trustgraph/query/graph_embeddings/qdrant/__init__.py

@@ -0,0 +1,3 @@
+
+from . service import *
+

trustgraph/query/graph_embeddings/qdrant/__main__.py

@@ -0,0 +1,7 @@
+#!/usr/bin/env python3
+
+from . hf import run
+
+if __name__ == '__main__':
+    run()
+

trustgraph/query/graph_embeddings/qdrant/service.py

@@ -0,0 +1,133 @@
+
+"""
+Graph embeddings query service.  Input is vector, output is list of
+entities
+"""
+
+from qdrant_client import QdrantClient
+from qdrant_client.models import PointStruct
+from qdrant_client.models import Distance, VectorParams
+import uuid
+
+from .... schema import GraphEmbeddingsRequest, GraphEmbeddingsResponse
+from .... schema import Error, Value
+from .... schema import graph_embeddings_request_queue
+from .... schema import graph_embeddings_response_queue
+from .... base import ConsumerProducer
+
+module = ".".join(__name__.split(".")[1:-1])
+
+default_input_queue = graph_embeddings_request_queue
+default_output_queue = graph_embeddings_response_queue
+default_subscriber = module
+default_store_uri = 'http://localhost:6333'
+
+class Processor(ConsumerProducer):
+
+    def __init__(self, **params):
+
+        input_queue = params.get("input_queue", default_input_queue)
+        output_queue = params.get("output_queue", default_output_queue)
+        subscriber = params.get("subscriber", default_subscriber)
+        store_uri = params.get("store_uri", default_store_uri)
+
+        super(Processor, self).__init__(
+            **params | {
+                "input_queue": input_queue,
+                "output_queue": output_queue,
+                "subscriber": subscriber,
+                "input_schema": GraphEmbeddingsRequest,
+                "output_schema": GraphEmbeddingsResponse,
+                "store_uri": store_uri,
+            }
+        )
+
+        self.client = QdrantClient(url=store_uri)
+
+    def create_value(self, ent):
+        if ent.startswith("http://") or ent.startswith("https://"):
+            return Value(value=ent, is_uri=True)
+        else:
+            return Value(value=ent, is_uri=False)
+        
+    def handle(self, msg):
+
+        try:
+
+            v = msg.value()
+
+            # Sender-produced ID
+            id = msg.properties()["id"]
+
+            print(f"Handling input {id}...", flush=True)
+
+            entities = set()
+
+            for vec in v.vectors:
+
+                dim = len(vec)
+                collection = "triples_" + str(dim)
+
+                search_result = self.client.query_points(
+                    collection_name=collection,
+                    query=vec,
+                    limit=v.limit,
+                    with_payload=True,
+                ).points
+
+                for r in search_result:
+                    ent = r.payload["entity"]
+                    entities.add(ent)
+
+            # Convert set to list
+            entities = list(entities)
+
+            ents2 = []
+
+            for ent in entities:
+                ents2.append(self.create_value(ent))
+
+            entities = ents2
+
+            print("Send response...", flush=True)
+            r = GraphEmbeddingsResponse(entities=entities, error=None)
+            self.producer.send(r, properties={"id": id})
+
+            print("Done.", flush=True)
+
+        except Exception as e:
+
+            print(f"Exception: {e}")
+
+            print("Send error response...", flush=True)
+
+            r = GraphEmbeddingsResponse(
+                error=Error(
+                    type = "llm-error",
+                    message = str(e),
+                ),
+                entities=None,
+            )
+
+            self.producer.send(r, properties={"id": id})
+
+            self.consumer.acknowledge(msg)
+
+    @staticmethod
+    def add_args(parser):
+
+        ConsumerProducer.add_args(
+            parser, default_input_queue, default_subscriber,
+            default_output_queue,
+        )
+
+        parser.add_argument(
+            '-t', '--store-uri',
+            default=default_store_uri,
+            help=f'Milvus store URI (default: {default_store_uri})'
+        )
+
+def run():
+
+    Processor.start(module, __doc__)
+

trustgraph/storage/graph_embeddings/qdrant/__init__.py

@@ -0,0 +1,3 @@
+
+from . write import *
+

trustgraph/storage/graph_embeddings/qdrant/__main__.py

@@ -0,0 +1,7 @@
+#!/usr/bin/env python3
+
+from . write import run
+
+if __name__ == '__main__':
+    run()
+

trustgraph/storage/graph_embeddings/qdrant/write.py

@@ -0,0 +1,102 @@
+
+"""
+Accepts entity/vector pairs and writes them to a Qdrant store.
+"""
+
+from qdrant_client import QdrantClient
+from qdrant_client.models import PointStruct
+from qdrant_client.models import Distance, VectorParams
+import uuid
+
+from .... schema import GraphEmbeddings
+from .... schema import graph_embeddings_store_queue
+from .... log_level import LogLevel
+from .... base import Consumer
+
+module = ".".join(__name__.split(".")[1:-1])
+
+default_input_queue = graph_embeddings_store_queue
+default_subscriber = module
+default_store_uri = 'http://localhost:6333'
+
+class Processor(Consumer):
+
+    def __init__(self, **params):
+
+        input_queue = params.get("input_queue", default_input_queue)
+        subscriber = params.get("subscriber", default_subscriber)
+        store_uri = params.get("store_uri", default_store_uri)
+
+        super(Processor, self).__init__(
+            **params | {
+                "input_queue": input_queue,
+                "subscriber": subscriber,
+                "input_schema": GraphEmbeddings,
+                "store_uri": store_uri,
+            }
+        )
+
+        self.last_collection = None
+        self.last_dim = None
+
+        self.client = QdrantClient(url=store_uri)
+
+    def handle(self, msg):
+
+        v = msg.value()
+
+        if v.entity.value == "" or v.entity.value is None: return
+
+        for vec in v.vectors:
+
+            dim = len(vec)
+            collection = "triples_" + str(dim)
+
+            if dim != self.last_dim:
+
+                if not self.client.collection_exists(collection):
+
+                    try:
+                        self.client.create_collection(
+                            collection_name=collection,
+                            vectors_config=VectorParams(
+                                size=dim, distance=Distance.DOT
+                            ),
+                        )
+                    except Exception as e:
+                        print("Qdrant collection creation failed")
+                        raise e
+
+                self.last_collection = collection
+                self.last_dim = dim
+
+            self.client.upsert(
+                collection_name=collection,
+                points=[
+                    PointStruct(
+                        id=str(uuid.uuid4()),
+                        vector=vec,
+                        payload={
+                            "entity": v.entity.value,
+                        }
+                    )
+                ]
+            )
+
+    @staticmethod
+    def add_args(parser):
+
+        Consumer.add_args(
+            parser, default_input_queue, default_subscriber,
+        )
+
+        parser.add_argument(
+            '-t', '--store-uri',
+            default=default_store_uri,
+            help=f'Qdrant store URI (default: {default_store_uri})'
+        )
+
+def run():
+
+    Processor.start(module, __doc__)
+