Feature/librarian (#307)

* Bring QDrant up-to-date

* Tables for data from queue outputs

- Pass single Pulsar client to everything in gateway & librarian
- Pulsar listener-name support in gateway
- PDF and text load working in librarian

* Complete Cassandra schema

* Add librarian support to templates

trustgraph-flow/trustgraph/gateway/agent.py

@@ -7,10 +7,10 @@ from . endpoint import ServiceEndpoint
 from . requestor import ServiceRequestor
 
 class AgentRequestor(ServiceRequestor):
-    def __init__(self, pulsar_host, timeout, auth):
+    def __init__(self, pulsar_client, timeout, auth):
 
         super(AgentRequestor, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=agent_request_queue,
             response_queue=agent_response_queue,
             request_schema=AgentRequest,

trustgraph-flow/trustgraph/gateway/dbpedia.py

@@ -7,10 +7,10 @@ from . endpoint import ServiceEndpoint
 from . requestor import ServiceRequestor
 
 class DbpediaRequestor(ServiceRequestor):
-    def __init__(self, pulsar_host, timeout, auth):
+    def __init__(self, pulsar_client, timeout, auth):
 
         super(DbpediaRequestor, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=dbpedia_lookup_request_queue,
             response_queue=dbpedia_lookup_response_queue,
             request_schema=LookupRequest,

trustgraph-flow/trustgraph/gateway/document_embeddings_load.py

@@ -15,17 +15,17 @@ from . serialize import to_subgraph
 class DocumentEmbeddingsLoadEndpoint(SocketEndpoint):
 
     def __init__(
-            self, pulsar_host, auth, path="/api/v1/load/document-embeddings",
+            self, pulsar_client, auth, path="/api/v1/load/document-embeddings",
     ):
 
         super(DocumentEmbeddingsLoadEndpoint, self).__init__(
             endpoint_path=path, auth=auth,
         )
 
-        self.pulsar_host=pulsar_host
+        self.pulsar_client=pulsar_client
 
         self.publisher = Publisher(
-            self.pulsar_host, document_embeddings_store_queue,
+            self.pulsar_client, document_embeddings_store_queue,
             schema=JsonSchema(DocumentEmbeddings)
         )
 

trustgraph-flow/trustgraph/gateway/document_embeddings_stream.py

@@ -14,17 +14,18 @@ from . serialize import serialize_document_embeddings
 class DocumentEmbeddingsStreamEndpoint(SocketEndpoint):
 
     def __init__(
-            self, pulsar_host, auth, path="/api/v1/stream/document-embeddings"
+            self, pulsar_client, auth,
+            path="/api/v1/stream/document-embeddings"
     ):
 
         super(DocumentEmbeddingsStreamEndpoint, self).__init__(
             endpoint_path=path, auth=auth,
         )
 
-        self.pulsar_host=pulsar_host
+        self.pulsar_client=pulsar_client
 
         self.subscriber = Subscriber(
-            self.pulsar_host, document_embeddings_store_queue,
+            self.pulsar_client, document_embeddings_store_queue,
             "api-gateway", "api-gateway",
             schema=JsonSchema(DocumentEmbeddings)
         )

trustgraph-flow/trustgraph/gateway/document_load.py

@@ -8,10 +8,10 @@ from . sender import ServiceSender
 from . serialize import to_subgraph
 
 class DocumentLoadSender(ServiceSender):
-    def __init__(self, pulsar_host):
+    def __init__(self, pulsar_client):
 
         super(DocumentLoadSender, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=document_ingest_queue,
             request_schema=Document,
         )

trustgraph-flow/trustgraph/gateway/document_rag.py

@@ -7,10 +7,10 @@ from . endpoint import ServiceEndpoint
 from . requestor import ServiceRequestor
 
 class DocumentRagRequestor(ServiceRequestor):
-    def __init__(self, pulsar_host, timeout, auth):
+    def __init__(self, pulsar_client, timeout, auth):
 
         super(DocumentRagRequestor, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=document_rag_request_queue,
             response_queue=document_rag_response_queue,
             request_schema=DocumentRagQuery,

trustgraph-flow/trustgraph/gateway/embeddings.py

@@ -7,10 +7,10 @@ from . endpoint import ServiceEndpoint
 from . requestor import ServiceRequestor
 
 class EmbeddingsRequestor(ServiceRequestor):
-    def __init__(self, pulsar_host, timeout, auth):
+    def __init__(self, pulsar_client, timeout, auth):
 
         super(EmbeddingsRequestor, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=embeddings_request_queue,
             response_queue=embeddings_response_queue,
             request_schema=EmbeddingsRequest,

trustgraph-flow/trustgraph/gateway/encyclopedia.py

@@ -7,10 +7,10 @@ from . endpoint import ServiceEndpoint
 from . requestor import ServiceRequestor
 
 class EncyclopediaRequestor(ServiceRequestor):
-    def __init__(self, pulsar_host, timeout, auth):
+    def __init__(self, pulsar_client, timeout, auth):
 
         super(EncyclopediaRequestor, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=encyclopedia_lookup_request_queue,
             response_queue=encyclopedia_lookup_response_queue,
             request_schema=LookupRequest,

trustgraph-flow/trustgraph/gateway/graph_embeddings_load.py

@@ -15,17 +15,17 @@ from . serialize import to_subgraph, to_value
 class GraphEmbeddingsLoadEndpoint(SocketEndpoint):
 
     def __init__(
-            self, pulsar_host, auth, path="/api/v1/load/graph-embeddings",
+            self, pulsar_client, auth, path="/api/v1/load/graph-embeddings",
     ):
 
         super(GraphEmbeddingsLoadEndpoint, self).__init__(
             endpoint_path=path, auth=auth,
         )
 
-        self.pulsar_host=pulsar_host
+        self.pulsar_client=pulsar_client
 
         self.publisher = Publisher(
-            self.pulsar_host, graph_embeddings_store_queue,
+            self.pulsar_client, graph_embeddings_store_queue,
             schema=JsonSchema(GraphEmbeddings)
         )
 

trustgraph-flow/trustgraph/gateway/graph_embeddings_query.py

@@ -8,10 +8,10 @@ from . requestor import ServiceRequestor
 from . serialize import serialize_value
 
 class GraphEmbeddingsQueryRequestor(ServiceRequestor):
-    def __init__(self, pulsar_host, timeout, auth):
+    def __init__(self, pulsar_client, timeout, auth):
 
         super(GraphEmbeddingsQueryRequestor, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=graph_embeddings_request_queue,
             response_queue=graph_embeddings_response_queue,
             request_schema=GraphEmbeddingsRequest,

trustgraph-flow/trustgraph/gateway/graph_embeddings_stream.py

@@ -14,17 +14,17 @@ from . serialize import serialize_graph_embeddings
 class GraphEmbeddingsStreamEndpoint(SocketEndpoint):
 
     def __init__(
-            self, pulsar_host, auth, path="/api/v1/stream/graph-embeddings"
+            self, pulsar_client, auth, path="/api/v1/stream/graph-embeddings"
     ):
 
         super(GraphEmbeddingsStreamEndpoint, self).__init__(
             endpoint_path=path, auth=auth,
         )
 
-        self.pulsar_host=pulsar_host
+        self.pulsar_client=pulsar_client
 
         self.subscriber = Subscriber(
-            self.pulsar_host, graph_embeddings_store_queue,
+            self.pulsar_client, graph_embeddings_store_queue,
             "api-gateway", "api-gateway",
             schema=JsonSchema(GraphEmbeddings)
         )

trustgraph-flow/trustgraph/gateway/graph_rag.py

@@ -7,10 +7,10 @@ from . endpoint import ServiceEndpoint
 from . requestor import ServiceRequestor
 
 class GraphRagRequestor(ServiceRequestor):
-    def __init__(self, pulsar_host, timeout, auth):
+    def __init__(self, pulsar_client, timeout, auth):
 
         super(GraphRagRequestor, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=graph_rag_request_queue,
             response_queue=graph_rag_response_queue,
             request_schema=GraphRagQuery,

trustgraph-flow/trustgraph/gateway/internet_search.py

@@ -7,10 +7,10 @@ from . endpoint import ServiceEndpoint
 from . requestor import ServiceRequestor
 
 class InternetSearchRequestor(ServiceRequestor):
-    def __init__(self, pulsar_host, timeout, auth):
+    def __init__(self, pulsar_client, timeout, auth):
 
         super(InternetSearchRequestor, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=internet_search_request_queue,
             response_queue=internet_search_response_queue,
             request_schema=LookupRequest,

trustgraph-flow/trustgraph/gateway/librarian.py

@@ -9,10 +9,10 @@ from . serialize import serialize_document_package, serialize_document_info
 from . serialize import to_document_package, to_document_info, to_criteria
 
 class LibrarianRequestor(ServiceRequestor):
-    def __init__(self, pulsar_host, timeout, auth):
+    def __init__(self, pulsar_client, timeout, auth):
 
         super(LibrarianRequestor, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=librarian_request_queue,
             response_queue=librarian_response_queue,
             request_schema=LibrarianRequest,
@@ -22,17 +22,19 @@ class LibrarianRequestor(ServiceRequestor):
 
     def to_request(self, body):
 
+        print("TRR")
         if "document" in body:
             dp = to_document_package(body["document"])
         else:
             dp = None
 
+        print("GOT")
         if "criteria" in body:
             criteria = to_criteria(body["criteria"])
         else:
             criteria = None
 
-        limit = int(body.get("limit", 10000))
+        print("ASLDKJ")
 
         return LibrarianRequest(
             operation = body.get("operation", None),

trustgraph-flow/trustgraph/gateway/mux.py

@@ -18,7 +18,7 @@ MAX_QUEUE_SIZE = 10
 class MuxEndpoint(SocketEndpoint):
 
     def __init__(
-            self, pulsar_host, auth,
+            self, pulsar_client, auth,
             services,
             path="/api/v1/socket",
     ):

trustgraph-flow/trustgraph/gateway/prompt.py

@@ -9,10 +9,10 @@ from . endpoint import ServiceEndpoint
 from . requestor import ServiceRequestor
 
 class PromptRequestor(ServiceRequestor):
-    def __init__(self, pulsar_host, timeout, auth):
+    def __init__(self, pulsar_client, timeout, auth):
 
         super(PromptRequestor, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=prompt_request_queue,
             response_queue=prompt_response_queue,
             request_schema=PromptRequest,

trustgraph-flow/trustgraph/gateway/requestor.py

@@ -14,7 +14,7 @@ class ServiceRequestor:
 
     def __init__(
             self,
-            pulsar_host,
+            pulsar_client,
             request_queue, request_schema,
             response_queue, response_schema,
             subscription="api-gateway", consumer_name="api-gateway",
@@ -22,12 +22,12 @@ class ServiceRequestor:
     ):
 
         self.pub = Publisher(
-            pulsar_host, request_queue,
-            schema=JsonSchema(request_schema)
+            pulsar_client, request_queue,
+            schema=JsonSchema(request_schema),
         )
 
         self.sub = Subscriber(
-            pulsar_host, response_queue,
+            pulsar_client, response_queue,
             subscription, consumer_name,
             JsonSchema(response_schema)
         )
@@ -53,9 +53,11 @@ class ServiceRequestor:
 
             q = self.sub.subscribe(id)
 
+            print("BOUT TO SEDN")
             await asyncio.to_thread(
                 self.pub.send, id, self.to_request(request)
             )
+            print("SENT")
 
             while True:
 

trustgraph-flow/trustgraph/gateway/sender.py

@@ -15,13 +15,13 @@ class ServiceSender:
 
     def __init__(
             self,
-            pulsar_host,
+            pulsar_client,
             request_queue, request_schema,
     ):
 
         self.pub = Publisher(
-            pulsar_host, request_queue,
-            schema=JsonSchema(request_schema)
+            pulsar_client, request_queue,
+            schema=JsonSchema(request_schema),
         )
 
     async def start(self):
@@ -53,4 +53,3 @@ class ServiceSender:
 
             return err
 
-

trustgraph-flow/trustgraph/gateway/serialize.py

@@ -126,7 +126,7 @@ def to_document_package(x):
 
     return DocumentPackage(
         metadata = to_subgraph(x["metadata"]),
-        document = base64.b64decode(x["document"].encode("utf-8")),
+        document = x.get("document", None),
         kind = x.get("kind", None),
         user = x.get("user", None),
         collection = x.get("collection", None),

trustgraph-flow/trustgraph/gateway/service.py

@@ -73,6 +73,11 @@ class Api:
         self.port = int(config.get("port", default_port))
         self.timeout = int(config.get("timeout", default_timeout))
         self.pulsar_host = config.get("pulsar_host", default_pulsar_host)
+        self.pulsar_listener = config.get("pulsar_listener", None)
+
+        self.pulsar_client = pulsar.Client(
+            self.pulsar_host, listener_name=self.pulsar_listener
+        )
 
         self.prometheus_url = config.get(
             "prometheus_url", default_prometheus_url,
@@ -91,58 +96,58 @@ class Api:
 
         self.services = {
             "text-completion": TextCompletionRequestor(
-                pulsar_host=self.pulsar_host, timeout=self.timeout,
+                pulsar_client=self.pulsar_client, timeout=self.timeout,
                 auth = self.auth,
             ),
             "prompt": PromptRequestor(
-                pulsar_host=self.pulsar_host, timeout=self.timeout,
+                pulsar_client=self.pulsar_client, timeout=self.timeout,
                 auth = self.auth,
             ),
             "graph-rag": GraphRagRequestor(
-                pulsar_host=self.pulsar_host, timeout=self.timeout,
+                pulsar_client=self.pulsar_client, timeout=self.timeout,
                 auth = self.auth,
             ),
             "document-rag": DocumentRagRequestor(
-                pulsar_host=self.pulsar_host, timeout=self.timeout,
+                pulsar_client=self.pulsar_client, timeout=self.timeout,
                 auth = self.auth,
             ),
             "triples-query": TriplesQueryRequestor(
-                pulsar_host=self.pulsar_host, timeout=self.timeout,
+                pulsar_client=self.pulsar_client, timeout=self.timeout,
                 auth = self.auth,
             ),
             "graph-embeddings-query": GraphEmbeddingsQueryRequestor(
-                pulsar_host=self.pulsar_host, timeout=self.timeout,
+                pulsar_client=self.pulsar_client, timeout=self.timeout,
                 auth = self.auth,
             ),
             "embeddings": EmbeddingsRequestor(
-                pulsar_host=self.pulsar_host, timeout=self.timeout,
+                pulsar_client=self.pulsar_client, timeout=self.timeout,
                 auth = self.auth,
             ),
             "agent": AgentRequestor(
-                pulsar_host=self.pulsar_host, timeout=self.timeout,
+                pulsar_client=self.pulsar_client, timeout=self.timeout,
                 auth = self.auth,
             ),
             "librarian": LibrarianRequestor(
-                pulsar_host=self.pulsar_host, timeout=self.timeout,
+                pulsar_client=self.pulsar_client, timeout=self.timeout,
                 auth = self.auth,
             ),
             "encyclopedia": EncyclopediaRequestor(
-                pulsar_host=self.pulsar_host, timeout=self.timeout,
+                pulsar_client=self.pulsar_client, timeout=self.timeout,
                 auth = self.auth,
             ),
             "dbpedia": DbpediaRequestor(
-                pulsar_host=self.pulsar_host, timeout=self.timeout,
+                pulsar_client=self.pulsar_client, timeout=self.timeout,
                 auth = self.auth,
             ),
             "internet-search": InternetSearchRequestor(
-                pulsar_host=self.pulsar_host, timeout=self.timeout,
+                pulsar_client=self.pulsar_client, timeout=self.timeout,
                 auth = self.auth,
             ),
             "document-load": DocumentLoadSender(
-                pulsar_host=self.pulsar_host,
+                pulsar_client=self.pulsar_client,
             ),
             "text-load": TextLoadSender(
-                pulsar_host=self.pulsar_host,
+                pulsar_client=self.pulsar_client,
             ),
         }
 
@@ -205,31 +210,31 @@ class Api:
                 requestor = self.services["text-load"],
             ),
             TriplesStreamEndpoint(
-                pulsar_host=self.pulsar_host,
+                pulsar_client=self.pulsar_client,
                 auth = self.auth,
             ),
             GraphEmbeddingsStreamEndpoint(
-                pulsar_host=self.pulsar_host,
+                pulsar_client=self.pulsar_client,
                 auth = self.auth,
             ),
             DocumentEmbeddingsStreamEndpoint(
-                pulsar_host=self.pulsar_host,
+                pulsar_client=self.pulsar_client,
                 auth = self.auth,
             ),
             TriplesLoadEndpoint(
-                pulsar_host=self.pulsar_host,
+                pulsar_client=self.pulsar_client,
                 auth = self.auth,
             ),
             GraphEmbeddingsLoadEndpoint(
-                pulsar_host=self.pulsar_host,
+                pulsar_client=self.pulsar_client,
                 auth = self.auth,
             ),
             DocumentEmbeddingsLoadEndpoint(
-                pulsar_host=self.pulsar_host,
+                pulsar_client=self.pulsar_client,
                 auth = self.auth,
             ),
             MuxEndpoint(
-                pulsar_host=self.pulsar_host,
+                pulsar_client=self.pulsar_client,
                 auth = self.auth,
                 services = self.services,
             ),
@@ -266,6 +271,11 @@ def run():
         help=f'Pulsar host (default: {default_pulsar_host})',
     )
 
+    parser.add_argument(
+        '--pulsar-listener',
+        help=f'Pulsar listener (default: none)',
+    )
+
     parser.add_argument(
         '-m', '--prometheus-url',
         default=default_prometheus_url,

trustgraph-flow/trustgraph/gateway/text_completion.py

@@ -7,10 +7,10 @@ from . endpoint import ServiceEndpoint
 from . requestor import ServiceRequestor
 
 class TextCompletionRequestor(ServiceRequestor):
-    def __init__(self, pulsar_host, timeout, auth):
+    def __init__(self, pulsar_client, timeout, auth):
 
         super(TextCompletionRequestor, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=text_completion_request_queue,
             response_queue=text_completion_response_queue,
             request_schema=TextCompletionRequest,

trustgraph-flow/trustgraph/gateway/text_load.py

@@ -8,10 +8,10 @@ from . sender import ServiceSender
 from . serialize import to_subgraph
 
 class TextLoadSender(ServiceSender):
-    def __init__(self, pulsar_host):
+    def __init__(self, pulsar_client):
 
         super(TextLoadSender, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=text_ingest_queue,
             request_schema=TextDocument,
         )

trustgraph-flow/trustgraph/gateway/triples_load.py

@@ -14,16 +14,16 @@ from . serialize import to_subgraph
 
 class TriplesLoadEndpoint(SocketEndpoint):
 
-    def __init__(self, pulsar_host, auth, path="/api/v1/load/triples"):
+    def __init__(self, pulsar_client, auth, path="/api/v1/load/triples"):
 
         super(TriplesLoadEndpoint, self).__init__(
             endpoint_path=path, auth=auth,
         )
 
-        self.pulsar_host=pulsar_host
+        self.pulsar_client=pulsar_client
 
         self.publisher = Publisher(
-            self.pulsar_host, triples_store_queue,
+            self.pulsar_client, triples_store_queue,
             schema=JsonSchema(Triples)
         )
 

trustgraph-flow/trustgraph/gateway/triples_query.py

@@ -8,10 +8,10 @@ from . requestor import ServiceRequestor
 from . serialize import to_value, serialize_subgraph
 
 class TriplesQueryRequestor(ServiceRequestor):
-    def __init__(self, pulsar_host, timeout, auth):
+    def __init__(self, pulsar_client, timeout, auth):
 
         super(TriplesQueryRequestor, self).__init__(
-            pulsar_host=pulsar_host,
+            pulsar_client=pulsar_client,
             request_queue=triples_request_queue,
             response_queue=triples_response_queue,
             request_schema=TriplesQueryRequest,

trustgraph-flow/trustgraph/gateway/triples_stream.py

@@ -13,16 +13,16 @@ from . serialize import serialize_triples
 
 class TriplesStreamEndpoint(SocketEndpoint):
 
-    def __init__(self, pulsar_host, auth, path="/api/v1/stream/triples"):
+    def __init__(self, pulsar_client, auth, path="/api/v1/stream/triples"):
 
         super(TriplesStreamEndpoint, self).__init__(
             endpoint_path=path, auth=auth,
         )
 
-        self.pulsar_host=pulsar_host
+        self.pulsar_client=pulsar_client
 
         self.subscriber = Subscriber(
-            self.pulsar_host, triples_store_queue,
+            self.pulsar_client, triples_store_queue,
             "api-gateway", "api-gateway",
             schema=JsonSchema(Triples)
         )