trustgraph-flow 0.21.2__tar.gz → 0.21.4__tar.gz

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: trustgraph-flow
-Version: 0.21.2
+Version: 0.21.4
 Summary: TrustGraph provides a means to run a pipeline of flexible AI processing components in a flexible means to achieve a processing pipeline.
 Home-page: https://github.com/trustgraph-ai/trustgraph
-Download-URL: https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v0.21.2.tar.gz
+Download-URL: https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v0.21.4.tar.gz
 Author: trustgraph.ai
 Author-email: security@trustgraph.ai
 Classifier: Programming Language :: Python :: 3

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/agent/react/service.py

@@ -156,14 +156,16 @@ class Processor(ConsumerProducer):
             subscriber=subscriber,
             input_queue=prompt_request_queue,
             output_queue=prompt_response_queue,
-            pulsar_host = self.pulsar_host
+            pulsar_host = self.pulsar_host,
+            pulsar_api_key=self.pulsar_api_key,
         )
 
         self.graph_rag = GraphRagClient(
             subscriber=subscriber,
             input_queue=graph_rag_request_queue,
             output_queue=graph_rag_response_queue,
-            pulsar_host = self.pulsar_host
+            pulsar_host = self.pulsar_host,
+            pulsar_api_key=self.pulsar_api_key,
         )
 
         # Need to be able to feed requests to myself
@@ -218,7 +220,7 @@ class Processor(ConsumerProducer):
 
             print(f"History: {history}", flush=True)
 
-            def think(x):
+            async def think(x):
 
                 print(f"Think: {x}", flush=True)
 
@@ -231,7 +233,7 @@ class Processor(ConsumerProducer):
 
                 await self.producer.send(r, properties={"id": id})
 
-            def observe(x):
+            async def observe(x):
 
                 print(f"Observe: {x}", flush=True)
 

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/document_rag.py

@@ -59,6 +59,7 @@ class DocumentRag:
     def __init__(
             self,
             pulsar_host="pulsar://pulsar:6650",
+            pulsar_api_key=None,
             pr_request_queue=None,
             pr_response_queue=None,
             emb_request_queue=None,
@@ -100,6 +101,7 @@ class DocumentRag:
             subscriber=module + "-de",
             input_queue=de_request_queue,
             output_queue=de_response_queue,
+            pulsar_api_key=pulsar_api_key,
         )            
 
         self.embeddings = EmbeddingsClient(
@@ -107,6 +109,7 @@ class DocumentRag:
             input_queue=emb_request_queue,
             output_queue=emb_response_queue,
             subscriber=module + "-emb",
+            pulsar_api_key=pulsar_api_key,
         )
 
         self.lang = PromptClient(
@@ -114,6 +117,7 @@ class DocumentRag:
             input_queue=pr_request_queue,
             output_queue=pr_response_queue,
             subscriber=module + "-de-prompt",
+            pulsar_api_key=pulsar_api_key,
         )
 
         if self.verbose:

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/embeddings/document_embeddings/embeddings.py

@@ -47,6 +47,7 @@ class Processor(ConsumerProducer):
 
         self.embeddings = EmbeddingsClient(
             pulsar_host=self.pulsar_host,
+            pulsar_api_key=self.pulsar_api_key,
             input_queue=emb_request_queue,
             output_queue=emb_response_queue,
             subscriber=module + "-emb",

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/extract/kg/definitions/extract.py

@@ -79,6 +79,7 @@ class Processor(ConsumerProducer):
 
         self.prompt = PromptClient(
             pulsar_host=self.pulsar_host,
+            pulsar_api_key=self.pulsar_api_key,
             input_queue=pr_request_queue,
             output_queue=pr_response_queue,
             subscriber = module + "-prompt",

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/extract/kg/relationships/extract.py

@@ -54,6 +54,7 @@ class Processor(ConsumerProducer):
 
         self.prompt = PromptClient(
             pulsar_host=self.pulsar_host,
+            pulsar_api_key=self.pulsar_api_key,
             input_queue=pr_request_queue,
             output_queue=pr_response_queue,
             subscriber = module + "-prompt",

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/extract/kg/topics/extract.py

@@ -52,6 +52,7 @@ class Processor(ConsumerProducer):
 
         self.prompt = PromptClient(
             pulsar_host=self.pulsar_host,
+            pulsar_api_key=self.pulsar_api_key,
             input_queue=pr_request_queue,
             output_queue=pr_response_queue,
             subscriber = module + "-prompt",

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/extract/object/row/extract.py

@@ -112,6 +112,7 @@ class Processor(ConsumerProducer):
 
         self.prompt = PromptClient(
             pulsar_host=self.pulsar_host,
+            pulsar_api_key=self.pulsar_api_key,
             input_queue=pr_request_queue,
             output_queue=pr_response_queue,
             subscriber = module + "-prompt",

trustgraph-flow-0.21.4/trustgraph/flow_version.py

@@ -0,0 +1 @@
+__version__ = "0.21.4"

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/gateway/document_embeddings_stream.py

@@ -27,7 +27,7 @@ class DocumentEmbeddingsStreamEndpoint(SocketEndpoint):
         self.subscriber = Subscriber(
             self.pulsar_client, document_embeddings_store_queue,
             "api-gateway", "api-gateway",
-            schema=JsonSchema(DocumentEmbeddings)
+            schema=JsonSchema(DocumentEmbeddings),
         )
 
     async def listener(self, ws, running):

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/gateway/librarian.py

@@ -53,7 +53,10 @@ class LibrarianRequestor(ServiceRequestor):
             response["document"] = serialize_document_package(message.document)
 
         if message.info:
-            response["info"] = serialize_document_info(message.info)
+            response["info"] = [
+                serialize_document_info(v)
+                for v in message.info
+            ]
         
         return response, True
 

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/gateway/serialize.py

@@ -84,6 +84,9 @@ def serialize_document_package(message):
 
     ret = {}
 
+    if message.id:
+        ret["id"] = message.id
+
     if message.metadata:
         ret["metadata"] = serialize_subgraph(message.metdata)
 
@@ -108,8 +111,8 @@ def serialize_document_info(message):
 
     ret = {}
 
-    if message.metadata:
-        ret["metadata"] = serialize_subgraph(message.metdata)
+    if message.id:
+        ret["id"] = message.id
 
     if message.kind:
         ret["kind"] = message.kind
@@ -120,25 +123,45 @@ def serialize_document_info(message):
     if message.collection:
         ret["collection"] = message.collection
 
+    if message.title:
+        ret["title"] = message.title
+
+    if message.comments:
+        ret["comments"] = message.comments
+
+    if message.time:
+        ret["time"] = message.time
+
+    if message.metadata:
+        ret["metadata"] = serialize_subgraph(message.metadata)
+
     return ret
 
 def to_document_package(x):
 
     return DocumentPackage(
-        metadata = to_subgraph(x["metadata"]),
-        document = x.get("document", None),
+        id = x.get("id", None),
         kind = x.get("kind", None),
         user = x.get("user", None),
         collection = x.get("collection", None),
+        title = x.get("title", None),
+        comments = x.get("comments", None),
+        time = x.get("time", None),
+        document = x.get("document", None),
+        metadata = to_subgraph(x["metadata"]),
     )
 
 def to_document_info(x):
 
     return DocumentInfo(
-        metadata = to_subgraph(x["metadata"]),
+        id = x.get("id", None),
         kind = x.get("kind", None),
         user = x.get("user", None),
         collection = x.get("collection", None),
+        title = x.get("title", None),
+        comments = x.get("comments", None),
+        time = x.get("time", None),
+        metadata = to_subgraph(x["metadata"]),
     )
 
 def to_criteria(x):

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/gateway/service.py

@@ -57,6 +57,7 @@ logger.setLevel(logging.INFO)
 
 default_pulsar_host = os.getenv("PULSAR_HOST", "pulsar://pulsar:6650")
 default_prometheus_url = os.getenv("PROMETHEUS_URL", "http://prometheus:9090")
+default_pulsar_api_key = os.getenv("PULSAR_API_KEY", None)
 default_timeout = 600
 default_port = 8088
 default_api_token = os.getenv("GATEWAY_SECRET", "")
@@ -73,11 +74,20 @@ 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_api_key = config.get(
+            "pulsar_api_key", default_pulsar_api_key
+        )
         self.pulsar_listener = config.get("pulsar_listener", None)
 
-        self.pulsar_client = pulsar.Client(
-            self.pulsar_host, listener_name=self.pulsar_listener
-        )
+        if self.pulsar_api_key:
+            self.pulsar_client = pulsar.Client(
+                self.pulsar_host, listener_name=self.pulsar_listener,
+                authentication=pulsar.AuthenticationToken(self.pulsar_api_key)
+            )
+        else:
+            self.pulsar_client = pulsar.Client(
+                self.pulsar_host, listener_name=self.pulsar_listener,
+            )
 
         self.prometheus_url = config.get(
             "prometheus_url", default_prometheus_url,
@@ -270,6 +280,12 @@ def run():
         default=default_pulsar_host,
         help=f'Pulsar host (default: {default_pulsar_host})',
     )
+    
+    parser.add_argument(
+        '--pulsar-api-key',
+        default=default_pulsar_api_key,
+        help=f'Pulsar API key',
+    )
 
     parser.add_argument(
         '--pulsar-listener',

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/graph_rag.py

@@ -161,6 +161,7 @@ class GraphRag:
     def __init__(
             self,
             pulsar_host="pulsar://pulsar:6650",
+            pulsar_api_key=None,
             pr_request_queue=None,
             pr_response_queue=None,
             emb_request_queue=None,
@@ -207,6 +208,7 @@ class GraphRag:
 
         self.ge_client = GraphEmbeddingsClient(
             pulsar_host=pulsar_host,
+            pulsar_api_key=pulsar_api_key,
             subscriber=module + "-ge",
             input_queue=ge_request_queue,
             output_queue=ge_response_queue,
@@ -214,6 +216,7 @@ class GraphRag:
 
         self.triples_client = TriplesQueryClient(
             pulsar_host=pulsar_host,
+            pulsar_api_key=pulsar_api_key,
             subscriber=module + "-tpl",
             input_queue=tpl_request_queue,
             output_queue=tpl_response_queue
@@ -221,6 +224,7 @@ class GraphRag:
 
         self.embeddings = EmbeddingsClient(
             pulsar_host=pulsar_host,
+            pulsar_api_key=pulsar_api_key,
             input_queue=emb_request_queue,
             output_queue=emb_response_queue,
             subscriber=module + "-emb",
@@ -234,6 +238,7 @@ class GraphRag:
 
         self.prompt = PromptClient(
             pulsar_host=pulsar_host,
+            pulsar_api_key=pulsar_api_key,
             input_queue=pr_request_queue,
             output_queue=pr_response_queue,
             subscriber=module + "-prompt",

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/librarian/librarian.py

@@ -1,4 +1,4 @@
-from .. schema import LibrarianRequest, LibrarianResponse, Error
+from .. schema import LibrarianRequest, LibrarianResponse, Error, Triple
 from .. knowledge import hash
 from .. exceptions import RequestError
 from . table_store import TableStore
@@ -26,7 +26,7 @@ class Librarian:
         self.load_document = load_document
         self.load_text = load_text
 
-    async def add(self, id, document):
+    async def add(self, document):
 
         if document.kind not in (
                 "text/plain", "application/pdf"
@@ -41,9 +41,9 @@ class Librarian:
         self.table_store.add(object_id, document)
 
         if document.kind == "application/pdf":
-            await self.load_document(id, document)
+            await self.load_document(document)
         elif document.kind == "text/plain":
-            await self.load_text(id, document)
+            await self.load_text(document)
 
         print("Add complete", flush=True)
 
@@ -53,6 +53,20 @@ class Librarian:
             info = None,
         )
 
+    async def list(self, user, collection):
+
+        print("list")
+
+        info = self.table_store.list(user, collection)
+
+        print(">>", info)
+
+        return LibrarianResponse(
+            error = None,
+            document = None,
+            info = info,
+        )
+
     def handle_triples(self, m):
         self.table_store.add_triples(m)
 

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/librarian/service.py

@@ -223,11 +223,11 @@ class Processor(ConsumerProducer):
 
             self.librarian.handle_document_embeddings(msg)
 
-    async def load_document(self, id, document):
+    async def load_document(self, document):
 
         doc = Document(
             metadata = Metadata(
-                id = id,
+                id = document.id,
                 metadata = document.metadata,
                 user = document.user,
                 collection = document.collection
@@ -237,14 +237,14 @@ class Processor(ConsumerProducer):
 
         self.document_load.send(None, doc)
 
-    async def load_text(self, id, document):
+    async def load_text(self, document):
 
         text = base64.b64decode(document.document)
         text = text.decode("utf-8")
 
         doc = TextDocument(
             metadata = Metadata(
-                id = id,
+                id = document.id,
                 metadata = document.metadata,
                 user = document.user,
                 collection = document.collection
@@ -259,19 +259,33 @@ class Processor(ConsumerProducer):
         if v.operation is None:
             raise RequestError("Null operation")
 
+        print("op", v.operation)
+
         if v.operation == "add":
             if (
-                    v.id and v.document and v.document.metadata and
+                    v.document and v.document.id and v.document.metadata and
                     v.document.document and v.document.kind
             ):
                 return partial(
                     self.librarian.add,
-                    id = v.id,
                     document = v.document,
                 )
             else:
                 raise RequestError("Invalid call")
 
+        if v.operation == "list":
+            print("list", v)
+            print(v.user)
+            if v.user:
+                return partial(
+                    self.librarian.list,
+                    user = v.user,
+                    collection = v.collection,
+                )
+            else:
+                print("BROK")
+                raise RequestError("Invalid call")
+
         raise RequestError("Invalid operation: " + v.operation)
 
     async def handle(self, msg):
@@ -298,6 +312,7 @@ class Processor(ConsumerProducer):
 
         try:
             resp = await func()
+            print("->", resp)
         except RequestError as e:
             resp = LibrarianResponse(
                 error = Error(
@@ -318,7 +333,7 @@ class Processor(ConsumerProducer):
             await self.send(resp, properties={"id": id})
             return
 
-        print("Send response...", flush=True)
+        print("Send response..!.", flush=True)
 
         await self.send(resp, properties={"id": id})
 

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/librarian/table_store.py

@@ -1,4 +1,5 @@
-from .. schema import LibrarianRequest, LibrarianResponse, Error
+from .. schema import LibrarianRequest, LibrarianResponse
+from .. schema import DocumentInfo, Error, Triple, Value
 from .. knowledge import hash
 from .. exceptions import RequestError
 
@@ -61,7 +62,7 @@ class TableStore:
             CREATE TABLE IF NOT EXISTS document (
                 user text,
                 collection text,
-                id uuid,
+                id text,
                 time timestamp,
                 title text,
                 comments text,
@@ -157,6 +158,20 @@ class TableStore:
             VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)
         """)
 
+        self.list_document_stmt = self.cassandra.prepare("""
+            SELECT
+                id, kind, user, collection, title, comments, time, metadata
+            FROM document
+            WHERE user = ?
+        """)
+
+        self.list_document_by_collection_stmt = self.cassandra.prepare("""
+            SELECT
+                id, kind, user, collection, title, comments, time, metadata
+            FROM document
+            WHERE user = ? AND collection = ?
+        """)
+
         self.insert_triples_stmt = self.cassandra.prepare("""
             INSERT INTO triples
             (
@@ -192,10 +207,9 @@ class TableStore:
             raise RequestError("Invalid document kind: " + document.kind)
 
         # Create random doc ID
-        doc_id = uuid.uuid4()
         when = int(time.time() * 1000)
 
-        print("Adding", object_id, doc_id)
+        print("Adding", document.id, object_id)
 
         metadata = [
             (
@@ -205,8 +219,6 @@ class TableStore:
             for v in document.metadata
         ]
 
-        # FIXME: doc_id should be the user-supplied ID???
-
         while True:
 
             try:
@@ -214,7 +226,7 @@ class TableStore:
                 resp = self.cassandra.execute(
                     self.insert_document_stmt,
                     (
-                        doc_id, document.user, document.collection,
+                        document.id, document.user, document.collection,
                         document.kind, object_id, when,
                         document.title, document.comments,
                         metadata
@@ -275,6 +287,73 @@ class TableStore:
                 print(f"{e}, retry...", flush=True)
                 time.sleep(1)
 
+    def list(self, user, collection=None):
+
+        print("LIST")
+        while True:
+
+            print("TRY")
+
+            print(self.list_document_stmt)
+            try:
+
+                if collection:
+                    resp = self.cassandra.execute(
+                        self.list_document_by_collection_stmt,
+                        (user, collection)
+                    )
+                else:
+                    resp = self.cassandra.execute(
+                        self.list_document_stmt,
+                        (user,)
+                    )
+                break
+
+                print("OK")
+
+            except Exception as e:
+                print("Exception:", type(e))
+                print(f"{e}, retry...", flush=True)
+                time.sleep(1)
+
+        print("OK2")
+
+        info = [
+            DocumentInfo(
+                id = row[0],
+                kind = row[1],
+                user = row[2],
+                collection = row[3],
+                title = row[4],
+                comments = row[5],
+                time = int(1000 * row[6].timestamp()),
+                metadata = [
+                    Triple(
+                        s=Value(value=m[0], is_uri=m[1]),
+                        p=Value(value=m[2], is_uri=m[3]),
+                        o=Value(value=m[4], is_uri=m[5])
+                    )
+                    for m in row[7]
+                ],
+            )
+            for row in resp
+        ]
+
+        print("OK3")
+
+        print(info[0])
+
+        print(info[0].user)
+        print(info[0].time)
+        print(info[0].kind)
+        print(info[0].collection)
+        print(info[0].title)
+        print(info[0].comments)
+        print(info[0].metadata)
+        print(info[0].metadata)
+
+        return info
+
     def add_graph_embeddings(self, m):
 
         when = int(time.time() * 1000)

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/model/prompt/generic/service.py

@@ -63,7 +63,8 @@ class Processor(ConsumerProducer):
             subscriber=subscriber,
             input_queue=tc_request_queue,
             output_queue=tc_response_queue,
-            pulsar_host = self.pulsar_host
+            pulsar_host = self.pulsar_host,
+            pulsar_api_key=self.pulsar_api_key,
         )
 
     def parse_json(self, text):

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/model/prompt/template/service.py

@@ -136,7 +136,8 @@ class Processor(ConsumerProducer):
             subscriber=subscriber,
             input_queue=tc_request_queue,
             output_queue=tc_response_queue,
-            pulsar_host = self.pulsar_host
+            pulsar_host = self.pulsar_host,
+            pulsar_api_key=self.pulsar_api_key,
         )
 
         # System prompt hack

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/processing/processing.py

@@ -49,11 +49,12 @@ class Processing:
             pulsar_host,
             log_level,
             file,
+            pulsar_api_key=None,
     ):
         self.pulsar_host = pulsar_host
         self.log_level = log_level
         self.file = file
-
+        self.pulsar_api_key = pulsar_api_key
         self.defs = load(open(file, "r"), Loader=Loader)
 
     def run(self):
@@ -68,6 +69,7 @@ class Processing:
 
             params = {
                 "pulsar_host": self.pulsar_host,
+                "pulsar_api_key": self.pulsar_api_key,
                 "log_level": str(self.log_level),
             }
 
@@ -125,12 +127,19 @@ def run():
     )
 
     default_pulsar_host = os.getenv("PULSAR_HOST", 'pulsar://pulsar:6650')
+    default_pulsar_api_key = os.getenv("PULSAR_API_KEY", None)
 
     parser.add_argument(
         '-p', '--pulsar-host',
         default=default_pulsar_host,
         help=f'Pulsar host (default: {default_pulsar_host})',
     )
+    
+    parser.add_argument(
+        '--pulsar-api-key',
+        default=default_pulsar_api_key,
+        help=f'Pulsar API key',
+    )
 
     parser.add_argument(
         '-l', '--log-level',
@@ -153,6 +162,7 @@ def run():
         try:
             p = Processing(
                 pulsar_host=args.pulsar_host,
+                pulsar_api_key=args.pulsar_api_key,
                 file=args.file,
                 log_level=args.log_level,
             )

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/retrieval/document_rag/rag.py

@@ -68,6 +68,7 @@ class Processor(ConsumerProducer):
 
         self.rag = DocumentRag(
             pulsar_host=self.pulsar_host,
+            pulsar_api_key=self.pulsar_api_key,
             pr_request_queue=pr_request_queue,
             pr_response_queue=pr_response_queue,
             emb_request_queue=emb_request_queue,

{trustgraph-flow-0.21.2 → trustgraph-flow-0.21.4}/trustgraph/retrieval/graph_rag/rag.py

@@ -82,6 +82,7 @@ class Processor(ConsumerProducer):
 
         self.rag = GraphRag(
             pulsar_host=self.pulsar_host,
+            pulsar_api_key=self.pulsar_api_key,
             pr_request_queue=pr_request_queue,
             pr_response_queue=pr_response_queue,
             emb_request_queue=emb_request_queue,