From a4dd1c8fa39c5fabd3522594304609766cf94b1b Mon Sep 17 00:00:00 2001
From: Cyber MacGeddon <cybermaggedon@gmail.com>
Date: Fri, 29 Nov 2024 17:18:00 +0000
Subject: [PATCH] Starting to hack a lookup service

---
 trustgraph-base/trustgraph/schema/__init__.py |   2 +
 trustgraph-base/trustgraph/schema/lookup.py   |  34 ++++++
 .../trustgraph/external/__init__.py           |   0
 .../trustgraph/external/wikipedia/__init__.py |   0
 .../trustgraph/external/wikipedia/service.py  | 100 ++++++++++++++++++
 5 files changed, 136 insertions(+)
 create mode 100644 trustgraph-base/trustgraph/schema/lookup.py
 create mode 100644 trustgraph-flow/trustgraph/external/__init__.py
 create mode 100644 trustgraph-flow/trustgraph/external/wikipedia/__init__.py
 create mode 100644 trustgraph-flow/trustgraph/external/wikipedia/service.py

diff --git a/trustgraph-base/trustgraph/schema/__init__.py b/trustgraph-base/trustgraph/schema/__init__.py
index 3196691b..be41b670 100644
--- a/trustgraph-base/trustgraph/schema/__init__.py
+++ b/trustgraph-base/trustgraph/schema/__init__.py
@@ -9,4 +9,6 @@ from . graph import *
 from . retrieval import *
 from . metadata import *
 from . agent import *
+from . lookup import *
+
 
diff --git a/trustgraph-base/trustgraph/schema/lookup.py b/trustgraph-base/trustgraph/schema/lookup.py
new file mode 100644
index 00000000..f8aaf5fc
--- /dev/null
+++ b/trustgraph-base/trustgraph/schema/lookup.py
@@ -0,0 +1,34 @@
+
+from pulsar.schema import Record, Bytes, String, Boolean, Integer, Array, Double
+
+from . types import Error, Value, Triple
+from . topic import topic
+from . metadata import Metadata
+
+############################################################################
+
+# Lookups
+
+class LookupRequest(Record):
+    kind = String()
+    term = String()
+
+class LookupResponse(Record):
+    text = String()
+
+wikipedia_lookup_request_queue = topic(
+    'encyclopedia', kind='non-persistent', namespace='request'
+)
+wikipedia_lookup_response_queue = topic(
+    'encyclopedia', kind='non-persistent', namespace='response', 
+)
+
+internet_search_request_queue = topic(
+    'internet-search', kind='non-persistent', namespace='request'
+)
+internet_search_response_queue = topic(
+    'internet-search', kind='non-persistent', namespace='response', 
+)
+
+############################################################################
+
diff --git a/trustgraph-flow/trustgraph/external/__init__.py b/trustgraph-flow/trustgraph/external/__init__.py
new file mode 100644
index 00000000..e69de29b
diff --git a/trustgraph-flow/trustgraph/external/wikipedia/__init__.py b/trustgraph-flow/trustgraph/external/wikipedia/__init__.py
new file mode 100644
index 00000000..e69de29b
diff --git a/trustgraph-flow/trustgraph/external/wikipedia/service.py b/trustgraph-flow/trustgraph/external/wikipedia/service.py
new file mode 100644
index 00000000..4b3b39c1
--- /dev/null
+++ b/trustgraph-flow/trustgraph/external/wikipedia/service.py
@@ -0,0 +1,100 @@
+
+"""
+Embeddings service, applies an embeddings model selected from HuggingFace.
+Input is text, output is embeddings vector.
+"""
+
+from langchain_huggingface import HuggingFaceEmbeddings
+
+from trustgraph.schema import EmbeddingsRequest, EmbeddingsResponse, Error
+from trustgraph.schema import embeddings_request_queue
+from trustgraph.schema import embeddings_response_queue
+from trustgraph.log_level import LogLevel
+from trustgraph.base import ConsumerProducer
+
+module = ".".join(__name__.split(".")[1:-1])
+
+default_input_queue = embeddings_request_queue
+default_output_queue = embeddings_response_queue
+default_subscriber = module
+default_model="all-MiniLM-L6-v2"
+
+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)
+        model = params.get("model", default_model)
+
+        super(Processor, self).__init__(
+            **params | {
+                "input_queue": input_queue,
+                "output_queue": output_queue,
+                "subscriber": subscriber,
+                "input_schema": EmbeddingsRequest,
+                "output_schema": EmbeddingsResponse,
+            }
+        )
+
+        self.embeddings = HuggingFaceEmbeddings(model_name=model)
+
+    def handle(self, msg):
+
+        v = msg.value()
+
+        # Sender-produced ID
+        id = msg.properties()["id"]
+
+        print(f"Handling input {id}...", flush=True)
+
+        try:
+
+            text = v.text
+            embeds = self.embeddings.embed_documents([text])
+
+            print("Send response...", flush=True)
+            r = EmbeddingsResponse(vectors=embeds, 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 = EmbeddingsResponse(
+                error=Error(
+                    type = "llm-error",
+                    message = str(e),
+                ),
+                response=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(
+            '-m', '--model',
+            default="all-MiniLM-L6-v2",
+            help=f'LLM model (default: all-MiniLM-L6-v2)'
+        )
+
+def run():
+
+    Processor.start(module, __doc__)
+