Feature/pkgsplit (#83)

* Starting to spawn base package
* More package hacking
* Bedrock and VertexAI
* Parquet split
* Updated templates
* Utils

trustgraph-flow/trustgraph/embeddings/ollama/__init__.py

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

trustgraph-flow/trustgraph/embeddings/ollama/__main__.py

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

trustgraph-flow/trustgraph/embeddings/ollama/processor.py

@@ -0,0 +1,84 @@
+
+"""
+Embeddings service, applies an embeddings model selected from HuggingFace.
+Input is text, output is embeddings vector.
+"""
+from langchain_community.embeddings import OllamaEmbeddings
+
+from ... schema import EmbeddingsRequest, EmbeddingsResponse
+from ... schema import embeddings_request_queue, embeddings_response_queue
+from ... log_level import LogLevel
+from ... 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="mxbai-embed-large"
+default_ollama = 'http://localhost:11434'
+
+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)
+
+        super(Processor, self).__init__(
+            **params | {
+                "input_queue": input_queue,
+                "output_queue": output_queue,
+                "subscriber": subscriber,
+                "input_schema": EmbeddingsRequest,
+                "output_schema": EmbeddingsResponse,
+            }
+        )
+
+        self.embeddings = OllamaEmbeddings(base_url=ollama, model=model)
+
+    def handle(self, msg):
+
+        v = msg.value()
+
+        # Sender-produced ID
+
+        id = msg.properties()["id"]
+
+        print(f"Handling input {id}...", flush=True)
+
+        text = v.text
+        embeds = self.embeddings.embed_query([text])
+
+        print("Send response...", flush=True)
+        r = EmbeddingsResponse(vectors=[embeds])
+
+        self.producer.send(r, properties={"id": id})
+
+        print("Done.", flush=True)
+
+    @staticmethod
+    def add_args(parser):
+
+        ConsumerProducer.add_args(
+            parser, default_input_queue, default_subscriber,
+            default_output_queue,
+        )
+
+        parser.add_argument(
+            '-m', '--model',
+            default=default_model,
+            help=f'Embeddings model (default: {default_model})'
+        )
+
+        parser.add_argument(
+            '-r', '--ollama',
+            default=default_ollama,
+            help=f'ollama (default: {default_ollama})'
+        )
+
+def run():
+
+    Processor.start(module, __doc__)
+

trustgraph-flow/trustgraph/embeddings/vectorize/__init__.py

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

trustgraph-flow/trustgraph/embeddings/vectorize/__main__.py

@@ -0,0 +1,6 @@
+
+from . vectorize import run
+
+if __name__ == '__main__':
+    run()
+

trustgraph-flow/trustgraph/embeddings/vectorize/vectorize.py

@@ -0,0 +1,103 @@
+
+"""
+Vectorizer, calls the embeddings service to get embeddings for a chunk.
+Input is text chunk, output is chunk and vectors.
+"""
+
+from ... schema import Chunk, ChunkEmbeddings
+from ... schema import chunk_ingest_queue, chunk_embeddings_ingest_queue
+from ... schema import embeddings_request_queue, embeddings_response_queue
+from ... clients.embeddings_client import EmbeddingsClient
+from ... log_level import LogLevel
+from ... base import ConsumerProducer
+
+module = ".".join(__name__.split(".")[1:-1])
+
+default_input_queue = chunk_ingest_queue
+default_output_queue = chunk_embeddings_ingest_queue
+default_subscriber = module
+
+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)
+        emb_request_queue = params.get(
+            "embeddings_request_queue", embeddings_request_queue
+        )
+        emb_response_queue = params.get(
+            "embeddings_response_queue", embeddings_response_queue
+        )
+
+        super(Processor, self).__init__(
+            **params | {
+                "input_queue": input_queue,
+                "output_queue": output_queue,
+                "embeddings_request_queue": emb_request_queue,
+                "embeddings_response_queue": emb_response_queue,
+                "subscriber": subscriber,
+                "input_schema": Chunk,
+                "output_schema": ChunkEmbeddings,
+            }
+        )
+
+        self.embeddings = EmbeddingsClient(
+            pulsar_host=self.pulsar_host,
+            input_queue=emb_request_queue,
+            output_queue=emb_response_queue,
+            subscriber=module + "-emb",
+        )
+
+    def emit(self, source, chunk, vectors):
+
+        r = ChunkEmbeddings(source=source, chunk=chunk, vectors=vectors)
+        self.producer.send(r)
+
+    def handle(self, msg):
+
+        v = msg.value()
+        print(f"Indexing {v.source.id}...", flush=True)
+
+        chunk = v.chunk.decode("utf-8")
+
+        try:
+
+            vectors = self.embeddings.request(chunk)
+
+            self.emit(
+                source=v.source,
+                chunk=chunk.encode("utf-8"),
+                vectors=vectors
+            )
+
+        except Exception as e:
+            print("Exception:", e, flush=True)
+
+        print("Done.", flush=True)
+
+    @staticmethod
+    def add_args(parser):
+
+        ConsumerProducer.add_args(
+            parser, default_input_queue, default_subscriber,
+            default_output_queue,
+        )
+
+        parser.add_argument(
+            '--embeddings-request-queue',
+            default=embeddings_request_queue,
+            help=f'Embeddings request queue (default: {embeddings_request_queue})',
+        )
+
+        parser.add_argument(
+            '--embeddings-response-queue',
+            default=embeddings_response_queue,
+            help=f'Embeddings request queue (default: {embeddings_response_queue})',
+        )
+
+def run():
+
+    Processor.start(module, __doc__)
+