Feature: Pinecone connector

.gitignore

@@ -189,3 +189,5 @@ terraform.rc
 
 .history
 
+# Ignore ChromaDB
+chroma.sqlite3

.vscode/launch.json

@@ -0,0 +1,20 @@
+{
+  // Use IntelliSense to learn about possible attributes.
+  // Hover to view descriptions of existing attributes.
+  // For more information, visit: https://go.microsoft.com/fwlink/?linkid=830387
+  "version": "0.2.0",
+  "configurations": [
+    {
+      "name": "Python: FastAPI",
+      "console": "integratedTerminal",
+      "type": "python",
+      "request": "launch",
+      "module": "uvicorn",
+      "envFile": "${workspaceFolder}/apps/semantic_search/.env",
+      "args": ["semantic_search:app", "--reload"],
+      "jinja": true,
+      "justMyCode": true,
+      "pythonArgs": ["-Xfrozen_modules=off"]
+    }
+  ]
+}

apps/semantic_search/.dockerignore

@@ -1,2 +1,3 @@
 poetry.lock
 .env.example
+chroma.sqlite3

apps/semantic_search/.env.example

@@ -62,4 +62,12 @@ CHROMA_COLLECTION= # Collection name to store the embeddings.
 SUPABASE_URL= # Your Supabase database URL
 SUPABASE_KEY= # Your Supabase API key
 SUPABASE_TABLE= # The database table used to save the embeddings.
-SUPABASE_FUNCTION= # The database function used to query the embeddings.
+SUPABASE_FUNCTION= # The database function used to query the embeddings.
+
+################################################
+
+## Pinecone provider (ONLY IF EMBEDDING_STORE is PINECONE)
+
+PINECONE_KEY= # Your Pinecone API key
+PINECONE_ENVIRONMENT= # The environment where your Pinecone project is.
+PINECONE_INDEX= # The Pinecone Index used to store and query the embeddings.

apps/semantic_search/.env.local.example

@@ -23,7 +23,7 @@ OPENAI_MODEL=text-embedding-ada-002  # Your OpenAI model. See: https://platform.
 
 # Embedding store configuration
 
-EMBEDDING_STORE=LOCAL # Can be LOCAL | CHROMA | SUPABASE
+EMBEDDING_STORE=LOCAL # Can be LOCAL | CHROMA | SUPABASE | PINECONE
 MATCH_THRESHOLD=0.5 # Threshold to consider two embeddings as a match. Value from 0 to 1
 
 ################################################

apps/semantic_search/pyproject.toml

@@ -19,6 +19,7 @@ python-dotenv = "^1.0.0"
 gradio-client = "^0.2.6"
 supabase = "^1.0.3"
 openai = "^0.27.8"
+pinecone-client = "^2.2.2"
 
 [tool.poetry.scripts]
 start = "semantic_search:start"

apps/semantic_search/stores/pinecone_client.py

@@ -0,0 +1,86 @@
+import uuid
+from typing import Any
+
+import pinecone
+from pinecone import Index, Vector
+from pydantic import BaseModel
+from stores.base import EmbeddingsStore, SearchResult, StoreRequest
+
+
+class PineconeEmbeddingsStoreSettings(BaseModel):
+    key: str
+    environment: str
+    index: str
+
+
+class PineconeMatch(BaseModel):
+    id: str
+    score: float
+    values: list[float]
+    metadata: dict[str, Any]
+
+
+class PineconeEmbeddingsStore(EmbeddingsStore):
+    def __init__(self, settings: PineconeEmbeddingsStoreSettings) -> None:
+        pinecone.init(api_key=settings.key, environment=settings.environment)
+
+        self.index = Index(settings.index)
+
+    def store(self, embeddings: list[StoreRequest]) -> list[str]:
+        self._validate_configuration()
+
+        pinecone_vectors: list[Vector] = []
+        for embedding in embeddings:
+            pinecone_metadata = embedding.metadata.copy()
+            pinecone_metadata["cluster_id"] = embedding.cluster_id
+            vector = Vector(
+                id=str(uuid.uuid4()),
+                values=embedding.embedding,
+                metadata=pinecone_metadata,
+            )
+            pinecone_vectors.append(vector)
+
+        result = self.index.upsert(vectors=pinecone_vectors)
+
+        result.get("upsertedData")
+        if result.get("upsertedData") == 0:
+            raise Exception("Error inserting: No rows added")
+
+        return [vector["id"] for vector in pinecone_vectors]
+
+    def search(
+        self,
+        embedding: list[float],
+        cluster_ids: list[str],
+        match_threshold: float = 0.8,
+        limit: int = 10,
+    ) -> list[SearchResult]:
+        self._validate_configuration()
+
+        query_response = self.index.query(
+            top_k=limit,
+            include_values=True,
+            include_metadata=True,
+            vector=embedding,
+            filter={"cluster_id": {"$in": cluster_ids}},
+        )
+
+        matches: list[PineconeMatch] = query_response.get("matches")
+        search_results: list[SearchResult] = []
+        for match in matches:
+            if match.score < match_threshold:
+                continue
+
+            cluster_id = match.metadata.pop("cluster_id")
+            search_result = SearchResult(
+                id=match.id,
+                metadata=match.metadata,
+                score=match.score,
+                cluster_id=cluster_id,
+            )
+            search_results.append(search_result)
+        return search_results
+
+    def _validate_configuration(self):
+        if not self.index:
+            raise ValueError("Pinecone index is required.")

apps/semantic_search/stores/provider.py

@@ -1,6 +1,10 @@
 from common.utils import get_env_or_fail
 from stores.base import EmbeddingsStore
 from stores.chroma import ChromaEmbeddingsStore, ChromaEmbeddingsStoreSettings
+from stores.pinecone_client import (
+    PineconeEmbeddingsStore,
+    PineconeEmbeddingsStoreSettings,
+)
 from stores.supabase_client import (
     SupabaseEmbeddingsStore,
     SupabaseEmbeddingsStoreSettings,
@@ -29,5 +33,13 @@ def get_client(name: str) -> EmbeddingsStore:
                     query_function=get_env_or_fail("SUPABASE_FUNCTION"),
                 )
             )
+        if name == "PINECONE":
+            return PineconeEmbeddingsStore(
+                PineconeEmbeddingsStoreSettings(
+                    key=get_env_or_fail("PINECONE_KEY"),
+                    environment=get_env_or_fail("PINECONE_ENVIRONMENT"),
+                    index=get_env_or_fail("PINECONE_INDEX"),
+                )
+            )
 
         raise Exception("Missing valid store client name")

website/docs/03_semantic_search/030_getting_started.mdx

@@ -39,6 +39,9 @@ The following table describes the different variables that can be set on the `.e
 | `SUPABASE_KEY` | :x: | The key to use when connecting to Supabase. This is only required if you are using the `SUPABASE` embedding store |
 | `SUPABASE_TABLE` | :x: | The table to use when storing embeddings. This is only required if you are using the `SUPABASE` embedding store |
 | `SUPABASE_FUNCTION` | :x: | The function to use when querying embeddings. This is only required if you are using the `SUPABASE` embedding store |
+| `PINECONE_KEY` | :x: | The key to use when connecting to Pinecone. This is only required if you are using the `PINECONE` embedding store |
+| `PINECONE_ENVIRONMENT` | :x: | The Pinecone's project environment. This is only required if you are using the `PINECONE` embedding store |
+| `PINECONE_INDEX` | :x: | The index (Pinecone database) where embeddings will be stored and queried. This is only required if you are using the `PINECONE` embedding store |
 
 
 

website/docs/03_semantic_search/032_embedding_stores.md

@@ -51,4 +51,21 @@ SUPABASE_FUNCTION="<SUPABASE_FUNCTION>"
 - `SUPABASE_TABLE` is the table where the embeddings are stored.
 - `SUPABASE_FUNCTION` is the function that is used to perform similarity searches.
 
-You can check out how to set up your Supabase database [here](https://supabase.com/blog/openai-embeddings-postgres-vector).
+You can check out how to set up your Supabase database [here](https://supabase.com/blog/openai-embeddings-postgres-vector).
+
+## Pinecone store
+
+Another alternative could be [Pinecone](https://www.pinecone.io/). It is a vector-oriented database created specifically for semanitc search. It is a cloud-based service, so you don't have to worry about setting up a database by yourself.
+
+```.env
+EMBEDDING_STORE="PINECONE"
+PINECONE_KEY="<PINECONE_KEY>"
+PINECONE_ENVIRONMENT="<PINECONE_ENVIRONMENT>"
+PINECONE_INDEX="<PINECONE_INDEX>"
+```
+
+- `PINECONE_KEY` is the access key of the Pinecone database.
+- `PINECONE_ENVIRONMENT` is the environment where the Pinecone project is placed.
+- `PINECONE_INDEX` is the Pinecone index (database) where embeddings are stored and queried.
+
+You can check out how to set up your Pinecone database [here](https://www.pinecone.io/).