[feat] Refactor VectorDB class hierarchy for flexibility

embedchain/vectordb/base_vector_db.py

@@ -0,0 +1,10 @@
+class BaseVectorDB:
+    def __init__(self):
+        self.client = self._get_or_create_db()
+        self.collection = self._get_or_create_collection()
+
+    def _get_or_create_db(self):
+        raise NotImplementedError
+
+    def _get_or_create_collection(self):
+        raise NotImplementedError

embedchain/vectordb/chroma_db.py

@@ -0,0 +1,26 @@
+import os
+import chromadb
+from base_vector_db import BaseVectorDB
+from chromadb.utils import embedding_functions
+
+openai_ef = embedding_functions.OpenAIEmbeddingFunction(
+    api_key=os.getenv("OPENAI_API_KEY"),
+    model_name="text-embedding-ada-002"
+)
+
+class ChromaDB(BaseVectorDB):
+    def __init__(self, db_dir):
+        self.client_settings = chromadb.config.Settings(
+            chroma_db_impl="duckdb+parquet",
+            persist_directory=db_dir,
+            anonymized_telemetry=False
+        )
+        super().__init__()
+
+    def _get_or_create_db(self):
+        return chromadb.Client(self.client_settings)
+
+    def _get_or_create_collection(self):
+        return self.client.get_or_create_collection(
+            'embedchain_store', embedding_function=openai_ef,
+        )