[ES2-1256] convert es2 to pyenvector

.github/workflows/release.yaml

@@ -192,7 +192,7 @@ jobs:
           MAX_ATTEMPTS=30
           SLEEP_SECONDS=10
 
-          echo "Attempting to install es2==${WHEEL_VERSION} from TestPyPI..."
+          echo "Attempting to install pyenvector==${WHEEL_VERSION} from TestPyPI..."
           ATTEMPTS=0
           while true; do
             ATTEMPTS=$((ATTEMPTS + 1))

.gitignore

@@ -39,10 +39,8 @@ keys/
 VECTORSTORE.md
 
 # External symlinks (local workspace references)
-es2-msa
 es2-msa/
-es2-deploy
-es2-deploy/
+envector-deployment/
 
 # Local helper scripts
 run_unit_tests.py

CONTRIBUTE.md

@@ -13,16 +13,16 @@ Thanks for your interest in improving the project! This guide covers local setup
 
 ## Testing
 - **Unit tests** (fakes only): `python run_unit_tests.py`
-- **Integration tests** (requires ES2 server + keys):
-  - Export `ES2_ADDRESS`, `ES2_KEY_PATH`, `ES2_KEY_ID`
-  - Optional: `ES2_USE_EMBEDDINGS=1`, `ES2_EMB_MODEL`, `ES2_USE_HF_DATASET=1`
+- **Integration tests** (requires EnVector server + keys):
+  - Export `ENVECTOR_ADDRESS`, `ENVECTOR_KEY_PATH`, `ENVECTOR_KEY_ID`
+  - Optional: `ENVECTOR_USE_EMBEDDINGS=1`, `ENVECTOR_EMB_MODEL`, `ENVECTOR_USE_HF_DATASET=1`
   - Run `pytest -m integration -s`
 
 Please run relevant tests before submitting a PR and mention coverage in the description.
 
 ## Development Guidelines
 - Keep code, comments, and docs in English.
-- Prefer the high-level `es2` SDK APIs; avoid direct gRPC/indexer calls unless required.
+- Prefer the high-level `pyenvector` SDK APIs; avoid direct gRPC/indexer calls unless required.
 - Keep changes focused and documented; update README or notebooks when behavior changes.
 - Follow existing formatting and type-hint conventions.
 

README.md

@@ -1,6 +1,6 @@
 # LangChain Envector Integration
 
-Encrypted vector search for LangChain using Envector (ES2), powered by homomorphic encryption (CKKS). This repo ships a LangChain-compatible VectorStore and retriever utilities built on the high-level `es2` Python SDK.
+Encrypted vector search for LangChain using Envector, powered by homomorphic encryption (CKKS). This repo ships a LangChain-compatible VectorStore and retriever utilities built on the high-level `pyenvector` Python SDK.
 
 ## Features
 - LangChain `VectorStore` interface with `similarity_search`, `from_texts`, etc.
@@ -13,10 +13,10 @@ Encrypted vector search for LangChain using Envector (ES2), powered by homomorph
   - `python3.11 -m venv .venv && source .venv/bin/activate`
 - Install runtime dependencies:
   - `pip install -U pip setuptools wheel`
-  - `pip install es2 langchain sentence-transformers`
+  - `pip install pyenvector langchain sentence-transformers`
 
 ## Usage Overview
-1. Configure Envector using `EnvectorConfig`, pointing to your ES2 endpoint and keys.
+1. Configure Envector using `EnvectorConfig`, pointing to your EnVector endpoint and keys.
 2. Initialize embeddings (or provide pre-computed vectors).
 3. Instantiate `Envector(config=cfg, embeddings=emb)` and call `add_texts`, `add_documents`, or use `as_retriever`.
 4. Run `similarity_search` or plug the retriever into your LangChain pipeline.
@@ -25,13 +25,13 @@ Encrypted vector search for LangChain using Envector (ES2), powered by homomorph
 
 ## Configuration
 Key dataclasses live in `libs/envector/config.py`:
-- `ConnectionConfig`: address or host/port for ES2.
+- `ConnectionConfig`: address or host/port for EnVector.
 - `KeyConfig`: key path, key ID, optional preset/eval mode.
 - `IndexSettings`: index name, dimension (32–4096), query encryption mode, optional output fields and fetch parameters.
 - `EnvectorConfig`: wraps the above and enables auto-creation via `create_if_missing`.
 
 ## Data Model
-- Each vector stores a single `metadata` string in ES2.
+- Each vector stores a single `metadata` string in EnVector.
 - To align with LangChain’s `Document`, inserts wrap data as JSON: `{"text": ..., "metadata": ...}`.
 - Retrieval unwraps JSON, returning `Document(page_content=text, metadata={...})`.
 - Client-side filtering requires the JSON envelope to include an object under `metadata`.
@@ -48,12 +48,12 @@ Key dataclasses live in `libs/envector/config.py`:
 
   cfg = EnvectorConfig(
       connection=ConnectionConfig(
-        address=ES2_ADDRESS, 
-        access_token=ES2_ACCESS_TOKEN
+        address=ENVECTOR_ADDRESS, 
+        access_token=ENVECTOR_ACCESS_TOKEN
       ),
       key=KeyConfig(
-        key_path=ES2_KEY_PATH, 
-        key_id=ES2_KEY_ID, 
+        key_path=ENVECTOR_KEY_PATH, 
+        key_id=ENVECTOR_KEY_ID, 
         preset="ip", 
         eval_mode="rmp"
       ),
@@ -100,18 +100,18 @@ Key dataclasses live in `libs/envector/config.py`:
   The methods `similarity_search` and `similarity_search_with_vector` (with `embeddings.embed_query()`) are also available to perform vector search.
 
 ## Troubleshooting
-- Connection issues: verify ES2 address and registered keys.
+- Connection issues: verify EnVector address and registered keys.
 - Embeddings mismatch: ensure embedding dimension equals `index.dim` when supplying vectors.
 - Unexpected raw strings: confirm inserts used the JSON envelope.
 - Key Issues: check key's metadata to sync with the registered key if facing any key issue.
 
-## Testing Without ES2
-- Run unit tests offline (no ES2 or SDK required):
+## Testing Without EnVector
+- Run unit tests offline (no EnVector or SDK required):
   - `python -m pytest -q -m "not integration"`
   - or `python scripts/run_unit_tests.py`
 - Run integration tests (requires server and keys):
-  - Export `ES2_ADDRESS`, `ES2_KEY_PATH`, `ES2_KEY_ID`
-  - Optional: `ES2_USE_EMBEDDINGS=1`, `ES2_EMB_MODEL`, `ES2_USE_HF_DATASET=1`
+  - Export `ENVECTOR_ADDRESS`, `ENVECTOR_KEY_PATH`, `ENVECTOR_KEY_ID`
+  - Optional: `ENVECTOR_USE_EMBEDDINGS=1`, `ENVECTOR_EMB_MODEL`, `ENVECTOR_USE_HF_DATASET=1`
   - `python -m pytest -q -m integration -s`
 
 ## Contributing

libs/envector/README.md

@@ -1,15 +1,15 @@
 # Envector (LangChain VectorStore)
 
-High-level VectorStore adaptor for Envector (ES2), using the `es2` SDK. Vectors are always encrypted on the server; the SDK performs required crypto client-side.
+High-level VectorStore adaptor for Envector, using the `pyenvector` SDK. Vectors are always encrypted on the server; the SDK performs required crypto client-side.
 
 Key points
-- Use high-level `es2.ES2` and `es2.Index`; avoid low-level `es2.api.Indexer`/gRPC.
+- Use high-level `pyenvector.EnvectorClient` and `pyenvector.Index`; avoid low-level `pyenvector.api.Indexer`/gRPC.
 - Index encryption is fixed to `cipher`. Query can be `plain` or `cipher`.
 - Metadata is stored as a single JSON string per item: `{id, text, metadata}`.
 
 Files
 - `config.py`: Configuration dataclasses (connection, key, index).
-- `client.py`: Initializes ES2 + index and returns an `Index` instance.
+- `client.py`: Initializes EnVector + index and returns an `Index` instance.
 - `vectorstore.py`: `Envector` VectorStore implementation.
 - `retriever.py`: Optional wrapper retriever.
 - `examples/`: Minimal examples.

libs/envector/examples/basic_usage.py

@@ -1,7 +1,7 @@
 """Basic usage example for Envector VectorStore.
 
 Requirements:
-- `es2`
+- `pyenvector`
 - `langchain` (version providing VectorStore APIs)
 - An embeddings backend, e.g. sentence-transformers
 """

libs/envector/examples/ingest_synthetic_1k.py

@@ -1,7 +1,7 @@
 """Ingest the synthetic 1K dataset into Envector.
 
 Requires:
-- ES2 server and keys.
+- EnVector server and keys.
 - Dataset at `data/synthetic_rag_1k.jsonl` (run scripts/make_synthetic_rag_dataset.py).
 
 Usage:

libs/envector/langchain_envector/__init__.py

@@ -1,6 +1,6 @@
 """Envector LangChain integration package.
 
-Provides a LangChain-compatible VectorStore that wraps the high-level `es2` SDK.
+Provides a LangChain-compatible VectorStore that wraps the high-level `pyenvector` SDK.
 All code and comments are in English as per project rules.
 """
 

libs/envector/langchain_envector/client.py

@@ -4,45 +4,45 @@
 
 
 class EnvectorClient:
-    """Thin convenience client around the high-level `es2` SDK.
+    """Thin convenience client around the high-level `pyenvector` SDK.
 
     - Establishes a connection
     - Initializes key and index configuration
     - Optionally creates the index if missing
-    - Provides access to the ES2 `Index` instance
+    - Provides access to the envector `Index` instance
     """
 
     def __init__(self, config: EnvectorConfig):
         self.config = config
-        self._es2 = None
+        self._ev = None
         self._index = None
 
     def init(self):
-        import es2
+        import pyenvector as ev
 
         c = self.config.connection
         k = self.config.key
         i = self.config.index
 
-        es2_client = es2.ES2()
+        ev_client = ev.EnvectorClient()
 
         # Connection
         if c.address:
-            es2_client.init_connect(address=c.address, access_token=c.access_token)
+            ev_client.init_connect(address=c.address, access_token=c.access_token)
         else:
             if not (c.host and c.port):
                 raise ValueError("Either address or host+port must be provided.")
-            es2_client.init_connect(
+            ev_client.init_connect(
                 host=c.host, port=c.port, access_token=c.access_token
             )
 
         # Key path baseline for Index
-        from es2.index import Index as _Index
+        from pyenvector.index import Index as _Index
 
         _Index.init_key_path(k.key_path)
 
         # Index config + key setup
-        es2_client.init_index_config(
+        ev_client.init_index_config(
             index_name=i.index_name,
             dim=i.dim,
             key_path=k.key_path,
@@ -59,13 +59,13 @@ def init(self):
 
         # Create index if missing
         if self.config.create_if_missing:
-            idx_list = es2_client.get_index_list()
+            idx_list = ev_client.get_index_list()
             if i.index_name not in idx_list:
-                es2_client.create_index(index_name=i.index_name, dim=i.dim)
+                ev_client.create_index(index_name=i.index_name, dim=i.dim)
 
         # Bind index instance
-        self._index = es2.Index(i.index_name)
-        self._es2 = es2_client
+        self._index = ev.Index(i.index_name)
+        self._ev = ev_client
         return self
 
     @property
@@ -75,7 +75,7 @@ def index(self):
         return self._index
 
     @property
-    def es2(self):
-        if self._es2 is None:
+    def ev(self):
+        if self._ev is None:
             raise RuntimeError("Client not initialized. Call init().")
-        return self._es2
+        return self._ev

libs/envector/langchain_envector/types.py

@@ -29,9 +29,9 @@ class SearchResult:
 
 
 def pack_metadata(text: str, metadata: Optional[Dict[str, Any]] = None) -> str:
-    """Pack text and metadata into a single JSON string field accepted by ES2.
+    """Pack text and metadata into a single JSON string field accepted by pyenvector.
 
-    ES2 metadata API stores lists of strings; we store a single JSON blob per item.
+    pyenvector metadata API stores lists of strings; we store a single JSON blob per item.
     Item-level IDs are not persisted/addressable.
     """
     import json
@@ -46,7 +46,7 @@ def pack_metadata(text: str, metadata: Optional[Dict[str, Any]] = None) -> str:
 def unpack_metadata(raw: Any) -> Dict[str, Any]:
     """Return metadata as a dict regardless of the raw payload type.
 
-    Recent ES2 versions may return decrypted metadata as a Python dict instead
+    Recent pyenvector versions may return decrypted metadata as a Python dict instead
     of the JSON string we originally stored. We normalise the payload here so
     downstream code always works with a dictionary.
     """
@@ -79,7 +79,7 @@ def unpack_metadata(raw: Any) -> Dict[str, Any]:
             if isinstance(data, dict):
                 return data
         except Exception:
-            # Some ES2 responses return Python-literal strings (single quotes).
+            # Some pyenvector responses return Python-literal strings (single quotes).
             try:
                 import ast
 

libs/envector/langchain_envector/vectorstore.py

@@ -38,10 +38,10 @@ def __init__(
 
 
 class Envector(VectorStore):  # type: ignore[misc]
-    """LangChain-compatible VectorStore adaptor for Envector (ES2).
+    """LangChain-compatible VectorStore adaptor for Envector.
 
-    This class wraps the high-level `es2` SDK. It does not use low-level
-    gRPC stubs or `es2.api.Indexer` directly.
+    This class wraps the high-level `pyenvector` SDK. It does not use low-level
+    gRPC stubs or `pyenvector.api.Indexer` directly.
     """
 
     def __init__(
@@ -89,7 +89,7 @@ def add_texts(
         # Prepare metadata JSON strings per item
         packed = [pack_metadata(t, m) for t, m in zip(texts, metadatas)]
 
-        # Insert using high-level ES2 Index
+        # Insert using high-level pyenvector Index
         result_ids = self.client.index.insert(data=vectors, metadata=packed)
 
         # Return ephemeral placeholders to satisfy VectorStore interface,
@@ -111,7 +111,7 @@ def _similarity_search_with_scores(
         results = self.client.index.search(
             query=embedding, top_k=top_k, output_fields=self.config.index.output_fields
         )
-        # ES2 Index.search returns a list for each query; we passed single query
+        # pyenvector Index.search returns a list for each query; we passed single query
         result = (
             results[0]
             if isinstance(results, list) and results and isinstance(results[0], list)
@@ -265,7 +265,7 @@ def add_documents(
         extracting `page_content` and `metadata` from each Document.
 
         Notes:
-        - Manual `ids` are ignored (ES2 does not support user-provided IDs).
+        - Manual `ids` are ignored (EnVector does not support user-provided IDs).
         - When `embeddings` is not configured, you must supply `vectors`.
         - Returns ephemeral IDs as produced by the client insert.
         """

pyproject.toml

@@ -7,19 +7,19 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "langchain-envector"
-version = "0.1.2"
-description = "LangChain VectorStore integration for Envector (ES2) encrypted vector search"
+version = "0.1.3"
+description = "LangChain VectorStore integration for Envector"
 readme = "README.md"
 license = {text = "MIT"}
 requires-python = ">=3.9,<3.14"
 authors = [
   { name = "Envector Contributors" }
 ]
 dependencies = [
-  "es2",
+  "pyenvector",
   "langchain>=0.2.0",
 ]
-keywords = ["langchain", "vectorstore", "homomorphic-encryption", "ckks", "encrypted-search", "envector", "es2"]
+keywords = ["langchain", "vectorstore", "homomorphic-encryption", "ckks", "encrypted-search", "envector", "pyenvector"]
 classifiers = [
   "Programming Language :: Python :: 3",
   "Programming Language :: Python :: 3.9",

pytest.ini

@@ -1,6 +1,6 @@
 [pytest]
 markers =
-    integration: tests that require a running ES2 server and the real es2 SDK
+    integration: tests that require a running EnVector server and the real EnVector SDK
 testpaths =
     tests