feat: add delete all documents to Pinecone DocumentStore (#2403)

* feat: add delete all documents to Pineconde DocumentStore

* add empty exception

---------

Co-authored-by: Michele Pangrazzi <xmikex83@gmail.com>

Author: ChinmayBansal

integrations/pinecone/src/haystack_integrations/document_stores/pinecone/document_store.py

@@ -11,6 +11,7 @@
 
 from pinecone import Pinecone, PineconeAsyncio, PodSpec, ServerlessSpec
 from pinecone.db_data import _Index, _IndexAsyncio
+from pinecone.exceptions import NotFoundException
 
 from .filters import _normalize_filters, _validate_filters
 
@@ -352,6 +353,30 @@ async def delete_documents_async(self, document_ids: List[str]) -> None:
         assert self._async_index is not None, "Index is not initialized"
         await self._async_index.delete(ids=document_ids, namespace=self.namespace)
 
+    def delete_all_documents(self) -> None:
+        """
+        Deletes all documents in the document store.
+        """
+        self._initialize_index()
+        assert self._index is not None, "Index is not initialized"
+        try:
+            self._index.delete(delete_all=True, namespace=self.namespace)
+        except NotFoundException:
+            # Namespace doesn't exist (empty collection), which is fine - nothing to delete
+            logger.debug("Namespace '{namespace}' not found. Nothing to delete.", namespace=self.namespace or "default")
+
+    async def delete_all_documents_async(self) -> None:
+        """
+        Asynchronously deletes all documents in the document store.
+        """
+        await self._initialize_async_index()
+        assert self._async_index is not None, "Index is not initialized"
+        try:
+            await self._async_index.delete(delete_all=True, namespace=self.namespace)
+        except NotFoundException:
+            # Namespace doesn't exist (empty collection), which is fine - nothing to delete
+            logger.debug("Namespace '{namespace}' not found. Nothing to delete.", namespace=self.namespace or "default")
+
     def _embedding_retrieval(
         self,
         query_embedding: List[float],

integrations/pinecone/tests/test_document_store.py

@@ -272,6 +272,19 @@ def test_write_documents_duplicate_skip(self, document_store: PineconeDocumentSt
     @pytest.mark.skip(reason="Pinecone creates a namespace only when the first document is written")
     def test_delete_documents_empty_document_store(self, document_store: PineconeDocumentStore): ...
 
+    def test_delete_all_documents(self, document_store: PineconeDocumentStore):
+        docs = [Document(content="first doc"), Document(content="second doc")]
+        document_store.write_documents(docs)
+        assert document_store.count_documents() == 2
+
+        document_store.delete_all_documents()
+        assert document_store.count_documents() == 0
+
+    def test_delete_all_documents_empty_collection(self, document_store: PineconeDocumentStore):
+        assert document_store.count_documents() == 0
+        document_store.delete_all_documents()
+        assert document_store.count_documents() == 0
+
     def test_embedding_retrieval(self, document_store: PineconeDocumentStore):
         query_embedding = [0.1] * 768
         most_similar_embedding = [0.8] * 768

integrations/pinecone/tests/test_document_store_async.py

@@ -66,6 +66,19 @@ async def test_delete_documents(self, document_store_async: PineconeDocumentStor
         await document_store_async.delete_documents_async([doc.id])
         assert await document_store_async.count_documents_async() == 0
 
+    async def test_delete_all_documents_async(self, document_store_async: PineconeDocumentStore):
+        docs = [Document(content="first doc"), Document(content="second doc")]
+        await document_store_async.write_documents_async(docs)
+        assert await document_store_async.count_documents_async() == 2
+
+        await document_store_async.delete_all_documents_async()
+        assert await document_store_async.count_documents_async() == 0
+
+    async def test_delete_all_documents_async_empty_collection(self, document_store_async: PineconeDocumentStore):
+        assert await document_store_async.count_documents_async() == 0
+        await document_store_async.delete_all_documents_async()
+        assert await document_store_async.count_documents_async() == 0
+
     async def test_embedding_retrieval(self, document_store_async: PineconeDocumentStore):
         query_embedding = [0.1] * 768
         most_similar_embedding = [0.8] * 768