Fix indexing errors for cases where the index does not have a column …

…doc_id (#23) * Fix indexing issues for new documents. Roll up adapter version * Merge with latest main * Update pdm.lock file --------- Signed-off-by: Gayathri <[email protected]>
Zipstack · Mar 22, 2024 · 8a63c09 · 8a63c09
1 parent d87222e
commit 8a63c09
Show file tree

Hide file tree

Showing 4 changed files with 249 additions and 9 deletions.
diff --git a/pdm.lock b/pdm.lock
diff --git a/pyproject.toml b/pyproject.toml
@@ -15,7 +15,7 @@ dependencies = [
     "python-magic~=0.4.27",
     "python-dotenv==1.0.0",
     # LLM Triad
-    "unstract-adapters~=0.5.0",
+    "unstract-adapters~=0.6.0",
     "llama-index==0.9.28",
     "tiktoken~=0.4.0",
     "transformers==4.37.0",
@@ -55,4 +55,4 @@ package-dir = "src"
 
 [tool.pdm.version]
 source = "file"
-path = "src/unstract/sdk/__init__.py"
+path = "src/unstract/sdk/__init__.py"
diff --git a/src/unstract/sdk/__init__.py b/src/unstract/sdk/__init__.py
@@ -1,4 +1,4 @@
-__version__ = "0.15.2"
+__version__ = "0.16.0"
 
 
 def get_sdk_version():

diff --git a/src/unstract/sdk/index.py b/src/unstract/sdk/index.py
@@ -172,9 +172,11 @@ def index_file(
             )
             raise SdkError(f"Error loading {vector_db}")
 
+        filter = [{"field_name": "doc_id", "operator": "=", "value": doc_id}]
         q = VectorStoreQuery(
             query_embedding=embedding_li.get_query_embedding(" "),
             doc_ids=[doc_id],
+            filters=filter,
         )
 
         doc_id_not_found = True