Fix indexing errors for cases where the index does not have a column doc_id (#23)

gaya3-zipstack · web-flow · commit 8a63c094c512 · 2024-03-22T14:54:03.000+05:30
* Fix indexing issues for new documents. Roll up adapter version

* Merge with latest main

* Update pdm.lock file

---------

Signed-off-by: Gayathri &lt;142381512+gaya3-zipstack@users.noreply.github.com&gt;
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

Original file line number	Diff line number	Diff line change
`@@ -1,4 +1,4 @@`
`1`		`-__version__ = "0.15.2"`
	`1`	`+__version__ = "0.16.0"`
`2`	`2`
`3`	`3`
`4`	`4`	`def get_sdk_version():`
Original file line number	Diff line number	Diff line change
`@@ -172,9 +172,11 @@ def index_file(`
`172`	`172`	`)`
`173`	`173`	`raise SdkError(f"Error loading {vector_db}")`
`174`	`174`
	`175`	`+ filter = [{"field_name": "doc_id", "operator": "=", "value": doc_id}]`
`175`	`176`	`q = VectorStoreQuery(`
`176`	`177`	`query_embedding=embedding_li.get_query_embedding(" "),`
`177`	`178`	`doc_ids=[doc_id],`
	`179`	`+ filters=filter,`
`178`	`180`	`)`
`179`	`181`
`180`	`182`	`doc_id_not_found = True`