Updated indexer,structure,text-extractor to use 0.13.0 SDK

Zipstack · Mar 1, 2024 · aeee5a7 · aeee5a7
1 parent 62bf9dd
commit aeee5a7
Show file tree

Hide file tree

Showing 7 changed files with 7 additions and 16 deletions.
diff --git a/tools/indexer/requirements.txt b/tools/indexer/requirements.txt
@@ -1,4 +1,4 @@
 # Add your dependencies here
 
 # Required for all unstract tools
-unstract-sdk~=0.11.1
+unstract-sdk~=0.13.0
diff --git a/tools/indexer/src/config/properties.json b/tools/indexer/src/config/properties.json
@@ -2,7 +2,7 @@
   "schemaVersion": "0.0.1",
   "displayName": "Document Indexer",
   "functionName": "document_indexer",
-  "toolVersion": "0.0.1",
+  "toolVersion": "0.0.2",
   "description": "Used to index documents in the specified vector DB / embedding.",
   "input": {
     "description": "Document that needs to be indexed"

diff --git a/tools/indexer/src/main.py b/tools/indexer/src/main.py
@@ -34,7 +34,7 @@ def run(
         tool_index = ToolIndex(tool=self)
         self.stream_log("Indexing document...")
         try:
-            tool_index.index_file(
+            index_key = tool_index.index_file(
                 tool_id=self.workflow_id,
                 embedding_type=settings[ToolSettingsKey.EMBEDDING_ADAPTER_ID],
                 vector_db=settings[ToolSettingsKey.VECTOR_DB_ADAPTER_ID],
@@ -47,15 +47,6 @@ def run(
             )
         except Exception as e:
             self.stream_error_and_exit(f"Error fetching data and indexing: {e}")
-        index_key = ToolIndex.generate_file_id(
-            tool_id=self.workflow_id,
-            file_hash=file_hash,
-            vector_db=settings[ToolSettingsKey.VECTOR_DB_ADAPTER_ID],
-            embedding=settings[ToolSettingsKey.EMBEDDING_ADAPTER_ID],
-            x2text=settings[ToolSettingsKey.X2TEXT_ADAPTER_ID],
-            chunk_size=settings[SettingsKeys.CHUNK_SIZE],
-            chunk_overlap=settings[SettingsKeys.CHUNK_OVERLAP],
-        )
         # Update GUI
         input_log = (
             "### Indexing file\n"

diff --git a/tools/structure/requirements.txt b/tools/structure/requirements.txt
@@ -1,4 +1,4 @@
 # Add your dependencies here
 
 # Required for all unstract tools
-unstract-sdk~=0.11.1
+unstract-sdk~=0.13.0
diff --git a/tools/structure/src/config/properties.json b/tools/structure/src/config/properties.json
@@ -2,7 +2,7 @@
   "schemaVersion": "0.0.1",
   "displayName": "Structure Tool",
   "functionName": "structure_tool",
-  "toolVersion": "0.0.1",
+  "toolVersion": "0.0.2",
   "description": "This is a template tool which can answer set of input prompts designed in the Prompt Studio",
   "input": {
     "description": "File that needs to be indexed and parsed for answers"

diff --git a/tools/text_extractor/requirements.txt b/tools/text_extractor/requirements.txt
@@ -1,4 +1,4 @@
 # Add your dependencies here
 
 # Required for all unstract tools
-unstract-sdk~=0.11.1
+unstract-sdk~=0.13.0
diff --git a/tools/text_extractor/src/config/properties.json b/tools/text_extractor/src/config/properties.json
@@ -2,7 +2,7 @@
   "schemaVersion": "0.0.1",
   "displayName": "Text Extractor",
   "functionName": "text_extractor",
-  "toolVersion": "0.0.2-v2",
+  "toolVersion": "0.0.2",
   "description": "The Text Extractor is a powerful tool designed to convert documents to its text form or Extract texts from documents",
   "input": {
     "description": "Document"