LLM-Driven Extraction of Unstructured Data — Built for API Deployments & ETL Pipeline Workflows
Python ai-agentsdata-engineeringdocument-aigenerative-aiidpjson-extractionllmmcp-serverocrpdf-extractionprompt-engineeringstructured-output
UN-3008 [FIX] Pass word-level confidence setting to Structure Tool (#1714)
Deepak K79ea7a348c33
diff --git a/backend/sample.env b/backend/sample.envindex 08658931e..1ecff5111 100644--- a/backend/sample.env+++ b/backend/sample.env@@ -89,9 +89,9 @@ PROMPT_STUDIO_FILE_PATH=/app/prompt-studio-data# Structure Tool Image (Runs prompt studio exported tools)# https://hub.docker.com/r/unstract/tool-structure-STRUCTURE_TOOL_IMAGE_URL="docker:unstract/tool-structure:0.0.94"+STRUCTURE_TOOL_IMAGE_URL="docker:unstract/tool-structure:0.0.95"STRUCTURE_TOOL_IMAGE_NAME="unstract/tool-structure"-STRUCTURE_TOOL_IMAGE_TAG="0.0.94"+STRUCTURE_TOOL_IMAGE_TAG="0.0.95"# Feature FlagsEVALUATION_SERVER_IP=unstract-fliptdiff --git a/tools/structure/src/config/properties.json b/tools/structure/src/config/properties.jsonindex 6726c9dc3..146691eb0 100644--- 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": "0.0.94",+ "toolVersion": "0.0.95","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/structure/src/constants.py b/tools/structure/src/constants.pyindex 2175a0a95..b4944e3f8 100644--- a/tools/structure/src/constants.py+++ b/tools/structure/src/constants.py@@ -63,6 +63,7 @@ class SettingsKeys:FILE_NAME = "file_name"FILE_HASH = "file_hash"ENABLE_HIGHLIGHT = "enable_highlight"+ ENABLE_WORD_CONFIDENCE = "enable_word_confidence"NAME = "name"INCLUDE_METADATA = "include_metadata"TABLE_SETTINGS = "table_settings"diff --git a/tools/structure/src/main.py b/tools/structure/src/main.pyindex 3900533cf..a097886c7 100644--- a/tools/structure/src/main.py+++ b/tools/structure/src/main.py@@ -160,6 +160,9 @@ class StructureTool(BaseTool):)challenge_llm: str = settings.get(SettingsKeys.CHALLENGE_LLM_ADAPTER_ID, "")is_highlight_enabled: bool = settings.get(SettingsKeys.ENABLE_HIGHLIGHT, False)+ is_word_confidence_enabled: bool = settings.get(+ SettingsKeys.ENABLE_WORD_CONFIDENCE, False+ )responder: PromptTool = PromptTool(tool=self,prompt_port=self.get_env_or_die(SettingsKeys.PROMPT_PORT),@@ -231,6 +234,7 @@ class StructureTool(BaseTool):tool_settings[SettingsKeys.ENABLE_SINGLE_PASS_EXTRACTION] = is_single_pass_enabledtool_settings[SettingsKeys.SUMMARIZE_AS_SOURCE] = is_summarization_enabledtool_settings[SettingsKeys.ENABLE_HIGHLIGHT] = is_highlight_enabled+ tool_settings[SettingsKeys.ENABLE_WORD_CONFIDENCE] = is_word_confidence_enabled_, file_name = os.path.split(input_file)if is_summarization_enabled:file_name = SettingsKeys.SUMMARIZE