diff --git a/mmv1/products/discoveryengine/DataStore.yaml b/mmv1/products/discoveryengine/DataStore.yaml index 61b868d539ab..700f768d8328 100644 --- a/mmv1/products/discoveryengine/DataStore.yaml +++ b/mmv1/products/discoveryengine/DataStore.yaml @@ -313,6 +313,14 @@ properties: description: | If true, the LLM based annotation is added to the image during parsing. required: false + - name: 'enableLlmLayoutParsing' + type: Boolean + description: | + If true, the pdf layout will be refined using an LLM. + - name: 'enableGetProcessedDocument' + type: Boolean + description: | + If true, the processed document will be made available for the GetProcessedDocument API. - name: 'structuredContentTypes' type: Array description: | @@ -401,6 +409,14 @@ properties: description: | If true, the LLM based annotation is added to the image during parsing. required: false + - name: 'enableLlmLayoutParsing' + type: Boolean + description: | + If true, the pdf layout will be refined using an LLM. + - name: 'enableGetProcessedDocument' + type: Boolean + description: | + If true, the processed document will be made available for the GetProcessedDocument API. - name: 'structuredContentTypes' type: Array description: | diff --git a/mmv1/templates/terraform/samples/services/discoveryengine/discoveryengine_datastore_document_processing_config_layout_full.tf.tmpl b/mmv1/templates/terraform/samples/services/discoveryengine/discoveryengine_datastore_document_processing_config_layout_full.tf.tmpl index b114fcb9544b..e56b5446e25f 100644 --- a/mmv1/templates/terraform/samples/services/discoveryengine/discoveryengine_datastore_document_processing_config_layout_full.tf.tmpl +++ b/mmv1/templates/terraform/samples/services/discoveryengine/discoveryengine_datastore_document_processing_config_layout_full.tf.tmpl @@ -9,12 +9,14 @@ resource "google_discovery_engine_data_store" "document_processing_config_layout document_processing_config { default_parsing_config { layout_parsing_config { - enable_table_annotation = true - enable_image_annotation = true - structured_content_types = ["shareholder-structure"] - exclude_html_elements = ["nav", "footer"] - exclude_html_classes = ["overlay", "screenreader"] - exclude_html_ids = ["cookie-banner"] + enable_table_annotation = true + enable_image_annotation = true + enable_llm_layout_parsing = true + enable_get_processed_document = true + structured_content_types = ["shareholder-structure"] + exclude_html_elements = ["nav", "footer"] + exclude_html_classes = ["overlay", "screenreader"] + exclude_html_ids = ["cookie-banner"] } } chunking_config { @@ -23,5 +25,14 @@ resource "google_discovery_engine_data_store" "document_processing_config_layout include_ancestor_headings = true } } + parsing_config_overrides { + file_type = "pdf" + layout_parsing_config { + enable_table_annotation = true + enable_image_annotation = true + enable_llm_layout_parsing = true + enable_get_processed_document = true + } + } } }