feat: Added a fields for image and table annotation output

Google APIs · copybara-github · commit 280bed6c2196 · 2026-04-01T11:25:19.000-07:00
feat: Update comments for proto fields

PiperOrigin-RevId: 893028748
diff --git a/google/cloud/documentai/v1/document.proto b/google/cloud/documentai/v1/document.proto
@@ -1,4 +1,4 @@
-// Copyright 2025 Google LLC
+// Copyright 2026 Google LLC
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -17,6 +17,7 @@ syntax = "proto3";
 package google.cloud.documentai.v1;
 
 import "google/api/field_behavior.proto";
+import "google/api/resource.proto";
 import "google/cloud/documentai/v1/barcode.proto";
 import "google/cloud/documentai/v1/geometry.proto";
 import "google/protobuf/timestamp.proto";
@@ -180,9 +181,9 @@ message Document {
 
       // Confidence of the current
       // [Layout][google.cloud.documentai.v1.Document.Page.Layout] within
-      // context of the object this layout is for. e.g. confidence can be for a
-      // single token, a table, a visual element, etc. depending on context.
-      // Range `[0, 1]`.
+      // context of the object this layout is for. For example, confidence can
+      // be for a single token, a table, a visual element, etc. depending on
+      // context. Range `[0, 1]`.
       float confidence = 2;
 
       // The bounding polygon for the
@@ -340,8 +341,8 @@ message Document {
       repeated DetectedLanguage detected_languages = 2;
     }
 
-    // Detected non-text visual elements e.g. checkbox, signature etc. on the
-    // page.
+    // Detected non-text visual elements, for example, checkbox, signature etc.
+    // on the page.
     message VisualElement {
       // [Layout][google.cloud.documentai.v1.Document.Page.Layout] for
       // [VisualElement][google.cloud.documentai.v1.Document.Page.VisualElement].
@@ -400,7 +401,7 @@ message Document {
     message FormField {
       // [Layout][google.cloud.documentai.v1.Document.Page.Layout] for the
       // [FormField][google.cloud.documentai.v1.Document.Page.FormField] name.
-      // e.g. `Address`, `Email`, `Grand total`, `Phone number`, etc.
+      // For example, `Address`, `Email`, `Grand total`, `Phone number`, etc.
       Layout field_name = 1;
 
       // [Layout][google.cloud.documentai.v1.Document.Page.Layout] for the
@@ -526,7 +527,7 @@ message Document {
     // A list of visually detected tokens on the page.
     repeated Token tokens = 8;
 
-    // A list of detected non-text visual elements e.g. checkbox,
+    // A list of detected non-text visual elements, for example, checkbox,
     // signature etc. on the page.
     repeated VisualElement visual_elements = 9;
 
@@ -586,6 +587,8 @@ message Document {
         // Float value.
         float float_value = 8;
 
+        // A signature - a graphical representation of a person's name,
+        // often used to sign a document.
         bool signature_value = 10;
       }
 
@@ -622,10 +625,11 @@ message Document {
     // [Document.text][google.cloud.documentai.v1.Document.text].
     TextAnchor text_anchor = 1 [(google.api.field_behavior) = OPTIONAL];
 
-    // Required. Entity type from a schema e.g. `Address`.
+    // Required. Entity type from a schema, for example, `Address`.
     string type = 2 [(google.api.field_behavior) = REQUIRED];
 
-    // Optional. Text value of the entity e.g. `1600 Amphitheatre Pkwy`.
+    // Optional. Text value of the entity, for example, `1600 Amphitheatre
+    // Pkwy`.
     string mention_text = 3 [(google.api.field_behavior) = OPTIONAL];
 
     // Optional. Deprecated.  Use `id` field instead.
@@ -643,9 +647,9 @@ message Document {
     string id = 7 [(google.api.field_behavior) = OPTIONAL];
 
     // Optional. Normalized entity value. Absent if the extracted value could
-    // not be converted or the type (e.g. address) is not supported for certain
-    // parsers. This field is also only populated for certain supported document
-    // types.
+    // not be converted or the type (for example, address) is not supported for
+    // certain parsers. This field is also only populated for certain supported
+    // document types.
     NormalizedValue normalized_value = 9
         [(google.api.field_behavior) = OPTIONAL];
 
@@ -854,7 +858,7 @@ message Document {
   message Revision {
     // Human Review information of the document.
     message HumanReview {
-      // Human review state. e.g. `requested`, `succeeded`, `rejected`.
+      // Human review state. For example, `requested`, `succeeded`, `rejected`.
       string state = 1;
 
       // A message providing more details about the current state of processing.
@@ -912,6 +916,12 @@ message Document {
     repeated Provenance provenance = 3 [deprecated = true];
   }
 
+  // Represents the annotation of a block or a chunk.
+  message Annotations {
+    // The description of the content with this annotation.
+    string description = 1;
+  }
+
   // Represents the parsed layout of a document as a collection of blocks that
   // the document is divided into.
   message DocumentLayout {
@@ -940,6 +950,9 @@ message Document {
         // A text block could further have child blocks.
         // Repeated blocks support further hierarchies and nested blocks.
         repeated DocumentLayoutBlock blocks = 3;
+
+        // Annotation of the text block.
+        Annotations annotations = 4;
       }
 
       // Represents a table type block.
@@ -952,6 +965,9 @@ message Document {
 
         // Table caption/title.
         string caption = 3;
+
+        // Annotation of the table block.
+        Annotations annotations = 4;
       }
 
       // Represents a row in a table.
@@ -990,6 +1006,37 @@ message Document {
         repeated DocumentLayoutBlock blocks = 1;
       }
 
+      // Represents an image type block.
+      message LayoutImageBlock {
+        // Source of the image.
+        oneof image_source {
+          // Optional. Asset id of the inline image. If set, find the image
+          // content in the blob_assets field.
+          string blob_asset_id = 4 [(google.api.field_behavior) = OPTIONAL];
+
+          // Optional. Google Cloud Storage uri of the image.
+          string gcs_uri = 5 [(google.api.field_behavior) = OPTIONAL];
+
+          // Optional. Data uri of the image.
+          // It is composed of four parts: a prefix (data:), a MIME type
+          // indicating the type of data, an optional base64 token if
+          // non-textual, and the data itself:
+          // data:[<mediatype>][;base64],<data>
+          string data_uri = 6 [(google.api.field_behavior) = OPTIONAL];
+        }
+
+        // Mime type of the image. An IANA published [media type (MIME type)]
+        // (https://www.iana.org/assignments/media-types/media-types.xhtml).
+        string mime_type = 1;
+
+        // Text extracted from the image using OCR or alt text describing the
+        // image.
+        string image_text = 2;
+
+        // Annotation of the image block.
+        Annotations annotations = 3;
+      }
+
       oneof block {
         // Block consisting of text content.
         LayoutTextBlock text_block = 2;
@@ -999,6 +1046,9 @@ message Document {
 
         // Block consisting of list content/structure.
         LayoutListBlock list_block = 4;
+
+        // Block consisting of image content.
+        LayoutImageBlock image_block = 7;
       }
 
       // ID of the block.
@@ -1046,6 +1096,48 @@ message Document {
         ChunkPageSpan page_span = 2;
       }
 
+      // The image chunk field in the chunk.
+      message ImageChunkField {
+        // Source of the image.
+        oneof image_source {
+          // Optional. Asset id of the inline image. If set, find the image
+          // content in the blob_assets field.
+          string blob_asset_id = 1 [(google.api.field_behavior) = OPTIONAL];
+
+          // Optional. Google Cloud Storage uri of the image.
+          string gcs_uri = 2 [(google.api.field_behavior) = OPTIONAL];
+
+          // Optional. Data uri of the image.
+          // It is composed of four parts: a prefix (data:), a MIME type
+          // indicating the type of data, an optional base64 token if
+          // non-textual, and the data itself:
+          // data:[<mediatype>][;base64],<data>
+          string data_uri = 3 [(google.api.field_behavior) = OPTIONAL];
+        }
+
+        // Annotation of the image chunk field.
+        Annotations annotations = 4;
+      }
+
+      // The table chunk field in the chunk.
+      message TableChunkField {
+        // Annotation of the table chunk field.
+        Annotations annotations = 1;
+      }
+
+      // The chunk field in the chunk. A chunk field could be one of the various
+      // types (for example, image, table) supported.
+      message ChunkField {
+        // The type of the chunk field.
+        oneof field_type {
+          // The image chunk field in the chunk.
+          ImageChunkField image_chunk_field = 1;
+
+          // The table chunk field in the chunk.
+          TableChunkField table_chunk_field = 2;
+        }
+      }
+
       // ID of the chunk.
       string chunk_id = 1;
 
@@ -1063,12 +1155,31 @@ message Document {
 
       // Page footers associated with the chunk.
       repeated ChunkPageFooter page_footers = 6;
+
+      // Chunk fields inside this chunk.
+      repeated ChunkField chunk_fields = 7;
     }
 
     // List of chunks.
     repeated Chunk chunks = 1;
   }
 
+  // Represents a blob asset. It's used to store the content of the inline blob
+  // in this document, for example, image bytes, such that it can be referenced
+  // by other fields in the document via asset ID.
+  message BlobAsset {
+    // Optional. The id of the blob asset.
+    string asset_id = 1 [(google.api.field_behavior) = OPTIONAL];
+
+    // Optional. The content of the blob asset, for example, image bytes.
+    bytes content = 2 [(google.api.field_behavior) = OPTIONAL];
+
+    // The mime type of the blob asset.
+    // An IANA published [media type (MIME
+    // type)](https://www.iana.org/assignments/media-types/media-types.xhtml).
+    string mime_type = 3;
+  }
+
   // The output of the validation given the document and the validation rules.
   message EntityValidationOutput {
     // Validation result for a single validation rule.
@@ -1186,6 +1297,12 @@ message Document {
   // Document chunked based on chunking config.
   ChunkedDocument chunked_document = 18;
 
+  // Optional. The blob assets in this document. This is used to store the
+  // content of the inline blobs in this document, for example, image bytes,
+  // such that it can be referenced by other fields in the document via asset
+  // id.
+  repeated BlobAsset blob_assets = 19 [(google.api.field_behavior) = OPTIONAL];
+
   // The entity validation output for the document. This is the validation
   // output for `document.entities` field.
   EntityValidationOutput entity_validation_output = 21;
@@ -1198,7 +1315,7 @@ message Document {
   // The entity revision ID that `document.entities` field is based on.
   // If this field is set and `entities_revisions` is not empty, the entities in
   // `document.entities` field are the entities in the entity revision with this
-  // ID and `document.entity_validation_output` field is the
+  // id and `document.entity_validation_output` field is the
   // `entity_validation_output` field in this entity revision.
   string entities_revision_id = 23;
 }
