feat: Added new messages EntityValidationOutput and EntitiesRevision

Google APIs · copybara-github · commit 1a87094dab8a · 2025-10-08T14:10:04.000-07:00
feat: Added new fields `entity_validation_output`, `entities_revisions`, and `entities_revision_id` to the `Document` message
docs: The field `training_method` in `TrainProcessorVersionRequest` is now marked as optional

PiperOrigin-RevId: 816853859
diff --git a/google/cloud/documentai/v1/document.proto b/google/cloud/documentai/v1/document.proto
@@ -1069,6 +1069,62 @@ message Document {
     repeated Chunk chunks = 1;
   }
 
+  // The output of the validation given the document and the validation rules.
+  message EntityValidationOutput {
+    // Validation result for a single validation rule.
+    message ValidationResult {
+      // The result of the validation rule.
+      enum ValidationResultType {
+        // The validation result type is unspecified.
+        VALIDATION_RESULT_TYPE_UNSPECIFIED = 0;
+
+        // The validation is valid.
+        VALIDATION_RESULT_TYPE_VALID = 1;
+
+        // The validation is invalid.
+        VALIDATION_RESULT_TYPE_INVALID = 2;
+
+        // The validation is skipped.
+        VALIDATION_RESULT_TYPE_SKIPPED = 3;
+
+        // The validation is not applicable.
+        VALIDATION_RESULT_TYPE_NOT_APPLICABLE = 4;
+      }
+
+      // The name of the validation rule.
+      string rule_name = 1;
+
+      // The description of the validation rule.
+      string rule_description = 2;
+
+      // The result of the validation rule.
+      ValidationResultType validation_result_type = 3;
+
+      // The detailed information of the running the validation process using
+      // the entity from the document based on the validation rule.
+      string validation_details = 4;
+    }
+
+    // The result of each validation rule.
+    repeated ValidationResult validation_results = 1;
+
+    // The overall result of the validation, true if all applicable rules are
+    // valid.
+    bool pass_all_rules = 2;
+  }
+
+  // Entity revision.
+  message EntitiesRevision {
+    // The revision id.
+    string revision_id = 1;
+
+    // The entities in this revision.
+    repeated Entity entities = 2;
+
+    // The entity validation output for this revision.
+    EntityValidationOutput entity_validation_output = 3;
+  }
+
   // Original source document from the user.
   oneof source {
     // Optional. Currently supports Google Cloud Storage URI of the form
@@ -1129,4 +1185,20 @@ message Document {
 
   // Document chunked based on chunking config.
   ChunkedDocument chunked_document = 18;
+
+  // The entity validation output for the document. This is the validation
+  // output for `document.entities` field.
+  EntityValidationOutput entity_validation_output = 21;
+
+  // A list of entity revisions. The entity revisions are appended to the
+  // document in the processing order. This field can be used for comparing the
+  // entity extraction results at different stages of the processing.
+  repeated EntitiesRevision entities_revisions = 22;
+
+  // 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
+  // `entity_validation_output` field in this entity revision.
+  string entities_revision_id = 23;
 }
diff --git a/google/cloud/documentai/v1/document_processor_service.proto b/google/cloud/documentai/v1/document_processor_service.proto
@@ -1008,8 +1008,8 @@ message TrainProcessorVersionRequest {
       TEMPLATE_BASED = 2;
     }
 
-    // Training method to use for CDE training.
-    TrainingMethod training_method = 3;
+    // Optional. Training method to use for CDE training.
+    TrainingMethod training_method = 3 [(google.api.field_behavior) = OPTIONAL];
   }
 
   // Options to control foundation model tuning of the processor.

Original file line number	Diff line number	Diff line change
`@@ -1008,8 +1008,8 @@ message TrainProcessorVersionRequest {`
`1008`	`1008`	`TEMPLATE_BASED = 2;`
`1009`	`1009`	`}`
`1010`	`1010`
`1011`		`- // Training method to use for CDE training.`
`1012`		`- TrainingMethod training_method = 3;`
	`1011`	`+ // Optional. Training method to use for CDE training.`
	`1012`	`+ TrainingMethod training_method = 3 [(google.api.field_behavior) = OPTIONAL];`
`1013`	`1013`	`}`
`1014`	`1014`
`1015`	`1015`	`// Options to control foundation model tuning of the processor.`