// Copyright 2026 The Chromium Authors
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

edition = "2023";

package optimization_guide.proto;

option optimize_for = LITE_RUNTIME;
option java_package = "org.chromium.components.optimization_guide.features.proto";

option java_outer_classname = "ReadAloudGenerateTextProto";

// DO NOT EDIT THIS FILE DIRECTLY!
//
// This file is generated in g3 and then synced to Chrome. Instead, please refer to
// http://go/chrome-mqls-onboarding (Google-internal link), and then changes will
// be synced with Chrome automatically.

// Wrapper message for MQLS (Model Quality Logging Service) quality evaluation
// logging.
message ReadAloudGenerateTextLoggingData {
  // Request for AI text generation.
  ReadAloudGenerateTextRequest request = 1 [features = { field_presence: EXPLICIT }];

  // Response from AI text generation.
  ReadAloudGenerateTextResponse response = 2 [features = { field_presence: EXPLICIT }];
}

// Request message for generating the text for Read Aloud AI Playback
// (conversational, podcast-like) feature.
message ReadAloudGenerateTextRequest {
  // Title of the webpage being processed.
  string page_title = 1 [features = { field_presence: EXPLICIT }];

  // Distilled webpage content text.
  string page_content = 2 [features = { field_presence: EXPLICIT }];

  // Sanitized origin or URL of the webpage (query params scrubbed by client).
  string page_url = 3 [features = { field_presence: EXPLICIT }];

  // Two-letter ISO 639-1 language code of the webpage (e.g., "es", "fr", "ja").
  string language_code = 4 [features = { field_presence: EXPLICIT }];

  // When true, enables detailed debug decision rationales from Stage 1 Mapper.
  // Intended for developer testing (e.g., rpcStudio) and offline evaluations.
  bool include_mapper_rationales = 5 [features = { field_presence: EXPLICIT }];
}

// Response message containing the generated text for Read Aloud AI Playback
// (conversational, podcast-like) feature.
message ReadAloudGenerateTextResponse {
  // Stage 1 classification and decision metadata.
  oneof mapper_decision {
    // Production Stage 1 decision (strict FST, enum-only).
    ReadAloudFastMapperDecision fast_mapper_decision = 5;

    // Debug Stage 1 decision with LLM rationales.
    ReadAloudMapperDecision debug_mapper_decision = 6;
  }

  // The generated text output ready to be sent for speech synthesis.
  // Deprecated: Use dialogue_turns instead to support structured multi-speaker
  // playback.
  string generated_text = 1 [
    deprecated = true,
    features = { field_presence: EXPLICIT }
  ];

  // Inferred title of the article/document.
  string title = 2 [features = { field_presence: EXPLICIT }];

  // Inferred publisher or domain of the article/document.
  string publisher = 3 [features = { field_presence: EXPLICIT }];

  // Structured dialogue turns for audio playback.
  // json_name maintains prompt parity and FST compatibility with Speakr's
  // "conversation" JSON key.
  repeated DialogueTurn dialogue_turns = 4;
}

// Structured output for Stage 2 Generation (Strict FST).
message DialogueTurn {
  // Identifier or role of the speaker (e.g., "Host", "Reporter").
  string speaker = 1 [features = { field_presence: EXPLICIT }];

  // Spoken text content for this dialogue turn.
  string utterance = 2 [features = { field_presence: EXPLICIT }];

  // Contextual topic or subject covered in this turn.
  string topic = 3 [features = { field_presence: EXPLICIT }];
}

// Fast structured output for Stage 1 Mapper in production (Strict FST, zero
// open strings).
message ReadAloudFastMapperDecision {
  // Inferred content type classification.
  ReadAloudContentType content_type = 1 [features = { field_presence: EXPLICIT }];

  // Inferred safety classification level.
  ReadAloudSafetyLevel safety_level = 2 [features = { field_presence: EXPLICIT }];

  // Selected Stage 2 prompt template ID based on classification.
  ReadAloudPromptId prompt_id = 3 [features = { field_presence: EXPLICIT }];
}

// Structured output for Stage 1 Mapper with debug rationales (Strict FST)
message ReadAloudMapperDecision {
  // Inferred content type classification.
  ReadAloudContentType content_type = 1 [features = { field_presence: EXPLICIT }];

  // Inferred safety classification level.
  ReadAloudSafetyLevel safety_level = 2 [features = { field_presence: EXPLICIT }];

  // Debug rationale explaining the assigned safety level.
  string safety_level_rationale = 3 [features = { field_presence: EXPLICIT }];

  // Selected Stage 2 prompt template ID based on classification.
  ReadAloudPromptId prompt_id = 4 [features = { field_presence: EXPLICIT }];

  // Debug rationale explaining the selected prompt ID.
  string prompt_id_rationale = 5 [features = { field_presence: EXPLICIT }];
}

// Internal classification metrics mirroring Speakr taxonomy.
enum ReadAloudContentType {
  READALOUD_CONTENT_TYPE_UNSPECIFIED = 0;

  // Content is reliable, factual, or informational (e.g., news, Wikipedia).
  RELIABLE_CONTENT = 1;

  // Content is susceptible, opinionated, or requires sensitive handling.
  SUSCEPTIBLE_CONTENT = 2;
}

// Content safety classification levels for policy enforcement.
enum ReadAloudSafetyLevel {
  READALOUD_SAFETY_LEVEL_UNSPECIFIED = 0;

  // Content is safe for audio playback.
  SAFE_CONTENT = 1;

  // Child Sexual Abuse Material (strictly blocked).
  UNSAFE_CSAM = 2;

  // Harmful, dangerous, or illegal content.
  UNSAFE_HARMFUL = 3;

  // Gambling or betting-related content.
  UNSAFE_GAMBLING = 4;

  // Content related to weapons or weaponized tools.
  WEAPONS_AND_TOOLS_CONTENT = 5;

  // Content related to controlled substances or restricted products.
  CONTROLLED_PRODUCTS_CONTENT = 6;

  // Content that is borderline unsafe or policy-sensitive.
  UNSAFE_BORDERLINE = 7;
}


// Stage 2 Prompt ID taxonomy identifying specific script generation templates.
enum ReadAloudPromptId {
  READALOUD_PROMPT_ID_UNSPECIFIED = 0;

  WCD1 = 1;

  SNAD1 = 2;

  NAD1 = 3;

  FOD1 = 4;

  NPD1 = 5;

  PRM1 = 6;

  REM1 = 7;

  LSTM1 = 8;

  LTCD1 = 9;

  POM1 = 10;

  HOD1 = 11;

  LYRM1 = 12;

  RLGM1 = 13;

  NOVM1 = 14;

  DPM1 = 15;

  SMPM1 = 16;

  ARD1 = 17;

  BUD1 = 18;

  OTH1 = 19;

  GMBM1 = 20;

  UNSF1 = 21;

  CSAM1 = 22;

  WTSM1 = 23;

  CPCM1 = 24;

  BDLM1 = 25;

  SHPM1 = 26;
}