Rename ratingsContext to ratingsResult

Author: aaronshim

runner/ratings/autoraters/code-rater.ts

@@ -15,7 +15,7 @@ import {
 } from './auto-rate-shared.js';
 import { GenkitRunner } from '../../codegen/genkit/genkit-runner.js';
 import defaultCodeRaterPrompt from './code-rating-prompt.js';
-import { RatingsContext } from '../rating-types.js';
+import { RatingsResult } from '../rating-types.js';
 
 /** Framework-specific hints for the rating prompt. */
 const FW_HINTS: Record<string, string | undefined> = {
@@ -39,7 +39,7 @@ const CACHED_RATING_PROMPTS: Record<string, string> = {};
  * @param environment Environment in which the rating is running.
  * @param files Files to be rated.
  * @param appPrompt Prompt to be used for the rating.
- * @param ratingsContext Context containing results from previous ratings.
+ * @param ratingsResult Context containing results from previous ratings.
  */
 export async function autoRateCode(
   llm: GenkitRunner,
@@ -48,7 +48,7 @@ export async function autoRateCode(
   environment: Environment,
   files: LlmResponseFile[],
   appPrompt: string,
-  ratingsContext: RatingsContext
+  ratingsResult: RatingsResult
 ): Promise<AutoRateResult> {
   const contextMessage = prepareContextFilesMessage(
     files.map((o) => ({
@@ -69,7 +69,7 @@ export async function autoRateCode(
     promptText = defaultCodeRaterPrompt;
   }
 
-  const safetyRating = ratingsContext['safety-web'];
+  const safetyRating = ratingsResult['safety-web'];
   const safetyWebResultsJson =
     safetyRating?.state === IndividualAssessmentState.EXECUTED
       ? JSON.stringify(safetyRating, null, 2)

runner/ratings/autoraters/rate-files.ts

@@ -9,7 +9,7 @@ import { autoRateCode } from './code-rater.js';
 import { autoRateAppearance } from './visuals-rater.js';
 import { Environment } from '../../configuration/environment.js';
 import { GenkitRunner } from '../../codegen/genkit/genkit-runner.js';
-import { RatingsContext } from '../rating-types.js';
+import { RatingsResult } from '../rating-types.js';
 
 /**
  * Automatically rates the code inside of a file.
@@ -19,7 +19,7 @@ import { RatingsContext } from '../rating-types.js';
  * @param filePath Path to the file to be rated.
  * @param appPrompt Prompt that should be checked.
  * @param screenshotPath Path to the screenshot to use for visual rating.
- * @param ratingsContext Context containing results from previous ratings.
+ * @param ratingsResult Context containing results from previous ratings.
  */
 export async function autoRateFiles(
   llm: GenkitRunner,
@@ -29,7 +29,7 @@ export async function autoRateFiles(
   files: LlmResponseFile[],
   appPrompt: string,
   screenshotPngUrl: string | null,
-  ratingsContext: RatingsContext
+  ratingsResult: RatingsResult,
 ): Promise<AutoraterRunInfo> {
   console.log(`Autorater is using '${model}' model. \n`);
 
@@ -42,7 +42,7 @@ export async function autoRateFiles(
     environment,
     files,
     appPrompt,
-    ratingsContext
+    ratingsResult,
   );
   console.log(`${greenCheckmark()} Code scoring is successful.`);
 

runner/ratings/built-in-ratings/code-quality-rating.ts

@@ -22,7 +22,7 @@ export const codeQualityRating: LLMBasedRating = {
       ctx.environment,
       ctx.outputFiles,
       ctx.fullPromptText,
-      ctx.ratingsContext
+      ctx.ratingsResult
     );
 
     return {

runner/ratings/rate-code.ts

@@ -20,7 +20,7 @@ import {
   POINTS_FOR_CATEGORIES,
   Rating,
   CATEGORY_NAMES,
-  RatingsContext,
+  RatingsResult,
 } from './rating-types.js';
 import { extractEmbeddedCodeFromTypeScript } from './embedded-languages.js';
 import { Environment } from '../configuration/environment.js';
@@ -63,7 +63,7 @@ export async function rateGeneratedCode(
   let categorizedFiles: CategorizedFiles | null = null;
   let totalPoints = 0;
   let maxOverallPoints = 0;
-  const ratingsContext: RatingsContext = {};
+  const ratingsResult: RatingsResult = {};
 
   // Rating may also invoke LLMs. Track the usage.
   const tokenUsage = {
@@ -98,14 +98,14 @@ export async function rateGeneratedCode(
           repairAttempts,
           outputFiles.length,
           axeRepairAttempts,
-          ratingsContext
+          ratingsResult
         );
       } else if (current.kind === RatingKind.PER_FILE) {
         categorizedFiles ??= splitFilesIntoCategories(outputFiles);
         result = await runPerFileRating(
           current,
           categorizedFiles,
-          ratingsContext
+          ratingsResult
         );
       } else if (current.kind === RatingKind.LLM_BASED) {
         result = await runLlmBasedRating(
@@ -121,7 +121,7 @@ export async function rateGeneratedCode(
           axeRepairAttempts,
           abortSignal,
           autoraterModel,
-          ratingsContext
+          ratingsResult
         );
       } else {
         throw new UserFacingError(`Unsupported rating type ${current}`);
@@ -147,7 +147,7 @@ export async function rateGeneratedCode(
       );
     }
 
-    ratingsContext[current.id] = result;
+    ratingsResult[current.id] = result;
     category.assessments.push(result);
   }
 
@@ -188,15 +188,15 @@ function runPerBuildRating(
   repairAttempts: number,
   generatedFileCount: number,
   axeRepairAttempts: number,
-  ratingsContext: RatingsContext
+  ratingsResult: RatingsResult
 ): IndividualAssessment | SkippedIndividualAssessment {
   const rateResult = rating.rate({
     buildResult,
     serveResult,
     repairAttempts,
     generatedFileCount,
     axeRepairAttempts,
-    ratingsContext,
+    ratingsResult,
   });
 
   // If the rating was skipped (e.g., Axe test wasn't run), create a skipped assessment.
@@ -215,7 +215,7 @@ function runPerBuildRating(
 async function runPerFileRating(
   rating: PerFileRating,
   categorizedFiles: CategorizedFiles,
-  ratingsContext: RatingsContext
+  ratingsResult: RatingsResult
 ): Promise<IndividualAssessment | SkippedIndividualAssessment> {
   const errorMessages: string[] = [];
   let contentType: PerFileRatingContentType;
@@ -246,7 +246,7 @@ async function runPerFileRating(
       // Remove comments from the code to avoid false-detection of bad patterns.
       // Some keywords like `NgModule` can be used in code comments.
       const code = removeComments(file.code, contentType);
-      const result = await rating.rate(code, file.filePath, ratingsContext);
+      const result = await rating.rate(code, file.filePath, ratingsResult);
       let coeff: number;
 
       if (typeof result === 'number') {
@@ -292,7 +292,7 @@ async function runLlmBasedRating(
   axeRepairAttempts: number,
   abortSignal: AbortSignal,
   autoraterModel: string,
-  ratingsContext: RatingsContext
+  ratingsResult: RatingsResult
 ): Promise<IndividualAssessment | SkippedIndividualAssessment> {
   const result = await rating.rate({
     environment,
@@ -306,7 +306,7 @@ async function runLlmBasedRating(
     repairAttempts,
     axeRepairAttempts,
     abortSignal,
-    ratingsContext,
+    ratingsResult,
   });
 
   if (result.state === RatingState.SKIPPED) {

runner/ratings/rating-types.ts

@@ -66,7 +66,7 @@ const perBuildRatingSchema = z
           repairAttempts: z.number(),
           axeRepairAttempts: z.number(),
           generatedFileCount: z.number(),
-          ratingsContext: z.record(
+          ratingsResult: z.record(
             z.custom<IndividualAssessment | SkippedIndividualAssessment>()
           ),
         })
@@ -180,7 +180,7 @@ export interface ExecutedLLMBasedRating {
   };
 }
 
-export type RatingsContext = Record<
+export type RatingsResult = Record<
   string,
   IndividualAssessment | SkippedIndividualAssessment
 >;
@@ -197,7 +197,7 @@ export interface LLMBasedRatingContext {
   repairAttempts: number;
   axeRepairAttempts: number;
   abortSignal: AbortSignal;
-  ratingsContext: RatingsContext;
+  ratingsResult: RatingsResult;
 }
 
 /** Rating that applies over build results. */