Add topics to advanced_runner comments output

GitOrigin-RevId: 0b0adb3a60e5427c46487a900ca8b5d8e0f40291

Author: alyssachvasta

runner-cli/advanced_runner.ts

@@ -35,7 +35,7 @@
 
 import { Command } from "commander";
 import { writeFileSync } from "fs";
-import { getCommentsFromCsv, getSummary } from "./runner_utils";
+import { concatTopics, getCommentsFromCsv, getSummary } from "./runner_utils";
 import { MajoritySummaryStats } from "../src/stats/majority_vote";
 import { TopicStats } from "../src/stats/summary_stats";
 import { RelativeContext } from "../src/tasks/summarization_subtasks/relative_context";
@@ -58,6 +58,8 @@ interface CommentWithScores {
   id: string;
   text: string;
   votes?: VoteInfo;
+  topics?: string;
+
   agreeRate?: number;
   disagreeRate?: number;
   passRate?: number;
@@ -114,7 +116,9 @@ function getCommentsWithScores(
       id: comment.id,
       text: comment.text,
       votes: comment.voteInfo,
+      topics: concatTopics(comment),
     };
+
     if (comment.voteInfo) {
       const commentWithVoteInfo = comment as CommentWithVoteInfo;
       commentWithScores.passRate = getTotalPassRate(comment.voteInfo, USE_PROBABILITY_ESTIMATES);

runner-cli/categorization_runner.ts

@@ -33,6 +33,7 @@ import { parse } from "csv-parse";
 import { createObjectCsvWriter } from "csv-writer";
 import * as fs from "fs";
 import * as path from "path";
+import { concatTopics } from "./runner_utils";
 
 type CommentCsvRow = {
   "comment-id": string;
@@ -150,31 +151,6 @@ function setTopics(csvRows: CommentCsvRow[], categorizedComments: Comment[]): Co
   return csvRowsWithTopics;
 }
 
-// Returns topics and subtopics concatenated together like
-// "Transportation:PublicTransit;Transportation:Parking;Technology:Internet"
-function concatTopics(comment: Comment): string {
-  const pairsArray = [];
-  for (const topic of comment.topics || []) {
-    if ("subtopics" in topic) {
-      for (const subtopic of topic.subtopics || []) {
-        if ("subtopics" in subtopic) {
-          if ("subtopics" in (subtopic as Topic)) {
-            for (const subsubtopic of subtopic.subtopics as Topic[]) {
-              pairsArray.push(`${topic.name}:${subtopic.name}:${subsubtopic.name}`);
-            }
-          }
-        } else {
-          pairsArray.push(`${topic.name}:${subtopic.name}`);
-        }
-      }
-    } else {
-      // handle case where no subtopics available
-      pairsArray.push(`${topic.name}`);
-    }
-  }
-  return pairsArray.join(";");
-}
-
 async function writeCsv(csvRows: CommentCsvRow[], outputFile: string) {
   // Expect that all objects have the same keys, and make id match header title
   const header: { id: string; title: string }[] = [];

runner-cli/runner_utils.ts

@@ -205,6 +205,31 @@ export function writeSummaryToHtml(summary: Summary, outputFile: string) {
   console.log(`Written summary to ${outputFile}`);
 }
 
+// Returns topics and subtopics concatenated together like
+// "Transportation:PublicTransit;Transportation:Parking;Technology:Internet"
+export function concatTopics(comment: Comment): string {
+  const pairsArray = [];
+  for (const topic of comment.topics || []) {
+    if ("subtopics" in topic) {
+      for (const subtopic of topic.subtopics || []) {
+        if ("subtopics" in subtopic && (subtopic.subtopics as Topic[]).length) {
+          if ("subtopics" in (subtopic as Topic)) {
+            for (const subsubtopic of subtopic.subtopics as Topic[]) {
+              pairsArray.push(`${topic.name}:${subtopic.name}:${subsubtopic.name}`);
+            }
+          }
+        } else {
+          pairsArray.push(`${topic.name}:${subtopic.name}`);
+        }
+      }
+    } else {
+      // handle case where no subtopics available
+      pairsArray.push(`${topic.name}`);
+    }
+  }
+  return pairsArray.join(";");
+}
+
 /**
  * Parse a topics string from the categorization_runner.ts into a (possibly) nested topics
  * array, omitting subtopics and subsubtopics if not present in the labels.