add long transcript seed data

Author: yujonglee

apps/desktop/src/components/devtool/seed/index.ts

@@ -1,6 +1,11 @@
 import type { SeedDefinition } from "./shared";
-import { curatedSeed, emptySeed, randomSeed } from "./versions";
+import { curatedSeed, emptySeed, longSeed, randomSeed } from "./versions";
 
 export { type SeedDefinition } from "./shared";
 
-export const seeds: SeedDefinition[] = [emptySeed, randomSeed, curatedSeed];
+export const seeds: SeedDefinition[] = [
+  emptySeed,
+  randomSeed,
+  longSeed,
+  curatedSeed,
+];

apps/desktop/src/components/devtool/seed/shared/builders.ts

@@ -19,7 +19,6 @@ import type {
   WordStorage,
 } from "@hypr/store";
 
-import { DEFAULT_USER_ID, id } from "../../../../utils";
 import { createCalendar } from "./calendar";
 import { createChatGroup, createChatMessage } from "./chat";
 import { createChatShortcut } from "./chat-shortcut";
@@ -274,36 +273,19 @@ export const buildTranscriptsForSessions = (
   const words: Record<string, WordStorage> = {};
 
   sessionIds.forEach((sessionId) => {
-    const transcriptId = id();
-    const createdAt = faker.date.recent({ days: 30 });
-    const startedAt = createdAt.getTime();
-
-    const transcript = generateTranscript();
-    let maxEndMs = 0;
-
-    transcript.words.forEach((word) => {
-      if (word.end_ms !== undefined && word.end_ms > maxEndMs) {
-        maxEndMs = word.end_ms;
-      }
-      const wordId = id();
-      words[wordId] = {
-        user_id: DEFAULT_USER_ID,
-        transcript_id: transcriptId,
-        text: word.text,
-        start_ms: word.start_ms,
-        end_ms: word.end_ms,
-        channel: word.channel,
-        created_at: faker.date.recent({ days: 30 }).toISOString(),
-      };
+    const result = generateTranscript({ sessionId });
+
+    if (!("transcript" in result)) {
+      throw new Error("Expected transcript metadata");
+    }
+
+    const { transcriptId, transcript, words: transcriptWords } = result;
+
+    Object.entries(transcriptWords).forEach(([wordId, word]) => {
+      words[wordId] = word;
     });
 
-    transcripts[transcriptId] = {
-      user_id: DEFAULT_USER_ID,
-      session_id: sessionId,
-      created_at: createdAt.toISOString(),
-      started_at: startedAt,
-      ended_at: maxEndMs > 0 ? startedAt + maxEndMs : undefined,
-    };
+    transcripts[transcriptId] = transcript;
   });
 
   return { transcripts, words };

apps/desktop/src/components/devtool/seed/shared/transcript.ts

@@ -1,9 +1,19 @@
 import { faker } from "@faker-js/faker";
 
-import type { WordStorage } from "@hypr/store";
+import type { Transcript, WordStorage } from "@hypr/store";
 
 import { DEFAULT_USER_ID, id } from "../../../../utils";
 
+type TranscriptWithMetadata = {
+  transcriptId: string;
+  transcript: Transcript;
+  words: Record<string, WordStorage>;
+};
+
+type TranscriptWordsOnly = {
+  words: Array<WordStorage>;
+};
+
 const selectWeighted = <T,>(choices: Array<{ weight: number; value: T }>): T =>
   faker.helpers.weightedArrayElement(choices);
 
@@ -80,15 +90,37 @@ const generateSentence = () => {
   return sentenceWords;
 };
 
-export const generateTranscript = () => {
+export function generateTranscript(options: {
+  turnCount?: { min: number; max: number };
+  days?: number;
+  sessionId: string;
+}): TranscriptWithMetadata;
+export function generateTranscript(options?: {
+  turnCount?: { min: number; max: number };
+  days?: number;
+  sessionId?: undefined;
+}): TranscriptWordsOnly;
+export function generateTranscript(options?: {
+  turnCount?: { min: number; max: number };
+  days?: number;
+  sessionId?: string;
+}): TranscriptWithMetadata | TranscriptWordsOnly {
+  const {
+    turnCount: turnCountRange = { min: 200, max: 500 },
+    days = 30,
+    sessionId,
+  } = options ?? {};
+
   const channelCount = 2;
-  const turnCount = faker.number.int({ min: 10, max: 20 });
+  const turnCount = faker.number.int(turnCountRange);
 
   const transcriptId = id();
   const words: Array<WordStorage> = [];
   let currentTimeMs = 0;
   let currentChannel = 0;
-  const createdAt = faker.date.recent({ days: 30 }).toISOString();
+  const createdAt = faker.date.recent({ days });
+  const createdAtStr = createdAt.toISOString();
+  const startedAt = createdAt.getTime();
 
   for (let turnIndex = 0; turnIndex < turnCount; turnIndex++) {
     const sentenceCount = selectWeighted([
@@ -119,7 +151,7 @@ export const generateTranscript = () => {
 
         words.push({
           user_id: DEFAULT_USER_ID,
-          created_at: createdAt,
+          created_at: createdAtStr,
           transcript_id: transcriptId,
           channel: currentChannel,
           text: ` ${text}`,
@@ -159,5 +191,25 @@ export const generateTranscript = () => {
     currentChannel = (currentChannel + 1) % channelCount;
   }
 
+  if (sessionId) {
+    const wordsRecord: Record<string, WordStorage> = {};
+    words.forEach((word) => {
+      const wordId = id();
+      wordsRecord[wordId] = word;
+    });
+
+    return {
+      transcriptId,
+      transcript: {
+        user_id: DEFAULT_USER_ID,
+        session_id: sessionId,
+        created_at: createdAtStr,
+        started_at: startedAt,
+        ended_at: startedAt + currentTimeMs,
+      },
+      words: wordsRecord,
+    };
+  }
+
   return { words };
-};
+}

apps/desktop/src/components/devtool/seed/versions/index.ts

@@ -1,3 +1,4 @@
 export { curatedSeed } from "./curated";
 export { emptySeed } from "./empty";
+export { longSeed } from "./long";
 export { randomSeed } from "./random";

apps/desktop/src/components/devtool/seed/versions/long.ts

@@ -0,0 +1,51 @@
+import { faker } from "@faker-js/faker";
+import type { Tables } from "tinybase/with-schemas";
+
+import type { Schemas } from "../../../../store/tinybase/main";
+import type { Store as PersistedStore } from "../../../../store/tinybase/main";
+import type { SeedDefinition } from "../shared";
+import {
+  createHuman,
+  createOrganization,
+  createSession,
+  generateTranscript,
+} from "../shared";
+
+faker.seed(789);
+
+const LONG_DATA = (() => {
+  const organization = createOrganization();
+  const human = createHuman(organization.id, true);
+  const session = createSession();
+
+  const result = generateTranscript({
+    turnCount: { min: 800, max: 1200 },
+    days: 7,
+    sessionId: session.id,
+  });
+
+  if (!("transcript" in result)) {
+    throw new Error("Expected transcript metadata");
+  }
+
+  const { transcriptId, transcript, words } = result;
+
+  return {
+    organizations: { [organization.id]: organization.data },
+    humans: { [human.id]: human.data },
+    sessions: { [session.id]: session.data },
+    transcripts: { [transcriptId]: transcript },
+    words,
+  } satisfies Tables<Schemas[0]>;
+})();
+
+export const longSeed: SeedDefinition = {
+  id: "long",
+  label: "Long",
+  run: (store: PersistedStore) => {
+    store.transaction(() => {
+      store.delTables();
+      store.setTables(LONG_DATA);
+    });
+  },
+};