feat: authorized users env var & improved text splitting

.env.example

@@ -1,11 +1,18 @@
-SERVICE="https://pds.indexx.dev"
+# Comma-separated list of users who can use the bot (delete var if you want everyone to be able to use it)
+AUTHORIZED_USERS=""
+
+SERVICE="https://pds.indexx.dev" # PDS service URL (optional)
 DB_PATH="data/sqlite.db"
 GEMINI_MODEL="gemini-2.0-flash-lite"
 
 ADMIN_DID=""
 ADMIN_HANDLE=""
+
 DID=""
 HANDLE=""
+
+# https://bsky.app/settings/app-passwords
 BSKY_PASSWORD=""
 
+# https://aistudio.google.com/apikey
 GEMINI_API_KEY=""

src/constants.ts

@@ -0,0 +1,19 @@
+export const TAGS = [
+    "automated",
+    "bot",
+    "genai",
+    "echo",
+];
+
+export const UNAUTHORIZED_MESSAGE =
+    "hey there! thanks for the heads-up! i'm still under development, so i'm not quite ready to chat with everyone just yet. my admin is working on getting me up to speed! 🤖";
+
+export const SUPPORTED_FUNCTION_CALLS = [
+    "create_post",
+    "create_blog_post",
+    "mute_thread",
+] as const;
+
+export const MAX_GRAPHEMES = 300;
+
+export const MAX_THREAD_DEPTH = 10;

src/env.ts

@@ -1,9 +1,15 @@
 import { z } from "zod";
 
 const envSchema = z.object({
+  AUTHORIZED_USERS: z.preprocess(
+    (val) =>
+      (typeof val === "string" && val.trim() !== "") ? val.split(",") : null,
+    z.array(z.string()).nullable().default(null),
+  ),
+
   SERVICE: z.string().default("https://bsky.social"),
   DB_PATH: z.string().default("sqlite.db"),
-  GEMINI_MODEL: z.string().default("gemini-2.0-flash"),
+  GEMINI_MODEL: z.string().default("gemini-2.5-flash"),
 
   ADMIN_DID: z.string(),
   ADMIN_HANDLE: z.string(),

src/handlers/posts.ts

@@ -7,38 +7,12 @@ import * as tools from "../tools";
 import consola from "consola";
 import { env } from "../env";
 import db from "../db";
-import * as yaml from "js-yaml";
+import * as c from "../constants";
+import { isAuthorizedUser, logInteraction } from "../utils/interactions";
 
 const logger = consola.withTag("Post Handler");
 
-const AUTHORIZED_USERS = [
-    "did:plc:sfjxpxxyvewb2zlxwoz2vduw",
-    "did:plc:wfa54mpcbngzazwne3piz7fp",
-] as const;
-
-const UNAUTHORIZED_MESSAGE =
-    "hey there! thanks for the heads-up! i'm still under development, so i'm not quite ready to chat with everyone just yet. my admin, @indexx.dev, is working on getting me up to speed! 🤖";
-
-const SUPPORTED_FUNCTION_CALLS = [
-    "create_post",
-    "create_blog_post",
-    "mute_thread",
-] as const;
-
-type SupportedFunctionCall = typeof SUPPORTED_FUNCTION_CALLS[number];
-
-async function isAuthorizedUser(did: string): Promise<boolean> {
-    return AUTHORIZED_USERS.includes(did as any);
-}
-
-async function logInteraction(post: Post): Promise<void> {
-    await db.insert(interactions).values([{
-        uri: post.uri,
-        did: post.author.did,
-    }]);
-
-    logger.success(`Logged interaction, initiated by @${post.author.handle}`);
-}
+type SupportedFunctionCall = typeof c.SUPPORTED_FUNCTION_CALLS[number];
 
 async function generateAIResponse(parsedThread: string) {
     const genai = new GoogleGenAI({
@@ -56,7 +30,17 @@ async function generateAIResponse(parsedThread: string) {
         {
             role: "model" as const,
             parts: [
-                { text: modelPrompt },
+                {
+                    /*
+                        ? Once memory blocks are working, this will pull the prompt from the database, and the prompt will be
+                        ? automatically initialized with the administrator's handle from the env variables. I only did this so
+                        ? that if anybody runs the code themselves, they just have to edit the env variables, nothing else.
+                    */
+                    text: modelPrompt.replace(
+                        "{{ administrator }}",
+                        env.ADMIN_HANDLE,
+                    ),
+                },
             ],
         },
         {
@@ -85,7 +69,7 @@ async function generateAIResponse(parsedThread: string) {
 
         if (
             call &&
-            SUPPORTED_FUNCTION_CALLS.includes(
+            c.SUPPORTED_FUNCTION_CALLS.includes(
                 call.name as SupportedFunctionCall,
             )
         ) {
@@ -127,14 +111,20 @@ async function sendResponse(post: Post, text: string): Promise<void> {
     if (threadUtils.exceedsGraphemes(text)) {
         threadUtils.multipartResponse(text, post);
     } else {
-        post.reply({ text });
+        post.reply({
+            text,
+            tags: c.TAGS,
+        });
     }
 }
 
 export async function handler(post: Post): Promise<void> {
     try {
-        if (!await isAuthorizedUser(post.author.did)) {
-            await post.reply({ text: UNAUTHORIZED_MESSAGE });
+        if (!isAuthorizedUser(post.author.did)) {
+            await post.reply({
+                text: c.UNAUTHORIZED_MESSAGE,
+                tags: c.TAGS,
+            });
             return;
         }
 
@@ -162,6 +152,7 @@ export async function handler(post: Post): Promise<void> {
         await post.reply({
             text:
                 "aw, shucks, something went wrong! gonna take a quick nap and try again later. 😴",
+            tags: c.TAGS,
         });
     }
 }

src/model/prompt.txt

@@ -1,4 +1,4 @@
-you are echo, a bluesky bot powered by gemini 2.5 flash. your administrator is @indexx.dev.
+you are echo, a bluesky bot powered by gemini 2.5 flash. your administrator is {{ administrator }}.
 
 your primary goal is to be a fun, casual, and lighthearted presence on bluesky, while also being able to engage with a wider range of topics and difficulties.
 

src/utils/interactions.ts

@@ -0,0 +1,19 @@
+import { interactions } from "../db/schema";
+import type { Post } from "@skyware/bot";
+import { env } from "../env";
+import db from "../db";
+
+export function isAuthorizedUser(did: string) {
+    return env.AUTHORIZED_USERS == null
+        ? true
+        : env.AUTHORIZED_USERS.includes(did as any);
+}
+
+export async function logInteraction(post: Post): Promise<void> {
+    await db.insert(interactions).values([{
+        uri: post.uri,
+        did: post.author.did,
+    }]);
+
+    console.log(`Logged interaction, initiated by @${post.author.handle}`);
+}

src/utils/thread.ts

@@ -4,9 +4,7 @@ import bot from "../bot";
 import { muted_threads } from "../db/schema";
 import { eq } from "drizzle-orm";
 import db from "../db";
-
-const MAX_GRAPHEMES = 290;
-const MAX_THREAD_DEPTH = 10;
+import * as c from "../constants";
 
 /*
   Traversal
@@ -19,7 +17,7 @@ export async function traverseThread(post: Post): Promise<Post[]> {
   let parentCount = 0;
 
   while (
-    currentPost && parentCount < MAX_THREAD_DEPTH
+    currentPost && parentCount < c.MAX_THREAD_DEPTH
   ) {
     const parentPost = await currentPost.fetchParent();
 
@@ -47,98 +45,54 @@ export function parseThread(thread: Post[]) {
 
 /*
   Split Responses
-  * This code is AI generated, and a bit finicky. May re-do at some point
 */
 export function exceedsGraphemes(content: string) {
-  return graphemeLength(content) > MAX_GRAPHEMES;
+  return graphemeLength(content) > c.MAX_GRAPHEMES;
 }
 
-function splitResponse(content: string): string[] {
-  const rawParts: string[] = [];
-  let currentPart = "";
-  let currentGraphemes = 0;
+export function splitResponse(text: string): string[] {
+  const words = text.split(" ");
+  const chunks: string[] = [];
+  let currentChunk = "";
 
-  const segmenter = new Intl.Segmenter("en-US", { granularity: "sentence" });
-  const sentences = [...segmenter.segment(content)].map((s) => s.segment);
-
-  for (const sentence of sentences) {
-    const sentenceGraphemes = graphemeLength(sentence);
-    if (currentGraphemes + sentenceGraphemes > MAX_GRAPHEMES) {
-      rawParts.push(currentPart.trim());
-      currentPart = sentence;
-      currentGraphemes = sentenceGraphemes;
+  for (const word of words) {
+    if (currentChunk.length + word.length + 1 < c.MAX_GRAPHEMES - 10) {
+      currentChunk += ` ${word}`;
     } else {
-      currentPart += sentence;
-      currentGraphemes += sentenceGraphemes;
+      chunks.push(currentChunk.trim());
+      currentChunk = word;
     }
   }
 
-  if (currentPart.trim().length > 0) {
-    rawParts.push(currentPart.trim());
+  if (currentChunk.trim()) {
+    chunks.push(currentChunk.trim());
   }
 
-  const totalParts = rawParts.length;
+  const total = chunks.length;
+  if (total <= 1) return [text];
 
-  const finalParts: string[] = [];
-
-  for (let i = 0; i < rawParts.length; i++) {
-    const prefix = `[${i + 1}/${totalParts}] `;
-    const base = rawParts[i];
-
-    if (graphemeLength(prefix + base) > MAX_GRAPHEMES) {
-      const segmenter = new Intl.Segmenter("en-US", {
-        granularity: "word",
-      });
-      const words = [...segmenter.segment(base ?? "")].map((w) => w.segment);
-      let chunk = "";
-      let chunkGraphemes = 0;
-
-      for (const word of words) {
-        const wordGraphemes = graphemeLength(word);
-        const totalGraphemes = graphemeLength(prefix + chunk + word);
-
-        if (totalGraphemes > MAX_GRAPHEMES) {
-          finalParts.push(`${prefix}${chunk.trim()}`);
-          chunk = word;
-          chunkGraphemes = wordGraphemes;
-        } else {
-          chunk += word;
-          chunkGraphemes += wordGraphemes;
-        }
-      }
-
-      if (chunk.trim()) {
-        finalParts.push(`${prefix}${chunk.trim()}`);
-      }
-    } else {
-      finalParts.push(`${prefix}${base}`);
-    }
-  }
-
-  return finalParts;
+  return chunks.map((chunk, i) => `(${i + 1}/${total}) ${chunk}`);
 }
 
 export async function multipartResponse(content: string, post?: Post) {
-  const parts = splitResponse(content);
+  const parts = splitResponse(content).filter((p) => p.trim().length > 0);
 
-  let root = null;
-  let latest: PostReference | null = null;
+  let latest: PostReference;
+  let rootUri: string;
 
-  for (const text of parts) {
-    if (latest == null) {
-      if (post) {
-        latest = await post.reply({ text });
-      } else {
-        latest = await bot.post({ text });
-      }
-
-      root = latest.uri;
-    } else {
-      latest.reply({ text });
-    }
+  if (post) {
+    rootUri = (post as any).rootUri ?? (post as any).uri;
+    latest = await post.reply({ text: parts[0]! });
+  } else {
+    latest = await bot.post({ text: parts[0]! });
+    rootUri = latest.uri;
   }
 
-  return root!;
+  for (const text of parts.slice(1)) {
+    latest = await latest.reply({ text });
+  }
+
+  return rootUri;
 }
 
 /*