@ztimson/ai-utils 0.6.5 → 0.6.7

package/dist/index.mjs

@@ -1,40 +1,38 @@
-import * as P from "node:os";
-import { objectMap as x, JSONAttemptParse as b, findByProp as S, JSONSanitize as w, clean as q, Http as $, consoleInterceptor as M, fn as E, ASet as A } from "@ztimson/utils";
-import { Anthropic as O } from "@anthropic-ai/sdk";
-import { OpenAI as v } from "openai";
-import { Worker as R } from "worker_threads";
-import { fileURLToPath as U } from "url";
-import { join as z, dirname as L } from "path";
-import { spawn as k } from "node:child_process";
-import { pipeline as N } from "@xenova/transformers";
-import * as W from "node:fs";
-import C from "wavefile";
-import { createWorker as D } from "tesseract.js";
+import * as S from "node:os";
+import { objectMap as _, JSONAttemptParse as g, findByProp as k, JSONSanitize as b, clean as T, Http as q, consoleInterceptor as P, fn as M, ASet as $ } from "@ztimson/utils";
+import { Anthropic as E } from "@anthropic-ai/sdk";
+import { OpenAI as A } from "openai";
+import { Worker as x } from "worker_threads";
+import { fileURLToPath as v } from "url";
+import { join as O, dirname as U } from "path";
+import R from "node:path";
+import { canDiarization as L } from "./asr.mjs";
+import { createWorker as N } from "tesseract.js";
 import "./embedder.mjs";
-import * as I from "cheerio";
-import { $ as F, $Sync as H } from "@ztimson/node-utils";
+import * as C from "cheerio";
+import { $ as D, $Sync as W } from "@ztimson/node-utils";
 class j {
 }
-class J extends j {
+class z extends j {
   constructor(r, e, t) {
-    super(), this.ai = r, this.apiToken = e, this.model = t, this.client = new O({ apiKey: e });
+    super(), this.ai = r, this.apiToken = e, this.model = t, this.client = new E({ apiKey: e });
   }
   client;
   toStandard(r) {
     const e = Date.now(), t = [];
-    for (let i of r)
-      if (typeof i.content == "string")
-        t.push({ timestamp: e, ...i });
+    for (let o of r)
+      if (typeof o.content == "string")
+        t.push({ timestamp: e, ...o });
       else {
-        const s = i.content?.filter((n) => n.type == "text").map((n) => n.text).join(`
+        const s = o.content?.filter((n) => n.type == "text").map((n) => n.text).join(`
 
 `);
-        s && t.push({ timestamp: e, role: i.role, content: s }), i.content.forEach((n) => {
+        s && t.push({ timestamp: e, role: o.role, content: s }), o.content.forEach((n) => {
           if (n.type == "tool_use")
             t.push({ timestamp: e, role: "tool", id: n.id, name: n.name, args: n.input, content: void 0 });
           else if (n.type == "tool_result") {
-            const o = t.findLast((a) => a.id == n.tool_use_id);
-            o && (o[n.is_error ? "error" : "content"] = n.content);
+            const i = t.findLast((m) => m.id == n.tool_use_id);
+            i && (i[n.is_error ? "error" : "content"] = n.content);
           }
         });
       }
@@ -55,9 +53,9 @@ class J extends j {
   }
   ask(r, e = {}) {
     const t = new AbortController();
-    return Object.assign(new Promise(async (i) => {
+    return Object.assign(new Promise(async (o) => {
       let s = this.fromStandard([...e.history || [], { role: "user", content: r, timestamp: Date.now() }]);
-      const n = e.tools || this.ai.options.llm?.tools || [], o = {
+      const n = e.tools || this.ai.options.llm?.tools || [], i = {
         model: e.model || this.model,
         max_tokens: e.max_tokens || this.ai.options.llm?.max_tokens || 4096,
         system: e.system || this.ai.options.llm?.system || "",
@@ -67,7 +65,7 @@ class J extends j {
           description: d.description,
           input_schema: {
             type: "object",
-            properties: d.args ? x(d.args, (c, l) => ({ ...l, required: void 0 })) : {},
+            properties: d.args ? _(d.args, (c, l) => ({ ...l, required: void 0 })) : {},
             required: d.args ? Object.entries(d.args).filter((c) => c[1].required).map((c) => c[0]) : []
           },
           fn: void 0
@@ -75,58 +73,58 @@ class J extends j {
         messages: s,
         stream: !!e.stream
       };
-      let a, m = !0;
+      let m, a = !0;
       do {
-        if (a = await this.client.messages.create(o).catch((c) => {
+        if (m = await this.client.messages.create(i).catch((c) => {
           throw c.message += `
 
 Messages:
 ${JSON.stringify(s, null, 2)}`, c;
         }), e.stream) {
-          m ? m = !1 : e.stream({ text: `
+          a ? a = !1 : e.stream({ text: `
 
-` }), a.content = [];
-          for await (const c of a) {
+` }), m.content = [];
+          for await (const c of m) {
             if (t.signal.aborted) break;
             if (c.type === "content_block_start")
-              c.content_block.type === "text" ? a.content.push({ type: "text", text: "" }) : c.content_block.type === "tool_use" && a.content.push({ type: "tool_use", id: c.content_block.id, name: c.content_block.name, input: "" });
+              c.content_block.type === "text" ? m.content.push({ type: "text", text: "" }) : c.content_block.type === "tool_use" && m.content.push({ type: "tool_use", id: c.content_block.id, name: c.content_block.name, input: "" });
             else if (c.type === "content_block_delta")
               if (c.delta.type === "text_delta") {
                 const l = c.delta.text;
-                a.content.at(-1).text += l, e.stream({ text: l });
-              } else c.delta.type === "input_json_delta" && (a.content.at(-1).input += c.delta.partial_json);
+                m.content.at(-1).text += l, e.stream({ text: l });
+              } else c.delta.type === "input_json_delta" && (m.content.at(-1).input += c.delta.partial_json);
             else if (c.type === "content_block_stop") {
-              const l = a.content.at(-1);
-              l.input != null && (l.input = l.input ? b(l.input, {}) : {});
+              const l = m.content.at(-1);
+              l.input != null && (l.input = l.input ? g(l.input, {}) : {});
             } else if (c.type === "message_stop")
               break;
           }
         }
-        const d = a.content.filter((c) => c.type === "tool_use");
+        const d = m.content.filter((c) => c.type === "tool_use");
         if (d.length && !t.signal.aborted) {
-          s.push({ role: "assistant", content: a.content });
+          s.push({ role: "assistant", content: m.content });
           const c = await Promise.all(d.map(async (l) => {
-            const h = n.find(S("name", l.name));
-            if (e.stream && e.stream({ tool: l.name }), !h) return { tool_use_id: l.id, is_error: !0, content: "Tool not found" };
+            const p = n.find(k("name", l.name));
+            if (e.stream && e.stream({ tool: l.name }), !p) return { tool_use_id: l.id, is_error: !0, content: "Tool not found" };
             try {
-              const u = await h.fn(l.input, e?.stream, this.ai);
-              return { type: "tool_result", tool_use_id: l.id, content: w(u) };
+              const u = await p.fn(l.input, e?.stream, this.ai);
+              return { type: "tool_result", tool_use_id: l.id, content: b(u) };
             } catch (u) {
               return { type: "tool_result", tool_use_id: l.id, is_error: !0, content: u?.message || u?.toString() || "Unknown" };
             }
           }));
-          s.push({ role: "user", content: c }), o.messages = s;
+          s.push({ role: "user", content: c }), i.messages = s;
         }
-      } while (!t.signal.aborted && a.content.some((d) => d.type === "tool_use"));
-      s.push({ role: "assistant", content: a.content.filter((d) => d.type == "text").map((d) => d.text).join(`
+      } while (!t.signal.aborted && m.content.some((d) => d.type === "tool_use"));
+      s.push({ role: "assistant", content: m.content.filter((d) => d.type == "text").map((d) => d.text).join(`
 
-`) }), s = this.toStandard(s), e.stream && e.stream({ done: !0 }), e.history && e.history.splice(0, e.history.length, ...s), i(s.at(-1)?.content);
+`) }), s = this.toStandard(s), e.stream && e.stream({ done: !0 }), e.history && e.history.splice(0, e.history.length, ...s), o(s.at(-1)?.content);
     }), { abort: () => t.abort() });
   }
 }
-class _ extends j {
-  constructor(r, e, t, i) {
-    super(), this.ai = r, this.host = e, this.token = t, this.model = i, this.client = new v(q({
+class w extends j {
+  constructor(r, e, t, o) {
+    super(), this.ai = r, this.host = e, this.token = t, this.model = o, this.client = new A(T({
       baseURL: e,
       apiKey: t
     }));
@@ -136,17 +134,17 @@ class _ extends j {
     for (let e = 0; e < r.length; e++) {
       const t = r[e];
       if (t.role === "assistant" && t.tool_calls) {
-        const i = t.tool_calls.map((s) => ({
+        const o = t.tool_calls.map((s) => ({
           role: "tool",
           id: s.id,
           name: s.function.name,
-          args: b(s.function.arguments, {}),
+          args: g(s.function.arguments, {}),
           timestamp: t.timestamp
         }));
-        r.splice(e, 1, ...i), e += i.length - 1;
+        r.splice(e, 1, ...o), e += o.length - 1;
       } else if (t.role === "tool" && t.content) {
-        const i = r.find((s) => t.tool_call_id == s.id);
-        i && (t.content.includes('"error":') ? i.error = t.content : i.content = t.content), r.splice(e, 1), e--;
+        const o = r.find((s) => t.tool_call_id == s.id);
+        o && (t.content.includes('"error":') ? o.error = t.content : o.content = t.content), r.splice(e, 1), e--;
       }
       r[e]?.timestamp || (r[e].timestamp = Date.now());
     }
@@ -167,7 +165,7 @@ class _ extends j {
           content: t.error || t.content
         });
       else {
-        const { timestamp: i, ...s } = t;
+        const { timestamp: o, ...s } = t;
         e.push(s);
       }
       return e;
@@ -175,31 +173,31 @@ class _ extends j {
   }
   ask(r, e = {}) {
     const t = new AbortController();
-    return Object.assign(new Promise(async (i, s) => {
+    return Object.assign(new Promise(async (o, s) => {
       e.system && e.history?.[0]?.role != "system" && e.history?.splice(0, 0, { role: "system", content: e.system, timestamp: Date.now() });
       let n = this.fromStandard([...e.history || [], { role: "user", content: r, timestamp: Date.now() }]);
-      const o = e.tools || this.ai.options.llm?.tools || [], a = {
+      const i = e.tools || this.ai.options.llm?.tools || [], m = {
         model: e.model || this.model,
         messages: n,
         stream: !!e.stream,
         max_tokens: e.max_tokens || this.ai.options.llm?.max_tokens || 4096,
         temperature: e.temperature || this.ai.options.llm?.temperature || 0.7,
-        tools: o.map((c) => ({
+        tools: i.map((c) => ({
           type: "function",
           function: {
             name: c.name,
             description: c.description,
             parameters: {
               type: "object",
-              properties: c.args ? x(c.args, (l, h) => ({ ...h, required: void 0 })) : {},
+              properties: c.args ? _(c.args, (l, p) => ({ ...p, required: void 0 })) : {},
               required: c.args ? Object.entries(c.args).filter((l) => l[1].required).map((l) => l[0]) : []
             }
           }
         }))
       };
-      let m, d = !0;
+      let a, d = !0;
       do {
-        if (m = await this.client.chat.completions.create(a).catch((l) => {
+        if (a = await this.client.chat.completions.create(m).catch((l) => {
           throw l.message += `
 
 Messages:
@@ -207,44 +205,38 @@ ${JSON.stringify(n, null, 2)}`, l;
         }), e.stream) {
           d ? d = !1 : e.stream({ text: `
 
-` }), m.choices = [{ message: { content: "", tool_calls: [] } }];
-          for await (const l of m) {
+` }), a.choices = [{ message: { content: "", tool_calls: [] } }];
+          for await (const l of a) {
             if (t.signal.aborted) break;
-            l.choices[0].delta.content && (m.choices[0].message.content += l.choices[0].delta.content, e.stream({ text: l.choices[0].delta.content })), l.choices[0].delta.tool_calls && (m.choices[0].message.tool_calls = l.choices[0].delta.tool_calls);
+            l.choices[0].delta.content && (a.choices[0].message.content += l.choices[0].delta.content, e.stream({ text: l.choices[0].delta.content })), l.choices[0].delta.tool_calls && (a.choices[0].message.tool_calls = l.choices[0].delta.tool_calls);
           }
         }
-        const c = m.choices[0].message.tool_calls || [];
+        const c = a.choices[0].message.tool_calls || [];
         if (c.length && !t.signal.aborted) {
-          n.push(m.choices[0].message);
-          const l = await Promise.all(c.map(async (h) => {
-            const u = o?.find(S("name", h.function.name));
-            if (e.stream && e.stream({ tool: h.function.name }), !u) return { role: "tool", tool_call_id: h.id, content: '{"error": "Tool not found"}' };
+          n.push(a.choices[0].message);
+          const l = await Promise.all(c.map(async (p) => {
+            const u = i?.find(k("name", p.function.name));
+            if (e.stream && e.stream({ tool: p.function.name }), !u) return { role: "tool", tool_call_id: p.id, content: '{"error": "Tool not found"}' };
             try {
-              const f = b(h.function.arguments, {}), y = await u.fn(f, e.stream, this.ai);
-              return { role: "tool", tool_call_id: h.id, content: w(y) };
+              const f = g(p.function.arguments, {}), y = await u.fn(f, e.stream, this.ai);
+              return { role: "tool", tool_call_id: p.id, content: b(y) };
             } catch (f) {
-              return { role: "tool", tool_call_id: h.id, content: w({ error: f?.message || f?.toString() || "Unknown" }) };
+              return { role: "tool", tool_call_id: p.id, content: b({ error: f?.message || f?.toString() || "Unknown" }) };
             }
           }));
-          n.push(...l), a.messages = n;
+          n.push(...l), m.messages = n;
         }
-      } while (!t.signal.aborted && m.choices?.[0]?.message?.tool_calls?.length);
-      n.push({ role: "assistant", content: m.choices[0].message.content || "" }), n = this.toStandard(n), e.stream && e.stream({ done: !0 }), e.history && e.history.splice(0, e.history.length, ...n), i(n.at(-1)?.content);
+      } while (!t.signal.aborted && a.choices?.[0]?.message?.tool_calls?.length);
+      n.push({ role: "assistant", content: a.choices[0].message.content || "" }), n = this.toStandard(n), e.stream && e.stream({ done: !0 }), e.history && e.history.splice(0, e.history.length, ...n), o(n.at(-1)?.content);
     }), { abort: () => t.abort() });
   }
 }
-class B {
+class I {
   constructor(r) {
-    this.ai = r, this.embedWorker = new R(z(L(U(import.meta.url)), "embedder.js")), this.embedWorker.on("message", ({ id: e, embedding: t }) => {
-      const i = this.embedQueue.get(e);
-      i && (i.resolve(t), this.embedQueue.delete(e));
-    }), r.options.llm?.models && Object.entries(r.options.llm.models).forEach(([e, t]) => {
-      this.defaultModel || (this.defaultModel = e), t.proto == "anthropic" ? this.models[e] = new J(this.ai, t.token, e) : t.proto == "ollama" ? this.models[e] = new _(this.ai, t.host, "not-needed", e) : t.proto == "openai" && (this.models[e] = new _(this.ai, t.host || null, t.token, e));
+    this.ai = r, r.options.llm?.models && Object.entries(r.options.llm.models).forEach(([e, t]) => {
+      this.defaultModel || (this.defaultModel = e), t.proto == "anthropic" ? this.models[e] = new z(this.ai, t.token, e) : t.proto == "ollama" ? this.models[e] = new w(this.ai, t.host, "not-needed", e) : t.proto == "openai" && (this.models[e] = new w(this.ai, t.host || null, t.token, e));
     });
   }
-  embedWorker = null;
-  embedQueue = /* @__PURE__ */ new Map();
-  embedId = 0;
   models = {};
   defaultModel;
   /**
@@ -256,22 +248,22 @@ class B {
   ask(r, e = {}) {
     const t = e.model || this.defaultModel;
     if (!this.models[t]) throw new Error(`Model does not exist: ${t}`);
-    let i = () => {
+    let o = () => {
     };
     return Object.assign(new Promise(async (s) => {
       if (e.history || (e.history = []), e.memory) {
         e.system = (e.system || "") + `
 You have passive persistent memory never make any mention of your memory capabilities and what you can/cannot remember
 `;
-        const o = async (m, d, c = 50) => {
-          const [l, h] = await Promise.all([
+        const i = async (a, d, c = 50) => {
+          const [l, p] = await Promise.all([
             d ? this.embedding(d) : Promise.resolve(null),
-            m ? this.embedding(m) : Promise.resolve(null)
+            a ? this.embedding(a) : Promise.resolve(null)
           ]);
-          return (e.memory || []).map((u) => ({ ...u, score: l ? this.cosineSimilarity(u.embeddings[0], l[0].embedding) : 1 })).filter((u) => u.score >= 0.8).map((u) => ({ ...u, score: h ? this.cosineSimilarity(u.embeddings[1], h[0].embedding) : u.score })).filter((u) => u.score >= 0.2).toSorted((u, f) => u.score - f.score).slice(0, c);
-        }, a = await o(r);
-        a.length && e.history.push({ role: "assistant", content: `Things I remembered:
-` + a.map((m) => `${m.owner}: ${m.fact}`).join(`
+          return (e.memory || []).map((u) => ({ ...u, score: l ? this.cosineSimilarity(u.embeddings[0], l[0].embedding) : 1 })).filter((u) => u.score >= 0.8).map((u) => ({ ...u, score: p ? this.cosineSimilarity(u.embeddings[1], p[0].embedding) : u.score })).filter((u) => u.score >= 0.2).toSorted((u, f) => u.score - f.score).slice(0, c);
+        }, m = await i(r);
+        m.length && e.history.push({ role: "assistant", content: `Things I remembered:
+` + m.map((a) => `${a.owner}: ${a.fact}`).join(`
 `) }), e.tools = [...e.tools || [], {
           name: "read_memory",
           description: "Check your long-term memory for more information",
@@ -280,32 +272,32 @@ You have passive persistent memory never make any mention of your memory capabil
             query: { type: "string", description: "Search memory based on a query, can be used with or without subject argument" },
             limit: { type: "number", description: "Result limit, default 5" }
           },
-          fn: (m) => {
-            if (!m.subject && !m.query) throw new Error("Either a subject or query argument is required");
-            return o(m.query, m.subject, m.limit || 5);
+          fn: (a) => {
+            if (!a.subject && !a.query) throw new Error("Either a subject or query argument is required");
+            return i(a.query, a.subject, a.limit || 5);
           }
         }];
       }
       const n = await this.models[t].ask(r, e);
       if (e.memory) {
-        const o = e.history?.findIndex((a) => a.role == "assistant" && a.content.startsWith("Things I remembered:"));
-        o != null && o >= 0 && e.history?.splice(o, 1);
+        const i = e.history?.findIndex((m) => m.role == "assistant" && m.content.startsWith("Things I remembered:"));
+        i != null && i >= 0 && e.history?.splice(i, 1);
       }
       if (e.compress || e.memory) {
-        let o = null;
+        let i = null;
         if (e.compress)
-          o = await this.ai.language.compressHistory(e.history, e.compress.max, e.compress.min, e), e.history.splice(0, e.history.length, ...o.history);
+          i = await this.ai.language.compressHistory(e.history, e.compress.max, e.compress.min, e), e.history.splice(0, e.history.length, ...i.history);
         else {
-          const a = e.history?.findLastIndex((m) => m.role == "user") ?? -1;
-          o = await this.ai.language.compressHistory(a != -1 ? e.history.slice(a) : e.history, 0, 0, e);
+          const m = e.history?.findLastIndex((a) => a.role == "user") ?? -1;
+          i = await this.ai.language.compressHistory(m != -1 ? e.history.slice(m) : e.history, 0, 0, e);
         }
         if (e.memory) {
-          const a = e.memory.filter((m) => !o.memory.some((d) => this.cosineSimilarity(m.embeddings[1], d.embeddings[1]) > 0.8)).concat(o.memory);
-          e.memory.splice(0, e.memory.length, ...a);
+          const m = e.memory.filter((a) => !i.memory.some((d) => this.cosineSimilarity(a.embeddings[1], d.embeddings[1]) > 0.8)).concat(i.memory);
+          e.memory.splice(0, e.memory.length, ...m);
         }
       }
       return s(n);
-    }), { abort: i });
+    }), { abort: o });
   }
   /**
    * Compress chat history to reduce context size
@@ -315,22 +307,22 @@ You have passive persistent memory never make any mention of your memory capabil
    * @param {LLMRequest} options LLM options
    * @returns {Promise<LLMMessage[]>} New chat history will summary at index 0
    */
-  async compressHistory(r, e, t, i) {
+  async compressHistory(r, e, t, o) {
     if (this.estimateTokens(r) < e) return { history: r, memory: [] };
     let s = 0, n = 0;
     for (let u of r.toReversed())
       if (n += this.estimateTokens(u.content), n < t) s++;
       else break;
     if (r.length <= s) return { history: r, memory: [] };
-    const o = r[0].role == "system" ? r[0] : null, a = s == 0 ? [] : r.slice(-s), m = (s == 0 ? r : r.slice(0, -s)).filter((u) => u.role === "assistant" || u.role === "user"), d = await this.json(`Create the smallest summary possible, no more than 500 tokens. Create a list of NEW facts (split by subject [pro]noun and fact) about what you learned from this conversation that you didn't already know or get from a tool call or system prompt. Focus only on new information about people, topics, or facts. Avoid generating facts about the AI. Match this format: {summary: string, facts: [[subject, fact]]}
+    const i = r[0].role == "system" ? r[0] : null, m = s == 0 ? [] : r.slice(-s), a = (s == 0 ? r : r.slice(0, -s)).filter((u) => u.role === "assistant" || u.role === "user"), d = await this.json(`Create the smallest summary possible, no more than 500 tokens. Create a list of NEW facts (split by subject [pro]noun and fact) about what you learned from this conversation that you didn't already know or get from a tool call or system prompt. Focus only on new information about people, topics, or facts. Avoid generating facts about the AI. Match this format: {summary: string, facts: [[subject, fact]]}
 
-${m.map((u) => `${u.role}: ${u.content}`).join(`
+${a.map((u) => `${u.role}: ${u.content}`).join(`
 
-`)}`, { model: i?.model, temperature: i?.temperature || 0.3 }), c = /* @__PURE__ */ new Date(), l = await Promise.all((d?.facts || [])?.map(async ([u, f]) => {
+`)}`, { model: o?.model, temperature: o?.temperature || 0.3 }), c = /* @__PURE__ */ new Date(), l = await Promise.all((d?.facts || [])?.map(async ([u, f]) => {
       const y = await Promise.all([this.embedding(u), this.embedding(`${u}: ${f}`)]);
       return { owner: u, fact: f, embeddings: [y[0][0].embedding, y[1][0].embedding], timestamp: c };
-    })), h = [{ role: "assistant", content: `Conversation Summary: ${d?.summary}`, timestamp: Date.now() }, ...a];
-    return o && h.splice(0, 0, o), { history: h, memory: l };
+    })), p = [{ role: "assistant", content: `Conversation Summary: ${d?.summary}`, timestamp: Date.now() }, ...m];
+    return i && p.splice(0, 0, i), { history: p, memory: l };
   }
   /**
    * Compare the difference between embeddings (calculates the angle between two vectors)
@@ -340,10 +332,10 @@ ${m.map((u) => `${u.role}: ${u.content}`).join(`
    */
   cosineSimilarity(r, e) {
     if (r.length !== e.length) throw new Error("Vectors must be same length");
-    let t = 0, i = 0, s = 0;
-    for (let o = 0; o < r.length; o++)
-      t += r[o] * e[o], i += r[o] * r[o], s += e[o] * e[o];
-    const n = Math.sqrt(i) * Math.sqrt(s);
+    let t = 0, o = 0, s = 0;
+    for (let i = 0; i < r.length; i++)
+      t += r[i] * e[i], o += r[i] * r[i], s += e[i] * e[i];
+    const n = Math.sqrt(o) * Math.sqrt(s);
     return n === 0 ? 0 : t / n;
   }
   /**
@@ -354,25 +346,25 @@ ${m.map((u) => `${u.role}: ${u.content}`).join(`
    * @returns {string[]} Chunked strings
    */
   chunk(r, e = 500, t = 50) {
-    const i = (a, m = "") => a ? Object.entries(a).flatMap(([d, c]) => {
-      const l = m ? `${m}${isNaN(+d) ? `.${d}` : `[${d}]`}` : d;
-      return typeof c == "object" && !Array.isArray(c) ? i(c, l) : `${l}: ${Array.isArray(c) ? c.join(", ") : c}`;
-    }) : [], n = (typeof r == "object" ? i(r) : r.split(`
-`)).flatMap((a) => [...a.split(/\s+/).filter(Boolean), `
-`]), o = [];
-    for (let a = 0; a < n.length; ) {
-      let m = "", d = a;
+    const o = (m, a = "") => m ? Object.entries(m).flatMap(([d, c]) => {
+      const l = a ? `${a}${isNaN(+d) ? `.${d}` : `[${d}]`}` : d;
+      return typeof c == "object" && !Array.isArray(c) ? o(c, l) : `${l}: ${Array.isArray(c) ? c.join(", ") : c}`;
+    }) : [], n = (typeof r == "object" ? o(r) : r.split(`
+`)).flatMap((m) => [...m.split(/\s+/).filter(Boolean), `
+`]), i = [];
+    for (let m = 0; m < n.length; ) {
+      let a = "", d = m;
       for (; d < n.length; ) {
-        const l = m + (m ? " " : "") + n[d];
+        const l = a + (a ? " " : "") + n[d];
         if (this.estimateTokens(l.replace(/\s*\n\s*/g, `
-`)) > e && m) break;
-        m = l, d++;
+`)) > e && a) break;
+        a = l, d++;
       }
-      const c = m.replace(/\s*\n\s*/g, `
+      const c = a.replace(/\s*\n\s*/g, `
 `).trim();
-      c && o.push(c), a = Math.max(d - t, d === a ? a + 1 : d);
+      c && i.push(c), m = Math.max(d - t, d === m ? m + 1 : d);
     }
-    return o;
+    return i;
   }
   /**
    * Create a vector representation of a string
@@ -382,18 +374,19 @@ ${m.map((u) => `${u.role}: ${u.content}`).join(`
    * @returns {Promise<Awaited<{index: number, embedding: number[], text: string, tokens: number}>[]>} Chunked embeddings
    */
   embedding(r, e = 500, t = 50) {
-    const i = (n) => new Promise((o, a) => {
-      const m = this.embedId++;
-      this.embedQueue.set(m, { resolve: o, reject: a }), this.embedWorker?.postMessage({
-        id: m,
-        text: n,
-        model: this.ai.options?.embedder || "bge-small-en-v1.5",
-        path: this.ai.options.path
-      });
+    const o = (n) => new Promise((i, m) => {
+      const a = new x(O(U(v(import.meta.url)), "embedder.js")), d = ({ embedding: l }) => {
+        a.terminate(), i(l);
+      }, c = (l) => {
+        a.terminate(), m(l);
+      };
+      a.on("message", d), a.on("error", c), a.on("exit", (l) => {
+        l !== 0 && m(new Error(`Worker exited with code ${l}`));
+      }), a.postMessage({ text: n, model: this.ai.options?.embedder || "bge-small-en-v1.5", path: this.ai.options.path });
     }), s = this.chunk(r, e, t);
-    return Promise.all(s.map(async (n, o) => ({
-      index: o,
-      embedding: await i(n),
+    return Promise.all(s.map(async (n, i) => ({
+      index: i,
+      embedding: await o(n),
       text: n,
       tokens: this.estimateTokens(n)
     })));
@@ -415,8 +408,8 @@ ${m.map((u) => `${u.role}: ${u.content}`).join(`
    */
   fuzzyMatch(r, ...e) {
     if (e.length < 2) throw new Error("Requires at least 2 strings to compare");
-    const t = (n, o = 10) => n.toLowerCase().split("").map((a, m) => a.charCodeAt(0) * (m + 1) % o / o).slice(0, o), i = t(r), s = e.map((n) => t(n)).map((n) => this.cosineSimilarity(i, n));
-    return { avg: s.reduce((n, o) => n + o, 0) / s.length, max: Math.max(...s), similarities: s };
+    const t = (n, i = 10) => n.toLowerCase().split("").map((m, a) => m.charCodeAt(0) * (a + 1) % i / i).slice(0, i), o = t(r), s = e.map((n) => t(n)).map((n) => this.cosineSimilarity(o, n));
+    return { avg: s.reduce((n, i) => n + i, 0) / s.length, max: Math.max(...s), similarities: s };
   }
   /**
    * Ask a question with JSON response
@@ -427,8 +420,8 @@ ${m.map((u) => `${u.role}: ${u.content}`).join(`
   async json(r, e) {
     let t = await this.ask(r, { system: "Respond using a JSON blob matching any provided examples", ...e });
     if (!t) return {};
-    const i = /```(?:.+)?\s*([\s\S]*?)```/.exec(t), s = i ? i[1].trim() : t;
-    return b(s, {});
+    const o = /```(?:.+)?\s*([\s\S]*?)```/.exec(t), s = o ? o[1].trim() : t;
+    return g(s, {});
   }
   /**
    * Create a summary of some text
@@ -441,101 +434,30 @@ ${m.map((u) => `${u.role}: ${u.content}`).join(`
     return this.ask(r, { system: `Generate a brief summary <= ${e} tokens. Output nothing else`, temperature: 0.3, ...t });
   }
 }
-class G {
+class J {
   constructor(r) {
     this.ai = r;
   }
-  whisperPipeline;
-  combineSpeakerTranscript(r, e) {
-    const t = /* @__PURE__ */ new Map();
-    let i = 0;
-    e.forEach((a) => {
-      t.has(a.speaker) || t.set(a.speaker, ++i);
-    });
-    const s = [];
-    let n = -1, o = "";
-    return r.forEach((a) => {
-      const m = a.timestamp[0], d = e.find((l) => m >= l.start && m <= l.end), c = d ? t.get(d.speaker) : 1;
-      c !== n ? (o && s.push(`[speaker ${n}]: ${o.trim()}`), n = c, o = a.text) : o += a.text;
-    }), o && s.push(`[speaker ${n}]: ${o.trim()}`), s.join(`
-`);
-  }
-  async canDiarization() {
-    return new Promise((r) => {
-      const e = k("python3", ["-c", "import pyannote.audio"]);
-      e.on("close", (t) => r(t === 0)), e.on("error", () => r(!1));
-    });
-  }
-  async runDiarization(r) {
-    if (!await this.canDiarization()) throw new Error("Pyannote is not installed: pip install pyannote.audio");
-    const e = `
-import sys
-import json
-from pyannote.audio import Pipeline
-
-os.environ['TORCH_HOME'] = "${this.ai.options.path}"
-pipeline = Pipeline.from_pretrained("pyannote/speaker-diarization-3.1")
-diarization = pipeline(sys.argv[1])
-
-segments = []
-for turn, _, speaker in diarization.itertracks(yield_label=True):
-    segments.append({
-        "start": turn.start,
-        "end": turn.end,
-        "speaker": speaker
-    })
-
-print(json.dumps(segments))
-`;
-    return new Promise((t, i) => {
-      let s = "";
-      const n = k("python3", ["-c", e, r]);
-      n.stdout.on("data", (o) => s += o.toString()), n.stderr.on("data", (o) => console.error(o.toString())), n.on("close", (o) => {
-        if (o === 0)
-          try {
-            t(JSON.parse(s));
-          } catch {
-            i(new Error("Failed to parse diarization output"));
-          }
-        else
-          i(new Error(`Python process exited with code ${o}`));
-      }), n.on("error", i);
-    });
-  }
   asr(r, e = {}) {
-    const { model: t = this.ai.options.asr || "whisper-base", speaker: i = !1 } = e;
+    const { model: t = this.ai.options.asr || "whisper-base", speaker: o = !1 } = e;
     let s = !1;
     const n = () => {
       s = !0;
-    }, o = new Promise(async (a, m) => {
-      try {
-        if (s || (this.whisperPipeline || (this.whisperPipeline = await N("automatic-speech-recognition", `Xenova/${t}`, { cache_dir: this.ai.options.path, quantized: !0 })), s)) return a(null);
-        const d = new C.WaveFile(W.readFileSync(r));
-        d.toBitDepth("32f"), d.toSampleRate(16e3);
-        const c = d.getSamples();
-        let l;
-        if (Array.isArray(c)) {
-          const y = c[0], T = c[1];
-          l = new Float32Array(y.length);
-          for (let g = 0; g < y.length; g++) l[g] = (y[g] + T[g]) / 2;
-        } else
-          l = c;
-        if (s) return a(null);
-        const h = await this.whisperPipeline(l, { return_timestamps: i ? "word" : !1 });
-        if (!i) return a(h.text?.trim() || null);
-        if (s) return a(null);
-        const u = await this.runDiarization(r);
-        if (s) return a(null);
-        const f = this.combineSpeakerTranscript(h.chunks || [], u);
-        a(f);
-      } catch (d) {
-        m(d);
-      }
+    }, i = new Promise((m, a) => {
+      const d = new x(R.join(import.meta.dirname, "asr.js")), c = ({ text: p, warning: u, error: f }) => {
+        d.terminate(), !s && (f ? a(new Error(f)) : (u && console.warn(u), m(p)));
+      }, l = (p) => {
+        d.terminate(), s || a(p);
+      };
+      d.on("message", c), d.on("error", l), d.on("exit", (p) => {
+        p !== 0 && !s && a(new Error(`Worker exited with code ${p}`));
+      }), d.postMessage({ file: r, model: t, speaker: o, modelDir: this.ai.options.path });
     });
-    return Object.assign(o, { abort: n });
+    return Object.assign(i, { abort: n });
   }
+  canDiarization = L;
 }
-class Q {
+class H {
   constructor(r) {
     this.ai = r;
   }
@@ -546,17 +468,17 @@ class Q {
    */
   ocr(r) {
     let e;
-    const t = new Promise(async (i) => {
-      e = await D(this.ai.options.ocr || "eng", 2, { cachePath: this.ai.options.path });
+    const t = new Promise(async (o) => {
+      e = await N(this.ai.options.ocr || "eng", 2, { cachePath: this.ai.options.path });
       const { data: s } = await e.recognize(r);
-      await e.terminate(), i(s.text.trim() || null);
+      await e.terminate(), o(s.text.trim() || null);
     });
     return Object.assign(t, { abort: () => e?.terminate() });
   }
 }
-class me {
+class oe {
   constructor(r) {
-    this.options = r, r.path || (r.path = P.tmpdir()), process.env.TRANSFORMERS_CACHE = r.path, this.audio = new G(this), this.language = new B(this), this.vision = new Q(this);
+    this.options = r, r.path || (r.path = S.tmpdir()), process.env.TRANSFORMERS_CACHE = r.path, this.audio = new J(this), this.language = new I(this), this.vision = new H(this);
   }
   /** Audio processing AI */
   audio;
@@ -565,38 +487,38 @@ class me {
   /** Vision processing AI */
   vision;
 }
-const K = {
+const F = {
   name: "cli",
   description: "Use the command line interface, returns any output",
   args: { command: { type: "string", description: "Command to run", required: !0 } },
-  fn: (p) => F`${p.command}`
-}, de = {
+  fn: (h) => D`${h.command}`
+}, ie = {
   name: "get_datetime",
   description: "Get current UTC date / time",
   args: {},
   fn: async () => (/* @__PURE__ */ new Date()).toUTCString()
-}, ue = {
+}, ae = {
   name: "exec",
   description: "Run code/scripts",
   args: {
     language: { type: "string", description: "Execution language", enum: ["cli", "node", "python"], required: !0 },
     code: { type: "string", description: "Code to execute", required: !0 }
   },
-  fn: async (p, r, e) => {
+  fn: async (h, r, e) => {
     try {
-      switch (p.type) {
+      switch (h.type) {
         case "bash":
-          return await K.fn({ command: p.code }, r, e);
+          return await F.fn({ command: h.code }, r, e);
         case "node":
-          return await V.fn({ code: p.code }, r, e);
+          return await G.fn({ code: h.code }, r, e);
         case "python":
-          return await X.fn({ code: p.code }, r, e);
+          return await B.fn({ code: h.code }, r, e);
       }
     } catch (t) {
       return { error: t?.message || t.toString() };
     }
   }
-}, pe = {
+}, ce = {
   name: "fetch",
   description: "Make HTTP request to URL",
   args: {
@@ -605,85 +527,86 @@ const K = {
     headers: { type: "object", description: "HTTP headers to send", default: {} },
     body: { type: "object", description: "HTTP body to send" }
   },
-  fn: (p) => new $({ url: p.url, headers: p.headers }).request({ method: p.method || "GET", body: p.body })
-}, V = {
+  fn: (h) => new q({ url: h.url, headers: h.headers }).request({ method: h.method || "GET", body: h.body })
+}, G = {
   name: "exec_javascript",
   description: "Execute commonjs javascript",
   args: {
     code: { type: "string", description: "CommonJS javascript", required: !0 }
   },
-  fn: async (p) => {
-    const r = M(null), e = await E({ console: r }, p.code, !0).catch((t) => r.output.error.push(t));
+  fn: async (h) => {
+    const r = P(null), e = await M({ console: r }, h.code, !0).catch((t) => r.output.error.push(t));
     return { ...r.output, return: e, stdout: void 0, stderr: void 0 };
   }
-}, X = {
+}, B = {
   name: "exec_javascript",
   description: "Execute commonjs javascript",
   args: {
     code: { type: "string", description: "CommonJS javascript", required: !0 }
   },
-  fn: async (p) => ({ result: H`python -c "${p.code}"` })
-}, he = {
+  fn: async (h) => ({ result: W`python -c "${h.code}"` })
+}, le = {
   name: "read_webpage",
   description: "Extract clean, structured content from a webpage. Use after web_search to read specific URLs",
   args: {
     url: { type: "string", description: "URL to extract content from", required: !0 },
     focus: { type: "string", description: 'Optional: What aspect to focus on (e.g., "pricing", "features", "contact info")' }
   },
-  fn: async (p) => {
-    const r = await fetch(p.url, { headers: { "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64)" } }).then((n) => n.text()).catch((n) => {
+  fn: async (h) => {
+    const r = await fetch(h.url, { headers: { "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64)" } }).then((n) => n.text()).catch((n) => {
       throw new Error(`Failed to fetch: ${n.message}`);
-    }), e = I.load(r);
+    }), e = C.load(r);
     e('script, style, nav, footer, header, aside, iframe, noscript, [role="navigation"], [role="banner"], .ad, .ads, .cookie, .popup').remove();
     const t = {
       title: e('meta[property="og:title"]').attr("content") || e("title").text() || "",
       description: e('meta[name="description"]').attr("content") || e('meta[property="og:description"]').attr("content") || ""
     };
-    let i = "";
+    let o = "";
     const s = ["article", "main", '[role="main"]', ".content", ".post", ".entry", "body"];
     for (const n of s) {
-      const o = e(n).first();
-      if (o.length && o.text().trim().length > 200) {
-        i = o.text();
+      const i = e(n).first();
+      if (i.length && i.text().trim().length > 200) {
+        o = i.text();
         break;
       }
     }
-    return i || (i = e("body").text()), i = i.replace(/\s+/g, " ").trim().slice(0, 8e3), { url: p.url, title: t.title.trim(), description: t.description.trim(), content: i, focus: p.focus };
+    return o || (o = e("body").text()), o = o.replace(/\s+/g, " ").trim().slice(0, 8e3), { url: h.url, title: t.title.trim(), description: t.description.trim(), content: o, focus: h.focus };
   }
-}, fe = {
+}, me = {
   name: "web_search",
   description: "Use duckduckgo (anonymous) to find find relevant online resources. Returns a list of URLs that works great with the `read_webpage` tool",
   args: {
     query: { type: "string", description: "Search string", required: !0 },
     length: { type: "string", description: "Number of results to return", default: 5 }
   },
-  fn: async (p) => {
-    const r = await fetch(`https://html.duckduckgo.com/html/?q=${encodeURIComponent(p.query)}`, {
+  fn: async (h) => {
+    const r = await fetch(`https://html.duckduckgo.com/html/?q=${encodeURIComponent(h.query)}`, {
       headers: { "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64)", "Accept-Language": "en-US,en;q=0.9" }
     }).then((s) => s.text());
     let e, t = /<a .*?href="(.+?)".+?<\/a>/g;
-    const i = new A();
+    const o = new $();
     for (; (e = t.exec(r)) !== null; ) {
       let s = /uddg=(.+)&amp?/.exec(decodeURIComponent(e[1]))?.[1];
-      if (s && (s = decodeURIComponent(s)), s && i.add(s), i.size >= (p.length || 5)) break;
+      if (s && (s = decodeURIComponent(s)), s && o.add(s), o.size >= (h.length || 5)) break;
     }
-    return i;
+    return o;
   }
 };
 export {
-  me as Ai,
-  J as Anthropic,
-  G as Audio,
-  K as CliTool,
-  de as DateTimeTool,
-  ue as ExecTool,
-  pe as FetchTool,
-  V as JSTool,
+  oe as Ai,
+  z as Anthropic,
+  J as Audio,
+  F as CliTool,
+  ie as DateTimeTool,
+  ae as ExecTool,
+  ce as FetchTool,
+  G as JSTool,
   j as LLMProvider,
-  _ as OpenAi,
-  X as PythonTool,
-  he as ReadWebpageTool,
-  Q as Vision,
-  fe as WebSearchTool
+  w as OpenAi,
+  B as PythonTool,
+  le as ReadWebpageTool,
+  H as Vision,
+  me as WebSearchTool,
+  L as canDiarization
 };
 //# sourceMappingURL=index.mjs.map