@cogitator-ai/channels 0.2.1

package/dist/inbound-debounce.js

@@ -0,0 +1,61 @@
+export class InboundDebouncer {
+    config;
+    onFlush;
+    buffers = new Map();
+    defaultDelay;
+    constructor(config, onFlush) {
+        this.config = config;
+        this.onFlush = onFlush;
+        this.defaultDelay = config.delayMs ?? 1500;
+    }
+    enqueue(msg) {
+        const key = `${msg.channelType}:${msg.channelId}:${msg.userId}`;
+        const delay = this.config.byChannel?.[msg.channelType] ?? this.defaultDelay;
+        const existing = this.buffers.get(key);
+        if (existing) {
+            clearTimeout(existing.timer);
+            existing.messages.push(msg);
+            existing.timer = setTimeout(() => void this.flush(key), delay);
+        }
+        else {
+            const timer = setTimeout(() => void this.flush(key), delay);
+            this.buffers.set(key, { messages: [msg], timer });
+        }
+    }
+    async flushAll() {
+        const keys = [...this.buffers.keys()];
+        await Promise.all(keys.map((key) => this.flush(key)));
+    }
+    dispose() {
+        for (const entry of this.buffers.values()) {
+            clearTimeout(entry.timer);
+        }
+        this.buffers.clear();
+    }
+    async flush(key) {
+        const entry = this.buffers.get(key);
+        if (!entry)
+            return;
+        this.buffers.delete(key);
+        const messages = entry.messages;
+        if (messages.length === 0)
+            return;
+        const first = messages[0];
+        const last = messages[messages.length - 1];
+        const merged = {
+            id: first.id,
+            channelType: first.channelType,
+            channelId: first.channelId,
+            userId: first.userId,
+            userName: first.userName,
+            groupId: first.groupId,
+            text: messages.map((m) => m.text).join('\n'),
+            raw: last.raw,
+        };
+        const allAttachments = messages.flatMap((m) => m.attachments ?? []);
+        if (allAttachments.length > 0)
+            merged.attachments = allAttachments;
+        await this.onFlush(merged);
+    }
+}
+//# sourceMappingURL=inbound-debounce.js.map

package/dist/inbound-debounce.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"inbound-debounce.js","sourceRoot":"","sources":["../src/inbound-debounce.ts"],"names":[],"mappings":"AAOA,MAAM,OAAO,gBAAgB;IAKR;IACA;IALX,OAAO,GAAG,IAAI,GAAG,EAAuB,CAAC;IAChC,YAAY,CAAS;IAEtC,YACmB,MAAsB,EACtB,OAAkD;QADlD,WAAM,GAAN,MAAM,CAAgB;QACtB,YAAO,GAAP,OAAO,CAA2C;QAEnE,IAAI,CAAC,YAAY,GAAG,MAAM,CAAC,OAAO,IAAI,IAAI,CAAC;IAC7C,CAAC;IAED,OAAO,CAAC,GAAmB;QACzB,MAAM,GAAG,GAAG,GAAG,GAAG,CAAC,WAAW,IAAI,GAAG,CAAC,SAAS,IAAI,GAAG,CAAC,MAAM,EAAE,CAAC;QAChE,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,SAAS,EAAE,CAAC,GAAG,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,YAAY,CAAC;QAE5E,MAAM,QAAQ,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QACvC,IAAI,QAAQ,EAAE,CAAC;YACb,YAAY,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;YAC7B,QAAQ,CAAC,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YAC5B,QAAQ,CAAC,KAAK,GAAG,UAAU,CAAC,GAAG,EAAE,CAAC,KAAK,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,KAAK,CAAC,CAAC;QACjE,CAAC;aAAM,CAAC;YACN,MAAM,KAAK,GAAG,UAAU,CAAC,GAAG,EAAE,CAAC,KAAK,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,KAAK,CAAC,CAAC;YAC5D,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,GAAG,EAAE,EAAE,QAAQ,EAAE,CAAC,GAAG,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC;QACpD,CAAC;IACH,CAAC;IAED,KAAK,CAAC,QAAQ;QACZ,MAAM,IAAI,GAAG,CAAC,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC;QACtC,MAAM,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;IACxD,CAAC;IAED,OAAO;QACL,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC;YAC1C,YAAY,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;QAC5B,CAAC;QACD,IAAI,CAAC,OAAO,CAAC,KAAK,EAAE,CAAC;IACvB,CAAC;IAEO,KAAK,CAAC,KAAK,CAAC,GAAW;QAC7B,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QACpC,IAAI,CAAC,KAAK;YAAE,OAAO;QACnB,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAEzB,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,CAAC;QAChC,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO;QAElC,MAAM,KAAK,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;QAC1B,MAAM,IAAI,GAAG,QAAQ,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QAE3C,MAAM,MAAM,GAAmB;YAC7B,EAAE,EAAE,KAAK,CAAC,EAAE;YACZ,WAAW,EAAE,KAAK,CAAC,WAAW;YAC9B,SAAS,EAAE,KAAK,CAAC,SAAS;YAC1B,MAAM,EAAE,KAAK,CAAC,MAAM;YACpB,QAAQ,EAAE,KAAK,CAAC,QAAQ;YACxB,OAAO,EAAE,KAAK,CAAC,OAAO;YACtB,IAAI,EAAE,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC;YAC5C,GAAG,EAAE,IAAI,CAAC,GAAG;SACd,CAAC;QAEF,MAAM,cAAc,GAAG,QAAQ,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,WAAW,IAAI,EAAE,CAAC,CAAC;QACpE,IAAI,cAAc,CAAC,MAAM,GAAG,CAAC;YAAE,MAAM,CAAC,WAAW,GAAG,cAAc,CAAC;QAEnE,MAAM,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;IAC7B,CAAC;CACF"}

package/dist/index.d.ts

@@ -0,0 +1,54 @@
+export { Gateway } from './gateway';
+export type { GatewayFullConfig } from './gateway';
+export { StreamBuffer } from './stream-buffer';
+export { HeartbeatScheduler } from './heartbeat';
+export type { HeartbeatConfig } from './heartbeat';
+export { StatusReactionTracker } from './status-reactions';
+export { InboundDebouncer } from './inbound-debounce';
+export { MessageQueue } from './message-queue';
+export { formatEnvelope, formatElapsed } from './envelope';
+export { TerminalChannel, terminalChannel } from './channels/terminal';
+export type { TerminalConfig } from './channels/terminal';
+export { WebChatChannel, webchatChannel } from './channels/webchat';
+export type { WebChatConfig } from './channels/webchat';
+export { TelegramChannel, telegramChannel } from './channels/telegram';
+export type { TelegramConfig } from './channels/telegram';
+export { DiscordChannel, discordChannel } from './channels/discord';
+export type { DiscordConfig } from './channels/discord';
+export { SlackChannel, slackChannel } from './channels/slack';
+export type { SlackConfig } from './channels/slack';
+export { WhatsAppChannel, whatsappChannel } from './channels/whatsapp';
+export type { WhatsAppChannelConfig } from './channels/whatsapp';
+export { PairingMiddleware, pairing } from './middleware/pairing';
+export type { PairingConfig } from './middleware/pairing';
+export { DmPolicyMiddleware, dmPolicy } from './middleware/dm-policy';
+export type { DmPolicyConfig, DmPolicyMode } from './middleware/dm-policy';
+export { RateLimitMiddleware, rateLimit } from './middleware/rate-limit';
+export type { RateLimitConfig } from './middleware/rate-limit';
+export { OwnerCommandsMiddleware, ownerCommands } from './middleware/owner-commands';
+export type { OwnerCommandsConfig, CommandLevel } from './middleware/owner-commands';
+export { createHookRegistry } from './hooks';
+export type { HookName, HookHandler, HookRegistry } from './hooks';
+export { AutoExtractMiddleware, autoExtract } from './middleware/auto-extract';
+export type { AutoExtractConfig, EntityExtractor, ExtractedEntities, } from './middleware/auto-extract';
+export { generateCapabilitiesDoc } from './capabilities';
+export type { CapabilitiesInput } from './capabilities';
+export { RuntimeBuilder } from './runtime-builder';
+export type { AssistantConfig, BuiltRuntime, RuntimeBuilderOpts } from './runtime-builder';
+export { SimpleTimerStore } from './simple-timer-store';
+export { MediaProcessor } from './media/media-processor';
+export type { MediaProcessResult, SttProvider } from './media/media-processor';
+export { LocalWhisper } from './media/whisper-local';
+export { createWhisperDownloadTool } from './media/whisper-tool';
+export { GroqSttProvider, OpenAISttProvider } from './media/whisper-api';
+export type { GroqSttConfig, OpenAISttConfig } from './media/whisper-api';
+export { DeepgramSttProvider } from './media/deepgram-stt';
+export type { DeepgramSttConfig } from './media/deepgram-stt';
+export { createSelfConfigTools } from './tools/self-config';
+export { AssistantConfigSchema, type AssistantConfigInput, type AssistantConfigOutput, } from './assistant-config';
+export { adaptMarkdown, chunkMessage, getPlatformLimit } from './formatters/markdown';
+export { chunkDiscordText } from './formatters/discord-chunker';
+export type { ChunkDiscordTextOpts } from './formatters/discord-chunker';
+export { markdownToWhatsApp } from './formatters/whatsapp-markdown';
+export type { Channel, ChannelMessage, ChannelUser, ChannelType, Attachment, SendOptions, StreamConfig, GatewayConfig, GatewayStats, GatewayMiddleware, MiddlewareContext, StatusPhase, StatusReactionConfig, DebounceConfig, EnvelopeConfig, QueueMode, } from '@cogitator-ai/types';
+//# sourceMappingURL=index.d.ts.map

package/dist/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AACpC,YAAY,EAAE,iBAAiB,EAAE,MAAM,WAAW,CAAC;AAEnD,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAE/C,OAAO,EAAE,kBAAkB,EAAE,MAAM,aAAa,CAAC;AACjD,YAAY,EAAE,eAAe,EAAE,MAAM,aAAa,CAAC;AAEnD,OAAO,EAAE,qBAAqB,EAAE,MAAM,oBAAoB,CAAC;AAC3D,OAAO,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AACtD,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC/C,OAAO,EAAE,cAAc,EAAE,aAAa,EAAE,MAAM,YAAY,CAAC;AAE3D,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AACvE,YAAY,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAE1D,OAAO,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,oBAAoB,CAAC;AACpE,YAAY,EAAE,aAAa,EAAE,MAAM,oBAAoB,CAAC;AAExD,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AACvE,YAAY,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAE1D,OAAO,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,oBAAoB,CAAC;AACpE,YAAY,EAAE,aAAa,EAAE,MAAM,oBAAoB,CAAC;AAExD,OAAO,EAAE,YAAY,EAAE,YAAY,EAAE,MAAM,kBAAkB,CAAC;AAC9D,YAAY,EAAE,WAAW,EAAE,MAAM,kBAAkB,CAAC;AAEpD,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AACvE,YAAY,EAAE,qBAAqB,EAAE,MAAM,qBAAqB,CAAC;AAEjE,OAAO,EAAE,iBAAiB,EAAE,OAAO,EAAE,MAAM,sBAAsB,CAAC;AAClE,YAAY,EAAE,aAAa,EAAE,MAAM,sBAAsB,CAAC;AAE1D,OAAO,EAAE,kBAAkB,EAAE,QAAQ,EAAE,MAAM,wBAAwB,CAAC;AACtE,YAAY,EAAE,cAAc,EAAE,YAAY,EAAE,MAAM,wBAAwB,CAAC;AAE3E,OAAO,EAAE,mBAAmB,EAAE,SAAS,EAAE,MAAM,yBAAyB,CAAC;AACzE,YAAY,EAAE,eAAe,EAAE,MAAM,yBAAyB,CAAC;AAE/D,OAAO,EAAE,uBAAuB,EAAE,aAAa,EAAE,MAAM,6BAA6B,CAAC;AACrF,YAAY,EAAE,mBAAmB,EAAE,YAAY,EAAE,MAAM,6BAA6B,CAAC;AAErF,OAAO,EAAE,kBAAkB,EAAE,MAAM,SAAS,CAAC;AAC7C,YAAY,EAAE,QAAQ,EAAE,WAAW,EAAE,YAAY,EAAE,MAAM,SAAS,CAAC;AAEnE,OAAO,EAAE,qBAAqB,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAC;AAC/E,YAAY,EACV,iBAAiB,EACjB,eAAe,EACf,iBAAiB,GAClB,MAAM,2BAA2B,CAAC;AAEnC,OAAO,EAAE,uBAAuB,EAAE,MAAM,gBAAgB,CAAC;AACzD,YAAY,EAAE,iBAAiB,EAAE,MAAM,gBAAgB,CAAC;AAExD,OAAO,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAC;AACnD,YAAY,EAAE,eAAe,EAAE,YAAY,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAC;AAE3F,OAAO,EAAE,gBAAgB,EAAE,MAAM,sBAAsB,CAAC;AAExD,OAAO,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAC;AACzD,YAAY,EAAE,kBAAkB,EAAE,WAAW,EAAE,MAAM,yBAAyB,CAAC;AAC/E,OAAO,EAAE,YAAY,EAAE,MAAM,uBAAuB,CAAC;AACrD,OAAO,EAAE,yBAAyB,EAAE,MAAM,sBAAsB,CAAC;AACjE,OAAO,EAAE,eAAe,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAC;AACzE,YAAY,EAAE,aAAa,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,mBAAmB,EAAE,MAAM,sBAAsB,CAAC;AAC3D,YAAY,EAAE,iBAAiB,EAAE,MAAM,sBAAsB,CAAC;AAE9D,OAAO,EAAE,qBAAqB,EAAE,MAAM,qBAAqB,CAAC;AAE5D,OAAO,EACL,qBAAqB,EACrB,KAAK,oBAAoB,EACzB,KAAK,qBAAqB,GAC3B,MAAM,oBAAoB,CAAC;AAE5B,OAAO,EAAE,aAAa,EAAE,YAAY,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AACtF,OAAO,EAAE,gBAAgB,EAAE,MAAM,8BAA8B,CAAC;AAChE,YAAY,EAAE,oBAAoB,EAAE,MAAM,8BAA8B,CAAC;AACzE,OAAO,EAAE,kBAAkB,EAAE,MAAM,gCAAgC,CAAC;AAEpE,YAAY,EACV,OAAO,EACP,cAAc,EACd,WAAW,EACX,WAAW,EACX,UAAU,EACV,WAAW,EACX,YAAY,EACZ,aAAa,EACb,YAAY,EACZ,iBAAiB,EACjB,iBAAiB,EACjB,WAAW,EACX,oBAAoB,EACpB,cAAc,EACd,cAAc,EACd,SAAS,GACV,MAAM,qBAAqB,CAAC"}

package/dist/index.js

@@ -0,0 +1,33 @@
+export { Gateway } from './gateway';
+export { StreamBuffer } from './stream-buffer';
+export { HeartbeatScheduler } from './heartbeat';
+export { StatusReactionTracker } from './status-reactions';
+export { InboundDebouncer } from './inbound-debounce';
+export { MessageQueue } from './message-queue';
+export { formatEnvelope, formatElapsed } from './envelope';
+export { TerminalChannel, terminalChannel } from './channels/terminal';
+export { WebChatChannel, webchatChannel } from './channels/webchat';
+export { TelegramChannel, telegramChannel } from './channels/telegram';
+export { DiscordChannel, discordChannel } from './channels/discord';
+export { SlackChannel, slackChannel } from './channels/slack';
+export { WhatsAppChannel, whatsappChannel } from './channels/whatsapp';
+export { PairingMiddleware, pairing } from './middleware/pairing';
+export { DmPolicyMiddleware, dmPolicy } from './middleware/dm-policy';
+export { RateLimitMiddleware, rateLimit } from './middleware/rate-limit';
+export { OwnerCommandsMiddleware, ownerCommands } from './middleware/owner-commands';
+export { createHookRegistry } from './hooks';
+export { AutoExtractMiddleware, autoExtract } from './middleware/auto-extract';
+export { generateCapabilitiesDoc } from './capabilities';
+export { RuntimeBuilder } from './runtime-builder';
+export { SimpleTimerStore } from './simple-timer-store';
+export { MediaProcessor } from './media/media-processor';
+export { LocalWhisper } from './media/whisper-local';
+export { createWhisperDownloadTool } from './media/whisper-tool';
+export { GroqSttProvider, OpenAISttProvider } from './media/whisper-api';
+export { DeepgramSttProvider } from './media/deepgram-stt';
+export { createSelfConfigTools } from './tools/self-config';
+export { AssistantConfigSchema, } from './assistant-config';
+export { adaptMarkdown, chunkMessage, getPlatformLimit } from './formatters/markdown';
+export { chunkDiscordText } from './formatters/discord-chunker';
+export { markdownToWhatsApp } from './formatters/whatsapp-markdown';
+//# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AAGpC,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAE/C,OAAO,EAAE,kBAAkB,EAAE,MAAM,aAAa,CAAC;AAGjD,OAAO,EAAE,qBAAqB,EAAE,MAAM,oBAAoB,CAAC;AAC3D,OAAO,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AACtD,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC/C,OAAO,EAAE,cAAc,EAAE,aAAa,EAAE,MAAM,YAAY,CAAC;AAE3D,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAGvE,OAAO,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,oBAAoB,CAAC;AAGpE,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAGvE,OAAO,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,oBAAoB,CAAC;AAGpE,OAAO,EAAE,YAAY,EAAE,YAAY,EAAE,MAAM,kBAAkB,CAAC;AAG9D,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAGvE,OAAO,EAAE,iBAAiB,EAAE,OAAO,EAAE,MAAM,sBAAsB,CAAC;AAGlE,OAAO,EAAE,kBAAkB,EAAE,QAAQ,EAAE,MAAM,wBAAwB,CAAC;AAGtE,OAAO,EAAE,mBAAmB,EAAE,SAAS,EAAE,MAAM,yBAAyB,CAAC;AAGzE,OAAO,EAAE,uBAAuB,EAAE,aAAa,EAAE,MAAM,6BAA6B,CAAC;AAGrF,OAAO,EAAE,kBAAkB,EAAE,MAAM,SAAS,CAAC;AAG7C,OAAO,EAAE,qBAAqB,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAC;AAO/E,OAAO,EAAE,uBAAuB,EAAE,MAAM,gBAAgB,CAAC;AAGzD,OAAO,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAC;AAGnD,OAAO,EAAE,gBAAgB,EAAE,MAAM,sBAAsB,CAAC;AAExD,OAAO,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAC;AAEzD,OAAO,EAAE,YAAY,EAAE,MAAM,uBAAuB,CAAC;AACrD,OAAO,EAAE,yBAAyB,EAAE,MAAM,sBAAsB,CAAC;AACjE,OAAO,EAAE,eAAe,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAC;AAEzE,OAAO,EAAE,mBAAmB,EAAE,MAAM,sBAAsB,CAAC;AAG3D,OAAO,EAAE,qBAAqB,EAAE,MAAM,qBAAqB,CAAC;AAE5D,OAAO,EACL,qBAAqB,GAGtB,MAAM,oBAAoB,CAAC;AAE5B,OAAO,EAAE,aAAa,EAAE,YAAY,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AACtF,OAAO,EAAE,gBAAgB,EAAE,MAAM,8BAA8B,CAAC;AAEhE,OAAO,EAAE,kBAAkB,EAAE,MAAM,gCAAgC,CAAC"}

package/dist/media/deepgram-stt.d.ts

@@ -0,0 +1,14 @@
+import type { SttProvider } from './media-processor';
+export interface DeepgramSttConfig {
+    apiKey: string;
+    model?: string;
+    language?: string;
+}
+export declare class DeepgramSttProvider implements SttProvider {
+    private readonly apiKey;
+    private readonly model;
+    private readonly language;
+    constructor(config: DeepgramSttConfig);
+    transcribe(buffer: Buffer, _mimeType: string): Promise<string>;
+}
+//# sourceMappingURL=deepgram-stt.d.ts.map

package/dist/media/deepgram-stt.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"deepgram-stt.d.ts","sourceRoot":"","sources":["../../src/media/deepgram-stt.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAErD,MAAM,WAAW,iBAAiB;IAChC,MAAM,EAAE,MAAM,CAAC;IACf,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,qBAAa,mBAAoB,YAAW,WAAW;IACrD,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAS;IAChC,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAS;IAC/B,OAAO,CAAC,QAAQ,CAAC,QAAQ,CAAqB;gBAElC,MAAM,EAAE,iBAAiB;IAM/B,UAAU,CAAC,MAAM,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC;CAwBrE"}

package/dist/media/deepgram-stt.js

@@ -0,0 +1,30 @@
+export class DeepgramSttProvider {
+    apiKey;
+    model;
+    language;
+    constructor(config) {
+        this.apiKey = config.apiKey;
+        this.model = config.model ?? 'nova-3';
+        this.language = config.language;
+    }
+    async transcribe(buffer, _mimeType) {
+        const params = new URLSearchParams({ model: this.model });
+        if (this.language)
+            params.set('language', this.language);
+        const res = await fetch(`https://api.deepgram.com/v1/listen?${params}`, {
+            method: 'POST',
+            headers: {
+                Authorization: `Token ${this.apiKey}`,
+                'Content-Type': 'audio/ogg',
+            },
+            body: new Uint8Array(buffer),
+        });
+        if (!res.ok) {
+            const body = await res.text();
+            throw new Error(`Deepgram STT failed (${res.status}): ${body}`);
+        }
+        const data = (await res.json());
+        return data.results.channels[0]?.alternatives[0]?.transcript?.trim() ?? '';
+    }
+}
+//# sourceMappingURL=deepgram-stt.js.map

package/dist/media/deepgram-stt.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"deepgram-stt.js","sourceRoot":"","sources":["../../src/media/deepgram-stt.ts"],"names":[],"mappings":"AAQA,MAAM,OAAO,mBAAmB;IACb,MAAM,CAAS;IACf,KAAK,CAAS;IACd,QAAQ,CAAqB;IAE9C,YAAY,MAAyB;QACnC,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC;QAC5B,IAAI,CAAC,KAAK,GAAG,MAAM,CAAC,KAAK,IAAI,QAAQ,CAAC;QACtC,IAAI,CAAC,QAAQ,GAAG,MAAM,CAAC,QAAQ,CAAC;IAClC,CAAC;IAED,KAAK,CAAC,UAAU,CAAC,MAAc,EAAE,SAAiB;QAChD,MAAM,MAAM,GAAG,IAAI,eAAe,CAAC,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,EAAE,CAAC,CAAC;QAC1D,IAAI,IAAI,CAAC,QAAQ;YAAE,MAAM,CAAC,GAAG,CAAC,UAAU,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;QAEzD,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,sCAAsC,MAAM,EAAE,EAAE;YACtE,MAAM,EAAE,MAAM;YACd,OAAO,EAAE;gBACP,aAAa,EAAE,SAAS,IAAI,CAAC,MAAM,EAAE;gBACrC,cAAc,EAAE,WAAW;aAC5B;YACD,IAAI,EAAE,IAAI,UAAU,CAAC,MAAM,CAAC;SAC7B,CAAC,CAAC;QAEH,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC;YACZ,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,IAAI,EAAE,CAAC;YAC9B,MAAM,IAAI,KAAK,CAAC,wBAAwB,GAAG,CAAC,MAAM,MAAM,IAAI,EAAE,CAAC,CAAC;QAClE,CAAC;QAED,MAAM,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,CAE7B,CAAC;QAEF,OAAO,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,YAAY,CAAC,CAAC,CAAC,EAAE,UAAU,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;IAC7E,CAAC;CACF"}

package/dist/media/index.d.ts

@@ -0,0 +1,7 @@
+export { MediaProcessor } from './media-processor';
+export type { MediaProcessResult, SttProvider } from './media-processor';
+export { LocalWhisper } from './whisper-local';
+export { createWhisperDownloadTool } from './whisper-tool';
+export { GroqSttProvider, OpenAISttProvider } from './whisper-api';
+export type { GroqSttConfig, OpenAISttConfig } from './whisper-api';
+//# sourceMappingURL=index.d.ts.map

package/dist/media/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/media/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAC;AACnD,YAAY,EAAE,kBAAkB,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AACzE,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC/C,OAAO,EAAE,yBAAyB,EAAE,MAAM,gBAAgB,CAAC;AAC3D,OAAO,EAAE,eAAe,EAAE,iBAAiB,EAAE,MAAM,eAAe,CAAC;AACnE,YAAY,EAAE,aAAa,EAAE,eAAe,EAAE,MAAM,eAAe,CAAC"}

package/dist/media/index.js

@@ -0,0 +1,5 @@
+export { MediaProcessor } from './media-processor';
+export { LocalWhisper } from './whisper-local';
+export { createWhisperDownloadTool } from './whisper-tool';
+export { GroqSttProvider, OpenAISttProvider } from './whisper-api';
+//# sourceMappingURL=index.js.map

package/dist/media/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/media/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAC;AAEnD,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC/C,OAAO,EAAE,yBAAyB,EAAE,MAAM,gBAAgB,CAAC;AAC3D,OAAO,EAAE,eAAe,EAAE,iBAAiB,EAAE,MAAM,eAAe,CAAC"}

package/dist/media/media-processor.d.ts

@@ -0,0 +1,20 @@
+import type { Attachment, ImageInput } from '@cogitator-ai/types';
+import type { LocalWhisper } from './whisper-local';
+export interface SttProvider {
+    transcribe(buffer: Buffer, mimeType: string): Promise<string>;
+}
+export interface MediaProcessResult {
+    images: ImageInput[];
+    transcribedText: string | null;
+    systemNotes: string[];
+}
+export declare class MediaProcessor {
+    private whisper;
+    private checkVision;
+    private sttProvider?;
+    constructor(whisper: LocalWhisper | null, checkVision: (modelId: string) => boolean, sttProvider?: (SttProvider | null) | undefined);
+    process(attachments: Attachment[], modelId: string): Promise<MediaProcessResult>;
+    private transcribeAudio;
+    private normalizeImageMime;
+}
+//# sourceMappingURL=media-processor.d.ts.map

package/dist/media/media-processor.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"media-processor.d.ts","sourceRoot":"","sources":["../../src/media/media-processor.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,UAAU,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAClE,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAEpD,MAAM,WAAW,WAAW;IAC1B,UAAU,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC;CAC/D;AAED,MAAM,WAAW,kBAAkB;IACjC,MAAM,EAAE,UAAU,EAAE,CAAC;IACrB,eAAe,EAAE,MAAM,GAAG,IAAI,CAAC;IAC/B,WAAW,EAAE,MAAM,EAAE,CAAC;CACvB;AAED,qBAAa,cAAc;IAEvB,OAAO,CAAC,OAAO;IACf,OAAO,CAAC,WAAW;IACnB,OAAO,CAAC,WAAW,CAAC;gBAFZ,OAAO,EAAE,YAAY,GAAG,IAAI,EAC5B,WAAW,EAAE,CAAC,OAAO,EAAE,MAAM,KAAK,OAAO,EACzC,WAAW,CAAC,GAAE,WAAW,GAAG,IAAI,aAAA;IAGpC,OAAO,CAAC,WAAW,EAAE,UAAU,EAAE,EAAE,OAAO,EAAE,MAAM,GAAG,OAAO,CAAC,kBAAkB,CAAC;YA2CxE,eAAe;IAqC7B,OAAO,CAAC,kBAAkB;CAQ3B"}

package/dist/media/media-processor.js

@@ -0,0 +1,80 @@
+export class MediaProcessor {
+    whisper;
+    checkVision;
+    sttProvider;
+    constructor(whisper, checkVision, sttProvider) {
+        this.whisper = whisper;
+        this.checkVision = checkVision;
+        this.sttProvider = sttProvider;
+    }
+    async process(attachments, modelId) {
+        const images = [];
+        let transcribedText = null;
+        const systemNotes = [];
+        const imageAttachments = attachments.filter((a) => a.type === 'image');
+        const audioAttachments = attachments.filter((a) => a.type === 'audio');
+        if (imageAttachments.length > 0) {
+            const hasVision = this.checkVision(modelId);
+            if (hasVision) {
+                for (const att of imageAttachments) {
+                    if (att.buffer) {
+                        const mimeType = this.normalizeImageMime(att.mimeType);
+                        images.push({
+                            data: att.buffer.toString('base64'),
+                            mimeType,
+                        });
+                    }
+                    else if (att.url) {
+                        images.push(att.url);
+                    }
+                }
+            }
+            else {
+                systemNotes.push('[System: the user sent an image but your model does not support image recognition. Let them know politely.]');
+            }
+        }
+        if (audioAttachments.length > 0) {
+            const firstAudio = audioAttachments[0];
+            if (firstAudio.buffer) {
+                transcribedText = await this.transcribeAudio(firstAudio.buffer, firstAudio.mimeType, systemNotes);
+            }
+        }
+        return { images, transcribedText, systemNotes };
+    }
+    async transcribeAudio(buffer, mimeType, systemNotes) {
+        if (this.sttProvider) {
+            try {
+                return await this.sttProvider.transcribe(buffer, mimeType);
+            }
+            catch (err) {
+                console.error('[media] STT provider failed:', err.message);
+                systemNotes.push('[System: the user sent a voice message but transcription failed due to a technical error.]');
+                return null;
+            }
+        }
+        if (this.whisper?.isModelDownloaded()) {
+            try {
+                return await this.whisper.transcribe(buffer, mimeType);
+            }
+            catch (err) {
+                console.error('[media] Whisper transcription failed:', err.message);
+                systemNotes.push('[System: the user sent a voice message but transcription failed due to a technical error.]');
+                return null;
+            }
+        }
+        systemNotes.push('[System: the user sent a voice message but speech recognition is not set up yet. ' +
+            'Suggest downloading it using the download_stt_model tool (~75MB, works offline). ' +
+            'Ask the user for confirmation first.]');
+        return null;
+    }
+    normalizeImageMime(mime) {
+        if (mime === 'image/png')
+            return 'image/png';
+        if (mime === 'image/gif')
+            return 'image/gif';
+        if (mime === 'image/webp')
+            return 'image/webp';
+        return 'image/jpeg';
+    }
+}
+//# sourceMappingURL=media-processor.js.map

package/dist/media/media-processor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"media-processor.js","sourceRoot":"","sources":["../../src/media/media-processor.ts"],"names":[],"mappings":"AAaA,MAAM,OAAO,cAAc;IAEf;IACA;IACA;IAHV,YACU,OAA4B,EAC5B,WAAyC,EACzC,WAAgC;QAFhC,YAAO,GAAP,OAAO,CAAqB;QAC5B,gBAAW,GAAX,WAAW,CAA8B;QACzC,gBAAW,GAAX,WAAW,CAAqB;IACvC,CAAC;IAEJ,KAAK,CAAC,OAAO,CAAC,WAAyB,EAAE,OAAe;QACtD,MAAM,MAAM,GAAiB,EAAE,CAAC;QAChC,IAAI,eAAe,GAAkB,IAAI,CAAC;QAC1C,MAAM,WAAW,GAAa,EAAE,CAAC;QAEjC,MAAM,gBAAgB,GAAG,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,OAAO,CAAC,CAAC;QACvE,MAAM,gBAAgB,GAAG,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,OAAO,CAAC,CAAC;QAEvE,IAAI,gBAAgB,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAChC,MAAM,SAAS,GAAG,IAAI,CAAC,WAAW,CAAC,OAAO,CAAC,CAAC;YAC5C,IAAI,SAAS,EAAE,CAAC;gBACd,KAAK,MAAM,GAAG,IAAI,gBAAgB,EAAE,CAAC;oBACnC,IAAI,GAAG,CAAC,MAAM,EAAE,CAAC;wBACf,MAAM,QAAQ,GAAG,IAAI,CAAC,kBAAkB,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;wBACvD,MAAM,CAAC,IAAI,CAAC;4BACV,IAAI,EAAE,GAAG,CAAC,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC;4BACnC,QAAQ;yBACT,CAAC,CAAC;oBACL,CAAC;yBAAM,IAAI,GAAG,CAAC,GAAG,EAAE,CAAC;wBACnB,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;oBACvB,CAAC;gBACH,CAAC;YACH,CAAC;iBAAM,CAAC;gBACN,WAAW,CAAC,IAAI,CACd,6GAA6G,CAC9G,CAAC;YACJ,CAAC;QACH,CAAC;QAED,IAAI,gBAAgB,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAChC,MAAM,UAAU,GAAG,gBAAgB,CAAC,CAAC,CAAC,CAAC;YACvC,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC;gBACtB,eAAe,GAAG,MAAM,IAAI,CAAC,eAAe,CAC1C,UAAU,CAAC,MAAM,EACjB,UAAU,CAAC,QAAQ,EACnB,WAAW,CACZ,CAAC;YACJ,CAAC;QACH,CAAC;QAED,OAAO,EAAE,MAAM,EAAE,eAAe,EAAE,WAAW,EAAE,CAAC;IAClD,CAAC;IAEO,KAAK,CAAC,eAAe,CAC3B,MAAc,EACd,QAAgB,EAChB,WAAqB;QAErB,IAAI,IAAI,CAAC,WAAW,EAAE,CAAC;YACrB,IAAI,CAAC;gBACH,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAC;YAC7D,CAAC;YAAC,OAAO,GAAG,EAAE,CAAC;gBACb,OAAO,CAAC,KAAK,CAAC,8BAA8B,EAAG,GAAa,CAAC,OAAO,CAAC,CAAC;gBACtE,WAAW,CAAC,IAAI,CACd,4FAA4F,CAC7F,CAAC;gBACF,OAAO,IAAI,CAAC;YACd,CAAC;QACH,CAAC;QAED,IAAI,IAAI,CAAC,OAAO,EAAE,iBAAiB,EAAE,EAAE,CAAC;YACtC,IAAI,CAAC;gBACH,OAAO,MAAM,IAAI,CAAC,OAAO,CAAC,UAAU,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAC;YACzD,CAAC;YAAC,OAAO,GAAG,EAAE,CAAC;gBACb,OAAO,CAAC,KAAK,CAAC,uCAAuC,EAAG,GAAa,CAAC,OAAO,CAAC,CAAC;gBAC/E,WAAW,CAAC,IAAI,CACd,4FAA4F,CAC7F,CAAC;gBACF,OAAO,IAAI,CAAC;YACd,CAAC;QACH,CAAC;QAED,WAAW,CAAC,IAAI,CACd,mFAAmF;YACjF,mFAAmF;YACnF,uCAAuC,CAC1C,CAAC;QACF,OAAO,IAAI,CAAC;IACd,CAAC;IAEO,kBAAkB,CACxB,IAAY;QAEZ,IAAI,IAAI,KAAK,WAAW;YAAE,OAAO,WAAW,CAAC;QAC7C,IAAI,IAAI,KAAK,WAAW;YAAE,OAAO,WAAW,CAAC;QAC7C,IAAI,IAAI,KAAK,YAAY;YAAE,OAAO,YAAY,CAAC;QAC/C,OAAO,YAAY,CAAC;IACtB,CAAC;CACF"}

package/dist/media/whisper-api.d.ts

@@ -0,0 +1,22 @@
+import type { SttProvider } from './media-processor';
+export interface GroqSttConfig {
+    apiKey: string;
+    model?: string;
+}
+export declare class GroqSttProvider implements SttProvider {
+    private readonly apiKey;
+    private readonly model;
+    constructor(config: GroqSttConfig);
+    transcribe(buffer: Buffer, mimeType: string): Promise<string>;
+}
+export interface OpenAISttConfig {
+    apiKey: string;
+    model?: string;
+}
+export declare class OpenAISttProvider implements SttProvider {
+    private readonly apiKey;
+    private readonly model;
+    constructor(config: OpenAISttConfig);
+    transcribe(buffer: Buffer, mimeType: string): Promise<string>;
+}
+//# sourceMappingURL=whisper-api.d.ts.map

package/dist/media/whisper-api.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"whisper-api.d.ts","sourceRoot":"","sources":["../../src/media/whisper-api.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAErD,MAAM,WAAW,aAAa;IAC5B,MAAM,EAAE,MAAM,CAAC;IACf,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED,qBAAa,eAAgB,YAAW,WAAW;IACjD,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAS;IAChC,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAS;gBAEnB,MAAM,EAAE,aAAa;IAK3B,UAAU,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC;CAsBpE;AAED,MAAM,WAAW,eAAe;IAC9B,MAAM,EAAE,MAAM,CAAC;IACf,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED,qBAAa,iBAAkB,YAAW,WAAW;IACnD,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAS;IAChC,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAS;gBAEnB,MAAM,EAAE,eAAe;IAK7B,UAAU,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC;CAsBpE"}

package/dist/media/whisper-api.js

@@ -0,0 +1,53 @@
+export class GroqSttProvider {
+    apiKey;
+    model;
+    constructor(config) {
+        this.apiKey = config.apiKey;
+        this.model = config.model ?? 'whisper-large-v3';
+    }
+    async transcribe(buffer, mimeType) {
+        const ext = mimeType.includes('ogg') ? 'ogg' : mimeType.includes('wav') ? 'wav' : 'ogg';
+        const blob = new Blob([new Uint8Array(buffer)], { type: mimeType });
+        const form = new FormData();
+        form.append('file', blob, `audio.${ext}`);
+        form.append('model', this.model);
+        const res = await fetch('https://api.groq.com/openai/v1/audio/transcriptions', {
+            method: 'POST',
+            headers: { Authorization: `Bearer ${this.apiKey}` },
+            body: form,
+        });
+        if (!res.ok) {
+            const body = await res.text();
+            throw new Error(`Groq STT failed (${res.status}): ${body}`);
+        }
+        const data = (await res.json());
+        return data.text.trim();
+    }
+}
+export class OpenAISttProvider {
+    apiKey;
+    model;
+    constructor(config) {
+        this.apiKey = config.apiKey;
+        this.model = config.model ?? 'whisper-1';
+    }
+    async transcribe(buffer, mimeType) {
+        const ext = mimeType.includes('ogg') ? 'ogg' : mimeType.includes('wav') ? 'wav' : 'ogg';
+        const blob = new Blob([new Uint8Array(buffer)], { type: mimeType });
+        const form = new FormData();
+        form.append('file', blob, `audio.${ext}`);
+        form.append('model', this.model);
+        const res = await fetch('https://api.openai.com/v1/audio/transcriptions', {
+            method: 'POST',
+            headers: { Authorization: `Bearer ${this.apiKey}` },
+            body: form,
+        });
+        if (!res.ok) {
+            const body = await res.text();
+            throw new Error(`OpenAI STT failed (${res.status}): ${body}`);
+        }
+        const data = (await res.json());
+        return data.text.trim();
+    }
+}
+//# sourceMappingURL=whisper-api.js.map

package/dist/media/whisper-api.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"whisper-api.js","sourceRoot":"","sources":["../../src/media/whisper-api.ts"],"names":[],"mappings":"AAOA,MAAM,OAAO,eAAe;IACT,MAAM,CAAS;IACf,KAAK,CAAS;IAE/B,YAAY,MAAqB;QAC/B,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC;QAC5B,IAAI,CAAC,KAAK,GAAG,MAAM,CAAC,KAAK,IAAI,kBAAkB,CAAC;IAClD,CAAC;IAED,KAAK,CAAC,UAAU,CAAC,MAAc,EAAE,QAAgB;QAC/C,MAAM,GAAG,GAAG,QAAQ,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,QAAQ,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC;QACxF,MAAM,IAAI,GAAG,IAAI,IAAI,CAAC,CAAC,IAAI,UAAU,CAAC,MAAM,CAAC,CAAC,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,CAAC;QAEpE,MAAM,IAAI,GAAG,IAAI,QAAQ,EAAE,CAAC;QAC5B,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,IAAI,EAAE,SAAS,GAAG,EAAE,CAAC,CAAC;QAC1C,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC;QAEjC,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,qDAAqD,EAAE;YAC7E,MAAM,EAAE,MAAM;YACd,OAAO,EAAE,EAAE,aAAa,EAAE,UAAU,IAAI,CAAC,MAAM,EAAE,EAAE;YACnD,IAAI,EAAE,IAAI;SACX,CAAC,CAAC;QAEH,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC;YACZ,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,IAAI,EAAE,CAAC;YAC9B,MAAM,IAAI,KAAK,CAAC,oBAAoB,GAAG,CAAC,MAAM,MAAM,IAAI,EAAE,CAAC,CAAC;QAC9D,CAAC;QAED,MAAM,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,CAAqB,CAAC;QACpD,OAAO,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;IAC1B,CAAC;CACF;AAOD,MAAM,OAAO,iBAAiB;IACX,MAAM,CAAS;IACf,KAAK,CAAS;IAE/B,YAAY,MAAuB;QACjC,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC;QAC5B,IAAI,CAAC,KAAK,GAAG,MAAM,CAAC,KAAK,IAAI,WAAW,CAAC;IAC3C,CAAC;IAED,KAAK,CAAC,UAAU,CAAC,MAAc,EAAE,QAAgB;QAC/C,MAAM,GAAG,GAAG,QAAQ,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,QAAQ,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC;QACxF,MAAM,IAAI,GAAG,IAAI,IAAI,CAAC,CAAC,IAAI,UAAU,CAAC,MAAM,CAAC,CAAC,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,CAAC;QAEpE,MAAM,IAAI,GAAG,IAAI,QAAQ,EAAE,CAAC;QAC5B,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,IAAI,EAAE,SAAS,GAAG,EAAE,CAAC,CAAC;QAC1C,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC;QAEjC,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,gDAAgD,EAAE;YACxE,MAAM,EAAE,MAAM;YACd,OAAO,EAAE,EAAE,aAAa,EAAE,UAAU,IAAI,CAAC,MAAM,EAAE,EAAE;YACnD,IAAI,EAAE,IAAI;SACX,CAAC,CAAC;QAEH,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC;YACZ,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,IAAI,EAAE,CAAC;YAC9B,MAAM,IAAI,KAAK,CAAC,sBAAsB,GAAG,CAAC,MAAM,MAAM,IAAI,EAAE,CAAC,CAAC;QAChE,CAAC;QAED,MAAM,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,CAAqB,CAAC;QACpD,OAAO,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;IAC1B,CAAC;CACF"}

package/dist/media/whisper-local.d.ts

@@ -0,0 +1,16 @@
+export declare class LocalWhisper {
+    private pipeline;
+    private loading;
+    readonly modelDir: string;
+    constructor(modelDir?: string);
+    isModelDownloaded(): boolean;
+    download(): Promise<void>;
+    transcribe(audioBuffer: Buffer, mimeType: string): Promise<string>;
+    private ensureLoaded;
+    private ensureDeps;
+    private loadTransformers;
+    private decodeOgg;
+    private decodeWav;
+    private resample;
+}
+//# sourceMappingURL=whisper-local.d.ts.map

package/dist/media/whisper-local.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"whisper-local.d.ts","sourceRoot":"","sources":["../../src/media/whisper-local.ts"],"names":[],"mappings":"AAQA,qBAAa,YAAY;IACvB,OAAO,CAAC,QAAQ,CAAiB;IACjC,OAAO,CAAC,OAAO,CAA8B;IAC7C,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;gBAEd,QAAQ,CAAC,EAAE,MAAM;IAI7B,iBAAiB,IAAI,OAAO;IAgBtB,QAAQ,IAAI,OAAO,CAAC,IAAI,CAAC;IAkBzB,UAAU,CAAC,WAAW,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC;YAgC1D,YAAY;YA0BZ,UAAU;YAoBV,gBAAgB;YAIhB,SAAS;IAoBvB,OAAO,CAAC,SAAS;IAyBjB,OAAO,CAAC,QAAQ;CAajB"}

package/dist/media/whisper-local.js

@@ -0,0 +1,155 @@
+import { existsSync, readdirSync } from 'node:fs';
+import { execSync } from 'node:child_process';
+import { join } from 'node:path';
+import { homedir } from 'node:os';
+const WHISPER_MODEL = 'Xenova/whisper-tiny';
+const SAMPLE_RATE = 16000;
+export class LocalWhisper {
+    pipeline = null;
+    loading = null;
+    modelDir;
+    constructor(modelDir) {
+        this.modelDir = modelDir ?? join(homedir(), '.cogitator', 'models');
+    }
+    isModelDownloaded() {
+        const jsCache = join(this.modelDir, 'Xenova', 'whisper-tiny', 'onnx');
+        if (existsSync(jsCache))
+            return true;
+        const pyCache = join(this.modelDir, 'models--Xenova--whisper-tiny', 'snapshots');
+        if (existsSync(pyCache)) {
+            try {
+                return readdirSync(pyCache).length > 0;
+            }
+            catch {
+                return false;
+            }
+        }
+        return false;
+    }
+    async download() {
+        await this.ensureDeps();
+        const transformers = await this.loadTransformers();
+        const { env, pipeline } = transformers;
+        env.cacheDir = this.modelDir;
+        env.allowLocalModels = true;
+        console.log(`[whisper] Downloading ${WHISPER_MODEL} to ${this.modelDir}...`);
+        this.pipeline = await pipeline('automatic-speech-recognition', WHISPER_MODEL, {
+            dtype: 'q8',
+        });
+        console.log('[whisper] Model downloaded and ready');
+    }
+    async transcribe(audioBuffer, mimeType) {
+        if (!this.pipeline) {
+            await this.ensureLoaded();
+        }
+        let pcmFloat32;
+        if (mimeType === 'audio/ogg' || mimeType === 'audio/opus') {
+            pcmFloat32 = await this.decodeOgg(audioBuffer);
+        }
+        else if (mimeType === 'audio/wav' || mimeType === 'audio/wave') {
+            pcmFloat32 = this.decodeWav(audioBuffer);
+        }
+        else {
+            pcmFloat32 = await this.decodeOgg(audioBuffer);
+        }
+        const pipe = this.pipeline;
+        const startToken = pipe.model.config.decoder_start_token_id;
+        const noTsToken = pipe.model.generation_config.no_timestamps_token_id;
+        const result = await pipe(pcmFloat32, {
+            decoder_input_ids: [startToken, noTsToken],
+        });
+        return result.text.trim();
+    }
+    async ensureLoaded() {
+        if (this.pipeline)
+            return;
+        if (this.loading) {
+            await this.loading;
+            return;
+        }
+        this.loading = (async () => {
+            const transformers = await this.loadTransformers();
+            const { env, pipeline } = transformers;
+            env.cacheDir = this.modelDir;
+            env.allowLocalModels = true;
+            this.pipeline = await pipeline('automatic-speech-recognition', WHISPER_MODEL, {
+                dtype: 'q8',
+            });
+        })();
+        await this.loading;
+        this.loading = null;
+    }
+    async ensureDeps() {
+        const missing = [];
+        try {
+            await import('@huggingface/transformers');
+        }
+        catch {
+            missing.push('@huggingface/transformers');
+        }
+        try {
+            await import('ogg-opus-decoder');
+        }
+        catch {
+            missing.push('ogg-opus-decoder');
+        }
+        if (missing.length === 0)
+            return;
+        console.log(`[whisper] Installing dependencies: ${missing.join(', ')}...`);
+        const cmd = `npm install --no-save ${missing.join(' ')}`;
+        execSync(cmd, { stdio: 'pipe', timeout: 120_000 });
+        console.log('[whisper] Dependencies installed');
+    }
+    async loadTransformers() {
+        return import('@huggingface/transformers');
+    }
+    async decodeOgg(buffer) {
+        const mod = (await import('ogg-opus-decoder'));
+        const OggOpusDecoder = mod.OggOpusDecoder;
+        const decoder = new OggOpusDecoder();
+        await decoder.ready;
+        const result = await decoder.decode(new Uint8Array(buffer));
+        decoder.free();
+        const samples = result.channelData[0];
+        if (result.sampleRate === SAMPLE_RATE)
+            return samples;
+        return this.resample(samples, result.sampleRate, SAMPLE_RATE);
+    }
+    decodeWav(buffer) {
+        const dataOffset = buffer.indexOf('data') + 8;
+        const bitsPerSample = buffer.readUInt16LE(34);
+        const sampleRate = buffer.readUInt32LE(24);
+        const numChannels = buffer.readUInt16LE(22);
+        const dataSize = buffer.readUInt32LE(dataOffset - 4);
+        const numSamples = dataSize / (bitsPerSample / 8) / numChannels;
+        const samples = new Float32Array(numSamples);
+        let offset = dataOffset;
+        for (let i = 0; i < numSamples; i++) {
+            if (bitsPerSample === 16) {
+                samples[i] = buffer.readInt16LE(offset) / 32768;
+                offset += 2 * numChannels;
+            }
+            else if (bitsPerSample === 32) {
+                samples[i] = buffer.readFloatLE(offset);
+                offset += 4 * numChannels;
+            }
+        }
+        if (sampleRate === SAMPLE_RATE)
+            return samples;
+        return this.resample(samples, sampleRate, SAMPLE_RATE);
+    }
+    resample(samples, fromRate, toRate) {
+        const ratio = fromRate / toRate;
+        const newLength = Math.round(samples.length / ratio);
+        const result = new Float32Array(newLength);
+        for (let i = 0; i < newLength; i++) {
+            const srcIdx = i * ratio;
+            const lo = Math.floor(srcIdx);
+            const hi = Math.min(lo + 1, samples.length - 1);
+            const frac = srcIdx - lo;
+            result[i] = samples[lo] * (1 - frac) + samples[hi] * frac;
+        }
+        return result;
+    }
+}
+//# sourceMappingURL=whisper-local.js.map