botinabox 2.14.0 → 2.15.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -1,223 +0,0 @@
1
- // src/channels/slack/transcribe.ts
2
- import { execFileSync } from "child_process";
3
- import { writeFileSync, unlinkSync, mkdirSync } from "fs";
4
- import { join } from "path";
5
- import { randomUUID } from "crypto";
6
- import os from "os";
7
- import { createRequire } from "module";
8
- var TEMP_DIR = join(os.tmpdir(), "botinabox-audio");
9
- async function transcribeAudio(audioBuffer, filename, opts) {
10
- let whisper;
11
- try {
12
- const require2 = createRequire(import.meta.url);
13
- const mod = require2("whisper-node");
14
- whisper = mod.whisper ?? mod.default ?? mod;
15
- } catch {
16
- console.warn("[botinabox] whisper-node not installed \u2014 voice transcription unavailable. Run: npm install whisper-node && npx whisper-node download");
17
- return null;
18
- }
19
- try {
20
- execFileSync("ffmpeg", ["-version"], { stdio: "ignore" });
21
- } catch {
22
- console.warn("[botinabox] ffmpeg not found \u2014 required for audio conversion. Install: brew install ffmpeg");
23
- return null;
24
- }
25
- const id = randomUUID().slice(0, 8);
26
- const ext = filename.split(".").pop() ?? "aac";
27
- mkdirSync(TEMP_DIR, { recursive: true });
28
- const inputPath = join(TEMP_DIR, `${id}.${ext}`);
29
- const wavPath = join(TEMP_DIR, `${id}.wav`);
30
- try {
31
- writeFileSync(inputPath, audioBuffer);
32
- execFileSync("ffmpeg", ["-y", "-i", inputPath, "-ar", "16000", "-ac", "1", "-c:a", "pcm_s16le", wavPath], {
33
- stdio: "ignore",
34
- timeout: 3e4
35
- });
36
- const segments = await whisper(wavPath, {
37
- modelName: opts?.modelName ?? "base.en",
38
- whisperOptions: {
39
- language: opts?.language ?? "auto"
40
- }
41
- });
42
- if (!segments || segments.length === 0) return null;
43
- return segments.map((s) => s.speech).join(" ").trim();
44
- } catch (err) {
45
- console.error("[botinabox] Transcription failed:", err);
46
- return null;
47
- } finally {
48
- try {
49
- unlinkSync(inputPath);
50
- } catch {
51
- }
52
- try {
53
- unlinkSync(wavPath);
54
- } catch {
55
- }
56
- }
57
- }
58
- async function downloadAudio(url, token) {
59
- try {
60
- const resp = await fetch(url, {
61
- headers: { Authorization: `Bearer ${token}` }
62
- });
63
- if (!resp.ok) {
64
- console.error(`[botinabox] Audio download failed: ${resp.status} ${resp.statusText}`);
65
- return null;
66
- }
67
- return Buffer.from(await resp.arrayBuffer());
68
- } catch (err) {
69
- console.error("[botinabox] Audio download error:", err);
70
- return null;
71
- }
72
- }
73
-
74
- // src/channels/slack/media-type.ts
75
- var FILETYPE_MAP = {
76
- // image
77
- jpg: "image",
78
- jpeg: "image",
79
- png: "image",
80
- gif: "image",
81
- webp: "image",
82
- heic: "image",
83
- svg: "image",
84
- bmp: "image",
85
- // video
86
- mp4: "video",
87
- mov: "video",
88
- webm: "video",
89
- avi: "video",
90
- mkv: "video",
91
- // audio (also handled by voice-message path — included for completeness)
92
- aac: "audio",
93
- m4a: "audio",
94
- mp3: "audio",
95
- wav: "audio",
96
- ogg: "audio",
97
- flac: "audio",
98
- // pdf
99
- pdf: "pdf",
100
- // doc
101
- gdoc: "doc",
102
- docx: "doc",
103
- doc: "doc",
104
- md: "doc",
105
- txt: "doc",
106
- rtf: "doc",
107
- // excel
108
- gsheet: "excel",
109
- xlsx: "excel",
110
- xls: "excel",
111
- csv: "excel",
112
- tsv: "excel",
113
- // presentation
114
- gslide: "presentation",
115
- pptx: "presentation",
116
- ppt: "presentation",
117
- key: "presentation",
118
- // html
119
- html: "html",
120
- htm: "html"
121
- };
122
- function slackFiletypeToMediaType(filetype) {
123
- if (!filetype) return "misc";
124
- return FILETYPE_MAP[filetype.toLowerCase()] ?? "misc";
125
- }
126
- var URL_REGEX = /https?:\/\/[^\s<>"')]+/g;
127
- function extractUrls(text) {
128
- if (!text) return [];
129
- const matches = text.match(URL_REGEX);
130
- if (!matches) return [];
131
- return Array.from(new Set(matches.map((u) => u.replace(/[.,;:!?)]+$/, ""))));
132
- }
133
-
134
- // src/channels/slack/inbound.ts
135
- var AUDIO_TYPES = /* @__PURE__ */ new Set(["aac", "mp4", "m4a", "ogg", "webm", "mp3", "wav"]);
136
- function extractVoiceTranscript(file) {
137
- const isAudio = file.subtype === "slack_audio" || AUDIO_TYPES.has(file.filetype ?? "");
138
- if (!isAudio) return null;
139
- const transcript = file.transcription?.preview?.content ?? (typeof file.preview === "string" ? file.preview : null);
140
- return transcript ?? null;
141
- }
142
- function parseSlackEvent(event) {
143
- const id = event.client_msg_id ?? event.ts ?? event.event_ts ?? `slack-${Date.now()}`;
144
- const channel = event.channel ?? "unknown";
145
- const from = event.user ?? "unknown";
146
- const threadId = event.thread_ts !== void 0 ? event.thread_ts : void 0;
147
- const receivedAt = event.ts ? new Date(parseFloat(event.ts) * 1e3).toISOString() : (/* @__PURE__ */ new Date()).toISOString();
148
- let body = event.text ?? "";
149
- if (event.subtype === "file_share" && event.files?.length) {
150
- for (const file of event.files) {
151
- const transcript = extractVoiceTranscript(file);
152
- if (transcript) {
153
- body = body ? `${body}
154
-
155
- [Voice message] ${transcript}` : `[Voice message] ${transcript}`;
156
- break;
157
- }
158
- }
159
- }
160
- if (event.subtype === "file_share" && event.files?.length && !body) {
161
- const hasAudio = event.files.some(
162
- (f) => f.subtype === "slack_audio" || AUDIO_TYPES.has(f.filetype ?? "")
163
- );
164
- if (hasAudio) {
165
- body = "[Voice message \u2014 no transcript available]";
166
- }
167
- }
168
- const attachments = [];
169
- if (event.subtype === "file_share" && event.files?.length) {
170
- for (const file of event.files) {
171
- const isAudio = file.subtype === "slack_audio" || AUDIO_TYPES.has(file.filetype ?? "");
172
- if (isAudio) continue;
173
- attachments.push({
174
- type: slackFiletypeToMediaType(file.filetype),
175
- url: file.url_private,
176
- mimeType: file.mimetype,
177
- filename: file.name ?? file.title,
178
- size: file.size
179
- });
180
- }
181
- }
182
- const urls = extractUrls(body);
183
- for (const url of urls) {
184
- attachments.push({ type: "link", url });
185
- }
186
- return {
187
- id,
188
- channel,
189
- from,
190
- body,
191
- threadId,
192
- attachments: attachments.length > 0 ? attachments : void 0,
193
- receivedAt,
194
- raw: event
195
- };
196
- }
197
- async function enrichVoiceMessage(msg, botToken) {
198
- if (!msg.body.includes("[Voice message \u2014 no transcript available]")) return msg;
199
- const raw = msg.raw;
200
- const files = raw?.files;
201
- if (!files?.length) return msg;
202
- const audioFile = files.find(
203
- (f) => f.subtype === "slack_audio" || AUDIO_TYPES.has(f.filetype ?? "")
204
- );
205
- if (!audioFile?.url_private) return msg;
206
- const buffer = await downloadAudio(audioFile.url_private, botToken);
207
- if (!buffer) return msg;
208
- const filename = audioFile.name ?? `voice.${audioFile.filetype ?? "aac"}`;
209
- const transcript = await transcribeAudio(buffer, filename);
210
- if (!transcript) return msg;
211
- return {
212
- ...msg,
213
- body: `[Voice message] ${transcript}`
214
- };
215
- }
216
-
217
- export {
218
- transcribeAudio,
219
- downloadAudio,
220
- extractVoiceTranscript,
221
- parseSlackEvent,
222
- enrichVoiceMessage
223
- };
@@ -1,224 +0,0 @@
1
- // src/channels/slack/transcribe.ts
2
- import { execFileSync } from "child_process";
3
- import { writeFileSync, unlinkSync, mkdirSync } from "fs";
4
- import { join } from "path";
5
- import { randomUUID } from "crypto";
6
- import os from "os";
7
- import { createRequire } from "module";
8
- var TEMP_DIR = join(os.tmpdir(), "botinabox-audio");
9
- async function transcribeAudio(audioBuffer, filename, opts) {
10
- let whisper;
11
- try {
12
- const require2 = createRequire(import.meta.url);
13
- const mod = require2("whisper-node");
14
- whisper = mod.whisper ?? mod.default ?? mod;
15
- } catch {
16
- console.warn("[botinabox] whisper-node not installed \u2014 voice transcription unavailable. Run: npm install whisper-node && npx whisper-node download");
17
- return null;
18
- }
19
- try {
20
- execFileSync("ffmpeg", ["-version"], { stdio: "ignore" });
21
- } catch {
22
- console.warn("[botinabox] ffmpeg not found \u2014 required for audio conversion. Install: brew install ffmpeg");
23
- return null;
24
- }
25
- const id = randomUUID().slice(0, 8);
26
- const ext = filename.split(".").pop() ?? "aac";
27
- mkdirSync(TEMP_DIR, { recursive: true });
28
- const inputPath = join(TEMP_DIR, `${id}.${ext}`);
29
- const wavPath = join(TEMP_DIR, `${id}.wav`);
30
- try {
31
- writeFileSync(inputPath, audioBuffer);
32
- execFileSync("ffmpeg", ["-y", "-i", inputPath, "-ar", "16000", "-ac", "1", "-c:a", "pcm_s16le", wavPath], {
33
- stdio: "ignore",
34
- timeout: 3e4
35
- });
36
- const segments = await whisper(wavPath, {
37
- modelName: opts?.modelName ?? "base.en",
38
- whisperOptions: {
39
- language: opts?.language ?? "auto"
40
- }
41
- });
42
- if (!segments || segments.length === 0) return null;
43
- return segments.map((s) => s.speech).join(" ").trim();
44
- } catch (err) {
45
- console.error("[botinabox] Transcription failed:", err);
46
- return null;
47
- } finally {
48
- try {
49
- unlinkSync(inputPath);
50
- } catch {
51
- }
52
- try {
53
- unlinkSync(wavPath);
54
- } catch {
55
- }
56
- }
57
- }
58
- async function downloadAudio(url, token) {
59
- try {
60
- const resp = await fetch(url, {
61
- headers: { Authorization: `Bearer ${token}` }
62
- });
63
- if (!resp.ok) {
64
- console.error(`[botinabox] Audio download failed: ${resp.status} ${resp.statusText}`);
65
- return null;
66
- }
67
- return Buffer.from(await resp.arrayBuffer());
68
- } catch (err) {
69
- console.error("[botinabox] Audio download error:", err);
70
- return null;
71
- }
72
- }
73
-
74
- // src/channels/slack/media-type.ts
75
- var FILETYPE_MAP = {
76
- // image
77
- jpg: "image",
78
- jpeg: "image",
79
- png: "image",
80
- gif: "image",
81
- webp: "image",
82
- heic: "image",
83
- svg: "image",
84
- bmp: "image",
85
- // video
86
- mp4: "video",
87
- mov: "video",
88
- webm: "video",
89
- avi: "video",
90
- mkv: "video",
91
- // audio (also handled by voice-message path — included for completeness)
92
- aac: "audio",
93
- m4a: "audio",
94
- mp3: "audio",
95
- wav: "audio",
96
- ogg: "audio",
97
- flac: "audio",
98
- // pdf
99
- pdf: "pdf",
100
- // doc
101
- gdoc: "doc",
102
- docx: "doc",
103
- doc: "doc",
104
- md: "doc",
105
- txt: "doc",
106
- rtf: "doc",
107
- // excel
108
- gsheet: "excel",
109
- xlsx: "excel",
110
- xls: "excel",
111
- csv: "excel",
112
- tsv: "excel",
113
- // presentation
114
- gslide: "presentation",
115
- pptx: "presentation",
116
- ppt: "presentation",
117
- key: "presentation",
118
- // html
119
- html: "html",
120
- htm: "html"
121
- };
122
- function slackFiletypeToMediaType(filetype) {
123
- if (!filetype) return "misc";
124
- return FILETYPE_MAP[filetype.toLowerCase()] ?? "misc";
125
- }
126
- var URL_REGEX = /https?:\/\/[^\s<>"')]+/g;
127
- function extractUrls(text) {
128
- if (!text) return [];
129
- const matches = text.match(URL_REGEX);
130
- if (!matches) return [];
131
- return Array.from(new Set(matches.map((u) => u.replace(/[.,;:!?)]+$/, ""))));
132
- }
133
-
134
- // src/channels/slack/inbound.ts
135
- var AUDIO_TYPES = /* @__PURE__ */ new Set(["aac", "mp4", "m4a", "ogg", "webm", "mp3", "wav"]);
136
- function extractVoiceTranscript(file) {
137
- const isAudio = file.subtype === "slack_audio" || AUDIO_TYPES.has(file.filetype ?? "");
138
- if (!isAudio) return null;
139
- const transcript = file.transcription?.preview?.content ?? (typeof file.preview === "string" ? file.preview : null);
140
- return transcript ?? null;
141
- }
142
- function parseSlackEvent(event) {
143
- const id = event.client_msg_id ?? event.ts ?? event.event_ts ?? `slack-${Date.now()}`;
144
- const channel = event.channel ?? "unknown";
145
- const from = event.user ?? "unknown";
146
- const isChannel = channel.startsWith("C") || channel.startsWith("G");
147
- const threadId = event.thread_ts ?? (isChannel ? event.ts : void 0);
148
- const receivedAt = event.ts ? new Date(parseFloat(event.ts) * 1e3).toISOString() : (/* @__PURE__ */ new Date()).toISOString();
149
- let body = event.text ?? "";
150
- if (event.subtype === "file_share" && event.files?.length) {
151
- for (const file of event.files) {
152
- const transcript = extractVoiceTranscript(file);
153
- if (transcript) {
154
- body = body ? `${body}
155
-
156
- [Voice message] ${transcript}` : `[Voice message] ${transcript}`;
157
- break;
158
- }
159
- }
160
- }
161
- if (event.subtype === "file_share" && event.files?.length && !body) {
162
- const hasAudio = event.files.some(
163
- (f) => f.subtype === "slack_audio" || AUDIO_TYPES.has(f.filetype ?? "")
164
- );
165
- if (hasAudio) {
166
- body = "[Voice message \u2014 no transcript available]";
167
- }
168
- }
169
- const attachments = [];
170
- if (event.subtype === "file_share" && event.files?.length) {
171
- for (const file of event.files) {
172
- const isAudio = file.subtype === "slack_audio" || AUDIO_TYPES.has(file.filetype ?? "");
173
- if (isAudio) continue;
174
- attachments.push({
175
- type: slackFiletypeToMediaType(file.filetype),
176
- url: file.url_private,
177
- mimeType: file.mimetype,
178
- filename: file.name ?? file.title,
179
- size: file.size
180
- });
181
- }
182
- }
183
- const urls = extractUrls(body);
184
- for (const url of urls) {
185
- attachments.push({ type: "link", url });
186
- }
187
- return {
188
- id,
189
- channel,
190
- from,
191
- body,
192
- threadId,
193
- attachments: attachments.length > 0 ? attachments : void 0,
194
- receivedAt,
195
- raw: event
196
- };
197
- }
198
- async function enrichVoiceMessage(msg, botToken) {
199
- if (!msg.body.includes("[Voice message \u2014 no transcript available]")) return msg;
200
- const raw = msg.raw;
201
- const files = raw?.files;
202
- if (!files?.length) return msg;
203
- const audioFile = files.find(
204
- (f) => f.subtype === "slack_audio" || AUDIO_TYPES.has(f.filetype ?? "")
205
- );
206
- if (!audioFile?.url_private) return msg;
207
- const buffer = await downloadAudio(audioFile.url_private, botToken);
208
- if (!buffer) return msg;
209
- const filename = audioFile.name ?? `voice.${audioFile.filetype ?? "aac"}`;
210
- const transcript = await transcribeAudio(buffer, filename);
211
- if (!transcript) return msg;
212
- return {
213
- ...msg,
214
- body: `[Voice message] ${transcript}`
215
- };
216
- }
217
-
218
- export {
219
- transcribeAudio,
220
- downloadAudio,
221
- extractVoiceTranscript,
222
- parseSlackEvent,
223
- enrichVoiceMessage
224
- };