@delofarag/ai-utils 1.4.3 → 1.6.0
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.
- package/README.md +354 -216
- package/dist/heart/agent.d.ts +1 -1
- package/dist/heart/agent.d.ts.map +1 -1
- package/dist/heart/agent.js +4 -9
- package/dist/heart/agent.js.map +1 -1
- package/dist/heart/chain.d.ts +1 -1
- package/dist/heart/chain.d.ts.map +1 -1
- package/dist/heart/chain.js +4 -4
- package/dist/heart/chain.js.map +1 -1
- package/dist/heart/chatbot.d.ts +1 -1
- package/dist/heart/memorychain.d.ts +2 -2
- package/dist/heart/memorychain.js +6 -6
- package/dist/heart/memorychain.js.map +1 -1
- package/dist/heart/tools/zodios/ZodiosToolRegistry.d.ts +1 -1
- package/dist/heart/tools/zodios/ZodiosToolRegistry.d.ts.map +1 -1
- package/dist/helpers/chatbot.d.ts +31 -0
- package/dist/helpers/chatbot.d.ts.map +1 -0
- package/dist/helpers/chatbot.js +71 -0
- package/dist/helpers/chatbot.js.map +1 -0
- package/dist/helpers/helpers.d.ts +25 -0
- package/dist/helpers/helpers.d.ts.map +1 -0
- package/dist/helpers/helpers.js +43 -0
- package/dist/helpers/helpers.js.map +1 -0
- package/dist/helpers/llms.d.ts +95 -0
- package/dist/helpers/llms.d.ts.map +1 -0
- package/dist/helpers/llms.js +62 -0
- package/dist/helpers/llms.js.map +1 -0
- package/dist/helpers/memory.d.ts +107 -0
- package/dist/helpers/memory.d.ts.map +1 -0
- package/dist/helpers/memory.js +497 -0
- package/dist/helpers/memory.js.map +1 -0
- package/dist/helpers/rag.d.ts +42 -0
- package/dist/helpers/rag.d.ts.map +1 -0
- package/dist/helpers/rag.js +66 -0
- package/dist/helpers/rag.js.map +1 -0
- package/dist/helpers.d.ts +3 -3
- package/dist/index.d.ts +10 -10
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +10 -10
- package/dist/index.js.map +1 -1
- package/dist/magic-funcs/answerers/ask.d.ts +18 -2
- package/dist/magic-funcs/answerers/ask.d.ts.map +1 -1
- package/dist/magic-funcs/answerers/ask.js +20 -9
- package/dist/magic-funcs/answerers/ask.js.map +1 -1
- package/dist/magic-funcs/answerers/websearch.d.ts +15 -0
- package/dist/magic-funcs/answerers/websearch.d.ts.map +1 -1
- package/dist/magic-funcs/answerers/websearch.js +15 -0
- package/dist/magic-funcs/answerers/websearch.js.map +1 -1
- package/dist/magic-funcs/evaluators/classify.d.ts +20 -1
- package/dist/magic-funcs/evaluators/classify.d.ts.map +1 -1
- package/dist/magic-funcs/evaluators/classify.js +22 -3
- package/dist/magic-funcs/evaluators/classify.js.map +1 -1
- package/dist/magic-funcs/evaluators/decide.d.ts +18 -0
- package/dist/magic-funcs/evaluators/decide.d.ts.map +1 -1
- package/dist/magic-funcs/evaluators/decide.js +21 -3
- package/dist/magic-funcs/evaluators/decide.js.map +1 -1
- package/dist/magic-funcs/optimizers/promptify.d.ts +18 -0
- package/dist/magic-funcs/optimizers/promptify.d.ts.map +1 -1
- package/dist/magic-funcs/optimizers/promptify.js +21 -3
- package/dist/magic-funcs/optimizers/promptify.js.map +1 -1
- package/dist/magic-funcs/optimizers/ragify.d.ts +16 -0
- package/dist/magic-funcs/optimizers/ragify.d.ts.map +1 -1
- package/dist/magic-funcs/optimizers/ragify.js +19 -3
- package/dist/magic-funcs/optimizers/ragify.js.map +1 -1
- package/dist/magic-funcs/parsers/extract.d.ts +23 -0
- package/dist/magic-funcs/parsers/extract.d.ts.map +1 -1
- package/dist/magic-funcs/parsers/extract.js +26 -3
- package/dist/magic-funcs/parsers/extract.js.map +1 -1
- package/dist/magic-funcs/parsers/rewrite.d.ts +19 -0
- package/dist/magic-funcs/parsers/rewrite.d.ts.map +1 -1
- package/dist/magic-funcs/parsers/rewrite.js +22 -2
- package/dist/magic-funcs/parsers/rewrite.js.map +1 -1
- package/dist/magic-funcs/parsers/structure.d.ts +22 -0
- package/dist/magic-funcs/parsers/structure.d.ts.map +1 -1
- package/dist/magic-funcs/parsers/structure.js +27 -4
- package/dist/magic-funcs/parsers/structure.js.map +1 -1
- package/dist/magic-funcs/parsers/summarize.d.ts +20 -0
- package/dist/magic-funcs/parsers/summarize.d.ts.map +1 -1
- package/dist/magic-funcs/parsers/summarize.js +23 -3
- package/dist/magic-funcs/parsers/summarize.js.map +1 -1
- package/dist/memory.d.ts +1 -1
- package/dist/modalities/image-gen/generateImages.d.ts +56 -0
- package/dist/modalities/image-gen/generateImages.d.ts.map +1 -0
- package/dist/modalities/image-gen/generateImages.js +66 -0
- package/dist/modalities/image-gen/generateImages.js.map +1 -0
- package/dist/modalities/image-gen/helpers.d.ts +15 -0
- package/dist/modalities/image-gen/helpers.d.ts.map +1 -0
- package/dist/modalities/image-gen/helpers.js +17 -0
- package/dist/modalities/image-gen/helpers.js.map +1 -0
- package/dist/modalities/openrouter.d.ts +11 -0
- package/dist/modalities/openrouter.d.ts.map +1 -0
- package/dist/modalities/openrouter.js +39 -0
- package/dist/modalities/openrouter.js.map +1 -0
- package/dist/modalities/speech/VoiceEngine.d.ts +3 -0
- package/dist/modalities/speech/VoiceEngine.d.ts.map +1 -0
- package/dist/modalities/speech/VoiceEngine.js +3 -0
- package/dist/modalities/speech/VoiceEngine.js.map +1 -0
- package/dist/modalities/speech/stt/helpers.d.ts +3 -0
- package/dist/modalities/speech/stt/helpers.d.ts.map +1 -0
- package/dist/modalities/speech/stt/helpers.js +28 -0
- package/dist/modalities/speech/stt/helpers.js.map +1 -0
- package/dist/modalities/speech/stt/stt.d.ts +113 -0
- package/dist/modalities/speech/stt/stt.d.ts.map +1 -0
- package/dist/modalities/speech/stt/stt.js +177 -0
- package/dist/modalities/speech/stt/stt.js.map +1 -0
- package/dist/modalities/speech/tts/helpers.d.ts +12 -0
- package/dist/modalities/speech/tts/helpers.d.ts.map +1 -0
- package/dist/modalities/speech/tts/helpers.js +10 -0
- package/dist/modalities/speech/tts/helpers.js.map +1 -0
- package/dist/modalities/speech/tts/tts.d.ts +115 -0
- package/dist/modalities/speech/tts/tts.d.ts.map +1 -0
- package/dist/modalities/speech/tts/tts.js +142 -0
- package/dist/modalities/speech/tts/tts.js.map +1 -0
- package/dist/modalities/vision/helpers.d.ts +3 -0
- package/dist/modalities/vision/helpers.d.ts.map +1 -0
- package/dist/modalities/vision/helpers.js +7 -0
- package/dist/modalities/vision/helpers.js.map +1 -0
- package/dist/modalities/vision/vision.d.ts +55 -0
- package/dist/modalities/vision/vision.d.ts.map +1 -0
- package/dist/modalities/vision/vision.js +71 -0
- package/dist/modalities/vision/vision.js.map +1 -0
- package/package.json +2 -2
|
@@ -0,0 +1,142 @@
|
|
|
1
|
+
import { getLLM } from "../../../helpers/llms";
|
|
2
|
+
import { getOpenRouterRuntime } from "../../openrouter";
|
|
3
|
+
import { createTTSPayload } from "./helpers";
|
|
4
|
+
/**
|
|
5
|
+
* Converts text to speech via OpenRouter's dedicated `/audio/speech` endpoint.
|
|
6
|
+
*
|
|
7
|
+
* Internally this function calls OpenRouter directly and returns raw audio bytes.
|
|
8
|
+
* If no `llm` is provided, it creates one with
|
|
9
|
+
* `getLLM({ provider: "openrouter", type: "tts" })` and reads
|
|
10
|
+
* `process.env.OPENROUTER_API_KEY`.
|
|
11
|
+
*
|
|
12
|
+
* Make sure `OPENROUTER_API_KEY` is set in your `.env`.
|
|
13
|
+
*
|
|
14
|
+
* @param params.llm Optional LLM instance from `getLLM(...)`.
|
|
15
|
+
* @param params.text Text that should be synthesized to audio.
|
|
16
|
+
* @param params.model Voice identifier to use for speech generation.
|
|
17
|
+
* @param params.responseFormat Audio response format (`mp3` or `pcm`).
|
|
18
|
+
* @param params.speed Playback speed multiplier (default `1.0`).
|
|
19
|
+
* @returns Promise with `audioBytes` and optional response metadata headers.
|
|
20
|
+
*
|
|
21
|
+
* @example
|
|
22
|
+
* CONFIG:
|
|
23
|
+
* ```ts
|
|
24
|
+
* tts({
|
|
25
|
+
* llm = getLLM({ provider: "openrouter", type: "tts" }),
|
|
26
|
+
* responseFormat = "mp3",
|
|
27
|
+
* speed = 1.0,
|
|
28
|
+
* text,
|
|
29
|
+
* model
|
|
30
|
+
* })
|
|
31
|
+
* ```
|
|
32
|
+
*
|
|
33
|
+
* @example
|
|
34
|
+
* ```ts
|
|
35
|
+
* const speech = await tts({
|
|
36
|
+
* text: "Hallo! Das ist ein Test.",
|
|
37
|
+
* model: "nova",
|
|
38
|
+
* responseFormat: "mp3"
|
|
39
|
+
* });
|
|
40
|
+
*
|
|
41
|
+
* console.log(speech.audioBytes.length);
|
|
42
|
+
* ```
|
|
43
|
+
*/
|
|
44
|
+
export async function tts({ llm = getLLM({ provider: "openrouter", type: "tts" }), responseFormat = "mp3", speed = 1.0, text, model }) {
|
|
45
|
+
const runtime = getOpenRouterRuntime(llm);
|
|
46
|
+
const payload = createTTSPayload({
|
|
47
|
+
model: runtime.model,
|
|
48
|
+
input: text,
|
|
49
|
+
voice: model,
|
|
50
|
+
response_format: responseFormat,
|
|
51
|
+
speed
|
|
52
|
+
});
|
|
53
|
+
const response = await fetch(`${runtime.baseURL}/audio/speech`, {
|
|
54
|
+
method: "POST",
|
|
55
|
+
headers: {
|
|
56
|
+
Authorization: `Bearer ${runtime.apiKey}`,
|
|
57
|
+
"Content-Type": "application/json"
|
|
58
|
+
},
|
|
59
|
+
body: JSON.stringify(payload)
|
|
60
|
+
});
|
|
61
|
+
if (!response.ok) {
|
|
62
|
+
const errorText = await response.text();
|
|
63
|
+
throw new Error(`OpenRouter TTS failed (${response.status}): ${errorText}`);
|
|
64
|
+
}
|
|
65
|
+
const audioBuffer = await response.arrayBuffer();
|
|
66
|
+
return {
|
|
67
|
+
audioBytes: new Uint8Array(audioBuffer),
|
|
68
|
+
contentType: response.headers.get("content-type") ?? undefined,
|
|
69
|
+
generationId: response.headers.get("x-generation-id") ?? undefined
|
|
70
|
+
};
|
|
71
|
+
}
|
|
72
|
+
/**
|
|
73
|
+
* Streams synthesized TTS audio as sequential chunks for phone socket pipelines
|
|
74
|
+
* (for example Twilio/Telnyx websocket media send loops).
|
|
75
|
+
*
|
|
76
|
+
* Internally this helper calls `tts()` once and splits the returned audio into
|
|
77
|
+
* chunked payloads that can be forwarded to a live call socket.
|
|
78
|
+
*
|
|
79
|
+
* @param params.llm Optional LLM instance from `getLLM(...)`.
|
|
80
|
+
* @param params.text Text that should be synthesized to audio.
|
|
81
|
+
* @param params.model Voice identifier for speech generation.
|
|
82
|
+
* @param params.responseFormat Audio response format (`mp3` or `pcm`).
|
|
83
|
+
* @param params.speed Playback speed multiplier.
|
|
84
|
+
* @param params.chunkSizeBytes Size of each emitted audio chunk.
|
|
85
|
+
* @param params.emitBase64 If true, emits base64 chunks; otherwise emits `Uint8Array`.
|
|
86
|
+
* @param params.onChunk Callback that receives each emitted chunk in order.
|
|
87
|
+
* @returns Promise with total emitted chunks and response metadata.
|
|
88
|
+
*
|
|
89
|
+
* @example
|
|
90
|
+
* CONFIG:
|
|
91
|
+
* ```ts
|
|
92
|
+
* streamTTSOverPhoneSocket({
|
|
93
|
+
* llm = getLLM({ provider: "openrouter", type: "tts" }),
|
|
94
|
+
speed = 1.0,
|
|
95
|
+
chunkSizeBytes = 3200,
|
|
96
|
+
emitBase64 = true,
|
|
97
|
+
responseFormat = "mp3",
|
|
98
|
+
text,
|
|
99
|
+
model,
|
|
100
|
+
onChunk
|
|
101
|
+
* })
|
|
102
|
+
* ```
|
|
103
|
+
*
|
|
104
|
+
* @example
|
|
105
|
+
* ```ts
|
|
106
|
+
* await streamTTSOverPhoneSocket({
|
|
107
|
+
model,
|
|
108
|
+
onChunk
|
|
109
|
+
* })
|
|
110
|
+
* ```
|
|
111
|
+
*
|
|
112
|
+
* @example
|
|
113
|
+
* ```ts
|
|
114
|
+
* await streamTTSOverPhoneSocket({
|
|
115
|
+
* text: "Willkommen beim Support.",
|
|
116
|
+
* model: "nova",
|
|
117
|
+
* onChunk: async (chunk) => socket.send(String(chunk))
|
|
118
|
+
* });
|
|
119
|
+
* ```
|
|
120
|
+
*/
|
|
121
|
+
export async function streamTTSOverPhoneSocket({ llm = getLLM({ provider: "openrouter", type: "tts" }), speed = 1.0, chunkSizeBytes = 3200, emitBase64 = true, responseFormat = "mp3", text, model, onChunk }) {
|
|
122
|
+
const result = await tts({
|
|
123
|
+
llm,
|
|
124
|
+
text,
|
|
125
|
+
model,
|
|
126
|
+
responseFormat,
|
|
127
|
+
speed
|
|
128
|
+
});
|
|
129
|
+
let index = 0;
|
|
130
|
+
for (let offset = 0; offset < result.audioBytes.length; offset += chunkSizeBytes) {
|
|
131
|
+
const chunk = result.audioBytes.slice(offset, offset + chunkSizeBytes);
|
|
132
|
+
const outgoing = emitBase64 ? Buffer.from(chunk).toString("base64") : chunk;
|
|
133
|
+
await onChunk(outgoing, index);
|
|
134
|
+
index += 1;
|
|
135
|
+
}
|
|
136
|
+
return {
|
|
137
|
+
totalChunks: index,
|
|
138
|
+
contentType: result.contentType,
|
|
139
|
+
generationId: result.generationId
|
|
140
|
+
};
|
|
141
|
+
}
|
|
142
|
+
//# sourceMappingURL=tts.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"tts.js","sourceRoot":"","sources":["../../../../src/modalities/speech/tts/tts.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,EAAE,MAAM,uBAAuB,CAAA;AAC9C,OAAO,EAAe,oBAAoB,EAAE,MAAM,kBAAkB,CAAA;AACpE,OAAO,EAAE,gBAAgB,EAAyC,MAAM,WAAW,CAAA;AAgBnF;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAuCG;AACH,MAAM,CAAC,KAAK,UAAU,GAAG,CAAC,EACtB,GAAG,GAAG,MAAM,CAAC,EAAE,QAAQ,EAAE,YAAY,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,EACrD,cAAc,GAAG,KAAK,EACtB,KAAK,GAAG,GAAG,EACX,IAAI,EACJ,KAAK,EACI;IACT,MAAM,OAAO,GAAG,oBAAoB,CAAC,GAAG,CAAC,CAAA;IAEzC,MAAM,OAAO,GAAG,gBAAgB,CAAC;QAC7B,KAAK,EAAE,OAAO,CAAC,KAAK;QACpB,KAAK,EAAC,IAAI;QACV,KAAK,EAAC,KAAK;QACX,eAAe,EAAE,cAAc;QAC/B,KAAK;KACR,CAAC,CAAA;IAEF,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,GAAG,OAAO,CAAC,OAAO,eAAe,EAAE;QAC5D,MAAM,EAAE,MAAM;QACd,OAAO,EAAE;YACL,aAAa,EAAE,UAAU,OAAO,CAAC,MAAM,EAAE;YACzC,cAAc,EAAE,kBAAkB;SACrC;QACD,IAAI,EAAE,IAAI,CAAC,SAAS,CAAC,OAAO,CAAC;KAChC,CAAC,CAAA;IAEF,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;QACf,MAAM,SAAS,GAAG,MAAM,QAAQ,CAAC,IAAI,EAAE,CAAA;QACvC,MAAM,IAAI,KAAK,CAAC,0BAA0B,QAAQ,CAAC,MAAM,MAAM,SAAS,EAAE,CAAC,CAAA;IAC/E,CAAC;IAED,MAAM,WAAW,GAAG,MAAM,QAAQ,CAAC,WAAW,EAAE,CAAA;IAChD,OAAO;QACH,UAAU,EAAE,IAAI,UAAU,CAAC,WAAW,CAAC;QACvC,WAAW,EAAE,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,cAAc,CAAC,IAAI,SAAS;QAC9D,YAAY,EAAE,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,iBAAiB,CAAC,IAAI,SAAS;KACrE,CAAA;AACL,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAgDG;AACH,MAAM,CAAC,KAAK,UAAU,wBAAwB,CAAC,EAC3C,GAAG,GAAG,MAAM,CAAC,EAAE,QAAQ,EAAE,YAAY,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,EACrD,KAAK,GAAG,GAAG,EACX,cAAc,GAAG,IAAI,EACrB,UAAU,GAAG,IAAI,EACjB,cAAc,GAAG,KAAK,EACtB,IAAI,EACJ,KAAK,EACL,OAAO,EACmB;IAC1B,MAAM,MAAM,GAAG,MAAM,GAAG,CAAC;QACrB,GAAG;QACH,IAAI;QACJ,KAAK;QACL,cAAc;QACd,KAAK;KACR,CAAC,CAAA;IAEF,IAAI,KAAK,GAAG,CAAC,CAAA;IACb,KAAK,IAAI,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC,UAAU,CAAC,MAAM,EAAE,MAAM,IAAI,cAAc,EAAE,CAAC;QAC/E,MAAM,KAAK,GAAG,MAAM,CAAC,UAAU,CAAC,KAAK,CAAC,MAAM,EAAE,MAAM,GAAG,cAAc,CAAC,CAAA;QACtE,MAAM,QAAQ,GAAG,UAAU,CAAC,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,KAAK,CAAA;QAC3E,MAAM,OAAO,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAA;QAC9B,KAAK,IAAI,CAAC,CAAA;IACd,CAAC;IAED,OAAO;QACH,WAAW,EAAE,KAAK;QAClB,WAAW,EAAE,MAAM,CAAC,WAAW;QAC/B,YAAY,EAAE,MAAM,CAAC,YAAY;KACpC,CAAA;AACL,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"helpers.d.ts","sourceRoot":"","sources":["../../../src/modalities/vision/helpers.ts"],"names":[],"mappings":"AAAA,MAAM,MAAM,gBAAgB,GAAG,MAAM,CAAA;AAErC,wBAAgB,oBAAoB,CAAC,KAAK,EAAE,gBAAgB,GAAG,MAAM,CAMpE"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"helpers.js","sourceRoot":"","sources":["../../../src/modalities/vision/helpers.ts"],"names":[],"mappings":"AAEA,MAAM,UAAU,oBAAoB,CAAC,KAAuB;IACxD,IAAI,KAAK,CAAC,UAAU,CAAC,SAAS,CAAC,IAAI,KAAK,CAAC,UAAU,CAAC,UAAU,CAAC,IAAI,KAAK,CAAC,UAAU,CAAC,aAAa,CAAC,EAAE,CAAC;QACjG,OAAO,KAAK,CAAA;IAChB,CAAC;IAED,OAAO,0BAA0B,KAAK,EAAE,CAAA;AAC5C,CAAC"}
|
|
@@ -0,0 +1,55 @@
|
|
|
1
|
+
import { OpenRouterVisionModel } from "../../helpers/llms";
|
|
2
|
+
import { LLMInstance } from "../openrouter";
|
|
3
|
+
import { VisionImageInput } from "./helpers";
|
|
4
|
+
type VisionOptions = {
|
|
5
|
+
llm?: LLMInstance;
|
|
6
|
+
prompt: string;
|
|
7
|
+
images: VisionImageInput[];
|
|
8
|
+
detail?: "auto" | "low" | "high";
|
|
9
|
+
model?: OpenRouterVisionModel;
|
|
10
|
+
};
|
|
11
|
+
/**
|
|
12
|
+
* Analyzes one or more images with a vision-capable OpenRouter model.
|
|
13
|
+
*
|
|
14
|
+
* Internally this function sends `image_url` content parts through OpenRouter's
|
|
15
|
+
* chat-completions-compatible interface. If no `llm` is provided, it builds one
|
|
16
|
+
* with `getLLM({ provider: "openrouter", type: "vision" })` and reads
|
|
17
|
+
* `process.env.OPENROUTER_API_KEY`.
|
|
18
|
+
*
|
|
19
|
+
* Make sure `OPENROUTER_API_KEY` is set in your `.env`.
|
|
20
|
+
*
|
|
21
|
+
* @param params.llm Optional LLM instance from `getLLM(...)`.
|
|
22
|
+
* @param params.prompt Instruction/question for the model about the provided images.
|
|
23
|
+
* @param params.images Array of image URLs, data URLs, or base64 strings.
|
|
24
|
+
* @param params.detail Vision detail level (`auto`, `low`, `high`).
|
|
25
|
+
* @param params.model Optional model override for this call.
|
|
26
|
+
* @returns Promise with normalized `text` and full provider response in `raw`.
|
|
27
|
+
*
|
|
28
|
+
* @example
|
|
29
|
+
* CONFIG:
|
|
30
|
+
* ```ts
|
|
31
|
+
* vision({
|
|
32
|
+
* llm = getLLM({ provider: "openrouter", type: "vision" }),
|
|
33
|
+
* prompt = "Describe the image in detail.",
|
|
34
|
+
* detail = "auto",
|
|
35
|
+
* images,
|
|
36
|
+
* model
|
|
37
|
+
* })
|
|
38
|
+
* ```
|
|
39
|
+
*
|
|
40
|
+
* @example
|
|
41
|
+
* ```ts
|
|
42
|
+
* const result = await vision({
|
|
43
|
+
* prompt: "What objects do you see?",
|
|
44
|
+
* images: ["https://example.com/cat.jpg"]
|
|
45
|
+
* });
|
|
46
|
+
*
|
|
47
|
+
* console.log(result.text);
|
|
48
|
+
* ```
|
|
49
|
+
*/
|
|
50
|
+
export declare function vision({ llm, prompt, detail, images, model }: VisionOptions): Promise<{
|
|
51
|
+
text: string;
|
|
52
|
+
raw: unknown;
|
|
53
|
+
}>;
|
|
54
|
+
export {};
|
|
55
|
+
//# sourceMappingURL=vision.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"vision.d.ts","sourceRoot":"","sources":["../../../src/modalities/vision/vision.ts"],"names":[],"mappings":"AACA,OAAO,EAAU,qBAAqB,EAAE,MAAM,oBAAoB,CAAA;AAClE,OAAO,EAAE,WAAW,EAAwB,MAAM,eAAe,CAAA;AACjE,OAAO,EAAwB,gBAAgB,EAAE,MAAM,WAAW,CAAA;AAElE,KAAK,aAAa,GAAG;IACjB,GAAG,CAAC,EAAE,WAAW,CAAA;IACjB,MAAM,EAAE,MAAM,CAAA;IACd,MAAM,EAAE,gBAAgB,EAAE,CAAA;IAC1B,MAAM,CAAC,EAAE,MAAM,GAAG,KAAK,GAAG,MAAM,CAAA;IAChC,KAAK,CAAC,EAAE,qBAAqB,CAAA;CAChC,CAAA;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAsCG;AACH,wBAAsB,MAAM,CAAC,EACzB,GAAwD,EACxD,MAAwC,EACxC,MAAe,EACf,MAAM,EACN,KAAK,EACR,EAAE,aAAa,GAAG,OAAO,CAAC;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,GAAG,EAAE,OAAO,CAAA;CAAE,CAAC,CA+BzD"}
|
|
@@ -0,0 +1,71 @@
|
|
|
1
|
+
import { HumanMessage } from "../../imports";
|
|
2
|
+
import { getLLM } from "../../helpers/llms";
|
|
3
|
+
import { getOpenRouterRuntime } from "../openrouter";
|
|
4
|
+
import { normalizeVisionImage } from "./helpers";
|
|
5
|
+
/**
|
|
6
|
+
* Analyzes one or more images with a vision-capable OpenRouter model.
|
|
7
|
+
*
|
|
8
|
+
* Internally this function sends `image_url` content parts through OpenRouter's
|
|
9
|
+
* chat-completions-compatible interface. If no `llm` is provided, it builds one
|
|
10
|
+
* with `getLLM({ provider: "openrouter", type: "vision" })` and reads
|
|
11
|
+
* `process.env.OPENROUTER_API_KEY`.
|
|
12
|
+
*
|
|
13
|
+
* Make sure `OPENROUTER_API_KEY` is set in your `.env`.
|
|
14
|
+
*
|
|
15
|
+
* @param params.llm Optional LLM instance from `getLLM(...)`.
|
|
16
|
+
* @param params.prompt Instruction/question for the model about the provided images.
|
|
17
|
+
* @param params.images Array of image URLs, data URLs, or base64 strings.
|
|
18
|
+
* @param params.detail Vision detail level (`auto`, `low`, `high`).
|
|
19
|
+
* @param params.model Optional model override for this call.
|
|
20
|
+
* @returns Promise with normalized `text` and full provider response in `raw`.
|
|
21
|
+
*
|
|
22
|
+
* @example
|
|
23
|
+
* CONFIG:
|
|
24
|
+
* ```ts
|
|
25
|
+
* vision({
|
|
26
|
+
* llm = getLLM({ provider: "openrouter", type: "vision" }),
|
|
27
|
+
* prompt = "Describe the image in detail.",
|
|
28
|
+
* detail = "auto",
|
|
29
|
+
* images,
|
|
30
|
+
* model
|
|
31
|
+
* })
|
|
32
|
+
* ```
|
|
33
|
+
*
|
|
34
|
+
* @example
|
|
35
|
+
* ```ts
|
|
36
|
+
* const result = await vision({
|
|
37
|
+
* prompt: "What objects do you see?",
|
|
38
|
+
* images: ["https://example.com/cat.jpg"]
|
|
39
|
+
* });
|
|
40
|
+
*
|
|
41
|
+
* console.log(result.text);
|
|
42
|
+
* ```
|
|
43
|
+
*/
|
|
44
|
+
export async function vision({ llm = getLLM({ provider: "openrouter", type: "vision" }), prompt = "Describe the image in detail.", detail = "auto", images, model }) {
|
|
45
|
+
const runtime = getOpenRouterRuntime(llm);
|
|
46
|
+
const content = [
|
|
47
|
+
{ type: "text", text: prompt },
|
|
48
|
+
...images.map((image) => ({
|
|
49
|
+
type: "image_url",
|
|
50
|
+
image_url: {
|
|
51
|
+
url: normalizeVisionImage(image),
|
|
52
|
+
detail
|
|
53
|
+
}
|
|
54
|
+
}))
|
|
55
|
+
];
|
|
56
|
+
const runtimeLLM = model && model !== runtime.model
|
|
57
|
+
? getLLM({
|
|
58
|
+
provider: "openrouter",
|
|
59
|
+
type: "vision",
|
|
60
|
+
model,
|
|
61
|
+
apikey: runtime.apiKey
|
|
62
|
+
})
|
|
63
|
+
: llm;
|
|
64
|
+
const response = await runtimeLLM.invoke([new HumanMessage({ content })]);
|
|
65
|
+
const text = typeof response.content === "string" ? response.content : JSON.stringify(response.content);
|
|
66
|
+
return {
|
|
67
|
+
text,
|
|
68
|
+
raw: response
|
|
69
|
+
};
|
|
70
|
+
}
|
|
71
|
+
//# sourceMappingURL=vision.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"vision.js","sourceRoot":"","sources":["../../../src/modalities/vision/vision.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,eAAe,CAAA;AAC5C,OAAO,EAAE,MAAM,EAAyB,MAAM,oBAAoB,CAAA;AAClE,OAAO,EAAe,oBAAoB,EAAE,MAAM,eAAe,CAAA;AACjE,OAAO,EAAE,oBAAoB,EAAoB,MAAM,WAAW,CAAA;AAUlE;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAsCG;AACH,MAAM,CAAC,KAAK,UAAU,MAAM,CAAC,EACzB,GAAG,GAAG,MAAM,CAAC,EAAE,QAAQ,EAAE,YAAY,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,EACxD,MAAM,GAAG,+BAA+B,EACxC,MAAM,GAAG,MAAM,EACf,MAAM,EACN,KAAK,EACO;IACZ,MAAM,OAAO,GAAG,oBAAoB,CAAC,GAAG,CAAC,CAAA;IAEzC,MAAM,OAAO,GAAG;QACZ,EAAE,IAAI,EAAE,MAAe,EAAE,IAAI,EAAE,MAAM,EAAE;QACvC,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;YACtB,IAAI,EAAE,WAAoB;YAC1B,SAAS,EAAE;gBACP,GAAG,EAAE,oBAAoB,CAAC,KAAK,CAAC;gBAChC,MAAM;aACT;SACJ,CAAC,CAAC;KACN,CAAA;IAED,MAAM,UAAU,GACZ,KAAK,IAAI,KAAK,KAAK,OAAO,CAAC,KAAK;QAC5B,CAAC,CAAC,MAAM,CAAC;YACH,QAAQ,EAAE,YAAY;YACtB,IAAI,EAAE,QAAQ;YACd,KAAK;YACL,MAAM,EAAE,OAAO,CAAC,MAAM;SACzB,CAAC;QACJ,CAAC,CAAC,GAAG,CAAA;IAEb,MAAM,QAAQ,GAAG,MAAM,UAAU,CAAC,MAAM,CAAC,CAAC,IAAI,YAAY,CAAC,EAAE,OAAO,EAAE,CAAC,CAAC,CAAC,CAAA;IACzE,MAAM,IAAI,GAAG,OAAO,QAAQ,CAAC,OAAO,KAAK,QAAQ,CAAC,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAA;IAEvG,OAAO;QACH,IAAI;QACJ,GAAG,EAAE,QAAQ;KAChB,CAAA;AACL,CAAC"}
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@delofarag/ai-utils",
|
|
3
|
-
"version": "1.
|
|
3
|
+
"version": "1.6.0",
|
|
4
4
|
"type": "module",
|
|
5
5
|
"main": "dist/index.js",
|
|
6
6
|
"types": "dist/index.d.ts",
|
|
@@ -34,7 +34,7 @@
|
|
|
34
34
|
"zodios": "^5.1.0"
|
|
35
35
|
},
|
|
36
36
|
"peerDependencies": {
|
|
37
|
-
"@delofarag/supabase-utils": "^1.
|
|
37
|
+
"@delofarag/supabase-utils": "^1.3.10"
|
|
38
38
|
},
|
|
39
39
|
"devDependencies": {
|
|
40
40
|
"@types/node": "^20.0.0",
|