utilitas 2001.1.84 → 2001.1.86

package/lib/alan.mjs

@@ -47,19 +47,19 @@ const [
     OPENAI, GOOGLE, OLLAMA, NOVA, DEEPSEEK_32, MD_CODE, CLOUD_OPUS_45, AUDIO,
     WAV, OPENAI_VOICE, GPT_REASONING_EFFORT, THINK, THINK_STR, THINK_END,
     TOOLS_STR, TOOLS_END, TOOLS, TEXT, OK, FUNC, GPT_52, GPT_51_CODEX,
-    GPT_5_IMAGE, GEMMA_3_27B, ANTHROPIC, ais, MAX_TOOL_RECURSION, LOG, name,
+    GPT_IMAGE_15, GEMMA_3_27B, ANTHROPIC, ais, MAX_TOOL_RECURSION, LOG, name,
     user, system, assistant, JSON_OBJECT, PROMPT_IS_REQUIRED, k, trimTailing,
     trimBeginning, GEMINI_30_PRO_IMAGE, IMAGE, JINA, JINA_DEEPSEARCH,
     SILICONFLOW, SF_DEEPSEEK_32, OPENROUTER_API, OPENROUTER, AUTO, TOOL, ONLINE,
     GEMINI_30_PRO, GEMINI_30_FLASH, IMAGEN_4_ULTRA, VEO_31, IMAGEN_4_UPSCALE,
-    ERROR_GENERATING, GEMINI_25_FLASH_TTS, GEMINI_25_PRO_TTS, wav,
+    ERROR_GENERATING, GEMINI_25_FLASH_TTS, GEMINI_25_PRO_TTS, wav, PNG_EXT,
     GPT_4O_MIMI_TTS, GPT_4O_TRANSCRIBE, INVALID_AUDIO, OGG_EXT, ELLIPSIS,
     TOP_LIMIT, ATTACHMENT, PROCESSING, CURSOR, LN1, LN2, TOP, DEEPSEEK
 ] = [
         'OpenAI', 'Google', 'Ollama', 'nova', 'deepseek-3.2-speciale', '```',
         'claude-opus-4.5', 'audio', 'wav', 'OPENAI_VOICE', 'medium', 'think',
         '<think>', '</think>', '<tools>', '</tools>', 'tools', 'text', 'OK',
-        'function', 'gpt-5.2', 'gpt-5.1-codex', 'gpt-5-image', 'gemma3:27b',
+        'function', 'gpt-5.2', 'gpt-5.1-codex', 'gpt-image-1.5', 'gemma3:27b',
         'Anthropic', [], 30, { log: true }, 'Alan', 'user', { role: 'system' },
         { role: 'assistant' }, 'json_object', 'Prompt is required.',
         x => 1000 * x, x => x.replace(/[\.\s]*$/, ''),
@@ -70,10 +70,10 @@ const [
         'gemini-3-pro-preview', 'gemini-3-flash-preview',
         'imagen-4.0-ultra-generate-001', 'veo-3.1-generate-preview',
         'imagen-4.0-upscale-preview', 'Error generating content.',
-        'gemini-2.5-flash-preview-tts', 'gemini-2.5-pro-tts', 'wav',
-        'gpt-4o-mini-tts', 'gpt-4o-transcribe', 'Invalid audio data.', 'ogg',
-        '...', 3, 'ATTACHMENT', { processing: true }, ' █', '\n\n', '\n\n\n',
-        'top', 'DeepSeek',
+        'gemini-2.5-flash-preview-tts', 'gemini-2.5-pro-preview-tts', 'wav',
+        'png', 'gpt-4o-mini-tts', 'gpt-4o-transcribe', 'Invalid audio data.',
+        'ogg', '...', 3, 'ATTACHMENT', { processing: true }, ' █', '\n\n',
+        '\n\n\n', 'top', 'DeepSeek',
     ];
 
 const [joinL1, joinL2]
@@ -152,13 +152,18 @@ const MODELS = {
         source: GOOGLE, maxInputTokens: 480,
         image: true, defaultProvider: GOOGLE,
     },
+    // [IMAGEN_4_UPSCALE]: {
+    //     source: GOOGLE, maxInputTokens: 0, image: true,
+    //     supportedMimeTypes: [MIME_PNG, MIME_JPEG], defaultProvider: GOOGLE,
+    // },
     [VEO_31]: {
         source: GOOGLE, maxInputTokens: 1024, attachmentTokenCost: 0,
         video: true, vision: true,
         supportedMimeTypes: [MIME_PNG, MIME_JPEG], defaultProvider: GOOGLE,
     },
-    [GPT_5_IMAGE]: {
-        ...OPENAI_RULES, label: 'gpt-image-1', image: true,
+    [GPT_IMAGE_15]: {
+        ...OPENAI_RULES, image: true, defaultProvider: OPENAI,
+        supportedMimeTypes: [MIME_PNG, MIME_JPEG, MIME_GIF, MIME_WEBP],
     },
     // models with code capabilities
     [GPT_51_CODEX]: { ...OPENAI_RULES },
@@ -996,11 +1001,23 @@ const promptGoogle = async (aiId, prompt, options = {}) => {
     const target_model = options?.model || model.name;
     const M = MODELS[target_model];
     prompt = ensureString(prompt, { trim: true });
+    // M.name === IMAGEN_4_ULTRA ||
     assertPrompt(prompt);
     M.tts && (prompt = `${options?.prompt || TTS_PROMPT}: ${prompt}`);
     prompt = trimText(prompt, { limit: M.maxInputTokens });
     if (M?.image) {
-        var resp = await client.models.generateImages({
+        var resp;
+        // if (M.name === IMAGEN_4_ULTRA) {
+        //     resp = await client.models.upscaleImage({
+        //         model: M.name, prompt, config: mergeAtoB(options?.config, {
+        //             numberOfImages: options?.n || 4, sampleImageSize: '2K',
+        //             includeRaiReason: true,
+        //             // "1:1" (default), "3:4", "4:3", "9:16", and "16:9"
+        //             aspectRatio: '16:9', personGeneration: 'allow_adult',
+        //         }),
+        //     });
+        // } else {
+        resp = await client.models.generateImages({
             model: M.name, prompt, config: mergeAtoB(options?.config, {
                 numberOfImages: options?.n || 4, sampleImageSize: '2K',
                 includeRaiReason: true,
@@ -1008,6 +1025,7 @@ const promptGoogle = async (aiId, prompt, options = {}) => {
                 aspectRatio: '16:9', personGeneration: 'allow_adult',
             }),
         });
+        // }
         var generated = resp?.generatedImages;
         assert(!resp?.error && generated?.filter(
             x => !x.raiFilteredReason
@@ -1113,7 +1131,40 @@ const promptOpenAI = async (aiId, prompt, options = {}) => {
     const target_model = options?.model || model.name;
     const M = MODELS[target_model];
     prompt = ensureString(prompt, { trim: true });
-    if (M?.audio) {
+    if (M?.image) {
+        assertPrompt(prompt);
+        prompt = trimText(prompt, { limit: M.maxInputTokens });
+        var resp = await client.images.generate({
+            model: M.name, moderation: 'low', output_format: 'png',
+            prompt, quality: 'high', size: '1536x1024',
+            // n = 1, // 1 ~ 10
+            // @TODO: 支持 streaming image generation
+            // https://platform.openai.com/docs/guides/image-generation?api=responses
+            // partial_images: 3, stream: true,
+            // @TODO: 支持 image editing {}
+            ...options?.params || {},
+        });
+        if (!options?.raw) {
+            // print(resp);
+            resp = {
+                text: '', images: await Promise.all(resp.data.map(
+                    async (x, i) => {
+                        const buffer = await convert(
+                            x.b64_json, { input: BASE64 }
+                        );
+                        return {
+                            data: await convert(buffer, {
+                                input: BUFFER, suffix: PNG_EXT, ...options || {}
+                            }),
+                            mimeType: (await getMime(
+                                buffer, `${i}.${resp.output_format}`
+                            ))?.mime || MIME_PNG,
+                        }
+                    })
+                ), model: packModelId([provider, M.source, M.name]),
+            };
+        }
+    } else if (M?.audio) {
         assertPrompt(prompt);
         const ins_prompt = options?.prompt || `${TTS_PROMPT}.`;
         prompt = trimText(prompt, {
@@ -1506,7 +1557,7 @@ export {
     GEMINI_30_FLASH,
     GEMINI_25_PRO_TTS,
     GEMINI_30_PRO_IMAGE,
-    GPT_5_IMAGE,
+    GPT_IMAGE_15,
     GPT_52,
     IMAGEN_4_ULTRA,
     INSTRUCTIONS,

package/lib/manifest.mjs

@@ -1,7 +1,7 @@
 const manifest = {
     "name": "utilitas",
     "description": "Just another common utility for JavaScript.",
-    "version": "2001.1.84",
+    "version": "2001.1.86",
     "private": false,
     "homepage": "https://github.com/Leask/utilitas",
     "main": "index.mjs",

package/package.json

@@ -1,7 +1,7 @@
 {
     "name": "utilitas",
     "description": "Just another common utility for JavaScript.",
-    "version": "2001.1.84",
+    "version": "2001.1.86",
     "private": false,
     "homepage": "https://github.com/Leask/utilitas",
     "main": "index.mjs",