[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8},"actionData","errors","root",{"_108":109,"_110":111},"routes/embed.x.$handle.$name.$",{"_9":10,"_11":12,"_13":14,"_15":16,"_17":18,"_19":20,"_21":-5},"handle","mdevslabs","name","mai","path","images.ts","project",{"_22":45,"_11":12,"_33":46,"_47":-5,"_48":42,"_49":50,"_51":50,"_52":53,"_54":55,"_56":57,"_58":59,"_60":32,"_61":-5,"_62":-5,"_63":64,"_65":66,"_67":68,"_69":70,"_71":72,"_73":-5,"_74":75,"_76":75,"_77":78,"_79":78,"_80":78,"_81":75,"_82":75,"_83":84,"_85":86,"_87":88,"_89":42},"file",{"_22":23,"_11":14,"_24":25,"_26":-5,"_27":28,"_29":30,"_31":32,"_33":34,"_35":36,"_37":38,"_39":40,"_41":42,"_43":-5,"_13":44},"permalink","https://www.val.town/x/mdevslabs/mai/code/images.ts","extension","id","01b444a4-db1c-486c-9c4c-3d111a9ed347","fileId","d2e42ca4-9fd5-11f1-866a-1607ee4eb77e","parentId","branchId","3345cb6c-eeb2-4b61-ab03-8e0abc2dcd73","branchVersion",120,"latest",true,"createdAt",["D",1788164141214],"version",8,"fileType","script","content","import type { Hono } from \"npm:hono@4\";\nimport {\n extractTierFromApiKey,\n extractToken,\n getDb,\n getTierDailyImageLimit,\n getTierImageRequestCost,\n getUserQuotaBoost,\n isPaidTier,\n verifyToken,\n} from \"./config.ts\";\n\nexport interface ImageModelItem {\n capabilities?: {\n async_task?: boolean;\n image_to_image?: boolean;\n inpainting?: boolean;\n text_to_image?: boolean;\n upscaling?: boolean;\n variation?: boolean;\n };\n created: number;\n description: string;\n endpoints?: string[];\n features: string[];\n form:\n | \"openai-standard\"\n | \"midjourney-pipeline\"\n | \"image-edit\"\n | \"image-upscale\";\n id: string;\n name: string;\n object: \"model\";\n owned_by: string;\n parent?: string | null;\n permission?: Array<{\n allow_create_engine: boolean;\n allow_fine_tuning: boolean;\n allow_logprobs: boolean;\n allow_sampling: boolean;\n allow_search_indices: boolean;\n allow_view: boolean;\n created: number;\n group: string | null;\n id: string;\n is_blocking: boolean;\n object: \"model_permission\";\n organization: string;\n }>;\n root?: string;\n}\n\nexport function getCometApiKey(): string {\n if (typeof Deno !== \"undefined\" && Deno.env) {\n return Deno.env.get(\"COMET_API_KEY\") || \"\";\n }\n if (typeof process !== \"undefined\" && process.env) {\n return process.env.COMET_API_KEY || \"\";\n }\n return \"\";\n}\n\nexport function normalizeImageSrc(url?: string | null): string {\n if (!url || typeof url !== \"string\") {\n return \"\";\n }\n const clean = url.trim().replace(/^[\"']|[\"']$/g, \"\");\n if (!clean) {\n return \"\";\n }\n\n if (\n clean.startsWith(\"http://\") ||\n clean.startsWith(\"https://\") ||\n clean.startsWith(\"data:\") ||\n clean.startsWith(\"blob:\") ||\n clean.startsWith(\"/\")\n ) {\n return clean;\n }\n\n let mime = \"image/png\";\n if (clean.startsWith(\"/9j/\")) {\n mime = \"image/jpeg\";\n } else if (clean.startsWith(\"R0lGOD\")) {\n mime = \"image/gif\";\n } else if (clean.startsWith(\"UklGR\")) {\n mime = \"image/webp\";\n } else if (clean.startsWith(\"PHN2Zy\") || clean.startsWith(\"PD94bWw\")) {\n mime = \"image/svg+xml\";\n }\n\n return `data:${mime};base64,${clean}`;\n}\n\n/**\n * Cache mémoire pour les modèles CometAPI (TTL = 60 secondes)\n */\nlet cachedCometImageModels: ImageModelItem[] = [];\nlet lastCometFetchTime = 0;\nconst CACHE_TTL_MS = 60_000;\n\n/**\n * Modèles de repli de haute qualité en cas de défaillance réseau temporaire de CometAPI\n */\nconst FALLBACK_IMAGE_MODELS: ImageModelItem[] = [\n {\n capabilities: {\n async_task: true,\n image_to_image: false,\n inpainting: false,\n text_to_image: true,\n },\n created: 1722470400,\n description:\n \"Modèle ultra-rapide en 4 étapes par Black Forest Labs (Text-to-Image).\",\n endpoints: [\"/v1/images/generations\"],\n features: [\"text-to-image\"],\n form: \"openai-standard\",\n id: \"black-forest-labs/flux-schnell\",\n name: \"FLUX.1 Schnell\",\n object: \"model\",\n owned_by: \"black-forest-labs\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: true,\n inpainting: true,\n text_to_image: true,\n },\n created: 1722470400,\n description:\n \"Modèle phare de synthèse photoréaliste et artistique par Black Forest Labs (Text-to-Image / Image-to-Image).\",\n endpoints: [\"/v1/images/generations\", \"/v1/images/edits\"],\n features: [\"text-to-image\", \"image-to-image\"],\n form: \"openai-standard\",\n id: \"black-forest-labs/flux-dev\",\n name: \"FLUX.1 Dev\",\n object: \"model\",\n owned_by: \"black-forest-labs\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: false,\n inpainting: false,\n text_to_image: true,\n },\n created: 1727827200,\n description:\n \"Sommet de la qualité visuelle, cohérence typographique et détails avancés par Black Forest Labs.\",\n endpoints: [\"/v1/images/generations\"],\n features: [\"text-to-image\"],\n form: \"openai-standard\",\n id: \"black-forest-labs/flux-pro\",\n name: \"FLUX 1.1 Pro\",\n object: \"model\",\n owned_by: \"black-forest-labs\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: false,\n inpainting: false,\n text_to_image: true,\n },\n created: 1730419200,\n description:\n \"Version Ultra haute résolution (jusqu'à 4K) avec photoréalisme extrême et détails de peau par Black Forest Labs.\",\n endpoints: [\"/v1/images/generations\"],\n features: [\"text-to-image\"],\n form: \"openai-standard\",\n id: \"black-forest-labs/flux-1.1-pro-ultra\",\n name: \"FLUX 1.1 Pro Ultra\",\n object: \"model\",\n owned_by: \"black-forest-labs\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: true,\n inpainting: true,\n text_to_image: true,\n },\n created: 1729641600,\n description:\n \"Modèle phare de 8 milliards de paramètres de Stability AI pour une variété stylistique maximale et une typographie nette.\",\n endpoints: [\"/v1/images/generations\", \"/v1/images/edits\"],\n features: [\"text-to-image\", \"image-to-image\"],\n form: \"openai-standard\",\n id: \"stabilityai/stable-diffusion-3.5-large\",\n name: \"Stable Diffusion 3.5 Large\",\n object: \"model\",\n owned_by: \"stabilityai\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: false,\n inpainting: false,\n text_to_image: true,\n },\n created: 1729641600,\n description:\n \"Version accélérée en 4 étapes de Stable Diffusion 3.5 Large par Stability AI.\",\n endpoints: [\"/v1/images/generations\"],\n features: [\"text-to-image\"],\n form: \"openai-standard\",\n id: \"stabilityai/stable-diffusion-3.5-large-turbo\",\n name: \"Stable Diffusion 3.5 Large Turbo\",\n object: \"model\",\n owned_by: \"stabilityai\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: true,\n inpainting: true,\n text_to_image: true,\n },\n created: 1690848000,\n description:\n \"Modèle SDXL 1.0 haute résolution de Stability AI pour la création artistique et le rendu réaliste.\",\n endpoints: [\"/v1/images/generations\", \"/v1/images/edits\"],\n features: [\"text-to-image\", \"image-to-image\"],\n form: \"openai-standard\",\n id: \"stabilityai/sdxl\",\n name: \"Stable Diffusion XL 1.0\",\n object: \"model\",\n owned_by: \"stabilityai\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: true,\n inpainting: true,\n text_to_image: true,\n upscaling: true,\n variation: true,\n },\n created: 1718064000,\n description:\n \"Génération stylisée haut de gamme avec esthétique cinématique et compréhension sémantique de pointe via Midjourney v6.\",\n endpoints: [\n \"/v1/images/generations\",\n \"/mj/submit/imagine\",\n \"/mj/submit/action\",\n \"/mj/task/:id/fetch\",\n ],\n features: [\"text-to-image\", \"image-to-image\", \"upscale\", \"variation\"],\n form: \"midjourney-pipeline\",\n id: \"midjourney/v6\",\n name: \"Midjourney v6\",\n object: \"model\",\n owned_by: \"midjourney\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: true,\n inpainting: true,\n text_to_image: true,\n upscaling: true,\n variation: true,\n },\n created: 1722384000,\n description:\n \"Dernière itération du moteur Midjourney v6.1 avec cohérence accrue des mains, textures et détails fins.\",\n endpoints: [\n \"/v1/images/generations\",\n \"/mj/submit/imagine\",\n \"/mj/submit/action\",\n \"/mj/task/:id/fetch\",\n ],\n features: [\"text-to-image\", \"image-to-image\", \"upscale\", \"variation\"],\n form: \"midjourney-pipeline\",\n id: \"midjourney/v6.1\",\n name: \"Midjourney v6.1\",\n object: \"model\",\n owned_by: \"midjourney\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: true,\n inpainting: false,\n text_to_image: true,\n },\n created: 1730419200,\n description:\n \"Génération vectorielle et matricielle spécialisée dans les logos, icônes, illustrations et design graphique.\",\n endpoints: [\"/v1/images/generations\"],\n features: [\"text-to-image\", \"image-to-image\"],\n form: \"openai-standard\",\n id: \"recraft-ai/recraft-v3\",\n name: \"Recraft V3\",\n object: \"model\",\n owned_by: \"recraft-ai\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: false,\n inpainting: false,\n text_to_image: true,\n },\n created: 1724284800,\n description:\n \"Modèle phare de rendu de texte dans l'image et composition graphique par Ideogram AI.\",\n endpoints: [\"/v1/images/generations\"],\n features: [\"text-to-image\"],\n form: \"openai-standard\",\n id: \"ideogram-ai/ideogram-v2\",\n name: \"Ideogram V2\",\n object: \"model\",\n owned_by: \"ideogram-ai\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: false,\n inpainting: false,\n text_to_image: true,\n },\n created: 1724284800,\n description:\n \"Version ultra-rapide d'Ideogram V2 optimisée pour les flux de production en temps réel.\",\n endpoints: [\"/v1/images/generations\"],\n features: [\"text-to-image\"],\n form: \"openai-standard\",\n id: \"ideogram-ai/ideogram-v2-turbo\",\n name: \"Ideogram V2 Turbo\",\n object: \"model\",\n owned_by: \"ideogram-ai\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: false,\n inpainting: false,\n text_to_image: true,\n },\n created: 1698796800,\n description:\n \"Modèle DALL-E 3 d'OpenAI pour la synthèse d'images haute fidélité avec reformulation automatique des prompts.\",\n endpoints: [\"/v1/images/generations\"],\n features: [\"text-to-image\"],\n form: \"openai-standard\",\n id: \"dall-e-3\",\n name: \"DALL-E 3\",\n object: \"model\",\n owned_by: \"openai\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: true,\n inpainting: true,\n text_to_image: true,\n },\n created: 1667260800,\n description:\n \"Modèle classique DALL-E 2 d'OpenAI pour la génération et l'édition rapide d'images.\",\n endpoints: [\n \"/v1/images/generations\",\n \"/v1/images/edits\",\n \"/v1/images/variations\",\n ],\n features: [\"text-to-image\", \"image-to-image\"],\n form: \"openai-standard\",\n id: \"dall-e-2\",\n name: \"DALL-E 2\",\n object: \"model\",\n owned_by: \"openai\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: false,\n inpainting: false,\n text_to_image: true,\n },\n created: 1723680000,\n description:\n \"Modèle de synthèse photoréaliste et typographique de Google DeepMind Imagen 3.\",\n endpoints: [\"/v1/images/generations\"],\n features: [\"text-to-image\"],\n form: \"openai-standard\",\n id: \"google/imagen-3\",\n name: \"Google Imagen 3\",\n object: \"model\",\n owned_by: \"google\",\n },\n {\n capabilities: {\n async_task: true,\n image_to_image: false,\n inpainting: false,\n text_to_image: true,\n },\n created: 1730000000,\n description: \"Modèle ultra-rapide de génération photoréaliste par Luma AI.\",\n endpoints: [\"/v1/images/generations\"],\n features: [\"text-to-image\"],\n form: \"openai-standard\",\n id: \"luma/photon\",\n name: \"Luma Photon\",\n object: \"model\",\n owned_by: \"luma\",\n },\n];\n\n/**\n * Déterminer la forme technique d'un modèle d'image CometAPI\n */\nfunction detectModelForm(\n modelId: string,\n _features?: string[],\n): ImageModelItem[\"form\"] {\n const lower = modelId.toLowerCase();\n if (\n lower.includes(\"midjourney\") || lower.includes(\"mj-\") ||\n lower.startsWith(\"mj/\")\n ) {\n return \"midjourney-pipeline\";\n }\n if (lower.includes(\"upscale\")) {\n return \"image-upscale\";\n }\n if (lower.includes(\"edit\") || lower.includes(\"inpaint\")) {\n return \"image-edit\";\n }\n return \"openai-standard\";\n}\n\n/**\n * Obtenir la liste ordonnée des identifiants candidats pour un modèle donné sur CometAPI\n */\nfunction getCometCandidateModelIds(requestedModel: string): string[] {\n const lower = (requestedModel || \"\").toLowerCase().trim();\n\n if (\n lower.includes(\"schnell\") || lower === \"flux\" ||\n lower.includes(\"flux-1-schnell\") || lower.includes(\"flux-schnell\")\n ) {\n return [\n \"flux-schnell\",\n \"black-forest-labs/flux-schnell\",\n \"flux-1-schnell\",\n \"black-forest-labs/flux-1-schnell\",\n \"dall-e-3\",\n ];\n }\n if (lower.includes(\"flux\") && lower.includes(\"dev\")) {\n return [\n \"flux-dev\",\n \"black-forest-labs/flux-dev\",\n \"flux-1-dev\",\n \"black-forest-labs/flux-1-dev\",\n \"dall-e-3\",\n ];\n }\n if (\n lower.includes(\"flux\") && (lower.includes(\"pro\") || lower.includes(\"ultra\"))\n ) {\n return [\n \"flux-pro\",\n \"black-forest-labs/flux-pro\",\n \"flux-1.1-pro\",\n \"black-forest-labs/flux-1.1-pro\",\n \"dall-e-3\",\n ];\n }\n if (\n lower.includes(\"dall-e-3\") || lower.includes(\"dalle-3\") ||\n lower === \"dall-e\"\n ) {\n return [\"dall-e-3\", \"openai/dall-e-3\"];\n }\n if (lower.includes(\"dall-e-2\") || lower.includes(\"dalle-2\")) {\n return [\"dall-e-2\", \"openai/dall-e-2\", \"dall-e-3\"];\n }\n if (lower.includes(\"midjourney\") || lower.includes(\"mj\")) {\n return [\n \"midjourney/v6.1\",\n \"midjourney/v6\",\n \"midjourney\",\n \"mj-v6\",\n \"dall-e-3\",\n ];\n }\n if (\n lower.includes(\"diffusion\") || lower.includes(\"sd3\") ||\n lower.includes(\"sd-3\")\n ) {\n return [\n \"stable-diffusion-3.5-large\",\n \"stabilityai/stable-diffusion-3.5-large\",\n \"sdxl\",\n \"stabilityai/sdxl\",\n \"dall-e-3\",\n ];\n }\n if (lower.includes(\"sdxl\")) {\n return [\"sdxl\", \"stabilityai/sdxl\", \"dall-e-3\"];\n }\n if (lower.includes(\"recraft\")) {\n return [\"recraft-v3\", \"recraft-ai/recraft-v3\", \"dall-e-3\"];\n }\n if (lower.includes(\"ideogram\")) {\n return [\n \"ideogram-v2\",\n \"ideogram-ai/ideogram-v2\",\n \"ideogram-v2-turbo\",\n \"dall-e-3\",\n ];\n }\n if (lower.includes(\"imagen\")) {\n return [\"google/imagen-3\", \"imagen-3\", \"dall-e-3\"];\n }\n if (lower.includes(\"photon\")) {\n return [\"luma/photon\", \"photon\", \"dall-e-3\"];\n }\n\n return [requestedModel, \"flux-schnell\", \"dall-e-3\"];\n}\n\n/**\n * Résolution des alias de modèles\n */\nfunction resolveImageModel(requestedModel: string): string {\n if (!requestedModel) return \"flux-schnell\";\n const candidates = getCometCandidateModelIds(requestedModel);\n return candidates[0] || requestedModel;\n}\n\n/**\n * Formatage d'un modèle d'image selon le schéma OpenAI Model Specification\n */\nfunction formatOpenAiImageModel(m: any): ImageModelItem {\n const modelId = m.id || \"image-model\";\n const org = m.owned_by || modelId.split(\"/\")[0] || \"cometapi\";\n const created = Number(m.created) ||\n Math.floor(Date.now() / 1000) - 86_400 * 30;\n const features = m.features ||\n m.supported_features ||\n (modelId.toLowerCase().includes(\"diffusion\") ||\n modelId.toLowerCase().includes(\"midjourney\")\n ? [\"text-to-image\", \"image-to-image\"]\n : [\"text-to-image\"]);\n\n const form = m.form || detectModelForm(modelId, features);\n const endpoints = m.endpoints || (\n form === \"midjourney-pipeline\"\n ? [\n \"/v1/images/generations\",\n \"/mj/submit/imagine\",\n \"/mj/submit/action\",\n \"/mj/task/:id/fetch\",\n ]\n : form === \"image-edit\"\n ? [\"/v1/images/edits\", \"/v1/images/generations\"]\n : form === \"image-upscale\"\n ? [\"/v1/images/upscale\"]\n : [\"/v1/images/generations\", \"/v1/images/edits\"]\n );\n\n return {\n capabilities: {\n async_task: true,\n image_to_image: features.includes(\"image-to-image\") ||\n form === \"image-edit\",\n inpainting: features.includes(\"inpainting\") ||\n features.includes(\"image-to-image\"),\n text_to_image: true,\n upscaling: features.includes(\"upscale\") || form === \"image-upscale\" ||\n form === \"midjourney-pipeline\",\n variation: features.includes(\"variation\") ||\n form === \"midjourney-pipeline\",\n },\n created,\n description: m.description ||\n `Modèle de génération d'images haute fidélité ${m.name || modelId}.`,\n endpoints,\n features,\n form,\n id: modelId,\n name: m.name || modelId,\n object: \"model\",\n owned_by: org,\n parent: null,\n permission: [\n {\n allow_create_engine: false,\n allow_fine_tuning: false,\n allow_logprobs: true,\n allow_sampling: true,\n allow_search_indices: false,\n allow_view: true,\n created,\n group: null,\n id: `modelperm-${modelId.replace(/[^a-zA-Z0-9]/g, \"-\")}`,\n is_blocking: false,\n object: \"model_permission\",\n organization: \"*\",\n },\n ],\n root: modelId,\n };\n}\n\n/**\n * Récupération dynamique et filtrage en temps réel des modèles d'images depuis CometAPI\n */\nasync function fetchLiveCometImageModels(): Promise {\n const now = Date.now();\n if (\n cachedCometImageModels.length > 0 && now - lastCometFetchTime < CACHE_TTL_MS\n ) {\n return cachedCometImageModels;\n }\n\n const cometApiKey = getCometApiKey();\n if (!cometApiKey) {\n return FALLBACK_IMAGE_MODELS;\n }\n\n try {\n const res = await fetch(\"https://api.cometapi.com/v1/models\", {\n headers: {\n Authorization: `Bearer ${cometApiKey}`,\n \"Content-Type\": \"application/json\",\n Accept: \"application/json\",\n },\n });\n\n if (!res.ok) {\n console.warn(`[CometAPI] /v1/models réponse non-OK (${res.status})`);\n return cachedCometImageModels.length > 0\n ? cachedCometImageModels\n : FALLBACK_IMAGE_MODELS;\n }\n\n const json = await res.json();\n const rawModels: any[] = json.data || json.models || [];\n\n // Filtrage dynamique multi-formes des modèles d'image\n const imageModels = rawModels.filter((m: any) => {\n if (!m || !m.id) return false;\n const mType = String(\n m.model_type ||\n m.type ||\n m.architecture?.modality ||\n m.object ||\n \"\",\n ).toLowerCase();\n const outputModalities = (m.architecture?.output_modalities || []).map((\n o: string,\n ) => String(o).toLowerCase());\n const features = (m.features || m.supported_features || []).map((\n f: string,\n ) => String(f).toLowerCase());\n const idLower = String(m.id).toLowerCase();\n\n const isImageModality = mType.includes(\"image\") ||\n outputModalities.includes(\"image\") ||\n mType.endsWith(\"->image\") ||\n mType.includes(\"text->image\");\n\n const hasImageFeature = features.includes(\"text-to-image\") ||\n features.includes(\"image-to-image\") ||\n features.includes(\"inpainting\") ||\n features.includes(\"image\");\n\n const matchesImageKeyword = idLower.includes(\"flux\") ||\n idLower.includes(\"diffusion\") ||\n idLower.includes(\"dall-e\") ||\n idLower.includes(\"midjourney\") ||\n idLower.includes(\"mj-\") ||\n idLower.includes(\"recraft\") ||\n idLower.includes(\"ideogram\") ||\n idLower.includes(\"imagen\") ||\n idLower.includes(\"photon\") ||\n idLower.includes(\"kling\") ||\n idLower.includes(\"kolors\") ||\n idLower.includes(\"sdxl\") ||\n idLower.includes(\"stable-diffusion\") ||\n idLower.startsWith(\"sd-\") ||\n idLower.includes(\"qwen-image\") ||\n idLower.includes(\"gpt-image\");\n\n return isImageModality || hasImageFeature || matchesImageKeyword;\n });\n\n if (imageModels.length > 0) {\n const formatted = imageModels.map(formatOpenAiImageModel);\n cachedCometImageModels = formatted;\n lastCometFetchTime = now;\n return formatted;\n }\n\n return cachedCometImageModels.length > 0\n ? cachedCometImageModels\n : FALLBACK_IMAGE_MODELS;\n } catch (err) {\n console.error(\"[CometAPI] Erreur de récupération des modèles:\", err);\n return cachedCometImageModels.length > 0\n ? cachedCometImageModels\n : FALLBACK_IMAGE_MODELS;\n }\n}\n\n/**\n * Exécuteur robuste multi-stratégie pour la génération d'images CometAPI\n */\nasync function callCometImageGeneration(params: {\n apiKey: string;\n isEdit?: boolean;\n isVariation?: boolean;\n mask?: string;\n model: string;\n n: number;\n negativePrompt?: string;\n prompt: string;\n quality?: string;\n responseFormat?: string;\n seed?: number;\n size: string;\n sourceImage?: string;\n style?: string;\n}): Promise<\n {\n data: Array<{ b64_json?: string; revised_prompt?: string; url?: string }>;\n usedModel: string;\n }\n> {\n const {\n apiKey,\n prompt,\n size,\n n,\n responseFormat,\n quality,\n style,\n negativePrompt,\n seed,\n sourceImage,\n mask,\n isEdit,\n isVariation,\n } = params;\n\n const candidateModels = getCometCandidateModelIds(params.model);\n let lastErrorText = \"\";\n\n for (const candidate of candidateModels) {\n // 1. Construire le payload de base minimal\n const basePayload: Record = {\n model: candidate,\n prompt,\n size: size || \"1024x1024\",\n };\n\n if (n && n > 1) {\n basePayload.n = Math.min(n, 4);\n }\n if (responseFormat && responseFormat === \"b64_json\") {\n basePayload.response_format = \"b64_json\";\n }\n if (quality) {\n basePayload.quality = quality;\n }\n if (style) {\n basePayload.style = style;\n }\n if (negativePrompt) {\n basePayload.negative_prompt = negativePrompt;\n }\n if (seed !== undefined) {\n basePayload.seed = seed;\n }\n if (sourceImage) {\n basePayload.image = sourceImage;\n }\n if (mask) {\n basePayload.mask = mask;\n }\n\n const endpointUrl = isEdit\n ? \"https://api.cometapi.com/v1/images/edits\"\n : isVariation\n ? \"https://api.cometapi.com/v1/images/variations\"\n : \"https://api.cometapi.com/v1/images/generations\";\n\n let currentPayload = { ...basePayload };\n\n for (let attempt = 0; attempt < 2; attempt++) {\n try {\n const res = await fetch(endpointUrl, {\n body: JSON.stringify(currentPayload),\n headers: {\n Authorization: `Bearer ${apiKey}`,\n \"Content-Type\": \"application/json\",\n Accept: \"application/json\",\n },\n method: \"POST\",\n });\n\n if (res.ok) {\n const json = await res.json().catch(() => ({}));\n const rawList = json.data || json.images || [];\n if (Array.isArray(rawList) && rawList.length > 0) {\n const formatted = rawList.map((img: any) => {\n const rawUrl = img.url || \"\";\n const b64 = img.b64_json || \"\";\n const resolved = normalizeImageSrc(rawUrl || b64);\n return {\n b64_json: b64 ||\n (responseFormat === \"b64_json\" && resolved.startsWith(\"data:\")\n ? resolved.split(\",\")[1]\n : undefined),\n revised_prompt: img.revised_prompt || prompt,\n url: resolved || rawUrl,\n };\n });\n return { data: formatted, usedModel: candidate };\n }\n }\n\n lastErrorText = await res.text().catch(() => \"\");\n console.warn(\n `[CometAPI] Modèle '${candidate}' tentative ${\n attempt + 1\n } échouée (${res.status}):`,\n lastErrorText,\n );\n\n // Si l'erreur indique un paramètre inconnu (ex: 'style', 'quality', 'seed', etc.)\n if (\n lastErrorText.includes(\"Unknown parameter\") ||\n lastErrorText.includes(\"unknown_parameter\")\n ) {\n // Supprimer les paramètres optionnels et re-tenter immédiatement\n currentPayload = {\n model: candidate,\n prompt,\n size: size || \"1024x1024\",\n };\n continue;\n }\n } catch (fetchErr: unknown) {\n lastErrorText = fetchErr instanceof Error\n ? fetchErr.message\n : \"Erreur réseau\";\n console.warn(\n `[CometAPI] Exception fetch '${candidate}':`,\n lastErrorText,\n );\n }\n break;\n }\n }\n\n // 2. Fallback alternatif : essayer via /v1/chat/completions sur CometAPI\n try {\n const chatRes = await fetch(\n \"https://api.cometapi.com/v1/chat/completions\",\n {\n body: JSON.stringify({\n messages: [\n {\n content: `Generate this image: ${prompt}`,\n role: \"user\",\n },\n ],\n model: \"dall-e-3\",\n }),\n headers: {\n Authorization: `Bearer ${apiKey}`,\n \"Content-Type\": \"application/json\",\n Accept: \"application/json\",\n },\n method: \"POST\",\n },\n );\n\n if (chatRes.ok) {\n const chatJson = await chatRes.json();\n const content = chatJson.choices?.[0]?.message?.content || \"\";\n const match = content.match(/!\\[.*?\\]\$(https?:\\/\\/[^\\s\$]+)\\)/) ||\n content.match(/(https?:\\/\\/[^\\s\\)]+\\.(?:png|jpg|jpeg|webp))/i);\n if (match && match[1]) {\n return {\n data: [\n {\n revised_prompt: prompt,\n url: normalizeImageSrc(match[1]),\n },\n ],\n usedModel: \"chat-image-fallback\",\n };\n }\n }\n } catch {}\n\n throw new Error(\n `Erreur fournisseur Comet API: ${\n lastErrorText ||\n \"Impossible de générer l'image après plusieurs tentatives\"\n }`,\n );\n}\n\n/**\n * Exécuteur et pont asynchrone pour les tâches Midjourney sur CometAPI\n */\nasync function executeMidjourneyImagineBridge(\n prompt: string,\n cometApiKey: string,\n options?: {\n aspectRatio?: string;\n base64Array?: string[];\n notifyHook?: string;\n },\n): Promise<\n { failReason?: string; imageUrl: string; progress?: string; taskId?: string }\n> {\n let finalPrompt = prompt.trim();\n if (options?.aspectRatio && !finalPrompt.includes(\"--ar\")) {\n finalPrompt = `${finalPrompt} --ar ${options.aspectRatio}`;\n }\n\n const payload: Record = {\n prompt: finalPrompt,\n };\n if (options?.base64Array && options.base64Array.length > 0) {\n payload.base64Array = options.base64Array;\n }\n if (options?.notifyHook) {\n payload.notifyHook = options.notifyHook;\n }\n\n try {\n const submitRes = await fetch(\n \"https://api.cometapi.com/mj/submit/imagine\",\n {\n body: JSON.stringify(payload),\n headers: {\n Authorization: `Bearer ${cometApiKey}`,\n \"Content-Type\": \"application/json\",\n Accept: \"application/json\",\n },\n method: \"POST\",\n },\n );\n\n if (submitRes.ok) {\n const submitJson = await submitRes.json();\n const taskId = submitJson.result || submitJson.task_id || submitJson.id;\n\n if (taskId) {\n // Polling actif jusqu'à succès (max 60s)\n const maxAttempts = 35;\n const pollIntervalMs = 1800;\n\n for (let attempt = 0; attempt < maxAttempts; attempt++) {\n await new Promise((resolve) => setTimeout(resolve, pollIntervalMs));\n\n try {\n const fetchRes = await fetch(\n `https://api.cometapi.com/mj/task/${taskId}/fetch`,\n {\n headers: {\n Authorization: `Bearer ${cometApiKey}`,\n \"Content-Type\": \"application/json\",\n Accept: \"application/json\",\n },\n },\n );\n\n if (!fetchRes.ok) continue;\n\n const taskJson = await fetchRes.json();\n const status = String(taskJson.status || \"\").toUpperCase();\n\n if (status === \"SUCCESS\") {\n const imgUrl = taskJson.imageUrl || taskJson.cdnImage ||\n taskJson.discordImage || taskJson.image_url;\n if (imgUrl) {\n return {\n imageUrl: imgUrl,\n progress: \"100%\",\n taskId,\n };\n }\n } else if (status === \"FAILURE\" || status === \"FAILED\") {\n break;\n }\n } catch {}\n }\n }\n }\n } catch (err) {\n console.warn(\"[CometAPI] Midjourney imagine bridge error:\", err);\n }\n\n // Si le bridge MJ natif échoue ou timeout, fallback transparent sur les générateurs d'images CometAPI\n const fallbackRes = await callCometImageGeneration({\n apiKey: cometApiKey,\n model: \"dall-e-3\",\n n: 1,\n prompt,\n size: \"1024x1024\",\n });\n\n return {\n imageUrl: fallbackRes.data[0]?.url || \"\",\n progress: \"100%\",\n taskId: \"fallback-standard\",\n };\n}\n\nexport function registerImageRoutes(app: Hono) {\n // ─────────────────────────────────────────────\n // GET /v1/models/images, /models/images & /v1/images/models (OpenAI Compatible)\n // ─────────────────────────────────────────────\n const handleGetImageModels = async (c: any) => {\n const userPlan = c.get(\"userPlan\");\n const shouldFilterFreeOnly = !isPaidTier(userPlan);\n\n try {\n let models = await fetchLiveCometImageModels();\n\n if (shouldFilterFreeOnly) {\n models = models.filter((m) => {\n const idLower = (m.id || \"\").toLowerCase();\n return idLower.includes(\"flux\") || idLower.includes(\"schnell\") ||\n idLower.includes(\"free\");\n });\n }\n\n return c.json({ data: models, object: \"list\" });\n } catch (_err) {\n let fallback = FALLBACK_IMAGE_MODELS;\n if (shouldFilterFreeOnly) {\n fallback = fallback.filter((m) => m.id.toLowerCase().includes(\"flux\"));\n }\n return c.json({\n data: fallback.map(formatOpenAiImageModel),\n object: \"list\",\n });\n }\n };\n\n app.get(\"/v1/models/images\", handleGetImageModels);\n app.get(\"/models/images\", handleGetImageModels);\n app.get(\"/v1/images/models\", handleGetImageModels);\n app.get(\"/images/models\", handleGetImageModels);\n\n // ─────────────────────────────────────────────\n // GET /v1/models/images/:id & /models/images/:id (OpenAI Model Detail)\n // ─────────────────────────────────────────────\n const handleGetSingleImageModel = async (c: any) => {\n const rawId = c.req.param(\"id\") || \"\";\n const resolvedId = resolveImageModel(rawId);\n\n const liveModels = await fetchLiveCometImageModels();\n const found = liveModels.find(\n (m) =>\n m.id.toLowerCase() === rawId.toLowerCase() ||\n m.id.toLowerCase() === resolvedId.toLowerCase() ||\n m.name.toLowerCase() === rawId.toLowerCase(),\n ) ||\n FALLBACK_IMAGE_MODELS.find(\n (m) =>\n m.id.toLowerCase() === rawId.toLowerCase() ||\n m.id.toLowerCase() === resolvedId.toLowerCase() ||\n m.name.toLowerCase() === rawId.toLowerCase(),\n ) || {\n created: Math.floor(Date.now() / 1000),\n description: `Modèle d'image ${rawId}.`,\n features: [\"text-to-image\"],\n form: detectModelForm(resolvedId, [\"text-to-image\"]),\n id: resolvedId,\n name: rawId,\n object: \"model\" as const,\n owned_by: resolvedId.split(\"/\")[0] || \"cometapi\",\n };\n\n return c.json(formatOpenAiImageModel(found));\n };\n\n app.get(\"/v1/models/images/:id\", handleGetSingleImageModel);\n app.get(\"/models/images/:id\", handleGetSingleImageModel);\n\n // ─────────────────────────────────────────────\n // GET /v1/images/usage & /images/usage\n // ─────────────────────────────────────────────\n const handleGetImageUsage = async (c: any) => {\n try {\n const token = extractToken(c.req.raw);\n let userId = c.get(\"userId\");\n let userPlan = c.get(\"userPlan\") || \"Free\";\n\n if (token) {\n try {\n const payload = await verifyToken(token);\n userId = (payload.sub as string) || userId;\n userPlan = (payload.tier as string) || userPlan;\n } catch {\n // Token non JWT ignoré\n }\n }\n\n const sql = getDb();\n\n if (token) {\n try {\n const keyRows = await sql`\n SELECT k.user_id, u.tier, u.email, u.username\n FROM mprojects_api_keys k\n LEFT JOIN users u ON k.user_id = u.id::text OR k.user_id = u.username OR k.user_id = u.email\n WHERE k.api_key = ${token}::text\n LIMIT 1\n `;\n if (keyRows.length > 0) {\n userId = keyRows[0].user_id;\n userPlan = keyRows[0].tier || userPlan;\n }\n } catch {\n // Erreur DB ignorée\n }\n }\n\n if (!userId) {\n return c.json({ error: \"Non authentifié.\" }, 401);\n }\n\n const [uRows, countRows] = await Promise.all([\n sql`SELECT tier FROM users WHERE id::text = ${userId}::text OR username = ${userId}::text OR email = ${userId}::text LIMIT 1`,\n sql`\n SELECT COALESCE(SUM(images_generated::numeric), 0) as images_generated \n FROM mprojects_daily_image_usage \n WHERE (\n user_id = ${userId}::text \n OR user_id IN (SELECT id::text FROM users WHERE id::text = ${userId}::text OR email = ${userId}::text OR username = ${userId}::text)\n OR user_id IN (SELECT email FROM users WHERE id::text = ${userId}::text OR email = ${userId}::text OR username = ${userId}::text)\n OR user_id IN (SELECT username FROM users WHERE id::text = ${userId}::text OR email = ${userId}::text OR username = ${userId}::text)\n ) AND usage_date = CURRENT_DATE \n LIMIT 1\n `.catch(() => []),\n ]);\n\n const keyTier = extractTierFromApiKey(token);\n const effectiveTier = keyTier || uRows[0]?.tier || userPlan || \"Free\";\n const usedToday = Number(countRows[0]?.images_generated || 0);\n const imageBoost = await getUserQuotaBoost(sql, userId, \"images\");\n const dailyLimit = getTierDailyImageLimit(effectiveTier) + imageBoost;\n\n const now = new Date();\n const tomorrowMidnight = new Date(\n Date.UTC(\n now.getUTCFullYear(),\n now.getUTCMonth(),\n now.getUTCDate() + 1,\n 0,\n 0,\n 0,\n ),\n );\n\n return c.json({\n dailyLimit,\n plan: effectiveTier,\n resetAt: tomorrowMidnight.toISOString(),\n usedToday,\n userId,\n });\n } catch (err: unknown) {\n const msg = err instanceof Error ? err.message : \"Erreur inconnue\";\n return c.json(\n {\n details: msg,\n error: \"Erreur lors de la récupération de l'usage image.\",\n },\n 500,\n );\n }\n };\n\n app.get(\"/v1/images/usage\", handleGetImageUsage);\n app.get(\"/images/usage\", handleGetImageUsage);\n\n // ─────────────────────────────────────────────\n // GET /v1/images/history & /images/history\n // ─────────────────────────────────────────────\n const handleGetImageHistory = async (c: any) => {\n try {\n const token = extractToken(c.req.raw);\n let userId = c.get(\"userId\");\n\n if (token) {\n try {\n const payload = await verifyToken(token);\n userId = payload.sub as string;\n } catch {\n // Token invalide ignoré\n }\n }\n\n if (!userId) {\n return c.json({ error: \"Non authentifié.\" }, 401);\n }\n\n const sql = getDb();\n const history = await sql`\n SELECT id, title, pinned, model, prompt, negative_prompt, width, height, image_url, status, created_at\n FROM mprojects_image_generations\n WHERE user_id = ${userId}::text\n ORDER BY pinned DESC, created_at DESC\n LIMIT 50\n `.catch(() => []);\n\n const formattedHistory = history.map((item: any) => ({\n ...item,\n image_url: normalizeImageSrc(item.image_url),\n pinned: Boolean(item.pinned),\n title: item.title || null,\n }));\n\n return c.json({ data: formattedHistory, success: true });\n } catch (err: unknown) {\n const msg = err instanceof Error ? err.message : \"Erreur inconnue\";\n return c.json(\n { details: msg, error: \"Erreur historique images.\" },\n 500,\n );\n }\n };\n\n const handleUpdateImageHistory = async (c: any) => {\n try {\n const token = extractToken(c.req.raw);\n let userId = c.get(\"userId\");\n\n if (token) {\n try {\n const payload = await verifyToken(token);\n userId = payload.sub as string;\n } catch {}\n }\n\n if (!userId) {\n return c.json({ error: \"Non authentifié.\" }, 401);\n }\n\n const id = c.req.param(\"id\") || c.req.query(\"id\");\n if (!id) {\n return c.json({ error: \"ID manquant.\" }, 400);\n }\n\n const body = await c.req.json().catch(() => ({}));\n const sql = getDb();\n const sets: string[] = [];\n const values: any[] = [];\n let idx = 1;\n\n if (body.title !== undefined) {\n sets.push(`title = $${idx++}`);\n values.push(\n body.title ? String(body.title).trim().slice(0, 200) : null,\n );\n }\n if (body.pinned !== undefined) {\n sets.push(`pinned = $${idx++}`);\n values.push(Boolean(body.pinned));\n }\n\n if (sets.length === 0) {\n return c.json({ success: true });\n }\n\n values.push(id, userId);\n const query = `UPDATE mprojects_image_generations SET ${\n sets.join(\", \")\n } WHERE id::text = $${idx++}::text AND user_id = $${idx}::text RETURNING id, title, pinned`;\n const result = await sql.unsafe(query, values);\n\n return c.json({ data: result[0] || null, success: true });\n } catch (err: unknown) {\n const msg = err instanceof Error ? err.message : \"Erreur inconnue\";\n return c.json({ details: msg, error: \"Erreur mise à jour image.\" }, 500);\n }\n };\n\n app.get(\"/v1/images/history\", handleGetImageHistory);\n app.get(\"/images/history\", handleGetImageHistory);\n app.patch(\"/v1/images/history/:id\", handleUpdateImageHistory);\n app.patch(\"/v1/images/history\", handleUpdateImageHistory);\n app.patch(\"/images/history/:id\", handleUpdateImageHistory);\n app.patch(\"/images/history\", handleUpdateImageHistory);\n\n // ─────────────────────────────────────────────\n // POST /v1/images/generations & Formes Multi-Modèles (OpenAI, Midjourney, Google SDK)\n // ─────────────────────────────────────────────\n const handleImageGeneration = async (c: any) => {\n try {\n const token = extractToken(c.req.raw);\n const authHeader = c.req.header(\"Authorization\");\n const headerApiKey = c.req.header(\"x-api-key\") ||\n c.req.header(\"X-API-Key\") ||\n c.req.header(\"x-goog-api-key\") ||\n c.req.header(\"X-Goog-Api-Key\");\n const apiKey = authHeader?.startsWith(\"Bearer \")\n ? authHeader.slice(7)\n : headerApiKey || null;\n\n let userId = c.get(\"userId\");\n let userPlan = c.get(\"userPlan\") || \"Free\";\n\n if (token) {\n try {\n const payload = await verifyToken(token);\n userId = payload.sub as string;\n userPlan = (payload.tier as string) || userPlan;\n } catch {\n // Token non JWT ignoré\n }\n }\n\n if (!userId) {\n return c.json(\n {\n error: {\n code: \"invalid_api_key\",\n message: \"Non authentifié. Clé API ou token requis.\",\n param: null,\n type: \"authentication_error\",\n },\n },\n 401,\n );\n }\n\n const body = await c.req.json().catch(() => ({}));\n const reqPath = c.req.path || \"\";\n const isGoogleFormat = reqPath.includes(\":generateImages\") ||\n reqPath.includes(\":predict\") ||\n Boolean(body.instances || body.numberOfImages || body.aspectRatio);\n\n const rawPrompt = body.prompt ||\n body.instances?.[0]?.prompt ||\n body.parameters?.prompt ||\n \"\";\n const prompt = typeof rawPrompt === \"string\" ? rawPrompt.trim() : \"\";\n\n const pathModel = reqPath\n .replace(/^\\/(v1beta|v1)\\/models\\//, \"\")\n .replace(/:(generateImages|predict).*$/, \"\");\n const requestedModel = body.model || pathModel ||\n \"black-forest-labs/flux-schnell\";\n\n const model = resolveImageModel(requestedModel);\n const modelForm = detectModelForm(model, []);\n\n // Résolution des dimensions et aspect ratio\n let width = 1024;\n let height = 1024;\n const aspectRatioStr = body.aspect_ratio || body.aspectRatio || \"\";\n\n if (aspectRatioStr) {\n const ar = String(aspectRatioStr).trim();\n if (ar === \"16:9\") {\n width = 1280;\n height = 720;\n } else if (ar === \"9:16\") {\n width = 720;\n height = 1280;\n } else if (ar === \"4:3\") {\n width = 1024;\n height = 768;\n } else if (ar === \"3:4\") {\n width = 768;\n height = 1024;\n } else if (ar === \"21:9\") {\n width = 1536;\n height = 640;\n }\n } else if (body.size) {\n const parts = String(body.size).split(\"x\");\n if (parts.length === 2) {\n width = Number.parseInt(parts[0], 10) || 1024;\n height = Number.parseInt(parts[1], 10) || 1024;\n }\n } else {\n if (body.width) width = Number(body.width);\n if (body.height) height = Number(body.height);\n }\n\n const negativePrompt = body.negative_prompt || \"\";\n const n = Math.max(\n 1,\n Math.min(Number(body.n || body.numberOfImages || 1), 4),\n );\n const responseFormat = body.response_format || \"url\";\n const quality = body.quality ? String(body.quality) : undefined;\n const style = body.style ? String(body.style) : undefined;\n\n if (!prompt) {\n return c.json(\n {\n error: {\n code: \"missing_prompt\",\n message:\n \"Le paramètre 'prompt' est obligatoire pour la génération d'images.\",\n param: \"prompt\",\n type: \"invalid_request_error\",\n },\n },\n 400,\n );\n }\n\n const sql = getDb();\n\n // Vérifier le forfait utilisateur réel\n const uRows = await sql`\n SELECT tier FROM users \n WHERE id::text = ${userId}::text OR username = ${userId}::text \n LIMIT 1\n `;\n const keyTier = extractTierFromApiKey(token);\n const effectiveTier = keyTier || uRows[0]?.tier || userPlan || \"Free\";\n const requestQuota = (c as any).get?.(\"requestQuota\") as\n | {\n apiKey: string | null;\n limit: number;\n remaining: number;\n used: number;\n }\n | undefined;\n\n if (!isPaidTier(effectiveTier)) {\n return c.json(\n {\n error: {\n code: \"image_generation_tier_restricted\",\n message:\n `La génération d'images via l'API est réservée aux abonnements payants (Plus, Pro, Max). Les clés API issues d'un compte Free ne sont pas autorisées à effectuer de requêtes de génération d'images.`,\n param: null,\n type: \"permission_error\",\n },\n },\n 403,\n );\n }\n\n const imageBoost = await getUserQuotaBoost(sql, userId, \"images\");\n const dailyLimit = getTierDailyImageLimit(effectiveTier) + imageBoost;\n const usageRows = await sql`\n SELECT images_generated \n FROM mprojects_daily_image_usage \n WHERE user_id = ${userId}::text AND usage_date = CURRENT_DATE \n LIMIT 1\n `;\n const currentDailyUsage = Number(usageRows[0]?.images_generated || 0);\n\n if (currentDailyUsage + n > dailyLimit) {\n return c.json(\n {\n error: {\n code: \"daily_image_quota_exceeded\",\n limit: dailyLimit,\n message:\n `Votre quota journalier de génération d'images est épuisé (${currentDailyUsage}/${dailyLimit} par jour pour le forfait ${effectiveTier}, ${n} image(s) demandée(s)). Réinitialisation automatique à minuit UTC.`,\n requested: n,\n type: \"quota_error\",\n used: currentDailyUsage,\n },\n },\n 429,\n );\n }\n\n // Le crédit API débité est le coût du tier multiplié par le nombre d'images demandées:\n // il doit rester assez pour la requête AVANT d'appeler le fournisseur.\n const totalRequestCost = getTierImageRequestCost(effectiveTier) * n;\n if (\n requestQuota &&\n requestQuota.used + totalRequestCost > requestQuota.limit\n ) {\n return c.json(\n {\n error: {\n code: \"insufficient_api_requests\",\n dailyLimit,\n limit: requestQuota.limit,\n message:\n `Requête refusée : ${totalRequestCost} crédits API sont nécessaires pour générer ${n} image(s) avec le forfait ${effectiveTier}, il n'en reste que ${requestQuota.remaining} sur ${requestQuota.limit} cette semaine.`,\n remaining: requestQuota.remaining,\n required: totalRequestCost,\n type: \"quota_error\",\n used: requestQuota.used,\n },\n },\n 429,\n );\n }\n\n // Appel de Comet API avec exécution multi-stratégie & gestion des erreurs\n const cometApiKey = getCometApiKey();\n let generatedImageUrl = \"\";\n let cometResultData: Array<\n { b64_json?: string; revised_prompt?: string; url?: string }\n > = [];\n let finalUsedModel = model;\n\n if (cometApiKey) {\n if (\n modelForm === \"midjourney-pipeline\" && !reqPath.includes(\"/edits\")\n ) {\n const mjRes = await executeMidjourneyImagineBridge(\n prompt,\n cometApiKey,\n {\n aspectRatio: aspectRatioStr ||\n (width !== height ? `${width}:${height}` : undefined),\n base64Array: body.image ? [body.image] : undefined,\n },\n );\n generatedImageUrl = normalizeImageSrc(mjRes.imageUrl);\n cometResultData = [{\n revised_prompt: prompt,\n url: generatedImageUrl,\n }];\n } else {\n const genResult = await callCometImageGeneration({\n apiKey: cometApiKey,\n isEdit: reqPath.includes(\"/edits\") ||\n Boolean(body.image && body.mask),\n isVariation: reqPath.includes(\"/variations\"),\n mask: body.mask,\n model,\n n,\n negativePrompt,\n prompt,\n quality,\n responseFormat,\n seed: body.seed,\n size: `${width}x${height}`,\n sourceImage: body.image || body.image_url,\n style,\n });\n\n cometResultData = genResult.data;\n finalUsedModel = genResult.usedModel;\n if (cometResultData.length > 0) {\n generatedImageUrl = cometResultData[0].url || \"\";\n }\n }\n } else {\n generatedImageUrl = `https://picsum.photos/seed/${\n encodeURIComponent(prompt.slice(0, 20))\n }/${width}/${height}`;\n cometResultData = [{ revised_prompt: prompt, url: generatedImageUrl }];\n }\n\n // Incrémentation du quota journalier (1 par image générée)\n await sql`\n INSERT INTO mprojects_daily_image_usage (user_id, usage_date, images_generated, updated_at)\n VALUES (${userId}::text, CURRENT_DATE, ${n}, NOW())\n ON CONFLICT (user_id, usage_date)\n DO UPDATE SET \n images_generated = mprojects_daily_image_usage.images_generated + ${n},\n updated_at = NOW()\n `;\n\n // Enregistrement dans l'historique\n await sql`\n INSERT INTO mprojects_image_generations (\n user_id, api_key, model, prompt, negative_prompt, width, height, image_url, status\n ) VALUES (\n ${userId}::text,\n ${apiKey || null},\n ${finalUsedModel}::text,\n ${prompt}::text,\n ${negativePrompt || null},\n ${width}::integer,\n ${height}::integer,\n ${generatedImageUrl}::text,\n 'completed'\n )\n `;\n\n // Débit du compteur de requêtes de la clé API (cout x n verifie avant l'appel)\n const debitKey = requestQuota?.apiKey || apiKey;\n if (debitKey) {\n await sql`\n UPDATE mprojects_api_keys\n SET request_count = request_count + ${totalRequestCost}, last_used_at = NOW()\n WHERE api_key = ${debitKey}::text\n `;\n (c as any).set?.(\"quotaDebitedByHandler\", true);\n }\n\n // Log de requête\n try {\n await sql`\n INSERT INTO mprojects_api_logs (api_key, endpoint, method, status_code, latency_ms, created_at)\n VALUES (${apiKey || \"anonymous\"}::text, ${\n reqPath || \"/v1/images/generations\"\n }::text, 'POST', 200, 1500, NOW())\n `;\n } catch {}\n\n // Format Google GenAI\n if (isGoogleFormat) {\n return c.json({\n generatedImages: cometResultData.map((img) => ({\n image: {\n imageBytes: img.b64_json || \"\",\n mimeType: \"image/png\",\n uri: img.url || generatedImageUrl,\n },\n })),\n });\n }\n\n // Format OpenAI Standard\n return c.json({\n created: Math.floor(Date.now() / 1000),\n data: cometResultData,\n model: finalUsedModel,\n usage: {\n daily_limit: dailyLimit,\n daily_used: currentDailyUsage + n,\n plan: effectiveTier,\n request_cost: totalRequestCost,\n requests_counted: totalRequestCost,\n },\n });\n } catch (err: unknown) {\n const msg = err instanceof Error ? err.message : \"Erreur inconnue\";\n console.error(\"[ImagesAPI] Erreur serveur génération:\", err);\n return c.json(\n {\n error: {\n code: \"image_generation_failed\",\n details: msg,\n message: `Erreur lors de la génération d'image : ${msg}`,\n type: \"api_error\",\n },\n },\n 500,\n );\n }\n };\n\n // ─────────────────────────────────────────────\n // POST & GET Routes Midjourney Dédiées (`/mj/*` et `/v1/mj/*`)\n // ─────────────────────────────────────────────\n const handleMidjourneyForward = async (c: any) => {\n const cometApiKey = getCometApiKey();\n if (!cometApiKey) {\n return c.json({\n error: \"Clé COMET_API_KEY non configurée sur le serveur.\",\n }, 500);\n }\n\n const subPath = c.req.path.replace(/^\\/v1\\/mj\\//, \"\").replace(\n /^\\/mj\\//,\n \"\",\n );\n const targetUrl = `https://api.cometapi.com/mj/${subPath}`;\n const method = c.req.method;\n\n try {\n const init: RequestInit = {\n headers: {\n Authorization: `Bearer ${cometApiKey}`,\n \"Content-Type\": \"application/json\",\n Accept: \"application/json\",\n },\n method,\n };\n\n if (method === \"POST\" || method === \"PUT\" || method === \"PATCH\") {\n const reqBody = await c.req.json().catch(() => ({}));\n init.body = JSON.stringify(reqBody);\n }\n\n const res = await fetch(targetUrl, init);\n const data = await res.json().catch(() => ({}));\n return c.json(data, res.status as any);\n } catch (err: unknown) {\n const msg = err instanceof Error ? err.message : \"Erreur inconnue\";\n return c.json(\n { error: \"Erreur proxy Midjourney CometAPI\", details: msg },\n 500,\n );\n }\n };\n\n app.post(\"/mj/submit/imagine\", handleMidjourneyForward);\n app.post(\"/v1/mj/submit/imagine\", handleMidjourneyForward);\n app.get(\"/mj/task/:id/fetch\", handleMidjourneyForward);\n app.get(\"/v1/mj/task/:id/fetch\", handleMidjourneyForward);\n app.post(\"/mj/submit/action\", handleMidjourneyForward);\n app.post(\"/v1/mj/submit/action\", handleMidjourneyForward);\n app.post(\"/mj/submit/describe\", handleMidjourneyForward);\n app.post(\"/v1/mj/submit/describe\", handleMidjourneyForward);\n app.post(\"/mj/submit/blend\", handleMidjourneyForward);\n app.post(\"/v1/mj/submit/blend\", handleMidjourneyForward);\n app.post(\"/mj/submit/change\", handleMidjourneyForward);\n app.post(\"/v1/mj/submit/change\", handleMidjourneyForward);\n app.post(\"/mj/submit/modal\", handleMidjourneyForward);\n app.post(\"/v1/mj/submit/modal\", handleMidjourneyForward);\n app.post(\"/mj/submit/shorten\", handleMidjourneyForward);\n app.post(\"/v1/mj/submit/shorten\", handleMidjourneyForward);\n app.post(\"/mj/submit/swap-face\", handleMidjourneyForward);\n app.post(\"/v1/mj/submit/swap-face\", handleMidjourneyForward);\n\n // ─────────────────────────────────────────────\n // POST & GET Routes Upscale, Edits, Variations & Tasks\n // ─────────────────────────────────────────────\n const handleImageUpscale = async (c: any) => {\n const cometApiKey = getCometApiKey();\n if (!cometApiKey) {\n return c.json({ error: \"Clé COMET_API_KEY non configurée.\" }, 500);\n }\n try {\n const body = await c.req.json().catch(() => ({}));\n const res = await fetch(\"https://api.cometapi.com/v1/images/upscale\", {\n body: JSON.stringify(body),\n headers: {\n Authorization: `Bearer ${cometApiKey}`,\n \"Content-Type\": \"application/json\",\n Accept: \"application/json\",\n },\n method: \"POST\",\n });\n const data = await res.json().catch(() => ({}));\n return c.json(data, res.status as any);\n } catch (err: unknown) {\n const msg = err instanceof Error ? err.message : \"Erreur inconnue\";\n return c.json({ error: \"Erreur upscale image\", details: msg }, 500);\n }\n };\n\n app.post(\"/v1/images/upscale\", handleImageUpscale);\n app.post(\"/images/upscale\", handleImageUpscale);\n\n const handleFetchImageTask = async (c: any) => {\n const taskId = c.req.param(\"id\");\n const cometApiKey = getCometApiKey();\n if (!cometApiKey) {\n return c.json({ error: \"Clé COMET_API_KEY non configurée.\" }, 500);\n }\n try {\n const res = await fetch(\n `https://api.cometapi.com/mj/task/${taskId}/fetch`,\n {\n headers: {\n Authorization: `Bearer ${cometApiKey}`,\n \"Content-Type\": \"application/json\",\n Accept: \"application/json\",\n },\n },\n );\n const data = await res.json().catch(() => ({}));\n return c.json(data, res.status as any);\n } catch (err: unknown) {\n const msg = err instanceof Error ? err.message : \"Erreur inconnue\";\n return c.json({ error: \"Erreur récupération tâche\", details: msg }, 500);\n }\n };\n\n app.get(\"/v1/images/tasks/:id\", handleFetchImageTask);\n app.get(\"/images/tasks/:id\", handleFetchImageTask);\n app.get(\"/v1/tasks/:id\", handleFetchImageTask);\n app.get(\"/tasks/:id\", handleFetchImageTask);\n\n // Routes OpenAI & Anthropic SDK\n app.post(\"/v1/images/generations\", handleImageGeneration);\n app.post(\"/images/generations\", handleImageGeneration);\n app.post(\"/v1/images\", handleImageGeneration);\n app.post(\"/images\", handleImageGeneration);\n app.post(\"/v1/images/edits\", handleImageGeneration);\n app.post(\"/images/edits\", handleImageGeneration);\n app.post(\"/v1/images/variations\", handleImageGeneration);\n app.post(\"/images/variations\", handleImageGeneration);\n\n // Routes Google GenAI / Gemini / Vertex SDK\n app.post(\"/v1beta/models/*:generateImages\", handleImageGeneration);\n app.post(\"/v1/models/*:generateImages\", handleImageGeneration);\n app.post(\"/v1beta/models/*:predict\", handleImageGeneration);\n app.post(\"/v1/models/*:predict\", handleImageGeneration);\n}","deleted",false,"label",[14],"200f8efb-f4ef-4117-ae26-ee737dbffb76",["D",1785339526786],"deletedAt","legacyEnvVars","privacy","public","httpPrivacy","description","Avec l'API d'mAI, récupérez des données sur les projets !","imageUrl","https://imagedelivery.net/iHX6Ovru0O7AjmyT5yZRoA/cdf5d0c3-f220-4dec-712e-87455a037c00/square512?exp=1789538399&sig=cd777e882462fce47bac2003be9a9fe1267d4472bdb46131ff02f8b27f89b2ac","userId","7ed13655-83c2-4934-a981-01578302df33","pinnedFileId","968028a2-8b63-11f1-b718-1607ee4eb77e","pinPreview","pinMigrated","screenshotVersion","likeCount",0,"orgId","019f99eb-d944-7332-99f8-2fbb8b1e7011","tursoName","200f8efbf4ef4117ae26ee737dbffb76","tursoHostname","200f8efbf4ef4117ae26ee737dbffb76-val-town.aws-us-east-2.turso.io","tursoMeasuredAt",["D",1789474235934],"pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_9":10,"_107":103},"org",{"_9":10,"_100":101,"_102":103,"_56":57,"_104":105,"_106":42},"branch",{"_35":90,"_11":91,"_22":28,"_92":45,"_93":-5,"_94":-5,"_95":-5,"_96":-5,"_97":-5,"_98":12,"_99":10},"isOwned",364,"main","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","displayName","mProjects","avatar","https://imagedelivery.net/iHX6Ovru0O7AjmyT5yZRoA/ec72623f-c02d-461a-9b1c-c2f416214400/square512?exp=1789538399&sig=b90c7895d89b2c8a9d976c298024ed06d954dd92e3be0b0905971776951516a4","tier","free","canRestrictVals","profileImageUrl","ctx",{"_83":-5,"_141":142,"_143":144,"_145":146,"_147":42,"_148":42,"_149":42,"_150":32,"_151":-5,"_152":-7,"_153":-7,"_154":64,"_155":42,"_156":42,"_157":-5,"_158":42,"_159":42,"_160":42,"_161":-5},"ENV",{"_112":113,"_114":115,"_116":91,"_117":118,"_119":120,"_121":122,"_123":124,"_125":126,"_127":128,"_129":130,"_131":132,"_133":134,"_135":136,"_137":138,"_139":140},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","e87eb04e707cf64e7e532c8e8ec395f209e17f15","RENDER_GIT_BRANCH","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","CLOUDFLARE_TURNSTILE_SITE_KEY","0x4AAAAAACtaIZJH7s9rBMZ6","GOOGLE_CLIENT_ID","875989696899-jtsqkhlbdv5jrlsk8631p7ipjg3c6o62.apps.googleusercontent.com","STRIPE_PUBLISHABLE_KEY","pk_live_51N2zvWEpZTTLipIkVEwXrUG9AT0yeOOEYxszarZT1jpsZV58Ly6wlg3SltZpHN26CUalOBmUSY7adQRQiSkCZz2500qSXNkf8Y","API_URL","https://api.val.town","orgs",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"]