#!/usr/bin/env python3 """ Phosor AI CLI Client — single-file, stdlib-only. Usage: python3 phosor_client.py [options] python3 phosor_client.py --help Environment: PHOSOR_API_KEY — API key for authentication (required for most commands) PHOSOR_BASE_URL — Base URL override (default: https://phosor.ai) """ import argparse import json import os import re import sys import time import uuid from pathlib import Path from urllib.parse import quote from urllib.request import Request, urlopen from urllib.error import HTTPError, URLError VERSION = "1.3.6" # keep equal to the VERSION file shipped in this package DEFAULT_BASE_URL = "https://phosor.ai" WORKSPACE_DIR = Path.home() / ".openclaw" / "workspace" PENDING_FILE = WORKSPACE_DIR / "phosor-pending.json" # Fields forwarded verbatim to /api/v1/inference/submit. # ONE list, used by both Client.submit() and the CLI dispatch — they used to carry # separate copies and adding a parameter meant remembering all of them. _SUBMIT_FIELDS = ( "width", "height", "num_frames", "frames_per_second", "seed", "negative_prompt", "image_url", "audio_url", "video_url", "lora_id", "lora_scale", "loras", "num_inference_steps", "guidance_scale", "model", "num_images", "strength", "output_format", # MiniMax H3 (duration-based, not frame-based) "duration", "resolution_tier", "aspect_ratio", "end_image_url", "reference_image_urls", "reference_video_urls", "reference_audio_urls", "use_ref_video_audio", # What each reference asset is, in the same order as the *_urls arrays. # Optional but recommended for Ref2VA: uploaded files are stored under # generated ids, so without this the model cannot tell which picture shows # which character named in your prompt. Keep your prompt as you wrote it — # the platform does the binding for you. "reference_image_labels", "reference_video_labels", "reference_audio_labels", # Qwen3-TTS (takes `text`, not `prompt`) "text", "speaker", "language", "temperature", "top_p", "top_k", "repetition_penalty", ) MODELS = { "models": { "minimax/h3/reference-to-video/turbo": { "id": "minimax/h3/reference-to-video/turbo", "name": "MiniMax H3 Reference-to-Video Turbo", "type": "reference-to-video", "billing": "per output second PLUS per reference input", "note": "MiniMax H3 Turbo; output 480p $0.006/s, 768p $0.009/s; first 5 images free, then $0.0025 each; audio free.", }, "wan/v2.2-a14b/text-to-video": { "id": "wan/v2.2-a14b/text-to-video", "name": "Wan 2.2 Text-to-Video 14B (FP8)", "type": "text-to-video", "input": { "prompt": {"type": "string", "required": True, "max_length": 2000}, "negative_prompt": {"type": "string", "required": False, "default": ""}, "width": {"type": "integer", "required": False, "default": 854}, "height": {"type": "integer", "required": False, "default": 480}, "num_frames": {"type": "integer", "required": False, "default": 81}, "frames_per_second": {"type": "integer", "required": False, "default": 16, "min": 4, "max": 30}, "num_inference_steps": {"type": "integer", "required": False, "default": 4, "min": 4, "max": 10}, "guidance_scale": {"type": "number", "required": False, "default": 1.0, "min": 1.0, "max": 3.5}, "seed": {"type": "integer", "required": False, "default": None}, "lora_id": {"type": "string", "required": False}, "loras": {"type": "array", "required": False, "description": "Multiple LoRAs: [{\"lora_id\": \"...\", \"lora_scale\": 1.0}]"}, "lora_scale": {"type": "number", "required": False, "default": 1.0, "min": 0.0, "max": 1.0}, }, "resolutions": { "480p_landscape": {"width": 854, "height": 480, "max_frames": 161}, "480p_portrait": {"width": 480, "height": 854, "max_frames": 161}, "720p_landscape": {"width": 1280, "height": 720, "max_frames": 161}, "720p_portrait": {"width": 720, "height": 1280, "max_frames": 161}, "1080p_landscape": {"width": 1920, "height": 1080, "max_frames": 153}, "1080p_portrait": {"width": 1080, "height": 1920, "max_frames": 153}, }, }, "wan/v2.2-a14b/image-to-video": { "id": "wan/v2.2-a14b/image-to-video", "name": "Wan 2.2 Image-to-Video 14B (FP16)", "type": "image-to-video", "input": { "prompt": {"type": "string", "required": True, "max_length": 2000}, "image_url": {"type": "string", "required": True, "description": "S3 key from upload-image"}, "negative_prompt": {"type": "string", "required": False, "default": ""}, "width": {"type": "integer", "required": False, "default": 854}, "height": {"type": "integer", "required": False, "default": 480}, "num_frames": {"type": "integer", "required": False, "default": 81}, "frames_per_second": {"type": "integer", "required": False, "default": 16, "min": 4, "max": 30}, "num_inference_steps": {"type": "integer", "required": False, "default": 4, "min": 4, "max": 10}, "guidance_scale": {"type": "number", "required": False, "default": 1.0, "min": 1.0, "max": 3.5}, "seed": {"type": "integer", "required": False, "default": None}, "lora_id": {"type": "string", "required": False}, "loras": {"type": "array", "required": False, "description": "Multiple LoRAs: [{\"lora_id\": \"...\", \"lora_scale\": 1.0}]"}, "lora_scale": {"type": "number", "required": False, "default": 1.0, "min": 0.0, "max": 1.0}, }, "resolutions": { "480p_landscape": {"width": 854, "height": 480, "max_frames": 161}, "480p_portrait": {"width": 480, "height": 854, "max_frames": 161}, "720p_landscape": {"width": 1280, "height": 720, "max_frames": 161}, "720p_portrait": {"width": 720, "height": 1280, "max_frames": 161}, "1080p_landscape": {"width": 1920, "height": 1080, "max_frames": 153}, "1080p_portrait": {"width": 1080, "height": 1920, "max_frames": 153}, }, }, "wan/v2.2-a14b/speech-to-video": { "id": "wan/v2.2-a14b/speech-to-video", "name": "Wan 2.2 Speech-to-Video 14B", "type": "speech-to-video", "input": { "prompt": {"type": "string", "required": True, "max_length": 2000}, "image_url": {"type": "string", "required": True, "description": "S3 key from upload-image (reference face image)"}, "audio_url": {"type": "string", "required": True, "description": "URL to audio file (WAV/MP3)"}, "negative_prompt": {"type": "string", "required": False, "default": ""}, "width": {"type": "integer", "required": False, "default": 854}, "height": {"type": "integer", "required": False, "default": 480}, "num_frames": {"type": "integer", "required": False, "default": 81}, "frames_per_second": {"type": "integer", "required": False, "default": 16, "min": 4, "max": 30}, "num_inference_steps": {"type": "integer", "required": False, "default": 4, "min": 4, "max": 10}, "guidance_scale": {"type": "number", "required": False, "default": 1.0, "min": 1.0, "max": 3.5}, "seed": {"type": "integer", "required": False, "default": None}, }, "resolutions": { "480p_landscape": {"width": 854, "height": 480, "max_frames": 161}, "480p_portrait": {"width": 480, "height": 854, "max_frames": 161}, "512p_square": {"width": 512, "height": 512, "max_frames": 161}, "720p_landscape": {"width": 1280, "height": 720, "max_frames": 161}, "720p_portrait": {"width": 720, "height": 1280, "max_frames": 161}, }, }, "wan/v2.2-a14b/animate": { "id": "wan/v2.2-a14b/animate", "name": "Wan 2.2 Animate 14B", "type": "animate", "input": { "prompt": {"type": "string", "required": True, "max_length": 2000}, "image_url": {"type": "string", "required": True, "description": "S3 key from upload-image (reference character image)"}, "video_url": {"type": "string", "required": True, "description": "URL to motion/pose reference video"}, "negative_prompt": {"type": "string", "required": False, "default": ""}, "width": {"type": "integer", "required": False, "default": 854}, "height": {"type": "integer", "required": False, "default": 480}, "num_frames": {"type": "integer", "required": False, "default": 81}, "frames_per_second": {"type": "integer", "required": False, "default": 16, "min": 4, "max": 30}, "num_inference_steps": {"type": "integer", "required": False, "default": 4, "min": 4, "max": 10}, "guidance_scale": {"type": "number", "required": False, "default": 1.0, "min": 1.0, "max": 3.5}, "seed": {"type": "integer", "required": False, "default": None}, }, "resolutions": { "480p_landscape": {"width": 854, "height": 480, "max_frames": 161}, "480p_portrait": {"width": 480, "height": 854, "max_frames": 161}, "512p_square": {"width": 512, "height": 512, "max_frames": 161}, "720p_landscape": {"width": 1280, "height": 720, "max_frames": 161}, "720p_portrait": {"width": 720, "height": 1280, "max_frames": 161}, }, }, "qwen-image/v2512/text-to-image": { "id": "qwen-image/v2512/text-to-image", "name": "Qwen Image 2512 Text-to-Image", "type": "text-to-image", "input": { "prompt": {"type": "string", "required": True, "max_length": 2000}, "negative_prompt": {"type": "string", "required": False, "default": ""}, "width": {"type": "integer", "required": False, "default": 1024}, "height": {"type": "integer", "required": False, "default": 1024}, "num_images": {"type": "integer", "required": False, "default": 1, "min": 1, "max": 4}, "num_inference_steps": {"type": "integer", "required": False, "default": 20, "min": 1, "max": 40}, "guidance_scale": {"type": "number", "required": False, "default": 7.5, "min": 1.0, "max": 20.0}, "seed": {"type": "integer", "required": False, "default": None}, "output_format": {"type": "string", "required": False, "default": "png", "enum": ["png", "jpeg"]}, "lora_id": {"type": "string", "required": False}, "lora_scale": {"type": "number", "required": False, "default": 1.0, "min": 0.0, "max": 1.0}, }, "resolutions": { "512x512": {"width": 512, "height": 512}, "1024x1024": {"width": 1024, "height": 1024}, "1024x768": {"width": 1024, "height": 768}, "768x1024": {"width": 768, "height": 1024}, "1280x768": {"width": 1280, "height": 768}, "768x1280": {"width": 768, "height": 1280}, }, }, "qwen-image/v2512/text-to-image/lora": { "id": "qwen-image/v2512/text-to-image/lora", "name": "Qwen Image 2512 Text-to-Image with LoRA", "type": "text-to-image", "input": { "prompt": {"type": "string", "required": True, "max_length": 2000}, "negative_prompt": {"type": "string", "required": False, "default": ""}, "width": {"type": "integer", "required": False, "default": 1024}, "height": {"type": "integer", "required": False, "default": 1024}, "num_images": {"type": "integer", "required": False, "default": 1, "min": 1, "max": 4}, "num_inference_steps": {"type": "integer", "required": False, "default": 20, "min": 1, "max": 40}, "guidance_scale": {"type": "number", "required": False, "default": 7.5, "min": 1.0, "max": 20.0}, "seed": {"type": "integer", "required": False, "default": None}, "output_format": {"type": "string", "required": False, "default": "png", "enum": ["png", "jpeg"]}, "lora_id": {"type": "string", "required": True}, "lora_scale": {"type": "number", "required": False, "default": 1.0, "min": 0.0, "max": 1.0}, }, "resolutions": { "512x512": {"width": 512, "height": 512}, "1024x1024": {"width": 1024, "height": 1024}, "1024x768": {"width": 1024, "height": 768}, "768x1024": {"width": 768, "height": 1024}, "1280x768": {"width": 1280, "height": 768}, "768x1280": {"width": 768, "height": 1280}, }, }, "z-image/turbo/text-to-image": { "id": "z-image/turbo/text-to-image", "name": "Z-Image Turbo Text-to-Image", "type": "text-to-image", "input": { "prompt": {"type": "string", "required": True, "max_length": 2000}, "negative_prompt": {"type": "string", "required": False, "default": ""}, "width": {"type": "integer", "required": False, "default": 1024}, "height": {"type": "integer", "required": False, "default": 1024}, "num_images": {"type": "integer", "required": False, "default": 1, "min": 1, "max": 4}, "num_inference_steps": {"type": "integer", "required": False, "default": 4, "min": 1, "max": 4}, "guidance_scale": {"type": "number", "required": False, "default": 1.0, "min": 1.0, "max": 20.0}, "seed": {"type": "integer", "required": False, "default": None}, "output_format": {"type": "string", "required": False, "default": "png", "enum": ["png", "jpeg"]}, "lora_id": {"type": "string", "required": False}, "lora_scale": {"type": "number", "required": False, "default": 1.0, "min": 0.0, "max": 1.0}, }, "resolutions": { "512x512": {"width": 512, "height": 512}, "1024x1024": {"width": 1024, "height": 1024}, "1024x768": {"width": 1024, "height": 768}, "768x1024": {"width": 768, "height": 1024}, "1280x768": {"width": 1280, "height": 768}, "768x1280": {"width": 768, "height": 1280}, }, }, "z-image/turbo/text-to-image/lora": { "id": "z-image/turbo/text-to-image/lora", "name": "Z-Image Turbo Text-to-Image with LoRA", "type": "text-to-image", "input": { "prompt": {"type": "string", "required": True, "max_length": 2000}, "negative_prompt": {"type": "string", "required": False, "default": ""}, "width": {"type": "integer", "required": False, "default": 1024}, "height": {"type": "integer", "required": False, "default": 1024}, "num_images": {"type": "integer", "required": False, "default": 1, "min": 1, "max": 4}, "num_inference_steps": {"type": "integer", "required": False, "default": 4, "min": 1, "max": 4}, "guidance_scale": {"type": "number", "required": False, "default": 1.0, "min": 1.0, "max": 20.0}, "seed": {"type": "integer", "required": False, "default": None}, "output_format": {"type": "string", "required": False, "default": "png", "enum": ["png", "jpeg"]}, "lora_id": {"type": "string", "required": True}, "lora_scale": {"type": "number", "required": False, "default": 1.0, "min": 0.0, "max": 1.0}, }, "resolutions": { "512x512": {"width": 512, "height": 512}, "1024x1024": {"width": 1024, "height": 1024}, "1024x768": {"width": 1024, "height": 768}, "768x1024": {"width": 768, "height": 1024}, "1280x768": {"width": 1280, "height": 768}, "768x1280": {"width": 768, "height": 1280}, }, }, "z-image/turbo/image-to-image": { "id": "z-image/turbo/image-to-image", "name": "Z-Image Turbo Image-to-Image", "type": "image-to-image", "input": { "prompt": {"type": "string", "required": True, "max_length": 2000}, "image_url": {"type": "string", "required": True, "description": "S3 key from upload-image"}, "negative_prompt": {"type": "string", "required": False, "default": ""}, "width": {"type": "integer", "required": False, "default": 1024}, "height": {"type": "integer", "required": False, "default": 1024}, "num_images": {"type": "integer", "required": False, "default": 1, "min": 1, "max": 4}, "strength": {"type": "number", "required": False, "default": 0.7, "min": 0.0, "max": 1.0}, "num_inference_steps": {"type": "integer", "required": False, "default": 4, "min": 1, "max": 4}, "guidance_scale": {"type": "number", "required": False, "default": 1.0, "min": 1.0, "max": 20.0}, "seed": {"type": "integer", "required": False, "default": None}, "output_format": {"type": "string", "required": False, "default": "png", "enum": ["png", "jpeg"]}, "lora_id": {"type": "string", "required": False}, "lora_scale": {"type": "number", "required": False, "default": 1.0, "min": 0.0, "max": 1.0}, }, "resolutions": { "512x512": {"width": 512, "height": 512}, "1024x1024": {"width": 1024, "height": 1024}, "1024x768": {"width": 1024, "height": 768}, "768x1024": {"width": 768, "height": 1024}, "1280x768": {"width": 1280, "height": 768}, "768x1280": {"width": 768, "height": 1280}, }, }, "z-image/turbo/image-to-image/lora": { "id": "z-image/turbo/image-to-image/lora", "name": "Z-Image Turbo Image-to-Image with LoRA", "type": "image-to-image", "input": { "prompt": {"type": "string", "required": True, "max_length": 2000}, "image_url": {"type": "string", "required": True, "description": "S3 key from upload-image"}, "negative_prompt": {"type": "string", "required": False, "default": ""}, "width": {"type": "integer", "required": False, "default": 1024}, "height": {"type": "integer", "required": False, "default": 1024}, "num_images": {"type": "integer", "required": False, "default": 1, "min": 1, "max": 4}, "strength": {"type": "number", "required": False, "default": 0.7, "min": 0.0, "max": 1.0}, "num_inference_steps": {"type": "integer", "required": False, "default": 4, "min": 1, "max": 4}, "guidance_scale": {"type": "number", "required": False, "default": 1.0, "min": 1.0, "max": 20.0}, "seed": {"type": "integer", "required": False, "default": None}, "output_format": {"type": "string", "required": False, "default": "png", "enum": ["png", "jpeg"]}, "lora_id": {"type": "string", "required": True}, "lora_scale": {"type": "number", "required": False, "default": 1.0, "min": 0.0, "max": 1.0}, }, "resolutions": { "512x512": {"width": 512, "height": 512}, "1024x1024": {"width": 1024, "height": 1024}, "1024x768": {"width": 1024, "height": 768}, "768x1024": {"width": 768, "height": 1024}, "1280x768": {"width": 1280, "height": 768}, "768x1280": {"width": 768, "height": 1280}, }, }, "minimax/h3/text-to-video": { "id": "minimax/h3/text-to-video", "name": "MiniMax H3 Text-to-Video", "type": "text-to-video", "billing": "per output second", "note": "Duration-based. Ignores num_frames / frames_per_second; output FPS is fixed at 24.", "input": { "prompt": {"type": "string", "required": True}, "resolution_tier": {"type": "string", "required": False, "default": "480p", "enum": ["480p", "768p"]}, "aspect_ratio": {"type": "string", "required": False, "default": "16:9", "enum": ["16:9", "4:3", "1:1", "3:4", "9:16"]}, "duration": {"type": "number", "required": False, "default": 5, "min": 4, "max": 15}, "seed": {"type": "integer", "required": False, "default": None}, }, "frame_sizes": { "480p": {"16:9": [832, 480], "4:3": [640, 480], "1:1": [480, 480], "3:4": [480, 640], "9:16": [480, 832]}, "768p": {"16:9": [1344, 768], "4:3": [1024, 768], "1:1": [768, 768], "3:4": [768, 1024], "9:16": [768, 1344]}, }, }, "minimax/h3/image-to-video": { "id": "minimax/h3/image-to-video", "name": "MiniMax H3 Image-to-Video", "type": "image-to-video", "billing": "per output second", "input": { "prompt": {"type": "string", "required": True}, "image_url": {"type": "string", "required": True, "description": "S3 key from upload-image"}, "end_image_url": {"type": "string", "required": False, "description": "Optional closing frame"}, "resolution_tier": {"type": "string", "required": False, "default": "480p", "enum": ["480p", "768p"]}, "aspect_ratio": {"type": "string", "required": False, "default": "16:9", "enum": ["16:9", "4:3", "1:1", "3:4", "9:16"]}, "duration": {"type": "number", "required": False, "default": 5, "min": 4, "max": 15}, "seed": {"type": "integer", "required": False, "default": None}, }, }, "minimax/h3/text-to-video/turbo": { "id": "minimax/h3/text-to-video/turbo", "name": "MiniMax H3 Text-to-Video Turbo", "type": "text-to-video", "billing": "per output second", "note": "MiniMax H3 Turbo; fixed 8 sampling steps; 480p $0.006/s, 768p $0.009/s.", }, "minimax/h3/image-to-video/turbo": { "id": "minimax/h3/image-to-video/turbo", "name": "MiniMax H3 Image-to-Video Turbo", "type": "image-to-video", "billing": "per output second", "note": "MiniMax H3 Turbo; fixed 8 sampling steps; 480p $0.006/s, 768p $0.009/s.", }, "minimax/h3/reference-to-video": { "id": "minimax/h3/reference-to-video", "name": "MiniMax H3 Reference-to-Video (Ref2VA)", "type": "reference-to-video", "billing": "per output second PLUS per reference input - see pricing.minimax_h3_ref2va", "note": "Refer to references positionally in the prompt as , , ...", "input": { "prompt": {"type": "string", "required": True}, "reference_image_urls": {"type": "array", "required": False}, "reference_video_urls": {"type": "array", "required": False}, "reference_audio_urls": {"type": "array", "required": False}, "use_ref_video_audio": {"type": "boolean", "required": False, "default": False}, "resolution_tier": {"type": "string", "required": False, "default": "480p", "enum": ["480p", "768p"]}, "aspect_ratio": {"type": "string", "required": False, "default": "16:9", "enum": ["16:9", "4:3", "1:1", "3:4", "9:16"]}, "duration": {"type": "number", "required": False, "default": 5, "min": 4, "max": 15}, "seed": {"type": "integer", "required": False, "default": None}, }, "requires": "at least one of reference_image_urls / reference_video_urls", "reference_limits": { "480p": {"max_images": 5, "max_images_if_images_only": 9, "max_videos": 3, "max_audios": 3}, "768p": {"max_images": 5, "max_images_if_images_only": 9, "max_videos": 3, "max_audios": 3}, "max_reference_video_total_seconds": 15.1, "max_reference_video_fps": 24, "max_reference_audio_seconds": 10.0, "max_reference_image_edge": 2048, "max_reference_image_aspect_ratio": 4.0, }, }, "wan/v2.2-a14b/text-to-video/turbo": { "id": "wan/v2.2-a14b/text-to-video/turbo", "name": "Wan 2.2 Text-to-Video 14B (turbo)", "type": "text-to-video", "note": "Ignores num_inference_steps / guidance_scale. 1080p allows 153 frames (standard caps at 81).", }, "wan/v2.2-a14b/image-to-video/turbo": { "id": "wan/v2.2-a14b/image-to-video/turbo", "name": "Wan 2.2 Image-to-Video 14B (turbo)", "type": "image-to-video", "note": "Ignores num_inference_steps / guidance_scale.", }, "openai/gpt-image-2/text-to-image": { "id": "openai/gpt-image-2/text-to-image", "name": "GPT Image 2 Text-to-Image", "type": "text-to-image", "note": ("Sizes: 1024x1024 | 1920x1072 / 1072x1920 (1080p) | 2560x1440 / 1440x2560 (2K); " "other sizes are rejected with 400. Always returns exactly 1 image " "(num_images ignored); also ignores num_inference_steps / guidance_scale / output_format."), "input": { "prompt": {"type": "string", "required": True}, "width": {"type": "integer", "required": False, "default": 1024}, "height": {"type": "integer", "required": False, "default": 1024}, "seed": {"type": "integer", "required": False}, }, # Mirrors the gateway's accepted resolution list for this model; a size that is # not listed here is rejected with 400. 1536x1024 / 1024x1536 are intentionally # absent: the provider honours them inconsistently and may silently return a # different size. 4K is not open yet. 1080 is not a multiple of 16 (the provider # requires multiples of 16), hence 1072. "resolutions": [[1024, 1024], [1920, 1072], [1072, 1920], [2560, 1440], [1440, 2560]], }, "flux2/dev/text-to-image": { "id": "flux2/dev/text-to-image", "name": "FLUX.2-dev Text-to-Image", "type": "text-to-image", "note": "Own resolution whitelist; always returns exactly 1 image (num_images and steps are ignored).", "input": { "prompt": {"type": "string", "required": True}, "width": {"type": "integer", "required": False, "default": 1024}, "height": {"type": "integer", "required": False, "default": 1024}, "seed": {"type": "integer", "required": False}, }, "resolutions": [[2048, 1536], [1536, 2048], [2048, 1152], [1152, 2048], [2048, 2048], [1024, 1024]], }, "flux2/dev/image-edit": { "id": "flux2/dev/image-edit", "name": "FLUX.2-dev Image Edit", "type": "image-edit", "note": "Always returns exactly 1 image.", "input": { "prompt": {"type": "string", "required": True}, "image_url": {"type": "string", "required": True}, "width": {"type": "integer", "required": False, "default": 1024}, "height": {"type": "integer", "required": False, "default": 1024}, }, "resolutions": [[2048, 1536], [1536, 2048], [2048, 1152], [1152, 2048], [2048, 2048], [1024, 1024]], }, "qwen3-tts/text-to-speech/1.7b": { "id": "qwen3-tts/text-to-speech/1.7b", "name": "Qwen3-TTS CustomVoice", "type": "text-to-speech", "note": "Keys off `text`, NOT `prompt`. Use the submit-tts command.", "input": { "text": {"type": "string", "required": True, "max_length": 500}, "speaker": {"type": "string", "required": False, "default": "Sohee"}, "language": {"type": "string", "required": False, "default": "Chinese"}, "seed": {"type": "integer", "required": False, "description": "-1 = random"}, "temperature": {"type": "number", "required": False, "default": 0.9, "min": 0.0, "max": 2.0}, "top_p": {"type": "number", "required": False, "default": 1.0}, "top_k": {"type": "integer", "required": False, "default": 50, "min": 1, "max": 200}, "repetition_penalty": {"type": "number", "required": False, "default": 1.05, "min": 1.0, "max": 2.0}, }, }, }, "frame_alignment": { "formula": "valid_frames = 1 + 4*k where k >= 1", "examples": [5, 9, 13, 17, 21, 25, 29, 33, 37, 41, 45, 49, 53, 57, 61, 65, 69, 73, 77, 81], }, "pricing": { "inference": { "480p": {"per_frame_usd": 0.0009375, "per_frame_credits": 0.009375}, "720p": {"per_frame_usd": 0.001875, "per_frame_credits": 0.01875}, "1080p": {"per_frame_usd": 0.0025, "per_frame_credits": 0.025}, }, "s2v": { "480p": {"per_frame_usd": 0.0009375, "per_frame_credits": 0.009375}, "512p": {"per_frame_usd": 0.0013125, "per_frame_credits": 0.013125}, "720p": {"per_frame_usd": 0.001875, "per_frame_credits": 0.01875}, }, "animate": { "480p": {"per_frame_usd": 0.00125, "per_frame_credits": 0.0125}, "512p": {"per_frame_usd": 0.00175, "per_frame_credits": 0.0175}, "720p": {"per_frame_usd": 0.0025, "per_frame_credits": 0.025}, }, "lora_multiplier": 1.2, "minimax_h3": { "480p": {"per_second_usd": 0.0045, "per_second_credits": 0.045}, "768p": {"per_second_usd": 0.012, "per_second_credits": 0.12}, }, "minimax_h3_ref2va": { "formula": "duration * base_rate + max(0, images - 5) * 0.0075 + ref_video_seconds * base_rate (audio free)", "480p": {"base_per_second_usd": 0.0045, "reference_video_per_second_usd": 0.0045}, "768p": {"base_per_second_usd": 0.012, "reference_video_per_second_usd": 0.012}, "reference_image_usd": 0.0075, "reference_image_free_count": 5, "reference_audio_usd": 0.0, "note": "Same base rate as plain H3 T2V/I2V since 2026-09-10.", }, "minimax_h3_turbo": { "480p": {"per_second_usd": 0.0015, "per_second_credits": 0.015}, "768p": {"per_second_usd": 0.004, "per_second_credits": 0.04}, }, "minimax_h3_ref2va_turbo": { "formula": "duration * turbo_base_rate + max(0, images - 5) * 0.0025 + ref_video_seconds * turbo_base_rate (audio free)", "480p": {"base_per_second_usd": 0.0015, "reference_video_per_second_usd": 0.0015}, "768p": {"base_per_second_usd": 0.004, "reference_video_per_second_usd": 0.004}, "reference_image_usd": 0.0025, "reference_image_free_count": 5, "reference_audio_usd": 0.0, }, "audio": {"qwen_tts_per_char_usd": 3e-05, "min_charge_usd": 0.003}, "image": { "gpt_image_2_t2i": {"per_image_usd": 0.03, "per_image_credits": 0.3}, "flux2_t2i": {"per_image_usd": 0.006, "per_image_credits": 0.06}, "flux2_image_edit": {"per_image_usd": 0.012, "per_image_credits": 0.12}, "qwen": {"per_image_usd": 0.015, "per_image_credits": 0.15}, "qwen_lora": {"per_image_usd": 0.018, "per_image_credits": 0.18}, "qwen_image_edit": {"per_image_usd": 0.003, "per_image_credits": 0.03}, "qwen_image_edit_turbo": {"per_image_usd": 0.005, "per_image_credits": 0.05}, "z_turbo": {"per_image_usd": 0.0025, "per_image_credits": 0.025}, "z_turbo_lora": {"per_image_usd": 0.003, "per_image_credits": 0.03}, }, "credits_per_usd": 10, "note": "Live prices are always available at runtime via `studio-pricing` (Image Studio) — this table is a static reference and can drift; treat it as indicative only.", }, "limits": { "max_loras_per_user": 20, "max_api_keys_per_user": 10, # Enforced against the COMBINED size of high_noise + low_noise, not per file: # exceeding it returns HTTP 413 "LoRA files too large (N MB). Maximum allowed: 2048 MB." "max_lora_upload_total_mb": 2048, "uploaded_lora_expiry_days": 1, "saved_lora_expiry_days": 7, "max_image_file_size_mb": 20, "rate_limit_requests_per_minute": 1000, }, } # ─── Image Studio catalog ────────────────────────────────────────────────── # Image Studio is a separate product surface (e-commerce product/model photography, # not the video/LoRA pipeline above) reached through the SAME gateway + API key, # under the `/api/v1/image-studio` prefix. All endpoints are async: POST returns # {request_id, status:"pending"}, poll GET /jobs/{request_id} until status is "done"/"error"/"cancelled". # Every image-generation endpoint accepts `model` (whitelist: "openai:gpt-image-2", # "phosor:qwen-image-edit-v2511") and is billed uniformly per output image; the four # analyze endpoints use a fixed model and a daily free quota before billing kicks in. # Real-time prices: call `studio-pricing` (GET /pricing) — do not hardcode from here. IMAGE_STUDIO = { "base_path": "/api/v1/image-studio", "billing": { "generation": "Flat per-image credit rate (same for every generation endpoint below), returned live by GET /pricing as `per_image_credits`. Partial success (e.g. 3 of 5 images generated) is billed only for the successes; the rest is auto-refunded.", "analyze": "First `analyze_daily_free_quota` analyze calls per user per day are free (shared across agent/product/model analyze); beyond that, `per_analyze_credits` per call (from GET /pricing).", }, "endpoints": { # ── Analyze (freemium, shared quota) ── "agent/analyze": {"method": "POST", "path": "/agent/analyze", "billing": "analyze", "fields": {"image_url": "required", "prompt": "optional", "language": "optional, default zh"}}, "product/analyze": {"method": "POST", "path": "/product/analyze", "billing": "analyze", "fields": {"image_url": "required", "prompt": "optional", "language": "optional, default zh"}}, "model/analyze": {"method": "POST", "path": "/model/analyze", "billing": "analyze", "fields": {"image_url": "required", "prompt": "optional", "language": "optional, default zh"}}, "product/reference/analyze": {"method": "POST", "path": "/product/reference/analyze", "billing": "free", "fields": {"source_type": "optional: web_url|image_url, default web_url", "url": "required if source_type=web_url", "image_url": "required if source_type=image_url", "prompt": "optional"}}, # ── Product image generation ── "product/suite": {"method": "POST", "path": "/product/suite", "billing": "generation", "model_selectable": True, "fields": {"product_image_url": "required", "layout_types": "optional list[str]", "count_per_type": "optional int, default 1", "custom_suggestions": "optional list", "product_info": "optional str", "same_style_reference": "optional dict", "brand_config": "optional dict", "aspect_ratio": "optional: 1:1|3:4|4:3|9:16|16:9", "gen_language": "optional", "model": "optional"}}, "product/scene-compose": {"method": "POST", "path": "/product/scene-compose", "billing": "generation", "model_selectable": True, "fields": {"composite_image_url": "required", "reference_image_url": "required", "similarity": "optional, default strong", "prompt": "optional", "count": "optional int 1-20, default 2", "aspect_ratio": "optional", "model": "optional"}}, "product/scene-variation": {"method": "POST", "path": "/product/scene-variation", "billing": "generation", "model_selectable": True, "fields": {"product_image_url": "required", "count": "optional int 1-20, default 2", "style": "optional: smart|realistic|minimal|studio|outdoor|lifestyle", "scene_description": "optional", "aspect_ratio": "optional", "model": "optional"}}, "product/remove-bg": {"method": "POST", "path": "/product/remove-bg", "billing": "generation", "model_selectable": True, "fields": {"image_url": "required", "count": "optional int, max 20, default 1", "aspect_ratio": "optional", "model": "optional"}}, "product/replace": {"method": "POST", "path": "/product/replace", "billing": "generation", "model_selectable": True, "fields": {"image_url": "required (scene)", "new_product_url": "optional (product to composite in)", "count": "optional int 1-4, default 1", "prompt": "optional", "model": "optional"}}, "product/inpaint": {"method": "POST", "path": "/product/inpaint", "billing": "generation", "model_selectable": True, "fields": {"image_url": "required", "mask_url": "required", "prompt": "optional fill description", "aspect_ratio": "optional", "count": "optional int, max 20", "model": "optional"}}, "product/erase": {"method": "POST", "path": "/product/erase", "billing": "generation", "model_selectable": True, "fields": {"image_url": "required", "mask_url": "required", "prompt": "optional removal note", "aspect_ratio": "optional", "count": "optional int, max 20", "model": "optional"}}, "product/handheld": {"method": "POST", "path": "/product/handheld", "billing": "generation", "model_selectable": True, "fields": {"image_url": "required", "pose_id": "optional: front_single_lift|side_single_grip|both_hands_cup|usage_scene|gift_display|outdoor_handheld", "prompt": "optional override", "aspect_ratio": "optional", "count": "optional int, max 20", "model": "optional"}}, "product/translate": {"method": "POST", "path": "/product/translate", "billing": "generation", "model_selectable": True, "fields": {"image_url": "required", "source_lang": "optional, default auto", "target_langs": "optional list[str], default [en] — one output image per language", "aspect_ratio": "optional, default source (keep input size)", "model": "optional"}}, "product/outpaint": {"method": "POST", "path": "/product/outpaint", "billing": "generation", "model_selectable": True, "fields": {"image_url": "required", "expand": "optional dict {top,bottom,left,right} in px, default all 0", "prompt": "optional expansion note", "model": "optional"}}, "product/recolor": {"method": "POST", "path": "/product/recolor", "billing": "generation", "model_selectable": True, "fields": {"image_url": "required", "target_color": "optional", "region": "optional, default the product", "prompt": "optional", "aspect_ratio": "optional", "count": "optional int, max 20", "model": "optional"}}, "product/enhance": {"method": "POST", "path": "/product/enhance", "billing": "generation", "model_selectable": True, "fields": {"image_url": "required", "enhance_type": "optional: smart|product|color|pattern|hands", "prompt": "optional", "aspect_ratio": "optional", "model": "optional"}}, "product/upscale": {"method": "POST", "path": "/product/upscale", "billing": "generation", "model_selectable": True, "fields": {"image_url": "required", "prompt": "optional", "model": "optional"}}, # ── Model (clothing/photography) image generation ── "model/clothing-suite": {"method": "POST", "path": "/model/clothing-suite", "billing": "generation", "model_selectable": True, "fields": {"clothing_image_urls": "required list[str], up to 5 used", "main_image_types": "optional dict {model_shot|grass_shot|selling_point|size_chart: count}", "aplus_types": "optional dict {standard_aplus|mobile_aplus|basic_aplus|custom_ratio: count}", "product_info": "optional", "brand_config": "optional dict", "same_style_reference": "optional dict", "aspect_ratio": "optional, default 3:4", "gen_language": "optional", "model": "optional"}}, "model/real-model-swap": {"method": "POST", "path": "/model/real-model-swap", "billing": "generation", "model_selectable": True, "fields": {"garment_image_urls": "required list[str]", "count": "optional int 1-20, default 2", "model_attrs": "optional dict {gender,age_group,ethnicity,skin_tone,hair_color}", "scene_id": "optional", "aspect_ratio": "optional", "model": "optional"}}, "model/mannequin-swap": {"method": "POST", "path": "/model/mannequin-swap", "billing": "generation", "model_selectable": True, "fields": {"garment_image_urls": "required list[str]", "count": "optional int 1-20, default 2", "model_attrs": "optional dict", "expression": "optional, default smile", "custom_prompt": "optional, max 500 chars", "aspect_ratio": "optional", "model": "optional"}}, "model/model-scene-swap": {"method": "POST", "path": "/model/model-scene-swap", "billing": "generation", "model_selectable": True, "fields": {"source_image_urls": "required list[str]", "count": "optional int 1-20, default 2", "scene_id": "optional", "scene_description": "optional", "scene_image_url": "optional (image-to-image scene reference)", "aspect_ratio": "optional", "model": "optional"}}, "model/ai-outfit": {"method": "POST", "path": "/model/ai-outfit", "billing": "generation", "model_selectable": True, "fields": {"clothing_image_urls": "required list[str], first 2 used", "count": "optional int 1-20, default 2", "model_attrs": "optional dict — REQUIRED for accurate results, not inferred from image alone", "aspect_ratio": "optional", "model": "optional"}}, "model/pose-variation": {"method": "POST", "path": "/model/pose-variation", "billing": "generation", "model_selectable": True, "fields": {"source_image_urls": "required list[str]", "pose_ids": "optional list[str], up to 3", "custom_pose_url": "optional reference pose image", "count": "optional int, per-pose count, default 1", "aspect_ratio": "optional", "model": "optional"}}, "model/ai-wearable": {"method": "POST", "path": "/model/ai-wearable", "billing": "generation", "model_selectable": True, "fields": {"accessory_image_url": "required", "count": "optional int 1-20, default 2", "wearable_type": "optional, default accessory", "model_attrs": "optional dict", "aspect_ratio": "optional", "model": "optional"}}, # ── Job / account ── "jobs": {"method": "GET", "path": "/jobs/{request_id}", "billing": "free"}, "cancel": {"method": "POST", "path": "/jobs/{request_id}/cancel", "billing": "free"}, "pricing": {"method": "GET", "path": "/pricing", "billing": "free"}, "my-works": {"method": "GET", "path": "/my-works", "billing": "free", "fields": {"task_type": "optional filter", "date_from": "optional YYYY-MM-DD", "date_to": "optional YYYY-MM-DD", "limit": "optional 1-100, default 20", "offset": "optional, default 0"}}, }, } # ─── Helpers ──────────────────────────────────────────────────────────────── def _json_out(data): """Print JSON to stdout and exit 0.""" print(json.dumps(data, indent=2, default=str)) sys.exit(0) def _error_out(message, code=1): """Print error JSON to stderr and exit non-zero.""" print(json.dumps({"error": message}, indent=2), file=sys.stderr) sys.exit(code) def _parse_json_arg(raw, flag_name, expected_type): """Parse a --flag JSON string argument; None passes through untouched.""" if raw is None: return None try: value = json.loads(raw) except json.JSONDecodeError: _error_out(f"{flag_name} must be valid JSON") if not isinstance(value, expected_type): _error_out(f"{flag_name} must be a JSON {expected_type}") return value # ─── PhosorClient ────────────────────────────────────────────────────────── class PhosorClient: """Core API wrapper — stdlib-only HTTP via urllib.""" def __init__(self, base_url=None, api_key=None, allow_http=None): resolved = (base_url or os.environ.get("PHOSOR_BASE_URL") or DEFAULT_BASE_URL).rstrip("/") # Production defaults to HTTPS (phosor.ai). Dev is the SAME API/endpoints/params — only the # base URL differs: a plain-http host:port of a deployment you run yourself. Allow http when explicitly opted in # (--allow-http / PHOSOR_ALLOW_HTTP=1) or when it's clearly a local gateway, so dev and prod # share this ONE client instead of anyone forking a patched copy. if allow_http is None: allow_http = os.environ.get("PHOSOR_ALLOW_HTTP", "").lower() in ("1", "true", "yes") _is_local = resolved.startswith(("http://localhost", "http://127.0.0.1")) if not resolved.startswith("https://") and not (allow_http or _is_local): _error_out(f"Base URL must use HTTPS (pass --allow-http or PHOSOR_ALLOW_HTTP=1 for a dev http gateway): {resolved}") self.base_url = resolved self.api_key = api_key or self._resolve_api_key() @staticmethod def _resolve_api_key(): key = os.environ.get("PHOSOR_API_KEY") if key: return key.strip() # Fallback: try reading from TOOLS.md in workspace tools_md = WORKSPACE_DIR / "TOOLS.md" if tools_md.exists(): for line in tools_md.read_text().splitlines(): if "PHOSOR_API_KEY" in line and "=" in line: candidate = line.split("=", 1)[1].strip().strip('"').strip("'") # Basic validation: reject suspiciously long or malformed keys if candidate and len(candidate) <= 256 and re.match(r'^[A-Za-z0-9_\-\.]+$', candidate): return candidate return None def _request(self, method, path, json_data=None, params=None, raise_http=False): """Make an authenticated HTTP request. Returns parsed JSON. With `raise_http=True` an HTTPError is re-raised instead of going through `_error_out` and exiting -- for callers that branch on the status code (check_key reports 401/403 as "invalid" rather than printing an error and exiting). Defaults to False; every other caller keeps its behaviour. """ url = f"{self.base_url}{path}" if params: qs = "&".join(f"{quote(str(k), safe='')}={quote(str(v), safe='')}" for k, v in params.items() if v is not None) if qs: url = f"{url}?{qs}" body = None if json_data is not None: body = json.dumps(json_data).encode("utf-8") headers = {"Content-Type": "application/json"} if self.api_key: headers["X-API-Key"] = self.api_key req = Request(url, data=body, headers=headers, method=method) try: with urlopen(req, timeout=120) as resp: raw = resp.read().decode("utf-8") return json.loads(raw) if raw else {} except HTTPError as e: if raise_http: raise raw = e.read().decode("utf-8", errors="replace") try: detail = json.loads(raw) except Exception: detail = {"raw": raw} _error_out(f"HTTP {e.code}: {detail}") except URLError as e: _error_out(f"Connection error: {e.reason}") @staticmethod def _sanitize_filename(name): """Remove characters unsafe for Content-Disposition filename.""" return re.sub(r'["\r\n\\]', '_', name) def _upload_multipart(self, path, file_path, field_name="image"): """Upload a file via multipart/form-data (stdlib-only).""" boundary = f"----PhosorBoundary{uuid.uuid4().hex}" file_path = Path(file_path) filename = self._sanitize_filename(file_path.name) # Guess content type ext = file_path.suffix.lower() ct_map = { ".jpg": "image/jpeg", ".jpeg": "image/jpeg", ".png": "image/png", ".webp": "image/webp", ".zip": "application/zip", ".safetensors": "application/octet-stream", } content_type = ct_map.get(ext, "application/octet-stream") file_data = file_path.read_bytes() body_parts = [] body_parts.append(f"--{boundary}\r\n".encode()) body_parts.append( f'Content-Disposition: form-data; name="{field_name}"; filename="{filename}"\r\n' f"Content-Type: {content_type}\r\n\r\n".encode() ) body_parts.append(file_data) body_parts.append(f"\r\n--{boundary}--\r\n".encode()) body = b"".join(body_parts) url = f"{self.base_url}{path}" headers = { "Content-Type": f"multipart/form-data; boundary={boundary}", } if self.api_key: headers["X-API-Key"] = self.api_key req = Request(url, data=body, headers=headers, method="POST") try: with urlopen(req, timeout=300) as resp: raw = resp.read().decode("utf-8") return json.loads(raw) if raw else {} except HTTPError as e: raw = e.read().decode("utf-8", errors="replace") try: detail = json.loads(raw) except Exception: detail = {"raw": raw} _error_out(f"HTTP {e.code}: {detail}") except URLError as e: _error_out(f"Connection error: {e.reason}") # ── Inference ─────────────────────────────────────────────────────────── def check_key(self): """Validate the API key against an endpoint that actually REQUIRES it. History, so nobody repeats it: this used to probe /api/v1/training/quotas, which stopped being public (2026-09-02). The first replacement was /api/v1/models — which is a **public** endpoint that answers 200 without any key, so `check-key` cheerfully reported `"status": "valid"` for `sk_total_garbage`. A key check whose probe does not require the key validates nothing. The probe must therefore be auth-required, cheap, and part of the public surface. GET /api/v1/loras?limit=1 is all three (inference/history is auth-required too but returns ~160KB, far too heavy for a liveness check). """ return self._request("GET", "/api/v1/loras", params={"limit": 1}, raise_http=True) def submit(self, prompt, **kwargs): """Submit an inference job (T2V, I2V, T2I, or I2I).""" payload = {"prompt": prompt} for k in _SUBMIT_FIELDS: if kwargs.get(k) is not None: payload[k] = kwargs[k] return self._request("POST", "/api/v1/inference/submit", json_data=payload) def submit_tts(self, text, **kwargs): """Submit a Qwen3-TTS job. Same endpoint as submit(), but TTS keys off `text` rather than `prompt` — sending a prompt here produces nothing. """ payload = {"model": kwargs.pop("model", None) or "qwen3-tts/text-to-speech/1.7b", "text": text} for k in ("speaker", "language", "seed", "temperature", "top_p", "top_k", "repetition_penalty"): if kwargs.get(k) is not None: payload[k] = kwargs[k] return self._request("POST", "/api/v1/inference/submit", json_data=payload) def status(self, request_id): """Get job status.""" return self._request("GET", f"/api/v1/inference/status/{request_id}") def result(self, request_id): """Get job result (video URL or image URL).""" return self._request("GET", f"/api/v1/inference/result/{request_id}") def history(self, limit=50): """Get job history for authenticated user.""" return self._request("GET", "/api/v1/inference/history", params={"limit": limit}) # ── Storage: Image ────────────────────────────────────────────────────── def upload_image(self, file_path): """Upload a local image file for I2V.""" return self._upload_multipart("/api/v1/storage/image/upload", file_path, field_name="image") def import_image(self, url, filename=None): """Import an image from a remote URL.""" payload = {"url": url} if filename: payload["filename"] = filename return self._request("POST", "/api/v1/storage/image/import", json_data=payload) # ── Storage: LoRA ─────────────────────────────────────────────────────── def upload_lora(self, high_noise_path, low_noise_path, name=None): """Upload a LoRA model (two .safetensors files: high_noise + low_noise).""" boundary = f"----PhosorBoundary{uuid.uuid4().hex}" high_path = Path(high_noise_path) low_path = Path(low_noise_path) body_parts = [] for field, fpath in [("high_noise_file", high_path), ("low_noise_file", low_path)]: safe_name = self._sanitize_filename(fpath.name) body_parts.append(f"--{boundary}\r\n".encode()) body_parts.append( f'Content-Disposition: form-data; name="{field}"; filename="{safe_name}"\r\n' f"Content-Type: application/octet-stream\r\n\r\n".encode() ) body_parts.append(fpath.read_bytes()) body_parts.append(b"\r\n") if name: safe_val = re.sub(r'[\r\n]', ' ', name) body_parts.append(f"--{boundary}\r\n".encode()) body_parts.append( f'Content-Disposition: form-data; name="name"\r\n\r\n{safe_val}\r\n'.encode() ) body_parts.append(f"--{boundary}--\r\n".encode()) body = b"".join(body_parts) url = f"{self.base_url}/api/v1/storage/lora/upload" headers = {"Content-Type": f"multipart/form-data; boundary={boundary}"} if self.api_key: headers["X-API-Key"] = self.api_key req = Request(url, data=body, headers=headers, method="POST") try: with urlopen(req, timeout=600) as resp: raw = resp.read().decode("utf-8") return json.loads(raw) if raw else {} except HTTPError as e: raw = e.read().decode("utf-8", errors="replace") try: detail = json.loads(raw) except Exception: detail = {"raw": raw} _error_out(f"HTTP {e.code}: {detail}") except URLError as e: _error_out(f"Connection error: {e.reason}") def import_lora(self, high_noise_url, low_noise_url=None, name=None): """Import a LoRA model from remote URLs (HTTPS .safetensors). Video LoRA: two files (high_noise_url + low_noise_url). Image LoRA: single file (high_noise_url only). """ payload = { "high_noise_url": high_noise_url, } if low_noise_url: payload["low_noise_url"] = low_noise_url if name: payload["name"] = name return self._request("POST", "/api/v1/storage/lora/import", json_data=payload) def list_models(self): """List available models.""" return self._request("GET", "/api/v1/models") # ── LoRA Management ───────────────────────────────────────────────────── def loras(self, limit=50, offset=0): """List LoRA models.""" return self._request("GET", "/api/v1/loras", params={"limit": limit, "offset": offset}) def lora_status(self, lora_id): """Get LoRA upload/import status.""" return self._request("GET", f"/api/v1/loras/{lora_id}/status") def save_lora(self, lora_id, name=None): """Activate a LoRA, extending its expiry to 7 days (status: trained → ready). `trained` is just the wire value for "finished processing" - it is set for uploaded and imported LoRAs alike. """ payload = {} if name: payload["name"] = name return self._request("POST", f"/api/v1/loras/{lora_id}/save", json_data=payload) def delete_lora(self, lora_id): """Delete a LoRA model (soft delete).""" return self._request("DELETE", f"/api/v1/loras/{lora_id}") # ── Utility ───────────────────────────────────────────────────────────── # ── Image Studio (product/model photography — separate product surface) ── IMAGE_STUDIO_PREFIX = "/api/v1/image-studio" def studio_call(self, method, path, json_data=None, params=None): """Generic Image Studio call for any endpoint in IMAGE_STUDIO['endpoints']. `path` is relative to /api/v1/image-studio, e.g. '/product/remove-bg'.""" if not path.startswith("/"): path = "/" + path return self._request(method.upper(), f"{self.IMAGE_STUDIO_PREFIX}{path}", json_data=json_data, params=params) def studio_analyze(self, target, image_url, prompt=None, url=None, language=None): """target: 'agent' | 'product' | 'model' | 'reference'.""" path_map = { "agent": "/agent/analyze", "product": "/product/analyze", "model": "/model/analyze", "reference": "/product/reference/analyze", } if target not in path_map: _error_out(f"--target must be one of {list(path_map)}") payload = {} if target == "reference": payload["source_type"] = "image_url" if image_url else "web_url" if image_url: payload["image_url"] = image_url if url: payload["url"] = url else: if image_url: payload["image_url"] = image_url if prompt: payload["prompt"] = prompt if language: payload["language"] = language return self.studio_call("POST", path_map[target], json_data=payload) def studio_suite(self, product_image_url, layout_types=None, count_per_type=None, custom_suggestions=None, product_info=None, aspect_ratio=None, gen_language=None, model=None, same_style_reference=None, layout_template_ids=None, layout_random_count=None, image_urls=None): payload = {"product_image_url": product_image_url} if image_urls: payload["image_urls"] = image_urls if layout_types: payload["layout_types"] = layout_types if count_per_type is not None: payload["count_per_type"] = count_per_type if custom_suggestions: payload["custom_suggestions"] = custom_suggestions # Lean layout selection: send template ids, backend resolves layout_prompt+template image. if layout_template_ids: payload["layout_template_ids"] = layout_template_ids if layout_random_count: payload["layout_random_count"] = layout_random_count if product_info: payload["product_info"] = product_info if aspect_ratio: payload["aspect_ratio"] = aspect_ratio if gen_language: payload["gen_language"] = gen_language if model: payload["model"] = model if same_style_reference: payload["same_style_reference"] = same_style_reference return self.studio_call("POST", "/product/suite", json_data=payload) def studio_layouts(self, module="product", type_filter=None): """List the layout template library. It is a STATIC asset served by the web front (https://phosor.ai) at /asset-library/..., not a /api/v1 gateway endpoint, and needs no API key. On a self-hosted deployment where the web front is not the same host:port as the gateway, set PHOSOR_WEB_BASE_URL to the web front; it is tried after --base-url. module: product|clothing.""" path = f"/asset-library/image-studio/references/layouts/{module}/templates.json" candidates = [self.base_url] web_base = (os.environ.get("PHOSOR_WEB_BASE_URL") or "").rstrip("/") if web_base and web_base != self.base_url: candidates.append(web_base) last = None for base in candidates: try: with urlopen(Request(base + path, headers={"Accept": "application/json"}), timeout=60) as resp: data = json.loads(resp.read()) if type_filter: data = [t for t in data if t.get("type") == type_filter] return data except Exception as e: # noqa: BLE001 last = e _error_out("could not fetch the layout template library -- it is served by the " "web front, not the gateway. Point --base-url at the site " f"(https://phosor.ai) or set PHOSOR_WEB_BASE_URL. detail: {last}") def layout_templates_to_suggestions(self, template_ids, module="product"): """Expand chosen layout template ids into custom_suggestions entries, exactly how the UI sends a manual template selection to /product/suite.""" LABELS = {"selling_point": "Selling point", "aplus": "Premium A+", "white_bg": "White background", "scene": "Scene", "closeup": "Close-up", "size_chart": "Size chart"} tpls = {t.get("id"): t for t in self.studio_layouts(module)} cs = [] for tid in template_ids: t = tpls.get(tid) if not t: _error_out(f"layout template id not found in {module} library: {tid}") cs.append({"id": tid, "label": LABELS.get(t.get("type"), t.get("type") or "Image"), "desc": t.get("layout_prompt"), "layout_prompt": t.get("layout_prompt"), "template_file": t.get("file"), "template_name": t.get("name"), "type": "layout_composite", "width": 1024, "height": 1024}) return cs def studio_clothing_suite(self, clothing_image_urls, main_image_types=None, aplus_types=None, product_info=None, brand_config=None, aspect_ratio=None, gen_language=None, model=None, same_style_reference=None): payload = {"clothing_image_urls": clothing_image_urls} if main_image_types: payload["main_image_types"] = main_image_types if aplus_types: payload["aplus_types"] = aplus_types if product_info: payload["product_info"] = product_info if brand_config: payload["brand_config"] = brand_config if aspect_ratio: payload["aspect_ratio"] = aspect_ratio if gen_language: payload["gen_language"] = gen_language if model: payload["model"] = model if same_style_reference: payload["same_style_reference"] = same_style_reference return self.studio_call("POST", "/model/clothing-suite", json_data=payload) def studio_cancel(self, request_id): """Cancel a running Image Studio generation. Billing follows how far each image got, not how many you receive: images still queued are refunded, images already generating are charged (they cannot be stopped), images already delivered bill once through the normal path. The response reports the split as refunded_queued / charged_running / already_done, and the task then polls as status "cancelled" - not an error. """ return self.studio_call("POST", f"/jobs/{request_id}/cancel") def studio_status(self, request_id): """Poll an Image Studio job. The async key is `request_id`, same name as inference but a **separate id space** - an Image Studio id is not valid on /api/v1/inference/status/... Earlier revisions called this `job_id`; that key is never in a response. """ return self.studio_call("GET", f"/jobs/{request_id}") def studio_pricing(self): """Live Image Studio pricing — always the source of truth over any static table.""" return self.studio_call("GET", "/pricing") def studio_my_works(self, task_type=None, limit=20, offset=0): params = {"limit": limit, "offset": offset} if task_type: params["task_type"] = task_type return self.studio_call("GET", "/my-works", params=params) # ─── PendingManager ──────────────────────────────────────────────────────── class PendingManager: """Track pending jobs locally at ~/.openclaw/workspace/phosor-pending.json.""" def __init__(self, path=None): self.path = Path(path) if path else PENDING_FILE self.path.parent.mkdir(parents=True, exist_ok=True) def _load(self): if self.path.exists(): try: return json.loads(self.path.read_text()) except (json.JSONDecodeError, OSError): return {} return {} def _save(self, data): self.path.write_text(json.dumps(data, indent=2, default=str)) def add(self, request_id, job_type, metadata=None): data = self._load() data[request_id] = { "job_type": job_type, "submitted_at": time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime()), **(metadata or {}), } self._save(data) def remove(self, request_id): data = self._load() data.pop(request_id, None) self._save(data) def list_pending(self): return self._load() def poll_all(self, client): """Poll all pending jobs, remove completed/failed, return results.""" data = self._load() results = {} to_remove = [] for rid, meta in data.items(): job_type = meta.get("job_type", "inference") try: if job_type == "image-studio": st = client.studio_status(rid) else: st = client.status(rid) results[rid] = st status_val = st.get("status") or st.get("job_status") or "" if status_val.upper() in ("COMPLETED", "FAILED", "NOT_FOUND", "DONE", "ERROR"): to_remove.append(rid) except SystemExit: results[rid] = {"error": "Failed to poll"} for rid in to_remove: self.remove(rid) return {"polled": len(data), "completed": len(to_remove), "results": results} # ─── CLI ──────────────────────────────────────────────────────────────────── def build_parser(): parser = argparse.ArgumentParser( prog="phosor_client", description="Phosor AI CLI — generate videos and images, bring your own LoRAs, manage models", ) parser.add_argument("--version", action="version", version=f"phosor-client {VERSION}") parser.add_argument("--base-url", default=None, help="API base URL override (default: https://phosor.ai)") parser.add_argument("--api-key", default=None, help="API key override") parser.add_argument("--allow-http", action="store_true", default=False, help="Allow a plain-http base URL (HTTPS is required otherwise). Also via PHOSOR_ALLOW_HTTP=1. localhost is always allowed.") sub = parser.add_subparsers(dest="command", help="Available commands") # ── check-key ─────────────────────────────────────────────────────── sub.add_parser("check-key", help="Validate API key") # ── submit ────────────────────────────────────────────────────────── p = sub.add_parser("submit", help="Submit inference job (T2V, I2V, T2I, or I2I)") p.add_argument("prompt", help="Text prompt") p.add_argument("--width", type=int, default=None) p.add_argument("--height", type=int, default=None) p.add_argument("--num-frames", type=int, default=None) p.add_argument("--fps", type=int, default=None, dest="frames_per_second") p.add_argument("--steps", type=int, default=None, dest="num_inference_steps", help="Inference steps (4-40, default 4)") p.add_argument("--guidance", type=float, default=None, dest="guidance_scale", help="Guidance scale (1.0-10.0 video, 1.0-20.0 image)") p.add_argument("--seed", type=int, default=None) p.add_argument("--negative-prompt", default=None) p.add_argument("--image-url", default=None, help="S3 key for I2V/I2I/S2V/Animate (from upload-image)") p.add_argument("--audio-url", default=None, help="Audio URL for S2V (speech-to-video)") p.add_argument("--video-url", default=None, help="Video URL for Animate (motion reference)") p.add_argument("--lora-id", default=None) p.add_argument("--lora-scale", type=float, default=None) p.add_argument("--loras", default=None, help='Multiple LoRAs as JSON: \'[{"lora_id":"...","lora_scale":1.0}]\'') p.add_argument("--model", default=None) p.add_argument("--num-images", type=int, default=None, dest="num_images", help="Number of images to generate, 1-4 (image models only)") p.add_argument("--strength", type=float, default=None, help="Transformation strength 0.0-1.0 (image-to-image only)") p.add_argument("--output-format", default=None, dest="output_format", choices=["png", "jpeg"], help="Output format (image models only)") # MiniMax H3 — duration-based, ignores width/height/num_frames/fps (output is fixed 24fps) p.add_argument("--duration", type=float, default=None, help="Output length in seconds, 4-15 (MiniMax H3; billed per output second)") p.add_argument("--resolution-tier", default=None, dest="resolution_tier", choices=["480p", "768p"], help="MiniMax H3 resolution tier (default 480p)") p.add_argument("--aspect", default=None, dest="aspect_ratio", choices=["16:9", "4:3", "1:1", "3:4", "9:16"], help="MiniMax H3 aspect ratio (default 16:9)") p.add_argument("--end-image-url", default=None, dest="end_image_url", help="Optional ending frame for minimax/h3/image-to-video") p.add_argument("--reference-image-urls", default=None, dest="reference_image_urls", help="Comma-separated reference images for minimax/h3/reference-to-video") p.add_argument("--reference-video-urls", default=None, dest="reference_video_urls", help="Comma-separated reference videos for minimax/h3/reference-to-video") p.add_argument("--reference-audio-urls", default=None, dest="reference_audio_urls", help="Comma-separated reference audio for minimax/h3/reference-to-video") p.add_argument("--use-ref-video-audio", action="store_true", default=None, dest="use_ref_video_audio", help="Also feed each reference video's own soundtrack as an audio reference") # ── status ────────────────────────────────────────────────────────── p = sub.add_parser("status", help="Get job status") p.add_argument("request_id", help="Job request ID") # ── result ────────────────────────────────────────────────────────── p = sub.add_parser("result", help="Get job result (video or image URL)") p.add_argument("request_id", help="Job request ID") # ── poll ──────────────────────────────────────────────────────────── sub.add_parser("poll", help="Poll all pending jobs") # ── list ──────────────────────────────────────────────────────────── sub.add_parser("list", help="List pending jobs (local tracking)") # ── history ───────────────────────────────────────────────────────── p = sub.add_parser("history", help="Get job history") p.add_argument("--limit", type=int, default=50) # ── upload-image ──────────────────────────────────────────────────── p = sub.add_parser("upload-image", help="Upload local image for I2V or I2I") p.add_argument("file", help="Path to image file (JPEG, PNG, WebP)") # ── import-image ──────────────────────────────────────────────────── p = sub.add_parser("import-image", help="Import image from URL") p.add_argument("url", help="Public image URL") p.add_argument("--filename", default=None) # ── upload-lora ───────────────────────────────────────────────────── p = sub.add_parser("upload-lora", help="Upload LoRA model (two .safetensors files)") p.add_argument("high_noise_file", help="Path to high_noise .safetensors file") p.add_argument("low_noise_file", help="Path to low_noise .safetensors file") p.add_argument("--name", default=None, help="LoRA model name") # ── import-lora ────────────────────────────────────────────────────── p = sub.add_parser("import-lora", help="Import LoRA from HTTPS URLs (video: 2 files, image: 1 file)") p.add_argument("high_noise_url", help="HTTPS URL to .safetensors (single file for image LoRA, high_noise for video)") p.add_argument("low_noise_url", nargs="?", default=None, help="HTTPS URL to low_noise .safetensors (video LoRA only)") p.add_argument("--name", default=None, help="LoRA model name") # ── submit-tts ────────────────────────────────────────────────────── p = sub.add_parser("submit-tts", help="Submit a text-to-speech job (Qwen3-TTS)") p.add_argument("text", help="Text to synthesize (max 500 characters)") p.add_argument("--speaker", default=None, help="Preset voice (default Sohee)") p.add_argument("--language", default=None, help="Language tag (default Chinese)") p.add_argument("--seed", type=int, default=None, help="-1 = random") p.add_argument("--temperature", type=float, default=None, help="0.0-2.0 (default 0.9)") p.add_argument("--top-p", type=float, default=None, dest="top_p", help="(0.0, 1.0] (default 1.0)") p.add_argument("--top-k", type=int, default=None, dest="top_k", help="1-200 (default 50)") p.add_argument("--repetition-penalty", type=float, default=None, dest="repetition_penalty", help="1.0-2.0 (default 1.05)") p.add_argument("--model", default=None, help="Override the TTS model id") # ── models ───────────────────────────────────────────────────────── sub.add_parser("models", help="List available models") # ── save-lora ─────────────────────────────────────────────────────── p = sub.add_parser("save-lora", help="Activate a LoRA, extending expiry to 7 days (trained → ready)") p.add_argument("lora_id", help="LoRA model UUID") p.add_argument("--name", default=None) # ── loras ─────────────────────────────────────────────────────────── p = sub.add_parser("loras", help="List LoRA models") p.add_argument("--limit", type=int, default=50) p.add_argument("--offset", type=int, default=0) # ── lora-status ───────────────────────────────────────────────────── p = sub.add_parser("lora-status", help="Get LoRA upload/import status") p.add_argument("lora_id", help="LoRA model UUID") # ── delete-lora ───────────────────────────────────────────────────── p = sub.add_parser("delete-lora", help="Delete a LoRA model") p.add_argument("lora_id", help="LoRA model UUID") # ── Image Studio (separate product surface: product/model photography) ── sub.add_parser("studio-features", help="List Image Studio endpoints, fields, and billing model") sub.add_parser("studio-pricing", help="Get live Image Studio pricing (source of truth over any static table)") p = sub.add_parser("studio-analyze", help="AI-analyze a product/garment image or reference URL (freemium)") p.add_argument("--target", required=True, choices=["agent", "product", "model", "reference"], help="agent=general Agent-image analysis, product=product suite, model=clothing suite, reference=analyze a reference URL/image") p.add_argument("--image-url", default=None, help="S3 key or URL of the image to analyze") p.add_argument("--url", default=None, help="Reference product page URL (only for --target reference with a web_url source)") p.add_argument("--prompt", default=None, help="Optional free-text hint for the analysis") p.add_argument("--language", default=None, choices=["zh", "en"]) p = sub.add_parser("studio-suite", help="Generate a product image suite (POST /product/suite)") p.add_argument("--image-url", required=True, dest="product_image_url", help="Product image S3 key/URL") p.add_argument("--layout-types", default=None, help="Comma-separated layout type ids") p.add_argument("--count-per-type", type=int, default=None) p.add_argument("--custom-suggestions", default=None, help="JSON array") p.add_argument("--product-info", default=None, help="Free-text product description/facts") p.add_argument("--aspect-ratio", default=None, choices=["1:1", "3:4", "4:3", "9:16", "16:9"]) p.add_argument("--gen-language", default=None) p.add_argument("--model", default=None) p.add_argument("--same-style-reference", default=None, help="JSON object (from studio-analyze --target reference)") p.add_argument("--template-ids", default=None, help="Comma-separated layout template ids (from studio-layouts) — sent as custom_suggestions like the UI's manual template pick") p.add_argument("--layout-template-ids", default=None, help="Comma-separated layout template ids sent as layout_template_ids — the LEAN way: backend resolves layout_prompt+template image from the id (recommended over --template-ids)") p.add_argument("--layout-random-count", type=int, default=None, help="Let the backend randomly pick N layout templates (no manual ids)") p.add_argument("--image-urls", dest="suite_image_urls", default=None, help="Comma-separated extra product view S3 URLs (multi-view input)") p = sub.add_parser("studio-layouts", help="List the layout template library (static asset; query, then select via studio-suite --template-ids)") p.add_argument("--module", default="product", choices=["product", "clothing"]) p.add_argument("--type", dest="type_filter", default=None, help="Filter by type: selling_point, aplus, white_bg, scene, closeup, size_chart …") p = sub.add_parser("studio-clothing-suite", help="Generate a model/clothing image suite (POST /model/clothing-suite)") p.add_argument("--image-urls", required=True, dest="clothing_image_urls", help="Comma-separated garment image S3 keys/URLs (up to 5 used)") p.add_argument("--main-image-types", default=None, help='JSON object, e.g. \'{"model_shot":2,"selling_point":1}\'') p.add_argument("--aplus-types", default=None, help='JSON object, e.g. \'{"standard_aplus":1}\'') p.add_argument("--product-info", default=None) p.add_argument("--brand-config", default=None, help="JSON object") p.add_argument("--aspect-ratio", default=None, help="Default 3:4") p.add_argument("--gen-language", default=None) p.add_argument("--model", default=None) p.add_argument("--same-style-reference", default=None, help="JSON object") p = sub.add_parser("studio-cancel", help="Cancel a running Image Studio generation (queued images refunded, generating ones charged)") p.add_argument("request_id", help="Image Studio request_id (UUID)") p = sub.add_parser("studio-status", help="Get Image Studio job status (separate id space; key is request_id)") p.add_argument("request_id", help="Image Studio request_id (UUID)") p = sub.add_parser("studio-my-works", help="List past Image Studio generations") p.add_argument("--task-type", default=None) p.add_argument("--limit", type=int, default=20) p.add_argument("--offset", type=int, default=0) p = sub.add_parser("studio-call", help="Generic Image Studio call — any endpoint from `studio-features` not covered by a dedicated command " "(remove-bg, replace, inpaint, erase, handheld, translate, outpaint, recolor, enhance, upscale, " "scene-compose, scene-variation, real-model-swap, mannequin-swap, model-scene-swap, ai-outfit, pose-variation, ai-wearable)") p.add_argument("method", choices=["GET", "POST", "PUT", "DELETE", "PATCH"]) p.add_argument("path", help="Path relative to /api/v1/image-studio, e.g. /product/remove-bg") p.add_argument("--json", dest="json_body", default=None, help="JSON request body") return parser def main(): parser = build_parser() args = parser.parse_args() if not args.command: parser.print_help() sys.exit(0) client = PhosorClient(base_url=args.base_url, api_key=args.api_key, allow_http=(args.allow_http or None)) pending = PendingManager() cmd = args.command # ── Offline commands ──────────────────────────────────────────────── if cmd == "models": _json_out(MODELS) if cmd == "studio-features": _json_out(IMAGE_STUDIO) if cmd == "list": _json_out(pending.list_pending()) # ── Commands requiring API key ────────────────────────────────────── if cmd == "check-key": if not client.api_key: _error_out("PHOSOR_API_KEY not set") # 401/403 must be reported as "invalid" -- the whole point of check-key is a # one-line verdict on the key, not a traceback. try: resp = client.check_key() except HTTPError as e: if e.code in (401, 403): _json_out({"status": "invalid", "http": e.code, "hint": "PHOSOR_API_KEY is missing, wrong, or revoked."}) return raise _json_out({"status": "valid", "loras_total": resp.get("total", resp.get("count"))}) if cmd == "submit": kwargs = {} for k in _SUBMIT_FIELDS: if k == "loras": continue # parsed from its own JSON string just below v = getattr(args, k, None) if v is not None: kwargs[k] = v # reference_*_urls arrive as comma-separated strings; the API wants arrays for k in ("reference_image_urls", "reference_video_urls", "reference_audio_urls"): if isinstance(kwargs.get(k), str): kwargs[k] = [u.strip() for u in kwargs[k].split(",") if u.strip()] # Parse --loras JSON string loras_str = getattr(args, "loras", None) if loras_str: try: kwargs["loras"] = json.loads(loras_str) except json.JSONDecodeError: _error_out("--loras must be valid JSON array, e.g. '[{\"lora_id\":\"...\",\"lora_scale\":1.0}]'") resp = client.submit(args.prompt, **kwargs) # Auto-track pending — detect job type from model rid = resp.get("request_id") or resp.get("requested_id") if rid: _model = kwargs.get("model", "") _IMAGE_MODEL_PREFIXES = ("qwen-image/", "z-image/") if any(_model.startswith(p) for p in _IMAGE_MODEL_PREFIXES): job_type = "i2i" if "image-to-image" in _model else "t2i" elif "speech-to-video" in _model: job_type = "s2v" elif "animate" in _model: job_type = "animate" else: job_type = "i2v" if kwargs.get("image_url") else "t2v" pending.add(rid, job_type, {"prompt": args.prompt[:80]}) _json_out(resp) if cmd == "submit-tts": kw = {k: getattr(args, k, None) for k in ("speaker", "language", "seed", "temperature", "top_p", "top_k", "repetition_penalty", "model")} resp = client.submit_tts(args.text, **kw) rid = resp.get("request_id") if rid: pending.add(rid, "inference", {"text": args.text[:80]}) _json_out(resp) if cmd == "status": _json_out(client.status(args.request_id)) if cmd == "result": _json_out(client.result(args.request_id)) if cmd == "poll": _json_out(pending.poll_all(client)) if cmd == "history": _json_out(client.history(limit=args.limit)) # ── Storage ───────────────────────────────────────────────────────── if cmd == "upload-image": _json_out(client.upload_image(args.file)) if cmd == "import-image": _json_out(client.import_image(args.url, filename=args.filename)) if cmd == "upload-lora": _json_out(client.upload_lora(args.high_noise_file, args.low_noise_file, name=args.name)) if cmd == "import-lora": _json_out(client.import_lora(args.high_noise_url, args.low_noise_url, name=args.name)) # ── LoRA Management ───────────────────────────────────────────────── if cmd == "save-lora": _json_out(client.save_lora(args.lora_id, name=args.name)) if cmd == "loras": _json_out(client.loras(limit=args.limit, offset=args.offset)) if cmd == "lora-status": _json_out(client.lora_status(args.lora_id)) if cmd == "delete-lora": _json_out(client.delete_lora(args.lora_id)) # ── Utility ───────────────────────────────────────────────────────── # ── Image Studio (separate product surface) ────────────────────────── if cmd == "studio-pricing": _json_out(client.studio_pricing()) if cmd == "studio-analyze": _json_out(client.studio_analyze( args.target, image_url=args.image_url, prompt=args.prompt, url=args.url, language=args.language, )) if cmd == "studio-layouts": tpls = client.studio_layouts(args.module, type_filter=args.type_filter) _json_out([{"id": t.get("id"), "name": t.get("name"), "name_en": t.get("name_en"), "type": t.get("type"), "platform": t.get("platform"), "random_safe": t.get("random_safe"), "file": t.get("file")} for t in tpls]) return if cmd == "studio-suite": layout_types = [t.strip() for t in args.layout_types.split(",") if t.strip()] if args.layout_types else None custom_suggestions = _parse_json_arg(args.custom_suggestions, "--custom-suggestions", list) or [] if getattr(args, "template_ids", None): ids = [x.strip() for x in args.template_ids.split(",") if x.strip()] custom_suggestions = custom_suggestions + client.layout_templates_to_suggestions(ids) custom_suggestions = custom_suggestions or None same_style_reference = _parse_json_arg(args.same_style_reference, "--same-style-reference", dict) layout_template_ids = [x.strip() for x in args.layout_template_ids.split(",") if x.strip()] if getattr(args, "layout_template_ids", None) else None suite_image_urls = [u.strip() for u in args.suite_image_urls.split(",") if u.strip()] if getattr(args, "suite_image_urls", None) else None resp = client.studio_suite( args.product_image_url, layout_types=layout_types, count_per_type=args.count_per_type, custom_suggestions=custom_suggestions, product_info=args.product_info, aspect_ratio=args.aspect_ratio, gen_language=args.gen_language, model=args.model, same_style_reference=same_style_reference, layout_template_ids=layout_template_ids, layout_random_count=getattr(args, "layout_random_count", None), image_urls=suite_image_urls, ) jid = resp.get("request_id") # the server returns request_id (job_id is never set) if jid: pending.add(jid, "image-studio", {"feature": "product/suite"}) _json_out(resp) if cmd == "studio-clothing-suite": clothing_image_urls = [u.strip() for u in args.clothing_image_urls.split(",") if u.strip()] main_image_types = _parse_json_arg(args.main_image_types, "--main-image-types", dict) aplus_types = _parse_json_arg(args.aplus_types, "--aplus-types", dict) brand_config = _parse_json_arg(args.brand_config, "--brand-config", dict) same_style_reference = _parse_json_arg(args.same_style_reference, "--same-style-reference", dict) resp = client.studio_clothing_suite( clothing_image_urls, main_image_types=main_image_types, aplus_types=aplus_types, product_info=args.product_info, brand_config=brand_config, aspect_ratio=args.aspect_ratio, gen_language=args.gen_language, model=args.model, same_style_reference=same_style_reference, ) jid = resp.get("request_id") # the server returns request_id (job_id is never set) if jid: pending.add(jid, "image-studio", {"feature": "model/clothing-suite"}) _json_out(resp) if cmd == "studio-status": _json_out(client.studio_status(args.request_id)) if cmd == "studio-cancel": _json_out(client.studio_cancel(args.request_id)) if cmd == "studio-my-works": _json_out(client.studio_my_works(task_type=args.task_type, limit=args.limit, offset=args.offset)) if cmd == "studio-call": body = _parse_json_arg(args.json_body, "--json", (dict, list)) resp = client.studio_call(args.method, args.path, json_data=body) jid = resp.get("request_id") if isinstance(resp, dict) else None if jid: pending.add(jid, "image-studio", {"feature": args.path.strip("/")}) _json_out(resp) if __name__ == "__main__": main()