import gradio as gr import asyncio from threading import RLock from pathlib import Path from huggingface_hub import InferenceClient server_timeout = 600 inference_timeout = 300 lock = RLock() loaded_models = {} model_info_dict = {} def to_list(s): return [x.strip() for x in s.split(",")] def list_sub(a, b): return [e for e in a if e not in b] def list_uniq(l): return sorted(set(l), key=l.index) def is_repo_name(s): import re return re.fullmatch(r'^[^/]+?/[^/]+?$', s) def find_model_list(author: str="", tags: list[str]=[], not_tag="", sort: str="last_modified", limit: int=30): from huggingface_hub import HfApi api = HfApi() default_tags = ["diffusers"] if not sort: sort = "last_modified" models = [] try: model_infos = api.list_models(author=author, pipeline_tag="text-to-image", tags=list_uniq(default_tags + tags), cardData=True, sort=sort, limit=limit * 5) except Exception as e: print(f"Error: Failed to list models.") print(e) return models for model in model_infos: if not model.private and not model.gated: if not_tag and not_tag in model.tags: continue models.append(model.id) if len(models) == limit: break return models def get_t2i_model_info_dict(repo_id: str): from huggingface_hub import HfApi api = HfApi() info = {"md": "None"} try: if not is_repo_name(repo_id) or not api.repo_exists(repo_id=repo_id): return info model = api.model_info(repo_id=repo_id) except Exception as e: print(f"Error: Failed to get {repo_id}'s info.") print(e) return info if model.private or model.gated: return info try: tags = model.tags except Exception as e: print(e) return info if not 'diffusers' in model.tags: return info if 'diffusers:StableDiffusionXLPipeline' in tags: info["ver"] = "SDXL" elif 'diffusers:StableDiffusionPipeline' in tags: info["ver"] = "SD1.5" elif 'diffusers:StableDiffusion3Pipeline' in tags: info["ver"] = "SD3" else: info["ver"] = "Other" info["url"] = f"https://huggingface.co/{repo_id}/" info["tags"] = model.card_data.tags if model.card_data and model.card_data.tags else [] info["downloads"] = model.downloads info["likes"] = model.likes info["last_modified"] = model.last_modified.strftime("lastmod: %Y-%m-%d") un_tags = ['text-to-image', 'stable-diffusion', 'stable-diffusion-api', 'safetensors', 'stable-diffusion-xl'] descs = [info["ver"]] + list_sub(info["tags"], un_tags) + [f'DLs: {info["downloads"]}'] + [f'❤: {info["likes"]}'] + [info["last_modified"]] info["md"] = f'Model Info: {", ".join(descs)} [Model Repo]({info["url"]})' return info def rename_image(image_path: str | None, model_name: str, save_path: str | None = None): from PIL import Image from datetime import datetime, timezone, timedelta if image_path is None: return None dt_now = datetime.now(timezone(timedelta(hours=9))) filename = f"{model_name.split('/')[-1]}_{dt_now.strftime('%Y%m%d_%H%M%S')}.png" try: if Path(image_path).exists(): png_path = "image.png" Image.open(image_path).convert('RGBA').save(png_path, "PNG") if save_path is not None: new_path = str(Path(png_path).resolve().rename(Path(save_path).resolve())) else: new_path = str(Path(png_path).resolve().rename(Path(filename).resolve())) return new_path else: return None except Exception as e: print(e) return None def save_gallery(image_path: str | None, images: list[tuple] | None): if images is None: images = [] files = [i[0] for i in images] if image_path is None: return images, files files.insert(0, str(image_path)) images.insert(0, (str(image_path), Path(image_path).stem)) return images, files # https://github.com/gradio-app/gradio/blob/main/gradio/external.py # https://huggingface.co/docs/huggingface_hub/package_reference/inference_client def load_from_model(model_name: str, hf_token: str = None): import httpx import huggingface_hub from gradio.exceptions import ModelNotFoundError model_url = f"https://huggingface.co/{model_name}" api_url = f"https://api-inference.huggingface.co/models/{model_name}" print(f"Fetching model from: {model_url}") headers = {"Authorization": f"Bearer {hf_token}"} if hf_token is not None else {} response = httpx.request("GET", api_url, headers=headers) if response.status_code != 200: raise ModelNotFoundError( f"Could not find model: {model_name}. If it is a private or gated model, please provide your Hugging Face access token (https://huggingface.co/settings/tokens) as the argument for the `hf_token` parameter." ) headers["X-Wait-For-Model"] = "true" client = huggingface_hub.InferenceClient(model=model_name, headers=headers, token=hf_token, timeout=server_timeout) inputs = gr.components.Textbox(label="Input") outputs = gr.components.Image(label="Output") fn = client.text_to_image def query_huggingface_inference_endpoints(*data, **kwargs): return fn(*data, **kwargs) interface_info = { "fn": query_huggingface_inference_endpoints, "inputs": inputs, "outputs": outputs, "title": model_name, } return gr.Interface(**interface_info) def load_model(model_name: str): global loaded_models global model_info_dict if model_name in loaded_models.keys(): return loaded_models[model_name] try: loaded_models[model_name] = load_from_model(model_name) print(f"Loaded: {model_name}") except Exception as e: if model_name in loaded_models.keys(): del loaded_models[model_name] print(f"Failed to load: {model_name}") print(e) return None try: model_info_dict[model_name] = get_t2i_model_info_dict(model_name) print(f"Assigned: {model_name}") except Exception as e: if model_name in model_info_dict.keys(): del model_info_dict[model_name] print(f"Failed to assigned: {model_name}") print(e) return loaded_models[model_name] def load_model_api(model_name: str): global loaded_models global model_info_dict if model_name in loaded_models.keys(): return loaded_models[model_name] try: client = InferenceClient(timeout=5) status = client.get_model_status(model_name) if status is None or status.framework != "diffusers" or status.state not in ["Loadable", "Loaded"]: print(f"Failed to load by API: {model_name}") return None else: loaded_models[model_name] = InferenceClient(model_name, timeout=server_timeout) print(f"Loaded by API: {model_name}") except Exception as e: if model_name in loaded_models.keys(): del loaded_models[model_name] print(f"Failed to load by API: {model_name}") print(e) return None try: model_info_dict[model_name] = get_t2i_model_info_dict(model_name) print(f"Assigned by API: {model_name}") except Exception as e: if model_name in model_info_dict.keys(): del model_info_dict[model_name] print(f"Failed to assigned by API: {model_name}") print(e) return loaded_models[model_name] def load_models(models: list): for model in models: load_model(model) positive_prefix = { "Pony": to_list("score_9, score_8_up, score_7_up"), "Pony Anime": to_list("source_anime, anime, score_9, score_8_up, score_7_up"), } positive_suffix = { "Common": to_list("highly detailed, masterpiece, best quality, very aesthetic, absurdres"), "Anime": to_list("anime artwork, anime style, studio anime, highly detailed"), } negative_prefix = { "Pony": to_list("score_6, score_5, score_4"), "Pony Anime": to_list("score_6, score_5, score_4, source_pony, source_furry, source_cartoon"), "Pony Real": to_list("score_6, score_5, score_4, source_anime, source_pony, source_furry, source_cartoon"), } negative_suffix = { "Common": to_list("lowres, (bad), bad hands, bad feet, text, error, fewer, extra, missing, worst quality, jpeg artifacts, low quality, watermark, unfinished, displeasing, oldest, early, chromatic aberration, signature, extra digits, artistic error, username, scan, [abstract]"), "Pony Anime": to_list("busty, ugly face, mutated hands, low res, blurry face, black and white, the simpsons, overwatch, apex legends"), "Pony Real": to_list("ugly, airbrushed, simple background, cgi, cartoon, anime"), } positive_all = negative_all = [] for k, v in (positive_prefix | positive_suffix).items(): positive_all = positive_all + v + [s.replace("_", " ") for s in v] positive_all = list_uniq(positive_all) for k, v in (negative_prefix | negative_suffix).items(): negative_all = negative_all + v + [s.replace("_", " ") for s in v] positive_all = list_uniq(positive_all) def recom_prompt(prompt: str = "", neg_prompt: str = "", pos_pre: list = [], pos_suf: list = [], neg_pre: list = [], neg_suf: list = []): def flatten(src): return [item for row in src for item in row] prompts = to_list(prompt) neg_prompts = to_list(neg_prompt) prompts = list_sub(prompts, positive_all) neg_prompts = list_sub(neg_prompts, negative_all) last_empty_p = [""] if not prompts and type != "None" else [] last_empty_np = [""] if not neg_prompts and type != "None" else [] prefix_ps = flatten([positive_prefix.get(s, []) for s in pos_pre]) suffix_ps = flatten([positive_suffix.get(s, []) for s in pos_suf]) prefix_nps = flatten([negative_prefix.get(s, []) for s in neg_pre]) suffix_nps = flatten([negative_suffix.get(s, []) for s in neg_suf]) prompt = ", ".join(list_uniq(prefix_ps + prompts + suffix_ps) + last_empty_p) neg_prompt = ", ".join(list_uniq(prefix_nps + neg_prompts + suffix_nps) + last_empty_np) return prompt, neg_prompt recom_prompt_type = { "None": ([], [], [], []), "Auto": ([], [], [], []), "Common": ([], ["Common"], [], ["Common"]), "Animagine": ([], ["Common", "Anime"], [], ["Common"]), "Pony": (["Pony"], ["Common"], ["Pony"], ["Common"]), "Pony Anime": (["Pony", "Pony Anime"], ["Common", "Anime"], ["Pony", "Pony Anime"], ["Common", "Pony Anime"]), "Pony Real": (["Pony"], ["Common"], ["Pony", "Pony Real"], ["Common", "Pony Real"]), } enable_auto_recom_prompt = False def insert_recom_prompt(prompt: str = "", neg_prompt: str = "", type: str = "None"): global enable_auto_recom_prompt if type == "Auto": enable_auto_recom_prompt = True else: enable_auto_recom_prompt = False pos_pre, pos_suf, neg_pre, neg_suf = recom_prompt_type.get(type, ([], [], [], [])) return recom_prompt(prompt, neg_prompt, pos_pre, pos_suf, neg_pre, neg_suf) def set_recom_prompt_preset(type: str = "None"): pos_pre, pos_suf, neg_pre, neg_suf = recom_prompt_type.get(type, ([], [], [], [])) return pos_pre, pos_suf, neg_pre, neg_suf def get_recom_prompt_type(): type = list(recom_prompt_type.keys()) type.remove("Auto") return type def get_positive_prefix(): return list(positive_prefix.keys()) def get_positive_suffix(): return list(positive_suffix.keys()) def get_negative_prefix(): return list(negative_prefix.keys()) def get_negative_suffix(): return list(negative_suffix.keys()) def get_tag_type(pos_pre: list = [], pos_suf: list = [], neg_pre: list = [], neg_suf: list = []): tag_type = "danbooru" words = pos_pre + pos_suf + neg_pre + neg_suf for word in words: if "Pony" in word: tag_type = "e621" break return tag_type def get_model_info_md(model_name: str): if model_name in model_info_dict.keys(): return model_info_dict[model_name].get("md", "") def change_model(model_name: str): load_model_api(model_name) return get_model_info_md(model_name) def warm_model(model_name: str): model = load_model_api(model_name) if model: try: print(f"Warming model: {model_name}") infer_body(model, " ") except Exception as e: print(e) # https://huggingface.co/docs/api-inference/detailed_parameters # https://huggingface.co/docs/huggingface_hub/package_reference/inference_client def infer_body(client: InferenceClient | gr.Interface, prompt: str, neg_prompt: str | None = None, height: int | None = None, width: int | None = None, steps: int | None = None, cfg: int | None = None): png_path = "image.png" kwargs = {} if height is not None and height >= 256: kwargs["height"] = height if width is not None and width >= 256: kwargs["width"] = width if steps is not None and steps >= 1: kwargs["num_inference_steps"] = steps if cfg is not None and cfg > 0: cfg = kwargs["guidance_scale"] = cfg try: if isinstance(client, InferenceClient): image = client.text_to_image(prompt=prompt, negative_prompt=neg_prompt, **kwargs) elif isinstance(client, gr.Interface): image = client.fn(prompt=prompt, negative_prompt=neg_prompt, **kwargs) else: return None image.save(png_path) return str(Path(png_path).resolve()) except Exception as e: print(e) return None async def infer(model_name: str, prompt: str, neg_prompt: str | None = None, height: int | None = None, width: int | None = None, steps: int | None = None, cfg: int | None = None, save_path: str | None = None, timeout: float = inference_timeout): import random noise = "" rand = random.randint(1, 500) for i in range(rand): noise += " " model = load_model(model_name) if not model: return None task = asyncio.create_task(asyncio.to_thread(infer_body, model, f"{prompt} {noise}", neg_prompt, height, width, steps, cfg)) await asyncio.sleep(0) try: result = await asyncio.wait_for(task, timeout=timeout) except (Exception, asyncio.TimeoutError) as e: print(e) print(f"Task timed out: {model_name}") if not task.done(): task.cancel() result = None if task.done() and result is not None: with lock: image = rename_image(result, model_name, save_path) return image return None def infer_fn(model_name: str, prompt: str, neg_prompt: str | None = None, height: int | None = None, width: int | None = None, steps: int | None = None, cfg: int | None = None, pos_pre: list = [], pos_suf: list = [], neg_pre: list = [], neg_suf: list = [], save_path: str | None = None): if model_name == 'NA': return None try: prompt, neg_prompt = recom_prompt(prompt, neg_prompt, pos_pre, pos_suf, neg_pre, neg_suf) loop = asyncio.new_event_loop() result = loop.run_until_complete(infer(model_name, prompt, neg_prompt, height, width, steps, cfg, save_path, inference_timeout)) except (Exception, asyncio.CancelledError) as e: print(e) print(f"Task aborted: {model_name}") result = None finally: loop.close() return result def infer_rand_fn(model_name_dummy: str, prompt: str, neg_prompt: str | None = None, height: int | None = None, width: int | None = None, steps: int | None = None, cfg: int | None = None, pos_pre: list = [], pos_suf: list = [], neg_pre: list = [], neg_suf: list = [], save_path: str | None = None): import random if model_name_dummy == 'NA': return None random.seed() model_name = random.choice(list(loaded_models.keys())) try: prompt, neg_prompt = recom_prompt(prompt, neg_prompt, pos_pre, pos_suf, neg_pre, neg_suf) loop = asyncio.new_event_loop() result = loop.run_until_complete(infer(model_name, prompt, neg_prompt, height, width, steps, cfg, save_path, inference_timeout)) except (Exception, asyncio.CancelledError) as e: print(e) print(f"Task aborted: {model_name}") result = None finally: loop.close() return result