Merge pull request #28 from cdb-boop/main

File download scans and urls

README.md

@@ -25,6 +25,7 @@ Designed to support desktop, mobile and multi-screen devices.
 - Press the "copy" button to copy a model to ComfyUI's clipboard or copy the embedding to the system clipboard. (Copying the embedding to the system clipboard requires a secure http connection.)
 - Press the "add" button to add the model to the ComfyUI graph or append the embedding to one or more selected nodes.
 - Press the "load workflow" button to try and load a workflow embedded in a model's preview image.
+- Press the "open model url" button to try and search the web and open a model's webpage.
 
 ### Download Tab
 
@@ -55,6 +56,7 @@ Designed to support desktop, mobile and multi-screen devices.
 - Read, edit and save notes. (Saved as a `.txt` file beside the model).
   - `Ctrl+s` or `⌘+S` to save a note when the textarea is in focus.
   - Autosave can be enabled in settings. (Note: Once the model info view is closed, the undo history is lost.)
+  - Automatically search the web for model info and save as notes with a single button.
 - Change or remove a model's preview image.
 - View training tags and use the random tag generator to generate prompt ideas. (Inspired by the one in A1111.)
 

__init__.py

@@ -8,6 +8,8 @@ import copy
 import importlib
 import re
 import base64
+import hashlib
+import markdownify
 
 from aiohttp import web
 import server
@@ -23,7 +25,7 @@ import folder_paths
 
 comfyui_model_uri = folder_paths.models_dir
 
-extension_uri = os.path.dirname(__file__)
+extension_uri = os.path.dirname(os.path.abspath(__file__))
 
 config_loader_path = os.path.join(extension_uri, 'config_loader.py')
 config_loader_spec = importlib.util.spec_from_file_location('config_loader', config_loader_path)
@@ -58,7 +60,8 @@ preview_extensions = ( # TODO: JavaScript does not know about this (x2 states)
     image_extensions + # order matters
     stable_diffusion_webui_civitai_helper_image_extensions
 )
-model_info_extension = ".txt"
+model_notes_extension = ".txt"
+model_info_extension = ".json"
 #video_extensions = (".avi", ".mp4", ".webm") # TODO: Requires ffmpeg or cv2. Cache preview frame?
 
 def split_valid_ext(s, *arg_exts):
@@ -71,6 +74,7 @@ def split_valid_ext(s, *arg_exts):
 
 _folder_names_and_paths = None # dict[str, tuple[list[str], list[str]]]
 def folder_paths_folder_names_and_paths(refresh = False):
+    # TODO: "diffusers" extension whitelist is ["folder"]
     global _folder_names_and_paths
     if refresh or _folder_names_and_paths is None:
         _folder_names_and_paths = {}
@@ -189,6 +193,7 @@ def ui_rules():
         Rule("model-show-add-button", True, bool),
         Rule("model-show-copy-button", True, bool),
         Rule("model-show-load-workflow-button", True, bool),
+        Rule("model-show-open-model-url-button", False, bool),
         Rule("model-info-button-on-left", False, bool),
         Rule("model-buttons-only-on-hover", False, bool),
         
@@ -232,6 +237,7 @@ def get_def_headers(url=""):
     if url.startswith("https://civitai.com/"):
         api_key = server_settings["civitai_api_key"]
         if (api_key != ""):
+            def_headers["Content-Type"] = f"application/json"
             def_headers["Authorization"] = f"Bearer {api_key}"
             url += "&" if "?" in url else "?" # not the most robust solution
             url += f"token={api_key}" # TODO: Authorization didn't work in the header
@@ -243,6 +249,246 @@ def get_def_headers(url=""):
     return def_headers
 
 
+def save_web_url(path, url):
+    with open(path, "w", encoding="utf-8") as f:
+        f.write(f"[InternetShortcut]\nURL={url}\n")
+
+
+def try_load_web_url(path):
+    with open(path, "r", encoding="utf-8") as f:
+        if f.readline() != "[InternetShortcut]\n": return ""
+        url = f.readline()
+        if not url.startswith("URL="): return ""
+        if not url.endswith("\n"): return ""
+        return url[4:len(url)-1]
+
+
+def hash_file(path, buffer_size=1024*1024):
+    sha256 = hashlib.sha256()
+    with open(path, 'rb') as f:
+        while True:
+            data = f.read(buffer_size)
+            if not data: break
+            sha256.update(data)
+    return sha256.hexdigest()
+
+
+class Civitai:
+    IMAGE_URL_SUBDIRECTORY_PREFIX = "https://civitai.com/images/"
+    IMAGE_URL_DOMAIN_PREFIX = "'https://image.civitai.com/"
+
+    @staticmethod
+    def image_subdirectory_url_to_image_url(image_url):
+        url_suffix = image_url[len(Civitai.IMAGE_URL_SUBDIRECTORY_PREFIX):]
+        image_id = re.search(r"^\d+", url_suffix).group(0)
+        image_id = str(int(image_id))
+        image_info_url = f"https://civitai.com/api/v1/images?imageId={image_id}"
+        def_headers = get_def_headers(image_info_url)
+        response = requests.get(
+            url=image_info_url, 
+            stream=False, 
+            verify=False, 
+            headers=def_headers, 
+            proxies=None, 
+            allow_redirects=False, 
+        )
+        if response.ok:
+            #content_type = response.headers.get("Content-Type")
+            info = response.json()
+            items = info["items"]
+            if len(items) == 0:
+                raise RuntimeError("Civitai /api/v1/images returned 0 items!")
+            return items[0]["url"]
+        else:
+            raise RuntimeError("Bad response from api/v1/images!")
+
+    @staticmethod
+    def image_domain_url_full_size(url, width = None):
+        result = re.search("/width=(\d+)", url)
+        if width is None:
+            i0 = result.span()[0]
+            i1 = result.span()[1]
+            return url[0:i0] + url[i1:]
+        else:
+            w = int(result.group(1))
+            return url.replace(str(w), str(width))
+
+    @staticmethod
+    def search_by_hash(sha256_hash):
+        url_api_hash = r"https://civitai.com/api/v1/model-versions/by-hash/" + sha256_hash
+        hash_response = requests.get(url_api_hash)
+        if hash_response.status_code != 200:
+            return {}
+        return hash_response.json() # model version info
+
+    @staticmethod
+    def search_by_model_id(model_id):
+        url_api_model = r"https://civitai.com/api/v1/models/" + str(model_id)
+        model_response = requests.get(url_api_model)
+        if model_response.status_code != 200:
+            return {}
+        return model_response.json() # model group info
+
+    @staticmethod
+    def get_model_url(model_version_info):
+        if len(model_version_info) == 0: return ""
+        model_id = model_version_info.get("modelId")
+        if model_id is None:
+            # there can be incomplete model info, so don't throw just in case
+            return ""
+        url = f"https://civitai.com/models/{model_id}"
+        version_id = model_version_info.get("id")
+        if version_id is not None:
+            url += f"?modelVersionId={version_id}"
+        return url
+
+    @staticmethod
+    def get_preview_urls(model_version_info, full_size=False):
+        images = model_version_info.get("images", None)
+        if images is None:
+            return []
+        preview_urls = []
+        for image_info in images:
+            url = image_info["url"]
+            if full_size:
+                url = Civitai.image_domain_url_full_size(url, image_info.get("width", None))
+            preview_urls.append(url)
+        return preview_urls
+
+    @staticmethod
+    def search_notes(model_version_info):
+        if len(model_version_info) == 0:
+            return ""
+        model_name = None
+        if "modelId" in model_version_info and "id" in model_version_info:
+            model_id = model_version_info.get("modelId")
+            model_version_id = model_version_info.get("id")
+
+            model_version_description = ""
+            model_trigger_words = []
+            model_info = Civitai.search_by_model_id(model_id)
+            if len(model_info) == 0: # can happen if model download is disabled
+                print("Model Manager WARNING: Unable to find Civitai 'modelId' " + str(model_id) + ". Try deleting .json file and trying again later!")
+                return ""
+            model_name = model_info.get("name")
+            model_description = model_info.get("description")
+            for model_version in model_info["modelVersions"]:
+                if model_version["id"] == model_version_id:
+                    model_version_description = model_version.get("description")
+                    model_trigger_words = model_version.get("trainedWords")
+                    break
+        elif "description" in model_version_info and "activation text" in model_version_info and "notes" in model_version_info:
+            # {'description': str, 'sd version': str, 'activation text': str, 'preferred weight': int, 'notes': str}
+            model_description = model_version_info.get("description")
+            activation_text = model_version_info.get("activation text")
+            if activation_text != "":
+                model_trigger_words = [activation_text]
+            else:
+                model_trigger_words = []
+            model_version_description = model_version_info.get("notes")
+        else:
+            return ""
+        model_description = model_description if model_description is not None else ""
+        model_trigger_words = model_trigger_words if model_trigger_words is not None else []
+        model_version_description = model_version_description if model_version_description is not None else ""
+        model_name = model_name if model_name is not None else "Model Description"
+
+        notes = ""
+        if len(model_trigger_words) > 0:
+            notes += "# Trigger Words\n\n"
+            model_trigger_words = [re.sub(",$", "", s.strip()) for s in model_trigger_words]
+            join_separator = ', '
+            for s in model_trigger_words:
+                if ',' in s:
+                    join_separator = '\n'
+                    break
+            if join_separator == '\n':
+                model_trigger_words = ["* " + s for s in model_trigger_words]
+            notes += join_separator.join(model_trigger_words)
+        if model_version_description != "":
+            if len(notes) > 0: notes += "\n\n"
+            notes += "# About this version\n\n"
+            notes += markdownify.markdownify(model_version_description)
+        if model_description != "":
+            if len(notes) > 0: notes += "\n\n"
+            notes += "# " + model_name + "\n\n"
+            notes += markdownify.markdownify(model_description)
+        return notes.strip()
+
+
+class ModelInfo:
+    @staticmethod
+    def search_by_hash(sha256_hash):
+        model_info = Civitai.search_by_hash(sha256_hash)
+        if len(model_info) > 0: return model_info
+        # TODO: search other websites
+        return {}
+
+    @staticmethod
+    def try_load_cached(model_path):
+        model_info_path = os.path.splitext(model_path)[0] + model_info_extension
+        if os.path.isfile(model_info_path):
+            with open(model_info_path, "r", encoding="utf-8") as f:
+                model_info = json.load(f)
+                return model_info
+        return {}
+
+    @staticmethod
+    def get_hash(model_info):
+        model_info = Civitai.get_hash(model_info)
+        if len(model_info) > 0: return model_info
+        # TODO: search other websites
+        return {}
+
+    @staticmethod
+    def search_info(model_path, cache=True, use_cached=True):
+        model_info = ModelInfo.try_load_cached(model_path)
+        if use_cached and len(model_info) > 0:
+            return model_info
+
+        sha256_hash = hash_file(model_path)
+        model_info = ModelInfo.search_by_hash(sha256_hash)
+        if cache and len(model_info) > 0:
+            model_info_path = os.path.splitext(model_path)[0] + model_info_extension
+            with open(model_info_path, "w", encoding="utf-8") as f:
+                json.dump(model_info, f, indent=4)
+                print("Saved file: " + model_info_path)
+
+        return model_info
+
+    @staticmethod
+    def get_url(model_info):
+        if len(model_info) == 0:
+            return ""
+        model_url = Civitai.get_model_url(model_info)
+        if model_url != "":
+            return model_url
+        # TODO: huggingface has <user>/<model> formats
+        # TODO: support other websites
+        return ""
+
+    @staticmethod
+    def search_notes(model_path):
+        assert(os.path.isfile(model_path))
+        model_info = ModelInfo.search_info(model_path, cache=True, use_cached=True) # assume cached is correct; re-download elsewhere
+        if len(model_info) == 0:
+            return ""
+        notes = Civitai.search_notes(model_info)
+        if len(notes) > 0 and not notes.isspace():
+            return notes
+        # TODO: search other websites
+        return ""
+
+    @staticmethod
+    def get_web_preview_urls(model_info, full_size=False):
+        if len(model_info) == 0:
+            return []
+        preview_urls = Civitai.get_preview_urls(model_info, full_size)
+        if len(preview_urls) > 0:
+            return preview_urls
+        # TODO: support other websites
+        return []
+
 @server.PromptServer.instance.routes.get("/model-manager/timestamp")
 async def get_timestamp(request):
     return web.json_response({ "timestamp": datetime.now().timestamp() })
@@ -327,9 +573,12 @@ def get_auto_thumbnail_format(original_format):
     return "JPEG" # default fallback
 
 
-@server.PromptServer.instance.routes.get("/model-manager/preview/get")
+@server.PromptServer.instance.routes.get("/model-manager/preview/get/{uri}")
 async def get_model_preview(request):
-    uri = request.query.get("uri")
+    uri = request.match_info["uri"]
+    if uri is None: # BUG: this should never happen
+        print(f"Invalid uri! Request url: {request.url}")
+        uri = "no-preview"
     quality = 75
     response_image_format = request.query.get("image-format", None)
     if isinstance(response_image_format, str):
@@ -451,42 +700,16 @@ async def get_image_extensions(request):
     return web.json_response(image_extensions)
 
 
-def download_model_preview(formdata):
-    path = formdata.get("path", None)
-    if type(path) is not str:
+def download_model_preview(path, image, overwrite):
+    if not os.path.isfile(path):
         raise ValueError("Invalid path!")
-    path, model_type = search_path_to_system_path(path)
-    model_type_extensions = folder_paths_get_supported_pt_extensions(model_type)
-    path_without_extension, _ = split_valid_ext(path, model_type_extensions)
+    path_without_extension = os.path.splitext(path)[0]
 
-    overwrite = formdata.get("overwrite", "true").lower()
-    overwrite = True if overwrite == "true" else False
-
-    image = formdata.get("image", None)
     if type(image) is str:
-        civitai_image_url = "https://civitai.com/images/"
-        if image.startswith(civitai_image_url):
-            image_id = re.search(r"^\d+", image[len(civitai_image_url):]).group(0)
-            image_id = str(int(image_id))
-            image_info_url = f"https://civitai.com/api/v1/images?imageId={image_id}"
-            def_headers = get_def_headers(image_info_url)
-            response = requests.get(
-                url=image_info_url, 
-                stream=False, 
-                verify=False, 
-                headers=def_headers, 
-                proxies=None, 
-                allow_redirects=False, 
-            )
-            if response.ok:
-                content_type = response.headers.get("Content-Type")
-                info = response.json()
-                items = info["items"]
-                if len(items) == 0:
-                    raise RuntimeError("Civitai /api/v1/images returned 0 items!")
-                image = items[0]["url"]
-            else:
-                raise RuntimeError("Bad response from api/v1/images!")
+        if image.startswith(Civitai.IMAGE_URL_SUBDIRECTORY_PREFIX):
+            image = Civitai.image_subdirectory_url_to_image_url(image)
+        if image.startswith(Civitai.IMAGE_URL_DOMAIN_PREFIX):
+            image = Civitai.image_domain_url_full_size(image)
         _, image_extension = split_valid_ext(image, image_extensions)
         if image_extension == "":
             raise ValueError("Invalid image type!")
@@ -514,17 +737,23 @@ def download_model_preview(formdata):
 
     # detect (and try to fix) wrong file extension
     image_format = None
-    with Image.open(image_path) as image:
-        image_format = image.format
-    image_dir_and_name, image_ext = os.path.splitext(image_path)
-    if not image_format_is_equal(image_format, image_ext):
-        corrected_image_path = image_dir_and_name + "." + image_format.lower()
-        if os.path.exists(corrected_image_path) and not overwrite:
-            print("WARNING: '" + image_path + "' has wrong extension!")
-        else:
-            os.rename(image_path, corrected_image_path)
-            print("Saved file: " + corrected_image_path)
-            image_path = corrected_image_path
+    try:
+        with Image.open(image_path) as image:
+            image_format = image.format
+        image_dir_and_name, image_ext = os.path.splitext(image_path)
+        if not image_format_is_equal(image_format, image_ext):
+            corrected_image_path = image_dir_and_name + "." + image_format.lower()
+            if os.path.exists(corrected_image_path) and not overwrite:
+                print("WARNING: '" + image_path + "' has wrong extension!")
+            else:
+                os.rename(image_path, corrected_image_path)
+                print("Saved file: " + corrected_image_path)
+                image_path = corrected_image_path
+    except Image.UnidentifiedImageError as e: #TODO: handle case where "image" is actually video
+        print("WARNING: '" + image_path + "' image format was unknown!")
+        os.remove(image_path)
+        print("Deleted file: " + image_path)
+        image_path = ""
     return image_path # return in-case need corrected path
 
 
@@ -532,7 +761,15 @@ def download_model_preview(formdata):
 async def set_model_preview(request):
     formdata = await request.post()
     try:
-        download_model_preview(formdata)
+        search_path = formdata.get("path", None)
+        model_path, model_type = search_path_to_system_path(search_path)
+
+        image = formdata.get("image", None)
+
+        overwrite = formdata.get("overwrite", "true").lower()
+        overwrite = True if overwrite == "true" else False
+
+        download_model_preview(model_path, image, overwrite)
         return web.json_response({ "success": True })
     except ValueError as e:
         print(e, file=sys.stderr, flush=True)
@@ -710,8 +947,8 @@ async def get_model_list(request):
             if image is not None:
                 raw_post = os.path.join(model_type, str(base_path_index), rel_path, image)
                 item["preview"] = {
-                    "path": urllib.parse.quote_plus(raw_post),
-                    "dateModified": urllib.parse.quote_plus(str(image_modified)),
+                    "path": raw_post,
+                    "dateModified": str(image_modified),
                 }
             model_items.append(item)
 
@@ -779,6 +1016,116 @@ async def get_directory_list(request):
     return web.json_response(dir_list)
 
 
+def try_download_and_save_model_info(model_file_path):
+    success = (0, 0, 0) #info, notes, url
+    head, _ = os.path.splitext(model_file_path)
+    model_info_path = head + model_info_extension
+    model_notes_path = head + model_notes_extension
+    model_url_path = head + ".url"
+    if os.path.exists(model_info_path) and os.path.exists(model_notes_path) and os.path.exists(model_url_path):
+        return success
+    print("Scanning " + model_file_path)
+
+    model_info = {}
+    model_info = ModelInfo.search_info(model_file_path, cache=True, use_cached=True)
+    if len(model_info) == 0:
+        return success
+    success[0] = 1
+
+    if not os.path.exists(model_notes_path):
+        notes = ModelInfo.search_notes(model_file_path)
+        if not notes.isspace() and notes != "":
+            try:
+                with open(model_notes_path, "w", encoding="utf-8") as f:
+                    f.write(notes)
+                print("Saved file: " + model_notes_path)
+                success[1] = 1
+            except Exception as e:
+                print(f"Failed to save {model_notes_path}!")
+                print(e, file=sys.stderr, flush=True)
+
+    if not os.path.exists(model_url_path):
+        web_url = ModelInfo.get_url(model_info)
+        if web_url is not None and web_url != "":
+            try:
+                save_web_url(model_url_path, web_url)
+                print("Saved file: " + model_url_path)
+                success[2] = 1
+            except Exception as e:
+                print(f"Failed to save {model_url_path}!")
+                print(e, file=sys.stderr, flush=True)
+    return success
+
+
+@server.PromptServer.instance.routes.post("/model-manager/models/scan")
+async def try_scan_download(request):
+    refresh = request.query.get("refresh", None) is not None
+    response = {
+        "success": False,
+        "infoCount": 0,
+        "notesCount": 0,
+        "urlCount": 0,
+    }
+    model_paths = folder_paths_folder_names_and_paths(refresh)
+    for _, (model_dirs, model_extension_whitelist) in model_paths.items():
+        for root_dir in model_dirs:
+            for root, dirs, files in os.walk(root_dir):
+                for file in files:
+                    file_name, file_extension = os.path.splitext(file)
+                    if file_extension not in model_extension_whitelist:
+                        continue
+                    model_file_path = root + os.path.sep + file
+                    savedInfo, savedNotes, savedUrl = try_download_and_save_model_info(model_file_path)
+                    response["infoCount"] += savedInfo
+                    response["notesCount"] += savedNotes
+                    response["urlCount"] += savedUrl
+
+    response["success"] = True
+    return web.json_response(response)
+
+@server.PromptServer.instance.routes.post("/model-manager/preview/scan")
+async def try_scan_download_previews(request):
+    refresh = request.query.get("refresh", None) is not None
+    response = {
+        "success": False,
+        "count": 0,
+    }
+    model_paths = folder_paths_folder_names_and_paths(refresh)
+    for _, (model_dirs, model_extension_whitelist) in model_paths.items():
+        for root_dir in model_dirs:
+            for root, dirs, files in os.walk(root_dir):
+                for file in files:
+                    file_name, file_extension = os.path.splitext(file)
+                    if file_extension not in model_extension_whitelist:
+                        continue
+                    model_file_path = root + os.path.sep + file
+                    model_file_head = os.path.splitext(model_file_path)[0]
+
+                    preview_exists = False
+                    for preview_extension in preview_extensions:
+                        preview_path = model_file_head + preview_extension
+                        if os.path.isfile(preview_path):
+                            preview_exists = True
+                            break
+                    if preview_exists:
+                        continue
+
+                    model_info = ModelInfo.try_load_cached(model_file_path) # NOTE: model info must already be downloaded
+                    web_previews = ModelInfo.get_web_preview_urls(model_info, True)
+                    if len(web_previews) == 0:
+                        continue
+                    saved_image_path = download_model_preview(
+                        model_file_path,
+                        image=web_previews[0],
+                        overwrite=False,
+                    )
+                    if os.path.isfile(saved_image_path):
+                        response["count"] += 1
+
+    response["success"] = True
+    return web.json_response(response)
+
+
 def download_file(url, filename, overwrite):
     if not overwrite and os.path.isfile(filename):
         raise ValueError("File already exists!")
@@ -885,13 +1232,13 @@ def bytes_to_size(total_bytes):
     return "{:.2f}".format(total_bytes / (1 << (i * 10))) + " " + units[i]
 
 
-@server.PromptServer.instance.routes.get("/model-manager/model/info")
-async def get_model_info(request):
+@server.PromptServer.instance.routes.get("/model-manager/model/info/{path}")
+async def get_model_metadata(request):
     result = { "success": False }
 
-    model_path = request.query.get("path", None)
+    model_path = request.match_info["path"]
     if model_path is None:
-        result["alert"] = "Missing model path!"
+        result["alert"] = "Invalid model path!"
         return web.json_response(result)
     model_path = urllib.parse.unquote(model_path)
 
@@ -900,16 +1247,16 @@ async def get_model_info(request):
         result["alert"] = "Invalid model path!"
         return web.json_response(result)
 
-    info = {}
+    data = {}
     comfyui_directory, name = os.path.split(model_path)
-    info["File Name"] = name
-    info["File Directory"] = comfyui_directory
-    info["File Size"] = bytes_to_size(os.path.getsize(abs_path))
+    data["File Name"] = name
+    data["File Directory"] = comfyui_directory
+    data["File Size"] = bytes_to_size(os.path.getsize(abs_path))
     stats = pathlib.Path(abs_path).stat()
     date_format = "%Y-%m-%d %H:%M:%S"
     date_modified = datetime.fromtimestamp(stats.st_mtime).strftime(date_format)
-    #info["Date Modified"] = date_modified
-    #info["Date Created"] = datetime.fromtimestamp(stats.st_ctime).strftime(date_format)
+    #data["Date Modified"] = date_modified
+    #data["Date Created"] = datetime.fromtimestamp(stats.st_ctime).strftime(date_format)
 
     model_extensions = folder_paths_get_supported_pt_extensions(model_type)
     abs_name , _ = split_valid_ext(abs_path, model_extensions)
@@ -919,36 +1266,36 @@ async def get_model_info(request):
         if os.path.isfile(maybe_preview):
             preview_path, _ = split_valid_ext(model_path, model_extensions)
             preview_modified = pathlib.Path(maybe_preview).stat().st_mtime_ns
-            info["Preview"] = {
-                "path": urllib.parse.quote_plus(preview_path + extension),
-                "dateModified": urllib.parse.quote_plus(str(preview_modified)),
+            data["Preview"] = {
+                "path": preview_path + extension,
+                "dateModified": str(preview_modified),
             }
             break
 
     header = get_safetensor_header(abs_path)
     metadata = header.get("__metadata__", None)
 
-    if metadata is not None and info.get("Preview", None) is None:
+    if metadata is not None and data.get("Preview", None) is None:
         thumbnail = metadata.get("modelspec.thumbnail")
         if thumbnail is not None:
             i0 = thumbnail.find("/") + 1
             i1 = thumbnail.find(";", i0)
             thumbnail_extension = "." + thumbnail[i0:i1]
             if thumbnail_extension in image_extensions:
-                info["Preview"] = {
+                data["Preview"] = {
                     "path": request.query["path"] + thumbnail_extension,
                     "dateModified": date_modified,
                 }
 
     if metadata is not None:
-        info["Base Training Model"] = metadata.get("ss_sd_model_name", "")
-        info["Base Model Version"] = metadata.get("ss_base_model_version", "")
-        info["Network Dimension"] = metadata.get("ss_network_dim", "")
-        info["Network Alpha"] = metadata.get("ss_network_alpha", "")
+        data["Base Training Model"] = metadata.get("ss_sd_model_name", "")
+        data["Base Model Version"] = metadata.get("ss_base_model_version", "")
+        data["Network Dimension"] = metadata.get("ss_network_dim", "")
+        data["Network Alpha"] = metadata.get("ss_network_alpha", "")
 
     if metadata is not None:
         training_comment = metadata.get("ss_training_comment", "")
-        info["Description"] = (
+        data["Description"] = (
             metadata.get("modelspec.description", "") + 
             "\n\n" + 
             metadata.get("modelspec.usage_hint", "") + 
@@ -956,12 +1303,17 @@ async def get_model_info(request):
             training_comment if training_comment != "None" else ""
         ).strip()
 
-    info_text_file = abs_name + model_info_extension
+    notes_file = abs_name + model_notes_extension
     notes = ""
-    if os.path.isfile(info_text_file):
-        with open(info_text_file, 'r', encoding="utf-8") as f:
+    if os.path.isfile(notes_file):
+        with open(notes_file, 'r', encoding="utf-8") as f:
             notes = f.read()
 
+    web_url_file = abs_name + ".url"
+    web_url = ""
+    if os.path.isfile(web_url_file):
+        web_url = try_load_web_url(web_url_file)
+
     if metadata is not None:
         img_buckets = metadata.get("ss_bucket_info", None)
         datasets = metadata.get("ss_datasets", None)
@@ -983,7 +1335,7 @@ async def get_model_info(request):
                 resolutions[str(x) + "x" + str(y)] = count
         resolutions = list(resolutions.items())
         resolutions.sort(key=lambda x: x[1], reverse=True)
-        info["Bucket Resolutions"] = resolutions
+        data["Bucket Resolutions"] = resolutions
 
     tags = None
     if metadata is not None:
@@ -997,21 +1349,82 @@ async def get_model_info(request):
         tags = list(tags.items())
         tags.sort(key=lambda x: x[1], reverse=True)
 
+    model_info = ModelInfo.try_load_cached(abs_path)
+    web_previews = ModelInfo.get_web_preview_urls(model_info, True)
+
     result["success"] = True
-    result["info"] = info
+    result["info"] = data
     if metadata is not None:
         result["metadata"] = metadata
     if tags is not None:
         result["tags"] = tags
     result["notes"] = notes
+    result["url"] = web_url
+    result["webPreviews"] = web_previews
     return web.json_response(result)
 
 
+@server.PromptServer.instance.routes.get("/model-manager/model/web-url")
+async def get_model_web_url(request):
+    result = { "success": False }
+
+    model_path = request.query.get("path", None)
+    if model_path is None:
+        result["alert"] = "Invalid model path!"
+        return web.json_response(result)
+    model_path = urllib.parse.unquote(model_path)
+
+    abs_path, model_type = search_path_to_system_path(model_path)
+    if abs_path is None:
+        result["alert"] = "Invalid model path!"
+        return web.json_response(result)
+
+    url_path = os.path.splitext(abs_path)[0] + ".url"
+    if os.path.isfile(url_path):
+        web_url = try_load_web_url(url_path)
+        if web_url != "":
+            result["success"] = True
+            return web.json_response({ "url": web_url })
+
+    model_info = ModelInfo.search_info(abs_path)
+    if len(model_info) == 0:
+        result["alert"] = "Unable to find model info!"
+        return web.json_response(result)
+    web_url = ModelInfo.get_url(model_info)
+    if web_url != "" and web_url is not None:
+        save_web_url(url_path, web_url)
+        result["success"] = True
+
+    return web.json_response({ "url": web_url })
+
+
 @server.PromptServer.instance.routes.get("/model-manager/system-separator")
 async def get_system_separator(request):
     return web.json_response(os.path.sep)
 
 
+@server.PromptServer.instance.routes.post("/model-manager/model/download/info")
+async def download_model_info(request):
+    result = { "success": False }
+
+    model_path = request.query.get("path", None)
+    if model_path is None:
+        result["alert"] = "Missing model path!"
+        return web.json_response(result)
+    model_path = urllib.parse.unquote(model_path)
+
+    abs_path, model_type = search_path_to_system_path(model_path)
+    if abs_path is None:
+        result["alert"] = "Invalid model path!"
+        return web.json_response(result)
+
+    model_info = ModelInfo.search_info(abs_path, cache=True, use_cached=False)
+    if len(model_info) > 0:
+        result["success"] = True
+
+    return web.json_response(result)
+
+
 @server.PromptServer.instance.routes.post("/model-manager/model/download")
 async def download_model(request):
     formdata = await request.post()
@@ -1026,6 +1439,7 @@ async def download_model(request):
         result["alert"] = "Invalid save path!"
         return web.json_response(result)
 
+    # download model
     download_uri = formdata.get("download")
     if download_uri is None:
         result["alert"] = "Invalid download url!"
@@ -1049,14 +1463,24 @@ async def download_model(request):
         result["alert"] = "Failed to download model!\n\n" + str(e)
         return web.json_response(result)
 
+    # download model info
+    model_info = ModelInfo.search_info(file_name, cache=True) # save json
+
+    # save url
+    url_file_path = os.path.splitext(file_name)[0] + ".url"
+    url = ModelInfo.get_url(model_info)
+    if url != "" and url is not None:
+        save_web_url(url_file_path, url)
+
+    # save image as model preview
     image = formdata.get("image")
     if image is not None and image != "":
         try:
-            download_model_preview({
-                "path": model_path + os.sep + name,
-                "image": image,
-                "overwrite": formdata.get("overwrite"),
-            })
+            download_model_preview(
+                file_name,
+                image,
+                formdata.get("overwrite"),
+            )
         except Exception as e:
             print(e, file=sys.stderr, flush=True)
             result["alert"] = "Failed to download preview!\n\n" + str(e)
@@ -1122,7 +1546,7 @@ async def move_model(request):
         return web.json_response(result)
 
     # TODO: this could overwrite existing files in destination; do a check beforehand?
-    for extension in preview_extensions + (model_info_extension,):
+    for extension in preview_extensions + (model_notes_extension,) + (model_info_extension,):
         old_file = old_file_without_extension + extension
         if os.path.isfile(old_file):
             new_file = new_file_without_extension + extension
@@ -1176,6 +1600,7 @@ async def delete_model(request):
         print("Deleted file: " + model_path)
 
         delete_same_name_files(path_and_name, preview_extensions)
+        delete_same_name_files(path_and_name, (model_notes_extension,))
         delete_same_name_files(path_and_name, (model_info_extension,))
 
     return web.json_response(result)
@@ -1200,7 +1625,7 @@ async def set_notes(request):
     model_path, model_type = search_path_to_system_path(model_path)
     model_extensions = folder_paths_get_supported_pt_extensions(model_type)
     file_path_without_extension, _ = split_valid_ext(model_path, model_extensions)
-    filename = os.path.normpath(file_path_without_extension + model_info_extension)
+    filename = os.path.normpath(file_path_without_extension + model_notes_extension)
     
     if dt_epoch is not None and os.path.exists(filename) and os.path.getmtime(filename) > dt_epoch:
         # discard late save
@@ -1221,12 +1646,52 @@ async def set_notes(request):
         except ValueError as e:
             print(e, file=sys.stderr, flush=True)
             result["alert"] = "Failed to save notes!\n\n" + str(e)
-            web.json_response(result)
+            return web.json_response(result)
 
     result["success"] = True
     return web.json_response(result)
 
 
+@server.PromptServer.instance.routes.post("/model-manager/notes/download")
+async def try_download_notes(request):
+    result = { "success": False }
+
+    model_path = request.query.get("path", None)
+    if model_path is None:
+        result["alert"] = "Missing model path!"
+        return web.json_response(result)
+    model_path = urllib.parse.unquote(model_path)
+
+    abs_path, model_type = search_path_to_system_path(model_path)
+    if abs_path is None:
+        result["alert"] = "Invalid model path!"
+        return web.json_response(result)
+
+    overwrite = request.query.get("overwrite", None)
+    overwrite = not (overwrite == "False" or overwrite == "false" or overwrite == None)
+    notes_path = os.path.splitext(abs_path)[0] + ".txt"
+    if not overwrite and os.path.isfile(notes_path):
+        result["alert"] = "Notes already exist!"
+        return web.json_response(result)
+
+    notes = ModelInfo.search_notes(abs_path)
+    if notes.isspace() or notes == "":
+        result["alert"] = "No notes found!"
+        return web.json_response(result)
+
+    try:
+        with open(notes_path, "w", encoding="utf-8") as f:
+            f.write(notes)
+        result["success"] = True
+    except ValueError as e:
+        print(e, file=sys.stderr, flush=True)
+        result["alert"] = "Failed to save notes!\n\n" + str(e)
+        return web.json_response(result)
+
+    result["notes"] = notes
+    return web.json_response(result)
+
+
 WEB_DIRECTORY = "web"
 NODE_CLASS_MAPPINGS = {}
 __all__ = ["NODE_CLASS_MAPPINGS"]

requirements.txt

@@ -0,0 +1 @@
+markdownify

web/model-manager.css

@@ -10,7 +10,7 @@
     position: fixed;
     overflow: hidden;
     width: 100%;
-    z-index: 1100;
+    z-index: 1100; /*needs to be below the dialog modal element*/
     
     /*override comfy-modal settings*/
     border-radius: 0;
@@ -23,6 +23,10 @@
     touch-action: manipulation;
 }
 
+.model-manager .model-manager-dialog {
+    z-index: 2001; /*needs to be above the model manager element*/
+}
+
 .model-manager .comfy-modal-content {
     width: 100%;
     gap: 16px;
@@ -249,6 +253,10 @@
     user-select: none;
 }
 
+.model-manager code {
+    text-wrap: wrap;
+}
+
 /* main content */
 .model-manager .model-manager-panel {
     color: var(--fg-color);
@@ -409,6 +417,11 @@
 	border-radius: 8px;
 }
 
+.model-manager .model-info-container .item {
+    width: fit-content;
+    height: 50vh;
+}
+
 .model-manager .item img {
     width: 100%;
     height: 100%;
@@ -416,15 +429,13 @@
     border-radius: 8px;
 }
 
-.model-manager .model-info-container .item {
-    width: auto;
-    height: auto;
-}
-.model-manager .model-info-container .item img {
+.model-manager .model-info-container .item img,
+.model-manager .model-preview-full {
     height: auto;
     width: auto;
     max-width: 100%;
     max-height: 50vh;
+    border-radius: 8px;
 }
 
 .model-manager .model-preview-button-left,