Automatically Tag and Import NovelAI Images for Hydrus

by MiyaNya/.yamii.

This is a script for generating "sidecars" which are files containing tags that Hydrus will import alongside your NovelAI generated images.

Prerequisites:

Using the script:

Create a folder to set up the process. DO NOT PUT IT IN THE SAME FOLDER AS HYDRUS.
Inside this folder, create a file named generate_sidecars.py
Add the code below inside it. Save the file.
Make a folder called import_images, and put your images there.
Run CMD in the folder with the py file. Then, execute py -m pip install pillow.
Once pillow is installed, py generate_sidecars.py
Now your import folder will be full of text files containing the tags for the images.

Importing to Hydrus:

file -> import files...
Add folder, find the import_images folder.
Add tags/urls with the import >>
At the top, click the sidecars tab.
add -> add -> a .txt sidecar
Check "remove file .ext?"
apply -> apply -> apply

Code

import os
from PIL import Image
import json
import re

def main():
    dirPath = ".\\import_images"   # rename this to your folder of choice
    
    filelist = os.listdir(dirPath)
    
    for root, dirs, files in reversed(list(os.walk(dirPath))):
        for filename in reversed(files):
    
            if not(filename.endswith(".png")):
                continue
            
            image_path = os.path.join(root, filename)
            sidecar_path = os.path.splitext(image_path)[0] + ".txt"

            if os.path.isfile(sidecar_path):
                continue
                
            im = Image.open(image_path)
            comment = json.loads(im.info['Comment']) if 'Comment' in im.info else ''
            
            tags = ''
            if 'actual_prompts' in comment:
                tags = comment["actual_prompts"]["prompt"]["base_caption"].lower()
                for char_prompt in comment["actual_prompts"]["prompt"]["char_captions"]:
                    tags += ', ' + char_prompt["char_caption"].lower()
            elif 'Description' in im.info:
                tags = im.info['Description'].lower()
                if 'v4_prompt' in comment and 'char_captions' in comment['v4_prompt']['caption']:
                    for char_prompt in comment['v4_prompt']['caption']['char_captions']:
                        tags += ', ' + char_prompt['char_caption'].lower()
            else:
                print(im.info)
                continue
            
            tags = tags.replace("[","")
            tags = tags.replace("]","")
            tags = tags.replace("{","")
            tags = tags.replace("}","")
            tags = re.sub(r'-\d*\.?\d*::.*?::', '', tags) # remove tags with negative emphasis
            tags = re.sub(r'-?\d*\.?\d*::', '', tags) # remove all ## and any number behind it
            #tags = re.sub(r'(?<!\d)\b(girl|boy)\b', '', tags)
            
            tag_array = [x.strip() for x in re.split(r'[.,|]', tags)]
            
            tag_array.append('model:' + im.info["Source"])
            tag_array.append('sampler:' + comment["sampler"])
            
            sidecar_text = '\n'.join(tag_array)
            
            print(sidecar_path)
            
            write_to_file(sidecar_path, sidecar_text)


def write_to_file (output_file, output_string):
    with open(output_file, 'w', encoding='utf-8') as file:
        file.write(output_string)
        
if __name__ == '__main__':
     main()

# by .yamii.```