[FORMAT SUPPORT] - Civitai Metadata (Mojibake Reversal)

[duskfallcrew]

Project repo or official website

https://github.com/Ktiseos-Nyx/stable-diffusion-prompt-reader/tree/master

Image file

Summary

The tool is actually Civitai, and i'm working on dataset tools to still use the Civitai formatter, on MINE https://github.com/Ktiseos-Nyx/Dataset-Tools/ (Which i yes am using your tool as ab ack end but i'm wokring to only vendor the formatters) - I got Civitai comfyUI and A1111 parsing to work.

There's a bunch of Mojibake going on, and ther'es not much more i can explain?

[duskfallcrew]

There's extra code that i dont know how to fix entirely that has to do with errors i'm getting,

In your sd_prompt_reader/format/civitai.py

Inside the parse() method, after parsing extra_meta_dict:

        # ... (existing parameter extraction) ...
        if "seed" in extra_meta_dict: self._parameter["seed"] = str(extra_meta_dict["seed"])

        # --- NEW: Extract Model Information ---
        model_name_or_id = "Unknown" # Default
        if "resources" in extra_meta_dict and isinstance(extra_meta_dict["resources"], list):
            for resource in extra_meta_dict["resources"]:
                if isinstance(resource, dict) and resource.get("type") == "model": # Or based on some other Civitai convention
                    model_name_or_id = str(resource.get("name", resource.get("modelVersionId", "Unknown")))
                    break # Found the primary model
                elif isinstance(resource, dict) and "modelVersionId" in resource and not model_name_or_id.startswith("urn:"):
                    # Fallback: take the first modelVersionId if no explicit model type,
                    # but prioritize URNs from main workflow if found later.
                    model_name_or_id = str(resource.get("modelVersionId", "Unknown"))
                    # Don't break here, main workflow might have better info.
        
        # Alternative/Fallback: Check the main workflow data for CheckpointLoaderSimple
        if self.workflow_data and (model_name_or_id == "Unknown" or not model_name_or_id.startswith("urn:")) :
            # Look for CheckpointLoaderSimple node more reliably
            for node_id, node_data in self.workflow_data.items():
                if isinstance(node_data, dict) and node_data.get("class_type") == "CheckpointLoaderSimple":
                    ckpt_name = node_data.get("inputs", {}).get("ckpt_name")
                    if ckpt_name:
                        model_name_or_id = str(ckpt_name)
                        break # Found it
        
        self._parameter["model"] = model_name_or_id
        # --- END NEW: Extract Model Information ---

        self._width = str(extra_meta_dict.get("width", getattr(self, "_width", 0)))
        self._height = str(extra_meta_dict.get("height", getattr(self, "_height", 0)))
        # ...

It's cause i broke tkinkter on my master lol:
Exception in Tkinter callback
Traceback (most recent call last):
File "/Users/duskfall/.pyenv/versions/3.10.16/lib/python3.10/tkinter/init.py", line 1921, in call
return self.func(*args)
File "/Users/duskfall/.venvsd/lib/python3.10/site-packages/sd_prompt_reader/app.py", line 523, in display_info
self.setting_box_parameter.update_text(self.image_data.parameter)
File "/Users/duskfall/.venvsd/lib/python3.10/site-packages/sd_prompt_reader/parameter_viewer.py", line 141, in update_text
self.setting_model.configure(text=str(parameter["model"]))
KeyError: 'model'
2

Because id o'nt want to break it further i don't want to make MORE changes without like knowing wtf i'm doing XD