Code to extract data from LLM characters

nodes/__init__.py

@@ -1,7 +1,8 @@
 from .image_glitcher import ImageGlitcher
 from .color_stylizer import ColorStylizer
-from .local_llm import QueryLocalLLM
+from .local_llm import QueryLocalLLM, ExtractCharacterInfo
 from .sdxl_resolution import SdxlResolution, SdxlResolutionToDimensions
+from .checkpoint_names import CheckpointNames
 
 NODE_CLASS_MAPPINGS = {
     ImageGlitcher.NAME: ImageGlitcher,
@@ -9,6 +10,8 @@ NODE_CLASS_MAPPINGS = {
     QueryLocalLLM.NAME: QueryLocalLLM,
     SdxlResolution.NAME: SdxlResolution,
     SdxlResolutionToDimensions.NAME: SdxlResolutionToDimensions,
+    CheckpointNames.NAME: CheckpointNames,
+    ExtractCharacterInfo.NAME: ExtractCharacterInfo,
 }
 
 # A dictionary that contains the friendly/humanly readable titles for the nodes
@@ -18,6 +21,8 @@ NODE_DISPLAY_NAME_MAPPINGS = {
     "QueryLocalLLM": "Query Local LLM",
     "SdxlResolution": "SDXL Resolution",
     "SdxlResolutionToDimensions": "SDXL Resolution To Dimensions",
+    "CheckPointNames": "Checkpoint Names",
+    "ExtractCharacterInfo": "Extract Character Info"
 }
 
 __all__ = ["NODE_CLASS_MAPPINGS", "NODE_DISPLAY_NAME_MAPPINGS"]

nodes/checkpoint_names.py

@@ -0,0 +1,24 @@
+import folder_paths
+
+class CheckpointNames:
+    NAME = "Checkpoint Names"
+    
+    def __init__(self):
+        pass
+
+    @classmethod
+    def INPUT_TYPES(s):
+        return {
+            "required": {
+                "ckpt_name": ("ckpt_name", folder_paths.get_filename_list("checkpoints"), "CKPT_NAME"),
+            }
+        }
+
+    RETURN_TYPES = ("STRING",)
+    RETURN_NAMES = ("name",)
+    FUNCTION = "getCheckpoints"
+    OUTPUT_NODE = False
+    CATEGORY = "CrasH Utils/Loaders"
+
+    def getCheckpoints(self, checkpoint_name):
+        return checkpoint_name

nodes/local_llm.py

@@ -1,5 +1,8 @@
 import requests
 import json
+from PIL import Image
+import base64
+import re
 
 class QueryLocalLLM:
     NAME = "Query Local LLM"
@@ -65,4 +68,43 @@ class QueryLocalLLM:
             return generatedText
         else:
             print(f"Error {response.status_code}: {response.text}")
-            return None
+            return None
+
+class ExtractCharacterInfo:
+    NAME = "Extract Character Information"
+    
+    def __init__(self):
+        pass
+
+    @classmethod
+    def INPUT_TYPES(s):
+        return {
+            "required": {
+                "path": ("STRING", { "multiline": False, "default": "" }),
+            },
+        }
+
+    RETURN_TYPES = ("STRING",)
+    RETURN_NAMES = ("character_data",)
+    FUNCTION = "parse_png_text"
+    OUTPUT_NODE = False
+    CATEGORY = "CrasH Utils/LLM"
+    
+    def parse_png_text(self, path):
+        # Open the image file
+        with Image.open(path) as img:
+            # PNG images can have multiple text chunks, get them all
+            text_chunks = [chunk for chunk in img.text.values()]
+
+            # If there are no text chunks, raise an error
+            if not text_chunks:
+                raise ValueError('No text data found in PNG image.')
+
+            # Decode the first text chunk from base64 to utf-8
+            try:
+                # This assumes the text is base64-encoded as in the JS example
+                decoded_text = base64.b64decode(text_chunks[0]).decode('utf-8')
+                print(decoded_text)
+                return decoded_text
+            except Exception as e:
+                raise ValueError('Could not decode the text chunk.') from e