Examining the path of torch.classes raised: Tried to instantiate class '__path__._path', but it does not exist! Ensure that it is registered via torch::class_

[uniqueSkeeter]

2024-12-25 17:48:45.815 Examining the path of torch.classes raised: Tried to instantiate class 'path.path', but it does not exist! Ensure that it is registered via torch::class
2024-12-25 17:49:33.237 Examining the path of torch.classes raised: Tried to instantiate class 'path.path', but it does not exist! Ensure that it is registered via torch::class
Recognizing layout: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:02<00:00, 1.11s/it]
Detecting bboxes: 0it [00:00, ?it/s]
Recognizing equations: 0it [00:00, ?it/s]
Recognizing tables: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 2.83it/s]
2024-12-25 17:49:55.155 Uncaught app execution
Traceback (most recent call last):
File "/root/anaconda3/envs/marker/lib/python3.10/site-packages/streamlit/runtime/scriptrunner/exec_code.py", line 88, in exec_func_with_error_handling
result = func()
File "/root/anaconda3/envs/marker/lib/python3.10/site-packages/streamlit/runtime/scriptrunner/script_runner.py", line 579, in code_to_exec
exec(code, module.dict)
File "/root/anaconda3/envs/marker/lib/python3.10/site-packages/marker_app.py", line 136, in
rendered = convert_pdf(
File "/root/anaconda3/envs/marker/lib/python3.10/site-packages/marker_app.py", line 37, in convert_pdf
return converter(fname)
File "/root/anaconda3/envs/marker/lib/python3.10/site-packages/marker/converters/pdf.py", line 109, in call
processor(document)
File "/root/anaconda3/envs/marker/lib/python3.10/site-packages/marker/processors/debug.py", line 60, in call
self.draw_layout_debug_images(document)
File "/root/anaconda3/envs/marker/lib/python3.10/site-packages/marker/processors/debug.py", line 109, in draw_layout_debug_images
self.render_on_image(line_bboxes, png_image, labels=line_text, color="black", draw_bbox=False, label_font_size=24)
File "/root/anaconda3/envs/marker/lib/python3.10/site-packages/marker/processors/debug.py", line 173, in render_on_image
label_font = ImageFont.truetype(font_path, label_font_size)
File "/root/anaconda3/envs/marker/lib/python3.10/site-packages/PIL/ImageFont.py", line 834, in truetype
return freetype(font)
File "/root/anaconda3/envs/marker/lib/python3.10/site-packages/PIL/ImageFont.py", line 831, in freetype
return FreeTypeFont(font, size, index, encoding, layout_engine)
File "/root/anaconda3/envs/marker/lib/python3.10/site-packages/PIL/ImageFont.py", line 257, in init
self.font = core.getfont(
OSError: locations (loca) table missing

[10179013]

I also encountered this problem

[flight505]

I am getting this as well, I am not sure were it is coming from. I have marker in a streamlit app with LightRAG, I noticed someone else mentioning the issue on the LightRAG git

[Godplayer]

I am also getting the same error

[Bob080812]

I am also getting the same error

[Mehdi-GASMI]

same for me

[Franky5831]

Same here

[paulo-maia]

ditto

[franklinthony]

Same for me

[ahming]

same for me

[VikParuchuri]

Examining the path of torch.classes raised: Tried to instantiate class 'path.path', but it does not exist! Ensure that it is registered via torch::class is a warning, and only appears in streamlit. I'm guessing it's because streamlit reloads everything, but I haven't looked into it deeply. Things still work fine with this warning, though.

Your traceback shows a separate error -

File "/root/anaconda3/envs/marker/lib/python3.10/site-packages/PIL/ImageFont.py", line 257, in init
self.font = core.getfont(
OSError: locations (loca) table missing

This is related to missing the debug image fonts. It will only appear when you check debug. I can look into this.

[flight505]

@VikParuchuri this is also rellated to running Marker in Streamlit. WARNING streamlit.runtime.scriptrunner_utils.script_run_context: Thread 'MainThread': missing ScriptRunContext! This warning can be ignored when running in bare mode. While the application and Marker is working as it should the warning is repeated 50 times and could not be suppressed. And also this warning -> WARNING streamlit:
Warning: to view a Streamlit app on a browser, use Streamlit in a file and
run it with the following command:

streamlit run [FILE_NAME] [ARGUMENTS] 

While I normally from streamlit from my IDE I tested in Iterm2 still, persists and using uv run app.py && streamlit run app.py creating a temp env and running directly in that. It is not causing any problems directly. I just cant track it down, are you familier with any of these issues?

class MarkerConverter(PDFConverter):
   """PDF converter using Marker"""
   
   def __init__(self):
       """Initialize Marker converter"""
       try:
           from marker.converters.pdf import PdfConverter
           from marker.models import create_model_dict
           from marker.config.parser import ConfigParser
           
           # Configure Marker settings with enhanced equation detection
           config = {
               "output_format": "markdown",
               "layout_analysis": True,
               "detect_equations": True,
               "equation_detection_confidence": 0.3,
               "detect_inline_equations": True,
               "detect_tables": True,
               "detect_lists": True,
               "detect_code_blocks": True,
               "detect_footnotes": True,
               "equation_output": "latex",
               "preserve_math": True,
               "equation_detection_mode": "aggressive",
               "equation_context_window": 3,
               "equation_pattern_matching": True,
               "equation_symbol_extraction": True,
               
               # Enhanced header handling
               "header_detection": {
                   "enabled": True,
                   "style": "atx",  # Use # style headers
                   "levels": {
                       "title": 1,    # Title uses single #
                       "section": 2,   # Sections use ##
                       "subsection": 3 # Subsections use ###
                   },
                   "remove_duplicate_markers": True
               },
               
               # Enhanced list handling
               "list_detection": {
                   "enabled": True,
                   "unordered_marker": "-",  # Use - for unordered lists
                   "ordered_marker": "1.",   # Use 1. for ordered lists
                   "preserve_numbers": True,  # Keep original list numbers
                   "indent_spaces": 2        # Use 2 spaces for indentation
               },
               
               # Layout and formatting
               "layout": {
                   "paragraph_breaks": True,
                   "line_spacing": 2,
                   "remove_redundant_whitespace": True,
                   "preserve_line_breaks": True,
                   "preserve_blank_lines": True
               },
               
               # Content preservation
               "preserve": {
                   "links": True,
                   "tables": True,
                   "images": True,
                   "footnotes": True,
                   "formatting": True,
                   "lists": True,
                   "headers": True
               },
               
               # Output settings
               "output": {
                   "format": "markdown",
                   "save_markdown": True,
                   "save_text": True,
                   "markdown_ext": ".md",
                   "text_ext": ".txt"
               }
           }
           
           config_parser = ConfigParser(config)
           
           # Initialize converter with config
           self._converter = PdfConverter(
               config=config_parser.generate_config_dict(),
               artifact_dict=create_model_dict(),
               processor_list=config_parser.get_processors(),
               renderer=config_parser.get_renderer()
           )
           
           logger.info("Marker initialized with optimized settings")
           
       except Exception as e:
           logger.error(f"Failed to initialize Marker: {str(e)}")
           print(colored(f"⚠️ Failed to initialize Marker: {str(e)}", "yellow"))
           raise
   
   def extract_text(self, file_path: str) -> str:
       """Extract text with semantic structure preservation"""
       try:
           # Process PDF with Marker
           rendered = self._converter(file_path)
           
           # Save markdown file
           markdown_path = str(Path(file_path).with_suffix('.md'))
           
           # Extract text from rendered output
           if hasattr(rendered, 'markdown'):
               text = rendered.markdown
               # Save markdown content ... 

[alibabadoufu]

Same here. I am getting the same error

[vandant1]

Same! I too get the same error while executing the the streamlit code. But here along with that I get following error and not understand why is it getting ?Error during detection: operands could not be broadcast together with shapes (1,4) () (3,)

[Daryl149]

I got this too, it seems marker and streamlit do not work well together.