[Bug]: Chinese font setup in `MineruParser.parse_text_file` never works

[Glinte]

Do you need to file an issue?

• I have searched the existing issues and this bug is not already filed.
• I believe this is a legitimate bug, not just a question or feature request.

Describe the bug

...because you are trying to initialize UnicodeCIDFonts with font names that are not supported, but no one ever noticed this is failed because you suppressed all the errors. You used ["SimSun", "SimHei", "Microsoft YaHei"] on Windows and ["STSong-Light", "STHeiti"] on MacOS, but the only supported fonts to be used in UnicodeCIDFont is one of the 6 hardcoded values. By a quick read of the UnicodeCIDFont class docstring, I believe CIDFont should be used instead.

RAG-Anything/raganything/mineru_parser.py

Lines 647 to 681 in 9c72526

	# Try to register a font that supports Chinese characters
	try:
	# Try to use system fonts that support Chinese
	import platform
	system = platform.system()
	if system == "Windows":
	# Try common Windows fonts
	for font_name in ["SimSun", "SimHei", "Microsoft YaHei"]:
	try:
	from reportlab.pdfbase.cidfonts import (
	UnicodeCIDFont,
	)
	pdfmetrics.registerFont(UnicodeCIDFont(font_name))
	normal_style.fontName = font_name
	heading_style.fontName = font_name
	break
	except Exception:
	continue
	elif system == "Darwin": # macOS
	for font_name in ["STSong-Light", "STHeiti"]:
	try:
	from reportlab.pdfbase.cidfonts import (
	UnicodeCIDFont,
	)
	pdfmetrics.registerFont(UnicodeCIDFont(font_name))
	normal_style.fontName = font_name
	heading_style.fontName = font_name
	break
	except Exception:
	continue
	except Exception:
	pass # Use default fonts if Chinese font setup fails

https://github.com/eduardocereto/reportlab/blob/98758940eeae30db80bbc9c555e42b8c89b86be8/src/reportlab/pdfbase/cidfonts.py#L390-L395

class UnicodeCIDFont(CIDFont):
    def __init__(self, face, isVertical=False, isHalfWidth=False):
        #pass
        try:
            lang, defaultEncoding = defaultUnicodeEncodings[face]
        except KeyError:
            raise KeyError("don't know anything about CID font %s" % face)

https://github.com/eduardocereto/reportlab/blob/98758940eeae30db80bbc9c555e42b8c89b86be8/src/reportlab/pdfbase/_cidfontdata.py#L130-L141

defaultUnicodeEncodings = {
    #we ddefine a default Unicode encoding for each face name;
    #this should be the most commonly used horizontal unicode encoding;
    #also define a 3-letter language code.
    'HeiseiMin-W3': ('jpn','UniJIS-UCS2-H'),
    'HeiseiKakuGo-W5': ('jpn','UniJIS-UCS2-H'),
    'STSong-Light': ('chs', 'UniGB-UCS2-H'),
    'MSung-Light': ('cht', 'UniGB-UCS2-H'),
    #'MHei-Medium': ('cht', 'UniGB-UCS2-H'),
    'HYSMyeongJo-Medium': ('kor', 'UniKS-UCS2-H'),
    'HYGothic-Medium': ('kor','UniKS-UCS2-H'),
    }

Steps to reproduce

No response

Expected Behavior

No response

LightRAG Config Used

Paste your config here

Logs and screenshots

No response

Additional Information

• LightRAG Version:
• Operating System:
• Python Version:
• Related Issues:

[onestardao]

this looks like No.14 – Bootstrap Ordering from our problem map.
the font initialization logic is being called before the supported font list is actually available or verified, so it falls back incorrectly every time. effectively, the setup step is firing before its dependency (valid font resource) is ready.

i’ve got a complete fix path for this pattern, including the correct sequence to load/verify fonts so the fallback only triggers when it truly needs to—let me know if you’d like the details.

[Glinte]

Unless you are patching reportlab at runtime (eg adding values to defaultUnicodeEncodings), your usage of UnicodeCIDFont is simply incorrect - it always raises KeyError. I no longer use RAGAnything so I personally do not need further information on this problem.

[onestardao]

OK got it :)

[wcccccccc]

这个问题其实还是存在的，当txt或者md中存在中文时，就会导致生成的pdf里的中文显示实心黑框。我是通过opencode辅助我在自己代码里进行打patch修复。如下：`WINDOWS_CHINESE_FONTS = [
("SimHei", "C:/Windows/Fonts/simhei.ttf"),
("SimSun", "C:/Windows/Fonts/simsun.ttc"),
]

def patch_raganything_chinese_fonts():
"""
Monkey Patch raganything 的中文字体支持
解决 Windows 上 UnicodeCIDFont 注册失败导致中文显示为方块的问题
"""
if platform.system() != "Windows":
return

try:
    from raganything.parser import Parser
    from reportlab.pdfbase import pdfmetrics
    from reportlab.pdfbase.ttfonts import TTFont
    from pathlib import Path as FilePath
    
    original_convert = Parser.convert_text_to_pdf.__func__
    
    @classmethod
    def patched_convert_text_to_pdf(cls, text_path, output_dir=None):
        font_registered = False
        registered_font_name = None
        
        for font_name, font_path in WINDOWS_CHINESE_FONTS:
            if os.path.exists(font_path):
                try:
                    if font_name not in pdfmetrics.getRegisteredFontNames():
                        pdfmetrics.registerFont(TTFont(font_name, font_path))
                    registered_font_name = font_name
                    font_registered = True
                    break
                except Exception:
                    continue
        
        if not font_registered:
            return original_convert(cls, text_path, output_dir)
        
        from reportlab.lib.pagesizes import A4
        from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer
        from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
        from reportlab.lib.units import inch
        
        text_path = FilePath(text_path)
        
        with open(text_path, "r", encoding="utf-8") as f:
            text_content = f.read()
        
        if output_dir:
            base_output_dir = FilePath(output_dir)
        else:
            base_output_dir = text_path.parent / "reportlab_output"
        
        base_output_dir.mkdir(parents=True, exist_ok=True)
        pdf_path = base_output_dir / f"{text_path.stem}.pdf"
        
        doc = SimpleDocTemplate(
            str(pdf_path),
            pagesize=A4,
            leftMargin=inch,
            rightMargin=inch,
            topMargin=inch,
            bottomMargin=inch,
        )
        
        styles = getSampleStyleSheet()
        normal_style = ParagraphStyle(
            'ChineseNormal',
            parent=styles['Normal'],
            fontName=registered_font_name,
            fontSize=10,
            leading=14,
        )
        heading_style = ParagraphStyle(
            'ChineseHeading',
            parent=styles['Heading1'],
            fontName=registered_font_name,
            fontSize=14,
            leading=18,
        )
        
        story = []
        
        if text_path.suffix.lower() == ".md":
            lines = text_content.split("\n")
            for line in lines:
                line = line.strip()
                if not line:
                    story.append(Spacer(1, 12))
                    continue
                
                if line.startswith("#"):
                    level = len(line) - len(line.lstrip("#"))
                    header_text = line.lstrip("#").strip()
                    if header_text:
                        header_style_temp = ParagraphStyle(
                            name=f"Heading{level}",
                            parent=heading_style,
                            fontSize=max(16 - level, 10),
                            spaceAfter=8,
                            spaceBefore=16 if level <= 2 else 12,
                        )
                        story.append(Paragraph(header_text, header_style_temp))
                else:
                    story.append(Paragraph(line, normal_style))
                    story.append(Spacer(1, 6))
        else:
            for line in text_content.split("\n"):
                if line.strip():
                    story.append(Paragraph(line, normal_style))
                    story.append(Spacer(1, 6))
        
        doc.build(story)
        
        return pdf_path
    
    Parser.convert_text_to_pdf = patched_convert_text_to_pdf
    
except Exception:
    pass

patch_raganything_chinese_fonts()`