pdf-reader

PDF Reader Skill (PyMuPDF)

Use PyMuPDF (import fitz) to read PDF files. Supports 3 extraction modes with increasing fidelity. Handles Chinese text, multi-page PDFs, and image-based PDFs.

When to Use

• Any time you need to read a .pdf file — always use this instead of the native Read tool
• User mentions: "read PDF", "读取PDF", "PDF内容", "提取PDF", "PDF text", "open PDF"
• User provides a file path ending in .pdf
• Any task requiring PDF content extraction

Extraction Modes

Choose the mode based on the downstream task:

Mode	Method	Output	Best for
text (default)	get_text()	Plain text	Simple reading, summarization
html	get_text("html")	HTML with CSS formatting	High-fidelity intermediate format, preserving bold/italic/font size/color
dict	get_text("dict")	Structured JSON-like dict	Maximum fidelity — font name, size, color, position, flags; ideal for PDF → .tex conversion

Mode selection guide:

• Just need to read content? → text
• Need to preserve formatting (bold, italic, headings, tables)? → html
• Converting PDF to .tex/.docx and need font size / layout info? → dict
• Not sure? → Use dict (most complete, can derive everything else)

Implementation

Step 1: Check PyMuPDF Installation

python -m pip install PyMuPDF 2>&1 | tail -3

Step 2: Extract PDF Content

Always set PYTHONIOENCODING=utf-8 to avoid GBK encoding issues on Windows.

---

Mode: text (default — plain text extraction)

PYTHONIOENCODING=utf-8 python -c "
import fitz, sys
pdf_path = sys.argv[1]
doc = fitz.open(pdf_path)
print(f'Total pages: {len(doc)}')
for page in doc:
    text = page.get_text()
    if text.strip():
        print(f'--- Page {page.number + 1} ---')
        print(text)
    else:
        print(f'--- Page {page.number + 1}: [empty/image-only] ---')
doc.close()
" "<PDF_PATH>"

With page range:

PYTHONIOENCODING=utf-8 python -c "
import fitz, sys
pdf_path = sys.argv[1]
start, end = int(sys.argv[2]) - 1, int(sys.argv[3])
doc = fitz.open(pdf_path)
print(f'Total pages: {len(doc)}, extracting pages {start+1}-{end}')
for i in range(start, min(end, len(doc))):
    text = doc[i].get_text()
    if text.strip():
        print(f'--- Page {i + 1} ---')
        print(text)
    else:
        print(f'--- Page {i + 1}: [empty/image-only] ---')
doc.close()
" "<PDF_PATH>" "<START>" "<END>"