PDF

PDF Processing

Step 1: Identify the Task

Match the user's request to one workflow. Default to Extract Text when unclear.

Step 2: Install Dependencies

Check what is available before writing code:

pip list 2>/dev/null | grep -iE "pypdf|pdfplumber|reportlab|pypdfium2|weasyprint|pytesseract|pdf2image"
which pdftotext qpdf pdftk pdfimages 2>/dev/null

Install only what the task needs. Do not install everything.

Step 3: Execute the Workflow

Use the code patterns in references/cookbook.md for implementation details. Use references/advanced-features.md for pypdfium2, pdf-lib (JS), and advanced CLI operations.

Validation Checkpoints

Run these checks at each stage:

After extraction:

text = page.extract_text()
if not text or len(text.strip()) < 10:
    # PDF may be scanned -- fall back to OCR
    print(f"Page {i+1}: no text found, switching to OCR")

After merge/split:

result = PdfReader("output.pdf")
print(f"Output has {len(result.pages)} pages")
# Verify page count matches expectation

After form fill: Run the validation scripts described in references/forms.md before delivering the output.

After markdown-to-PDF:

import os
output_size = os.path.getsize("output.pdf")
print(f"Generated PDF: {output_size} bytes")
if output_size < 500:
    print("Warning: PDF seems too small, check for conversion errors")

Step 4: Deliver Results

• Report what was done: page count, text length, file size.
• If OCR was used, warn the user about potential accuracy issues.
• For form fills, note which fields were populated and which were skipped.

Steps:

1. Install pdfplumber and pandas.
2. Extract tables from each page.
3. Validate each table has rows before saving.
4. Combine and export to Excel.

import pdfplumber
import pandas as pd

with pdfplumber.open("document.pdf") as pdf:
    all_tables = []
    for i, page in enumerate(pdf.pages):
        tables = page.extract_tables()
        for table in tables:
            if table and len(table) > 1:
                df = pd.DataFrame(table[1:], columns=table[0])
                all_tables.append(df)
                print(f"Page {i+1}: extracted table with {len(df)} rows")
    if all_tables:
        combined = pd.concat(all_tables, ignore_index=True)
        combined.to_excel("tables.xlsx", index=False)
        print(f"Saved {len(combined)} total rows to tables.xlsx")
    else:
        print("No tables found in this PDF")

from pypdf import PdfWriter, PdfReader

writer = PdfWriter()
for pdf_file in ["doc1.pdf", "doc2.pdf", "doc3.pdf"]:
    reader = PdfReader(pdf_file)
    for page in reader.pages:
        writer.add_page(page)
    print(f"Added {len(reader.pages)} pages from {pdf_file}")

with open("merged.pdf", "wb") as output:
    writer.write(output)

# Verify
result = PdfReader("merged.pdf")
print(f"Merged PDF has {len(result.pages)} pages")

import pytesseract
from pdf2image import convert_from_path

images = convert_from_path("scanned.pdf", dpi=300)
text = ""
for i, image in enumerate(images):
    page_text = pytesseract.image_to_string(image)
    text += page_text
    print(f"Page {i+1}: extracted {len(page_text)} characters")

with open("extracted.txt", "w") as f:
    f.write(text)
print(f"Total: {len(text)} characters. Review for OCR errors.")

# macOS may need: export DYLD_LIBRARY_PATH=$(brew --prefix)/lib
python scripts/md_to_pdf.py report.md report.pdf

Features: A4 pages, proper margins, table/code support, Chinese font fallback. For batch conversion: python scripts/batch_convert.py *.md --output-dir ./pdfs/

Follow the complete workflow in references/forms.md. Summary:

1. Check if the PDF has fillable fields: python scripts/check_fillable_fields.py form.pdf
2. If fillable: extract field info, create values JSON, run fill script.
3. If not fillable: convert to images, identify fields visually, create bounding boxes, validate, fill with annotations.

References

Scripts