pdf

PDF Document Processing Expert Real Skill

One-sentence description

Comprehensive PDF document processing tool, supporting text and table extraction, PDF creation, document operations, and form filling.

PDF Processing Guide - Comprehensive PDF manipulation toolkit for extracting text and tables, creating new PDFs, merging/splitting documents, and handling forms.

---

Use cases

Applicable：

• Extract text and table data from PDF and export to Excel
• Create new PDF documents (reports, certificates, invoices)
• Merge, split, rotate PDF documents
• Fill PDF forms (government forms, application forms)
• Batch process multiple PDF files

Not applicable：

• Scanned PDFs (requires OCR first / requires OCR first)
• Complex PDF editing (use professional software / use professional software)

---

Core workflow

用户需求 → 判断任务类型 → 选择工具 → 执行操作 → 验证输出
User request → Task type classification → Tool selection → Execute → Validate

Task Classification (Task Classification)：

Extracting Data (Extracting Data)

• Text extraction (Text extraction) → pdfplumber.extract_text()
  • Better accuracy than pypdf for text extraction
• Table extraction (Table extraction) → pdfplumber.extract_tables() + pandas
  • Intelligent table structure recognition
  • Export to Excel, CSV, JSON

Creating PDFs (Creating PDFs)

• Simple documents (Simple documents) → reportlab.pdfgen.canvas
  • Quick document creation with text and graphics
• Complex layouts (Complex layouts) → reportlab.platypus
  • Paragraphs, Tables, Images with advanced styling

PDF Operations (PDF Operations)

• Merge (Merge) → pypdf.PdfWriter
  • Combine multiple PDFs into one
• Split (Split) → pypdf.PdfWriter (save each page separately)
• Rotate (Rotate) → page.rotate()
• Metadata (Metadata) → pypdf.PdfReader.metadata / PdfWriter.add_metadata()

Form Processing (Form Processing)

• View fields (View fields) → See forms.md
• Fill forms (Fill forms) → See forms.md
• Batch processing (Batch processing) → Loop + form filling API

---

Task completion criteria

Must satisfy(none may be missing):

• The PDF file can open normally
• The extracted data is complete and accurate
• The table structure is preserved correctly (rows and columns correspond)
• The created PDF content is clear and readable

Quality rating：

• ⭐⭐⭐⭐⭐ Excellent - Complete data + Standardized format + Automated batch processing
• ⭐⭐⭐ Pass - Correct data + Usable PDF
• ⭐ Fail - Missing data or corrupted PDF

---

Reference materials (for AI use)

Type	Path	Description
Core documents	docs/00-SKILL-完整处理指南.md	Complete PDF processing guide (~300 lines)
Form processing	docs/forms.md	PDF form filling detailed tutorial
Advanced reference	docs/reference.md	JavaScript libraries and advanced features

---

Key Principles (AI must-read / Critical Principles)

1. Prefer pdfplumber for Extraction (Prefer pdfplumber)

• Text extraction: More accurate than pypdf
• Table recognition: Strong table recognition capability
• Layout preservation: Preserves layout information
• Complex PDF: Better for complex PDF processing

2. Table Data Normalization (Table data normalization)

# Best practice for table extraction
import pdfplumber
import pandas as pd

all_tables = []
with pdfplumber.open("document.pdf") as pdf:
    for page in pdf.pages:
        tables = page.extract_tables()
        for table in tables:
            if table:  # Check not empty
                df = pd.DataFrame(table[1:], columns=table[0])  # First row as header
                all_tables.append(df)

# Combine and clean
combined = pd.concat(all_tables, ignore_index=True)
combined.to_excel("output.xlsx", index=False)

3. Scanned PDFs Require OCR (Scanned PDFs require OCR)

• Identification: If text extraction returns empty or garbled text
• Processing: Must use OCR tools first (Tesseract, Adobe Acrobat, etc.)
• Inform the user: Clearly inform user that OCR pre-processing is needed

4. Memory Management for Large Files (Memory management for large files)

• Batch processing for multiple PDFs
• Process pages sequentially rather than loading entire document
• Use generators when possible

---

Quick Commands (Quick Commands)

Text Extraction (Text extraction)

import pdfplumber

with pdfplumber.open("document.pdf") as pdf:
    for page in pdf.pages:
        text = page.extract_text()
        print(text)

Table to Excel (Table to Excel)

import pdfplumber
import pandas as pd

all_tables = []
with pdfplumber.open("document.pdf") as pdf:
    for page in pdf.pages:
        tables = page.extract_tables()
        for table in tables:
            if table:
                df = pd.DataFrame(table[1:], columns=table[0])
                all_tables.append(df)

combined = pd.concat(all_tables, ignore_index=True)
combined.to_excel("extracted_tables.xlsx", index=False)

Merge PDFs (Merge PDFs)

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)

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

Split PDF (Split PDF)

from pypdf import PdfReader, PdfWriter

reader = PdfReader("large_document.pdf")
for i, page in enumerate(reader.pages):
    writer = PdfWriter()
    writer.add_page(page)
    with open(f"page_{i+1}.pdf", "wb") as output:
        writer.write(output)

Create PDF (Create PDF)

from reportlab.lib.pagesizes import letter
from reportlab.pdfgen import canvas

c = canvas.Canvas("hello.pdf", pagesize=letter)
width, height = letter
c.drawString(100, height - 100, "Hello World!")
c.save()

Extract Metadata (Extract metadata)

from pypdf import PdfReader

reader = PdfReader("document.pdf")
meta = reader.metadata
print(f"Title: {meta.title}")
print(f"Author: {meta.author}")
print(f"Pages: {len(reader.pages)}")

---

Dependencies Installation (Dependencies Installation)

Python Dependencies

pip install -r requirements.txt
# Includes: pypdf, pdfplumber, reportlab, pandas, openpyxl

---

Common Scenarios (Common Scenarios)

Scenario 1: Batch Extract Invoice Tables (Batch extract invoice tables)

import pdfplumber
import pandas as pd
import glob

all_data = []
for pdf_file in glob.glob("invoices/*.pdf"):
    with pdfplumber.open(pdf_file) as pdf:
        for page in pdf.pages:
            tables = page.extract_tables()
            for table in tables:
                if table:
                    df = pd.DataFrame(table[1:], columns=table[0])
                    df['source_file'] = pdf_file
                    all_data.append(df)

combined = pd.concat(all_data, ignore_index=True)
combined.to_excel("all_invoices.xlsx", index=False)

Scenario 2: Create Report with Table

from reportlab.lib.pagesizes import letter
from reportlab.platypus import SimpleDocTemplate, Table, TableStyle, Paragraph
from reportlab.lib.styles import getSampleStyleSheet

doc = SimpleDocTemplate("report.pdf", pagesize=letter)
styles = getSampleStyleSheet()
story = []

# Title
story.append(Paragraph("Sales Report", styles['Title']))

# Table
data = [
    ['Product', 'Q1', 'Q2', 'Q3', 'Q4'],
    ['Product A', '1000', '1200', '1100', '1300'],
    ['Product B', '800', '900', '950', '1000']
]
t = Table(data)
t.setStyle(TableStyle([
    ('BACKGROUND', (0, 0), (-1, 0), 'grey'),
    ('TEXTCOLOR', (0, 0), (-1, 0), 'white'),
    ('ALIGN', (0, 0), (-1, -1), 'CENTER'),
    ('GRID', (0, 0), (-1, -1), 1, 'black')
]))
story.append(t)

doc.build(story)