Pdf Paper Translator

3. Structured Extraction and Translation

Pass the extracted text to an LLM to generate the translated summary. You must ensure the following rules are met in the LLM prompt:

• Academic Precision Translation: The translation must completely align with the Traditional Chinese academic context and terminology. Strictly avoid stiff or literal word-for-word translation.
• Structured Extraction Fields: You must explicitly extract exactly the following five fields:
  1. 論文標題 (Paper Title)
  2. 研究背景 (Background)
  3. 核心方法 (Core Methods)
  4. 實驗結果 (Experimental Results)
  5. 結論 (Conclusion)

Output Requirements

• CSV 垂直佈局：產出的 paper_analysis.csv 必須採用「兩欄式垂直佈局」（第一欄為項目名稱，第二欄為內容），避免在編輯器中產生長距離的橫向捲動。
• 長文本處理：寫入 CSV 時，必須保留內容內部的換行符號 (\n)，並確保所有欄位內容都使用雙引號 ("") 包覆，防止檔案解析跑版。
• 新增 Markdown 報告：除了 CSV，必須額外生成一個 report.md，使用 Markdown 的層級標題（#、##）與清單，將論文重點整理成易於閱讀的報告。

4. Save as CSV

Write the structured output into a CSV file.

• Output Filename: Must be exactly paper_analysis.csv.
• CSV Formatting: Always include a header row containing exactly: 論文標題,研究背景,核心方法,實驗結果,結論.
• Prevent Formatting Breakage (防止跑版): Use Python's built-in csv module (e.g., csv.DictWriter) with encoding="utf-8-sig" (to ensure Excel compatibility). The csv module guarantees that fields containing internal commas, quotes, or newlines are properly wrapped in double quotes, preventing structure breakage.

Example Script Outline

import os
import csv
import sys
# import fitz  # PyMuPDF

def check_environment(pdf_path):
    # 1. Verify input file exists
    if not os.path.exists(pdf_path):
        print(f"Error: File '{pdf_path}' not found.")
        sys.exit(1)
        
    # 2. Check dependencies
    try:
        import fitz
    except ImportError:
        print("Required library not found. Please install: pip install PyMuPDF")
        sys.exit(1)

def save_to_csv(extracted_data, output_file="paper_analysis.csv"):
    headers = ["論文標題", "研究背景", "核心方法", "實驗結果", "結論"]
    file_exists = os.path.exists(output_file)
    
    with open(output_file, mode='a' if file_exists else 'w', encoding='utf-8-sig', newline='') as f:
        writer = csv.DictWriter(f, fieldnames=headers)
        if not file_exists:
            writer.writeheader()
        writer.writerow(extracted_data)

# Main Execution Flow
# 1. check_environment(pdf_path)
# 2. text = read_pdf(pdf_path)
# 3. extracted_data = analyze_and_translate_with_llm(text)
# 4. save_to_csv(extracted_data)