Python+ReportLab自动生成PDF翻译质量审计报告：从数据到可视化的完整方案

在企业级PDF翻译流程中，质量审计是关键环节。本文介绍如何用Python+ReportLab搭建一套自动化的翻译质量审计报告生成系统，从数据采集到PDF可视化输出，提供完整可运行的代码方案。

一、为什么需要翻译质量审计报告？

在企业级PDF翻译场景中，质量管控通常面临以下挑战：

• 翻译一致性：同一术语在不同页面翻译不一致
• 格式完整性：翻译后表格错位、图片丢失
• 术语准确性：专业术语翻译错误
• 完整性检查：段落遗漏、内容截断

传统的人工逐页检查效率低下，且容易遗漏。自动化审计报告可以：

1. 量化翻译质量指标
2. 定位具体问题位置
3. 形成可追溯的质量档案

二、系统架构设计

┌─────────────┐    ┌─────────────┐    ┌─────────────┐
│  原文PDF    │───→│  翻译PDF    │───→│  质量审计   │
└─────────────┘    └─────────────┘    └──────┬──────┘
                                             │
                              ┌──────────────┼──────────────┐
                              ↓              ↓              ↓
                        ┌─────────┐   ┌──────────┐  ┌──────────┐
                        │术语一致性│   │格式完整性│  │内容完整性│
                        └────┬────┘   └────┬─────┘  └────┬─────┘
                             └─────────────┴─────────────┘
                                           ↓
                                    ┌────────────┐
                                    │  审计报告PDF │
                                    └────────────┘

三、核心模块实现

3.1 术语一致性检查

import re
from collections import defaultdict
from difflib import SequenceMatcher

class TerminologyChecker:
    def __init__(self, glossary_path=None):
        """
        初始化术语检查器
        glossary_path: 术语表路径，格式为 JSON {"英文术语": "标准中文翻译"}
        """
        self.glossary = {}
        if glossary_path:
            import json
            with open(glossary_path, 'r', encoding='utf-8') as f:
                self.glossary = json.load(f)
        self.found_terms = defaultdict(list)
    
    def extract_text_from_pdf(self, pdf_path):
        """从PDF中提取文本，返回 {page_num: [text_blocks]} """
        try:
            import fitz  # PyMuPDF
            doc = fitz.open(pdf_path)
            pages = {}
            for i, page in enumerate(doc):
                text = page.get_text()
                pages[i+1] = text.split('\n')
            doc.close()
            return pages
        except ImportError:
            print("请安装 PyMuPDF: pip install PyMuPDF")
            return {}
    
    def check_consistency(self, translated_pdf):
        """检查术语翻译一致性"""
        pages = self.extract_text_from_pdf(translated_pdf)
        inconsistencies = []
        
        for page_num, lines in pages.items():
            for line_num, line in enumerate(lines, 1):
                for en_term, std_cn in self.glossary.items():
                    if en_term.lower() in line.lower():
                        # 发现英文术语出现在译文中，可能是未翻译
                        inconsistencies.append({
                            'type': 'untranslated_term',
                            'page': page_num,
                            'line': line_num,
                            'term': en_term,
                            'context': line.strip()[:100]
                        })
        
        return inconsistencies

3.2 格式完整性检查

class FormatChecker:
    def __init__(self):
        self.issues = []
    
    def check_page_count(self, original_pdf, translated_pdf):
        """检查页数是否一致"""
        import fitz
        orig_doc = fitz.open(original_pdf)
        trans_doc = fitz.open(translated_pdf)
        
        orig_pages = len(orig_doc)
        trans_pages = len(trans_doc)
        
        if orig_pages != trans_pages:
            self.issues.append({
                'type': 'page_count_mismatch',
                'severity': 'high',
                'detail': f'原文{orig_pages}页，译文{trans_pages}页'
            })
        
        orig_doc.close()
        trans_doc.close()
        return self.issues
    
    def check_image_integrity(self, pdf_path):
        """检查图片是否完整"""
        import fitz
        doc = fitz.open(pdf_path)
        
        for i, page in enumerate(doc):
            images = page.get_images()
            for img_index, img in enumerate(images):
                xref = img[0]
                base_image = doc.extract_image(xref)
                if base_image["width"] == 0 or base_image["height"] == 0:
                    self.issues.append({
                        'type': 'corrupted_image',
                        'severity': 'medium',
                        'page': i+1,
                        'detail': f'第{i+1}页第{img_index+1}张图片损坏'
                    })
        
        doc.close()
        return self.issues

3.3 用ReportLab生成审计报告PDF

from reportlab.lib import colors
from reportlab.lib.pagesizes import A4
from reportlab.platypus import SimpleDocTemplate, Table, TableStyle, Paragraph, Spacer, Image
from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
from reportlab.lib.units import cm
from reportlab.pdfbase import pdfmetrics
from reportlab.pdfbase.ttfonts import TTFont
from datetime import datetime
import os

class AuditReportGenerator:
    def __init__(self, output_path):
        """
        初始化报告生成器
        output_path: 输出PDF路径
        """
        self.output_path = output_path
        self.doc = SimpleDocTemplate(
            output_path,
            pagesize=A4,
            rightMargin=72,
            leftMargin=72,
            topMargin=72,
            bottomMargin=18
        )
        
        # 注册中文字体（使用系统自带字体）
        try:
            pdfmetrics.registerFont(TTFont('SimSun', 'simsun.ttc'))
            self.font_name = 'SimSun'
        except:
            # 如果找不到宋体，使用默认字体
            self.font_name = 'Helvetica'
        
        self.styles = getSampleStyleSheet()
        self._setup_custom_styles()
    
    def _setup_custom_styles(self):
        """设置自定义样式"""
        self.styles.add(ParagraphStyle(
            name='ChineseTitle',
            fontName=self.font_name,
            fontSize=18,
            leading=22,
            alignment=1,  # 居中
            spaceAfter=12
        ))
        
        self.styles.add(ParagraphStyle(
            name='ChineseHeading',
            fontName=self.font_name,
            fontSize=14,
            leading=18,
            spaceAfter=6,
            textColor=colors.HexColor('#2c3e50')
        ))
        
        self.styles.add(ParagraphStyle(
            name='ChineseBody',
            fontName=self.font_name,
            fontSize=10,
            leading=14,
            spaceAfter=6
        ))
    
    def generate_summary_table(self, stats):
        """生成汇总统计表格"""
        data = [
            ['指标', '数值', '状态'],
            ['术语一致性', f"{stats.get('term_consistency', 0)}%", 
             '通过' if stats.get('term_consistency', 0) > 95 else '警告'],
            ['格式完整性', f"{stats.get('format_integrity', 0)}%",
             '通过' if stats.get('format_integrity', 0) > 95 else '警告'],
            ['页数一致性', '一致' if stats.get('page_match', False) else '不一致',
             '通过' if stats.get('page_match', False) else '错误'],
            ['图片完整性', f"{stats.get('image_integrity', 0)}%",
             '通过' if stats.get('image_integrity', 0) > 95 else '警告'],
        ]
        
        table = Table(data, colWidths=[4*cm, 3*cm, 2.5*cm])
        table.setStyle(TableStyle([
            ('BACKGROUND', (0, 0), (-1, 0), colors.HexColor('#3498db')),
            ('TEXTCOLOR', (0, 0), (-1, 0), colors.whitesmoke),
            ('ALIGN', (0, 0), (-1, -1), 'CENTER'),
            ('FONTNAME', (0, 0), (-1, 0), self.font_name),
            ('FONTSIZE', (0, 0), (-1, 0), 11),
            ('BOTTOMPADDING', (0, 0), (-1, 0), 12),
            ('BACKGROUND', (0, 1), (-1, -1), colors.HexColor('#ecf0f1')),
            ('GRID', (0, 0), (-1, -1), 1, colors.grey),
            ('FONTNAME', (0, 1), (-1, -1), self.font_name),
            ('FONTSIZE', (0, 1), (-1, -1), 10),
            ('ROWBACKGROUNDS', (0, 1), (-1, -1), [colors.white, colors.HexColor('#f8f9fa')])
        ]))
        
        return table
    
    def generate_detail_table(self, issues):
        """生成问题明细表格"""
        if not issues:
            return Paragraph("未发现问题", self.styles['ChineseBody'])
        
        data = [['序号', '类型', '严重程度', '位置', '详情']]
        for i, issue in enumerate(issues[:50], 1):  # 最多显示50条
            data.append([
                str(i),
                issue.get('type', '未知'),
                issue.get('severity', '未知'),
                f"第{issue.get('page', '?')}页",
                issue.get('detail', '')[:50]
            ])
        
        table = Table(data, colWidths=[1.2*cm, 2.5*cm, 1.8*cm, 1.8*cm, 5*cm])
        table.setStyle(TableStyle([
            ('BACKGROUND', (0, 0), (-1, 0), colors.HexColor('#e74c3c')),
            ('TEXTCOLOR', (0, 0), (-1, 0), colors.whitesmoke),
            ('ALIGN', (0, 0), (-1, -1), 'LEFT'),
            ('FONTNAME', (0, 0), (-1, 0), self.font_name),
            ('FONTSIZE', (0, 0), (-1, 0), 10),
            ('BOTTOMPADDING', (0, 0), (-1, 0), 10),
            ('GRID', (0, 0), (-1, -1), 0.5, colors.grey),
            ('FONTNAME', (0, 1), (-1, -1), self.font_name),
            ('FONTSIZE', (0, 1), (-1, -1), 9),
            ('VALIGN', (0, 0), (-1, -1), 'MIDDLE'),
        ]))
        
        return table
    
    def create_report(self, original_file, translated_file, stats, issues):
        """
        生成完整审计报告
        
        Args:
            original_file: 原文PDF路径
            translated_file: 译文PDF路径
            stats: 统计字典
            issues: 问题列表
        """
        story = []
        
        # 标题
        story.append(Paragraph("PDF翻译质量审计报告", self.styles['ChineseTitle']))
        story.append(Spacer(1, 0.3*cm))
        
        # 基本信息
        story.append(Paragraph(f"原文档：{os.path.basename(original_file)}", self.styles['ChineseBody']))
        story.append(Paragraph(f"译文档：{os.path.basename(translated_file)}", self.styles['ChineseBody']))
        story.append(Paragraph(f"审计时间：{datetime.now().strftime('%Y-%m-%d %H:%M:%S')}", self.styles['ChineseBody']))
        story.append(Spacer(1, 0.5*cm))
        
        # 汇总统计
        story.append(Paragraph("一、质量指标汇总", self.styles['ChineseHeading']))
        story.append(self.generate_summary_table(stats))
        story.append(Spacer(1, 0.5*cm))
        
        # 问题明细
        story.append(Paragraph("二、问题明细", self.styles['ChineseHeading']))
        story.append(self.generate_detail_table(issues))
        story.append(Spacer(1, 0.5*cm))
        
        # 结论
        story.append(Paragraph("三、审计结论", self.styles['ChineseHeading']))
        if not issues:
            conclusion = "本次审计未发现明显问题，译文质量良好。"
        else:
            high_count = sum(1 for i in issues if i.get('severity') == 'high')
            if high_count == 0:
                conclusion = f"本次审计发现 {len(issues)} 个轻微问题，建议优化后使用。"
            else:
                conclusion = f"本次审计发现 {len(issues)} 个问题，其中 {high_count} 个严重问题，必须修正后方可使用。"
        
        story.append(Paragraph(conclusion, self.styles['ChineseBody']))
        
        # 生成PDF
        self.doc.build(story)
        print(f"审计报告已生成：{self.output_path}")


# ============ 使用示例 ============
if __name__ == "__main__":
    # 模拟统计数据
    stats = {
        'term_consistency': 97.5,
        'format_integrity': 100.0,
        'page_match': True,
        'image_integrity': 100.0
    }
    
    # 模拟问题列表
    issues = [
        {
            'type': '术语不一致',
            'severity': 'medium',
            'page': 3,
            'detail': '"Machine Learning" 在第3页译为"机器学习"，第15页译为"机器进修"'
        },
        {
            'type': '格式异常',
            'severity': 'low',
            'page': 7,
            'detail': '表格列宽与原文不一致'
        }
    ]
    
    # 生成报告
    generator = AuditReportGenerator("translation_audit_report.pdf")
    generator.create_report(
        original_file="original.pdf",
        translated_file="translated.pdf",
        stats=stats,
        issues=issues
    )

四、完整运行流程

4.1 环境准备

# 创建虚拟环境
python -m venv venv
source venv/bin/activate  # Linux/Mac
# 或 venv\Scripts\activate  # Windows

# 安装依赖
pip install PyMuPDF reportlab

4.2 术语表准备

创建 glossary.json：

{
    "Machine Learning": "机器学习",
    "Deep Learning": "深度学习",
    "Natural Language Processing": "自然语言处理",
    "PDF": "便携式文档格式",
    "API": "应用程序接口"
}

4.3 运行审计

# main.py
from pathlib import Path

def run_audit(original_pdf, translated_pdf, glossary_path=None):
    """运行完整审计流程"""
    
    # 1. 术语一致性检查
    term_checker = TerminologyChecker(glossary_path)
    term_issues = term_checker.check_consistency(translated_pdf)
    
    # 2. 格式检查
    format_checker = FormatChecker()
    format_checker.check_page_count(original_pdf, translated_pdf)
    format_checker.check_image_integrity(translated_pdf)
    format_issues = format_checker.issues
    
    # 3. 合并问题列表
    all_issues = term_issues + format_issues
    
    # 4. 统计指标
    stats = {
        'term_consistency': 100 - len(term_issues) * 0.5,
        'format_integrity': 100 - len(format_issues) * 2,
        'page_match': not any(i['type'] == 'page_count_mismatch' for i in format_issues),
        'image_integrity': 100 - sum(1 for i in format_issues if i['type'] == 'corrupted_image') * 5
    }
    
    # 5. 生成报告
    output_path = f"audit_{Path(translated_pdf).stem}.pdf"
    generator = AuditReportGenerator(output_path)
    generator.create_report(original_pdf, translated_pdf, stats, all_issues)
    
    return output_path

# 执行
if __name__ == "__main__":
    report = run_audit(
        original_pdf="input.pdf",
        translated_pdf="output.pdf",
        glossary_path="glossary.json"
    )
    print(f"审计完成：{report}")

五、扩展建议

5.1 集成到CI/CD流程

# .github/workflows/translation-audit.yml
name: Translation Quality Audit
on:
  push:
    paths:
      - 'docs/translated/**'

jobs:
  audit:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v3
      - name: Setup Python
        uses: actions/setup-python@v4
        with:
          python-version: '3.10'
      - name: Run Audit
        run: |
          pip install -r requirements.txt
          python main.py --original docs/original/ --translated docs/translated/
      - name: Upload Report
        uses: actions/upload-artifact@v3
        with:
          name: audit-reports
          path: audit_*.pdf

5.2 与PDFTranslator集成

PDFTranslator.org 提供翻译后的PDF下载，你可以将上述审计流程作为后处理步骤：

1. 下载翻译后的PDF
2. 运行审计脚本
3. 自动生成质量报告
4. 根据报告评分决定是否重新翻译

六、总结

本文提供了一套完整的PDF翻译质量审计方案，核心要点：

1. 术语一致性检查：基于预定义术语表，确保专业词汇翻译统一
2. 格式完整性检查：验证页数、图片、布局是否与原文一致
3. 自动化报告生成：用ReportLab输出专业PDF报告
4. 可扩展架构：易于集成到现有工作流

完整代码已开源在GitHub（示例仓库），欢迎Star和PR。

---

相关工具推荐：

• PDFTranslator.org --- 在线PDF翻译工具，支持保留原始格式
• PyMuPDF --- Python PDF处理库
• ReportLab --- Python PDF生成库