用 Ollama 搭一个本地 AI 翻译工具，文档翻译 + 术语管理，零成本零隐私风险

# 安装 Ollama
curl -fsSL https://ollama.com/install.sh | sh

# 拉取翻译模型，qwen2.5-7b 翻译质量很好，8GB 内存就能跑
ollama pull qwen2.5:7b

{
  "source_lang": "en",
  "target_lang": "zh",
  "terms": {
    "latency": "延迟",
    "throughput": "吞吐量",
    "idempotent": "幂等的",
    "eventual consistency": "最终一致性",
    "circuit breaker": "熔断器",
    "rate limiting": "速率限制",
    "microservice": "微服务"
  }
}

#!/usr/bin/env python3
"""本地 AI 翻译工具：支持 .docx / .pdf 文档批量翻译"""
import json
import sys
from pathlib import Path
import ollama

# ============ 配置 ============
BASE_DIR = Path(__file__).parent
CONFIG_DIR = BASE_DIR / "config"
TERMS_PATH = CONFIG_DIR / "terms.json"
OUTPUT_DIR = BASE_DIR / "output"
OUTPUT_DIR.mkdir(exist_ok=True)

# ============ 加载术语表 ============
def load_terms() -> dict:
    with open(TERMS_PATH) as f:
        cfg = json.load(f)
    return cfg.get("terms", {})

# ============ 文档读取 ============
def read_docx(path: Path) -> str:
    """读取 .docx 文档内容"""
    from docx import Document
    doc = Document(path)
    return "\n".join(p.text for p in doc.paragraphs if p.text.strip())

def read_pdf(path: Path) -> str:
    """读取 .pdf 文档内容"""
    import PyPDF2
    text_parts = []
    with open(path, "rb") as f:
        reader = PyPDF2.PdfReader(f)
        for page in reader.pages:
            t = page.extract_text()
            if t:
                text_parts.append(t)
    return "\n".join(text_parts)

def read_text(path: Path) -> str:
    return path.read_text(encoding="utf-8")

def read_document(path: Path) -> str:
    if path.suffix == ".docx":
        return read_docx(path)
    elif path.suffix == ".pdf":
        return read_pdf(path)
    elif path.suffix == ".txt":
        return read_text(path)
    else:
        raise ValueError(f"不支持的文件类型: {path.suffix}")

# ============ 翻译核心 ============
def translate_chunk(text: str, terms: dict, model: str = "qwen2.5:7b") -> str:
    """翻译一段文本，插入术语表保证一致性"""
    # 构建术语提示
    term_hint = ""
    if terms:
        lines = [f"- {k} → {v}" for k, v in terms.items()]
        term_hint = "\n请优先使用以下术语翻译：\n" + "\n".join(lines)

    prompt = f"""你是一个专业的技术翻译助手。请将以下英文文本翻译成中文，要求：
1. 翻译准确、流畅，符合中文技术文档的表达习惯
2. 专业术语保持一致{term_hint}
3. 保留原文的段落结构和格式
4. 直接输出翻译结果，不要加任何解释

原文：
{text}

中文翻译：
"""
    response = ollama.chat(model=model, messages=[{"role": "user", "content": prompt}])
    return response["message"]["content"].strip()

# ============ 批量翻译 ============
def translate_document(doc_path: Path, model: str = "qwen2.5:7b", chunk_size: int = 2000):
    """翻译整个文档，分段处理避免超出上下文限制"""
    terms = load_terms()
    raw_text = read_document(doc_path)

    if not raw_text.strip():
        print(f"⚠️ 文档内容为空：{doc_path.name}")
        return

    print(f"📄 正在翻译：{doc_path.name}（共 {len(raw_text)} 字）")

    # 分段翻译
    chunks = [raw_text[i:i + chunk_size] for i in range(0, len(raw_text), chunk_size)]
    translated_parts = []

    for i, chunk in enumerate(chunks):
        print(f"  段落 {i + 1}/{len(chunks)} ...")
        translated = translate_chunk(chunk, terms, model)
        translated_parts.append(translated)

    full_translation = "\n\n".join(translated_parts)

    # 输出到文件
    out_name = doc_path.stem + "_translated" + doc_path.suffix
    out_path = OUTPUT_DIR / out_name
    out_path.write_text(full_translation, encoding="utf-8")
    print(f"✅ 翻译完成，已保存：{out_path}")
    return out_path

# ============ 命令行入口 ============
if __name__ == "__main__":
    import argparse
    parser = argparse.ArgumentParser(description="本地 AI 翻译工具")
    parser.add_argument("document", nargs="?", help="要翻译的文档路径（.docx/.pdf/.txt）")
    parser.add_argument("--batch", "-b", action="store_true", help="批量翻译 documents/ 目录下的所有文档")
    parser.add_argument("--model", "-m", default="qwen2.5:7b", help="Ollama 模型名称")
    args = parser.parse_args()

    if args.batch:
        docs = list((BASE_DIR / "documents").glob("*"))
        docs = [d for d in docs if d.suffix in (".docx", ".pdf", ".txt")]
        if not docs:
            print("📁 documents/ 目录下没有找到可翻译的文档")
            sys.exit(1)
        print(f"🔍 发现 {len(docs)} 个待翻译文档\n")
        for doc in docs:
            translate_document(doc, args.model)
            print()
    elif args.document:
        doc_path = Path(args.document)
        if not doc_path.exists():
            print(f"❌ 文件不存在：{doc_path}")
            sys.exit(1)
        translate_document(doc_path, args.model)
    else:
        parser.print_help()

pip install -r requirements.txt

python translator.py documents/product_spec.docx

python translator.py --batch