69 lines
2.1 KiB
Python
69 lines
2.1 KiB
Python
from mcp.server.fastmcp import FastMCP
|
|
import fitz
|
|
import httpx
|
|
import json
|
|
import os
|
|
|
|
mcp = FastMCP("pdf")
|
|
|
|
with open(os.path.join(os.path.dirname(__file__), '../server_config.json'), 'r', encoding='utf-8') as f:
|
|
config = json.load(f)
|
|
|
|
OLLAMA_URL = config["OLLAMA_URL"]
|
|
OLLAMA_MODEL = config["OLLAMA_MODEL"]
|
|
|
|
async def translate_text(text: str) -> str:
|
|
prompt = f"请将以下内容翻译成流畅、准确的中文，仅输出翻译结果：{text}"
|
|
payload = {
|
|
"model": OLLAMA_MODEL,
|
|
"prompt": prompt,
|
|
}
|
|
async with httpx.AsyncClient() as client:
|
|
try:
|
|
resp = await client.post(OLLAMA_URL, json=payload, timeout=120.0)
|
|
resp.raise_for_status()
|
|
lines = resp.text.strip().splitlines()
|
|
responses = []
|
|
for line in lines:
|
|
try:
|
|
result = json.loads(line)
|
|
if "response" in result and result["response"]:
|
|
responses.append(result["response"])
|
|
except Exception:
|
|
continue
|
|
if responses:
|
|
return "".join(responses)
|
|
return "翻译失败：无有效返回内容"
|
|
except Exception as e:
|
|
return f"翻译失败: {e}"
|
|
|
|
def extract_pdf_text(pdf_path: str) -> str:
|
|
try:
|
|
doc = fitz.open(pdf_path)
|
|
text = "\n".join(page.get_text() for page in doc)
|
|
doc.close()
|
|
return text
|
|
except Exception as e:
|
|
return f"PDF解析失败: {e}"
|
|
|
|
@mcp.tool()
|
|
async def translate_pdf(pdf_path: str) -> str:
|
|
"""
|
|
读取PDF文件内容并翻译成中文。
|
|
Args:
|
|
pdf_path: PDF文件的绝对路径
|
|
"""
|
|
text = extract_pdf_text(pdf_path)
|
|
if text.startswith("PDF解析失败"):
|
|
return text
|
|
max_len = 2000
|
|
chunks = [text[i:i+max_len] for i in range(0, len(text), max_len)]
|
|
translated = []
|
|
for chunk in chunks:
|
|
zh = await translate_text(chunk)
|
|
translated.append(zh)
|
|
return "\n".join(translated)
|
|
|
|
if __name__ == "__main__":
|
|
mcp.run(transport='stdio')
|