FORMAT_CONVERSION_MAXCOMPUTE/format_convert/libreoffice_interface.py

import base64
import json
import os
import re
import shutil
import subprocess
import sys
import traceback
import psutil
sys.path.append(os.path.dirname(os.path.abspath(__file__)) + "/../")
from format_convert.judge_platform import get_platform
import logging
from format_convert.utils import my_subprocess_call
from flask import Flask, request
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')


def monitor_libreoffice():
    try:
        # logging.info("=========================================")
        logging.info("into monitor_libreoffice")
        # logging.info("------------------------------MEM top 10")
        os.system("ps aux|head -1;ps aux|grep -v PID|sort -rn -k +4|head")
        #
        # logging.info("--------------------------soffice process")
        # os.system("ps -ef | grep soffice")

        pids = psutil.pids()
        for pid in pids:
            try:
                process = psutil.Process(pid)
                # if process.username() == "appuser":
                if re.search("soffice", process.exe()):
                    if str(pid) == str(globals().get("soffice_pid")):
                        logging.info("process " + str(pid) + str(process.exe()))
                        comm = "kill -9 " + str(pid)
                        os.system(comm)
                        logging.info("killed soffice" + str(pid))
                elif re.search("unrar", process.exe()):
                    logging.info("process " + str(pid) + str(process.exe()))
                    comm = "kill -9 " + str(pid)
                    os.system(comm)
                    logging.info("killed unrar" + str(pid))

            except TimeoutError:
                raise TimeoutError
            except:
                continue
        # logging.info("=========================================")
    except TimeoutError:
        raise TimeoutError


# @timeout_decorator.timeout(120, timeout_exception=TimeoutError, use_signals=False)
def office_convert(src_path, dest_path, target_format, retry_times=1):
    try:
        logging.info("into office_convert")
        # print("src_path", src_path)
        uid1 = src_path.split(os.sep)[-1].split(".")[0]
        dest_file_path = dest_path + uid1 + "." + target_format
        src_format = src_path.split(".")[-1]

        # 重试转换
        for i in range(retry_times):
            # 调用Win下的libreoffice子进程
            if get_platform() == "Windows":
                soffice = 'C:\\Program Files\\LibreOfficeDev 5\\program\\soffice.exe'
                comm_list = [soffice, '--headless', '--convert-to', target_format, src_path,
                             '--outdir', dest_path+os.sep]

                try:
                    p = subprocess.call(comm_list, timeout=30*(i+2))
                except:
                    continue

            # 调用Linux下的libreoffice子进程
            else:
                # 先杀libreoffice进程
                # monitor_libreoffice()

                # 再调用转换
                libreoffice_dir = 'soffice'
                comm_list = [libreoffice_dir, '--headless', '--convert-to', target_format, src_path,
                             '--outdir', dest_path+os.sep]

                comm = ''
                for c in comm_list:
                    comm += c + ' '
                # logging.info("office_convert command" + comm)
                try:
                    # p = subprocess.call(comm_list, timeout=30*(i+2))
                    # os.system(comm)
                    pid, p_code = my_subprocess_call(comm_list, timeout=30*(i+1))
                    logging.info("subprocess code " + str(p_code))
                    globals().update({"soffice_pid": pid})
                except TimeoutError:
                    return [-5]
                except Exception as e:
                    print(1, src_format + ' to ' + target_format + ' Failed! Retry...', i, 'times')
                    traceback.print_exc()
                    continue

            # 执行失败，重试
            if not os.path.exists(dest_file_path):
                print(2, src_format + ' to ' + target_format + ' Failed! Retry...', i, 'times')
                continue
            # 执行成功，跳出循环
            else:
                break

        # 重试后还未成功
        if not os.path.exists(dest_file_path):
            logging.info(str(3) + src_format + ' to ' + target_format + " failed!")
            return [-3]

        logging.info("out office_convert")
        return dest_file_path
    except TimeoutError:
        return [-5]


# 接口配置
app = Flask(__name__)


@app.route('/soffice', methods=['POST'])
def _office_convert():
    src_path = None
    try:
        logging.info("into office_convert")

        if not request.form:
            logging.info("office_convert no data!")
            return {"data": []}

        src_path = request.form.get("src_path")
        dest_path = request.form.get("dest_path")
        file_b64 = request.form.get("file")
        file_bytes = base64.b64decode(file_b64)
        target_format = request.form.get("target_format")
        # retry_times = int(request.form.get("retry_times"))

        uid1 = src_path.split(os.sep)[-1].split(".")[0]
        dest_file_path = dest_path + uid1 + "." + target_format
        src_format = src_path.split(".")[-1]

        if not os.path.exists(os.path.dirname(src_path)):
            os.makedirs(os.path.dirname(src_path), mode=0o777)
        with open(src_path, "wb") as f:
            f.write(file_bytes)

        # 调用Win下的libreoffice子进程
        if get_platform() == "Windows":
            soffice = 'C:\\Program Files\\LibreOfficeDev 5\\program\\soffice.exe'
            comm_list = [soffice, '--headless', '--convert-to', target_format, src_path,
                         '--outdir', dest_path+os.sep]
            p = subprocess.call(comm_list, timeout=10)

        # 调用Linux下的libreoffice子进程
        else:
            # 再调用转换
            libreoffice_dir = 'soffice'
            comm_list = [libreoffice_dir, '--headless', '--convert-to', target_format, src_path,
                         '--outdir', dest_path+os.sep]
            comm = ''
            for c in comm_list:
                comm += c + ' '
            logging.info("office_convert command" + comm)

            # p = subprocess.call(comm_list, timeout=30*(i+2))
            # os.system(comm)
            pid, p_code = my_subprocess_call(comm_list, timeout=10)
            logging.info("subprocess code " + str(p_code))

        # 重试后还未成功
        if not os.path.exists(dest_file_path):
            logging.info(str(3) + src_format + ' to ' + target_format + " failed!")
            return {"data": [-3]}

        logging.info("out office_convert")
        with open(dest_file_path, "rb") as f:
            file_bytes = f.read()
        base64_stream = base64.b64encode(file_bytes)

        # temp_dir = "/data/fangjiasheng/format_conversion_maxcompute/format_convert/temp/"
        # if os.path.exists(temp_dir):
        #     shutil.rmtree(temp_dir)

        print("base64_stream", type(base64_stream))
        return {"data": str(file_bytes)}
    except TimeoutError:
        return {"data": [-5]}
    except:
        traceback.print_exc()
        return {"data": [-1]}
    finally:
        if src_path is not None:
            file_dir = os.path.dirname(src_path)
            if os.path.exists(file_dir):
                logging.info("delete " + str(file_dir))
                shutil.rmtree(file_dir)


if __name__ == "__main__":
    port = 16000
    os.system("service cron start")
    app.run(host='0.0.0.0', port=port, threaded=False, debug=False)