Python把word文档批量写入MySQL数据库

话不多说了，希望大家喜欢，我也是新手。

首先第一步 ：电脑磁盘中的xls，xlsx，doc，docx文件找出来文件名，路径保存到另一个Excel文件：

BASE_DIR = Path(__file__).resolve().parent.parent
data = []
i = 0
e = 0
r = 0
t = 0


workbookXls = xlwt.Workbook(encoding="utf-8")
worksheetXls = workbookXls.add_sheet("xls")
workbookXlsx = xlwt.Workbook(encoding="utf-8")
worksheetXlsx = workbookXlsx.add_sheet("xlsx")
workbookDoc  = xlwt.Workbook(encoding="utf-8")
worksheetDoc = workbookDoc.add_sheet("doc")
workbookDocx = xlwt.Workbook(encoding="utf-8")
worksheetDocx = workbookDocx.add_sheet("a1")
for disk in psutil.disk_partitions():
    # 获取磁盘列表['C:\\', 'D:\\', 'E:\\', 'F:\\', 'G:\\', 'H:\\', 'I:\\', 'J:\\', 'K:\\', 'L:\\', 'M:\\']
    disk_list = disk.device
    for root, dirs, files in os.walk(disk_list):
        for a_files in files:
            suffex_list = [".xls", ".xlsx", ".doc", ".docx"]
            file_suffex_name = os.path.splitext(a_files)[1].lower()  # 获取文件后缀名 .xlsx等
            file_s_name_ = os.path.splitext(a_files)[0].lower()  # 获取文件名
            for a_suffex_list in suffex_list:
                if a_suffex_list == file_suffex_name and re.search("[\u4e00-\u9fa5]", file_s_name_) and not re.search(
                        "[\$|~]", root) and not re.search("[\$|~]", file_s_name_):
                    # data.append(os.path.join(root, a_files))
                    # print(file_suffex_name)
                    # print(a_files)
                    # print(root)
                    # print(os.path.join(root, a_files))
                    path = os.path.join(root, a_files)  # 获取整个文件地址与文件名
                    if file_suffex_name == ".xls":
                        i += 1
                    elif file_suffex_name == ".xlsx":
                        e += 1
                    elif file_suffex_name == ".doc":
                        r += 1
                    elif file_suffex_name == ".docx":
                        t += 1
                    #print(path)
    workbookXls.save("__$file_list_xls.xls")
    workbookXlsx.save("__$file_list_xlsx.xls")
    workbookDoc.save("__$file_list_doc1.xls")
    workbookDocx.save("__$file_list_docx.xls")

第二部，利用刚才保存过的文件路径和名称，读取doc，docx文件内容转换成HTML，然后写入MySQL数据库：

import time
from pydocx import PyDocX
import os
from win32com import client
import pythoncom
import xlrd
from pathlib import Path
import pymysql
import random


conn = pymysql.connect(host="127.0.0.1", user="root", password="root", db="mysite", port=3306)
BASE_DIR = Path(__file__).resolve().parent.parent

xls_path = "__$file_list_doc1.xls"
x_excel = xlrd.open_workbook(xls_path)
table = x_excel.sheet_by_index(0)
table_length = table.nrows  # 返回列的长度
indexnum =29648

for num in range(int(indexnum), int(table_length)):
    try:
        excel_file_path = table.cell(num, 1).value
        excel_file_root = str(table.cell(num, 2).value)
        excel_file_name = str(table.cell(num, 3).value)
        print(excel_file_path)
        as_docx_path = os.path.join(BASE_DIR, 'temp',
                                    "{}_{}.docx".format(time.strftime("%Y%m%d%H%M%S", time.localtime()), random.random()))

        as_docx_path_html = os.path.join(BASE_DIR, 'temp',
                                         "{}.html".format(
                                             time.strftime("%Y%m%d%H%M", time.localtime())))
        if os.path.splitext(excel_file_name)[1] == ".doc":
            pythoncom.CoInitialize()
            word = client.Dispatch("Word.Application")
            doc = word.Documents.Open(excel_file_path)
            doc.SaveAs(as_docx_path, 12, False, '', False, '', '', False, False, False, False)
            doc.Close()
            html = PyDocX.to_html(as_docx_path)
            f = open(as_docx_path_html, "w", encoding="utf-8")
            f.write(html)
            f.close()
            # return HttpResponse(as_docx_path_html)

        else:
            html = PyDocX.to_html(excel_file_path)
            f = open(as_docx_path_html, "w", encoding="utf-8")
            f.write(html)
            f.close()
            # return HttpResponse(as_docx_path_html)
        f = open(as_docx_path_html, 'r', encoding='utf-8')
        f = f.read()
        cursor = conn.cursor()
        table_name = "docx"
        sql_str_extis = "SHOW TABLES LIKE '{}'".format(table_name)
        if cursor.execute(sql_str_extis) == 0:
            sql_str = "CREATE TABLE `{}` (`id` int(99) NOT NULL AUTO_INCREMENT,`username` text DEFAULT NULL,`filename` text DEFAULT NULL,`pinyin` text DEFAULT NULL,`time` text DEFAULT NULL,`text` longtext,PRIMARY KEY (`id`)) ENGINE=MyISAM DEFAULT CHARSET=utf8;".format(
                table_name)
            cursor.execute(sql_str)
        insert_time = time.strftime("%Y%m%d%H%M", time.localtime())
        r = cursor.execute("INSERT INTO `{}` (`username`, `filename`, `pinyin`, `time`, `text`) VALUES ('adminstator', '{}', '_  ', '{}', '{}')".format(
                table_name, excel_file_name, insert_time, f))

        d = {"file": excel_file_path, "sql_response": r}
        print(d)
    except:
        continue

还有剩下的是Excel内容一行一行的写入数据库比较麻烦，所有下次给大家分享。。