清除从word粘贴过来多余html代码

原创

iteyer 2021-07-30 15:10:05 ©著作权

文章标签 html cms function class qt 文章分类 JavaScript 前端开发

©著作权归作者所有：来自51CTO博客作者iteyer的原创作品，请联系作者获取转载授权，否则将追究法律责任

function CleanWord(html)
{
html = REReplaceNocase(html,’ s* ’,’
’,’all’);
html = REReplaceNocase(html,’ .*? ’,’ ’,’all’);
html = REReplaceNocase(html,’s*mso-[^:]+:[^;"]+;?’,’
’,’all’);
html = REReplaceNocase(html,’s*MARGIN: 0cm 0cm 0pts*;’,’
’,’all’);
html = REReplaceNocase(html,’s*MARGIN: 0cm 0cm 0pts*"’,’"’,’all’);
html = REReplaceNocase(html,’s*TEXT-INDENT: 0cms*;’,’
’,’all’);
html = REReplaceNocase(html,’s*TEXT-INDENT: 0cms*"’,’"’,’all’);
html = REReplaceNocase(html,’s*TEXT-ALIGN: [^s;]+;?"’,’"’,’all’);
html = REReplaceNocase(html,’s*PAGE-BREAK-BEFORE: [^s;]+;?"’,’"’,’all’);
html = REReplaceNocase(html,’s*FONT-VARIANT: [^s;]+;?"’,’"’,’all’);
html = REReplaceNocase(html,’s*tab-stops:[^;"]*;?’,’
’,’all’);
html = REReplaceNocase(html,’s*tab-stops:[^"]*’,’
’,’all’);
html = REReplaceNocase(html,’s*face="[^"]*"’,’
’,’all’);
html = REReplaceNocase(html,’s*face=[^ >]*’,’
’,’all’);
html = REReplaceNocase(html,’s*FONT-FAMILY:[^;"]*;?’,’
’,’all’);
html = REReplaceNocase(html,’<(w[^>]*) class=([^ |>]*)([^>]*)’,’< ’,’all’);
html = REReplaceNocase(html,’<(w[^>]*) style="([^"]*)"([^>]*)’,’< ’,’all’);
html = REReplaceNocase(html,’s*style="s*"’,’
’,’all’);
html = REReplaceNocase(html,’ ]*>s* s*’,’ ’,’all’);
html = REReplaceNocase(html,’ ]*>’,’
’,’all’);
html = REReplaceNocase(html,’<(w[^>]*) lang=([^ |>]*)([^>]*)’,’< ’,’all’);
html = REReplaceNocase(html,’ (.*?)’,’ ’,’all’);
html = REReplaceNocase(html,’ (.*?)’,’ ’,’all’);
html = REReplaceNocase(html,’ ]*>’,’
’,’all’);
html = REReplaceNocase(html,’ ]*>’,’
’,’all’);
html = REReplaceNocase(html,’ s* ’,’
’,’all’);
html = REReplaceNocase(html,’ ]*)>’,’

’,’all’);
html = REReplaceNocase(html,’ ]*)>’,’

’,’all’);
html = REReplaceNocase(html,’’,’

’,’all’);
html = REReplaceNocase(html,’<(U|I|STRIKE)> ’,’ ’,’all’);
html = REReplaceNocase(html,’<([^s>]+)[^>]*>s* ’,’
’,’all’);
html = REReplaceNocase(html,’<([^s>]+)[^>]*>s* ’,’
’,’all’);
html = REReplaceNocase(html,’<([^s>]+)[^>]*>s* ’,’
’,’all’);
html = REReplaceNocase(html,’( ]*>.*?)(

)’,’

’,’all’);
return html;
}