清除从word粘贴过来多余html代码

最新推荐文章于 2021-06-15 14:15:01 发布

原创最新推荐文章于 2021-06-15 14:15:01 发布 · 2.4k 阅读

CC 4.0 BY-SA版权

文章标签：

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,’</??xml[^>]*>’,’
’,’all’);
html = REReplaceNocase(html,’</?w+:[^>]*>’,’
’,’all’);
html = REReplaceNocase(html,’s*’,’
’,’all’);
html = REReplaceNocase(html,’]*)>’,’

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