|
|
|
|
 |
|
在祐 / 2010-01-21 / 오후 3:27:59 / 24.10.174.180
조회수 : 628
|
| [추천] [퍼온글] 자바스트립트 HTML 태그 제거 |
private String getText(String content) {
Pattern SCRIPTS = Pattern.compile("<(no)?script[^>]*>.*?</(no)?script>",Pattern.DOTALL);
Pattern STYLE = Pattern.compile("<style[^>]*>.*</style>",Pattern.DOTALL);
Pattern TAGS = Pattern.compile("<(\"[^\"]*\"|\'[^\']*\'|[^\'\">])*>");
Pattern nTAGS = Pattern.compile("<\\w+\\s+[^<]*\\s*>");
Pattern ENTITY_REFS = Pattern.compile("&[^;]+;");
Pattern WHITESPACE = Pattern.compile("\\s\\s+");
Matcher m;
m = SCRIPTS.matcher(content);
content = m.replaceAll("");
m = STYLE.matcher(content);
content = m.replaceAll("");
m = TAGS.matcher(content);
content = m.replaceAll("");
m = ENTITY_REFS.matcher(content);
content = m.replaceAll("");
m = WHITESPACE.matcher(content);
content = m.replaceAll(" ");
return content;
}
|
|
|
