抓取html 里面的标签,过滤html里的标签获取纯文本

首先定义宏

#define REGEX_SCRIPT @"

#define REGEX_STYLE @"

#define REGEX_HTML @"<[^>]+>"

#define REGEX_SPACE @"\\s*|\t|\r|\n"

#define REGEX_DEL_P @"?[a|A][^>]*>"

之后就是一层层的过滤

- (NSString *)regexString:(NSString *)htmlString{

NSRegularExpression *reExFirst = [NSRegularExpression regularExpressionWithPattern:REGEX_SCRIPT options:0 error:nil];

htmlString = [reExFirst stringByReplacingMatchesInString:htmlString options:NSMatchingReportProgress range:NSMakeRange(0, htmlString.length) withTemplate:@""];

NSRegularExpression *reExSecond = [NSRegularExpression regularExpressionWithPattern:REGEX_STYLE options:0 error:nil];

htmlString = [reExSecond stringByReplacingMatchesInString:htmlString options:NSMatchingReportProgress range:NSMakeRange(0, htmlString.length) withTemplate:@""];

NSRegularExpression *reExThird = [NSRegularExpression regularExpressionWithPattern:REGEX_HTML options:0 error:nil];

htmlString = [reExThird stringByReplacingMatchesInString:htmlString options:NSMatchingReportProgress range:NSMakeRange(0, htmlString.length) withTemplate:@""];

NSRegularExpression *reExFourth = [NSRegularExpression regularExpressionWithPattern:REGEX_SPACE options:0 error:nil];

htmlString = [reExFourth stringByReplacingMatchesInString:htmlString options:NSMatchingReportProgress range:NSMakeRange(0, htmlString.length) withTemplate:@""];

NSRegularExpression *reExFifth = [NSRegularExpression regularExpressionWithPattern:REGEX_DEL_P options:0 error:nil];

htmlString = [reExFifth stringByReplacingMatchesInString:htmlString options:NSMatchingReportProgress range:NSMakeRange(0, htmlString.length) withTemplate:@""];

return htmlString;

}