Update HtmlFormatter.kt

优化
pull/1123/head
bushixuanqi 3 years ago committed by GitHub
parent 0c5c7e06b8
commit 2312e24f0c
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
  1. 72
      app/src/main/java/io/legado/app/utils/HtmlFormatter.kt

@ -11,10 +11,10 @@ object HtmlFormatter {
fun format(html: String?, otherRegex: Regex = otherHtmlRegex): String { fun format(html: String?, otherRegex: Regex = otherHtmlRegex): String {
html ?: return "" html ?: return ""
return html.replace(wrapHtmlRegex, "\n") return html.replace(wrapHtmlRegex, "\n")
.replace(otherRegex, "") .replace(otherRegex, "")
.replace("\\s*\\n+\\s*".toRegex(), "\n  ") .replace("\\s*\\n+\\s*".toRegex(), "\n  ")
.replace("^[\\n\\s]+".toRegex(), "  ") .replace("^[\\n\\s]+".toRegex(), "  ")
.replace("[\\n\\s]+$".toRegex(), "") .replace("[\\n\\s]+$".toRegex(), "")
} }
fun formatKeepImg(html: String?) = format(html, notImgHtmlRegex) fun formatKeepImg(html: String?) = format(html, notImgHtmlRegex)
@ -22,59 +22,43 @@ object HtmlFormatter {
fun formatKeepImg(html: String?, redirectUrl: URL?): String { fun formatKeepImg(html: String?, redirectUrl: URL?): String {
html ?: return "" html ?: return ""
val keepImgHtml = html.replace(wrapHtmlRegex, "\n") val keepImgHtml = html.replace(wrapHtmlRegex, "\n")
.replace(notImgHtmlRegex, "") .replace(notImgHtmlRegex, "")
.replace("[\\n\\s]+\$|^[\\n\\s]*".toRegex(), "") .replace("[\\n\\s]+\$|^[\\n\\s]*".toRegex(), "")
.replace("\\s*\\n+\\s*".toRegex(), "\n") .replace("\\s*\\n+\\s*".toRegex(), "\n")
val sb = StringBuffer() val sb = StringBuffer()
val hasDataType:Boolean //是否有数据属性 val hasDataType:Boolean //是否有数据属性
//图片有data-开头的数据属性时优先用数据属性作为src,没有数据属性时匹配src //图片有data-开头的数据属性时优先用数据属性作为src,没有数据属性时匹配src
val imgPattern = Pattern.compile( val imgPattern = Pattern.compile(
if(keepImgHtml.matches("<img[^>]*data-".toRegex())) { if(keepImgHtml.matches("<img[^>]*data-".toRegex())) {
hasDataType = true hasDataType = true
"<img[^>]*data-[^=]*= *\"([^\"])\"[^>]*>" "<img[^>]*data-[^=]*= *\"([^\"])\"[^>]*>"
} }
else { else {
hasDataType = false hasDataType = false
"<img[^>]*src *= *\"([^\"{]+(?:\\{(?:[^{}]|\\{[^{}]*\\})*\\})?)\"[^>]*>" "<img[^>]*src *= *\"([^\"{]+(?:\\{(?:[^{}]|\\{[^{}]*\\})*\\})?)\"[^>]*>"
}, Pattern.CASE_INSENSITIVE }, Pattern.CASE_INSENSITIVE
) )
val matcher = imgPattern.matcher(keepImgHtml) val matcher = imgPattern.matcher(keepImgHtml)
var appendPos = 0 var appendPos = 0
if(matcher.find()){ while(matcher.find()){
if(hasDataType || matcher.group(1)!!.indexOf(',') == -1) { //图片无参 val url = matcher.group(1)!!
val urlBefore = url.substringBefore(',')
do{ sb.append( keepImgHtml.substring(appendPos, matcher.start()).replace("\n","\n  ") ) //缩进换行下个非图片段落
sb.append(keepImgHtml.substring(appendPos, matcher.start()).replace("\n","\n  ")) //非图片部分换行缩进 sb.append(
sb.append( "<img src=\"${ "<img src=\"${
NetworkUtils.getAbsoluteURL(redirectUrl,matcher.group(1)!!) NetworkUtils.getAbsoluteURL(
}\">" )
appendPos = matcher.end()
}while (matcher.find())
}else{ //图片有参
do{
val url = matcher.group(1)!!
val urlBefore = url.substringBefore(',')
sb.append( keepImgHtml.substring(appendPos, matcher.start()).replace("\n","\n  ") ) //缩进换行下个非图片段落
sb.append(
"<img src=\"${
NetworkUtils.getAbsoluteURL(
redirectUrl, redirectUrl,
urlBefore urlBefore
) )
}${ }${
url.substring(urlBefore.length) url.substring(urlBefore.length)
}\">" }\">"
) )
appendPos = matcher.end() appendPos = matcher.end()
}while(matcher.find())
}
} }
if (appendPos < keepImgHtml.length) { if (appendPos < keepImgHtml.length) {

Loading…
Cancel
Save