Merge pull request #1058 from bushixuanqi/patch-37

Update HtmlFormatter.kt
pull/1059/head
kunfei 3 years ago committed by GitHub
commit fcdc167d3f
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
  1. 54
      app/src/main/java/io/legado/app/utils/HtmlFormatter.kt

@ -2,55 +2,49 @@ package io.legado.app.utils
import io.legado.app.model.analyzeRule.AnalyzeUrl
import java.net.URL
import java.util.regex.Pattern
import io.legado.app.constant.AppPattern
object HtmlFormatter {
private val wrapHtmlRegex = "</?(?:div|p|br|hr|h\\d|article|dd|dl)[^>]*>".toRegex()
private val notImgHtmlRegex = "</?(?!img)[a-zA-Z]+(?=[ >])[^<>]*>".toRegex()
private val otherHtmlRegex = "</?[a-zA-Z]+(?=[ >])[^<>]*>".toRegex()
private val imgPattern = Pattern.compile("<img [^>]*src=.*?\"(.*?(?:,\\{.*\\})?)\".*?>")
fun format(html: String?): String {
fun format(html: String?, otherRegex: Regex = otherHtmlRegex): String {
html ?: return ""
return html.replace(wrapHtmlRegex, "\n")
.replace(otherHtmlRegex, "")
.replace("\\s*\\n+\\s*".toRegex(), "\n  ")
.replace(otherRegex, "")
.replace("^[\\n\\s]+".toRegex(), "  ")
.replace("[\\n\\s]+$".toRegex(), "")
}
fun formatKeepImg(html: String?): String {
html ?: return ""
return html.replace(wrapHtmlRegex, "\n")
.replace(notImgHtmlRegex, "")
.replace("\\s*\\n+\\s*".toRegex(), "\n  ")
.replace("^[\\n\\s]+".toRegex(), "  ")
.replace("[\\n\\s]+$".toRegex(), "")
}
fun formatKeepImg(html: String?) = format(html,notImgHtmlRegex)
fun formatKeepImg(html: String?, redirectUrl: URL?): String {
html ?: return ""
var formatHtml = formatKeepImg(html)
val sb = StringBuffer()
val matcher = imgPattern.matcher(html)
var appendPos = 0
while (matcher.find()) {
val urlArray = matcher.group(1)!!.split(AnalyzeUrl.splitUrlRegex)
var url = NetworkUtils.getAbsoluteURL(redirectUrl, urlArray[0])
if (urlArray.size > 1) {
url = "$url,${urlArray[1]}"
while (appendPos < formatHtml.length) {
val matcher = AppPattern.imgPattern.matcher(formatHtml)
if(matcher.find()) {
val urlArray = matcher.group(1)!!.split(AnalyzeUrl.splitUrlRegex)
var url = NetworkUtils.getAbsoluteURL(redirectUrl, urlArray[0])
if (urlArray.size > 1) {
url = "$url,${urlArray[1]}"
}
sb.append(formatHtml.substring(appendPos, matcher.start()))
sb.append("<img src=\"$url\" >")
appendPos = matcher.end()
formatHtml = formatHtml.substring(appendPos, formatHtml.length)
appendPos = 0
} else {
sb.append(formatHtml)
appendPos = formatHtml.length
}
sb.append(html.substring(appendPos, matcher.start()))
sb.append("<img src=\"$url\" >")
appendPos = matcher.end()
}
if (appendPos < html.length) {
sb.append(html.substring(appendPos, html.length))
}
return sb.replace(wrapHtmlRegex, "\n")
.replace(notImgHtmlRegex, "")
.replace("\\s*\\n+\\s*".toRegex(), "\n  ")
.replace("^[\\n\\s]+".toRegex(), "  ")
.replace("[\\n\\s]+$".toRegex(), "")
return sb.toString()
}
}
Loading…
Cancel
Save