From 9dea01580a827edbed53f20cf90fb2ed43f4ff9d Mon Sep 17 00:00:00 2001
From: gedoor <gekunfei@live.com>
Date: Sun, 18 Jul 2021 22:22:39 +0800
Subject: [PATCH] HtmlFormatter.kt

---
 .../java/io/legado/app/utils/HtmlFormatter.kt | 90 ++++++++++++-------
 1 file changed, 59 insertions(+), 31 deletions(-)

diff --git a/app/src/main/java/io/legado/app/utils/HtmlFormatter.kt b/app/src/main/java/io/legado/app/utils/HtmlFormatter.kt
index 4ba022a73..8d086b55c 100644
--- a/app/src/main/java/io/legado/app/utils/HtmlFormatter.kt
+++ b/app/src/main/java/io/legado/app/utils/HtmlFormatter.kt
@@ -22,20 +22,21 @@ object HtmlFormatter {
 
     fun formatKeepImg(html: String?, redirectUrl: URL?): String {
         html ?: return ""
-        val keepImgHtml = html.replace(wrapHtmlRegex, "\n")
-            .replace(notImgHtmlRegex, "")
-            .replace("\\n\\s*$|^\\s*\\n".toRegex(), "")
-            .replace("\\n\\s*\\n".toRegex(), "\n")
+        val keepImgHtml = formatKeepImg(html)
 
         var str = StringBuffer()
         var endPos = 0
         var hasMatch = true //普通图片标签是否还未处理过
         var hasMatchX = false //是否存在带参数或带数据属性的图片标签
         var pos = -1
-        val list = listOf("<img[^>]*src *= *\"([^\"{]*\\{(?:[^{}]|\\{[^}]+\\})+\\})\"[^>]*>","<img[^>]*data-[^=]*= *\"([^\"]*)\"[^>]*>","<img[^>]*src *= *\"([^\"]*)\"[^>]*>") //优先匹配用户处理过所以带参数的图片标签，其次匹配带数据属性的图片标签
-        while(++pos<3){
-            if(pos == 2) {
-                if(hasMatchX)break //普通图片标签只在不存在存在带参数或带数据属性的图片标签的时候匹配
+        val list = listOf(
+            "<img[^>]*src *= *\"([^\"{]*\\{(?:[^{}]|\\{[^}]+\\})+\\})\"[^>]*>",
+            "<img[^>]*data-[^=]*= *\"([^\"]*)\"[^>]*>",
+            "<img[^>]*src *= *\"([^\"]*)\"[^>]*>"
+        ) //优先匹配用户处理过所以带参数的图片标签，其次匹配带数据属性的图片标签
+        while (++pos < 3) {
+            if (pos == 2) {
+                if (hasMatchX) break //普通图片标签只在不存在存在带参数或带数据属性的图片标签的时候匹配
                 else hasMatch = false //匹配带参数或带数据属性的图片标签同时也会格式化普通图片标签
             }
             var appendPos = 0
@@ -48,22 +49,38 @@ object HtmlFormatter {
 
                     if (hasMatch) { //格式化不带参数和数据属性的普通图片标签
                         var appendPos0 = 0
-                        val matcher0 = Pattern.compile("<img[^>]*src *= *\"([^\"]*)\"[^>]*>", Pattern.CASE_INSENSITIVE).matcher(strBefore) //格式化普通图片标签
+                        val matcher0 = Pattern.compile(
+                            "<img[^>]*src *= *\"([^\"]*)\"[^>]*>",
+                            Pattern.CASE_INSENSITIVE
+                        ).matcher(strBefore) //格式化普通图片标签
                         while (matcher0.find()) {
-                            sb.append(strBefore.substring(appendPos0, matcher0.start()),"<img src=\"${NetworkUtils.getAbsoluteURL(redirectUrl,matcher0.group(1)!!)}\">")
+                            sb.append(
+                                strBefore.substring(appendPos0, matcher0.start()),
+                                "<img src=\"${
+                                    NetworkUtils.getAbsoluteURL(
+                                        redirectUrl,
+                                        matcher0.group(1)!!
+                                    )
+                                }\">"
+                            )
                             appendPos0 = matcher0.end()
                         }
                         strBefore = strBefore.substring(appendPos0, strBefore.length)
                     }
 
-                    sb.append( strBefore,"<img src=\"${
-                        if (pos == 0) {
-                            val url = matcher.group(1)!!
-                            val urlMatcher = AnalyzeUrl.paramPattern.matcher(url)
-                            if(urlMatcher.find()) NetworkUtils.getAbsoluteURL(redirectUrl,url.substring(0,urlMatcher.start())) + ',' + url.substring(urlMatcher.end())
-                            else NetworkUtils.getAbsoluteURL(redirectUrl,url)
-                        }else NetworkUtils.getAbsoluteURL(redirectUrl,matcher.group(1)!!)
-                    }\">")
+                    sb.append(
+                        strBefore, "<img src=\"${
+                            if (pos == 0) {
+                                val url = matcher.group(1)!!
+                                val urlMatcher = AnalyzeUrl.paramPattern.matcher(url)
+                                if (urlMatcher.find()) NetworkUtils.getAbsoluteURL(
+                                    redirectUrl,
+                                    url.substring(0, urlMatcher.start())
+                                ) + ',' + url.substring(urlMatcher.end())
+                                else NetworkUtils.getAbsoluteURL(redirectUrl, url)
+                            } else NetworkUtils.getAbsoluteURL(redirectUrl, matcher.group(1)!!)
+                        }\">"
+                    )
 
                     appendPos = matcher.end()
                 } while (matcher.find())
@@ -79,19 +96,30 @@ object HtmlFormatter {
 
         if (endPos < keepImgHtml.length) {
             str.append(
-                    if(hasMatchX){ //处理末尾的普通图片标签
-                        var appendPos0 = 0
-                        val strBefore = keepImgHtml.substring(endPos, keepImgHtml.length)
-                        val matcher0 =
-                            Pattern.compile("<img[^>]*src *= *\"([^\"]*)\"[^>]*>", Pattern.CASE_INSENSITIVE)
-                                .matcher(strBefore) //格式化普通图片标签
-                        while (matcher0.find()) {
-                            str.append(strBefore.substring(appendPos0, matcher0.start()),"<img src=\"${ NetworkUtils.getAbsoluteURL( redirectUrl, matcher0.group(1)!! )  }\">")
-                            appendPos0 = matcher0.end()
-                        }
-                        strBefore.substring( appendPos0, strBefore.length )
-                    }else keepImgHtml.substring( endPos, keepImgHtml.length )
-                    )  //缩进图片之后的非空白段落
+                if (hasMatchX) { //处理末尾的普通图片标签
+                    var appendPos0 = 0
+                    val strBefore = keepImgHtml.substring(endPos, keepImgHtml.length)
+                    val matcher0 =
+                        Pattern.compile(
+                            "<img[^>]*src *= *\"([^\"]*)\"[^>]*>",
+                            Pattern.CASE_INSENSITIVE
+                        )
+                            .matcher(strBefore) //格式化普通图片标签
+                    while (matcher0.find()) {
+                        str.append(
+                            strBefore.substring(appendPos0, matcher0.start()),
+                            "<img src=\"${
+                                NetworkUtils.getAbsoluteURL(
+                                    redirectUrl,
+                                    matcher0.group(1)!!
+                                )
+                            }\">"
+                        )
+                        appendPos0 = matcher0.end()
+                    }
+                    strBefore.substring(appendPos0, strBefore.length)
+                } else keepImgHtml.substring(endPos, keepImgHtml.length)
+            )  //缩进图片之后的非空白段落
         }
         return str.toString()
     }