Merge pull request #1126 from bushixuanqi/master

优化图片格式化

app/src/main/java/io/legado/app/model/webBook/BookContent.kt

@@ -52,7 +52,7 @@ object BookContent {
         var contentData = analyzeContent(
             book, baseUrl, redirectUrl, body, contentRule, bookChapter, bookSource, mNextChapterUrl
         )
-        content.append(contentData.content).append("\n")
+        content.append(contentData.content)
         if (contentData.nextUrl.size == 1) {
             var nextUrl = contentData.nextUrl[0]
             while (nextUrl.isNotEmpty() && !nextUrlList.contains(nextUrl)) {
@@ -74,7 +74,7 @@ object BookContent {
                     )
                     nextUrl =
                         if (contentData.nextUrl.isNotEmpty()) contentData.nextUrl[0] else ""
-                    content.append(contentData.content).append("\n")
+                    content.append("\n").append(contentData.content)
                 }
             }
             Debug.log(bookSource.bookSourceUrl, "◇本章总页数:${nextUrlList.size}")
@@ -98,11 +98,10 @@ object BookContent {
                 }
                 asyncArray.forEach { coroutine ->
                     scope.ensureActive()
-                    content.append(coroutine.await()).append("\n")
+                    content.append("\n").append(coroutine.await())
                 }
             }
         }
-        content.deleteCharAt(content.length - 1)
         var contentStr = content.toString()
         val replaceRegex = contentRule.replaceRegex
         if (!replaceRegex.isNullOrEmpty()) {
@@ -150,4 +149,4 @@ object BookContent {
         }
         return ContentData(content, nextUrlList)
     }
-}
+}

app/src/main/java/io/legado/app/utils/HtmlFormatter.kt

@@ -11,10 +11,10 @@ object HtmlFormatter {
     fun format(html: String?, otherRegex: Regex = otherHtmlRegex): String {
         html ?: return ""
         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)
@@ -22,47 +22,37 @@ object HtmlFormatter {
     fun formatKeepImg(html: String?, redirectUrl: URL?): String {
         html ?: return ""
         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 hasDataType:Boolean //是否有数据属性
+        var endPos = 0
-
+        for(pattern in listOf("<img[^>]*data-[^=]*= *\"([^\"])\"[^>]*>","<img[^>]*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
+            var appendPos = 0
-        val imgPattern = Pattern.compile(
+            val matcher = Pattern.compile(pattern, Pattern.CASE_INSENSITIVE).matcher(keepImgHtml)
-                if(keepImgHtml.matches("<img[^>]*data-".toRegex())) {
+            while (matcher.find()) {
-                    hasDataType = true
+                val url = matcher.group(1)!!
-                    "<img[^>]*data-[^=]*= *\"([^\"])\"[^>]*>"
+                val urlBefore = url.substringBefore(',')
-                }
+                val strBefore = keepImgHtml.substring(appendPos, matcher.start())
-                else {
+                sb.append(
-                    hasDataType = false
+                    if(strBefore.isBlank()) strBefore else strBefore.replace("\n", "\n　　") //缩进图片之间的非空白段落
-                    "<img[^>]*src *= *\"([^\"{]+(?:\\{(?:[^{}]|\\{[^{}]*\\})*\\})?)\"[^>]*>"
+                )
-                }, Pattern.CASE_INSENSITIVE
+                sb.append(
-        )
-
-        val matcher = imgPattern.matcher(keepImgHtml)
-        var appendPos = 0
-
-        while(matcher.find()){
-            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()
+            }
+            if(appendPos != 0)endPos = appendPos
         }
 
-        if (appendPos < keepImgHtml.length) {
+        if (endPos < keepImgHtml.length) {
-            sb.append( keepImgHtml.substring( appendPos, keepImgHtml.length ).replace("\n","\n　　") ) //缩进换行下个非图片段落
+            sb.append( keepImgHtml.substring( endPos, keepImgHtml.length ).replace("\n","\n　　") ) //缩进图片之后的非空白段落
         }
         return sb.toString()
     }