mirror of
https://github.com/201206030/novel-plus.git
synced 2025-07-01 15:26:37 +00:00
Compare commits
6 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
| efb136e3be | |||
| 7955db0e3c | |||
| 60dc28c5ed | |||
| 1534220f0c | |||
| 0830f6ffeb | |||
| adc83db64e |
@ -5,7 +5,7 @@
|
||||
|
||||
<groupId>com.java2nb</groupId>
|
||||
<artifactId>novel-admin</artifactId>
|
||||
<version>5.1.4</version>
|
||||
<version>5.1.5</version>
|
||||
<packaging>jar</packaging>
|
||||
|
||||
<name>novel-admin</name>
|
||||
|
||||
@ -5,7 +5,7 @@
|
||||
<parent>
|
||||
<artifactId>novel</artifactId>
|
||||
<groupId>com.java2nb</groupId>
|
||||
<version>5.1.4</version>
|
||||
<version>5.1.5</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
|
||||
@ -5,7 +5,7 @@
|
||||
<parent>
|
||||
<artifactId>novel</artifactId>
|
||||
<groupId>com.java2nb</groupId>
|
||||
<version>5.1.4</version>
|
||||
<version>5.1.5</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
|
||||
@ -100,6 +100,22 @@ public class CrawlParser {
|
||||
.replaceAll("<p>\\s*</p>", "")
|
||||
.replaceAll("<p>", "")
|
||||
.replaceAll("</p>", "<br/>");
|
||||
// 小说简介过滤
|
||||
String filterDesc = ruleBean.getFilterDesc();
|
||||
if (StringUtils.isNotBlank(filterDesc)) {
|
||||
String[] filterRules = filterDesc.replace("\r\n", "\n").split("\n");
|
||||
for (String filterRule : filterRules) {
|
||||
if (StringUtils.isNotBlank(filterRule)) {
|
||||
desc = desc.replaceAll(filterRule, "");
|
||||
}
|
||||
}
|
||||
}
|
||||
// 去除小说简介前后空格
|
||||
desc = desc.trim();
|
||||
// 去除小说简介末尾冗余的小说名
|
||||
if (desc.endsWith(bookName)) {
|
||||
desc = desc.substring(0, desc.length() - bookName.length());
|
||||
}
|
||||
//设置书籍简介
|
||||
book.setBookDesc(desc);
|
||||
if (StringUtils.isNotBlank(ruleBean.getStatusPatten())) {
|
||||
@ -151,7 +167,7 @@ public class CrawlParser {
|
||||
}
|
||||
|
||||
public boolean parseBookIndexAndContent(String sourceBookId, Book book, RuleBean ruleBean,
|
||||
Map<Integer, BookIndex> existBookIndexMap, CrawlBookChapterHandler handler) throws InterruptedException{
|
||||
Map<Integer, BookIndex> existBookIndexMap, CrawlBookChapterHandler handler) throws InterruptedException {
|
||||
|
||||
Date currentDate = new Date();
|
||||
|
||||
@ -238,6 +254,8 @@ public class CrawlParser {
|
||||
}
|
||||
}
|
||||
}
|
||||
// 去除小说内容末尾的所有换行
|
||||
content = removeTrailingBrTags(content);
|
||||
//插入章节目录和章节内容
|
||||
BookIndex bookIndex = new BookIndex();
|
||||
bookIndex.setIndexName(indexName);
|
||||
@ -314,4 +332,12 @@ public class CrawlParser {
|
||||
return false;
|
||||
|
||||
}
|
||||
|
||||
/**
|
||||
* 删除字符串末尾的所有 <br> 类似标签（允许各种空格）
|
||||
*/
|
||||
public static String removeTrailingBrTags(String str) {
|
||||
return str.replaceAll("(?i)(?:\\s*<\\s*br\\s*/?\\s*>)++(?:\\s|\\u3000)*$", "");
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
@ -45,6 +45,7 @@ public class RuleBean {
|
||||
private String visitCountPatten;
|
||||
private String descStart;
|
||||
private String descEnd;
|
||||
private String filterDesc;
|
||||
private String upadateTimePatten;
|
||||
private String upadateTimeFormatPatten;
|
||||
private String bookIndexUrl;
|
||||
|
||||
@ -118,6 +118,9 @@
|
||||
示例：<b></p></b>
|
||||
<li><input type="text" id="descEnd" class="s_input icon_key" placeholder="小说简介结束截取字符串：">
|
||||
</li>
|
||||
示例：<b><span\s+class="allshow">([^/]+)</span></b>
|
||||
<li><textarea id="filterDesc"
|
||||
placeholder="过滤简介（多个内容换行）" rows="5" cols="52"></textarea></li>
|
||||
示例：<b>更新：(\d+-\d+-\d+\s\d+:\d+:\d+)</a></b>
|
||||
<li><input type="text" id="upadateTimePatten" class="s_input icon_key"
|
||||
placeholder="小说更新时间的正则表达式："></li>
|
||||
@ -338,6 +341,9 @@
|
||||
|
||||
crawlRule.descEnd = descEnd;
|
||||
|
||||
var filterDesc = $("#filterDesc").val();
|
||||
crawlRule.filterDesc = filterDesc;
|
||||
|
||||
var upadateTimePatten = $("#upadateTimePatten").val();
|
||||
|
||||
if (upadateTimePatten.length > 0) {
|
||||
|
||||
@ -119,6 +119,9 @@
|
||||
示例：<b></p></b>
|
||||
<li><input type="text" id="descEnd" class="s_input icon_key" placeholder="小说简介结束截取字符串：">
|
||||
</li>
|
||||
示例：<b><span\s+class="allshow">([^/]+)</span></b>
|
||||
<li><textarea id="filterDesc"
|
||||
placeholder="过滤简介（多个内容换行）" rows="5" cols="52"></textarea></li>
|
||||
示例：<b>更新：(\d+-\d+-\d+\s\d+:\d+:\d+)</a></b>
|
||||
<li><input type="text" id="upadateTimePatten" class="s_input icon_key"
|
||||
placeholder="小说更新时间的正则表达式："></li>
|
||||
@ -266,6 +269,7 @@
|
||||
$("#visitCountPatten").val(crawlRule.visitCountPatten);
|
||||
$("#descStart").val(crawlRule.descStart);
|
||||
$("#descEnd").val(crawlRule.descEnd);
|
||||
$("#filterDesc").val(crawlRule.filterDesc);
|
||||
$("#upadateTimePatten").val(crawlRule.upadateTimePatten);
|
||||
$("#upadateTimeFormatPatten").val(crawlRule.upadateTimeFormatPatten);
|
||||
$("#bookIndexUrl").val(crawlRule.bookIndexUrl);
|
||||
@ -424,6 +428,9 @@
|
||||
|
||||
crawlRule.descEnd = descEnd;
|
||||
|
||||
var filterDesc = $("#filterDesc").val();
|
||||
crawlRule.filterDesc = filterDesc;
|
||||
|
||||
var upadateTimePatten = $("#upadateTimePatten").val();
|
||||
|
||||
if (upadateTimePatten.length > 0) {
|
||||
|
||||
@ -5,7 +5,7 @@
|
||||
<parent>
|
||||
<artifactId>novel</artifactId>
|
||||
<groupId>com.java2nb</groupId>
|
||||
<version>5.1.4</version>
|
||||
<version>5.1.5</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
|
||||
Reference in New Issue
Block a user