增加百书斋源

novel-front/pom.xml

@@ -10,7 +10,7 @@
     </parent>
     <groupId>xyz.zinglizingli</groupId>
     <artifactId>novel-front</artifactId>
-    <version>2.2.0.beta</version>
+    <version>2.3.0.beta</version>
     <name>novel-front</name>
     <description>小说精品楼-前台web网站</description>
 

novel-front/src/main/java/xyz/zinglizingli/books/core/config/CrawlBaishuzhaiConfig.java

@@ -0,0 +1,29 @@
+package xyz.zinglizingli.books.core.config;
+
+import lombok.extern.slf4j.Slf4j;
+import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
+import org.springframework.boot.context.properties.ConfigurationProperties;
+import org.springframework.context.annotation.Bean;
+import org.springframework.context.annotation.Configuration;
+import org.springframework.context.annotation.Primary;
+import xyz.zinglizingli.books.core.crawl.BaseHtmlCrawlSource;
+import xyz.zinglizingli.books.core.crawl.BiquCrawlSource;
+
+/**
+ * @author 11797
+ */
+@Slf4j
+@Configuration
+public class CrawlBaishuzhaiConfig {
+
+
+    @Bean
+    @Primary //必须加此注解，不然报错，下一个类则不需要添加
+    @ConfigurationProperties(prefix = "baishuzhai.crawlsource") // prefix值必须是application.yml中对应属性的前缀
+    @ConditionalOnProperty(prefix = "crawl.website",name = "type",havingValue = "4")
+    public BaseHtmlCrawlSource dingdianCrawlSource() {
+        return new BiquCrawlSource();
+    }
+
+
+}

novel-front/src/main/java/xyz/zinglizingli/books/core/crawl/BiquCrawlSource.java

@@ -91,7 +91,13 @@ public class BiquCrawlSource extends BaseHtmlCrawlSource {
                                 Matcher updateTimeMatch = updateTimePatten.matcher(body);
                                 if (updateTimeMatch.find()) {
                                     String updateTimeStr = updateTimeMatch.group(1);
-                                    SimpleDateFormat format = new SimpleDateFormat("yy-MM-dd HH:mm:ss");
+                                    SimpleDateFormat format ;
+                                    if(updateTimeStr.length()>10){
+
+                                        format = new SimpleDateFormat("yy-MM-dd HH:mm:ss");
+                                    }else{
+                                        format = new SimpleDateFormat("yy-MM-dd");
+                                    }
                                     updateTime = format.parse(updateTimeStr);
                                     if(!newCat2Date.containsKey(i)) {
                                         newCat2Date.put(i, updateTime);
@@ -159,7 +165,13 @@ public class BiquCrawlSource extends BaseHtmlCrawlSource {
                                 Matcher updateTimeMatch = updateTimePatten.matcher(body);
                                 if (updateTimeMatch.find()) {
                                     String updateTimeStr = updateTimeMatch.group(1);
-                                    SimpleDateFormat format = new SimpleDateFormat("yy-MM-dd HH:mm:ss");
+                                    SimpleDateFormat format ;
+                                    if(updateTimeStr.length()>10){
+
+                                        format = new SimpleDateFormat("yy-MM-dd HH:mm:ss");
+                                    }else{
+                                        format = new SimpleDateFormat("yy-MM-dd");
+                                    }
                                     Date updateTime = format.parse(updateTimeStr);
                                     Pattern picPatten = compile(getPicPattern());
                                     Matcher picMather = picPatten.matcher(body);

novel-front/src/main/resources/application-crawl.yml

@@ -54,6 +54,23 @@ dingdian:
     catalog-url-pattern: <a\s+href="(/ddk\d+/all.html)">查看完整目录</a>
     catalog-pattern: <a\s+style=""\s+href="(/ddk\d+/\d+\.html)">([^/]+)</a>
 
+
+baishuzhai:
+  crawlsource:
+    index-url: https://m.baishuzhai.com
+    list-page-url: https://m.baishuzhai.com/sort/{0}/{1}.html
+    book-url-pattern: href="/(ibook/\d+/\d+)/"
+    score-pattern: <div\s+class="score">(\d+\.\d+)分</div>
+    book-name-pattern: <p class="title">([^/]+)</p>
+    author-pattern: 作者：([^/]+)<
+    status-pattern: 状态：([^/]+)</li>
+    cat-pattern: 类别：([^/]+)</li>
+    update-time-pattern: 更新：(\d+-\d+-\d+)</li>
+    pic-pattern: <img src="([^>]+)"\s+onerror="this.src=
+    intro-pattern: class="review">([^/]+)</p>
+    catalog-url-pattern: <a\s+href="(/ibook/\d+/\d+/all\.html)">查看完整目录</a>
+    catalog-pattern: <a\s+style=""\s+href="(/ibook/\d+/\d+/\d+\.html)">([^/]+)</a>
+
 biquge:
   crawlsource:
     index-url: http://m.biquge.info

novel-front/src/main/resources/application.yml

@@ -84,10 +84,10 @@ books:
   #小说的更新间隔（分）
   updatePeriod: 1
 
-#爬取的网站名称类型 1：笔趣岛 ，2：笔趣塔,3:顶点小说  更多网站解析中，敬请期待
+#爬取的网站名称类型 1：笔趣岛 ，2：笔趣塔,3:顶点小说 ，4：百书斋 更多网站解析中，敬请期待
 crawl:
   website:
-    type: 2
+    type: 4
 
 
 

script/crawlbook/application.yml

@@ -1,14 +1,15 @@
 server: {port: 8083}
 spring:
-  datasource: {url: 'jdbc:mysql://47.106.243.172:3306/novel?useUnicode=true&characterEncoding=utf-8&useSSL=false&allowPublicKeyRetrieval=true&serverTimezone=Asia/Shanghai',
+  datasource: {url: 'jdbc:mysql://127.0.0.1:3306/books?useUnicode=true&characterEncoding=utf-8&useSSL=false&allowPublicKeyRetrieval=true&serverTimezone=Asia/Shanghai',
-    username: novel, password: novel!8888}
+    username: root, password: test123456}
 mybatis:
   mapper-locations: classpath:mybatis/mapping/*.xml
   type-aliases-package: xyz.zinglizingli.books.po
   configuration: {log-impl: org.apache.ibatis.logging.stdout.StdOutImpl}
 mysql: {charset: utf8mb4}
-books: {lowestScore: '9.0'}
+books: {lowestScore: 9.0}
 crawl:
-  website: {type: '2'}
+  website: {type: '4'}
   soft-novel: '0'
   manhua: '0'
+logging: {config: 'classpath:logback-boot.xml'}