mirror of
https://github.com/201206030/novel-plus.git
synced 2025-04-27 01:30:51 +00:00
36 lines
2.7 KiB
Java
36 lines
2.7 KiB
Java
/*
|
||
Navicat MySQL Data Transfer
|
||
|
||
Source Server : localhost
|
||
Source Server Version : 50624
|
||
Source Host : localhost:3306
|
||
Source Database : novel_biz
|
||
|
||
Target Server Type : MYSQL
|
||
Target Server Version : 50624
|
||
File Encoding : 65001
|
||
|
||
Date: 2020-05-02 12:40:12
|
||
*/
|
||
|
||
SET FOREIGN_KEY_CHECKS=0;
|
||
|
||
-- ----------------------------
|
||
-- Table structure for crawl_source
|
||
-- ----------------------------
|
||
DROP TABLE IF EXISTS `crawl_source`;
|
||
CREATE TABLE `crawl_source` (
|
||
`id` int(11) NOT NULL AUTO_INCREMENT COMMENT '主键',
|
||
`source_name` varchar(50) DEFAULT NULL COMMENT '源站名',
|
||
`crawl_rule` varchar(2048) DEFAULT NULL COMMENT '爬取规则（json串）',
|
||
`source_status` tinyint(1) DEFAULT '0' COMMENT '爬虫源状态，0：关闭，1：开启',
|
||
`create_time` datetime DEFAULT NULL COMMENT '创建时间',
|
||
`update_time` datetime DEFAULT NULL COMMENT '更新时间',
|
||
PRIMARY KEY (`id`)
|
||
) ENGINE=InnoDB AUTO_INCREMENT=3 DEFAULT CHARSET=utf8mb4 COMMENT='爬虫源表';
|
||
|
||
-- ----------------------------
|
||
-- Records of crawl_source
|
||
-- ----------------------------
|
||
INSERT INTO `crawl_source` VALUES ('2', '百书斋', '{\r\n \"bookListUrl\": \"https://m.baishuzhai.com/blhb/{catId}/{page}.html\",\r\n \"catIdRule\": {\r\n \"catId1\": \"1\",\r\n \"catId2\": \"2\",\r\n \"catId3\": \"3\",\r\n \"catId4\": \"4\",\r\n \"catId5\": \"5\",\r\n \"catId6\": \"6\",\r\n \"catId7\": \"7\"\r\n },\r\n \"bookIdPatten\": \"href=\\\"/ibook/(\\\\d+/\\\\d+)/\\\"\",\r\n \"pagePatten\": \"value=\\\"(\\\\d+)/\\\\d+\\\"\",\r\n \"totalPagePatten\": \"value=\\\"\\\\d+/(\\\\d+)\\\"\",\r\n \"bookDetailUrl\": \"https://m.baishuzhai.com/ibook/{bookId}/\",\r\n \"bookNamePatten\": \"<span class=\\\"title\\\">([^/]+)</span>\",\r\n \"authorNamePatten\": \">作者：([^/]+)<\",\r\n \"picUrlPatten\": \"<img src=\\\"([^>]+)\\\"\\\\s+onerror=\\\"this.src=\",\r\n \"statusPatten\": \"状态：([^/]+)</li>\",\r\n \"bookStatusRule\": {\r\n \"连载\": 0,\r\n \"完成\": 1\r\n },\r\n \"scorePatten\": \"<div\\\\s+class=\\\"score\\\">(\\\\d+\\\\.\\\\d+)分</div>\",\r\n \"descStart\": \"<p class=\\\"review\\\">\",\r\n \"descEnd\": \"</p>\",\r\n \"upadateTimePatten\": \"更新：(\\\\d+-\\\\d+-\\\\d+)</li>\",\r\n \"upadateTimeFormatPatten\": \"yy-MM-dd\",\r\n \"bookIndexUrl\": \"https://m.baishuzhai.com/ibook/{bookId}/all.html\",\r\n \"indexIdPatten\": \"<a\\\\s+style=\\\"\\\"\\\\s+href=\\\"/ibook/\\\\d+/\\\\d+/(\\\\d+)\\\\.html\\\">[^/]+</a>\",\r\n \"indexNamePatten\": \"<a\\\\s+style=\\\"\\\"\\\\s+href=\\\"/ibook/\\\\d+/\\\\d+/\\\\d+\\\\.html\\\">([^/]+)</a>\",\r\n \"bookContentUrl\": \"https://baishuzhai.com/ibook/{bookId}/{indexId}.html\",\r\n \"contentStart\": \"id=\\\"content\\\">\",\r\n \"contentEnd\": \"<script>\"\r\n}', '1', '2020-05-01 14:22:50', '2020-05-01 14:22:50');
|