简要说明:运用到的框架是webmagic mongonDB做数据库落地数据
maven 依赖jar以及版本
<dependency>
<groupId>us.codecraft</groupId>
<artifactId>webmagic-core</artifactId>
<version>0.7.3</version>
</dependency>
<dependency>
<groupId>us.codecraft</groupId>
<artifactId>webmagic-extension</artifactId>
<version>0.7.3</version>
</dependency>
入库实体类:
package com.huoli.trip.operation.core.service;
import com.sleepycat.persist.model.Entity;
import lombok.*;
import org.springframework.data.mongodb.core.mapping.Document;
import org.springframework.data.mongodb.core.mapping.Field;
import java.io.Serializable;
@AllArgsConstructor
@NoArgsConstructor
@Setter
@Getter
@Builder
@Entity
@Document(collection = "trip_test")
public class AddressDb implements Serializable {
// 省市区ID
@Field("code")
private String code;
// 省市区名称
@Field("name")
private String name;
// 等级类型
@Field("level")
private Short level;
}
爬取逻辑:
package com.huoli.trip.operation.core.service;
import com.xiaoleilu.hutool.util.CharsetUtil;
import com.xiaoleilu.hutool.util.CollectionUtil;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.data.mongodb.core.MongoTemplate;
import org.springframework.data.mongodb.core.query.Criteria;
import org.springframework.data.mongodb.core.query.Query;
import org.springframework.stereotype.Component;
import us.codecraft.webmagic.Page;
import us.codecraft.webmagic.Site;
import us.codecraft.webmagic.processor.PageProcessor;
import java.util.List;
@Component
public class CityProcessor implements PageProcessor {
@Autowired
private MongoTemplate mongoTemplate;
private static final String TJSJ_CITY_BASE_URL = "http://www.stats.gov.cn/tjsj/tjbz/tjyqhdmhcxhfdm/2019/";
private static final String TJSJ_CITY_WEB_URL = "^http://www.stats.gov.cn/tjsj/tjbz/tjyqhdmhcxhfdm/2019/[A-Za-z]+.html$";
// 获取市信息
private static final String PROVINCE_URL = "^http://www.stats.gov.cn/tjsj/tjbz/tjyqhdmhcxhfdm/2019/\\d{1,2}.html$";
// 获取区县信息
private static final String CITY_URL = "^http://www.stats.gov.cn/tjsj/tjbz/tjyqhdmhcxhfdm/2019/\\d{1,2}/\\d{1,4}.html$";
private Site site = Site.me()
.setDomain("http://www.stats.gov.cn")
.setSleepTime(5000)
.setRetryTimes(3)
.setCharset(CharsetUtil.GBK)
.setUserAgent("Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/75.0.3770.100 Safari/537.36");
public void process(Page page) {
List<String> urlList;
String baseUrl;
if (page.getUrl().regex(TJSJ_CITY_WEB_URL).match()) {
// 获取省份code列表
List<String> provinceCodes = page.getHtml().regex("<td><a href=\\\"(.{1,30}).html\\\">.*?<br></a></td>").all();
// 获取省份信息列表
List<String> provinceNames = page.getHtml().regex("<td><a href=\\\".*?.html\\\">(.{1,30})<br></a></td>").all();
AddressDb addressDb;
for (int i = 0, n = provinceCodes.size(); i < n; i++) {
addressDb = AddressDb.builder()
.code(provinceCodes.get(i)+"0000")
.name(provinceNames.get(i))
.level((short) 1).build();
mongoTemplate.insert(addressDb);
}
urlList = page.getHtml().xpath("//*[@class=\"provincetr\"]/td/a/@href").all();
if (CollectionUtil.isNotEmpty(urlList)) {
urlList.stream().forEach( str -> page.addTargetRequest(TJSJ_CITY_BASE_URL + str));
}
} else if (page.getUrl().regex(PROVINCE_URL).match()) {
// 获取数据库中省份Code和省份信息
String code = page.getUrl().regex("http://www.stats.gov.cn/tjsj/tjbz/tjyqhdmhcxhfdm/2019/(.{2,}).html").toString()+"0000";
Query query = new Query(new Criteria().and("code").is(code));
AddressDb provinceInfo = mongoTemplate.findOne(query, AddressDb.class);
if (provinceInfo != null) {
AddressDb addressDb;
// 获取城市信息
List<String> cityInfos = page.getHtml().regex("<td><a href=\\\".*?.html\\\">(.{1,30})</a></td>").all();
for (int i = 0, n = cityInfos.size(); i < n; i++) {
addressDb = AddressDb.builder()
.code(cityInfos.get(i).substring(0, 6))
.name(cityInfos.get(++i))
.level((short) 2).build();
mongoTemplate.insert(addressDb);
}
}
urlList = page.getHtml().xpath("//*[@class=\"citytr\"]/td/a/@href").all();
if (CollectionUtil.isNotEmpty(urlList)) {
urlList.stream().distinct().forEach( str -> page.addTargetRequest(TJSJ_CITY_BASE_URL + str));
}
} else if (page.getUrl().regex(CITY_URL).match()) {
// 获取数据库中城市Code和信息
String code = page.getUrl().regex("http://www.stats.gov.cn/tjsj/tjbz/tjyqhdmhcxhfdm/2019/\\d{1,}/(.{2,}).html").toString() + "00";
Query query = new Query(new Criteria().and("code").is(code));
AddressDb cityInfo = mongoTemplate.findOne(query, AddressDb.class);
// 获取区县信息
List<String> countyInfos = page.getHtml().regex("<td><a href=\\\".*?.html\\\">(.{1,30})</a></td>").all();
if (cityInfo != null) {
AddressDb addressDb;
for (int i = 0, n = countyInfos.size(); i < n; i++) {
addressDb = AddressDb.builder()
.code(countyInfos.get(i).substring(0, 6))
.name(countyInfos.get(++i))
.level((short) 3).build();
mongoTemplate.insert(addressDb);
}
}
baseUrl = page.getUrl().regex("^http://www.stats.gov.cn/tjsj/tjbz/tjyqhdmhcxhfdm/2019/\\d{1,2}/").toString();
urlList = page.getHtml().xpath("//*[@class=\"countytr\"]/td/a/@href").all();
if (CollectionUtil.isNotEmpty(urlList)) {
urlList.stream().distinct().forEach( str -> page.addTargetRequest(baseUrl + str));
}
}
}
public Site getSite() {
return site;
}
}
控制器入口:
package com.huoli.trip.operation.core.controller;
import com.huoli.trip.operation.core.service.CityProcessor;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RestController;
import us.codecraft.webmagic.Spider;
import us.codecraft.webmagic.downloader.HttpClientDownloader;
@RestController
public class NovelController {
@Autowired
private CityProcessor cityProcessor;
@RequestMapping("getCityData")
public void crawelCity(String url) {
HttpClientDownloader httpClientDownloader = new HttpClientDownloader();
Spider.create(cityProcessor).addUrl("http://www.stats.gov.cn/tjsj/tjbz/tjyqhdmhcxhfdm/2019/index.html").setDownloader(httpClientDownloader).thread(10).run();
}
}
上图为落地数据document结构,根据源数据表,设计mongo项目中业务需求需要的表结构
@Test public void testCity(){ CityRequest cityRequest = new CityRequest(); cityRequest.setCode("440100".substring(0,4)); List<AreaPO> areaPOS = cityDao.queryAreaList(cityRequest); System.out.println(areaPOS); // CityRequest cityRequest = new CityRequest(); // Criteria criteria = Criteria.where("country").exists(true); // MatchOperation matchOperation = Aggregation.match(criteria); // ArrayList<MatchOperation> matchOperations = Lists.newArrayList(matchOperation); // Aggregation aggregation = Aggregation.newAggregation(matchOperations); // AggregationResults<ProvincePO> output = mongoTemplate.aggregate(aggregation, Constants.TRIP_BASE_PROVINCE, ProvincePO.class); // // output.getMappedResults().forEach(t->{ // t.getProvincePOS().forEach(cityPO ->{ // System.out.println(cityPO.getId()+" 输出了 "+cityPO.getName()); // }); // }); /** * 源 */ // List<TripTest> all = mongoTemplate.findAll(TripTest.class); /** * 区 */ // Map<String, List<TripTest>> cityMap = all.stream().filter(tripTest -> !tripTest.getId().substring(2).equals("0000")&&(tripTest.getName().contains("区")||tripTest.getName().contains("县"))).collect(Collectors.groupingBy(tripTest -> tripTest.getId().substring(0, 4))); // Set<Map.Entry<String, List<TripTest>>> entries = cityMap.entrySet(); // for (Map.Entry<String, List<TripTest>> map: entries){ // List<AreaPO> areaPOS = new ArrayList<>(); // List<TripTest> value = map.getValue(); // value.forEach(tripTest -> { // AreaPO areaPO = new AreaPO(); // areaPO.setId(tripTest.getId()); // areaPO.setLevel(4); // areaPO.setName(tripTest.getName().replaceAll("\\u00A0","").trim()); // areaPO.setCityId(map.getKey()); // areaPOS.add(areaPO); // }); // AreaPO areaPO = new AreaPO(); // areaPO.setCityId(map.getKey()); // areaPO.setAreaPOS(areaPOS); // mongoTemplate.insert(areaPO); // } // Query query = new Query(); // query.addCriteria(Criteria.where("level").is(1)); // Update update = new Update(); // update.set("areas",areaPOS); // mongoTemplate.upsert(query, update, BaseCityPO.class, Constants.TRIP_BASE_CITY); /** * 城市 */ // Map<String, List<TripTest>> cityMap = all.stream().filter(tripTest -> tripTest.getName().contains("市")).collect(Collectors.groupingBy(tripTest -> tripTest.getId().substring(0, 2))); // Set<Map.Entry<String, List<TripTest>>> entries = cityMap.entrySet(); // // for (Map.Entry<String, List<TripTest>> map: entries){ // List<BaseCityPO> cities = new ArrayList<>(); // List<TripTest> value = map.getValue(); // value.forEach(tripTest -> { // BaseCityPO city = new BaseCityPO(); // city.setId(tripTest.getId()); // city.setLevel(3); // city.setName(tripTest.getName().replaceAll("\\u00A0","").trim()); // city.setProvinceId(map.getKey()); // cities.add(city); // }); // BaseCityPO cityPO = new BaseCityPO(); // cityPO.setProvinceId(map.getKey()); // cityPO.setCityPOS(cities); // mongoTemplate.insert(cityPO); // } //更新操作 // Query query = new Query(); // query.addCriteria(Criteria.where("level").is(1)); // Update update = new Update(); // update.set("citys",cities); // mongoTemplate.upsert(query, update, BaseCityPO.class, Constants.TRIP_BASE_CITY); /** * 省 */ // String country = "中国"; // all.forEach(tripTest -> { // if(tripTest.getName().contains("省")){ // ProvincePO province = new ProvincePO(); // province.setId(tripTest.getId()); // province.setLevel(2); // province.setName(tripTest.getName()); // province.setCountry(country); // mongoTemplate.insert(province); // } // }); }