package weli.v2;
import org.apache.commons.lang3.StringUtils;
import us.codecraft.webmagic.Page;
import us.codecraft.webmagic.Site;
import us.codecraft.webmagic.Spider;
import us.codecraft.webmagic.processor.PageProcessor;
import us.codecraft.webmagic.selector.Html;
import us.codecraft.webmagic.selector.Selectable;
import java.util.*;
/**
* 根据给定综述介绍页
* 爬取参数页
*/
public class ZolPageProcessor4_ implements PageProcessor {
@Override
public void process(Page page) {
page.setCharset("utf-8");
Map<String, Selectable> map = new HashMap<String, Selectable>() {
{
put("上市日期", null);
put("电商报价", null);
put("主屏分辨率", null);
put("RAM容量", null);
put("ROM容量", null);
put("CPU型号", null);
put("电池容量", null);
put("主屏尺寸", null);
put("手机重量", null);
put("手机尺寸", null);
put("操作系统", null);
put("出厂系统内核", null);
}
};
Html html = page.getHtml();
html.xpath("//div[@class=\"detailed-parameters\"]/table/tbody/tr")
.nodes()
.forEach((selectable) -> {
for (String s1 : map.keySet()) {
if (selectable.regex("[\\d\\D]*" + s1 + "[\\d\\D]*").match()) {
map.put(s1, selectable);
}
}
});
String url = page.getUrl().toString();
String brand = html.xpath("//div[@class=\"product-model page-title clearfix\"]/h1/text()").toString();
Phone phone = new Phone().setBrand(brand).setUrl(url);
if (map.get("上市日期") != null) {
String timeToMarket = map.get("上市日期").xpath("//td/span/text()").toString();
if (StringUtils.isEmpty(timeToMarket)) {
timeToMarket = map.get("上市日期").xpath("//td/span/a/text()").toString();
}
phone.setTimeToMarket(timeToMarket);
}
if (map.get("电商报价") != null) {
String price = map.get("电商报价").xpath("//td/span/text()").toString();
if (StringUtils.isEmpty(price)) {
price = map.get("电商报价").xpath("//td/span/a/text()").toString();
phone.setPrice(price);
}
}
if (map.get("主屏分辨率") != null) {
String resolutionRatio = map.get("主屏分辨率").xpath("//td/span/text()").toString();
if (StringUtils.isEmpty(resolutionRatio)) {
resolutionRatio = map.get("主屏分辨率").xpath("//td/span/a/text()").toString();
}
phone.setResolutionRatio(resolutionRatio);
}
if (map.get("RAM容量") != null) {
String ram = map.get("RAM容量").xpath("//td/span/text()").toString();
if (StringUtils.isEmpty(ram)) {
ram = map.get("RAM容量").xpath("//td/span/a/text()").toString();
}
phone.setRam(ram);
}
if (map.get("ROM容量") != null) {
String rom = map.get("ROM容量").xpath("//td/span/text()").toString();
if (StringUtils.isEmpty(rom)) {
rom = map.get("ROM容量").xpath("//td/span/a/text()").toString();
}
phone.setRom(rom);
}
if (map.get("CPU型号") != null) {
String cpu = map.get("CPU型号").xpath("//td/span/text()").toString();
if (StringUtils.isEmpty(cpu)) {
cpu = map.get("CPU型号").xpath("//td/span/a/text()").toString();
}
phone.setCpu(cpu);
}
if (map.get("主屏尺寸") != null) {
String screen = map.get("主屏尺寸").xpath("//td/span/text()").toString();
if (StringUtils.isEmpty(screen)) {
screen = map.get("主屏尺寸").xpath("//td/span/a/text()").toString();
}
phone.setScreen(screen);
}
if (map.get("手机重量") != null) {
String weight = map.get("手机重量").xpath("//td/span/text()").toString();
if (StringUtils.isEmpty(weight)) {
weight = map.get("手机重量").xpath("//td/span/a/text()").toString();
}
phone.setWeight(weight);
}
if (map.get("手机尺寸") != null) {
String size = map.get("手机尺寸").xpath("//td/span/text()").toString();
if (StringUtils.isEmpty(size)) {
size = map.get("手机尺寸").xpath("//td/span/a/text()").toString();
}
phone.setSize(size);
}
if (map.get("电池容量") != null) {
String battery = map.get("电池容量").xpath("//td/span/text()").toString();
if (StringUtils.isEmpty(battery)) {
battery = map.get("电池容量").xpath("//td/span/a/text()").toString();
}
phone.setBattery(battery);
}
if (map.get("操作系统") != null || map.get("出厂系统内核") != null) {
phone.setOs(Optional.ofNullable(map.get("操作系统"))
.orElseGet(() -> map.get("出厂系统内核"))
.xpath("//td/span/text()")
.toString());
}
phone.format();
page.putField(url, phone);
}
public Phone process(Html html, Phone phone) {
Map<String, Selectable> map = new HashMap<String, Selectable>() {
{
put("上市日期", null);
put("电商报价", null);
put("主屏分辨率", null);
put("RAM容量", null);
put("ROM容量", null);
put("CPU型号", null);
put("电池容量", null);
put("主屏尺寸", null);
put("手机重量", null);
put("手机尺寸", null);
put("操作系统", null);
put("出厂系统内核", null);
}
};
html.xpath("//div[@class=\"detailed-parameters\"]/table/tbody/tr")
.nodes()
.forEach((selectable) -> {
for (String s1 : map.keySet()) {
if (selectable.regex("[\\d\\D]*" + s1 + "[\\d\\D]*").match()) {
map.put(s1, selectable);
}
}
});
String brand = html.xpath("//div[@class=\"product-model page-title clearfix\"]/h1/text()").toString();
phone.setBrand(brand);
if (map.get("上市日期") != null) {
String timeToMarket = map.get("上市日期").xpath("//td/span/text()").toString();
if (StringUtils.isEmpty(timeToMarket)) {
timeToMarket = map.get("上市日期").xpath("//td/span/a/text()").toString();
}
phone.setTimeToMarket(timeToMarket);
}
// if (map.get("电商报价") != null) {
// String price = map.get("电商报价").xpath("//td/span/text()").toString();
// if (StringUtils.isEmpty(price)){
// price = map.get("电商报价").xpath("//td/span/a/text()").toString();
// phone.setPrice(price);
// }
// }
if (map.get("主屏分辨率") != null) {
String resolutionRatio = map.get("主屏分辨率").xpath("//td/span/text()").toString();
if (StringUtils.isEmpty(resolutionRatio)) {
resolutionRatio = map.get("主屏分辨率").xpath("//td/span/a/text()").toString();
webmagic_Demo.rar （59个子文件） 
webmagic_Demo 
pom.xml 2KB webmagic_Demo.iml 81B target test-classes weli AppTest.class 455B generated-test-sources test-annotations classes META-INF webmagic_Demo.kotlin_module 16B weli v3 Process1.class 2KB Process2.class 4KB Process3.class 3KB v2 ZolPageProcessor1.class 3KB ZolPageProcessor4_$1.class 1KB ZolPageProcessor2.class 3KB ZolPageProcessor4_$2.class 1KB ZolPageProcessor4_.class 8KB Phone.class 8KB HttpClient.class 6KB Utils.class 402B ZolPageProcessor.class 2KB etl BrandSeparateVersion.class 5KB BrandSeparateVersion$1.class 1KB SqlPipeline.class 4KB ZolPageProcessor3.class 3KB v1 ZolPageProcessor1.class 3KB ZolPageProcessor2.class 3KB ZolPageProcessor3_.class 3KB Phone.class 2KB ZolPageProcessor.class 2KB SqlPipeline.class 3KB generated-sources annotations src test java weli AppTest.java 296B main java weli v3 Process1.java 1KB Process2.java 1KB Process3.java 1KB v2 ZolPageProcessor1.java 1KB ZolPageProcessor.java 1KB ZolPageProcessor2.java 2KB ZolPageProcessor3.java 2KB SqlPipeline.java 2KB HttpClient.java 6KB ZolPageProcessor4_.java 11KB etl BrandSeparateVersion.java 3KB Phone.java 2KB v1 ZolPageProcessor1.java 1KB ZolPageProcessor3_.java 2KB ZolPageProcessor.java 1KB ZolPageProcessor2.java 2KB SqlPipeline.java 1KB Phone.java 219B .idea dataSources.xml 811B uiDesigner.xml 9KB $PRODUCT_WORKSPACE_FILE$ 1KB $CACHE_FILE$ 2KB misc.xml 513B encodings.xml 262B compiler.xml 610B workspace.xml 10KB dataSources.local.xml 2KB sqldialects.xml 223B dataSources 8e758d00-9aee-4be1-b771-b2f8c26912e8 storage_v2 _src_ schema information_schema.FNRwLQ.meta 76B e96f1308-909f-4634-8af1-c488bc544902 storage_v2 _src_ schema information_schema.FNRwLQ.meta 76B e96f1308-909f-4634-8af1-c488bc544902.xml 37KB 8e758d00-9aee-4be1-b771-b2f8c26912e8.xml 35KB资源评论
_+_
- 粉丝: 95
- 资源: 11
最新资源
- new_bird_c-c语言入门
- christmasTree-圣诞树html网页代码
- working-shell脚本入门——流程控制
- hadoop_install-sqoop数据导入
- ThinkCMF-mysql安装
- BigData-Notes-sqoop的安装与配置
- C语言-leetcode题解之28-implement-strstr.c
- C语言-leetcode题解之27-remove-element.c
- C语言-leetcode题解之26-remove-duplicates-from-sorted-array.c
- C语言-leetcode题解之24-swap-nodes-in-pairs.c
资源上传下载、课程学习等过程中有任何疑问或建议，欢迎提出宝贵意见哦~我们会及时处理！
点击此处反馈
