添加了pullandbear、unqlo、zara的数据爬虫，待完善数据格式化

src/main/java/com/diaoyun/zion/chinafrica/bis/impl/PullandbearSpider.java

+package com.diaoyun.zion.chinafrica.bis.impl;
+
+import com.diaoyun.zion.chinafrica.bis.IItemSpider;
+import com.diaoyun.zion.chinafrica.enums.PlatformEnum;
+import com.diaoyun.zion.master.util.HttpClientUtil;
+import net.sf.json.JSONArray;
+import net.sf.json.JSONObject;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+import org.springframework.stereotype.Component;
+
+import java.io.IOException;
+import java.net.URISyntaxException;
+import java.util.concurrent.ExecutionException;
+import java.util.concurrent.TimeoutException;
+
+/**
+ * 西班牙年轻时尚品牌-pullandbear 数据爬虫
+ *
+ * TODO 图片路径未处理
+ * 图片路径为：
+ */
+@Component("pullandbearSpider")
+public class PullandbearSpider implements IItemSpider {
+
+    private static Logger logger = LoggerFactory.getLogger(PullandbearSpider.class);
+
+    //PullandBear商品详情
+    private static final String pullandbearUrl="https://www.pullandbear.cn/itxrest/2/catalog/store/24009528/20309423/category/0/product/";
+
+    @Override
+    public JSONObject captureItem(String targetUrl) throws URISyntaxException, IOException, ExecutionException, InterruptedException, TimeoutException {
+        // 提取链接中的商品 id 信息
+        String pId = targetUrl.substring(targetUrl.lastIndexOf("p")+1, targetUrl.lastIndexOf(".html"));
+        // 组成链接并获取商品详情信息的 Json
+        targetUrl = pullandbearUrl + pId + "/detail?languageId=-7&appId=1";
+        // 获取网页内容
+        String content = HttpClientUtil.getContentByUrl(targetUrl, PlatformEnum.PULLANDBEAR.getValue());
+        // 转换为Json格式
+        JSONObject json = JSONObject.fromObject(content);
+        // 简化Json格式
+        json = getMainData(json, content);
+        return json;
+    }
+
+    /**
+     * 简化Json格式
+     * @param json
+     * @param content
+     * @return
+     */
+    private JSONObject getMainData(JSONObject json, String content) {
+        json = json.getJSONArray("bundleProductSummaries").getJSONObject(0);
+        // 商品id
+        String productId = json.getString("id");
+        // 商品名称
+        String productName = json.getString("name");
+        JSONArray colors = json.getJSONObject("detail").getJSONArray("colors");
+        // 返回数据
+        JSONObject returnJson = new JSONObject();
+        returnJson.put("id", productId);
+        returnJson.put("name", productName);
+        returnJson.put("data", colors);
+        return returnJson;
+    }
+
+}
+
+

src/main/java/com/diaoyun/zion/chinafrica/bis/impl/UniqloSpider.java

+package com.diaoyun.zion.chinafrica.bis.impl;
+
+import com.diaoyun.zion.chinafrica.bis.IItemSpider;
+import com.diaoyun.zion.chinafrica.enums.PlatformEnum;
+import com.diaoyun.zion.master.util.HttpClientUtil;
+import net.sf.json.JSONArray;
+import net.sf.json.JSONObject;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+import org.springframework.stereotype.Component;
+
+import java.io.IOException;
+import java.net.URISyntaxException;
+import java.util.concurrent.ExecutionException;
+import java.util.concurrent.TimeoutException;
+
+/**
+ * 优衣库数据爬虫
+ *
+ * TODO 图片路径未处理
+ * 图片路径为："https://www.uniqlo.cn/hmall/test/" + 商品id + "/sku/40/" + 商品图片id + ".jpg"
+ */
+@Component("uniqloSpider")
+public class UniqloSpider implements IItemSpider {
+    private static Logger logger = LoggerFactory.getLogger(UniqloSpider.class);
+
+    // 优衣库数据爬虫
+    private static final String uniqloUrl = "https://www.uniqlo.cn/data/products/spu/";
+
+    @Override
+    public JSONObject captureItem(String targetUrl) throws URISyntaxException, IOException, ExecutionException, InterruptedException, TimeoutException {
+        // 获取商品 id
+        String[] split = targetUrl.split("productCode=");
+        String pId = split[1];
+        // 获取商品详情的json链接
+        targetUrl = uniqloUrl + "zh_CN/" + pId + ".json";
+        // 获取网页内容
+        String content = HttpClientUtil.getContentByUrl(targetUrl, PlatformEnum.UNIQLO.getValue());
+        // 转换为 json
+        JSONObject json = JSONObject.fromObject(content);
+        // 商品id
+        String pName = json.getJSONObject("summary").getString("name");
+        // 商品价格
+        String pPrice = json.getJSONObject("summary").getString("originPrice");
+        // 格式化数据
+        JSONArray rowsJson = json.getJSONArray("rows");
+        JSONObject returnJson = new JSONObject();
+        returnJson.put("name", pName);
+        returnJson.put("price", pPrice);
+        returnJson.elementOpt("data", rowsJson);
+        return returnJson;
+    }
+
+}

src/main/java/com/diaoyun/zion/chinafrica/bis/impl/ZaraSpider.java

+package com.diaoyun.zion.chinafrica.bis.impl;
+
+import com.diaoyun.zion.chinafrica.bis.IItemSpider;
+import com.diaoyun.zion.chinafrica.enums.PlatformEnum;
+import com.diaoyun.zion.master.util.HttpClientUtil;
+import net.sf.json.JSONArray;
+import net.sf.json.JSONObject;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+import org.springframework.stereotype.Component;
+
+import java.io.IOException;
+import java.net.URISyntaxException;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.concurrent.ExecutionException;
+import java.util.concurrent.TimeoutException;
+
+/**
+ * Zara西班牙时尚品牌数据爬虫
+ * TODO 数据未处理完全
+ */
+@Component("zaraSpider")
+public class ZaraSpider implements IItemSpider {
+    private static Logger logger = LoggerFactory.getLogger(ZaraSpider.class);
+
+    //西班牙时尚品牌数据爬虫
+    private static final String zaraUrl = "https://www.zara.cn/cn/zh/";
+
+    @Override
+    public JSONObject captureItem(String targetUrl) throws URISyntaxException, IOException, ExecutionException, InterruptedException, TimeoutException {
+        // 获取url中的网页内容
+        String content = HttpClientUtil.getContentByUrl(targetUrl, PlatformEnum.ZARA.getValue());
+        // 截取主要的 Json 内容
+        String jsonDataStr = getDataJson(content, "dataLayer", ";window.zara.viewPayload")
+                .replace("dataLayer = ", "");
+        // 转换为 Json 格式
+        JSONObject jsonObject = JSONObject.fromObject(jsonDataStr);
+        return jsonObject;
+    }
+
+    /**
+     * 根据首位字符串内容进行截取
+     * @param jsonStr
+     * @param startStr 起始字符串
+     * @param lastStr 结尾字符串(不包含)
+     * @return
+     */
+    private static String getDataJson(String jsonStr, String startStr, String lastStr) {
+        int startIndex = jsonStr.indexOf(startStr);
+        int lastIndex = jsonStr.lastIndexOf(lastStr);
+        return jsonStr.substring(startIndex, lastIndex);
+    }
+
+    public static void main(String[] args) throws IOException, URISyntaxException {
+        // URL链接
+        String targetUrl = "https://www.zara.cn/cn/zh/%E5%BA%9C%E7%BB%B8%E9%95%BF%E7%89%88%E8%A1%AC%E8%A1%AB-p08053157.html?v1=31979171&v2=1319321";
+        // 获取网页内容
+        String content = HttpClientUtil.getContentByUrl(targetUrl, PlatformEnum.ZARA.getValue());
+        // 截取主要的商品数据
+        int labelHeadIndex = content.indexOf("dataLayer");
+        int labelTailIndex = content.lastIndexOf(";window.zara.viewPayload");
+        String abv = content.substring(labelHeadIndex, labelTailIndex).replace("dataLayer = ", "");
+        System.err.println(abv);
+
+        // 转换为 Json 格式
+        JSONObject json = JSONObject.fromObject(abv);
+//        System.err.println(json);
+
+        // product 对象节点
+        JSONObject responseData = json.getJSONObject("product");
+//        System.err.println(responseData);
+
+        // detail 对象节点
+        JSONObject details = responseData.getJSONObject("detail");
+//        System.err.println(details);
+
+        // colors 数组节点
+        JSONArray colorsArray = details.getJSONArray("colors");
+        Map<Integer, JSONObject> colorMap = new HashMap<>();
+        for (int i = 0; i < colorsArray.size(); i++) {
+            colorMap.put(i, colorsArray.getJSONObject(i));
+        }
+        System.out.println(colorMap);
+
+        // TODO 取出颜色属性
+
+
+        // sizes 数组节点
+        Map<Integer, JSONObject> sizesMap = new HashMap<>();
+        for(Map.Entry<Integer, JSONObject> entry : colorMap.entrySet()){
+            // 遍历出每个 colors 对象节点
+            JSONObject colors = entry.getValue();
+            JSONArray sizesArray = colors.getJSONArray("sizes");
+            for (int i = 0; i < sizesArray.size(); i++) {
+                sizesMap.put(i, sizesArray.getJSONObject(i));
+            }
+        }
+        System.out.println(sizesMap);
+
+        // TODO 取出价格和尺码属性
+
+
+        // TODO 取出图片属性(图片实体类未知)
+    }
+}

src/main/java/com/diaoyun/zion/chinafrica/enums/PlatformEnum.java

@@ -12,7 +12,10 @@ public enum PlatformEnum implements EnumItemable<PlatformEnum> {
 
 	TB("淘宝", "tb"),
 	TM("天猫", "tm"),
+	PULLANDBEAR("Pullandbear","pullandbear"),
 	GAP("GAP", "gap"),
+	ZARA("Zara", "zara"),
+	UNIQLO("优衣库", "uniqlo"),
 	NIKE("NIKE", "nike"),
 	UN("未知", "un");
 

src/main/java/com/diaoyun/zion/chinafrica/factory/ItemSpiderFactory.java

@@ -20,10 +20,22 @@ public class ItemSpiderFactory {
                 iItemSpider= (IItemSpider) SpringContextUtil.getBean("tmItemSpider");
                 break;
             }
+            case "pullandbear":{
+                iItemSpider= (IItemSpider) SpringContextUtil.getBean("pullandbearSpider");
+                break;
+            }
             case "gap":{
                 iItemSpider= (IItemSpider) SpringContextUtil.getBean("gapItemSpider");
                 break;
             }
+            case "zara":{
+                iItemSpider= (IItemSpider) SpringContextUtil.getBean("zaraSpider");
+                break;
+            }
+            case "uniqlo":{
+                iItemSpider= (IItemSpider) SpringContextUtil.getBean("uniqloSpider");
+                break;
+            }
             case "nike":{
                 iItemSpider= (IItemSpider) SpringContextUtil.getBean("nikeItemSpider");
                 break;

src/main/java/com/diaoyun/zion/chinafrica/service/impl/SpiderServiceImpl.java

@@ -43,14 +43,20 @@ public class SpiderServiceImpl implements SpiderService {
 
     private PlatformEnum judgeUrlType(String targetUrl) {
         PlatformEnum platformEnum = PlatformEnum.UN;
-        if(targetUrl.contains("taobao.com")&&(targetUrl.contains("item.htm")||targetUrl.contains("detail.htm"))) {
-            platformEnum=PlatformEnum.TB;
-        } else if(targetUrl.contains("tmall.com/item.htm")) {
+        if (targetUrl.contains("taobao.com")&&(targetUrl.contains("item.htm")||targetUrl.contains("detail.htm"))) {
+            platformEnum = PlatformEnum.TB;
+        } else if (targetUrl.contains("tmall.com/item.htm")) {
             platformEnum=PlatformEnum.TM;
-        } else if(targetUrl.contains("www.gap.cn/pdp/")) {
-            platformEnum=PlatformEnum.GAP;
-        } else if(targetUrl.contains("www.nike.com/cn/t/")) {
-            platformEnum=PlatformEnum.NIKE;
+        } else if (targetUrl.contains("https://www.pullandbear.cn")) {
+            platformEnum = PlatformEnum.PULLANDBEAR;
+        } else if (targetUrl.contains("www.gap.cn/pdp/")) {
+            platformEnum = PlatformEnum.GAP;
+        } else if (targetUrl.contains("zara.cn")) {
+            platformEnum = platformEnum.ZARA;
+        } else if (targetUrl.contains("uniqlo.cn/product-detail.html")) {
+            platformEnum = platformEnum.UNIQLO;
+        } else if (targetUrl.contains("www.nike.com/cn/t/")) {
+            platformEnum = platformEnum.NIKE;
         }
         return platformEnum;
     }