增加了新的爬虫，重新排布了爬虫模块的代码

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

@@ -19,6 +19,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 import static com.diaoyun.zion.master.util.SpiderUtil.exchangeRate;
 
@@ -45,6 +46,17 @@ public class AberCrombieFitchSpider implements IItemSpider {
         return resultJson;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      * @param content 主要的页面数据

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

@@ -17,6 +17,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 import static com.diaoyun.zion.master.util.SpiderUtil.exchangeRate;
 
@@ -49,6 +50,17 @@ public class AdidasSpider implements IItemSpider {
         return resultJson;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+    
     /**
      * 格式化返回数据
      * @param content 主要的页面数据

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

@@ -17,6 +17,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 import static com.diaoyun.zion.master.util.SpiderUtil.exchangeRate;
 
@@ -52,6 +53,17 @@ public class CoachSpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      * @param dataMap 主要的Json数据

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

@@ -17,6 +17,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 import static com.diaoyun.zion.master.util.SpiderUtil.exchangeRate;
 
@@ -45,6 +46,17 @@ public class EspritSpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      *

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

@@ -19,6 +19,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 import static com.diaoyun.zion.master.util.SpiderUtil.exchangeRate;
 
@@ -48,6 +49,17 @@ public class GucciSpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      * @param content 主要的网页内容

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

@@ -23,6 +23,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 import static com.diaoyun.zion.master.util.SpiderUtil.exchangeRate;
 
@@ -51,6 +52,17 @@ public class HmSpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      *

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

@@ -17,6 +17,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 import static com.diaoyun.zion.master.util.SpiderUtil.exchangeRate;
 
@@ -49,6 +50,17 @@ public class LeviSpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      * @param dataMap 主要的Json数据

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

@@ -20,6 +20,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 /**
  * MajeSpider 数据爬虫
@@ -44,6 +45,17 @@ public class MajeSpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      * @param content 主要的页面数据

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

@@ -18,6 +18,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 /**
  * Massimo Dutti 数据爬虫
@@ -49,6 +50,17 @@ public class MassimoduttiSpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      * @param dataMap 主要的 json 数据

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

@@ -16,6 +16,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 import static com.diaoyun.zion.master.util.SpiderUtil.exchangeRate;
 
@@ -51,6 +52,17 @@ public class MocoSpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      *

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

@@ -20,6 +20,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 /**
  * OchirlySpider 数据爬虫
@@ -45,6 +46,17 @@ public class OchirlySpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      * @param content 主要的页面数据

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

@@ -18,6 +18,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 /**
  * Oysho 数据爬虫
@@ -52,6 +53,17 @@ public class OyshoSpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      *

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

@@ -18,6 +18,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 /**
  * 西班牙年轻时尚品牌-PullAndBear 数据爬虫
@@ -35,16 +36,31 @@ public class PullandbearSpider implements IItemSpider {
      */
     @Override
     public JSONObject captureItem(String targetUrl) throws URISyntaxException, IOException, ExecutionException, InterruptedException, TimeoutException {
+        JSONObject resultJson = new JSONObject();
+        if (!urlPattern(targetUrl)) {
+            resultJson.put("message", "不是商品的详情页路径");
+            return resultJson;
+        }
         String pId = targetUrl.substring(targetUrl.lastIndexOf("p")+1, targetUrl.lastIndexOf(".html"));
         targetUrl = "https://www.pullandbear.cn/itxrest/2/catalog/store/24009528/20309423/category/0/product/" + pId + "/detail?languageId=-7&appId=1";
         String content = HttpClientUtil.getContentByUrl(targetUrl, PlatformEnum.PULLANDBEAR.getValue());
-        JSONObject resultJson = JSONObject.fromObject(content);
+        resultJson = JSONObject.fromObject(content);
         ProductResponse productResponse = formatProductResponse(resultJson, pId);
         resultJson = JSONObject.fromObject(productResponse);
         TranslateHelper.translateProductResponse(resultJson);
         return resultJson;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "^.*pullandbear\\.cn/.*-c\\d{5,10}p\\d{9,10}.html\\?cS=\\d*";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化 PullAndBear 返回数据
      * @see com.diaoyun.zion.chinafrica.bis.impl.PullandbearSpider

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

@@ -20,6 +20,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 /**
  * Revolve  数据爬虫
@@ -44,6 +45,17 @@ public class RevolveSpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      * @param content 主要的网页内容

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

@@ -17,6 +17,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 /**
  * 优衣库数据爬虫
@@ -56,6 +57,17 @@ public class UniqloSpider implements IItemSpider {
         return resultJson;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 返回格式化数据
      * @param dataMap 调用优衣库网页接口接收的主要商品数据

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

@@ -16,6 +16,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 import static com.diaoyun.zion.master.util.SpiderUtil.exchangeRate;
 
@@ -51,6 +52,17 @@ public class UrbanRevivoSpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化返回数据
      * @param dataMap 主要的 json 数据

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

@@ -17,6 +17,7 @@ import java.net.URISyntaxException;
 import java.util.*;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Pattern;
 
 import static com.diaoyun.zion.master.util.SpiderUtil.exchangeRate;
 
@@ -47,6 +48,17 @@ public class ZaraSpider implements IItemSpider {
         return resultObj;
     }
 
+    /**
+     * 正则匹配是否为商品详情页的链接
+     * TODO 正则编写
+     * @param targetUrl url路径
+     * @return 匹配失败则返回错误信息
+     */
+    private boolean urlPattern(String targetUrl) {
+        String regex = "";
+        return Pattern.matches(regex, targetUrl);
+    }
+
     /**
      * 格式化 Zara 返回数据
      * @param dataMap 主要的 json 数据

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

@@ -76,31 +76,31 @@ public class SpiderServiceImpl implements SpiderService {
             platformEnum = PlatformEnum.HM;
         } else if(targetUrl.contains("adidas.com") && targetUrl.contains("item")) {
             platformEnum=PlatformEnum.ADIDAS;
-        } else if(targetUrl.contains("http://www.lily.sh.cn/webapp/wcs/stores/servlet/lilystore")) {
+        } else if(targetUrl.contains("lily")) {
             platformEnum=PlatformEnum.LILY;
         } else if(targetUrl.contains("eifini")) {
             platformEnum=PlatformEnum.EIFINI;
-        } else if(targetUrl.contains("wap.ur") && targetUrl.contains("product")) {
+        } else if(targetUrl.contains("wap")) {
             platformEnum=PlatformEnum.URBANREVIVO;
         } else if(targetUrl.contains("abercrombie")) {
             platformEnum=PlatformEnum.ABERCROMBIEFITCH;
-        } else if(targetUrl.contains("ochirly.com") && targetUrl.contains("p/mobile/")) {
+        } else if(targetUrl.contains("ochirly.com")) {
             platformEnum=PlatformEnum.OCHIRLY;
-        } else if(targetUrl.contains("esprit") && targetUrl.contains("product") && targetUrl.contains("styleNo") && targetUrl.contains("skucode")) {
+        } else if(targetUrl.contains("esprit")) {
             platformEnum=PlatformEnum.ESPRIT;
-        } else if(targetUrl.contains("levi.com") && targetUrl.contains("product") && targetUrl.contains("styleNo")) {
+        } else if(targetUrl.contains("levi.com")) {
             platformEnum=PlatformEnum.LEVI;
         } else if(targetUrl.contains("moco.com/moco/")) {
             platformEnum=PlatformEnum.MOCO;
-        } else if (targetUrl.contains("massimodutti") && targetUrl.contains("colorId") && targetUrl.contains("categoryId")) {
+        } else if (targetUrl.contains("massimodutti")) {
             platformEnum = PlatformEnum.MASSIMODUTTI;
         } else if (targetUrl.contains("coach")) {
             platformEnum = PlatformEnum.COACH;
         } else if (targetUrl.contains("revolve")) {
             platformEnum = PlatformEnum.REVOLVE;
-        } else if (targetUrl.contains("vans.com") && targetUrl.contains("wap/product")) {
+        } else if (targetUrl.contains("vans.com")) {
             platformEnum = PlatformEnum.VANS;
-        } else if (targetUrl.contains("oysho") && (targetUrl.contains("origenId") || targetUrl.contains("colorId")) ) {
+        } else if (targetUrl.contains("oysho")) {
             platformEnum = PlatformEnum.OYSHO;
         } else if (targetUrl.contains("stradivarius")) {
             platformEnum = PlatformEnum.STRADIVARIUS;
@@ -110,11 +110,11 @@ public class SpiderServiceImpl implements SpiderService {
             platformEnum = PlatformEnum.GUCCI;
         } else if (targetUrl.contains("burberry.com")) {
             platformEnum = PlatformEnum.BURBERRY;
-        } else if (targetUrl.contains("prada.com") && targetUrl.contains("products")) {
+        } else if (targetUrl.contains("prada.com")) {
             platformEnum = PlatformEnum.PRADA;
         } else if (targetUrl.contains("fendi")) {
             platformEnum = PlatformEnum.FENDI;
-        } else if (targetUrl.contains("apple") && targetUrl.contains("buy")) {
+        } else if (targetUrl.contains("apple")) {
             platformEnum = PlatformEnum.APPLE;
         } else if (targetUrl.contains("louisvuitton")) {
             platformEnum = PlatformEnum.LOUISVUITTON;