newselectors

webmagic-core/pom.xml

@@ -25,6 +25,12 @@
             <artifactId>commons-lang3</artifactId>
         </dependency>
 
+        <dependency>
+            <groupId>us.codecraft</groupId>
+            <artifactId>xsoup</artifactId>
+            <version>0.0.1-SNAPSHOT</version>
+        </dependency>
+
         <dependency>
             <groupId>log4j</groupId>
             <artifactId>log4j</artifactId>

webmagic-core/src/main/java/us/codecraft/webmagic/selector/BaseElementSelector.java

@@ -0,0 +1,23 @@
+package us.codecraft.webmagic.selector;
+
+import org.jsoup.Jsoup;
+
+import java.util.List;
+
+/**
+ * @author code4crafter@gmail.com
+ * @since 0.2.2
+ */
+public abstract class BaseElementSelector implements Selector,ElementSelector {
+
+    @Override
+    public String select(String text) {
+        return select(Jsoup.parse(text));
+    }
+
+    @Override
+    public List<String> selectList(String text) {
+        return selectList(Jsoup.parse(text));
+    }
+
+}

webmagic-core/src/main/java/us/codecraft/webmagic/selector/CssSelector.java

@@ -1,8 +1,6 @@
 package us.codecraft.webmagic.selector;
 
 import org.apache.commons.collections.CollectionUtils;
-import org.jsoup.Jsoup;
-import org.jsoup.nodes.Document;
 import org.jsoup.nodes.Element;
 import org.jsoup.select.Elements;
 
@@ -15,7 +13,7 @@ import java.util.List;
  * @author code4crafter@gmail.com <br>
  * @since 0.1.0
  */
-public class CssSelector implements Selector {
+public class CssSelector extends BaseElementSelector {
 
     private String selectorText;
 
@@ -30,16 +28,6 @@ public class CssSelector implements Selector {
         this.attrName = attrName;
     }
 
-    @Override
-    public String select(String text) {
-        Document doc = Jsoup.parse(text);
-        Elements elements = doc.select(selectorText);
-        if (CollectionUtils.isEmpty(elements)) {
-            return null;
-        }
-        return getValue(elements.get(0));
-    }
-
     private String getValue(Element element) {
         if (attrName == null) {
             return element.outerHtml();
@@ -51,9 +39,17 @@ public class CssSelector implements Selector {
     }
 
     @Override
-    public List<String> selectList(String text) {
+    public String select(Element element) {
+        Elements elements = element.select(selectorText);
+        if (CollectionUtils.isEmpty(elements)) {
+            return null;
+        }
+        return getValue(elements.get(0));
+    }
+
+    @Override
+    public List<String> selectList(Element doc) {
         List<String> strings = new ArrayList<String>();
-        Document doc = Jsoup.parse(text);
         Elements elements = doc.select(selectorText);
         if (CollectionUtils.isNotEmpty(elements)) {
             for (Element element : elements) {

webmagic-core/src/main/java/us/codecraft/webmagic/selector/ElementSelector.java

@@ -0,0 +1,32 @@
+package us.codecraft.webmagic.selector;
+
+import org.jsoup.nodes.Element;
+
+import java.util.List;
+
+/**
+ * Selector(extractor) for html elements.<br>
+ *
+ * @author code4crafter@gmail.com <br>
+ * @since 0.2.2
+ */
+public interface ElementSelector {
+
+    /**
+     * Extract single result in text.<br>
+     * If there are more than one result, only the first will be chosen.
+     *
+     * @param element
+     * @return result
+     */
+    public String select(Element element);
+
+    /**
+     * Extract all results in text.<br>
+     *
+     * @param element
+     * @return results
+     */
+    public List<String> selectList(Element element);
+
+}

webmagic-core/src/main/java/us/codecraft/webmagic/selector/XsoupSelector.java

@@ -0,0 +1,32 @@
+package us.codecraft.webmagic.selector;
+
+import org.jsoup.nodes.Element;
+import us.codecraft.xsoup.XPathEvaluator;
+import us.codecraft.xsoup.Xsoup;
+
+import java.util.List;
+
+/**
+ * XPath selector based on Xsoup.<br>
+ *
+ * @author code4crafter@gmail.com <br>
+ * @since 0.2.2
+ */
+public class XsoupSelector extends BaseElementSelector {
+
+    private XPathEvaluator xPathEvaluator;
+
+    public XsoupSelector(String xpathStr) {
+        this.xPathEvaluator = Xsoup.compile(xpathStr);
+    }
+
+    @Override
+    public String select(Element element) {
+        return xPathEvaluator.evaluate(element).get();
+    }
+
+    @Override
+    public List<String> selectList(Element element) {
+        return xPathEvaluator.evaluate(element).list();
+    }
+}