Merge branch 'release/0.9.0'

code4craft · Jun 22, 2023 · fd4a136 · fd4a136
2 parents 9e59b37 + 3688226
commit fd4a136
Show file tree

Hide file tree

Showing 25 changed files with 418 additions and 187 deletions.
diff --git a/README-zh.md b/README-zh.md
@@ -39,12 +39,12 @@ webmagic使用maven管理依赖，在项目中添加对应的依赖即可使用w
 <dependency>
     <groupId>us.codecraft</groupId>
     <artifactId>webmagic-core</artifactId>
-    <version>0.7.5</version>
+    <version>${webmagic.version}</version>
 </dependency>
 <dependency>
     <groupId>us.codecraft</groupId>
     <artifactId>webmagic-extension</artifactId>
-    <version>0.7.5</version>
+    <version>${webmagic.version}</version>
 </dependency>
 ```
         

diff --git a/README.md b/README.md
@@ -25,12 +25,12 @@ Add dependencies to your pom.xml:
 <dependency>
     <groupId>us.codecraft</groupId>
     <artifactId>webmagic-core</artifactId>
-    <version>0.7.5</version>
+    <version>${webmagic.version}</version>
 </dependency>
 <dependency>
     <groupId>us.codecraft</groupId>
     <artifactId>webmagic-extension</artifactId>
-    <version>0.7.5</version>
+    <version>${webmagic.version}</version>
 </dependency>
 ```
         

diff --git a/pom.xml b/pom.xml
@@ -1,7 +1,7 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
     <groupId>us.codecraft</groupId>
-    <version>0.8.0</version>
+    <version>0.9.0</version>
     <modelVersion>4.0.0</modelVersion>
     <packaging>pom</packaging>
     <properties>
@@ -124,7 +124,7 @@
             <dependency>
                 <groupId>us.codecraft</groupId>
                 <artifactId>xsoup</artifactId>
-                <version>0.3.6</version>
+                <version>0.3.7</version>
             </dependency>
             <dependency>
                 <groupId>com.alibaba</groupId>

diff --git a/webmagic-core/pom.xml b/webmagic-core/pom.xml
@@ -3,7 +3,7 @@
     <parent>
         <groupId>us.codecraft</groupId>
         <artifactId>webmagic-parent</artifactId>
-        <version>0.8.0</version>
+        <version>0.9.0</version>
     </parent>
     <modelVersion>4.0.0</modelVersion>
 

diff --git a/webmagic-core/src/main/java/us/codecraft/webmagic/Page.java b/webmagic-core/src/main/java/us/codecraft/webmagic/Page.java
@@ -20,7 +20,7 @@
  * {@link #getHtml()}  get content of current page                 <br>
  * {@link #putField(String, Object)}  save extracted result            <br>
  * {@link #getResultItems()} get extract results to be used in {@link us.codecraft.webmagic.pipeline.Pipeline}<br>
- * {@link #addTargetRequests(java.util.List)} {@link #addTargetRequest(String)} add urls to fetch                 <br>
+ * {@link #addTargetRequests(Iterable)} {@link #addTargetRequest(String)} add urls to fetch                 <br>
  *
  * @author [email protected] <br>
  * @see us.codecraft.webmagic.downloader.Downloader
@@ -52,7 +52,7 @@ public class Page {
     private List<Request> targetRequests = new ArrayList<Request>();
 
     private String charset;
-    
+
     public Page() {
     }
 
@@ -108,7 +108,8 @@ public Json getJson() {
      * @deprecated since 0.4.0
      * The html is parse just when first time of calling {@link #getHtml()}, so use {@link #setRawText(String)} instead.
      */
-    public void setHtml(Html html) {
+    @Deprecated
+	public void setHtml(Html html) {
         this.html = html;
     }
 
@@ -121,7 +122,7 @@ public List<Request> getTargetRequests() {
      *
      * @param requests requests
      */
-    public void addTargetRequests(List<String> requests) {
+    public void addTargetRequests(Iterable<String> requests) {
         for (String s : requests) {
             if (StringUtils.isBlank(s) || s.equals("#") || s.startsWith("javascript:")) {
                 continue;
@@ -137,7 +138,7 @@ public void addTargetRequests(List<String> requests) {
      * @param requests requests
      * @param priority priority
      */
-    public void addTargetRequests(List<String> requests, long priority) {
+    public void addTargetRequests(Iterable<String> requests, long priority) {
         for (String s : requests) {
             if (StringUtils.isBlank(s) || s.equals("#") || s.startsWith("javascript:")) {
                 continue;

diff --git a/webmagic-core/src/main/java/us/codecraft/webmagic/Site.java b/webmagic-core/src/main/java/us/codecraft/webmagic/Site.java
@@ -28,6 +28,8 @@ public class Site {
 
     private String charset;
 
+    private String defaultCharset;
+
     private int sleepTime = 5000;
 
     private int retryTimes = 0;
@@ -168,6 +170,30 @@ public String getCharset() {
         return charset;
     }
 
+    /**
+     * Set default charset of page.
+     *
+     * When charset detect failed, use this default charset.
+     *
+     * @param defaultCharset the default charset
+     * @return this
+     * @since 0.9.0
+     */
+    public Site setDefaultCharset(String defaultCharset) {
+        this.defaultCharset = defaultCharset;
+        return this;
+    }
+
+    /**
+     * The default charset if charset detected failed.
+     *
+     * @return the defulat charset
+     * @since 0.9.0
+     */
+    public String getDefaultCharset() {
+        return defaultCharset;
+    }
+
     public int getTimeOut() {
         return timeOut;
     }

diff --git a/webmagic-core/src/main/java/us/codecraft/webmagic/downloader/HttpClientDownloader.java b/webmagic-core/src/main/java/us/codecraft/webmagic/downloader/HttpClientDownloader.java
@@ -4,6 +4,7 @@
 import java.nio.charset.Charset;
 import java.util.HashMap;
 import java.util.Map;
+import java.util.Optional;
 
 import org.apache.commons.io.IOUtils;
 import org.apache.http.HttpResponse;
@@ -76,7 +77,7 @@ public Page download(Request request, Task task) {
         }
         CloseableHttpResponse httpResponse = null;
         CloseableHttpClient httpClient = getHttpClient(task.getSite());
-        Proxy proxy = proxyProvider != null ? proxyProvider.getProxy(task) : null;
+        Proxy proxy = proxyProvider != null ? proxyProvider.getProxy(request, task) : null;
         HttpClientRequestContext requestContext = httpUriRequestConverter.convert(request, task.getSite(), proxy);
         Page page = Page.fail();
         try {
@@ -116,7 +117,7 @@ protected Page handleResponse(Request request, String charset, HttpResponse http
         page.setBytes(bytes);
         if (!request.isBinaryContent()) {
             if (charset == null) {
-                charset = getHtmlCharset(contentType, bytes);
+                charset = getHtmlCharset(contentType, bytes, task);
             }
             page.setCharset(charset);
             page.setRawText(new String(bytes, charset));
@@ -131,11 +132,11 @@ protected Page handleResponse(Request request, String charset, HttpResponse http
         return page;
     }
 
-    private String getHtmlCharset(String contentType, byte[] contentBytes) throws IOException {
+    private String getHtmlCharset(String contentType, byte[] contentBytes, Task task) throws IOException {
         String charset = CharsetUtils.detectCharset(contentType, contentBytes);
         if (charset == null) {
-            charset = Charset.defaultCharset().name();
-            logger.warn("Charset autodetect failed, use {} as charset. Please specify charset in Site.setCharset()", Charset.defaultCharset());
+            charset = Optional.ofNullable(task.getSite().getDefaultCharset()).orElseGet(Charset.defaultCharset()::name);
+            logger.info("Charset autodetect failed, use {} as charset.", task.getSite().getDefaultCharset());
         }
         return charset;
     }

diff --git a/webmagic-core/src/main/java/us/codecraft/webmagic/downloader/HttpClientGenerator.java b/webmagic-core/src/main/java/us/codecraft/webmagic/downloader/HttpClientGenerator.java
@@ -1,16 +1,5 @@
 package us.codecraft.webmagic.downloader;
 
-import java.io.IOException;
-import java.security.KeyManagementException;
-import java.security.NoSuchAlgorithmException;
-import java.security.cert.CertificateException;
-import java.security.cert.X509Certificate;
-import java.util.Map;
-
-import javax.net.ssl.SSLContext;
-import javax.net.ssl.TrustManager;
-import javax.net.ssl.X509TrustManager;
-
 import org.apache.commons.lang3.JavaVersion;
 import org.apache.commons.lang3.SystemUtils;
 import org.apache.http.HttpException;
@@ -22,28 +11,32 @@
 import org.apache.http.config.SocketConfig;
 import org.apache.http.conn.socket.ConnectionSocketFactory;
 import org.apache.http.conn.socket.PlainConnectionSocketFactory;
-import org.apache.http.conn.ssl.DefaultHostnameVerifier;
 import org.apache.http.conn.ssl.SSLConnectionSocketFactory;
-import org.apache.http.impl.client.BasicCookieStore;
-import org.apache.http.impl.client.CloseableHttpClient;
-import org.apache.http.impl.client.DefaultHttpRequestRetryHandler;
-import org.apache.http.impl.client.HttpClientBuilder;
-import org.apache.http.impl.client.HttpClients;
+import org.apache.http.impl.client.*;
 import org.apache.http.impl.conn.PoolingHttpClientConnectionManager;
 import org.apache.http.impl.cookie.BasicClientCookie;
 import org.apache.http.protocol.HttpContext;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
-
 import us.codecraft.webmagic.Site;
 
+import javax.net.ssl.SSLContext;
+import javax.net.ssl.TrustManager;
+import javax.net.ssl.X509TrustManager;
+import java.io.IOException;
+import java.security.KeyManagementException;
+import java.security.NoSuchAlgorithmException;
+import java.security.cert.CertificateException;
+import java.security.cert.X509Certificate;
+import java.util.Map;
+
 /**
  * @author [email protected] <br>
  * @since 0.4.0
  */
 public class HttpClientGenerator {
 
-	private transient Logger logger = LoggerFactory.getLogger(getClass());
+    private transient Logger logger = LoggerFactory.getLogger(getClass());
 
     private PoolingHttpClientConnectionManager connectionManager;
 
@@ -61,21 +54,20 @@ private SSLConnectionSocketFactory buildSSLConnectionSocketFactory() {
             SSLContext sslContext = createIgnoreVerifySSL();
             String[] supportedProtocols;
             if (SystemUtils.isJavaVersionAtLeast(JavaVersion.JAVA_11)) {
-                supportedProtocols = new String[] { "SSLv3", "TLSv1", "TLSv1.1", "TLSv1.2", "TLSv1.3" };
+                supportedProtocols = new String[]{"SSLv3", "TLSv1", "TLSv1.1", "TLSv1.2", "TLSv1.3"};
             } else {
-                supportedProtocols = new String[] { "SSLv3", "TLSv1", "TLSv1.1", "TLSv1.2" };
+                supportedProtocols = new String[]{"SSLv3", "TLSv1", "TLSv1.1", "TLSv1.2"};
             }
             logger.debug("supportedProtocols: {}", String.join(", ", supportedProtocols));
             return new SSLConnectionSocketFactory(sslContext, supportedProtocols,
                     null,
-                    new DefaultHostnameVerifier()); // 优先绕过安全证书
-        } catch (KeyManagementException e) {
-            logger.error("ssl connection fail", e);
-        } catch (NoSuchAlgorithmException e) {
+                    //不进行主机校验
+                    (host, sslSession) -> true); // 优先绕过安全证书
+        } catch (KeyManagementException | NoSuchAlgorithmException e) {
             logger.error("ssl connection fail", e);
         }
         return SSLConnectionSocketFactory.getSocketFactory();
-	}
+    }
 
     private SSLContext createIgnoreVerifySSL() throws NoSuchAlgorithmException, KeyManagementException {
         // 实现一个X509TrustManager接口，用于绕过验证，不用修改里面的方法
@@ -97,9 +89,9 @@ public X509Certificate[] getAcceptedIssuers() {
         };
 
         SSLContext sc = SSLContext.getInstance("TLS");
-        sc.init(null, new TrustManager[] { trustManager }, null);
+        sc.init(null, new TrustManager[]{trustManager}, null);
         return sc;
-	}
+    }
 
     public HttpClientGenerator setPoolSize(int poolSize) {
         connectionManager.setMaxTotal(poolSize);

diff --git a/webmagic-core/src/main/java/us/codecraft/webmagic/proxy/ProxyProvider.java b/webmagic-core/src/main/java/us/codecraft/webmagic/proxy/ProxyProvider.java
@@ -1,6 +1,7 @@
 package us.codecraft.webmagic.proxy;
 
 import us.codecraft.webmagic.Page;
+import us.codecraft.webmagic.Request;
 import us.codecraft.webmagic.Task;
 
 /**
@@ -23,7 +24,23 @@ public interface ProxyProvider {
      * Get a proxy for task by some strategy.
      * @param task the download task
      * @return proxy 
+     * @deprecated Use {@link #getProxy(Request, Task)} instead.
      */
-    Proxy getProxy(Task task);
+    @Deprecated
+    default Proxy getProxy(Task task) {
+        throw new UnsupportedOperationException();
+    }
+
+    /**
+     * Returns a proxy for the request.
+     *
+     * @param request the request
+     * @param task the download task
+     * @return proxy
+     * @since 0.9.0
+     */
+    default Proxy getProxy(Request request, Task task) {
+        return this.getProxy(task);
+    }
 
 }
diff --git a/webmagic-core/src/main/java/us/codecraft/webmagic/proxy/SimpleProxyProvider.java b/webmagic-core/src/main/java/us/codecraft/webmagic/proxy/SimpleProxyProvider.java
@@ -1,6 +1,7 @@
 package us.codecraft.webmagic.proxy;
 
 import us.codecraft.webmagic.Page;
+import us.codecraft.webmagic.Request;
 import us.codecraft.webmagic.Task;
 
 import java.util.ArrayList;
@@ -44,7 +45,7 @@ public void returnProxy(Proxy proxy, Page page, Task task) {
     }
 
     @Override
-    public Proxy getProxy(Task task) {
+    public Proxy getProxy(Request request, Task task) {
         return proxies.get(incrForLoop());
     }
 

diff --git a/webmagic-core/src/main/java/us/codecraft/webmagic/selector/BaseElementSelector.java b/webmagic-core/src/main/java/us/codecraft/webmagic/selector/BaseElementSelector.java
@@ -3,6 +3,7 @@
 import org.jsoup.Jsoup;
 import org.jsoup.nodes.Document;
 import org.jsoup.nodes.Element;
+import us.codecraft.webmagic.utils.BaseSelectorUtils;
 
 import java.util.ArrayList;
 import java.util.List;
@@ -13,16 +14,9 @@
  */
 public abstract class BaseElementSelector implements Selector, ElementSelector {
     private Document parse(String text) {
-        if (text == null) {
-            return null;
-        }
-
         // Jsoup could not parse <tr></tr> or <td></td> tag directly
         // https://stackoverflow.com/questions/63607740/jsoup-couldnt-parse-tr-tag
-        if ((text.startsWith("<tr>") && text.endsWith("</tr>"))
-                || (text.startsWith("<td>") && text.endsWith("</td>"))) {
-            text = "<table>" + text + "</table>";
-        }
+        text = BaseSelectorUtils.preParse(text);
         return Jsoup.parse(text);
     }
 

diff --git a/webmagic-core/src/main/java/us/codecraft/webmagic/utils/BaseSelectorUtils.java b/webmagic-core/src/main/java/us/codecraft/webmagic/utils/BaseSelectorUtils.java
@@ -0,0 +1,23 @@
+package us.codecraft.webmagic.utils;
+
+/**
+ * @author hooy
+ */
+public class BaseSelectorUtils {
+
+    /**
+     * Jsoup/HtmlCleaner could not parse "tr" or "td" tag directly
+     * https://stackoverflow.com/questions/63607740/jsoup-couldnt-parse-tr-tag
+     *
+     * @param text - the html string
+     * @return text
+     */
+    public static String preParse(String text) {
+        if (((text.startsWith("<tr>") || text.startsWith("<tr ")) && text.endsWith("</tr>"))
+                || ((text.startsWith("<td>") || text.startsWith("<td ")) && text.endsWith("</td>"))) {
+            text = "<table>" + text + "</table>";
+        }
+        return text;
+    }
+
+}
diff --git a/webmagic-core/src/test/java/us/codecraft/webmagic/SiteTest.java b/webmagic-core/src/test/java/us/codecraft/webmagic/SiteTest.java
@@ -0,0 +1,17 @@
+package us.codecraft.webmagic;
+
+import static org.junit.Assert.assertEquals;
+
+import java.nio.charset.StandardCharsets;
+
+import org.junit.Test;
+
+public class SiteTest {
+
+    @Test
+    public void test() {
+        Site site = Site.me().setDefaultCharset(StandardCharsets.UTF_8.name());
+        assertEquals(StandardCharsets.UTF_8.name(), site.getDefaultCharset());
+    }
+
+}