TestSelenium爬虫小说测试

dolyw · Apr 20, 2022 · 3ea26f6 · 3ea26f6
1 parent 7b10b28
commit 3ea26f6
Showing 1 changed file with 35 additions and 8 deletions.
diff --git a/TestWeb/src/test/java/com/wang/other/TestSelenium.java b/TestWeb/src/test/java/com/wang/other/TestSelenium.java
@@ -102,7 +102,7 @@ public void test02() throws Exception {
     }
 
     @Test
-    public void test03() throws Exception {
+    public void test03() throws Exception{
         String fileName = "D:\\Program Files (x86)\\Reader_v1.9.3.2\\天命大反派.txt";
         if (!FileUtil.exist(fileName)) {
             File file = new File(fileName);
@@ -134,18 +134,19 @@ public void test03() throws Exception {
         WebDriver chromeDriver = new ChromeDriver(chromeOptions);
 
         TestSelenium.text(chromeDriver, "https://www.xs123.org/xs/33/33112/21404025.html", fileName);
+        // TestSelenium.text(chromeDriver, "https://www.xs123.org/xs/33/33112/70349769.html", fileName);
         // TestSelenium.text(chromeDriver, "https://www.xs123.org/xs/33/33112/70963309.html", fileName);
 
         chromeDriver.close();
     }
 
-    public static void text(WebDriver chromeDriver, String url, String fileName) {
+    public static void text(WebDriver chromeDriver, String url, String fileName) throws Exception {
         chromeDriver.get(url);
+        Thread.sleep(100);
         WebElement boxWebElement = chromeDriver.findElement(By.className("box_con"));
         WebElement titleElement = boxWebElement.findElement(By.tagName("h1"));
 
-        // System.out.println(titleElement.getText());
-        if (titleElement.getText().contains("950")) {
+        if (titleElement.getText().contains("1030")) {
             return;
         }
         if ("玄幻：我！天命大反派".equals(titleElement.getText())) {
@@ -154,17 +155,43 @@ public static void text(WebDriver chromeDriver, String url, String fileName) {
 
         // 标题
         List<String> lines = new ArrayList<>();
-        lines.add("第" + titleElement.getText().substring(0,4) + "章 " + titleElement.getText().substring(4));
-        lines.add("");
+        String title = "第" + titleElement.getText().substring(0,4).trim() + "章 " + titleElement.getText().substring(4).trim();
+        title.replaceAll("/?", "");
+        title.replaceAll("：", "");
+        System.out.println(title);
+        lines.add(title);
 
         // 正文
         WebElement conWebElement = chromeDriver.findElement(By.id("content"));
         String con = conWebElement.getText();
-        // con = con.replaceAll("<br/>", "/r/n");
-        lines.add(con);
+        con = con.replaceAll("<br/>", "");
+        con = con.replaceAll("\n", "");
+        con = con.replaceAll(",", "，");
+        String[] conArray = con.split("。");
+        for (String text : conArray) {
+            if (text.length() > 50) {
+                String[] conArray2 = text.split("，");
+                for (int i = 0; i < conArray2.length; i++) {
+                    if (i + 2 < conArray2.length) {
+                        lines.add(conArray2[i] + "，" + conArray2[++i] + "，" + conArray2[++i] + "，");
+                    } else if (i + 1 < conArray2.length) {
+                        lines.add(conArray2[i] + "，" + conArray2[++i] + "，");
+                    } else {
+                        lines.add(conArray2[i] + "。");
+                    }
+                    lines.add("");
+                }
+                lines.add("");
+            } else {
+                lines.add(text + "。");
+                lines.add("");
+            }
+        }
 
         FileUtil.appendUtf8Lines(lines, fileName);
 
+        Thread.sleep(10);
+
         WebElement btnWebElement = chromeDriver.findElement(By.className("bottem2"));
         List<WebElement> btnListWebElement = btnWebElement.findElements(By.tagName("a"));