This commit is contained in:
Niko 2026-05-22 09:53:07 +08:00
parent 2518b922af
commit b6f285bf68

View File

@ -29,9 +29,8 @@ public class EtsScraper {
private static final String USERNAME = "sccw";
private static final String PASSWORD = "slife@123";
private static final Path SCREENSHOT_DIR = Path.of("screenshots");
private static final String OLLAMA_URL = "http://10.0.1.39:11434";
private static final String OLLAMA_URL = "http://127.0.0.1:11434";
private static final String OLLAMA_MODEL = "qwen3-vl:4b";
private static final String PROXY_HOST = "http://127.0.0.1:8081";
public static void main(String[] args) throws Exception {
// Parse CLI arguments
@ -39,6 +38,7 @@ public class EtsScraper {
String proxyUser = null;
String proxyPass = null;
String dateStr = null;
Integer repeatInterval = null;
for (int i = 0; i < args.length; i++) {
switch (args[i]) {
@ -57,6 +57,9 @@ public class EtsScraper {
case "-d":
dateStr = args[++i];
break;
case "-r":
repeatInterval = Integer.parseInt(args[++i]);
break;
default:
System.err.println("[-] Unknown option: " + args[i]);
printHelp();
@ -77,13 +80,42 @@ public class EtsScraper {
System.err.println("[-] Invalid date format: " + dateStr + ", expected yyyy-MM-dd");
return;
}
if (repeatInterval != null) {
// Polling mode: run every N seconds
System.out.println("[*] Polling mode: running every " + repeatInterval + " seconds (target date: " + dateStr + ")");
while (true) {
try {
runScraper(proxyHost, proxyUser, proxyPass, targetDate, true);
} catch (Exception e) {
System.err.println("[-] Scraper failed: " + e.getMessage());
}
System.out.println("[*] Sleeping " + repeatInterval + "s before next run...");
sleep(repeatInterval * 1000L);
}
} else {
// Single-shot mode: run once
runScraper(proxyHost, proxyUser, proxyPass, targetDate, false);
}
}
private static void runScraper(String proxyHost, String proxyUser, String proxyPass,
java.time.LocalDate targetDate, boolean isPolling) throws Exception {
String dateStrFormatted = targetDate.format(java.time.format.DateTimeFormatter.ofPattern("yyyy-MM-dd"));
String dateStrFileName = targetDate.format(DateTimeFormatter.ofPattern("yyyyMMdd_HHmm"));
// In one-shot mode, use target date as filename (for "file exists" check)
// In polling mode, use current time as filename to always re-run
String dateStrFileName;
if (isPolling) {
dateStrFileName = java.time.LocalDateTime.now()
.format(DateTimeFormatter.ofPattern("yyyyMMdd_HHmm"));
} else {
dateStrFileName = targetDate.format(DateTimeFormatter.ofPattern("yyyyMMdd_HHmm"));
}
Path savedFile = Path.of("downloads").resolve("三联单列表_" + dateStrFileName + ".xls");
if (java.nio.file.Files.exists(savedFile) && java.nio.file.Files.size(savedFile) > 0) {
if (!isPolling && java.nio.file.Files.exists(savedFile) && java.nio.file.Files.size(savedFile) > 0) {
System.out.println("[+] File already exists: " + savedFile);
System.out.println("[+] File size: " + java.nio.file.Files.size(savedFile) + " bytes");
autoImportBill(savedFile, proxyHost, proxyUser, proxyPass);
@ -96,6 +128,8 @@ public class EtsScraper {
System.err.println("Failed to create directories: " + e.getMessage());
}
System.out.println("[*] --- Scraper run at " + LocalDateTime.now().format(DateTimeFormatter.ofPattern("yyyyMMdd_HHmmss")) + " ---");
try (Playwright playwright = Playwright.create()) {
Browser browser = playwright.chromium().launch(
new BrowserType.LaunchOptions()
@ -128,8 +162,6 @@ public class EtsScraper {
// Close notification dialog FIRST (before filling credentials)
closeNotificationDialog(page);
// screenshot(page, "after_close_dialog");
// Download captcha image
downloadCaptcha(page);
@ -143,8 +175,6 @@ public class EtsScraper {
System.out.println("[+] Login successful!");
sleep(2000);
// screenshot(page, "after_login");
System.out.println("[+] Page title: " + page.title());
System.out.println("[+] Page URL: " + page.url());
@ -152,8 +182,6 @@ public class EtsScraper {
System.out.println("[*] Clicking 三联单 menu...");
page.locator("#module_2094F683-C542-4904-B33E-0D227C4DE199").first().click();
sleep(3000);
// screenshot(page, "after_sanliandan");
System.out.println("[+] 三联单 page title: " + page.title());
// 设置日期筛选
@ -186,7 +214,6 @@ public class EtsScraper {
// 点击查询按钮等待列表加载
if (queryBtnExists) {
System.out.println("[*] Clicking query button...");
// 等待列表内容出现
page.waitForResponse("https://101.227.180.215/SHCityEnvCW/Services/CWSServ.asmx/ThreeBillQueryBiTripList", () -> {
page.locator("#Search_ThreeBillList_Button").first().click();
});
@ -195,13 +222,9 @@ public class EtsScraper {
System.out.println("[!] Query button not found");
}
// screenshot(page, "after_query");
// 点击导出按钮
if (page.locator("#Export_ThreeBillList_Button").count() > 0) {
System.out.println("[*] Clicking export button...");
// 设置下载目录
// 点击主导出按钮打开对话框再用 JS click 触发对话框内导出按钮
Download dl = page.waitForDownload(new Page.WaitForDownloadOptions().setTimeout(300000),
() -> {
page.locator("#Export_ThreeBillList_Button").first().click();
@ -220,11 +243,9 @@ public class EtsScraper {
autoImportBill(savedFile, proxyHost, proxyUser, proxyPass);
}
}
// screenshot(page, "after_export");
System.out.println("[+] Query and export completed!");
} else {
System.out.println("[-] Login failed. Check screenshots/ for debugging.");
// screenshot(page, "login_failed");
}
} finally {
browser.close();
@ -232,6 +253,7 @@ public class EtsScraper {
}
}
private static void printHelp() {
System.out.println("""
ETS 三联单爬虫 - 导出并导入三联单 Excel 数据
@ -243,10 +265,12 @@ public class EtsScraper {
-u <user> ets-proxy 用户名
-p <pass> ets-proxy 密码
-d <date> 查询日期格式 yyyy-MM-dd
-r <seconds> 定时执行间隔不传则只执行一次
-h 显示此帮助信息
示例:
java -jar ets-playwright.jar -s https://api.ets.niko.red -u admin -p 123456 -d 2026-05-04
java -jar ets-playwright.jar -s http://127.0.0.1:8081 -u sccw -p slife@123 -d 2026-05-21
java -jar ets-playwright.jar -s http://127.0.0.1:8081 -u sccw -p slife@123 -d 2026-05-21 -r 300
""");
}