103 lines
4.3 KiB
Java
103 lines
4.3 KiB
Java
import cn.book.bus.service.crawler.LocalHtmlCrawlerService;
|
|
import cn.book.bus.domain.CrawlSourceConfig;
|
|
import cn.book.bus.domain.Fiction;
|
|
import cn.book.bus.dto.crawler.ChapterData;
|
|
|
|
import java.util.List;
|
|
import java.util.Map;
|
|
|
|
/**
|
|
* 简化的本地HTML解析测试
|
|
* 不依赖Spring框架，直接测试核心功能
|
|
*/
|
|
public class SimpleLocalHtmlTest {
|
|
|
|
public static void main(String[] args) {
|
|
System.out.println("=== 简化版本地HTML解析测试 ===");
|
|
|
|
// 创建服务实例
|
|
LocalHtmlCrawlerService localHtmlCrawlerService = new LocalHtmlCrawlerService();
|
|
|
|
// 检查本地HTML文件是否存在
|
|
boolean hasFiles = localHtmlCrawlerService.hasLocalHtmlFiles();
|
|
System.out.println("本地HTML文件存在: " + hasFiles);
|
|
|
|
if (!hasFiles) {
|
|
System.out.println("❌ 未找到本地HTML文件，请确保 html/ 目录下有 czbooks_catalog.html 和 czbooks_chapter.html 文件");
|
|
return;
|
|
}
|
|
|
|
// 创建测试小说对象
|
|
Fiction fiction = new Fiction();
|
|
fiction.setId(1);
|
|
fiction.setFictionName("我的老婆是頂流天后");
|
|
|
|
// 创建测试源配置
|
|
CrawlSourceConfig source = new CrawlSourceConfig();
|
|
source.setSourceName("local_html");
|
|
source.setRequiresLocalHtml(true);
|
|
source.setSelectors(Map.of(
|
|
"title", "span.title",
|
|
"author", "span.author a",
|
|
"intro", "div.description",
|
|
"chapterList", "ul#chapter-list li a",
|
|
"chapterTitle", "div.name",
|
|
"chapterContent", "div.content"
|
|
));
|
|
|
|
System.out.println("开始解析本地HTML文件...");
|
|
|
|
// 执行解析
|
|
List<ChapterData> chapters = localHtmlCrawlerService.parseFromLocalHtml(fiction, source);
|
|
|
|
if (chapters != null && !chapters.isEmpty()) {
|
|
System.out.println("✅ 解析成功！");
|
|
System.out.println("总章节数: " + chapters.size());
|
|
|
|
// 显示前几章的信息
|
|
int displayCount = Math.min(10, chapters.size()); // 只显示前10章
|
|
for (int i = 0; i < displayCount; i++) {
|
|
ChapterData chapter = chapters.get(i);
|
|
System.out.println((i + 1) + ". " + chapter.getTitle());
|
|
|
|
if (chapter.getTextContent() != null && !chapter.getTextContent().isEmpty()) {
|
|
System.out.println(" 内容长度: " + chapter.getTextContent().length() + " 字符");
|
|
// 显示内容的前100个字符
|
|
String preview = chapter.getTextContent().substring(0,
|
|
Math.min(100, chapter.getTextContent().length()));
|
|
System.out.println(" 预览: " + preview + "...");
|
|
} else {
|
|
System.out.println(" 内容: (暂无具体章节文件)");
|
|
}
|
|
System.out.println();
|
|
}
|
|
|
|
if (chapters.size() > displayCount) {
|
|
System.out.println("... 还有 " + (chapters.size() - displayCount) + " 章未显示");
|
|
}
|
|
|
|
// 统计有内容的章节数
|
|
long chaptersWithContent = chapters.stream()
|
|
.filter(c -> c.getTextContent() != null && !c.getTextContent().isEmpty())
|
|
.count();
|
|
System.out.println("有内容的章节数: " + chaptersWithContent);
|
|
|
|
// 显示一些统计信息
|
|
System.out.println("\n=== 统计信息 ===");
|
|
System.out.println("小说名称: " + fiction.getFictionName());
|
|
System.out.println("总章节数: " + chapters.size());
|
|
System.out.println("有内容章节数: " + chaptersWithContent);
|
|
if (chaptersWithContent > 0) {
|
|
int totalWords = chapters.stream()
|
|
.filter(c -> c.getTextContent() != null)
|
|
.mapToInt(c -> c.getTextContent().length())
|
|
.sum();
|
|
System.out.println("总字数: " + totalWords);
|
|
}
|
|
} else {
|
|
System.out.println("❌ 解析失败或没有找到章节");
|
|
}
|
|
|
|
System.out.println("=== 测试完成 ===");
|
|
}
|
|
} |