如何将《吃喝玩乐之重生1997》全部章节网址和全文爬下来

# 定义小说页面URL
$novelUrl = "http://www./105790648/"

# 设置请求头，模拟浏览器访问
$headers = @{
    "User-Agent" = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
    "Accept" = "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8"
    "Accept-Language" = "zh-CN,zh;q=0.9,en;q=0.8"
}

try {
    Write-Host "正在下载网页内容..." -ForegroundColor Yellow
   
    # 使用Invoke-WebRequest下载网页内容
    $response = Invoke-WebRequest -Uri $novelUrl -Headers $headers -UseBasicParsing
   
    # 获取HTML内容
    $htmlContent = $response.Content
   
    Write-Host "网页下载成功！" -ForegroundColor Green
   
    # 创建HTML文档对象
    $html = New-Object -ComObject "HTMLFile"
    $html.IHTMLDocument2_write($htmlContent)
   
    # 查找所有章节链接
    $chapterLinks = $html.getElementsByTagName("a") | Where-Object {
        $_.href -like "*/105790648/*.html" -and $_.title -match "^第\d+章"
    }
   
    Write-Host "找到 $($chapterLinks.Count) 个章节链接" -ForegroundColor Cyan
   
    # 提取章节信息
    $chapters = @()
    foreach ($link in $chapterLinks) {
        $chapters += [PSCustomObject]@{
            ChapterNumber = if ($link.title -match "第(\d+)章") { [int]$matches[1] } else { 0 }
            Title = $link.title
            URL = $link.href
        }
    }
   
    # 去重并按照章节号排序
    $uniqueChapters = $chapters | Sort-Object ChapterNumber -Unique
   
    # 输出结果
    Write-Host "《吃喝玩乐之重生1997》章节列表（共 $($uniqueChapters.Count) 章）" -ForegroundColor Green
    Write-Host "=" * 80
   
    # 显示所有章节
    $uniqueChapters | Format-Table -Property ChapterNumber, Title, URL -AutoSize
   
    # 导出到CSV文件
    $csvPath = "吃喝玩乐之重生1997_章节列表.csv"
    $uniqueChapters | Export-Csv -Path $csvPath -NoTypeInformation -Encoding UTF8
    Write-Host "章节信息已导出到: $csvPath" -ForegroundColor Yellow
   
    # 显示统计信息
    Write-Host "`n统计信息：" -ForegroundColor Cyan
    Write-Host "总章节数: $($uniqueChapters.Count)" -ForegroundColor White
    Write-Host "第一章: $(($uniqueChapters | Sort-Object ChapterNumber | Select-Object -First 1).Title)" -ForegroundColor White
    Write-Host "最新章: $(($uniqueChapters | Sort-Object ChapterNumber -Descending | Select-Object -First 1).Title)" -ForegroundColor White
   
} catch {
    Write-Host "错误: $($_.Exception.Message)" -ForegroundColor Red
    Write-Host "可能的原因：" -ForegroundColor Red
    Write-Host "1. 网络连接问题" -ForegroundColor Red
    Write-Host "2. 网站访问限制" -ForegroundColor Red
    Write-Host "3. 网页结构发生变化" -ForegroundColor Red
}

SET DEFAULT TO (ADDBS(JUSTPATH(SYS(16))))
DECLARE long PostMessageA IN user32 long,long,long,long
IF !FILE("吃喝玩乐之重生1997.dbf")
    CREATE TABLE 吃喝玩乐之重生1997 (title C(100), addr C(254), txt M)
    USE
ENDIF
USE 吃喝玩乐之重生1997 ALIAS tu
of = CREATEOBJECT("form1")
of.show(1)
CLOSE TABLES ALL
CLEAR ALL
RETURN

DEFINE CLASS form1 as Form
    width = 800
    height = 600
    AutoCenter = .T.
    AllowOutput = .f.
    ADD OBJECT but as commandbutton WITH left=10,top=10,width=100,height=22,caption="开始"
    ADD OBJECT grd as grid WITH left=10,top=40,width=250,height=550,RecordSource="tu",AllowCellSelection=.f.
    ADD OBJECT edt as editbox WITH left=280,top=40,width=510,height=550
    ADD OBJECT web as Olecontrol with OleClass="Shell.Explorer.2",left=-100

    PROCEDURE Destroy
        UNBINDEVENTS(this.hWnd)
    ENDPROC
   
    PROCEDURE Init
        this.web.Silent = .t.
        BINDEVENT(this.hWnd, 0x401, this, "myMessage")
    ENDPROC

    PROCEDURE but.click
        ZAP IN "tu"
        thisform.edt.value = ""
        thisform.web.navigate("http://www./105790648")
    ENDPROC
   
    PROCEDURE grd.click
        thisform.edt.value = tu.txt
        thisform.edt.SelStart = 0
    ENDPROC

    PROCEDURE web.documentComplete(pdisp, url)
        IF (SYS(3095, pdisp) == SYS(3095, this))
            PostMessageA(thisform.hWnd, 0x401, 0, 0)
        ENDIF
    ENDPROC

    FUNCTION myMessage(hWnd, uMsg, wParam, lParam)
        dom = this.web.document
        IF VARTYPE(dom) != "O"
            this.Enabled = .t.
            RETURN
        ENDIF
        IF EMPTY(tu.addr)
            #if 1   && 0 or 1
                * ie版本高试用
            lis = dom.getElementsByClassName("line3")
            FOR EACH li IN lis
                INSERT INTO tu VALUES (li.firstChild.innertext, li.firstChild.href, "")
            ENDFOR
            #else
                * ie版本低试用
            lis = dom.getElementsByTagName("li")
            FOR EACH li IN lis
                IF li.classname == "line3"
                    INSERT INTO tu VALUES (;
                        li.getElementsByTagName("a").item[0].innertext,;
                        li.getElementsByTagName("a").item[0].href, "";
                    )
                ENDIF
            ENDFOR
            #endif
            GO TOP IN "tu"
            this.grd.setfocus
            this.web.navigate(ALLTRIM(tu.addr))
            RETURN     
        ENDIF
        REPLACE tu.txt WITH ALLTRIM(tu.title)+ 0h0D0A0D0A + dom.getElementById("ad").innertext
        ? ALLTRIM(tu.title)
        SKIP IN "tu"
        IF !EOF("tu")
            this.grd.setfocus
            this.web.navigate(ALLTRIM(tu.addr))
            RETURN
        ENDIF
        GO TOP IN "tu"
        this.grd.setfocus
        this.Enabled = .t.
    ENDFUNC
ENDDEFINE