哔哩哔哩 爬取UP主专栏图片

爬取专栏图片.ps1

param (
    [ValidateRange(1, 999999999)]
    [int]
    $userId,
    [ValidateNotNullOrEmpty()]
    [string]
    $saveDir
)

#$sessdata = ''
$PSStyle.Progress.UseOSCIndicator = $true
$PSStyle.Progress.View = 'Classic'
$ErrorActionPreference = 'Stop'

if (!(Test-Path $saveDir)) {
    $null = mkdir $saveDir
}

$page = 0
$totalPagesCount = 1
$articles = $()
do {
    $page++
    $response = Invoke-RestMethod -Method Get `
        -Uri 'https://api.bilibili.com/x/space/article' `
        -Body @{ mid = $userId; pn = $page }
    #-Headers @{ SESSDATA = $sessdata }
    $articles += $response.data.articles
    $totalPagesCount = [int]($response.data.count / $response.data.ps)
    Write-Progress '爬取专栏' -Id 0 -Status "进行中 [$page/$totalPagesCount] 页 / 专栏 [$($articles.Count)] 个 ->" -PercentComplete ($page / $totalPagesCount * 100)
} while ($page -lt $totalPagesCount)

Write-Progress '爬取专栏' -Id 0 -Status '已完成'

$imageRegex = [regex]'<figure.*?data-src="(?<imageLink>//i0.hdslb.com/bfs/article/.*?)".*?<figcaption.*?>(?<imageDesc>.*?)</figcaption>.*?</figure>'

function NormalizePath([string] $path) {
    $sb = [System.Text.StringBuilder]::new()
    foreach ($char in $path.ToCharArray()) {
        if ($char -notmatch '[\x00-\x1F\x7F"\*\/:<>\?\\\|]') {
            $null = $sb.Append($char)
        }
        else {
            $null = $sb.AppendFormat("%{0:X2}", [int]$char)
        }
    }
    $sb.ToString()
}

$articleCount = 0;
$addedCount = 0;
$ignoredCount = 0;
foreach ($article in $articles) {
    $articleCount++
    Write-Progress '爬取图片' -Id 1 -Status "已处理 [$articleCount/$($articles.Count)] 个专栏 ->" -PercentComplete ($articleCount / $articles.Count * 100)
    Write-Progress '添加下载' -Id 2 -ParentId 1 -Status "添加 $addedCount 个 | 忽略 $ignoredCount 个"

    $articleSaveDir = "$saveDir\$(NormalizePath $article.title)"
    if (Test-Path $articleSaveDir) {
        $ignoredCount++
        continue
    }
    else {
        $addedCount++
        $null = mkdir $articleSaveDir
    }

    do {
        $failed = $false
        $job = Start-Job {
            $ProgressPreference = 'SilentlyContinue';
            Invoke-WebRequest "https://www.bilibili.com/read/cv$input/"
        } -InputObject $article.id
        $null = Wait-Job $job -Timeout 5
        if ($job.State -ne 'Completed') {
            Stop-Job $job
            $failed = $true
        }
    } until (!$failed)
    $result = Receive-Job $job

    $imageCount = 0
    foreach ($match in $imageRegex.Matches($result.Content)) {
        $imageCount++
        $imageLink = "https:$($match.Groups['imageLink'].Value)"

        $imageDesc = $match.Groups['imageDesc'].Value
        $index = $imageDesc.IndexOf('<br')
        if ($index -ne -1) {
            $imageDesc = $imageDesc.Remove($index)
        }
        if ($imageDesc -eq '') {
            $imageDesc = $imageCount.ToString()
        }

        $fileName = (NormalizePath $imageDesc) + '.jpg'
        
        $ProgressPreference = 'SilentlyContinue'
        $null = Invoke-WebRequest 'http://127.0.0.1:6800/jsonrpc' `
            -Method Post `
            -Headers @{
            "Content-Type" = "application/json; charset=utf-8"
        } `
            -Body ([System.Text.Encoding]::UTF8.GetBytes((ConvertTo-Json @{
                        jsonrpc = '2.0';
                        method  = 'aria2.addUri';
                        id      = 'wdnmd';
                        params  = @($imageLink),
                        @{dir = $articleSaveDir; out = $fileName }
                    })))
        $ProgressPreference = 'Continue'
    }
}

Write-Progress '爬取图片' -Id 1 -Status "已完成"