Add mgstage scraper functions (#151)

javinizer · Dec 13, 2020 · a0caea7 · a0caea7
1 parent a29c166
commit a0caea7
Show file tree

Hide file tree

Showing 3 changed files with 437 additions and 0 deletions.
diff --git a/src/Javinizer/Private/Scraper.Mgstage.ps1 b/src/Javinizer/Private/Scraper.Mgstage.ps1
@@ -0,0 +1,303 @@
+function Get-MgstageId {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        $id = (((($Webrequest.Content -split '<th>品番：<\/th>')[1] -split '<\/td>')[0]) -split '<td>')[1]
+
+        if ($id -eq '') {
+            $id = $null
+        }
+
+        Write-Output $Id
+    }
+}
+
+function Get-MgstageTitle {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        $title = (($Webrequest.Content -split '<title>')[1] -split '<\/title>')[0]
+        $title = Convert-HtmlCharacter -String $title
+
+        if ($title -eq '') {
+            $title = $null
+        }
+
+        Write-Output $Title
+    }
+}
+
+function Get-MgstageDescription {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        if ($Webrequest.Content -match '<p class="txt introduction">') {
+            $description = (($Webrequest.Content -split '<p class="txt introduction">')[1] -split '<\/p>')[0]
+            $description = Convert-HtmlCharacter -String $description
+        } else {
+            $description = $null
+        }
+
+        Write-Output $description
+    }
+}
+
+function Get-MgstageReleaseDate {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        $releaseDate = (((($Webrequest.Content -split '<th>配信開始日：<\/th>')[1] -split '<\/td>')[0]) -split '<td>')[1]
+        $releaseDate = Get-Date $releaseDate -Format "yyyy-MM-dd"
+
+        Write-Output $releaseDate
+    }
+}
+
+function Get-MgstageReleaseYear {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        $releaseYear = Get-MgstageReleaseDate -WebRequest $Webrequest
+        $releaseYear = ($releaseYear -split '-')[0]
+
+        Write-Output $releaseYear
+    }
+}
+
+function Get-MgstageRuntime {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        $length = ((($Webrequest.Content -split '<th>収録時間：<\/th>')[1] -split '<\/td>')[0] -split '<td>')[1]
+        $length = ($length -replace 'min').Trim()
+
+        Write-Output $length
+    }
+}
+
+function Get-MgstageMaker {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        $maker = (((($Webrequest.Content -split '<th>メーカー：<\/th>')[1] -split '<\/td>')[0] -split '>')[2] -split '<\/a')[0]
+        $maker = Convert-HtmlCharacter -String $maker
+
+        if ($maker -eq '') {
+            $maker = $null
+        }
+
+        Write-Output $maker
+    }
+}
+
+function Get-MgstageLabel {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest,
+
+        [Parameter()]
+        [Object]$Replace
+    )
+
+    process {
+        $label = (((($Webrequest.Content -split '<th>レーベル：<\/th>')[1] -split '<\/td>')[0] -split '>')[2] -split '<\/a')[0]
+        $label = Convert-HtmlCharacter -String $label
+
+        if ($label -eq '') {
+            $label = $null
+        }
+
+        Write-Output $label
+    }
+}
+
+function Get-MgstageSeries {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest,
+
+        [Parameter()]
+        [Object]$Replace
+    )
+
+    process {
+        $series = (((($Webrequest.Content -split '<th>シリーズ：<\/th>')[1] -split '<\/td>')[0] -split '>')[2] -split '<\/a')[0]
+        $series = Convert-HtmlCharacter -String $series
+
+        if ($series -eq '') {
+            $series = $null
+        }
+
+        Write-Output $series
+    }
+}
+
+function Get-MgstageRating {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        try {
+            $rating = ($Webrequest.Content | Select-String -Pattern '<span class="star_.*"><\/span>(.*)').Matches.Groups[1].Value
+        } catch {
+            return
+        }
+
+        # Multiply the rating value by 2 to conform to 1-10 rating standard
+        $newRating = [Decimal]$rating * 2
+        $newRating = [Math]::Round($newRating, 1)
+
+        if ($newRating -eq 0) {
+            $rating = $null
+        } else {
+            $rating = $newRating.ToString()
+        }
+
+        try {
+            $ratingCount = (($Webrequest.Content | Select-String -Pattern '\((\d) 件\)').Matches.Groups[1].Value).ToString()
+        } catch {
+            return
+        }
+
+        if ($ratingCount -eq 0) {
+            $ratingObject = $null
+        } else {
+            $ratingObject = [PSCustomObject]@{
+                Rating = $rating
+                Votes  = $ratingCount
+            }
+        }
+
+        Write-Output $ratingObject
+    }
+}
+
+function Get-MgstageGenre {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        $genreArray = @()
+        $genreHtml = ((($Webrequest.Content -split '<th>ジャンル：<\/th>')[1] -split '<\/td>')[0]) -split '<a href="\/search\/search\.php\?image_word_ids\[\]=.*">' | ForEach-Object { ($_ -replace '<td>' -replace '<\/a>').Trim() } | Where-Object { $_ -ne '' }
+
+        foreach ($genre in $genreHtml) {
+            $genre = Convert-HtmlCharacter -String $genre
+            $genreArray += $genre
+        }
+
+        if ($genreArray.Count -eq 0) {
+            $genreArray = $null
+        }
+
+        Write-Output $genreArray
+    }
+}
+
+function Get-MgstageActress {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        $movieActressObject = @()
+        $movieActress = (((($Webrequest.Content -split '<th>出演：<\/th>')[1] -split '<\/td>')[0]) -replace '<td>' -replace '<\/a>' -replace '<a href="\/search\/search\.php\?image_word_ids\[\]=.*">') -split '\n' `
+        | ForEach-Object { ($_).Trim() } | Where-Object {$_ -ne ''}
+
+        foreach ($actress in $movieActress) {
+            # Match if the name contains Japanese characters
+            if ($actress -match '[\u3040-\u309f]|[\u30a0-\u30ff]|[\uff66-\uff9f]|[\u4e00-\u9faf]') {
+                $movieActressObject += [PSCustomObject]@{
+                    LastName     = $null
+                    FirstName    = $null
+                    JapaneseName = $actress
+                    ThumbUrl     = $null
+                }
+            } else {
+                $movieActressObject += [PSCustomObject]@{
+                    LastName     = ($actress -split ' ')[1] -replace '\\', ''
+                    FirstName    = ($actress -split ' ')[0] -replace '\\', ''
+                    JapaneseName = $null
+                    ThumbUrl     = $null
+                }
+            }
+        }
+
+        if ($movieActressObject.Count -eq 0) {
+            $movieActressObject = $null
+        }
+
+        Write-Output $movieActressObject
+    }
+}
+
+function Get-MgstageCoverUrl {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        try {
+            #$coverUrl = ($Webrequest.Content | Select-String -Pattern '<img src="(.*)" width=".*" height=".*" class="enlarge_image"').Matches.Groups[1].Value
+            $coverUrl = ($Webrequest.Content | Select-String -Pattern 'class="link_magnify" href="(.*\.jpg)"').Matches.Groups[1].Value
+        } catch {
+            return
+        }
+
+        Write-Output $coverUrl
+    }
+}
+
+function Get-MgstageScreenshotUrl {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        try {
+            $screenshotUrl = ( $Webrequest.Content | Select-String -Pattern 'class="sample_image" href="(.*.jpg)"' -AllMatches ).Matches | ForEach-Object { $_.Groups[1].Value }
+        } catch {
+            return
+        }
+
+        Write-Output $screenshotUrl
+    }
+}
+
+function Get-MgstageTrailerUrl {
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true)]
+        [Object]$Webrequest
+    )
+
+    process {
+        Write-Output $trailerUrl
+    }
+}
diff --git a/src/Javinizer/Public/Get-MgstageData.ps1 b/src/Javinizer/Public/Get-MgstageData.ps1
@@ -0,0 +1,52 @@
+#Requires -PSEdition Core
+
+function Get-MgstageData {
+    [CmdletBinding()]
+    param (
+        [Parameter(Mandatory = $true, Position = 0, ValueFromPipeline = $true, ValueFromPipelineByPropertyName = $true)]
+        [String]$Url
+    )
+
+    begin {
+        $session = New-Object Microsoft.PowerShell.Commands.WebRequestSession
+        $cookie = New-Object System.Net.Cookie
+        $cookie.Name = 'adc'
+        $cookie.Value = '1'
+        $cookie.Domain = '.mgstage.com'
+        $session.Cookies.Add($cookie)
+    }
+
+    process {
+        $movieDataObject = @()
+
+        try {
+            Write-JVLog -Write:$script:JVLogWrite -LogPath $script:JVLogPath -WriteLevel $script:JVLogWriteLevel -Level Debug -Message "[$($MyInvocation.MyCommand.Name)] Performing [GET] on URL [$Url]"
+            $webRequest = Invoke-WebRequest -Uri $Url -Method Get -WebSession $session -Verbose:$false
+        } catch {
+            Write-JVLog -Write:$script:JVLogWrite -LogPath $script:JVLogPath -WriteLevel $script:JVLogWriteLevel -Level Error -Message "[$($MyInvocation.MyCommand.Name)] Error [GET] on URL [$Url]: $PSItem" -Action 'Continue'
+        }
+
+        $movieDataObject = [PSCustomObject]@{
+            Source        = 'mgstageja'
+            Url           = $Url
+            Id            = Get-MgstageId -WebRequest $webRequest
+            Title         = Get-MgstageTitle -WebRequest $webRequest
+            Description   = Get-MgstageDescription -WebRequest $webRequest
+            ReleaseDate   = Get-MgstageReleaseDate -WebRequest $webRequest
+            ReleaseYear   = Get-MgstageReleaseYear -WebRequest $webRequest
+            Runtime       = Get-MgstageRuntime -WebRequest $webRequest
+            Maker         = Get-MgstageMaker -WebRequest $webRequest
+            Label         = Get-MgstageLabel -WebRequest $webRequest
+            Series        = Get-MgstageSeries -WebRequest $webRequest
+            Rating        = Get-MGstageRating -WebRequest $webRequest
+            Actress       = Get-MgstageActress -WebRequest $webRequest
+            Genre         = Get-MgstageGenre -WebRequest $webRequest
+            CoverUrl      = Get-MgstageCoverUrl -WebRequest $webRequest
+            ScreenshotUrl = Get-MgstageScreenshotUrl -WebRequest $webRequest
+            TrailerUrl    = Get-MgstageTrailerUrl -WebRequest $webRequest
+        }
+
+        Write-JVLog -Write:$script:JVLogWrite -LogPath $script:JVLogPath -WriteLevel $script:JVLogWriteLevel -Level Debug -Message "[$($MyInvocation.MyCommand.Name)] Mgstage data object: $($movieDataObject | ConvertTo-Json -Depth 32 -Compress)"
+        Write-Output $movieDataObject
+    }
+}