Regalamiunsorriso/scripts/trim-dump-via-temp-mysql.ps1

param(
    [string]$SourceDump = (Join-Path $PSScriptRoot '..\db\pg-model-seed-trimmed-20260421.sql'),
    [string]$OverlayDump = (Join-Path $PSScriptRoot '..\db\pg-local-model-fixtures-overlay-20260422.sql'),
    [string]$OutputDump = (Join-Path $PSScriptRoot '..\db\pg-local-purpose-seed-20260422.sql'),
    [int[]]$KeepRaceIds = @(1018547, 1018557),
    [int[]]$KeepUserIds = @(2),
    [string]$ContainerName = 'regalami-dump-trim-mysql-temp',
    [string]$VolumeName = 'regalami-dump-trim-mysql-temp-data',
    [string]$DatabaseName = 'pgtrim',
    [string]$RootPassword = 'root'
)

$ErrorActionPreference = 'Stop'

function Invoke-DockerCapture {
    param(
        [Parameter(Mandatory = $true)]
        [string[]]$DockerArgs
    )

    $output = & docker @DockerArgs 2>&1
    if ($LASTEXITCODE -ne 0) {
        throw (("docker " + ($DockerArgs -join ' ')) + " failed:`n" + ($output -join "`n"))
    }

    return $output
}

function Invoke-DockerQuiet {
    param(
        [Parameter(Mandatory = $true)]
        [string[]]$DockerArgs
    )

    & docker @DockerArgs | Out-Null
    if ($LASTEXITCODE -ne 0) {
        throw "docker $($DockerArgs -join ' ') failed"
    }
}

function Wait-ForMysqlReady {
    param(
        [Parameter(Mandatory = $true)]
        [string]$Name,
        [Parameter(Mandatory = $true)]
        [string]$Password
    )

    for ($attempt = 0; $attempt -lt 180; $attempt++) {
        $logs = & docker logs $Name 2>&1
        if ($LASTEXITCODE -eq 0 -and ($logs -join "`n") -match 'ready for connections.*port: 3306') {
            return
        }

        Start-Sleep -Seconds 2
    }

    throw "MySQL in container $Name did not become ready in time."
}

function Invoke-MysqlQuery {
    param(
        [Parameter(Mandatory = $true)]
        [string]$Query,
        [switch]$SkipDatabase
    )

    $dockerArgs = @('exec', '-e', "MYSQL_PWD=$RootPassword", $ContainerName, 'mysql', '-N', '-B', '-uroot')
    if (-not $SkipDatabase) {
        $dockerArgs += @('-D', $DatabaseName)
    }
    $dockerArgs += @('-e', $Query)
    return Invoke-DockerCapture -DockerArgs $dockerArgs
}

function Get-SqlScalar {
    param(
        [Parameter(Mandatory = $true)]
        [string]$Query,
        [switch]$SkipDatabase
    )

    $result = Invoke-MysqlQuery -Query $Query -SkipDatabase:$SkipDatabase
    if (-not $result) {
        return ''
    }

    return ($result | Select-Object -First 1).Trim()
}

function Quote-Identifier {
    param([string]$Name)
    return '`' + $Name.Replace('`', '``') + '`'
}

function ConvertTo-SqlIntList {
    param(
        [Parameter(Mandatory = $true)]
        [int[]]$Values
    )

    $dedupedValues = $Values | Sort-Object -Unique
    if (-not $dedupedValues -or $dedupedValues.Count -eq 0) {
        throw 'At least one numeric keep id is required.'
    }

    return ($dedupedValues | ForEach-Object { [string]$_ }) -join ','
}

if (-not (Test-Path $SourceDump)) {
    throw "Source dump not found: $SourceDump"
}

$dbDirectory = Split-Path -Parent (Resolve-Path $SourceDump).Path
$sourceFileName = Split-Path -Leaf $SourceDump
$outputFileName = Split-Path -Leaf $OutputDump
$overlayResolvedPath = $null
$overlayFileName = ''
$overlayDirectory = $dbDirectory
$overlayMount = @()
if ($OverlayDump -and (Test-Path $OverlayDump)) {
    $overlayResolvedPath = (Resolve-Path $OverlayDump).Path
    $overlayFileName = Split-Path -Leaf $overlayResolvedPath
    $overlayDirectory = Split-Path -Parent $overlayResolvedPath
    if ($overlayDirectory -ne $dbDirectory) {
        $overlayMount = @('-v', "${overlayDirectory}:/workspace/overlay")
    }
}
$importErrorLogFileName = [System.IO.Path]::GetFileNameWithoutExtension($sourceFileName) + '.import-errors.log'
$importErrorLogPath = Join-Path $dbDirectory $importErrorLogFileName
$sourceSizeBytes = (Get-Item $SourceDump).Length
$keepRaceListSql = ConvertTo-SqlIntList -Values $KeepRaceIds
$keepUserListSql = ConvertTo-SqlIntList -Values $KeepUserIds

Write-Host "Using source dump: $SourceDump"
Write-Host "Source size: $sourceSizeBytes bytes"
Write-Host "Import error log: $importErrorLogPath"
if ($overlayResolvedPath) {
    Write-Host "Using overlay dump: $overlayResolvedPath"
}
Write-Host "KeepRaceIdsRequested=$keepRaceListSql"
Write-Host "KeepUserIdsRequested=$keepUserListSql"

try {
    & docker rm -f $ContainerName 1>$null 2>$null
    & docker volume rm $VolumeName 1>$null 2>$null

    $dockerRunArgs = @(
        'run', '-d', '--name', $ContainerName,
        '-e', "MYSQL_ROOT_PASSWORD=$RootPassword",
        '-e', 'MYSQL_ROOT_HOST=%',
        '-v', "${dbDirectory}:/workspace/db",
        '-v', "${VolumeName}:/var/lib/mysql"
    )
    if ($overlayMount.Count -gt 0) {
        $dockerRunArgs += $overlayMount
    }
    $dockerRunArgs += @(
        'mysql:8.4',
        '--max_allowed_packet=1G',
        '--net_read_timeout=600',
        '--net_write_timeout=600'
    )

    Invoke-DockerCapture -DockerArgs $dockerRunArgs | Out-Null
    Write-Host 'Temporary MySQL container started.'

    Wait-ForMysqlReady -Name $ContainerName -Password $RootPassword
    Write-Host 'Temporary MySQL is ready.'

    Invoke-MysqlQuery -SkipDatabase -Query "DROP DATABASE IF EXISTS $DatabaseName; CREATE DATABASE $DatabaseName CHARACTER SET utf8mb4 COLLATE utf8mb4_0900_ai_ci;" | Out-Null
    Write-Host 'Temporary database created.'

    Write-Host "Importing source dump into temporary MySQL database..."
    if (Test-Path $importErrorLogPath) {
        Remove-Item $importErrorLogPath -Force
    }
    Invoke-DockerQuiet -DockerArgs @(
        'exec', $ContainerName, 'sh', '-lc',
        "mysql --force -uroot -p$RootPassword $DatabaseName < /workspace/db/$sourceFileName 2> /workspace/db/$importErrorLogFileName"
    )

    if ($overlayResolvedPath) {
        $overlayContainerPath = if ($overlayDirectory -eq $dbDirectory) {
            "/workspace/db/$overlayFileName"
        }
        else {
            "/workspace/overlay/$overlayFileName"
        }

        Write-Host "Importing overlay dump into temporary MySQL database..."
        Invoke-DockerQuiet -DockerArgs @(
            'exec', $ContainerName, 'sh', '-lc',
            "mysql --force -uroot -p$RootPassword $DatabaseName < $overlayContainerPath"
        )
    }

    $beforeGara = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM gara;')
    $beforePuntoFoto = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM punto_foto;')
    $beforeFoto = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM foto;')
    $beforeLogFoto = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM log_foto;')
    $usersExists = (Get-SqlScalar -SkipDatabase -Query "SELECT COUNT(*) FROM information_schema.tables WHERE table_schema = '$DatabaseName' AND table_name = 'users';") -eq '1'
    $beforeUsers = 0
    if ($usersExists) {
        $beforeUsers = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM users;')
    }

    Invoke-MysqlQuery -Query @"
DROP TABLE IF EXISTS keep_gara_ids;
CREATE TABLE keep_gara_ids AS
SELECT id_gara
FROM gara
WHERE id_gara IN ($keepRaceListSql)
ORDER BY FIELD(id_gara, $keepRaceListSql);

DROP TABLE IF EXISTS keep_punto_foto_ids;
CREATE TABLE keep_punto_foto_ids AS
SELECT id_puntoFoto
FROM punto_foto
WHERE id_gara IN (SELECT id_gara FROM keep_gara_ids);

DROP TABLE IF EXISTS keep_foto_ids;
CREATE TABLE keep_foto_ids AS
SELECT id_foto
FROM foto
WHERE id_gara IN (SELECT id_gara FROM keep_gara_ids);
"@ | Out-Null

    $keptGaraIds = Invoke-MysqlQuery -Query 'SELECT id_gara FROM keep_gara_ids ORDER BY id_gara;'
    $missingRaceIds = @()
    foreach ($requestedRaceId in ($KeepRaceIds | Sort-Object -Unique)) {
        if (-not (($keptGaraIds | ForEach-Object { $_.Trim() }) -contains [string]$requestedRaceId)) {
            $missingRaceIds += $requestedRaceId
        }
    }
    if ($missingRaceIds.Count -gt 0) {
        throw "One or more requested race ids are missing from the imported data: $($missingRaceIds -join ',')"
    }

    $cleanupStats = [System.Collections.Generic.List[string]]::new()

    Invoke-MysqlQuery -Query 'DELETE FROM log_foto;' | Out-Null
    $cleanupStats.Add('log_foto:deleted-all')

    $garaDependentTables = Invoke-MysqlQuery -SkipDatabase -Query @"
SELECT table_name
FROM information_schema.columns
WHERE table_schema = '$DatabaseName'
  AND column_name = 'id_gara'
  AND table_name NOT IN ('gara', 'foto', 'punto_foto', 'keep_gara_ids', 'keep_punto_foto_ids', 'keep_foto_ids', 'keep_user_ids')
ORDER BY table_name;
"@

    foreach ($tableName in $garaDependentTables) {
        if (-not $tableName) {
            continue
        }
        $trimmedTableName = $tableName.Trim()
        $quotedTable = Quote-Identifier $trimmedTableName
        $deletedRows = [int64](Get-SqlScalar -Query "SELECT COUNT(*) FROM $quotedTable WHERE id_gara IS NOT NULL AND id_gara NOT IN (SELECT id_gara FROM keep_gara_ids);")
        if ($deletedRows -gt 0) {
            Invoke-MysqlQuery -Query "DELETE FROM $quotedTable WHERE id_gara IS NOT NULL AND id_gara NOT IN (SELECT id_gara FROM keep_gara_ids);" | Out-Null
            $cleanupStats.Add("$($trimmedTableName):$deletedRows")
        }
    }

    $fotoDependentTables = Invoke-MysqlQuery -SkipDatabase -Query @"
SELECT table_name
FROM information_schema.columns
WHERE table_schema = '$DatabaseName'
  AND column_name = 'id_foto'
  AND table_name NOT IN ('foto', 'log_foto', 'keep_foto_ids', 'keep_gara_ids', 'keep_punto_foto_ids', 'keep_user_ids')
ORDER BY table_name;
"@

    foreach ($tableName in $fotoDependentTables) {
        if (-not $tableName) {
            continue
        }
        $trimmedTableName = $tableName.Trim()
        $quotedTable = Quote-Identifier $trimmedTableName
        $deletedRows = [int64](Get-SqlScalar -Query "SELECT COUNT(*) FROM $quotedTable WHERE id_foto IS NOT NULL AND id_foto NOT IN (SELECT id_foto FROM keep_foto_ids);")
        if ($deletedRows -gt 0) {
            Invoke-MysqlQuery -Query "DELETE FROM $quotedTable WHERE id_foto IS NOT NULL AND id_foto NOT IN (SELECT id_foto FROM keep_foto_ids);" | Out-Null
            $cleanupStats.Add("$($trimmedTableName):$deletedRows")
        }
    }

    $deletedFotoRows = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM foto WHERE id_foto NOT IN (SELECT id_foto FROM keep_foto_ids);')
    if ($deletedFotoRows -gt 0) {
        Invoke-MysqlQuery -Query 'DELETE FROM foto WHERE id_foto NOT IN (SELECT id_foto FROM keep_foto_ids);' | Out-Null
        $cleanupStats.Add("foto:$deletedFotoRows")
    }

    $deletedPuntoFotoRows = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM punto_foto WHERE id_gara NOT IN (SELECT id_gara FROM keep_gara_ids);')
    if ($deletedPuntoFotoRows -gt 0) {
        Invoke-MysqlQuery -Query 'DELETE FROM punto_foto WHERE id_gara NOT IN (SELECT id_gara FROM keep_gara_ids);' | Out-Null
        $cleanupStats.Add("punto_foto:$deletedPuntoFotoRows")
    }

    $deletedGaraRows = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM gara WHERE id_gara NOT IN (SELECT id_gara FROM keep_gara_ids);')
    if ($deletedGaraRows -gt 0) {
        Invoke-MysqlQuery -Query 'DELETE FROM gara WHERE id_gara NOT IN (SELECT id_gara FROM keep_gara_ids);' | Out-Null
        $cleanupStats.Add("gara:$deletedGaraRows")
    }

    $keptUserIds = @()
    if ($usersExists) {
        Invoke-MysqlQuery -Query @"
DROP TABLE IF EXISTS keep_user_ids;
CREATE TABLE keep_user_ids AS
SELECT id_users
FROM users
WHERE id_users IN ($keepUserListSql)
ORDER BY FIELD(id_users, $keepUserListSql);
"@ | Out-Null

        $keptUserIds = Invoke-MysqlQuery -Query 'SELECT id_users FROM keep_user_ids ORDER BY id_users;'
        $missingUserIds = @()
        foreach ($requestedUserId in ($KeepUserIds | Sort-Object -Unique)) {
            if (-not (($keptUserIds | ForEach-Object { $_.Trim() }) -contains [string]$requestedUserId)) {
                $missingUserIds += $requestedUserId
            }
        }
        if ($missingUserIds.Count -gt 0) {
            throw "One or more requested user ids are missing from the imported data: $($missingUserIds -join ',')"
        }

        $usersDependentTables = Invoke-MysqlQuery -SkipDatabase -Query @"
SELECT table_name
FROM information_schema.columns
WHERE table_schema = '$DatabaseName'
  AND column_name = 'id_users'
  AND table_name NOT IN ('users', 'keep_user_ids', 'keep_gara_ids', 'keep_punto_foto_ids', 'keep_foto_ids')
ORDER BY table_name;
"@

        foreach ($tableName in $usersDependentTables) {
            if (-not $tableName) {
                continue
            }
            $trimmedTableName = $tableName.Trim()
            $quotedTable = Quote-Identifier $trimmedTableName
            $deletedRows = [int64](Get-SqlScalar -Query "SELECT COUNT(*) FROM $quotedTable WHERE id_users IS NOT NULL AND id_users NOT IN (SELECT id_users FROM keep_user_ids);")
            if ($deletedRows -gt 0) {
                Invoke-MysqlQuery -Query "DELETE FROM $quotedTable WHERE id_users IS NOT NULL AND id_users NOT IN (SELECT id_users FROM keep_user_ids);" | Out-Null
                $cleanupStats.Add("$($trimmedTableName):$deletedRows")
            }
        }

        $deletedUsers = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM users WHERE id_users NOT IN (SELECT id_users FROM keep_user_ids);')
        if ($deletedUsers -gt 0) {
            Invoke-MysqlQuery -Query 'DELETE FROM users WHERE id_users NOT IN (SELECT id_users FROM keep_user_ids);' | Out-Null
            $cleanupStats.Add("users:$deletedUsers")
        }
    }

    $afterGara = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM gara;')
    $afterPuntoFoto = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM punto_foto;')
    $afterFoto = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM foto;')
    $afterLogFoto = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM log_foto;')
    $afterUsers = 0
    if ($usersExists) {
        $afterUsers = [int64](Get-SqlScalar -Query 'SELECT COUNT(*) FROM users;')
    }

    if (Test-Path $OutputDump) {
        Remove-Item $OutputDump -Force
    }

    Write-Host "Exporting curated dump..."
    Invoke-DockerQuiet -DockerArgs @(
        'exec', $ContainerName, 'sh', '-lc',
        "mysqldump -uroot -p$RootPassword --default-character-set=utf8mb4 --single-transaction --routines --triggers --set-gtid-purged=OFF $DatabaseName > /workspace/db/$outputFileName"
    )

    if (-not (Test-Path $OutputDump)) {
        throw "Curated dump was not created: $OutputDump"
    }

    $outputSizeBytes = (Get-Item $OutputDump).Length

    Write-Host "OriginalSizeBytes=$sourceSizeBytes"
    Write-Host "CuratedSizeBytes=$outputSizeBytes"
    Write-Host "BeforeGara=$beforeGara"
    Write-Host "AfterGara=$afterGara"
    Write-Host "BeforePuntoFoto=$beforePuntoFoto"
    Write-Host "AfterPuntoFoto=$afterPuntoFoto"
    Write-Host "BeforeFoto=$beforeFoto"
    Write-Host "AfterFoto=$afterFoto"
    Write-Host "BeforeLogFoto=$beforeLogFoto"
    Write-Host "AfterLogFoto=$afterLogFoto"
    Write-Host "BeforeUsers=$beforeUsers"
    Write-Host "AfterUsers=$afterUsers"
    Write-Host ("KeptGaraIds=" + (($keptGaraIds | ForEach-Object { $_.Trim() }) -join ','))
    if ($usersExists) {
        Write-Host ("KeptUserIds=" + (($keptUserIds | ForEach-Object { $_.Trim() }) -join ','))
    }
    Write-Host ("DependencyCleanup=" + (($cleanupStats | Sort-Object) -join ';'))
    if (Test-Path $importErrorLogPath) {
        $importErrorCount = (Get-Item $importErrorLogPath).Length
        Write-Host "ImportErrorLogBytes=$importErrorCount"
    }
}
finally {
    & docker rm -f $ContainerName 1>$null 2>$null
    & docker volume rm $VolumeName 1>$null 2>$null
}