Sharepoint-Toolbox/SharepointToolbox/Services/Export/StorageCsvExportService.cs

using SharepointToolbox.Core.Models;
using SharepointToolbox.Localization;
using System.IO;
using System.Text;

namespace SharepointToolbox.Services.Export;

/// <summary>
/// Exports a flat list of StorageNode objects to a UTF-8 BOM CSV.
/// Compatible with Microsoft Excel (BOM signals UTF-8 encoding).
/// </summary>
public class StorageCsvExportService
{
    /// <summary>
    /// Builds a single-section CSV: header row plus one row per
    /// <see cref="StorageNode"/> with library, site, file count, total size
    /// (MB), version size (MB), and last-modified date.
    /// </summary>
    public string BuildCsv(IReadOnlyList<StorageNode> nodes)
    {
        var T = TranslationSource.Instance;
        var sb = new StringBuilder();

        // Header
        sb.AppendLine($"{T["report.col.library"]},{T["report.col.site"]},{T["report.stat.files"]},{T["report.col.total_size_mb"]},{T["report.col.version_size_mb"]},{T["report.col.last_modified"]}");

        foreach (var node in nodes)
        {
            sb.AppendLine(string.Join(",",
                Csv(node.Name),
                Csv(node.SiteTitle),
                node.TotalFileCount.ToString(),
                FormatMb(node.TotalSizeBytes),
                FormatMb(node.VersionSizeBytes),
                node.LastModified.HasValue
                    ? Csv(node.LastModified.Value.ToString("yyyy-MM-dd"))
                    : string.Empty));
        }

        return sb.ToString();
    }

    /// <summary>Writes the library-level CSV to <paramref name="filePath"/> with UTF-8 BOM.</summary>
    public async Task WriteAsync(IReadOnlyList<StorageNode> nodes, string filePath, CancellationToken ct)
    {
        var csv = BuildCsv(nodes);
        await ExportFileWriter.WriteCsvAsync(filePath, csv, ct);
    }

    /// <summary>
    /// Builds a CSV with library details followed by a file-type breakdown section.
    /// </summary>
    public string BuildCsv(IReadOnlyList<StorageNode> nodes, IReadOnlyList<FileTypeMetric> fileTypeMetrics)
    {
        var T = TranslationSource.Instance;
        var sb = new StringBuilder();

        // Library details
        sb.AppendLine($"{T["report.col.library"]},{T["report.col.site"]},{T["report.stat.files"]},{T["report.col.total_size_mb"]},{T["report.col.version_size_mb"]},{T["report.col.last_modified"]}");
        foreach (var node in nodes)
        {
            sb.AppendLine(string.Join(",",
                Csv(node.Name),
                Csv(node.SiteTitle),
                node.TotalFileCount.ToString(),
                FormatMb(node.TotalSizeBytes),
                FormatMb(node.VersionSizeBytes),
                node.LastModified.HasValue
                    ? Csv(node.LastModified.Value.ToString("yyyy-MM-dd"))
                    : string.Empty));
        }

        // File type breakdown
        if (fileTypeMetrics.Count > 0)
        {
            sb.AppendLine();
            sb.AppendLine($"{T["report.col.file_type"]},{T["report.col.size_mb"]},{T["report.col.file_count"]}");
            foreach (var m in fileTypeMetrics)
            {
                string label = string.IsNullOrEmpty(m.Extension) ? T["report.text.no_extension"] : m.Extension;
                sb.AppendLine(string.Join(",", Csv(label), FormatMb(m.TotalSizeBytes), m.FileCount.ToString()));
            }
        }

        return sb.ToString();
    }

    /// <summary>Writes the two-section CSV (libraries + file-type breakdown) with UTF-8 BOM.</summary>
    public async Task WriteAsync(IReadOnlyList<StorageNode> nodes, IReadOnlyList<FileTypeMetric> fileTypeMetrics, string filePath, CancellationToken ct)
    {
        var csv = BuildCsv(nodes, fileTypeMetrics);
        await ExportFileWriter.WriteCsvAsync(filePath, csv, ct);
    }

    /// <summary>
    /// Writes storage metrics with optional per-site partitioning.
    /// Single → one file. BySite → one file per SiteTitle. File-type metrics
    /// are replicated across all partitions because the tenant-level scan
    /// does not retain per-site breakdowns.
    /// </summary>
    public Task WriteAsync(
        IReadOnlyList<StorageNode> nodes,
        IReadOnlyList<FileTypeMetric> fileTypeMetrics,
        string basePath,
        ReportSplitMode splitMode,
        CancellationToken ct)
        => ReportSplitHelper.WritePartitionedAsync(
            nodes, basePath, splitMode,
            PartitionBySite,
            (part, path, c) => WriteAsync(part, fileTypeMetrics, path, c),
            ct);

    /// <summary>
    /// Splits the flat StorageNode list into per-site slices while preserving
    /// the DFS hierarchy (each root library followed by its indented descendants).
    /// Siblings sharing a SiteTitle roll up into the same partition.
    /// </summary>
    internal static IEnumerable<(string Label, IReadOnlyList<StorageNode> Partition)> PartitionBySite(
        IReadOnlyList<StorageNode> nodes)
    {
        var buckets = new Dictionary<string, List<StorageNode>>(StringComparer.OrdinalIgnoreCase);
        string currentSite = string.Empty;
        foreach (var node in nodes)
        {
            if (node.IndentLevel == 0)
                currentSite = string.IsNullOrWhiteSpace(node.SiteTitle)
                    ? ReportSplitHelper.DeriveSiteLabel(node.Url)
                    : node.SiteTitle;
            if (!buckets.TryGetValue(currentSite, out var list))
            {
                list = new List<StorageNode>();
                buckets[currentSite] = list;
            }
            list.Add(node);
        }
        return buckets.Select(kv => (kv.Key, (IReadOnlyList<StorageNode>)kv.Value));
    }

    // ── Helpers ───────────────────────────────────────────────────────────────

    private static string FormatMb(long bytes)
        => (bytes / (1024.0 * 1024.0)).ToString("F2");

    /// <summary>RFC 4180 CSV field quoting with formula-injection guard.</summary>
    private static string Csv(string value) => CsvSanitizer.EscapeMinimal(value);
}