wabbajack/Wabbajack.Lib/AInstaller.cs

using System;
using System.Collections.Generic;
using System.IO;
using System.IO.Compression;
using System.Linq;
using System.Threading;
using System.Threading.Tasks;
using Alphaleonis.Win32.Filesystem;
using Wabbajack.Common;
using Wabbajack.Lib.Downloaders;
using Wabbajack.VirtualFileSystem;
using Directory = Alphaleonis.Win32.Filesystem.Directory;
using File = Alphaleonis.Win32.Filesystem.File;
using FileInfo = Alphaleonis.Win32.Filesystem.FileInfo;
using Path = Alphaleonis.Win32.Filesystem.Path;

namespace Wabbajack.Lib
{
    public abstract class AInstaller : ABatchProcessor
    {
        public bool IgnoreMissingFiles { get; internal set; } = false;

        public AbsolutePath OutputFolder { get; private set; }
        public AbsolutePath DownloadFolder { get; private set; }

        public abstract ModManager ModManager { get; }

        public AbsolutePath ModListArchive { get; private set; }
        public ModList ModList { get; private set; }
        public Dictionary<Hash, AbsolutePath> HashedArchives { get; set; }
        
        public SystemParameters SystemParameters { get; set; }

        public AInstaller(AbsolutePath archive, ModList modList, AbsolutePath outputFolder, AbsolutePath downloadFolder, SystemParameters parameters)
        {
            ModList = modList;
            ModListArchive = archive;
            OutputFolder = outputFolder;
            DownloadFolder = downloadFolder;
            SystemParameters = parameters;
        }

        public void Info(string msg)
        {
            Utils.Log(msg);
        }

        public void Status(string msg)
        {
            Queue.Report(msg, Percent.Zero);
        }

        public void Error(string msg)
        {
            Utils.Log(msg);
            throw new Exception(msg);
        }

        public async Task<byte[]> LoadBytesFromPath(RelativePath path)
        {
            await using var fs = new FileStream((string)ModListArchive, FileMode.Open, FileAccess.Read, FileShare.Read);
            using var ar = new ZipArchive(fs, ZipArchiveMode.Read);
            await using var ms = new MemoryStream();
            var entry = ar.GetEntry((string)path);
            await using (var e = entry.Open())
            {
                await e.CopyToAsync(ms);
            }

            return ms.ToArray();
        }

        public static ModList LoadFromFile(AbsolutePath path)
        {
            using var fs = new FileStream((string)path, FileMode.Open, FileAccess.Read, FileShare.Read);
            using var ar = new ZipArchive(fs, ZipArchiveMode.Read);
            var entry = ar.GetEntry("modlist");
            if (entry == null)
            {
                entry = ar.GetEntry("modlist.json");
                using (var e = entry.Open())
                    return e.FromJSON<ModList>();
            }
            using (var e = entry.Open())
                return e.ReadAsMessagePack<ModList>();
        }

        /// <summary>
        ///     We don't want to make the installer index all the archives, that's just a waste of time, so instead
        ///     we'll pass just enough information to VFS to let it know about the files we have.
        /// </summary>
        protected async Task PrimeVFS()
        {
            VFS.AddKnown(ModList.Directives.OfType<FromArchive>().Select(d => d.ArchiveHashPath), HashedArchives);
            await VFS.BackfillMissing();
        }

        public void BuildFolderStructure()
        {
            Info("Building Folder Structure");
            ModList.Directives
                .Select(d => OutputFolder.Combine(d.To.Parent))
                .Distinct()
                .Do(f => f.CreateDirectory());
        }

        public async Task InstallArchives()
        {
            Info("Installing Archives");
            Info("Grouping Install Files");
            var grouped = ModList.Directives
                .OfType<FromArchive>()
                .GroupBy(e => e.ArchiveHashPath.BaseHash)
                .ToDictionary(k => k.Key);
            var archives = ModList.Archives
                .Select(a => new { Archive = a, AbsolutePath = HashedArchives.GetOrDefault(a.Hash) })
                .Where(a => a.AbsolutePath != null)
                .ToList();

            Info("Installing Archives");
            await archives.PMap(Queue, UpdateTracker,a => InstallArchive(Queue, a.Archive, a.AbsolutePath, grouped[a.Archive.Hash]));
        }

        private async Task InstallArchive(WorkQueue queue, Archive archive, AbsolutePath absolutePath, IGrouping<Hash, FromArchive> grouping)
        {
            Status($"Extracting {archive.Name}");

            List<FromArchive> vFiles = grouping.Select(g =>
            {
                var file = VFS.Index.FileForArchiveHashPath(g.ArchiveHashPath);
                g.FromFile = file;
                return g;
            }).ToList();

            var onFinish = await VFS.Stage(vFiles.Select(f => f.FromFile).Distinct());


            Status($"Copying files for {archive.Name}");

            async ValueTask CopyFile(AbsolutePath from, AbsolutePath to, bool useMove)
            {
                if (to.Exists)
                {
                    if (to.IsReadOnly)
                        to.IsReadOnly = false;
                    to.Delete();
                }

                if (from.Exists)
                {
                    if (from.IsReadOnly)
                        from.IsReadOnly = false;
                }


                if (useMove)
                    from.MoveTo(to);
                else
                    from.CopyTo(to);
                // If we don't do this, the file will use the last-modified date of the file when it was compressed
                // into an archive, which isn't really what we want in the case of files installed archives
                to.LastModified = DateTime.Now;
            }

            await vFiles.GroupBy(f => f.FromFile)
                  .PDoIndexed(queue, async (idx, group) =>
            {
                Utils.Status("Installing files", Percent.FactoryPutInRange(idx, vFiles.Count));
                var firstDest = OutputFolder.Combine(group.First().To);
                await CopyFile(group.Key.StagedPath, firstDest, true);
                
                foreach (var copy in group.Skip(1))
                {
                    await CopyFile(firstDest, OutputFolder.Combine(copy.To), false);
                }

            });

            Status("Unstaging files");
            onFinish();

            // Now patch all the files from this archive
            await grouping.OfType<PatchedFromArchive>()
                .PMap(queue, async toPatch =>
                {
                    await using var patchStream = new MemoryStream();
                    Status($"Patching {toPatch.To.FileName}");
                    // Read in the patch data

                    byte[] patchData = await LoadBytesFromPath(toPatch.PatchID);

                    var toFile = OutputFolder.Combine(toPatch.To);
                    var oldData = new MemoryStream(await toFile.ReadAllBytesAsync());

                    // Remove the file we're about to patch
                    toFile.Delete();

                    // Patch it
                    await using (var outStream = toFile.Create())
                    {
                        Utils.ApplyPatch(oldData, () => new MemoryStream(patchData), outStream);
                    }

                    Status($"Verifying Patch {toPatch.To.FileName}");
                    var resultSha = await toFile.FileHashAsync();
                    if (resultSha != toPatch.Hash)
                        throw new InvalidDataException($"Invalid Hash for {toPatch.To} after patching");
                });
        }

        public async Task DownloadArchives()
        {
            var missing = ModList.Archives.Where(a => !HashedArchives.ContainsKey(a.Hash)).ToList();
            Info($"Missing {missing.Count} archives");

            Info("Getting Nexus API Key, if a browser appears, please accept");

            var dispatchers = missing.Select(m => m.State.GetDownloader()).Distinct();

            await Task.WhenAll(dispatchers.Select(d => d.Prepare()));

            await DownloadMissingArchives(missing);
        }

        public async Task DownloadMissingArchives(List<Archive> missing, bool download = true)
        {
            if (download)
            {
                foreach (var a in missing.Where(a => a.State.GetType() == typeof(ManualDownloader.State)))
                {
                    var outputPath = DownloadFolder.Combine(a.Name);
                    await a.State.Download(a, outputPath);
                }
            }

            await missing.Where(a => a.State.GetType() != typeof(ManualDownloader.State))
                .PMap(Queue, async archive =>
                {
                    Info($"Downloading {archive.Name}");
                    var outputPath = DownloadFolder.Combine(archive.Name);

                    if (download)
                    {
                        if (outputPath.Exists)
                        {
                            var origName = Path.GetFileNameWithoutExtension(archive.Name);
                            var ext = Path.GetExtension(archive.Name);
                            var uniqueKey = archive.State.PrimaryKeyString.StringSha256Hex();
                            outputPath = DownloadFolder.Combine(origName + "_" + uniqueKey + "_" + ext);
                            outputPath.Delete();
                        }
                    }

                    return await DownloadArchive(archive, download, outputPath);
                });
        }

        public async Task<bool> DownloadArchive(Archive archive, bool download, AbsolutePath? destination = null)
        {
            try
            {
                if (destination == null) 
                    destination = DownloadFolder.Combine(archive.Name);
                await DownloadDispatcher.DownloadWithPossibleUpgrade(archive, destination.Value);
            }
            catch (Exception ex)
            {
                Utils.Log($"Download error for file {archive.Name}");
                Utils.Log(ex.ToString());
                return false;
            }

            return false;
        }

        public async Task HashArchives()
        {
            var hashResults = await DownloadFolder.EnumerateFiles()
                .Where(e => e.Extension != Consts.HashFileExtension)
                .PMap(Queue, async e => (await e.FileHashCachedAsync(), e));
            HashedArchives = hashResults
                .OrderByDescending(e => e.Item2.LastModified)
                .GroupBy(e => e.Item1)
                .Select(e => e.First())
                .ToDictionary(e => e.Item1, e => e.Item2);
        }

        /// <summary>
        /// Disabled
        /// </summary>
        public void ValidateFreeSpace()
        {
            return;
            // Disabled, caused more problems than it was worth.
            /* 
                DiskSpaceInfo DriveInfo(string path)
            {
                return Volume.GetDiskFreeSpace(Volume.GetUniqueVolumeNameForPath(path));
            }

            var paths = new[] {(OutputFolder, ModList.InstallSize),
                               (DownloadFolder, ModList.DownloadSize),
                               (Directory.GetCurrentDirectory(), ModList.ScratchSpaceSize)};
            paths.GroupBy(f => DriveInfo(f.Item1).DriveName)
                .Do(g =>
                {
                    var required = g.Sum(i => i.Item2);
                    var contains = g.Sum(folder =>
                        Directory.EnumerateFiles(folder.Item1, "*", DirectoryEnumerationOptions.Recursive)
                            .Sum(file => new FileInfo(file).Length));
                    var available = DriveInfo(g.Key).FreeBytesAvailable;
                    if (required - contains > available)
                        throw new NotEnoughDiskSpaceException(
                            $"This ModList requires {required.ToFileSizeString()} on {g.Key} but only {available.ToFileSizeString()} is available.");
                });
            */
        }


        /// <summary>
        /// The user may already have some files in the OutputFolder. If so we can go through these and
        /// figure out which need to be updated, deleted, or left alone
        /// </summary>
        public async Task OptimizeModlist()
        {
            Utils.Log("Optimizing ModList directives");
            
            // Clone the ModList so our changes don't modify the original data
            ModList = ModList.Clone();
            
            var indexed = ModList.Directives.ToDictionary(d => d.To);

            UpdateTracker.NextStep("Looking for files to delete");
            await OutputFolder.EnumerateFiles()
                .PMap(Queue, UpdateTracker, f =>
                {
                    var relativeTo = f.RelativeTo(OutputFolder);
                    Utils.Status($"Checking if ModList file {relativeTo}");
                    if (indexed.ContainsKey(relativeTo) || f.InFolder(DownloadFolder))
                        return;

                    Utils.Log($"Deleting {relativeTo} it's not part of this ModList");
                    f.Delete();
                });

            Utils.Log("Cleaning empty folders");
            var expectedFolders = indexed.Keys
                .Select(f => f.RelativeTo(OutputFolder))
                // We ignore the last part of the path, so we need a dummy file name
                .Append(DownloadFolder.Combine("_"))
                .Where(f => f.InFolder(OutputFolder))
                .SelectMany(path =>
                {
                    // Get all the folders and all the folder parents
                    // so for foo\bar\baz\qux.txt this emits ["foo", "foo\\bar", "foo\\bar\\baz"]
                    var split = ((string)path.RelativeTo(OutputFolder)).Split('\\');
                    return Enumerable.Range(1, split.Length - 1).Select(t => string.Join("\\", split.Take(t)));
                })
               .Distinct()
                .Select(p => OutputFolder.Combine(p))
               .ToHashSet();

            try
            {
                var toDelete = OutputFolder.EnumerateDirectories(true)
                    .Where(p => !expectedFolders.Contains(p))
                    .OrderByDescending(p => ((string)p).Length)
                    .ToList();
                foreach (var dir in toDelete)
                {
                    await dir.DeleteDirectory();
                }
            }
            catch (Exception)
            {
                // ignored because it's not worth throwing a fit over
                Utils.Log("Error when trying to clean empty folders. This doesn't really matter.");
            }

            UpdateTracker.NextStep("Looking for unmodified files");
            (await indexed.Values.PMap(Queue, UpdateTracker, async d =>
            {
                // Bit backwards, but we want to return null for 
                // all files we *want* installed. We return the files
                // to remove from the install list.
                Status($"Optimizing {d.To}");
                var path = OutputFolder.Combine(d.To);
                if (!path.Exists) return null;

                if (path.Size != d.Size) return null;
                
                return await path.FileHashAsync() == d.Hash ? d : null;
            }))
              .Where(d => d != null)
              .Do(d => indexed.Remove(d.To));

            UpdateTracker.NextStep("Updating ModList");
            Utils.Log($"Optimized {ModList.Directives.Count} directives to {indexed.Count} required");
            var requiredArchives = indexed.Values.OfType<FromArchive>()
                .GroupBy(d => d.ArchiveHashPath.BaseHash)
                .Select(d => d.Key)
                .ToHashSet();
            
            ModList.Archives = ModList.Archives.Where(a => requiredArchives.Contains(a.Hash)).ToList();
            ModList.Directives = indexed.Values.ToList();

        }
    }

    public class NotEnoughDiskSpaceException : Exception
    {
        public NotEnoughDiskSpaceException(string s) : base(s)
        {
        }
    }
}