| 1 | using System.Globalization; |
| 2 | using System.Text; |
| 3 | using HybridCodebaseIndex.Core.Indexing; |
| 4 | using Microsoft.Data.Sqlite; |
| 5 | |
| 6 | namespace HybridCodebaseIndex.Core; |
| 7 | |
| 8 | internal static partial class SqliteFtsIndex |
| 9 | { |
| 10 | private static ReindexSummary ReindexIncremental( |
| 11 | string workspaceRoot, |
| 12 | string dbPath, |
| 13 | IReadOnlyList<ICodebaseIndexReindexObserver>? reindexObservers, |
| 14 | CancellationToken cancellationToken) |
| 15 | { |
| 16 | var sw = System.Diagnostics.Stopwatch.StartNew(); |
| 17 | workspaceRoot = Path.GetFullPath(workspaceRoot.TrimEnd(Path.DirectorySeparatorChar)); |
| 18 | |
| 19 | var filesIndexed = 0; |
| 20 | var skippedLarge = 0; |
| 21 | var skippedBinary = 0; |
| 22 | var skippedExcluded = 0; |
| 23 | var skippedSample = new List<SkippedPath>(capacity: 64); |
| 24 | var skippedReasonCounts = new Dictionary<string, int>(StringComparer.OrdinalIgnoreCase); |
| 25 | var skippedTopPathPrefixes = new Dictionary<string, int>(StringComparer.OrdinalIgnoreCase); |
| 26 | |
| 27 | using var conn = new SqliteConnection($"Data Source={dbPath};Mode=ReadWriteCreate"); |
| 28 | conn.Open(); |
| 29 | |
| 30 | EnsureMetaTable(conn); |
| 31 | EnsureChunksTable(conn); |
| 32 | EnsureFileStateTable(conn); |
| 33 | |
| 34 | try |
| 35 | { |
| 36 | UpsertMeta(conn, "reindex_state", "running"); |
| 37 | UpsertMeta(conn, "reindex_started_at", DateTime.UtcNow.ToString("O", CultureInfo.InvariantCulture)); |
| 38 | |
| 39 | var settings = IndexSettings.TryLoadFromIndexDirectory(Path.GetDirectoryName(dbPath)!); |
| 40 | var extensions = settings.GetEffectiveExtensions(); |
| 41 | var maxBytes = settings.GetEffectiveMaxIndexedFileBytes(); |
| 42 | var chunkLines = settings.GetEffectiveChunkLines(); |
| 43 | var overlapLines = settings.GetEffectiveChunkOverlapLines(); |
| 44 | var probeBytes = settings.GetEffectiveBinaryProbeBytes(); |
| 45 | |
| 46 | var roots = new List<string>(capacity: 1 + settings.ExtraIncludeRoots.Count) { workspaceRoot }; |
| 47 | foreach (var extra in settings.ExtraIncludeRoots) |
| 48 | { |
| 49 | var p = Path.GetFullPath(Path.Combine(workspaceRoot, extra)); |
| 50 | if (Directory.Exists(p)) |
| 51 | roots.Add(p); |
| 52 | } |
| 53 | |
| 54 | var extSet = new HashSet<string>(extensions, StringComparer.OrdinalIgnoreCase); |
| 55 | |
| 56 | var excludeRootFullPaths = new List<string>(settings.ExcludeRoots.Count); |
| 57 | foreach (var rel in settings.ExcludeRoots) |
| 58 | { |
| 59 | var p = rel?.Trim(); |
| 60 | if (string.IsNullOrWhiteSpace(p)) |
| 61 | continue; |
| 62 | if (Path.IsPathRooted(p) || p.Contains("..", StringComparison.Ordinal)) |
| 63 | continue; |
| 64 | var abs = Path.GetFullPath(Path.Combine(workspaceRoot, p)); |
| 65 | if (Directory.Exists(abs)) |
| 66 | excludeRootFullPaths.Add(abs.TrimEnd(Path.DirectorySeparatorChar) + Path.DirectorySeparatorChar); |
| 67 | } |
| 68 | |
| 69 | var candidates = new List<string>(capacity: 8192); |
| 70 | foreach (var root in roots) |
| 71 | candidates.AddRange(WorkspaceScanner.EnumerateIndexableFiles(root, extSet, settings.ExcludePathSegments, excludeRootFullPaths)); |
| 72 | |
| 73 | var gitIgnore = GitIgnoreRules.TryLoad(workspaceRoot, settings.IgnoreFiles); |
| 74 | |
| 75 | var seen = new HashSet<string>(StringComparer.Ordinal); |
| 76 | |
| 77 | using var tx = conn.BeginTransaction(); |
| 78 | |
| 79 | foreach (var absolute in candidates) |
| 80 | { |
| 81 | cancellationToken.ThrowIfCancellationRequested(); |
| 82 | |
| 83 | if (WorkspaceScanner.ShouldExcludePath(absolute, settings.ExcludePathSegments)) |
| 84 | { |
| 85 | skippedExcluded++; |
| 86 | AddSkip(skippedSample, skippedReasonCounts, skippedTopPathPrefixes, WorkspaceScanner.RelativePath(workspaceRoot, absolute), "denylist"); |
| 87 | continue; |
| 88 | } |
| 89 | |
| 90 | var rel = WorkspaceScanner.RelativePath(workspaceRoot, absolute).Replace("\\", "/", StringComparison.Ordinal); |
| 91 | seen.Add(rel); |
| 92 | |
| 93 | if (GitIgnoreRules.IsIgnored(gitIgnore, rel)) |
| 94 | { |
| 95 | skippedExcluded++; |
| 96 | AddSkip(skippedSample, skippedReasonCounts, skippedTopPathPrefixes, rel, "gitignore"); |
| 97 | continue; |
| 98 | } |
| 99 | |
| 100 | FileInfo info; |
| 101 | try |
| 102 | { |
| 103 | info = new FileInfo(absolute); |
| 104 | if (info.Length > maxBytes) |
| 105 | { |
| 106 | skippedLarge++; |
| 107 | AddSkip(skippedSample, skippedReasonCounts, skippedTopPathPrefixes, rel, "too_large"); |
| 108 | continue; |
| 109 | } |
| 110 | } |
| 111 | catch |
| 112 | { |
| 113 | skippedExcluded++; |
| 114 | AddSkip(skippedSample, skippedReasonCounts, skippedTopPathPrefixes, rel, "io_error"); |
| 115 | continue; |
| 116 | } |
| 117 | |
| 118 | var lastWriteUtcTicks = info.LastWriteTimeUtc.Ticks; |
| 119 | if (!IsFileChanged(conn, tx, rel, info.Length, lastWriteUtcTicks)) |
| 120 | continue; |
| 121 | |
| 122 | using var fs = info.OpenRead(); |
| 123 | var probeSize = (int)Math.Min(probeBytes, info.Length); |
| 124 | var probe = new byte[probeSize]; |
| 125 | var read = fs.ReadAtLeast(probe.AsSpan(0, probeSize), probeSize, throwOnEndOfStream: false); |
| 126 | if (WorkspaceScanner.LooksBinary(probe.AsSpan(0, read))) |
| 127 | { |
| 128 | skippedBinary++; |
| 129 | AddSkip(skippedSample, skippedReasonCounts, skippedTopPathPrefixes, rel, "binary"); |
| 130 | continue; |
| 131 | } |
| 132 | |
| 133 | fs.Position = 0; |
| 134 | using var reader = new StreamReader(fs, Encoding.UTF8, detectEncodingFromByteOrderMarks: true); |
| 135 | var text = reader.ReadToEnd(); |
| 136 | |
| 137 | // Replace chunks for this path. |
| 138 | DeleteChunksForPath(conn, tx, rel); |
| 139 | |
| 140 | var ext = Path.GetExtension(absolute); |
| 141 | NotifyFileIndexed(reindexObservers, workspaceRoot, rel, absolute, ext, text, lastWriteUtcTicks); |
| 142 | var header = BuildArtifactAugmentationHeader(workspaceRoot, absolute, rel, ext, text); |
| 143 | var chunks = WorkspaceScanner.ChunkByLines(text, chunkLines, overlapLines); |
| 144 | var anyChunk = false; |
| 145 | foreach (var (lineStart, lineEnd, body) in chunks) |
| 146 | { |
| 147 | var augmentedBody = lineStart == 1 && header.Length > 0 ? header + body : body; |
| 148 | InsertChunk(conn, tx, rel, ext, lineStart, lineEnd, augmentedBody); |
| 149 | anyChunk = true; |
| 150 | } |
| 151 | |
| 152 | if (anyChunk) |
| 153 | filesIndexed++; |
| 154 | |
| 155 | UpsertFileState(conn, tx, rel, info.Length, lastWriteUtcTicks); |
| 156 | } |
| 157 | |
| 158 | // Remove deleted files. |
| 159 | foreach (var stale in EnumerateStalePaths(conn, tx, seen)) |
| 160 | { |
| 161 | DeleteChunksForPath(conn, tx, stale); |
| 162 | DeleteFileState(conn, tx, stale); |
| 163 | } |
| 164 | |
| 165 | tx.Commit(); |
| 166 | |
| 167 | UpsertMeta(conn, "format_version", FormatVersion.ToString(CultureInfo.InvariantCulture)); |
| 168 | UpsertMeta(conn, "indexed_at", DateTime.UtcNow.ToString("O", CultureInfo.InvariantCulture)); |
| 169 | UpsertMeta(conn, "workspace_root", workspaceRoot); |
| 170 | UpsertMeta(conn, "reindex_error", ""); |
| 171 | UpsertMeta(conn, "reindex_error_at", ""); |
| 172 | UpsertMeta(conn, "reindex_state", "idle"); |
| 173 | UpsertMeta(conn, "reindex_started_at", ""); |
| 174 | } |
| 175 | catch (Exception ex) |
| 176 | { |
| 177 | try |
| 178 | { |
| 179 | UpsertMeta(conn, "reindex_state", "error"); |
| 180 | UpsertMeta(conn, "reindex_error", ex.GetType().Name + ": " + ex.Message); |
| 181 | UpsertMeta(conn, "reindex_error_at", DateTime.UtcNow.ToString("O", CultureInfo.InvariantCulture)); |
| 182 | } |
| 183 | catch |
| 184 | { |
| 185 | // ignore |
| 186 | } |
| 187 | |
| 188 | throw; |
| 189 | } |
| 190 | |
| 191 | sw.Stop(); |
| 192 | return new ReindexSummary( |
| 193 | FormatVersion, |
| 194 | dbPath, |
| 195 | filesIndexed, |
| 196 | skippedLarge, |
| 197 | skippedBinary, |
| 198 | skippedExcluded, |
| 199 | skippedReasonCounts, |
| 200 | TopPrefixes(skippedTopPathPrefixes), |
| 201 | skippedSample, |
| 202 | sw.Elapsed); |
| 203 | } |
| 204 | } |
| 205 | |
| 206 | |