tiny.smg.churn

1const std = @import("std");2const git = @import("git.zig");3const graph_mod = @import("graph.zig");4const view = @import("view.zig");5const sys = @import("sys");6const text = @import("text/root.zig");78pub const Entity = struct {9    name: []const u8,10    touches: usize,11};1213pub const File = struct {14    path: []const u8,15    touches: usize,16};1718pub const Result = struct {19    total_commits: usize,20    time_range: []const u8,21    entities: []const Entity,22    files: []const File,23};2425const Range = struct {26    file: []const u8,27    start: i64,28    end: i64,29    name: []const u8,30};3132const Hunk = struct {33    commit: []const u8,34    file: []const u8,35    start: i64,36    end: i64,37};3839const Touch = struct {40    commit: []const u8,41    name: []const u8,42};4344pub fn compute(allocator: std.mem.Allocator, graph: graph_mod.Graph, root: []const u8, days: i64) !Result {45    const time_range = try std.fmt.allocPrint(allocator, "last {d} days", .{days});46    const since = try std.fmt.allocPrint(allocator, "{d}.days.ago", .{days});47    const output = try git.run(allocator, root, &.{ "git", "log", "--unified=0", "--no-color", "-p", "--since", since }, 16 * 1024 * 1024);48    if (!output.ok) return .{ .total_commits = 0, .time_range = time_range, .entities = &.{}, .files = &.{} };49    const ranges = try buildFileIndex(allocator, graph);50    const hunks = try parseUnifiedDiff(allocator, output.stdout);51    var commits: std.ArrayList([]const u8) = .empty;52    var files: std.ArrayList(File) = .empty;53    var touches: std.ArrayList(Touch) = .empty;54    for (hunks) |hunk| {55        try appendUniqueString(allocator, &commits, hunk.commit);56        try addFileTouch(allocator, &files, hunk.file);57        for (ranges) |range| {58            if (!std.mem.eql(u8, range.file, hunk.file)) continue;59            if (range.start <= hunk.end and hunk.start <= range.end) try appendUniqueTouch(allocator, &touches, .{ .commit = hunk.commit, .name = range.name });60        }61    }62    var entities: std.ArrayList(Entity) = .empty;63    for (touches.items) |touch| try addEntityTouch(allocator, &entities, touch.name);64    std.mem.sort(Entity, entities.items, {}, entityLess);65    std.mem.sort(File, files.items, {}, fileLess);66    return .{67        .total_commits = commits.items.len,68        .time_range = time_range,69        .entities = try entities.toOwnedSlice(allocator),70        .files = try files.toOwnedSlice(allocator),71    };72}7374pub fn empty(allocator: std.mem.Allocator, days: i64) !Result {75    return .{76        .total_commits = 0,77        .time_range = try std.fmt.allocPrint(allocator, "last {d} days", .{days}),78        .entities = &.{},79        .files = &.{},80    };81}8283fn buildFileIndex(allocator: std.mem.Allocator, graph: graph_mod.Graph) ![]const Range {84    var out: std.ArrayList(Range) = .empty;85    for (try view.allNodes(graph, allocator, null)) |node| {86        const file = node.file orelse continue;87        const line = node.line orelse continue;88        const end = node.end_line orelse line;89        try out.append(allocator, .{ .file = file, .start = line, .end = end, .name = node.name });90    }91    std.mem.sort(Range, out.items, {}, rangeLess);92    return try out.toOwnedSlice(allocator);93}9495fn parseUnifiedDiff(allocator: std.mem.Allocator, output: []const u8) ![]const Hunk {96    var hunks: std.ArrayList(Hunk) = .empty;97    var current_commit: []const u8 = "";98    var current_file: []const u8 = "";99    var lines = std.mem.splitScalar(u8, output, '\n');100    while (lines.next()) |line| {101        if (std.mem.startsWith(u8, line, "commit ")) {102            const raw = text.trim(line[7..]);103            current_commit = try allocator.dupe(u8, raw[0..@min(raw.len, 12)]);104        } else if (std.mem.startsWith(u8, line, "+++ b/")) {105            current_file = try allocator.dupe(u8, line[6..]);106        } else if (std.mem.startsWith(u8, line, "@@ ") and current_commit.len != 0 and current_file.len != 0) {107            var parts = std.mem.splitScalar(u8, line, ' ');108            _ = parts.next();109            _ = parts.next();110            const new_range = parts.next() orelse continue;111            if (new_range.len == 0 or new_range[0] != '+') continue;112            const parsed = try parseRange(new_range[1..]);113            if (parsed.start <= 0) continue;114            try hunks.append(allocator, .{115                .commit = current_commit,116                .file = current_file,117                .start = parsed.start,118                .end = parsed.start + @max(parsed.count - 1, 0),119            });120        }121    }122    return try hunks.toOwnedSlice(allocator);123}124125const ParsedRange = struct {126    start: i64,127    count: i64,128};129130fn parseRange(value: []const u8) !ParsedRange {131    if (std.mem.indexOfScalar(u8, value, ',')) |index| {132        return .{133            .start = try std.fmt.parseInt(i64, value[0..index], 10),134            .count = try std.fmt.parseInt(i64, value[index + 1 ..], 10),135        };136    }137    return .{ .start = try std.fmt.parseInt(i64, value, 10), .count = 1 };138}139140fn appendUniqueString(allocator: std.mem.Allocator, list: *std.ArrayList([]const u8), value: []const u8) !void {141    for (list.items) |item| if (std.mem.eql(u8, item, value)) return;142    try list.append(allocator, value);143}144145fn appendUniqueTouch(allocator: std.mem.Allocator, list: *std.ArrayList(Touch), value: Touch) !void {146    for (list.items) |item| {147        if (std.mem.eql(u8, item.commit, value.commit) and std.mem.eql(u8, item.name, value.name)) return;148    }149    try list.append(allocator, value);150}151152fn addFileTouch(allocator: std.mem.Allocator, list: *std.ArrayList(File), path: []const u8) !void {153    for (list.items) |*item| {154        if (std.mem.eql(u8, item.path, path)) {155            item.touches += 1;156            return;157        }158    }159    try list.append(allocator, .{ .path = path, .touches = 1 });160}161162fn addEntityTouch(allocator: std.mem.Allocator, list: *std.ArrayList(Entity), name: []const u8) !void {163    for (list.items) |*item| {164        if (std.mem.eql(u8, item.name, name)) {165            item.touches += 1;166            return;167        }168    }169    try list.append(allocator, .{ .name = name, .touches = 1 });170}171172fn rangeLess(_: void, a: Range, b: Range) bool {173    const file = std.mem.order(u8, a.file, b.file);174    if (file != .eq) return file == .lt;175    if (a.start != b.start) return a.start < b.start;176    return std.mem.lessThan(u8, a.name, b.name);177}178179fn entityLess(_: void, a: Entity, b: Entity) bool {180    if (a.touches != b.touches) return a.touches > b.touches;181    return std.mem.lessThan(u8, a.name, b.name);182}183184fn fileLess(_: void, a: File, b: File) bool {185    if (a.touches != b.touches) return a.touches > b.touches;186    return std.mem.lessThan(u8, a.path, b.path);187}188189test "parse ranges like python churn parser" {190    try std.testing.expectEqual(ParsedRange{ .start = 10, .count = 5 }, try parseRange("10,5"));191    try std.testing.expectEqual(ParsedRange{ .start = 42, .count = 1 }, try parseRange("42"));192}193194test "parse unified diff hunks like python churn parser" {195    var arena = std.heap.ArenaAllocator.init(std.testing.allocator);196    defer arena.deinit();197    const allocator = arena.allocator();198    const hunks = try parseUnifiedDiff(allocator,199        \\commit aaa111200        \\diff --git a/a.py b/a.py201        \\--- a/a.py202        \\+++ b/a.py203        \\@@ -1,2 +1,3 @@ something204        \\@@ -20,1 +21,1 @@ other205        \\commit bbb222206        \\diff --git a/b.py b/b.py207        \\--- a/b.py208        \\+++ b/b.py209        \\@@ -5,0 +5,2 @@ thing210        \\211    );212    try std.testing.expectEqual(@as(usize, 3), hunks.len);213    try std.testing.expectEqualStrings("aaa111", hunks[0].commit);214    try std.testing.expectEqualStrings("a.py", hunks[0].file);215    try std.testing.expectEqual(@as(i64, 1), hunks[0].start);216    try std.testing.expectEqual(@as(i64, 3), hunks[0].end);217    try std.testing.expectEqualStrings("aaa111", hunks[1].commit);218    try std.testing.expectEqualStrings("a.py", hunks[1].file);219    try std.testing.expectEqualStrings("bbb222", hunks[2].commit);220    try std.testing.expectEqualStrings("b.py", hunks[2].file);221}222223test "churn deduplicates entity hunks per commit" {224    var arena = std.heap.ArenaAllocator.init(std.testing.allocator);225    defer arena.deinit();226    const allocator = arena.allocator();227    var graph = graph_mod.init(allocator);228    try graph_mod.addNode(&graph, .{ .name = "mod.foo", .type = "function", .file = "a.py", .line = 1, .end_line = 15 });229    const ranges = try buildFileIndex(allocator, graph);230    const hunks = try parseUnifiedDiff(allocator,231        \\commit aaa111232        \\diff --git a/a.py b/a.py233        \\--- a/a.py234        \\+++ b/a.py235        \\@@ -2,1 +2,2 @@ def foo():236        \\@@ -8,1 +9,2 @@ def foo():237        \\commit bbb222238        \\diff --git a/a.py b/a.py239        \\--- a/a.py240        \\+++ b/a.py241        \\@@ -5,1 +6,2 @@ def foo():242        \\243    );244    var touches: std.ArrayList(Touch) = .empty;245    for (hunks) |hunk| {246        for (ranges) |range| {247            if (std.mem.eql(u8, range.file, hunk.file) and range.start <= hunk.end and hunk.start <= range.end) {248                try appendUniqueTouch(allocator, &touches, .{ .commit = hunk.commit, .name = range.name });249            }250        }251    }252    var entities: std.ArrayList(Entity) = .empty;253    for (touches.items) |touch| try addEntityTouch(allocator, &entities, touch.name);254    try std.testing.expectEqual(@as(usize, 1), entities.items.len);255    try std.testing.expectEqualStrings("mod.foo", entities.items[0].name);256    try std.testing.expectEqual(@as(usize, 2), entities.items[0].touches);257}258259test "compute maps git hunks to graph entities" {260    var arena = std.heap.ArenaAllocator.init(std.testing.allocator);261    defer arena.deinit();262    const allocator = arena.allocator();263    const root = try std.fmt.allocPrint(allocator, "/tmp/smg-churn-test-{x}-{x}", .{ @as(u64, @intCast(@max(0, sys.time.realMilliTimestamp()))), @intFromPtr(&arena) });264    defer sys.fs.deleteTree(root) catch {};265    try sys.fs.createDirPath(try std.fs.path.join(allocator, &.{ root, "src" }));266    try expectGit(allocator, root, &.{ "git", "init", "-q" });267    try expectGit(allocator, root, &.{ "git", "config", "user.email", "test@example.com" });268    try expectGit(allocator, root, &.{ "git", "config", "user.name", "Test User" });269    const path = try std.fs.path.join(allocator, &.{ root, "src", "app.py" });270    try text.writeFile(path,271        \\def hot():272        \\    value = 1273        \\    return value274        \\275        \\def cold():276        \\    return 0277        \\278    );279    try expectGit(allocator, root, &.{ "git", "add", "src/app.py" });280    try expectGit(allocator, root, &.{ "git", "commit", "-q", "-m", "initial" });281    try text.writeFile(path,282        \\def hot():283        \\    value = 2284        \\    extra = value + 1285        \\    return extra286        \\287        \\def cold():288        \\    return 0289        \\290    );291    try expectGit(allocator, root, &.{ "git", "add", "src/app.py" });292    try expectGit(allocator, root, &.{ "git", "commit", "-q", "-m", "touch hot" });293    var graph = graph_mod.init(allocator);294    try graph_mod.addNode(&graph, .{ .name = "src.app.hot", .type = "function", .file = "src/app.py", .line = 1, .end_line = 4 });295    try graph_mod.addNode(&graph, .{ .name = "src.app.cold", .type = "function", .file = "src/app.py", .line = 6, .end_line = 7 });296    const result = try compute(allocator, graph, root, 90);297    try std.testing.expectEqual(@as(usize, 2), result.total_commits);298    try std.testing.expectEqualStrings("last 90 days", result.time_range);299    try std.testing.expectEqual(@as(usize, 2), result.entities.len);300    try std.testing.expectEqualStrings("src.app.hot", result.entities[0].name);301    try std.testing.expectEqual(@as(usize, 2), result.entities[0].touches);302    try std.testing.expectEqualStrings("src.app.cold", result.entities[1].name);303    try std.testing.expectEqual(@as(usize, 1), result.entities[1].touches);304    try std.testing.expectEqual(@as(usize, 1), result.files.len);305    try std.testing.expectEqualStrings("src/app.py", result.files[0].path);306    try std.testing.expectEqual(@as(usize, 2), result.files[0].touches);307}308309fn expectGit(allocator: std.mem.Allocator, root: []const u8, argv: []const []const u8) !void {310    const output = try git.run(allocator, root, argv, 1024 * 1024);311    try std.testing.expect(output.ok);312}

1pub const churn = @import("churn.zig");