docs/superpowers/plans/2026-04-19-imgdiff-extraction.md
Ref: Size: 18.1 KiB
# Imgdiff Extraction Implementation Plan
> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
**Goal:** Extract the pure RMSE + max-pixel + diff-heatmap math from `src/tools/imgdiff.zig` into a new reusable module `src/imgdiff.zig`, so the scenario runner (later plans) can call the same code as the existing CLI.
**Architecture:** Pure-math module with `compare`, `makeDiffImage`, `DiffResult`, and the two default threshold constants (`RMSE_DEFAULT`, `PIXEL_MAX_DEFAULT`). The existing CLI at `src/tools/imgdiff.zig` shrinks to a thin `main` + env-var parsing that imports the new module. Zero behavior change for the CLI.
**Tech Stack:** Zig 0.15, existing `png` module.
**Reference spec:** `docs/superpowers/specs/2026-04-19-scenario-runner-design.md` — implementation phasing step 1 of 3.
---
## Reference facts (grep once, reuse)
- Existing CLI + math lives in `src/tools/imgdiff.zig` (152 lines). The pure math is the `compare` function + `DiffResult` struct + `makeDiffImage` helper. The CLI-side logic is `main`, `readFloatEnv`, env-var defaults (`RMSE` default `0.005`, `PIXEL_MAX` default `0.125`), argument parsing, PNG encode/decode, and stderr reporting.
- Build wiring in `build.zig:368-388` creates `imgdiff_mod` (executable) and `imgdiff_test_mod` (tests). Both use `src/tools/imgdiff.zig` as `root_source_file`.
- `png.Image` struct is `{ width: u32, height: u32, pixels: []u8 }` (BGRA/RGBA 4 bytes-per-pixel).
- This plan has no Vulkan, no Wayland, no main-loop touchpoints. Pure refactor.
- Zig's `zig build test` runs all test modules registered on `test_step`. The existing `imgdiff_tests` contributes two tests; this plan adds one more to the new module.
---
## File structure after this plan
- **`src/imgdiff.zig`** (NEW) — pure module. Public: `DiffResult`, `compare(a, b)`, `makeDiffImage(alloc, a, b)`, `RMSE_DEFAULT`, `PIXEL_MAX_DEFAULT`. Inline unit tests stay with the code.
- **`src/tools/imgdiff.zig`** (MODIFIED, shrunk) — CLI only. `main`, `readFloatEnv`. Imports `imgdiff` module for the math + constants.
- **`build.zig`** (MODIFIED) — add a new `imgdiff_mod` module (name collision with the existing variable; we'll rename — see task 2); wire it as an import on both the CLI exe and its test module.
---
## Task 1: Create `src/imgdiff.zig` as a pure module
**Files:**
- Create: `src/imgdiff.zig`
- Modify: `build.zig:368-388` — register the new module and its tests
**Goal of this task:** Stand up the new module alongside the existing CLI. The CLI keeps working unchanged. The new module has its own tests and runs green.
- [ ] **Step 1: Create `src/imgdiff.zig` with the pure math + constants + three tests.**
```zig
//! Pure-math PNG image diff used by both the `imgdiff` CLI
//! (src/tools/imgdiff.zig) and the scenario runner (to be added later).
//! Threshold constants live here so both callers agree.
const std = @import("std");
const png = @import("png");
pub const RMSE_DEFAULT: f64 = 0.005;
pub const PIXEL_MAX_DEFAULT: f64 = 0.125;
pub const DiffResult = struct {
rmse: f64, // [0, 1]
max_pixel: f64, // [0, 1]
pixel_count: usize,
};
pub fn compare(a: png.Image, b: png.Image) !DiffResult {
if (a.width != b.width or a.height != b.height) return error.DimensionsDiffer;
std.debug.assert(a.pixels.len == b.pixels.len);
const px_count = @as(usize, a.width) * a.height;
var sum_sq: f64 = 0;
var max_d: f64 = 0;
var i: usize = 0;
while (i < px_count) : (i += 1) {
const off = i * 4;
const dr = (@as(f64, @floatFromInt(a.pixels[off + 0])) - @as(f64, @floatFromInt(b.pixels[off + 0]))) / 255.0;
const dg = (@as(f64, @floatFromInt(a.pixels[off + 1])) - @as(f64, @floatFromInt(b.pixels[off + 1]))) / 255.0;
const db = (@as(f64, @floatFromInt(a.pixels[off + 2])) - @as(f64, @floatFromInt(b.pixels[off + 2]))) / 255.0;
const d_sq = (dr * dr + dg * dg + db * db) / 3.0;
sum_sq += d_sq;
const d = @sqrt(d_sq);
if (d > max_d) max_d = d;
}
return .{
.rmse = @sqrt(sum_sq / @as(f64, @floatFromInt(px_count))),
.max_pixel = max_d,
.pixel_count = px_count,
};
}
pub fn makeDiffImage(alloc: std.mem.Allocator, a: png.Image, b: png.Image) !png.Image {
// Side-by-side: [actual | reference | delta-heatmap]
const w = a.width * 3;
const h = a.height;
const pixels = try alloc.alloc(u8, w * h * 4);
var y: u32 = 0;
while (y < h) : (y += 1) {
const row_off = @as(usize, y) * w * 4;
const a_off = @as(usize, y) * a.width * 4;
@memcpy(pixels[row_off .. row_off + a.width * 4], a.pixels[a_off .. a_off + a.width * 4]);
@memcpy(pixels[row_off + a.width * 4 .. row_off + 2 * a.width * 4], b.pixels[a_off .. a_off + a.width * 4]);
var x: u32 = 0;
while (x < a.width) : (x += 1) {
const off = a_off + x * 4;
const dr = (@as(f64, @floatFromInt(a.pixels[off + 0])) - @as(f64, @floatFromInt(b.pixels[off + 0]))) / 255.0;
const dg = (@as(f64, @floatFromInt(a.pixels[off + 1])) - @as(f64, @floatFromInt(b.pixels[off + 1]))) / 255.0;
const db = (@as(f64, @floatFromInt(a.pixels[off + 2])) - @as(f64, @floatFromInt(b.pixels[off + 2]))) / 255.0;
const d = @sqrt((dr * dr + dg * dg + db * db) / 3.0);
const brightness: u8 = @intFromFloat(@min(255.0, d * 255.0 * 2.0));
const dst = row_off + 2 * a.width * 4 + x * 4;
pixels[dst + 0] = brightness;
pixels[dst + 1] = brightness;
pixels[dst + 2] = brightness;
pixels[dst + 3] = 255;
}
}
return .{ .width = w, .height = h, .pixels = pixels };
}
test "identical images produce zero RMSE" {
var pixels_a = [_]u8{ 10, 20, 30, 255, 40, 50, 60, 255 };
var pixels_b = [_]u8{ 10, 20, 30, 255, 40, 50, 60, 255 };
const a = png.Image{ .width = 2, .height = 1, .pixels = &pixels_a };
const b = png.Image{ .width = 2, .height = 1, .pixels = &pixels_b };
const r = try compare(a, b);
try std.testing.expectEqual(@as(f64, 0.0), r.rmse);
try std.testing.expectEqual(@as(f64, 0.0), r.max_pixel);
}
test "fully saturated difference produces rmse=1.0 and max=1.0" {
var pixels_a = [_]u8{ 0, 0, 0, 255 };
var pixels_b = [_]u8{ 255, 255, 255, 255 };
const a = png.Image{ .width = 1, .height = 1, .pixels = &pixels_a };
const b = png.Image{ .width = 1, .height = 1, .pixels = &pixels_b };
const r = try compare(a, b);
try std.testing.expectApproxEqAbs(@as(f64, 1.0), r.rmse, 1e-9);
try std.testing.expectApproxEqAbs(@as(f64, 1.0), r.max_pixel, 1e-9);
}
test "makeDiffImage: width triples, heatmap brighter where delta is larger" {
const alloc = std.testing.allocator;
// 2x1 image pair where one column differs more than the other.
var pixels_a = [_]u8{ 0, 0, 0, 255, 100, 100, 100, 255 };
var pixels_b = [_]u8{ 0, 0, 0, 255, 200, 200, 200, 255 };
const a = png.Image{ .width = 2, .height = 1, .pixels = &pixels_a };
const b = png.Image{ .width = 2, .height = 1, .pixels = &pixels_b };
const out = try makeDiffImage(alloc, a, b);
defer alloc.free(out.pixels);
try std.testing.expectEqual(@as(u32, 6), out.width);
try std.testing.expectEqual(@as(u32, 1), out.height);
// Heatmap is the last third. Column 0 (no delta) should be black;
// column 1 (delta) should be non-zero.
const heatmap_offset = 4 * 4; // bytes before heatmap (2 images × 2 pixels × 4 bytes = 16)
try std.testing.expectEqual(@as(u8, 0), out.pixels[heatmap_offset + 0]); // R at heatmap col 0
try std.testing.expect(out.pixels[heatmap_offset + 4] > 0); // R at heatmap col 1
}
```
- [ ] **Step 2: Wire `src/imgdiff.zig` as a module in `build.zig`.**
Current state of `build.zig:368-388`:
```zig
// imgdiff — standalone PNG comparison tool
const imgdiff_mod = b.createModule(.{
.root_source_file = b.path("src/tools/imgdiff.zig"),
.target = target,
.optimize = optimize,
});
imgdiff_mod.addImport("png", png_mod);
const imgdiff_exe = b.addExecutable(.{
.name = "imgdiff",
.root_module = imgdiff_mod,
});
b.installArtifact(imgdiff_exe);
const imgdiff_test_mod = b.createModule(.{
.root_source_file = b.path("src/tools/imgdiff.zig"),
.target = target,
.optimize = optimize,
});
imgdiff_test_mod.addImport("png", png_mod);
const imgdiff_tests = b.addTest(.{ .root_module = imgdiff_test_mod });
test_step.dependOn(&b.addRunArtifact(imgdiff_tests).step);
```
Replace with:
```zig
// imgdiff — pure library used by the CLI + the scenario runner
const imgdiff_lib_mod = b.createModule(.{
.root_source_file = b.path("src/imgdiff.zig"),
.target = target,
.optimize = optimize,
});
imgdiff_lib_mod.addImport("png", png_mod);
const imgdiff_lib_test_mod = b.createModule(.{
.root_source_file = b.path("src/imgdiff.zig"),
.target = target,
.optimize = optimize,
});
imgdiff_lib_test_mod.addImport("png", png_mod);
const imgdiff_lib_tests = b.addTest(.{ .root_module = imgdiff_lib_test_mod });
test_step.dependOn(&b.addRunArtifact(imgdiff_lib_tests).step);
// imgdiff — standalone PNG comparison CLI
const imgdiff_mod = b.createModule(.{
.root_source_file = b.path("src/tools/imgdiff.zig"),
.target = target,
.optimize = optimize,
});
imgdiff_mod.addImport("png", png_mod);
imgdiff_mod.addImport("imgdiff", imgdiff_lib_mod);
const imgdiff_exe = b.addExecutable(.{
.name = "imgdiff",
.root_module = imgdiff_mod,
});
b.installArtifact(imgdiff_exe);
const imgdiff_test_mod = b.createModule(.{
.root_source_file = b.path("src/tools/imgdiff.zig"),
.target = target,
.optimize = optimize,
});
imgdiff_test_mod.addImport("png", png_mod);
imgdiff_test_mod.addImport("imgdiff", imgdiff_lib_mod);
const imgdiff_tests = b.addTest(.{ .root_module = imgdiff_test_mod });
test_step.dependOn(&b.addRunArtifact(imgdiff_tests).step);
```
The library module (`imgdiff_lib_mod`) is a separate build unit from the CLI module (`imgdiff_mod`). Both reach `png`; the CLI additionally imports the library. Both have their tests registered.
At this point the CLI still has its own copy of `compare` / `makeDiffImage` / tests — Task 2 removes them. This task only *adds* the new module. Intermediate state: duplication between `src/imgdiff.zig` and `src/tools/imgdiff.zig`. That's fine for one commit; Task 2 removes the duplication.
- [ ] **Step 3: Build and run tests to verify the new module compiles and its tests pass.**
Run:
```bash
cd /home/xanderle/code/rad/waystty
zig build test 2>&1 | tail -30
```
Expected:
- Build succeeds.
- All existing tests still pass.
- Three new tests (from `src/imgdiff.zig`) show up in the count.
If the test runner reports a test-count mismatch relative to before, that's the expected addition of three new tests. If there are *fewer* tests, the old `src/tools/imgdiff.zig` tests didn't get picked up — investigate before committing.
- [ ] **Step 4: Commit.**
```bash
git add src/imgdiff.zig build.zig
git commit -m "$(cat <<'EOF'
imgdiff: add src/imgdiff.zig pure-math module
Copies compare + DiffResult + makeDiffImage into a standalone
module so it can be reused by the upcoming scenario runner.
CLI at src/tools/imgdiff.zig is untouched this commit; dedup
happens in the next commit.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
EOF
)"
```
---
## Task 2: Shrink `src/tools/imgdiff.zig` to a thin CLI
**Files:**
- Modify: `src/tools/imgdiff.zig` — remove `DiffResult`, `compare`, `makeDiffImage`, inline tests, replace with `@import("imgdiff")` usage.
- [ ] **Step 1: Replace `src/tools/imgdiff.zig` with a CLI-only version.**
Full new contents of `src/tools/imgdiff.zig`:
```zig
const std = @import("std");
const png = @import("png");
const imgdiff = @import("imgdiff");
pub fn main() !void {
var gpa: std.heap.DebugAllocator(.{}) = .init;
defer _ = gpa.deinit();
const alloc = gpa.allocator();
const args = try std.process.argsAlloc(alloc);
defer std.process.argsFree(alloc, args);
if (args.len < 3) {
std.debug.print("usage: imgdiff <actual.png> <reference.png> [diff.png]\n", .{});
std.process.exit(2);
}
const actual_path = args[1];
const reference_path = args[2];
const diff_path: ?[]const u8 = if (args.len >= 4) args[3] else null;
const rmse_max = readFloatEnv("WAYSTTY_TEST_RMSE_MAX", imgdiff.RMSE_DEFAULT);
const pixel_max = readFloatEnv("WAYSTTY_TEST_PIXEL_MAX", imgdiff.PIXEL_MAX_DEFAULT);
const actual_bytes = try std.fs.cwd().readFileAlloc(alloc, actual_path, 64 * 1024 * 1024);
defer alloc.free(actual_bytes);
const reference_bytes = try std.fs.cwd().readFileAlloc(alloc, reference_path, 64 * 1024 * 1024);
defer alloc.free(reference_bytes);
var actual = try png.decode(alloc, actual_bytes);
defer actual.deinit(alloc);
var reference = try png.decode(alloc, reference_bytes);
defer reference.deinit(alloc);
if (actual.width != reference.width or actual.height != reference.height) {
std.debug.print("FAIL: dimensions differ ({}x{} vs {}x{})\n", .{ actual.width, actual.height, reference.width, reference.height });
std.process.exit(3);
}
const r = try imgdiff.compare(actual, reference);
const pass = r.rmse <= rmse_max and r.max_pixel <= pixel_max;
if (pass) {
std.debug.print("OK: {s} RMSE={d:.4}% worst={d:.4}%\n", .{ reference_path, r.rmse * 100.0, r.max_pixel * 100.0 });
std.process.exit(0);
}
std.debug.print("FAIL: {s}\n RMSE: {d:.4}% (max {d:.4}%)\n worst pixel: {d:.4}% (max {d:.4}%)\n", .{ reference_path, r.rmse * 100.0, rmse_max * 100.0, r.max_pixel * 100.0, pixel_max * 100.0 });
if (diff_path) |p| {
const diff_img = try imgdiff.makeDiffImage(alloc, actual, reference);
defer alloc.free(diff_img.pixels);
var buf: std.ArrayList(u8) = .empty;
defer buf.deinit(alloc);
try png.encode(alloc, diff_img, buf.writer(alloc));
const out = try std.fs.cwd().createFile(p, .{ .truncate = true });
defer out.close();
try out.writeAll(buf.items);
std.debug.print(" diff: {s}\n", .{p});
}
std.debug.print(" actual: {s}\n", .{actual_path});
std.process.exit(1);
}
fn readFloatEnv(name: []const u8, default: f64) f64 {
const val = std.posix.getenv(name) orelse return default;
return std.fmt.parseFloat(f64, val) catch default;
}
```
What's gone from the old file: `pub const DiffResult`, `pub fn compare`, `fn makeDiffImage`, both `test "identical images ..."` and `test "fully saturated ..."` blocks. What's new: the `imgdiff` import and calls to `imgdiff.compare` / `imgdiff.makeDiffImage` / `imgdiff.RMSE_DEFAULT` / `imgdiff.PIXEL_MAX_DEFAULT`.
- [ ] **Step 2: Build everything.**
Run:
```bash
cd /home/xanderle/code/rad/waystty
zig build 2>&1 | tail -20
```
Expected: build succeeds, `imgdiff` binary still links.
- [ ] **Step 3: Run unit tests.**
Run:
```bash
zig build test 2>&1 | tail -30
```
Expected: all tests pass. Note: the `imgdiff_tests` test module (the CLI's test module) now has zero tests inside it because we moved them all — that's fine; it just means the test runner reports zero tests for that module while the new `imgdiff_lib_tests` reports three. The total count should be **unchanged minus 2 plus 3 = net +1** relative to before Task 1.
- [ ] **Step 4: Smoke test the CLI against an existing golden.**
Run:
```bash
cd /home/xanderle/code/rad/waystty
zig build
./zig-out/bin/imgdiff tests/golden/reference/basic_ascii.png tests/golden/reference/basic_ascii.png
```
Expected: `OK: tests/golden/reference/basic_ascii.png RMSE=0.0000% worst=0.0000%`, exit code 0. Identical image to itself → zero diff.
Then a failing-diff smoke:
```bash
./zig-out/bin/imgdiff tests/golden/reference/basic_ascii.png tests/golden/reference/bold_colors.png || echo "exit=$?"
```
Expected: `FAIL: ...` output, exit code 1 (or 3 if dimensions differ — `basic_ascii.png` and `bold_colors.png` are both 80×24 grids so dimensions match; exit 1 is the normal-diff fail).
If both these smoke cases behave the same as before the refactor, the CLI has not regressed.
- [ ] **Step 5: Run the full render-test suite (which uses imgdiff CLI).**
```bash
zig build test-render 2>&1 | tail -20
```
Expected: all golden scripts pass (`basic_ascii`, `bold_colors`, `box_drawing` at least, as of spec-writing time). If this step fails, the CLI refactor broke the existing regression gate — investigate before committing.
- [ ] **Step 6: Commit.**
```bash
git add src/tools/imgdiff.zig
git commit -m "$(cat <<'EOF'
imgdiff: shrink CLI to thin wrapper over src/imgdiff.zig
Removes duplicated pure-math (compare, makeDiffImage, DiffResult,
inline tests) from the CLI. CLI now imports the library module
for both the comparison and the default threshold constants.
No behavior change. test-render still passes against existing
golden PNGs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
EOF
)"
```
---
## Post-task verification
- [ ] `zig build` passes with no warnings.
- [ ] `zig build test` passes (three `imgdiff_lib_tests` + zero `imgdiff_tests` + everything else unchanged).
- [ ] `zig build test-render` still passes against existing goldens.
- [ ] `./zig-out/bin/imgdiff` CLI output format is byte-identical to pre-refactor (re-run the two smoke cases in Task 2 Step 4 if unsure).
- [ ] `src/imgdiff.zig` is importable by future code as `@import("imgdiff")` on any module that declares it — this is what plan 2 (scenario parser) will rely on.
---
## Self-review coverage check
Spec "Implementation phasing" step 1 says: move pure math into `src/imgdiff.zig`, keep CLI as thin wrapper. Task 1 adds the module and its tests; Task 2 shrinks the CLI. Post-task verification confirms no CLI behavior change. ✅ covered.
Spec "Files touched" lists `src/imgdiff.zig` as new and `src/tools/imgdiff.zig` as the thin wrapper. ✅ matches.
Spec doesn't require any changes outside imgdiff for plan 1 — plan 2 and plan 3 will consume `src/imgdiff.zig`.