Initial commit

2026-06-29 15:11:48 -07:00
commit 66dfdcc58d
404 changed files with 45970 additions and 0 deletions
@@ -0,0 +1,107 @@
+#!/usr/bin/env python3
+"""
+diff.py — Pixel diff two PNGs, producing a highlighted diff image and stats.
+
+Uses the `pixelmatch` Python port (pip install pixelmatch pillow). Returns JSON to stdout:
+  { diff_pct, diff_image_path, worst_regions: [ { x, y, width, height } ] }
+
+Worst regions are computed via connected-component analysis on the diff mask, sorted by area desc.
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import sys
+from pathlib import Path
+
+try:
+    from PIL import Image
+    from pixelmatch.contrib.PIL import pixelmatch
+except ImportError:
+    print(json.dumps({"error": "Missing dependencies. pip install pixelmatch pillow numpy"}), file=sys.stderr)
+    sys.exit(1)
+
+try:
+    import numpy as np
+    from scipy import ndimage
+    HAVE_SCIPY = True
+except ImportError:
+    HAVE_SCIPY = False
+
+
+def resize_to_match(a: Image.Image, b: Image.Image) -> tuple[Image.Image, Image.Image]:
+    if a.size == b.size:
+        return a, b
+    target = (min(a.size[0], b.size[0]), min(a.size[1], b.size[1]))
+    return a.resize(target), b.resize(target)
+
+
+def worst_regions_from_diff(diff_img: Image.Image, top_n: int = 5) -> list[dict]:
+    if not HAVE_SCIPY:
+        return []
+    arr = np.array(diff_img)
+    if arr.ndim == 3:
+        mask = (arr[..., :3].sum(axis=-1) > 0).astype(np.uint8)
+    else:
+        mask = (arr > 0).astype(np.uint8)
+    labels, n = ndimage.label(mask)
+    if n == 0:
+        return []
+    regions = []
+    for i in range(1, n + 1):
+        ys, xs = np.where(labels == i)
+        if len(xs) < 20:
+            continue
+        regions.append({
+            "x": int(xs.min()),
+            "y": int(ys.min()),
+            "width": int(xs.max() - xs.min() + 1),
+            "height": int(ys.max() - ys.min() + 1),
+            "area": int(len(xs)),
+        })
+    regions.sort(key=lambda r: r["area"], reverse=True)
+    out = []
+    for r in regions[:top_n]:
+        r.pop("area", None)
+        out.append(r)
+    return out
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--before", required=True)
+    parser.add_argument("--after", required=True)
+    parser.add_argument("--diff-out", required=True)
+    parser.add_argument("--threshold", type=float, default=0.1)
+    parser.add_argument("--json", action="store_true")
+    args = parser.parse_args()
+
+    a = Image.open(args.before).convert("RGBA")
+    b = Image.open(args.after).convert("RGBA")
+    a, b = resize_to_match(a, b)
+    diff = Image.new("RGBA", a.size, (0, 0, 0, 0))
+
+    mismatched = pixelmatch(a, b, diff, threshold=args.threshold, includeAA=False)
+    total = a.size[0] * a.size[1]
+    diff_pct = (mismatched / total) * 100.0 if total else 0.0
+
+    Path(args.diff_out).parent.mkdir(parents=True, exist_ok=True)
+    diff.save(args.diff_out)
+
+    worst = worst_regions_from_diff(diff)
+
+    result = {
+        "diff_pct": round(diff_pct, 3),
+        "diff_image_path": args.diff_out,
+        "worst_regions": worst,
+        "size": {"width": a.size[0], "height": a.size[1]},
+    }
+    if args.json:
+        print(json.dumps(result))
+    else:
+        print(f"diff_pct={result['diff_pct']}% diff_image={result['diff_image_path']} regions={len(worst)}")
+
+
+if __name__ == "__main__":
+    main()