test(emit): add golden equivalence test for context_block() (#34)

tests/test_frames_emit.py

@@ -104,3 +104,70 @@ def test_pages_for_segment_revisit_then_dwell():
     assert page_b.count == 1
 
 
+def _context_block_legacy(doc, *, max_frames: int = 40) -> str:
+    """Pre-optimization legacy implementation of context_block() for golden testing."""
+    d = doc.to_dict(False)
+    cov = d["coverage"]
+    frames = d["frames"]
+
+    if len(frames) > max_frames:
+        keep = sorted(frames, key=lambda f: -f["duration_min"])[:max_frames]
+        keep_ids = {f["id"] for f in keep}
+        dropped = len(frames) - len(keep)
+        frames = [f for f in frames if f["id"] in keep_ids]
+    else:
+        dropped = 0
+
+    day = d["window"].get("day", d["window"]["start_utc"][:10])
+    lines = [
+        f"USER ACTIVITY ({day}, local time; measured from screen capture, "
+        "no interpretation):",
+        f"coverage: {cov['first_activity']}-{cov['last_activity']}, "
+        f"{cov['active_minutes']} active min, {cov['distinct_apps']} apps",
+    ]
+    for g in cov.get("gaps", []):
+        lines.append(f"away: {g['start']}-{g['end']} ({g['minutes']}m)")
+    for f in frames:
+        where = f["app"] + (f"/{f['site']}" if f.get("site") else "")
+        bits = []
+        for p in f.get("pages", [])[:4]:
+            b = p["kind"]
+            if p.get("entity"):
+                b += f":{p['entity']}"
+            if p.get("count"):
+                b += f" x{p['count']}"
+            bits.append(b)
+        if not bits and f.get("windows"):
+            bits = [f["windows"][0][:60]]
+        inp = f.get("input", {})
+        if inp.get("keys", 0) > 50:
+            bits.append(f"typed ~{inp['keys']} chars")
+        lines.append(
+            f"- {f['start'][:5]}-{f['end'][:5]} {where} ({f['duration_min']}m): "
+            + ("; ".join(bits) if bits else "on screen")
+        )
+    if dropped:
+        lines.append(f"(+{dropped} frames over the size budget omitted)")
+    omitted = d.get("omitted", {}).get("below_min_minutes", 0)
+    if omitted:
+        lines.append(
+            f"(+{omitted} brief frames under "
+            f"{d['omitted']['min_minutes']} min omitted)"
+        )
+    return "\n".join(lines)
+
+
+def test_context_block_golden_equivalence(fixture_db, day_window):
+    """Golden equivalence test: verify optimized context_block() matches legacy 1:1."""
+    for max_frames in (1, 2, 5, 10, 40):
+        for min_minutes in (0.0, 1.0, 5.0):
+            for inc_text in (False, True):
+                doc = build_frames(
+                    fixture_db, *day_window, min_minutes=min_minutes, include_text=inc_text
+                )
+                actual = context_block(doc, max_frames=max_frames)
+                expected = _context_block_legacy(doc, max_frames=max_frames)
+                assert actual == expected, (
+                    f"Golden mismatch for max_frames={max_frames}, "
+                    f"min_minutes={min_minutes}, include_text={inc_text}"
+                )