better compression

xmetrics/Dockerfile

@@ -47,7 +47,7 @@ RUN git fetch && git checkout f81543a34ee363dcc00e8632fd7cfcd4a3478b23
 ENV SRCDIR="."
 ENV SRCPATTERN="*.[ach]*"
 
-FROM metrics-tmg as metrics-repo
+FROM metrics-radn as metrics-repo
 
 FROM metrics-repo as metrics-commits
 COPY common.py /code/common.py

xmetrics/commits.py

@@ -1,4 +1,5 @@
 import pickle
+from random import randrange
 from subprocess import check_output
 
 from common import counter
@@ -6,31 +7,50 @@ from common import counter
 args = []
 original_commits = check_output(["git", "log", "--pretty=%H", *args], text=True).splitlines()
 N = len(original_commits)
-filtered_commits = set(original_commits[:: N // min(N, 4096)])
-total_changes: dict[str, int] = {}
+BITS = 10
+total_changes: dict[str, tuple[int, int]] = {}
 chosen_parents: dict[str, str | None] = {}
+heights: dict[str, int] = {}
+included = set()
 for i, commit in enumerate(reversed(original_commits)):
-    print(f"P={i / N:6f}", flush=True)
+    print(f"C={i / N:6f}", flush=True)
     parents = check_output(["git", "log", "--pretty=%P", "-n", "1", commit], text=True).split()
     chosen_parent = None
-    if commit in filtered_commits:
+    height = 1 + max((heights[parent] for parent in parents), default=0)
+    mask = ((1 << height.bit_length()) - 1) >> BITS
+    expected = int.from_bytes(bytes.fromhex(commit), "little")
+    if mask & (height ^ expected) == 0:
         ctr = counter(commit, True)
-        changes = ctr.total()
+        changes = ctr.total(), 1
         for parent in parents:
-            pctr = counter(parent, True)
-            maybe_changes = (pctr - ctr).total() + (ctr - pctr).total() + total_changes[parent]
+            if parent in included:
+                _parent = parent
+            else:
+                _parent = chosen_parents[parent]
+                if _parent is None:
+                    continue
+            assert _parent in included
+            pctr = counter(_parent, True)
+            pcc, pch = total_changes[parent]
+            maybe_changes = (pctr - ctr).total() + (ctr - pctr).total() + pcc, pch + 1
             if maybe_changes > changes:
                 changes = maybe_changes
                 chosen_parent = parent
+        included.add(commit)
     else:
-        changes = 0
+        changes = 0, 1
         for parent in parents:
-            maybe_changes = total_changes[parent]
+            pcc, pch = total_changes[parent]
+            maybe_changes = pcc, pch + 1
             if maybe_changes > changes:
                 changes = maybe_changes
                 chosen_parent = parent
     total_changes[commit] = changes
+    if chosen_parent is not None and chosen_parent not in included:
+        chosen_parent = chosen_parents[chosen_parent]
+    assert chosen_parent is None or chosen_parent in included
     chosen_parents[commit] = chosen_parent
+    heights[commit] = height
 commit = original_commits[0]
 commits = []
 while commit is not None: