import pickle
from collections import Counter

from common import counter

with open("commits.dat", "rb") as file:
    commits: list[str] = pickle.load(file)
entries = []
last_ctr = Counter()
last_cor = []
cors = []
C = min(len(commits), 720)
for i, commit in enumerate(reversed(commits)):
    print(f"P={i/len(commits):6f}", flush=True)
    print("running", commit, flush=True)
    current_ctr = counter(commit, True)
    added = current_ctr - last_ctr
    deleted = last_ctr - current_ctr
    entries.append((i, current_ctr.total(), added.total(), deleted.total()))
    current_cor = []
    common_ctr = current_ctr & last_ctr
    for j, line in reversed(last_cor):
        if common_ctr[line]:
            common_ctr[line] -= 1
            current_cor.append((j, line))
    current_cor.reverse()
    c = i * C // len(commits)
    for line in added.elements():
        current_cor.append((c, line))
    cor_ctr = Counter(j for j, _ in current_cor)
    assert len(current_cor) == cor_ctr.total() == current_ctr.total()
    while c >= len(cors):
        cors.append([0] * i)
    for j, cor in enumerate(cors):
        cor.append(cor_ctr[j])
    last_ctr = current_ctr
    last_cor = current_cor
with open("/code/metrics.dat", "wb") as file:
    pickle.dump({"entries": entries, "cors": cors}, file)