"""Appendix D (revision point 4): for every pre-registration, the commit that first added it, the
results file its own run wrote, the commit that first added that file, and the order of the two
in git - "before" when the document is a strict ancestor of the results.  The mapping document ->
own results file is declared below (read off the sweep script each run used); the commits and
the order are read from git.

    python -u check_prereg_order.py        -> markdown rows and a summary on stdout
"""
from __future__ import annotations

import subprocess

OWN_RESULTS = {
    "PREREG_logarithmic_2026-09-20.md": "sweep_log.json",
    "PREREG_prestep_gate_2026-09-21.md": "sweep_gate.json",
    "PREREG_slope_gate_2026-09-21.md": "sweep_slope.json",
    "PREREG_points_and_direction_2026-09-21.md": "sweep_points.json",
    "PREREG_similarity_per_slot_2026-09-21.md": "sweep_order.json",
    "PREREG_scale_selection_2026-09-21.md": "sweep_cap.json",
    "PREREG_decision_place_2026-09-21.md": "sweep_decision.json",
    "PREREG_step_extrapolation_2026-09-21.md": "results_diag_step_extrapolation.txt",
    "PREREG_extrapolate_in_method_2026-09-21.md": "sweep_extrap.json",
    "PREREG_extrapolation_gates_2026-09-21.md": "sweep_extrap2.json",
    "PREREG_empty_step_2026-09-21.md": "results_diag_empty_step.txt",
    "PREREG_outcome_memory_2026-09-21.md": "results_diag_outcome_memory.txt",
    "PREREG_flat_regime_2026-09-21.md": "sweep_flat.json",
    "PREREG_flat_twice_2026-09-21.md": "sweep_flat2.json",
    "PREREG_flat_creep_2026-09-21.md": "sweep_flatcreep.json",
    "PREREG_flat_stick_2026-09-21.md": "sweep_flatstick.json",
    "PREREG_flat_exit_2026-09-21.md": "sweep_flatexit.json",
    "PREREG_flat_judge_2026-09-21.md": "sweep_flatjudge.json",
    "PREREG_flat_analogy_2026-09-21.md": "sweep_flatanalog.json",
    "PREREG_flat_analogy_mem_2026-09-21.md": None,      # decided on the smoke (stated in it)
    "PREREG_defect_contra_flat_2026-09-21.md": "sweep_defcontra.json",
    "PREREG_flat_order_point_2026-09-21.md": "sweep_orderpt.json",
    "PREREG_flat_exp_limit_2026-09-21.md": "sweep_explimit.json",
    "PREREG_half_contra_2026-09-21.md": "sweep_half.json",
    "PREREG_out_of_sample_2026-09-21.md": "sweep_oos.json",
    "PREREG_roots_jl_2026-09-21.md": "sweep_rootsjl.json",
    "PREREG_applied_2026-09-21.md": "sweep_applied.json",
    "PREREG_close_sub_2026-09-21.md": "sweep_fourth_fourth.json",
    "PREREG_modab_2026-09-21.md": "sweep_modab.json",
    "PREREG_controls_2026-09-22.md": "sweep_controls.json",
}


def first_commit(path):
    out = subprocess.run(["git", "log", "--diff-filter=A", "--format=%h", "--", path],
                         capture_output=True, text=True, check=False).stdout.split()
    return out[-1] if out else None


def order(pre, res):
    if pre == res:
        return "same commit"
    r = subprocess.run(["git", "merge-base", "--is-ancestor", pre, res], check=False).returncode
    return "before" if r == 0 else "after"


def rows():
    out = []
    for doc, res in OWN_RESULTS.items():
        hp = first_commit(doc)
        if res is None:
            out.append((doc, hp, "—", "—", "decided on the smoke run, no results file"))
            continue
        hr = first_commit(res)
        out.append((doc, hp, res, hr, order(hp, hr) if hr else "results not committed"))
    return out


def main():
    rs = rows()
    for doc, hp, res, hr, o in rs:
        print(f"| `{doc[7:-3]}` | `{hp}` | `{res}` | `{hr}` | {o} |")
    n = {k: sum(1 for r in rs if r[4] == k) for k in ("before", "same commit", "after")}
    print(f"\n{len(rs)} pre-registrations: {n['before']} before their results, "
          f"{n['same commit']} in the same commit, {n['after']} after")
    print("prereg order done")


if __name__ == "__main__":
    main()