savegame fix + chart updates

experiment_runner/analysis/statistical_game_analysis.py

@@ -26,6 +26,7 @@ from __future__ import annotations
 
 import logging
 import re
+import json
 from pathlib import Path
 from typing import List
 
@@ -64,7 +65,16 @@ def _numeric_columns(df: pd.DataFrame, extra_exclude: set[str] | None = None) ->
 
 
 def _phase_sort_key(ph: str) -> tuple[int, int]:
-    """Convert 'S1901M' → (1901, 0)."""
+    """
+    Sort key that keeps normal phases chronological and forces the literal
+    string 'COMPLETED' to the very end.
+
+    • 'S1901M' → (1901, 0)
+    • 'COMPLETED' → (9999, 9)
+    """
+    if ph.upper() == "COMPLETED":
+        return (9999, 9)               # always last
+
     year = int(ph[1:5]) if len(ph) >= 5 and ph[1:5].isdigit() else 0
     season = _SEASON_ORDER.get(ph[0], 9)
     return year, season
@@ -75,6 +85,27 @@ def _phase_index(series: pd.Series) -> pd.Series:
     mapping = {ph: i for i, ph in enumerate(uniq)}
     return series.map(mapping)
 
+def _map_game_id_to_run_dir(exp_dir: Path) -> dict[str, str]:
+    """
+    Reads each runs/run_xxxxx/lmvsgame.json file and returns
+    {game_id_string: 'run_xxxxx'}.
+    """
+    mapping: dict[str, str] = {}
+    runs_root = exp_dir / "runs"
+    for run_dir in runs_root.glob("run_*"):
+        json_path = run_dir / "lmvsgame.json"
+        if not json_path.exists():
+            continue
+        try:
+            with json_path.open(encoding="utf-8") as fh:
+                data = json.load(fh)
+            gid = str(data.get("id", ""))          # use top-level "id"
+            if gid:
+                mapping[gid] = run_dir.name
+        except Exception:                          # corrupt / unreadable → skip
+            continue
+    return mapping
+
 
 # ───────────────────────── plots ────────────────────────────
 def _plot_game_level(all_games: pd.DataFrame, plot_dir: Path) -> None:
@@ -130,7 +161,15 @@ def _plot_game_level(all_games: pd.DataFrame, plot_dir: Path) -> None:
 
 
 
-def _plot_phase_level(all_phase: pd.DataFrame, plot_dir: Path) -> None:
+def _plot_phase_level(
+    all_phase: pd.DataFrame,
+    plot_dir: Path,
+    title_suffix: str = "",
+) -> None:
+    """
+    Plots aggregated phase metrics.  If *title_suffix* is supplied it is
+    appended to each chart title — handy for per-run plots.
+    """
     if all_phase.empty:
         return
     plot_dir.mkdir(parents=True, exist_ok=True)
@@ -147,10 +186,10 @@ def _plot_phase_level(all_phase: pd.DataFrame, plot_dir: Path) -> None:
     )
 
     n_phases = agg["phase_index"].nunique()
-    fig_base_width = max(8, n_phases * 0.1 + 4)  # 0.45 in per label + padding
+    fig_w = max(8, n_phases * 0.1 + 4)
 
     for col in num_cols:
-        plt.figure(figsize=(fig_base_width, 6))
+        plt.figure(figsize=(fig_w, 6))
         sns.lineplot(
             data=agg,
             x="phase_index",
@@ -169,13 +208,40 @@ def _plot_phase_level(all_phase: pd.DataFrame, plot_dir: Path) -> None:
             rotation=90,
             fontsize=8,
         )
+        title = col.replace("_", " ").title()
+        if title_suffix:
+            title = f"{title} – {title_suffix}"
+        plt.title(title)
         plt.xlabel("Game Phase")
-        plt.title(col.replace("_", " ").title())
         plt.tight_layout()
         plt.savefig(plot_dir / f"{_sanitize(col)}.png", dpi=140)
         plt.close()
 
 
+def _plot_phase_level_per_game(
+    all_phase: pd.DataFrame,
+    root_dir: Path,
+    gameid_to_rundir: dict[str, str],
+) -> None:
+    """
+    Writes one folder of phase-plots per iteration.
+    Folder name and chart titles use the run directory (e.g. run_00003).
+    """
+    if all_phase.empty or "game_id" not in all_phase.columns:
+        return
+
+    for game_id, sub in all_phase.groupby("game_id"):
+        run_label = gameid_to_rundir.get(str(game_id), f"game_{_sanitize(str(game_id))}")
+        target = root_dir / run_label
+
+        # ── critical change: drop global phase_index so we rebuild a dense one ──
+        sub = sub.copy().drop(columns=["phase_index"], errors="ignore")
+
+        _plot_phase_level(sub, target, title_suffix=run_label)
+
+
+
+
 # ───────────────────────── entry-point ─────────────────────────
 def run(experiment_dir: Path, ctx: dict) -> None:  # pylint: disable=unused-argument
     root = experiment_dir / "analysis" / "statistical_game_analysis"
@@ -210,5 +276,12 @@ def run(experiment_dir: Path, ctx: dict) -> None:  # pylint: disable=unused-argu
     sns.set_theme(style="whitegrid")
     _plot_game_level(all_game_df, plots_root / "game")
     _plot_phase_level(all_phase_df, plots_root / "phase")
+    game_map = _map_game_id_to_run_dir(experiment_dir)
+    _plot_phase_level_per_game(
+        all_phase_df,
+        plots_root / "phase_by_game",
+        game_map,
+    )
+
 
     log.info("statistical_game_analysis: plots written → %s", plots_root)