Ajoute le graphique des personnages représentés

README.md

@@ -233,3 +233,10 @@ Le script relit `data/intermediate/sets_enriched.csv`, `data/intermediate/parts_
 
 - `data/intermediate/minifig_counts_by_set.csv` : `set_num`, `set_id`, `name`, `year`, `minifig_count`
 - `figures/step21/minifigs_per_set.png` : diagramme en barres horizontales (ordre décroissant) du nombre de minifigs par set filtré
+
+### Étape 22 : personnages représentés par les minifigs
+
+1. `source .venv/bin/activate`
+2. `python -m scripts.plot_minifig_characters`
+
+Le script lit `data/intermediate/minifigs_by_set.csv`, compte le nombre de minifigs distinctes par personnage (`known_character` + `fig_num`), écrit `data/intermediate/minifig_characters_counts.csv`, puis trace `figures/step22/minifig_characters.png` (barres horizontales triées).

lib/plots/minifig_characters.py

@@ -0,0 +1,42 @@
+"""Graphique du nombre de minifigs par personnage."""
+
+from pathlib import Path
+from typing import List
+
+import matplotlib.pyplot as plt
+
+from lib.filesystem import ensure_parent_dir
+from lib.rebrickable.stats import read_rows
+
+
+def load_counts(path: Path) -> List[dict]:
+    """Charge le CSV des comptes par personnage."""
+    return read_rows(path)
+
+
+def plot_minifigs_per_character(counts_path: Path, destination_path: Path) -> None:
+    """Trace un diagramme en barres horizontales du nombre de minifigs par personnage."""
+    rows = load_counts(counts_path)
+    characters = [row["known_character"] for row in rows]
+    counts = [int(row["minifig_count"]) for row in rows]
+    positions = list(range(len(rows)))
+    height = max(6, len(rows) * 0.22)
+
+    fig, ax = plt.subplots(figsize=(12, height))
+    bars = ax.barh(positions, counts, color="#1f77b4", edgecolor="#0d0d0d", linewidth=0.6)
+    ax.set_yticks(positions)
+    ax.set_yticklabels(characters)
+    ax.invert_yaxis()
+    ax.set_xlabel("Nombre de minifigs distinctes")
+    ax.set_title("Minifigs par personnage (thèmes filtrés)")
+    ax.grid(True, axis="x", linestyle="--", alpha=0.25)
+    max_value = max(counts) if counts else 0
+    ax.set_xlim(0, max_value + 1)
+    for index, bar in enumerate(bars):
+        value = counts[index]
+        ax.text(value + 0.1, bar.get_y() + bar.get_height() / 2, str(value), va="center", fontsize=8)
+
+    ensure_parent_dir(destination_path)
+    fig.tight_layout()
+    fig.savefig(destination_path, dpi=160)
+    plt.close(fig)

lib/rebrickable/minifig_characters.py

@@ -0,0 +1,41 @@
+"""Agrégation des minifigs par personnage représenté."""
+
+from collections import defaultdict
+from pathlib import Path
+from typing import Dict, Iterable, List, Sequence
+
+from lib.rebrickable.stats import read_rows
+from lib.filesystem import ensure_parent_dir
+import csv
+
+
+def load_minifigs_by_set(path: Path) -> List[dict]:
+    """Charge le CSV minifigs_by_set."""
+    return read_rows(path)
+
+
+def aggregate_by_character(rows: Iterable[dict]) -> List[dict]:
+    """Compte les minifigs distinctes par personnage (fig_num unique)."""
+    fig_nums_by_character: Dict[str, set] = defaultdict(set)
+    for row in rows:
+        character = row["known_character"].strip()
+        fig_num = row["fig_num"].strip()
+        if character == "" or fig_num == "":
+            continue
+        fig_nums_by_character[character].add(fig_num)
+    aggregates: List[dict] = []
+    for character, fig_nums in fig_nums_by_character.items():
+        aggregates.append({"known_character": character, "minifig_count": len(fig_nums)})
+    aggregates.sort(key=lambda r: (-r["minifig_count"], r["known_character"]))
+    return aggregates
+
+
+def write_character_counts(path: Path, rows: Sequence[dict]) -> None:
+    """Écrit le CSV des comptes par personnage."""
+    ensure_parent_dir(path)
+    fieldnames = ["known_character", "minifig_count"]
+    with path.open("w", newline="") as csv_file:
+        writer = csv.DictWriter(csv_file, fieldnames=fieldnames)
+        writer.writeheader()
+        for row in rows:
+            writer.writerow(row)

scripts/plot_minifig_characters.py

@@ -0,0 +1,23 @@
+"""Trace le nombre de minifigs distinctes par personnage."""
+
+from pathlib import Path
+
+from lib.plots.minifig_characters import plot_minifigs_per_character
+from lib.rebrickable.minifig_characters import aggregate_by_character, load_minifigs_by_set, write_character_counts
+
+
+MINIFIGS_BY_SET_PATH = Path("data/intermediate/minifigs_by_set.csv")
+COUNTS_PATH = Path("data/intermediate/minifig_characters_counts.csv")
+DESTINATION_PATH = Path("figures/step22/minifig_characters.png")
+
+
+def main() -> None:
+    """Construit le CSV de comptage par personnage et trace le graphique."""
+    rows = load_minifigs_by_set(MINIFIGS_BY_SET_PATH)
+    aggregates = aggregate_by_character(rows)
+    write_character_counts(COUNTS_PATH, aggregates)
+    plot_minifigs_per_character(COUNTS_PATH, DESTINATION_PATH)
+
+
+if __name__ == "__main__":
+    main()

tests/test_minifig_characters.py

@@ -0,0 +1,37 @@
+"""Tests de l'agrégation des minifigs par personnage."""
+
+from pathlib import Path
+
+from lib.rebrickable.minifig_characters import aggregate_by_character, write_character_counts
+
+
+def test_aggregate_by_character_counts_unique_figs() -> None:
+    """Compter les minifigs distinctes par personnage en excluant les noms vides."""
+    aggregates = aggregate_by_character(
+        [
+            {"set_num": "123-1", "part_num": "head-a", "known_character": "Owen Grady", "fig_num": "fig-owen-1"},
+            {"set_num": "124-1", "part_num": "head-b", "known_character": "Owen Grady", "fig_num": "fig-owen-1"},
+            {"set_num": "125-1", "part_num": "head-c", "known_character": "Owen Grady", "fig_num": "fig-owen-2"},
+            {"set_num": "126-1", "part_num": "head-d", "known_character": "Figurant", "fig_num": "fig-guard-1"},
+            {"set_num": "128-1", "part_num": "head-f", "known_character": "Figurant", "fig_num": "fig-guard-1"},
+            {"set_num": "129-1", "part_num": "head-g", "known_character": "", "fig_num": "fig-guard-2"},
+        ]
+    )
+
+    assert aggregates == [
+        {"known_character": "Owen Grady", "minifig_count": 2},
+        {"known_character": "Figurant", "minifig_count": 1},
+    ]
+
+
+def test_write_character_counts_outputs_csv(tmp_path: Path) -> None:
+    """Écrit le CSV des comptes par personnage."""
+    destination = tmp_path / "counts.csv"
+    rows = [
+        {"known_character": "A", "minifig_count": 2},
+        {"known_character": "B", "minifig_count": 1},
+    ]
+
+    write_character_counts(destination, rows)
+
+    assert destination.read_text() == "known_character,minifig_count\nA,2\nB,1\n"

tests/test_minifig_characters_plot.py

@@ -0,0 +1,25 @@
+"""Tests du graphique minifigs par personnage."""
+
+import matplotlib
+from pathlib import Path
+
+from lib.plots.minifig_characters import plot_minifigs_per_character
+
+
+matplotlib.use("Agg")
+
+
+def test_plot_minifigs_per_character(tmp_path: Path) -> None:
+    """Génère l'image de comptage par personnage."""
+    counts_path = tmp_path / "counts.csv"
+    destination = tmp_path / "figures" / "step22" / "minifig_characters.png"
+    counts_path.write_text(
+        "known_character,minifig_count\n"
+        "Owen Grady,2\n"
+        "Figurant,1\n"
+    )
+
+    plot_minifigs_per_character(counts_path, destination)
+
+    assert destination.exists()
+    assert destination.stat().st_size > 0