Ajoute l’analyse des catégories de pièces

scripts/compute_part_categories.py

@@ -0,0 +1,50 @@
+"""Calcule la répartition des pièces par catégories."""
+
+from pathlib import Path
+
+from lib.rebrickable.part_categories import (
+    build_category_totals,
+    build_global_totals,
+    enrich_categories_with_sets,
+    load_category_names,
+    load_parts_catalog,
+    load_sets_enriched,
+    write_categories_by_set,
+    write_categories_by_year,
+    write_categories_global,
+)
+from lib.rebrickable.stats import read_rows
+
+
+PARTS_PATH = Path("data/intermediate/parts_filtered.csv")
+PARTS_CATALOG_PATH = Path("data/raw/parts.csv")
+PART_CATEGORIES_PATH = Path("data/raw/part_categories.csv")
+SETS_ENRICHED_PATH = Path("data/intermediate/sets_enriched.csv")
+CATEGORIES_BY_SET_PATH = Path("data/intermediate/part_categories_by_set.csv")
+CATEGORIES_BY_YEAR_PATH = Path("data/intermediate/part_categories_by_year.csv")
+CATEGORIES_GLOBAL_PATH = Path("data/intermediate/part_categories_global.csv")
+
+
+def main() -> None:
+    """Construit les agrégats par set, par année et globaux."""
+    parts_rows = read_rows(PARTS_PATH)
+    parts_catalog = load_parts_catalog(PARTS_CATALOG_PATH)
+    category_names = load_category_names(PART_CATEGORIES_PATH)
+    sets_lookup = load_sets_enriched(SETS_ENRICHED_PATH)
+    grouped_parts = {}
+    for row in parts_rows:
+        set_rows = grouped_parts.get(row["set_num"])
+        if set_rows is None:
+            set_rows = []
+            grouped_parts[row["set_num"]] = set_rows
+        set_rows.append(row)
+    categories_by_set_raw, categories_by_year = build_category_totals(grouped_parts, parts_catalog, category_names)
+    categories_by_set = enrich_categories_with_sets(categories_by_set_raw, sets_lookup)
+    categories_global = build_global_totals(categories_by_set)
+    write_categories_by_set(CATEGORIES_BY_SET_PATH, categories_by_set)
+    write_categories_by_year(CATEGORIES_BY_YEAR_PATH, categories_by_year)
+    write_categories_global(CATEGORIES_GLOBAL_PATH, categories_global)
+
+
+if __name__ == "__main__":
+    main()

scripts/plot_part_categories.py

@@ -0,0 +1,27 @@
+"""Trace les graphiques de répartition par catégories de pièces."""
+
+from pathlib import Path
+
+from lib.plots.part_categories import (
+    plot_part_categories_heatmap,
+    plot_structural_share_timeline,
+    plot_top_part_categories_area,
+)
+
+
+CATEGORIES_BY_YEAR_PATH = Path("data/intermediate/part_categories_by_year.csv")
+CATEGORIES_GLOBAL_PATH = Path("data/intermediate/part_categories_global.csv")
+AREA_DESTINATION = Path("figures/step29/top_part_categories_area.png")
+HEATMAP_DESTINATION = Path("figures/step29/part_categories_heatmap.png")
+STRUCTURAL_DESTINATION = Path("figures/step29/structural_share_timeline.png")
+
+
+def main() -> None:
+    """Génère les visuels de répartition par catégorie."""
+    plot_top_part_categories_area(CATEGORIES_BY_YEAR_PATH, CATEGORIES_GLOBAL_PATH, AREA_DESTINATION)
+    plot_part_categories_heatmap(CATEGORIES_BY_YEAR_PATH, HEATMAP_DESTINATION)
+    plot_structural_share_timeline(CATEGORIES_BY_YEAR_PATH, STRUCTURAL_DESTINATION)
+
+
+if __name__ == "__main__":
+    main()