feat(recipe-generation): final export script med coverage + dedup + kostnad

- export-verified.ts exporterar nu även editorial (status=editorial) - Lägger till coverage-matrix, dedup-stats och summary.md - Inkluderar total kostnad för runda 1 + runda 2
2026-08-10 00:40:01 +07:00
parent fc3d9daac8
commit 2fda1a5785
1 changed files with 149 additions and 11 deletions
@@ -1,15 +1,34 @@
 import { createDatabase, closeDatabase } from "@app/database";
+import { SEED_INGREDIENTS } from "@app/database/seed";
 import * as fs from "node:fs/promises";
 import * as path from "node:path";

+const publicDir = "/mnt/c/Users/Public";
+
+interface RecipeWithIngredients {
+  id: string;
+  slug: string;
+  titleSv: string;
+  titleEn: string | null;
+  cuisine: string | null;
+  mealType: string | null;
+  dietVariant: string | null;
+  sourceType: string;
+  verificationStatus: string;
+}
+
+function normalizeDietVariant(raw: string | null): string {
+  return raw ?? "standard";
+}
+
 async function main() {
  const dbUrl = process.env.DATABASE_URL;
  if (!dbUrl) throw new Error("DATABASE_URL saknas");
  const { db } = createDatabase(dbUrl);

-  const recipes = await db.query.recipes.findMany({
-    where: (t, { eq }) => eq(t.verificationStatus, "verified"),
-  });
+  const recipes = (await db.query.recipes.findMany({
+    where: (t, { inArray }) => inArray(t.verificationStatus, ["verified", "editorial"]),
+  })) as RecipeWithIngredients[];

  const enriched = [];
  for (const r of recipes) {
@@ -22,15 +41,134 @@ async function main() {
    enriched.push({ ...r, ingredients, steps });
  }

-  const publicDir = "/mnt/c/Users/Public";
-  const outPath = path.join(publicDir, "cibello-verified-119-result.json");
-  await fs.writeFile(outPath, JSON.stringify({
-    exportedAt: new Date().toISOString(),
-    count: enriched.length,
-    recipes: enriched,
-  }, null, 2), "utf-8");
+  // ── Coverage matrix ──────────────────────────────────────────────────────
+  const mainIds = Array.from(new Set(SEED_INGREDIENTS.map((i) => i.id)));
+  const matrix = new Map<string, Map<string, number>>();
+
+  for (const r of recipes) {
+    const ingredients = await db.query.recipeIngredients.findMany({
+      where: (t, { eq }) => eq(t.recipeId, r.id),
+    });
+    const mainIngs = ingredients.filter((ing) => mainIds.includes(ing.canonicalIngredientId));
+    const diet = normalizeDietVariant(r.dietVariant);
+    for (const ing of mainIngs) {
+      if (!matrix.has(ing.canonicalIngredientId)) matrix.set(ing.canonicalIngredientId, new Map());
+      const cell = matrix.get(ing.canonicalIngredientId)!;
+      cell.set(diet, (cell.get(diet) ?? 0) + 1);
+    }
+  }
+
+  const coverageMatrix = mainIds.map((id) => {
+    const ing = SEED_INGREDIENTS.find((i) => i.id === id);
+    const cell = matrix.get(id) ?? new Map<string, number>();
+    return {
+      ingredientId: id,
+      nameSv: ing?.nameSv ?? id,
+      category: ing?.category ?? "unknown",
+      total: Array.from(cell.values()).reduce((s, v) => s + v, 0),
+      byDiet: Object.fromEntries(cell),
+    };
+  }).sort((a, b) => a.total - b.total);
+
+  const thinCells = coverageMatrix.filter((c) => c.total <= 2);
+
+  // ── Dedup / similarity stats ─────────────────────────────────────────────
+  const slugs = recipes.map((r) => r.slug);
+  const uniqueSlugs = new Set(slugs);
+  const titleWords = recipes.map((r) =>
+    r.titleSv.toLowerCase().replace(/[^\w\s]/g, "").split(/\s+/).filter((w) => w.length > 3),
+  );
+  const titleSignatures = titleWords.map((words) => words.slice(0, 4).sort().join(" "));
+  const sigCounts = new Map<string, number>();
+  for (const sig of titleSignatures) sigCounts.set(sig, (sigCounts.get(sig) ?? 0) + 1);
+  const suspectedDupes = Array.from(sigCounts.entries()).filter(([, n]) => n > 1);
+
+  // ── Cost summary ─────────────────────────────────────────────────────────
+  // Round 1 from earlier log summary; round 2 from the combined logs below.
+  const round1CostUsd = 0.0612;
+  const round2CostUsd = 0.0546; // g1 0.0040 + g2 0.0169 + g3 0.0337
+  const totalCostUsd = round1CostUsd + round2CostUsd;
+
+  const summary = {
+    exportedAt: new Date().toISOString(),
+    totalVerified: recipes.length,
+    editorialRecipes: recipes.filter((r) => r.sourceType === "own_editorial").length,
+    aiAssistedRecipes: recipes.filter((r) => r.sourceType === "ai_assisted_reviewed").length,
+    editorialRecipesByStatus: recipes.filter((r) => r.verificationStatus === "editorial").length,
+    verifiedRecipesByStatus: recipes.filter((r) => r.verificationStatus === "verified").length,
+    dedupStats: {
+      totalSlugs: slugs.length,
+      uniqueSlugs: uniqueSlugs.size,
+      duplicateSlugs: slugs.length - uniqueSlugs.size,
+      suspectedTitleSignatureDuplicates: suspectedDupes.length,
+      suspectedTitleSignatures: suspectedDupes.map(([sig, n]) => ({ signature: sig, count: n })),
+    },
+    coverageStats: {
+      totalCanonicalIngredients: mainIds.length,
+      ingredientsWithVerifiedRecipes: coverageMatrix.filter((c) => c.total > 0).length,
+      thinCellsCount: thinCells.length,
+      thinnestCells: thinCells.slice(0, 10),
+    },
+    costUsd: {
+      round1: round1CostUsd,
+      round2: round2CostUsd,
+      total: totalCostUsd,
+    },
+  };
+
+  await fs.writeFile(
+    path.join(publicDir, "cibello-verified-final-result.json"),
+    JSON.stringify({ exportedAt: new Date().toISOString(), count: enriched.length, recipes: enriched }, null, 2),
+    "utf-8",
+  );
+
+  await fs.writeFile(
+    path.join(publicDir, "cibello-verified-final-coverage-matrix.json"),
+    JSON.stringify({ exportedAt: new Date().toISOString(), coverageMatrix }, null, 2),
+    "utf-8",
+  );
+
+  await fs.writeFile(
+    path.join(publicDir, "cibello-verified-final-dedup-stats.json"),
+    JSON.stringify({ exportedAt: new Date().toISOString(), ...summary.dedupStats }, null, 2),
+    "utf-8",
+  );
+
+  await fs.writeFile(
+    path.join(publicDir, "cibello-verified-final-summary.md"),
+    `# Cibello Receptkatalog – Final Verified Set
+
+- **Exporterad:** ${summary.exportedAt}
+- **Totalt godkända recept (verified + editorial):** ${summary.totalVerified}
+  - Editorial: ${summary.editorialRecipes} (status=editorial)
+  - AI-assisted & reviewed: ${summary.aiAssistedRecipes} (status=verified)
+- **Kostnad:** $${summary.costUsd.total.toFixed(4)} (runda 1: $${summary.costUsd.round1.toFixed(4)}, runda 2: $${summary.costUsd.round2.toFixed(4)})
+
+## Dedup-statistik
+
+- Unika slugs: ${summary.dedupStats.uniqueSlugs} / ${summary.dedupStats.totalSlugs}
+- Dublett-slugs: ${summary.dedupStats.duplicateSlugs}
+- Misstänkta titel-signatur-dubbletter: ${summary.dedupStats.suspectedTitleSignatureDuplicates}
+
+## Täckningsmatris (tunnaste cellerna)
+
+| Huvudingrediens | Kategori | Totalt | Fördelning |
+|---|---|---:|---|
+${thinCells.slice(0, 15).map((c) => `| ${c.nameSv} | ${c.category} | ${c.total} | ${JSON.stringify(c.byDiet)} |`).join("\n")}
+
+## Granskning
+
+Setet är redo för allergen- + närings-sanity av granskare.
+`,
+    "utf-8",
+  );
+
+  console.error(`[export-verified] Exported ${recipes.length} verified recipes`);
+  console.error(`  result.json        -> ${path.join(publicDir, "cibello-verified-final-result.json")}`);
+  console.error(`  coverage-matrix    -> ${path.join(publicDir, "cibello-verified-final-coverage-matrix.json")}`);
+  console.error(`  dedup-stats        -> ${path.join(publicDir, "cibello-verified-final-dedup-stats.json")}`);
+  console.error(`  summary.md         -> ${path.join(publicDir, "cibello-verified-final-summary.md")}`);

-  console.error(`[export-verified] Exported ${recipes.length} verified recipes to ${outPath}`);
  await closeDatabase();
 }