asin-check/src/stalker/stalker-analyze.ts

import { db } from "../db/index.ts";
import { persistLlmResults, refreshRunStats } from "../db/persistence.ts";
import { sql } from "drizzle-orm";
import { normalizeAsin } from "../asin.ts";
import { analyzeProducts } from "../integrations/llm.ts";
import { fetchSpApiPricingAndFees } from "../integrations/sp-api.ts";
import type {
  AnalysisResult,
  EnrichedProduct,
  KeepaData,
  ProductRecord,
  SellabilityInfo,
} from "../types.ts";

const LLM_BATCH_SIZE = 5;
const LLM_BATCH_DELAY_MS = 5_000;

type Args = {
  stalkerRunId: number;
  analysisRunId: number;
  asins: string[];
  useClaude: boolean;
};

type InventoryRow = {
  inventoryItemId: number;
  asin: string;
  productTitle: string | null;
  brand: string | null;
  categoryTree: string | null;
  currentPrice: number | null;
  avgPrice90d: number | null;
  salesRank: number | null;
  monthlySold: number | null;
  sellerCount: number | null;
  amazonIsSeller: boolean | null;
  canSell: boolean | null;
  sellabilityStatus: SellabilityInfo["sellabilityStatus"] | null;
  sellabilityReason: string | null;
};

function readFlagValue(args: string[], flag: string): string | undefined {
  const index = args.indexOf(flag);
  if (index === -1) return undefined;
  return args[index + 1];
}

function parseArgs(argv = process.argv.slice(2)): Args {
  const stalkerRunId = Number(readFlagValue(argv, "--stalker-run-id"));
  const analysisRunId = Number(readFlagValue(argv, "--analysis-run-id"));
  const useClaude = argv.includes("--claude");
  const asins = (readFlagValue(argv, "--asins") ?? "")
    .split(",")
    .map((asin) => normalizeAsin(asin))
    .filter((asin): asin is string => asin !== null);

  if (!Number.isInteger(stalkerRunId) || stalkerRunId <= 0) {
    throw new Error("--stalker-run-id must be a positive integer");
  }
  if (!Number.isInteger(analysisRunId) || analysisRunId <= 0) {
    throw new Error("--analysis-run-id must be a positive integer");
  }
  if (asins.length === 0) throw new Error("Missing --asins");

  return { stalkerRunId, analysisRunId, asins, useClaude };
}

function wait(ms: number): Promise<void> {
  return new Promise((resolve) => setTimeout(resolve, ms));
}

function parseCategoryTree(value: string | null): string[] {
  return value ? value.split(" > ").filter(Boolean) : [];
}

function toProductRecord(row: InventoryRow): ProductRecord {
  const categoryTree = parseCategoryTree(row.categoryTree);
  return {
    asin: row.asin,
    name: row.productTitle ?? row.asin,
    brand: row.brand ?? undefined,
    category: categoryTree.join(" > ") || undefined,
    unitCost: 0,
    amazonRank: row.salesRank ?? undefined,
    sellingPriceFromSheet: row.currentPrice ?? undefined,
    avgPrice90FromSheet: row.avgPrice90d ?? undefined,
  };
}

function toKeepaData(row: InventoryRow): KeepaData {
  return {
    currentPrice: row.currentPrice,
    avgPrice90: row.avgPrice90d,
    minPrice90: null,
    maxPrice90: null,
    salesRank: row.salesRank,
    salesRankAvg90: null,
    salesRankDrops30: null,
    salesRankDrops90: null,
    sellerCount: row.sellerCount,
    amazonIsSeller: row.amazonIsSeller,
    amazonBuyboxSharePct90d: null,
    buyBoxSeller: null,
    buyBoxPrice: null,
    buyBoxAvg90: null,
    monthlySold: row.monthlySold,
    categoryTree: parseCategoryTree(row.categoryTree),
  };
}

function toSellability(row: InventoryRow): SellabilityInfo {
  return {
    canSell: row.canSell,
    sellabilityStatus: row.sellabilityStatus ?? "unknown",
    sellabilityReason: row.sellabilityReason ?? undefined,
  };
}

async function loadInventoryRows(
  stalkerRunId: number,
  asins: string[],
): Promise<InventoryRow[]> {
  if (asins.length === 0) return [];
  return db.execute(
    sql<InventoryRow>`SELECT DISTINCT ON (inventory.product_asin)
      inventory.id AS "inventoryItemId",
      inventory.product_asin AS asin,
      product.name AS "productTitle",
      product.brand,
      product.category AS "categoryTree",
      observation.current_price AS "currentPrice",
      observation.avg_price_90d AS "avgPrice90d",
      observation.sales_rank AS "salesRank",
      observation.monthly_sold AS "monthlySold",
      observation.seller_count AS "sellerCount",
      observation.amazon_is_seller AS "amazonIsSeller",
      observation.can_sell AS "canSell",
      observation.sellability_status AS "sellabilityStatus",
      observation.sellability_reason AS "sellabilityReason"
    FROM stalker_inventory_items inventory
    JOIN products product ON product.asin = inventory.product_asin
    JOIN product_observations observation ON observation.id = inventory.observation_id
    WHERE inventory.run_id = ${stalkerRunId}
      AND observation.can_sell = true
      AND observation.sellability_status = 'available'
      AND inventory.product_asin = ANY(${asins})
    ORDER BY inventory.product_asin, observation.fetched_at DESC`,
  );
}

async function buildEnrichedProducts(
  rows: InventoryRow[],
): Promise<EnrichedProduct[]> {
  const enriched: EnrichedProduct[] = [];

  for (const row of rows) {
    const sellability = toSellability(row);
    const spApi = await fetchSpApiPricingAndFees(
      row.asin,
      sellability,
      row.currentPrice,
    );

    enriched.push({
      record: toProductRecord(row),
      keepa: toKeepaData(row),
      spApi,
      fetchedAt: new Date().toISOString(),
    });
  }

  return enriched;
}

async function insertProductAnalysisResults(
  runId: number,
  results: AnalysisResult[],
  sourceInventoryIds: Map<string, number>,
): Promise<void> {
  if (results.length === 0) return;
  await persistLlmResults(runId, results, {
    source: "stalker_analysis",
    metadataSource: "catalog",
    sourceInventoryIds,
  });
}

async function refreshAnalysisRun(runId: number): Promise<void> {
  await refreshRunStats(runId);
}

async function analyzeInBatches(
  products: EnrichedProduct[],
  useClaude: boolean,
): Promise<AnalysisResult[]> {
  const results: AnalysisResult[] = [];

  for (let i = 0; i < products.length; i += LLM_BATCH_SIZE) {
    const batch = products.slice(i, i + LLM_BATCH_SIZE);
    const batchNumber = Math.floor(i / LLM_BATCH_SIZE) + 1;
    const totalBatches = Math.ceil(products.length / LLM_BATCH_SIZE);
    console.log(
      `Stalker analysis: LLM batch ${batchNumber}/${totalBatches} (${batch.length} product(s)).`,
    );

    if (i > 0) {
      await wait(LLM_BATCH_DELAY_MS);
    }

    let verdicts;
    try {
      verdicts = await analyzeProducts(batch, { useClaude });
    } catch (error) {
      console.warn(
        `Stalker analysis: LLM batch ${batchNumber} failed: ${
          error instanceof Error ? error.message : String(error)
        }`,
      );
      verdicts = null;
    }

    for (let j = 0; j < batch.length; j++) {
      const product = batch[j];
      if (!product) continue;
      results.push({
        product,
        verdict: verdicts?.[j] ?? {
          asin: product.record.asin,
          verdict: "SKIP",
          confidence: 0,
          reasoning: "LLM analysis failed or returned no verdict",
        },
      });
    }
  }

  return results;
}

async function main(): Promise<void> {
  const args = parseArgs();

  const rows = await loadInventoryRows(args.stalkerRunId, args.asins);
  if (rows.length === 0) {
    console.log("Stalker analysis: no sellable inventory rows to analyze.");
    return;
  }

  console.log(`Stalker analysis: analyzing ${rows.length} sellable ASIN(s).`);
  const enriched = await buildEnrichedProducts(rows);
  const results = await analyzeInBatches(enriched, args.useClaude);
  const sourceInventoryIds = new Map(
    rows.map((row) => [row.asin, row.inventoryItemId]),
  );
  await insertProductAnalysisResults(
    args.analysisRunId,
    results,
    sourceInventoryIds,
  );
  await refreshAnalysisRun(args.analysisRunId);
}

if (import.meta.main) {
  main().catch((error) => {
    console.error(error instanceof Error ? error.message : String(error));
    process.exit(1);
  });
}