feat: enhance README and index.ts for large file processing and output management
This commit is contained in:
Victor Noguera
213
src/index.ts
213
src/index.ts
@@ -4,15 +4,18 @@ import { fetchSellabilityBatch, fetchSpApiPricingAndFees } from "./sp-api.ts";
|
||||
import { connectCache, getCache, setCache, disconnectCache } from "./cache.ts";
|
||||
import { analyzeProducts } from "./llm.ts";
|
||||
import { printResults, writeResultsCsv } from "./writer.ts";
|
||||
import path from "node:path";
|
||||
import type {
|
||||
EnrichedProduct,
|
||||
AnalysisResult,
|
||||
KeepaData,
|
||||
ProductRecord,
|
||||
SellabilityInfo,
|
||||
SpApiData,
|
||||
} from "./types.ts";
|
||||
|
||||
const LLM_BATCH_SIZE = 5;
|
||||
const INPUT_BATCH_SIZE = 50;
|
||||
|
||||
function parseArgs(): { inputFile: string; outputFile?: string } {
|
||||
const args = process.argv.slice(2);
|
||||
@@ -29,41 +32,67 @@ function parseArgs(): { inputFile: string; outputFile?: string } {
|
||||
return { inputFile, outputFile };
|
||||
}
|
||||
|
||||
async function main() {
|
||||
const { inputFile, outputFile } = parseArgs();
|
||||
|
||||
console.log("Connecting to Redis...");
|
||||
await connectCache();
|
||||
|
||||
// Phase 1: Read input file
|
||||
console.log(`\nReading ${inputFile}...`);
|
||||
const products = readProducts(inputFile);
|
||||
|
||||
if (products.length === 0) {
|
||||
console.error("No valid products found in input file.");
|
||||
process.exit(1);
|
||||
function chunkArray<T>(items: T[], chunkSize: number): T[][] {
|
||||
const chunks: T[][] = [];
|
||||
for (let i = 0; i < items.length; i += chunkSize) {
|
||||
chunks.push(items.slice(i, i + chunkSize));
|
||||
}
|
||||
return chunks;
|
||||
}
|
||||
|
||||
// Phase 2: Check cache for all ASINs
|
||||
function resolveBaseOutputPath(inputFile: string, outputFile?: string): string {
|
||||
if (outputFile) return outputFile;
|
||||
|
||||
const parsedInput = path.parse(inputFile);
|
||||
return path.join(parsedInput.dir, `${parsedInput.name}_results.xlsx`);
|
||||
}
|
||||
|
||||
function buildChunkOutputPath(
|
||||
baseOutputPath: string,
|
||||
chunkIndex: number,
|
||||
): string {
|
||||
const parsed = path.parse(baseOutputPath);
|
||||
const extension = parsed.ext || ".xlsx";
|
||||
const chunkSuffix = String(chunkIndex + 1).padStart(3, "0");
|
||||
return path.join(
|
||||
parsed.dir,
|
||||
`${parsed.name}_part_${chunkSuffix}${extension}`,
|
||||
);
|
||||
}
|
||||
|
||||
async function processProductChunk(
|
||||
products: ProductRecord[],
|
||||
): Promise<AnalysisResult[]> {
|
||||
// Phase 2: Check cache for all ASINs in chunk
|
||||
console.log(`\nChecking cache for ${products.length} products...`);
|
||||
const cached = new Map<string, EnrichedProduct>();
|
||||
const excludedCachedAsins = new Set<string>();
|
||||
const uncachedProducts: ProductRecord[] = [];
|
||||
|
||||
for (const p of products) {
|
||||
const hit = await getCache(p.asin);
|
||||
if (hit) {
|
||||
console.log(` [cache hit] ${p.asin}`);
|
||||
cached.set(p.asin, hit);
|
||||
if (hit.spApi.sellabilityStatus === "available") {
|
||||
console.log(` [cache hit] ${p.asin}`);
|
||||
cached.set(p.asin, hit);
|
||||
} else {
|
||||
excludedCachedAsins.add(p.asin);
|
||||
console.log(
|
||||
` [exclude cached] ${p.asin} — status=${hit.spApi.sellabilityStatus}`,
|
||||
);
|
||||
}
|
||||
} else {
|
||||
uncachedProducts.push(p);
|
||||
}
|
||||
}
|
||||
console.log(`${cached.size} cached, ${uncachedProducts.length} to fetch`);
|
||||
console.log(
|
||||
`${cached.size} cached available, ${excludedCachedAsins.size} cached excluded, ${uncachedProducts.length} to fetch`,
|
||||
);
|
||||
|
||||
// Phase 3: Sellability gate — check all uncached ASINs before anything else
|
||||
// Phase 3: Sellability gate — check uncached ASINs before anything else
|
||||
const sellabilityMap = new Map<string, SellabilityInfo>();
|
||||
const sellableProducts: ProductRecord[] = [];
|
||||
const skippedProducts: ProductRecord[] = [];
|
||||
const availableProducts: ProductRecord[] = [];
|
||||
const unavailableProducts: ProductRecord[] = [];
|
||||
|
||||
if (uncachedProducts.length > 0) {
|
||||
console.log(
|
||||
@@ -81,50 +110,46 @@ async function main() {
|
||||
};
|
||||
sellabilityMap.set(p.asin, info);
|
||||
|
||||
// Keep: available, restricted (can request approval), unknown (proceed cautiously)
|
||||
// Skip: not_available with canSell explicitly false
|
||||
if (
|
||||
info.sellabilityStatus === "not_available" &&
|
||||
info.canSell === false
|
||||
) {
|
||||
skippedProducts.push(p);
|
||||
// Keep only ASINs that are explicitly available.
|
||||
if (info.sellabilityStatus === "available") {
|
||||
availableProducts.push(p);
|
||||
console.log(
|
||||
` [skip] ${p.asin} — ${info.sellabilityReason ?? "not available"}`,
|
||||
` [available] ${p.asin} — status=${info.sellabilityStatus}`,
|
||||
);
|
||||
} else {
|
||||
sellableProducts.push(p);
|
||||
unavailableProducts.push(p);
|
||||
console.log(
|
||||
` [sellable] ${p.asin} — status=${info.sellabilityStatus}`,
|
||||
` [exclude] ${p.asin} — status=${info.sellabilityStatus}, reason=${info.sellabilityReason ?? "n/a"}`,
|
||||
);
|
||||
}
|
||||
}
|
||||
|
||||
console.log(
|
||||
`\nSellability gate: ${sellableProducts.length} sellable, ${skippedProducts.length} skipped`,
|
||||
`\nSellability gate: ${availableProducts.length} available, ${unavailableProducts.length} excluded`,
|
||||
);
|
||||
}
|
||||
|
||||
// Phase 4: Keepa batch fetch — only for sellable (uncached) ASINs
|
||||
// Phase 4: Keepa batch fetch — only for available (uncached) ASINs
|
||||
let keepaResults = new Map<string, KeepaData>();
|
||||
if (sellableProducts.length > 0) {
|
||||
console.log(`\nFetching ${sellableProducts.length} ASINs from Keepa...`);
|
||||
if (availableProducts.length > 0) {
|
||||
console.log(`\nFetching ${availableProducts.length} ASINs from Keepa...`);
|
||||
try {
|
||||
keepaResults = await fetchKeepaDataBatch(
|
||||
sellableProducts.map((p) => p.asin),
|
||||
availableProducts.map((p) => p.asin),
|
||||
);
|
||||
} catch (err) {
|
||||
console.warn(`Keepa batch fetch failed: ${err}`);
|
||||
}
|
||||
}
|
||||
|
||||
// Phase 5: SP-API pricing + fees — only for sellable ASINs
|
||||
// Phase 5: SP-API pricing + fees — only for available ASINs
|
||||
console.log(
|
||||
`\nFetching pricing & fees for ${sellableProducts.length} ASINs...`,
|
||||
`\nFetching pricing & fees for ${availableProducts.length} ASINs...`,
|
||||
);
|
||||
const spApiResults = new Map<string, import("./types.ts").SpApiData>();
|
||||
const spApiResults = new Map<string, SpApiData>();
|
||||
|
||||
// Concurrency-limited pricing+fees fetches
|
||||
const pricingQueue = [...sellableProducts];
|
||||
const pricingQueue = [...availableProducts];
|
||||
let pricingDone = 0;
|
||||
|
||||
async function fetchNextPricing(): Promise<void> {
|
||||
@@ -140,16 +165,16 @@ async function main() {
|
||||
|
||||
spApiResults.set(p.asin, spApi);
|
||||
pricingDone++;
|
||||
if (pricingDone % 10 === 0 || pricingDone === sellableProducts.length) {
|
||||
if (pricingDone % 10 === 0 || pricingDone === availableProducts.length) {
|
||||
console.log(
|
||||
` [pricing] ${pricingDone}/${sellableProducts.length} fetched`,
|
||||
` [pricing] ${pricingDone}/${availableProducts.length} fetched`,
|
||||
);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
const pricingWorkers = Array.from(
|
||||
{ length: Math.min(5, sellableProducts.length || 1) },
|
||||
{ length: Math.min(5, availableProducts.length || 1) },
|
||||
() => fetchNextPricing(),
|
||||
);
|
||||
await Promise.all(pricingWorkers);
|
||||
@@ -157,9 +182,13 @@ async function main() {
|
||||
// Phase 6: Build enriched products
|
||||
console.log(`\nEnriching products...`);
|
||||
const enriched: EnrichedProduct[] = [];
|
||||
const autoSkipResults: AnalysisResult[] = [];
|
||||
const availableAsins = new Set(availableProducts.map((ap) => ap.asin));
|
||||
|
||||
for (const p of products) {
|
||||
if (excludedCachedAsins.has(p.asin)) {
|
||||
continue;
|
||||
}
|
||||
|
||||
// Cached products — already enriched
|
||||
const cachedProduct = cached.get(p.asin);
|
||||
if (cachedProduct) {
|
||||
@@ -167,38 +196,12 @@ async function main() {
|
||||
continue;
|
||||
}
|
||||
|
||||
// Skipped products — not sellable, auto-SKIP
|
||||
if (skippedProducts.some((sp) => sp.asin === p.asin)) {
|
||||
const sellability = sellabilityMap.get(p.asin)!;
|
||||
const product: EnrichedProduct = {
|
||||
record: p,
|
||||
keepa: null,
|
||||
spApi: {
|
||||
fbaFee: 0,
|
||||
fbmFee: 0,
|
||||
referralFeePercent: 15,
|
||||
estimatedSalePrice: 0,
|
||||
...sellability,
|
||||
},
|
||||
fetchedAt: new Date().toISOString(),
|
||||
};
|
||||
autoSkipResults.push({
|
||||
product,
|
||||
verdict: {
|
||||
asin: p.asin,
|
||||
verdict: "SKIP",
|
||||
confidence: 100,
|
||||
reasoning:
|
||||
`Not sellable: ${sellability.sellabilityReason ?? sellability.sellabilityStatus}`.slice(
|
||||
0,
|
||||
100,
|
||||
),
|
||||
},
|
||||
});
|
||||
// Exclude products that are not explicitly available.
|
||||
if (!availableAsins.has(p.asin)) {
|
||||
continue;
|
||||
}
|
||||
|
||||
// Sellable products — full enrichment
|
||||
// Available products — full enrichment
|
||||
const keepa = keepaResults.get(p.asin) ?? null;
|
||||
const spApi = spApiResults.get(p.asin) ?? {
|
||||
fbaFee: 5.0,
|
||||
@@ -229,7 +232,7 @@ async function main() {
|
||||
}
|
||||
}
|
||||
|
||||
// Phase 7: LLM analysis in batches — only for enriched (sellable + cached) products
|
||||
// Phase 7: LLM analysis in batches — only for enriched available products
|
||||
console.log(
|
||||
`\nAnalyzing ${enriched.length} products via LLM (batch size: ${LLM_BATCH_SIZE})...\n`,
|
||||
);
|
||||
@@ -274,16 +277,66 @@ async function main() {
|
||||
}
|
||||
}
|
||||
|
||||
// Merge: LLM-analyzed results + auto-skipped results
|
||||
const allResults = [...results, ...autoSkipResults];
|
||||
return results;
|
||||
}
|
||||
|
||||
printResults(allResults);
|
||||
async function main() {
|
||||
const { inputFile, outputFile } = parseArgs();
|
||||
|
||||
if (outputFile) {
|
||||
writeResultsCsv(allResults, outputFile);
|
||||
console.log("Connecting to Redis...");
|
||||
await connectCache();
|
||||
|
||||
try {
|
||||
// Phase 1: Read input file
|
||||
console.log(`\nReading ${inputFile}...`);
|
||||
const products = readProducts(inputFile);
|
||||
|
||||
if (products.length === 0) {
|
||||
console.error("No valid products found in input file.");
|
||||
process.exit(1);
|
||||
}
|
||||
|
||||
const productChunks = chunkArray(products, INPUT_BATCH_SIZE);
|
||||
const hasMultipleChunks = productChunks.length > 1;
|
||||
const shouldWriteChunkFiles = hasMultipleChunks;
|
||||
const resolvedBaseOutputPath = resolveBaseOutputPath(inputFile, outputFile);
|
||||
const allResults: AnalysisResult[] = [];
|
||||
|
||||
if (hasMultipleChunks) {
|
||||
console.log(
|
||||
`\nLarge input detected (${products.length} products). Processing in chunks of ${INPUT_BATCH_SIZE}.`,
|
||||
);
|
||||
console.log(
|
||||
`Chunk outputs will be written as numbered files using base: ${resolvedBaseOutputPath}`,
|
||||
);
|
||||
}
|
||||
|
||||
for (let chunkIndex = 0; chunkIndex < productChunks.length; chunkIndex++) {
|
||||
const chunk = productChunks[chunkIndex]!;
|
||||
console.log(
|
||||
`\n=== Input chunk ${chunkIndex + 1}/${productChunks.length} (${chunk.length} products) ===`,
|
||||
);
|
||||
|
||||
const chunkResults = await processProductChunk(chunk);
|
||||
allResults.push(...chunkResults);
|
||||
|
||||
if (shouldWriteChunkFiles) {
|
||||
const chunkOutputPath = buildChunkOutputPath(
|
||||
resolvedBaseOutputPath,
|
||||
chunkIndex,
|
||||
);
|
||||
writeResultsCsv(chunkResults, chunkOutputPath);
|
||||
}
|
||||
}
|
||||
|
||||
printResults(allResults);
|
||||
|
||||
if (!hasMultipleChunks && outputFile) {
|
||||
writeResultsCsv(allResults, outputFile);
|
||||
}
|
||||
} finally {
|
||||
await disconnectCache();
|
||||
}
|
||||
|
||||
await disconnectCache();
|
||||
}
|
||||
|
||||
main().catch((err) => {
|
||||
|
||||
Reference in New Issue
Block a user