chore(bikegear): add scrape-only cache command
This commit is contained in:
parent
6891ebf4bf
commit
78bc7c5140
@ -7,6 +7,7 @@
|
||||
"postinstall": "playwright install chromium",
|
||||
"setup": "playwright install chromium",
|
||||
"setup:ubuntu": "playwright install --with-deps chromium",
|
||||
"scrape:bikegear-cache": "node scripts/scrapeBikegearCache.js",
|
||||
"test:bikegear-proxy": "node scripts/testBikegearProxy.js",
|
||||
"start": "node server.js",
|
||||
"dev": "node server.js"
|
||||
|
||||
25
scripts/scrapeBikegearCache.js
Normal file
25
scripts/scrapeBikegearCache.js
Normal file
@ -0,0 +1,25 @@
|
||||
require("dotenv").config();
|
||||
|
||||
const bikegear = require("../src/business-logic/import-pipeline/sources/bikegear");
|
||||
|
||||
async function main() {
|
||||
process.env.BIKEGEAR_CACHE_ONLY = "false";
|
||||
process.env.BIKEGEAR_CACHE_HOURS = "0";
|
||||
process.env.BIKEGEAR_FETCH_MODE = process.env.BIKEGEAR_FETCH_MODE || "browser";
|
||||
process.env.BIKEGEAR_REQUEST_DELAY = process.env.BIKEGEAR_REQUEST_DELAY || "5000";
|
||||
|
||||
const runPaths = bikegear.paths();
|
||||
console.log("[BIKEGEAR-CACHE] Starting BikeGear scrape-only run.");
|
||||
console.log(`[BIKEGEAR-CACHE] Output: ${runPaths.aggregatedJson}`);
|
||||
|
||||
const summary = await bikegear.fetchWebsiteData({ paths: runPaths });
|
||||
const total = summary?.analysis?.totalProductsUnique ?? summary?.analysis?.detailSuccess ?? 0;
|
||||
|
||||
console.log(`[BIKEGEAR-CACHE] Done. Cached products: ${total}`);
|
||||
console.log(`[BIKEGEAR-CACHE] File ready: ${summary.outputJsonPath}`);
|
||||
}
|
||||
|
||||
main().catch((error) => {
|
||||
process.exitCode = 1;
|
||||
console.error(`[BIKEGEAR-CACHE] Failed: ${error.message}`);
|
||||
});
|
||||
Loading…
x
Reference in New Issue
Block a user