Merge pull request #674 from mendableai/nsc/map-pagination

v1/map improvements + higher limits
This commit is contained in:
Nicolas 2024-09-19 11:31:23 -04:00 committed by GitHub
commit ef71940429
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 129 additions and 36 deletions

View File

@ -19,8 +19,15 @@ import { billTeam } from "../../services/billing/credit_billing";
import { logJob } from "../../services/logging/log_job";
import { performCosineSimilarity } from "../../lib/map-cosine";
import { Logger } from "../../lib/logger";
import Redis from "ioredis";
configDotenv();
const redis = new Redis(process.env.REDIS_URL);
// Max Links that /map can return
const MAX_MAP_LIMIT = 5000;
// Max Links that "Smart /map" can return
const MAX_FIRE_ENGINE_RESULTS = 1000;
export async function mapController(
req: RequestWithAuth<{}, MapResponse, MapRequest>,
@ -30,8 +37,7 @@ export async function mapController(
req.body = mapRequestSchema.parse(req.body);
const limit : number = req.body.limit ?? 5000;
const limit: number = req.body.limit ?? MAX_MAP_LIMIT;
const id = uuidv4();
let links: string[] = [req.body.url];
@ -47,24 +53,61 @@ export async function mapController(
const crawler = crawlToCrawler(id, sc);
const sitemap = req.body.ignoreSitemap ? null : await crawler.tryGetSitemap();
if (sitemap !== null) {
sitemap.map((x) => {
links.push(x.url);
});
}
let urlWithoutWww = req.body.url.replace("www.", "");
let mapUrl = req.body.search
? `"${req.body.search}" site:${urlWithoutWww}`
: `site:${req.body.url}`;
// www. seems to exclude subdomains in some cases
const mapResults = await fireEngineMap(mapUrl, {
// limit to 100 results (beta)
numResults: Math.min(limit, 100),
const resultsPerPage = 100;
const maxPages = Math.ceil(Math.min(MAX_FIRE_ENGINE_RESULTS, limit) / resultsPerPage);
const cacheKey = `fireEngineMap:${mapUrl}`;
const cachedResult = await redis.get(cacheKey);
let allResults: any[];
let pagePromises: Promise<any>[];
if (cachedResult) {
allResults = JSON.parse(cachedResult);
} else {
const fetchPage = async (page: number) => {
return fireEngineMap(mapUrl, {
numResults: resultsPerPage,
page: page,
});
};
pagePromises = Array.from({ length: maxPages }, (_, i) => fetchPage(i + 1));
allResults = await Promise.all(pagePromises);
await redis.set(cacheKey, JSON.stringify(allResults), "EX", 24 * 60 * 60); // Cache for 24 hours
}
// Parallelize sitemap fetch with serper search
const [sitemap, ...searchResults] = await Promise.all([
req.body.ignoreSitemap ? null : crawler.tryGetSitemap(),
...(cachedResult ? [] : pagePromises),
]);
if (!cachedResult) {
allResults = searchResults;
}
if (sitemap !== null) {
sitemap.forEach((x) => {
links.push(x.url);
});
}
let mapResults = allResults
.flat()
.filter((result) => result !== null && result !== undefined);
const minumumCutoff = Math.min(MAX_MAP_LIMIT, limit);
if (mapResults.length > minumumCutoff) {
mapResults = mapResults.slice(0, minumumCutoff);
}
if (mapResults.length > 0) {
if (req.body.search) {
@ -88,13 +131,15 @@ export async function mapController(
links = performCosineSimilarity(links, searchQuery);
}
links = links.map((x) => {
links = links
.map((x) => {
try {
return checkAndUpdateURLForMap(x).url.trim()
return checkAndUpdateURLForMap(x).url.trim();
} catch (_) {
return null;
}
}).filter(x => x !== null);
})
.filter((x) => x !== null);
// allows for subdomains to be included
links = links.filter((x) => isSameDomain(x, req.body.url));
@ -107,8 +152,10 @@ export async function mapController(
// remove duplicates that could be due to http/https or www
links = removeDuplicateUrls(links);
billTeam(req.auth.team_id, 1).catch(error => {
Logger.error(`Failed to bill team ${req.auth.team_id} for 1 credit: ${error}`);
billTeam(req.auth.team_id, 1).catch((error) => {
Logger.error(
`Failed to bill team ${req.auth.team_id} for 1 credit: ${error}`
);
// Optionally, you could notify an admin or add to a retry queue here
});
@ -140,3 +187,51 @@ export async function mapController(
scrape_id: req.body.origin?.includes("website") ? id : undefined,
});
}
// Subdomain sitemap url checking
// // For each result, check for subdomains, get their sitemaps and add them to the links
// const processedUrls = new Set();
// const processedSubdomains = new Set();
// for (const result of links) {
// let url;
// let hostParts;
// try {
// url = new URL(result);
// hostParts = url.hostname.split('.');
// } catch (e) {
// continue;
// }
// console.log("hostParts", hostParts);
// // Check if it's a subdomain (more than 2 parts, and not 'www')
// if (hostParts.length > 2 && hostParts[0] !== 'www') {
// const subdomain = hostParts[0];
// console.log("subdomain", subdomain);
// const subdomainUrl = `${url.protocol}//${subdomain}.${hostParts.slice(-2).join('.')}`;
// console.log("subdomainUrl", subdomainUrl);
// if (!processedSubdomains.has(subdomainUrl)) {
// processedSubdomains.add(subdomainUrl);
// const subdomainCrawl = crawlToCrawler(id, {
// originUrl: subdomainUrl,
// crawlerOptions: legacyCrawlerOptions(req.body),
// pageOptions: {},
// team_id: req.auth.team_id,
// createdAt: Date.now(),
// plan: req.auth.plan,
// });
// const subdomainSitemap = await subdomainCrawl.tryGetSitemap();
// if (subdomainSitemap) {
// subdomainSitemap.forEach((x) => {
// if (!processedUrls.has(x.url)) {
// processedUrls.add(x.url);
// links.push(x.url);
// }
// });
// }
// }
// }
// }

View File

@ -36,17 +36,15 @@ export async function getLinksFromSitemap(
const root = parsed.urlset || parsed.sitemapindex;
if (root && root.sitemap) {
for (const sitemap of root.sitemap) {
if (sitemap.loc && sitemap.loc.length > 0) {
await getLinksFromSitemap({ sitemapUrl: sitemap.loc[0], allUrls, mode });
}
}
const sitemapPromises = root.sitemap
.filter(sitemap => sitemap.loc && sitemap.loc.length > 0)
.map(sitemap => getLinksFromSitemap({ sitemapUrl: sitemap.loc[0], allUrls, mode }));
await Promise.all(sitemapPromises);
} else if (root && root.url) {
for (const url of root.url) {
if (url.loc && url.loc.length > 0 && !WebCrawler.prototype.isFile(url.loc[0])) {
allUrls.push(url.loc[0]);
}
}
const validUrls = root.url
.filter(url => url.loc && url.loc.length > 0 && !WebCrawler.prototype.isFile(url.loc[0]))
.map(url => url.loc[0]);
allUrls.push(...validUrls);
}
} catch (error) {
Logger.debug(`Error processing sitemapUrl: ${sitemapUrl} | Error: ${error.message}`);