enhance news crawler configuration with new sources and improved request headers
This commit is contained in:
+47
-17
@@ -19,6 +19,8 @@ const TRACKING_PARAM_PATTERNS = [
|
||||
const LISTING_PATH_HINT = /(archive|archives|latest|topic|topics|section|sections|category|categories|news|world|business|politics|technology|tech|markets|economy|page|tag|tags)/i;
|
||||
const ARTICLE_DATE_PATH = /\/\d{4}\/\d{2}\/\d{2}(?:\/|$)|\/\d{4}\/\d{2}(?:\/|$)/;
|
||||
const ARTICLE_PATH_HINT = /(\/article\/|\/articles\/|\/news\/|\/story\/|\/stories\/)/i;
|
||||
const ARTICLE_PATH_STRONG_HINT = /\/\d{4}\/\d{2}\/\d{2}\//;
|
||||
const LISTING_ARTICLE_FALSE_POSITIVE_PATH = /(\/category\/|\/tag\/|\/latest(?:\/|$)|\/topics?(?:\/|$)|\/sections?(?:\/|$))/i;
|
||||
const BLOCKED_PATH_HINT = /(\/search(?:\/|$)|\/login(?:\/|$)|\/account(?:\/|$)|\/video(?:\/|$)|\/videos(?:\/|$)|\/podcast(?:\/|$)|\/podcasts(?:\/|$)|\/live(?:\/|$))/i;
|
||||
|
||||
function decodeHtmlEntities(value) {
|
||||
@@ -267,30 +269,36 @@ function selectPubDate(meta, jsonLdArticle, html) {
|
||||
function scorePage(pageUrl, meta, html, jsonLdArticle, links) {
|
||||
let articleScore = 0;
|
||||
let listingScore = 0;
|
||||
const pathname = new URL(pageUrl).pathname;
|
||||
const hasArticleDatePath = ARTICLE_DATE_PATH.test(pageUrl);
|
||||
const hasArticlePathHint = ARTICLE_PATH_HINT.test(pageUrl);
|
||||
const hasStrongArticlePath = ARTICLE_PATH_STRONG_HINT.test(pathname);
|
||||
const hasListingFalsePositivePath = LISTING_ARTICLE_FALSE_POSITIVE_PATH.test(pathname);
|
||||
const paragraphTextLength = extractParagraphTextLength(html);
|
||||
const headlineLinks = links.filter(({ text }) => text.length >= 25 && text.length <= 180).length;
|
||||
|
||||
if (jsonLdArticle) {
|
||||
articleScore += 3;
|
||||
articleScore += 4;
|
||||
}
|
||||
|
||||
if (String(meta.get('og:type') || '').toLowerCase() === 'article') {
|
||||
articleScore += 2;
|
||||
if (String(meta.get('og:type') || '').toLowerCase() === 'article' && !hasListingFalsePositivePath) {
|
||||
articleScore += 1;
|
||||
}
|
||||
|
||||
if (meta.get('article:published_time') || meta.get('og:article:published_time') || extractTimeDatetime(html)) {
|
||||
articleScore += 2;
|
||||
if ((meta.get('article:published_time') || meta.get('og:article:published_time') || extractTimeDatetime(html)) && !hasListingFalsePositivePath) {
|
||||
articleScore += 1;
|
||||
}
|
||||
|
||||
if (/<article\b/i.test(html)) {
|
||||
articleScore += 1;
|
||||
}
|
||||
|
||||
if (ARTICLE_DATE_PATH.test(pageUrl) || ARTICLE_PATH_HINT.test(pageUrl)) {
|
||||
articleScore += 1;
|
||||
if (hasArticleDatePath || hasArticlePathHint) {
|
||||
articleScore += 2;
|
||||
}
|
||||
|
||||
if (extractH1(html) && extractParagraphTextLength(html) >= 500) {
|
||||
articleScore += 1;
|
||||
if (extractH1(html) && paragraphTextLength >= 500) {
|
||||
articleScore += 2;
|
||||
}
|
||||
|
||||
if (links.length >= 20) {
|
||||
@@ -301,15 +309,23 @@ function scorePage(pageUrl, meta, html, jsonLdArticle, links) {
|
||||
listingScore += 2;
|
||||
}
|
||||
|
||||
if (LISTING_PATH_HINT.test(new URL(pageUrl).pathname)) {
|
||||
if (LISTING_PATH_HINT.test(pathname)) {
|
||||
listingScore += 1;
|
||||
}
|
||||
|
||||
if (articleScore > 0) {
|
||||
listingScore -= 2;
|
||||
if (hasListingFalsePositivePath) {
|
||||
listingScore += 3;
|
||||
}
|
||||
|
||||
return { articleScore, listingScore };
|
||||
if (articleScore > 0) {
|
||||
listingScore -= 1;
|
||||
}
|
||||
|
||||
const isArticleCandidate = articleScore >= 4
|
||||
&& articleScore > listingScore
|
||||
&& (Boolean(jsonLdArticle) || hasStrongArticlePath || hasArticlePathHint || paragraphTextLength >= 500);
|
||||
|
||||
return { articleScore, listingScore, isArticleCandidate };
|
||||
}
|
||||
|
||||
function shouldQueueLink(url) {
|
||||
@@ -396,6 +412,20 @@ function buildDefaultSeeds(feedUrl) {
|
||||
}
|
||||
}
|
||||
|
||||
function normalizeLimit(value, fallback, minimum, maximum) {
|
||||
const numeric = Number(value);
|
||||
|
||||
if (numeric === -1) {
|
||||
return Number.POSITIVE_INFINITY;
|
||||
}
|
||||
|
||||
if (!Number.isFinite(numeric)) {
|
||||
return fallback;
|
||||
}
|
||||
|
||||
return Math.max(minimum, Math.min(numeric, maximum));
|
||||
}
|
||||
|
||||
function normalizeSite(site) {
|
||||
const allowedHosts = unique((site.allowedHosts || []).map((host) => String(host || '').toLowerCase()).filter(Boolean));
|
||||
const seeds = unique((site.seeds || [])
|
||||
@@ -407,8 +437,8 @@ function normalizeSite(site) {
|
||||
label: String(site.label || '').trim(),
|
||||
allowedHosts,
|
||||
seeds,
|
||||
maxPages: Math.max(1, Math.min(Number(site.maxPages) || 15, 500)),
|
||||
maxDepth: Math.max(0, Math.min(Number(site.maxDepth) || 1, 5)),
|
||||
maxPages: normalizeLimit(site.maxPages, 15, 1, 500),
|
||||
maxDepth: normalizeLimit(site.maxDepth, 1, 0, 5),
|
||||
requestTimeout: Math.max(1000, Math.min(Number(site.requestTimeout) || 15000, 30000)),
|
||||
};
|
||||
}
|
||||
@@ -515,9 +545,9 @@ async function crawlSite(site) {
|
||||
? canonicalizeUrl(canonicalHref, current.url, normalizedSite.allowedHosts) || current.url
|
||||
: current.url;
|
||||
const links = extractLinks(html, canonicalUrl, normalizedSite.allowedHosts);
|
||||
const { articleScore, listingScore } = scorePage(canonicalUrl, meta, html, jsonLdArticle, links);
|
||||
const { listingScore, isArticleCandidate } = scorePage(canonicalUrl, meta, html, jsonLdArticle, links);
|
||||
|
||||
if (articleScore >= 3 && !discoveredArticleUrls.has(canonicalUrl)) {
|
||||
if (isArticleCandidate && !discoveredArticleUrls.has(canonicalUrl)) {
|
||||
const title = normalizeText(selectTitle(meta, jsonLdArticle, html));
|
||||
if (title) {
|
||||
discoveredArticleUrls.add(canonicalUrl);
|
||||
|
||||
Reference in New Issue
Block a user