cli-jaw 2.0.5 → 2.0.6

package/dist/src/browser/adaptive-fetch/browser-session.js

@@ -0,0 +1,97 @@
+// @ts-nocheck
+// Mirrored from agbrowse adaptive-fetch v2; keep runtime behavior aligned while cli-jaw mirror remains experimental.
+import { getFetchBrowserPage, closeFetchBrowserPage } from './browser-runtime.js';
+import { validateFetchUrl } from './safety.js';
+/**
+ * @param {{ browserDeps?: any }} [options]
+ * @returns {boolean}
+ */
+export function isUserSessionAvailable(options = {}) {
+    const deps = options?.browserDeps || {};
+    return typeof deps.getPage === 'function';
+}
+/**
+ * @param {any[]} candidates
+ * @param {{ browserSession?: string, browserSessionRaw?: string, browserDeps?: any }} options
+ * @returns {boolean|'prompt'}
+ */
+export function shouldTryUserSession(candidates, options) {
+    const rawSession = options.browserSessionRaw || options.browserSession;
+    if (rawSession === 'user' || rawSession === 'interactive')
+        return true;
+    const hasChallenge = candidates.some(c => c.challenge?.type === 'challenge' ||
+        c.challenge?.type === 'auth_required' ||
+        c.challenge?.type === 'paywall');
+    if (hasChallenge && isUserSessionAvailable(options))
+        return 'prompt';
+    return false;
+}
+/**
+ * @param {string} url
+ * @param {{ browserDeps?: any, timeoutMs?: number, selector?: string|null, allowPrivateNetwork?: boolean }} options
+ * @returns {Promise<{ source: string, finalUrl: string, title: string, text: string, contentType: string, status: number, ok: boolean, session: string, evidence: string[], warnings: string[], safetyFlags: string[] }>}
+ */
+export async function navigateInUserSession(url, options) {
+    const pageRef = await getFetchBrowserPage({
+        browserDeps: options.browserDeps,
+        browserSession: 'existing',
+    });
+    try {
+        const page = pageRef.page;
+        let navStatus = 200;
+        let navOk = true;
+        if (typeof page.goto === 'function') {
+            const response = await page.goto(url, { waitUntil: 'networkidle', timeout: options.timeoutMs || 15000 });
+            if (response) {
+                navStatus = typeof response.status === 'function' ? response.status() : (response.status || 200);
+                navOk = navStatus >= 200 && navStatus < 400;
+            }
+        }
+        const title = typeof page.title === 'function' ? await page.title() : '';
+        let text = '';
+        if (options.selector && typeof page.locator === 'function') {
+            text = await page.locator(options.selector).first().innerText({ timeout: 2000 }).catch(() => '');
+        }
+        else if (typeof page.evaluate === 'function') {
+            text = await page.evaluate(() => document.body?.innerText || '');
+        }
+        const finalUrl = typeof page.url === 'function' ? page.url() : url;
+        const warnings = [];
+        try {
+            validateFetchUrl(finalUrl, { allowPrivateNetwork: options.allowPrivateNetwork });
+        }
+        catch {
+            warnings.push('user-session-redirected-to-private-url');
+            return {
+                source: 'browser_user',
+                finalUrl,
+                title,
+                text: '',
+                contentType: 'text/html',
+                status: navStatus,
+                ok: false,
+                session: 'user',
+                evidence: ['user-session-render', 'private-url-rejected'],
+                warnings,
+                safetyFlags: ['user_session_used'],
+            };
+        }
+        return {
+            source: 'browser_user',
+            finalUrl,
+            title,
+            text,
+            contentType: 'text/html',
+            status: navStatus,
+            ok: navOk,
+            session: 'user',
+            evidence: ['user-session-render'],
+            warnings,
+            safetyFlags: ['user_session_used'],
+        };
+    }
+    finally {
+        await closeFetchBrowserPage(pageRef);
+    }
+}
+//# sourceMappingURL=browser-session.js.map

package/dist/src/browser/adaptive-fetch/browser-session.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"browser-session.js","sourceRoot":"","sources":["../../../../src/browser/adaptive-fetch/browser-session.ts"],"names":[],"mappings":"AAAA,cAAc;AACd,qHAAqH;AAErH,OAAO,EAAE,mBAAmB,EAAE,qBAAqB,EAAE,MAAM,sBAAsB,CAAC;AAClF,OAAO,EAAE,gBAAgB,EAAE,MAAM,aAAa,CAAC;AAE/C;;;GAGG;AACH,MAAM,UAAU,sBAAsB,CAAC,OAAO,GAAG,EAAE;IAC/C,MAAM,IAAI,GAAG,OAAO,EAAE,WAAW,IAAI,EAAE,CAAC;IACxC,OAAO,OAAO,IAAI,CAAC,OAAO,KAAK,UAAU,CAAC;AAC9C,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,oBAAoB,CAAC,UAAU,EAAE,OAAO;IACpD,MAAM,UAAU,GAAG,OAAO,CAAC,iBAAiB,IAAI,OAAO,CAAC,cAAc,CAAC;IACvE,IAAI,UAAU,KAAK,MAAM,IAAI,UAAU,KAAK,aAAa;QAAE,OAAO,IAAI,CAAC;IACvE,MAAM,YAAY,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CACrC,CAAC,CAAC,SAAS,EAAE,IAAI,KAAK,WAAW;QACjC,CAAC,CAAC,SAAS,EAAE,IAAI,KAAK,eAAe;QACrC,CAAC,CAAC,SAAS,EAAE,IAAI,KAAK,SAAS,CAClC,CAAC;IACF,IAAI,YAAY,IAAI,sBAAsB,CAAC,OAAO,CAAC;QAAE,OAAO,QAAQ,CAAC;IACrE,OAAO,KAAK,CAAC;AACjB,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,qBAAqB,CAAC,GAAG,EAAE,OAAO;IACpD,MAAM,OAAO,GAAG,MAAM,mBAAmB,CAAC;QACtC,WAAW,EAAE,OAAO,CAAC,WAAW;QAChC,cAAc,EAAE,UAAU;KAC7B,CAAC,CAAC;IACH,IAAI,CAAC;QACD,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,CAAC;QAC1B,IAAI,SAAS,GAAG,GAAG,CAAC;QACpB,IAAI,KAAK,GAAG,IAAI,CAAC;QACjB,IAAI,OAAO,IAAI,CAAC,IAAI,KAAK,UAAU,EAAE,CAAC;YAClC,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,IAAI,CAAC,GAAG,EAAE,EAAE,SAAS,EAAE,aAAa,EAAE,OAAO,EAAE,OAAO,CAAC,SAAS,IAAI,KAAK,EAAE,CAAC,CAAC;YACzG,IAAI,QAAQ,EAAE,CAAC;gBACX,SAAS,GAAG,OAAO,QAAQ,CAAC,MAAM,KAAK,UAAU,CAAC,CAAC,CAAC,QAAQ,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,MAAM,IAAI,GAAG,CAAC,CAAC;gBACjG,KAAK,GAAG,SAAS,IAAI,GAAG,IAAI,SAAS,GAAG,GAAG,CAAC;YAChD,CAAC;QACL,CAAC;QACD,MAAM,KAAK,GAAG,OAAO,IAAI,CAAC,KAAK,KAAK,UAAU,CAAC,CAAC,CAAC,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QACzE,IAAI,IAAI,GAAG,EAAE,CAAC;QACd,IAAI,OAAO,CAAC,QAAQ,IAAI,OAAO,IAAI,CAAC,OAAO,KAAK,UAAU,EAAE,CAAC;YACzD,IAAI,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,KAAK,EAAE,CAAC,SAAS,CAAC,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,EAAE,CAAC,CAAC;QACrG,CAAC;aAAM,IAAI,OAAO,IAAI,CAAC,QAAQ,KAAK,UAAU,EAAE,CAAC;YAC7C,IAAI,GAAG,MAAM,IAAI,CAAC,QAAQ,CAAC,GAAG,EAAE,CAAC,QAAQ,CAAC,IAAI,EAAE,SAAS,IAAI,EAAE,CAAC,CAAC;QACrE,CAAC;QACD,MAAM,QAAQ,GAAG,OAAO,IAAI,CAAC,GAAG,KAAK,UAAU,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC;QACnE,MAAM,QAAQ,GAAG,EAAE,CAAC;QACpB,IAAI,CAAC;YACD,gBAAgB,CAAC,QAAQ,EAAE,EAAE,mBAAmB,EAAE,OAAO,CAAC,mBAAmB,EAAE,CAAC,CAAC;QACrF,CAAC;QAAC,MAAM,CAAC;YACL,QAAQ,CAAC,IAAI,CAAC,wCAAwC,CAAC,CAAC;YACxD,OAAO;gBACH,MAAM,EAAE,cAAc;gBACtB,QAAQ;gBACR,KAAK;gBACL,IAAI,EAAE,EAAE;gBACR,WAAW,EAAE,WAAW;gBACxB,MAAM,EAAE,SAAS;gBACjB,EAAE,EAAE,KAAK;gBACT,OAAO,EAAE,MAAM;gBACf,QAAQ,EAAE,CAAC,qBAAqB,EAAE,sBAAsB,CAAC;gBACzD,QAAQ;gBACR,WAAW,EAAE,CAAC,mBAAmB,CAAC;aACrC,CAAC;QACN,CAAC;QACD,OAAO;YACH,MAAM,EAAE,cAAc;YACtB,QAAQ;YACR,KAAK;YACL,IAAI;YACJ,WAAW,EAAE,WAAW;YACxB,MAAM,EAAE,SAAS;YACjB,EAAE,EAAE,KAAK;YACT,OAAO,EAAE,MAAM;YACf,QAAQ,EAAE,CAAC,qBAAqB,CAAC;YACjC,QAAQ;YACR,WAAW,EAAE,CAAC,mBAAmB,CAAC;SACrC,CAAC;IACN,CAAC;YAAS,CAAC;QACP,MAAM,qBAAqB,CAAC,OAAO,CAAC,CAAC;IACzC,CAAC;AACL,CAAC"}

package/dist/src/browser/adaptive-fetch/challenge-detector.js

@@ -0,0 +1,88 @@
+// @ts-nocheck
+// Mirrored from agbrowse adaptive-fetch v2; keep runtime behavior aligned while cli-jaw mirror remains experimental.
+import { findBoundaryMarkers } from './validators.js';
+import { WAF_PROFILES, scoreProfile } from './waf-profiles.js';
+/**
+ * @param {{ url?: string, status?: number, text?: string, title?: string }} input
+ */
+export function detectChallengeMarkers(input = {}) {
+    return findBoundaryMarkers(`${input.url || ''}\n${input.status || ''}\n${input.title || ''}\n${input.text || ''}`);
+}
+/**
+ * @param {{ kind: string }[]} markers
+ */
+export function classifyAccessBoundary(markers = []) {
+    if (markers.some(marker => marker.kind === 'auth'))
+        return 'auth_required';
+    if (markers.some(marker => marker.kind === 'paywall'))
+        return 'paywall';
+    if (markers.some(marker => marker.kind === 'challenge'))
+        return 'challenge';
+    return null;
+}
+/**
+ * @param {{ status?: number, headers?: Record<string, string>, body?: string, cookies?: string[] }} response
+ */
+export function detectWafChallenge(response) {
+    const signals = {
+        cookies: response.cookies || parseCookieNames(response.headers || {}),
+        headers: response.headers || {},
+        body: response.body?.substring(0, 50000) ?? '',
+        status: response.status || 0,
+    };
+    const matches = WAF_PROFILES
+        .map(p => ({ profile: p, score: scoreProfile(p, signals) }))
+        .filter(m => m.score > 0)
+        .sort((a, b) => b.score - a.score);
+    return {
+        detected: matches.length > 0,
+        profiles: matches,
+        primary: matches[0] ?? null,
+        signals: { cookieCount: signals.cookies.length, status: signals.status },
+    };
+}
+/**
+ * @param {string} text
+ */
+export function detectLoginWall(text) {
+    const markers = findBoundaryMarkers(text);
+    const authMarkers = markers.filter(m => m.kind === 'auth');
+    return { detected: authMarkers.length > 0, markers: authMarkers };
+}
+/**
+ * @param {string} text
+ */
+export function detectPaywall(text) {
+    const markers = findBoundaryMarkers(text);
+    const paywallMarkers = markers.filter(m => m.kind === 'paywall');
+    return { detected: paywallMarkers.length > 0, markers: paywallMarkers };
+}
+/**
+ * @param {{ status?: number, headers?: Record<string, string>, body?: string, cookies?: string[] }} response
+ */
+export function classifyChallengeType(response) {
+    const waf = detectWafChallenge(response);
+    const textContent = (response.body || '').substring(0, 50000);
+    const login = detectLoginWall(textContent);
+    const paywall = detectPaywall(textContent);
+    if (waf.detected)
+        return { type: 'challenge', ...waf };
+    if (login.detected)
+        return { type: /** @type {const} */ ('auth_required'), ...login };
+    if (paywall.detected)
+        return { type: /** @type {const} */ ('paywall'), ...paywall };
+    return { type: null };
+}
+/**
+ * @param {Record<string, string>} headers
+ */
+function parseCookieNames(headers) {
+    const setCookie = headers['set-cookie'] || headers['Set-Cookie'] || '';
+    if (!setCookie)
+        return [];
+    return setCookie.split(/,(?=[^;]*=)/).map(c => {
+        const name = c.trim().split('=')[0];
+        return name || '';
+    }).filter(Boolean);
+}
+//# sourceMappingURL=challenge-detector.js.map

package/dist/src/browser/adaptive-fetch/challenge-detector.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"challenge-detector.js","sourceRoot":"","sources":["../../../../src/browser/adaptive-fetch/challenge-detector.ts"],"names":[],"mappings":"AAAA,cAAc;AACd,qHAAqH;AAErH,OAAO,EAAE,mBAAmB,EAAE,MAAM,iBAAiB,CAAC;AACtD,OAAO,EAAE,YAAY,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AAE/D;;GAEG;AACH,MAAM,UAAU,sBAAsB,CAAC,KAAK,GAAG,EAAE;IAC7C,OAAO,mBAAmB,CAAC,GAAG,KAAK,CAAC,GAAG,IAAI,EAAE,KAAK,KAAK,CAAC,MAAM,IAAI,EAAE,KAAK,KAAK,CAAC,KAAK,IAAI,EAAE,KAAK,KAAK,CAAC,IAAI,IAAI,EAAE,EAAE,CAAC,CAAC;AACvH,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,sBAAsB,CAAC,OAAO,GAAG,EAAE;IAC/C,IAAI,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,CAAC,IAAI,KAAK,MAAM,CAAC;QAAE,OAAO,eAAe,CAAC;IAC3E,IAAI,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,CAAC,IAAI,KAAK,SAAS,CAAC;QAAE,OAAO,SAAS,CAAC;IACxE,IAAI,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,CAAC,IAAI,KAAK,WAAW,CAAC;QAAE,OAAO,WAAW,CAAC;IAC5E,OAAO,IAAI,CAAC;AAChB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,kBAAkB,CAAC,QAAQ;IACvC,MAAM,OAAO,GAAG;QACZ,OAAO,EAAE,QAAQ,CAAC,OAAO,IAAI,gBAAgB,CAAC,QAAQ,CAAC,OAAO,IAAI,EAAE,CAAC;QACrE,OAAO,EAAE,QAAQ,CAAC,OAAO,IAAI,EAAE;QAC/B,IAAI,EAAE,QAAQ,CAAC,IAAI,EAAE,SAAS,CAAC,CAAC,EAAE,KAAK,CAAC,IAAI,EAAE;QAC9C,MAAM,EAAE,QAAQ,CAAC,MAAM,IAAI,CAAC;KAC/B,CAAC;IAEF,MAAM,OAAO,GAAG,YAAY;SACvB,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,KAAK,EAAE,YAAY,CAAC,CAAC,EAAE,OAAO,CAAC,EAAE,CAAC,CAAC;SAC3D,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC;SACxB,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC;IAEvC,OAAO;QACH,QAAQ,EAAE,OAAO,CAAC,MAAM,GAAG,CAAC;QAC5B,QAAQ,EAAE,OAAO;QACjB,OAAO,EAAE,OAAO,CAAC,CAAC,CAAC,IAAI,IAAI;QAC3B,OAAO,EAAE,EAAE,WAAW,EAAE,OAAO,CAAC,OAAO,CAAC,MAAM,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE;KAC3E,CAAC;AACN,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,eAAe,CAAC,IAAI;IAChC,MAAM,OAAO,GAAG,mBAAmB,CAAC,IAAI,CAAC,CAAC;IAC1C,MAAM,WAAW,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,MAAM,CAAC,CAAC;IAC3D,OAAO,EAAE,QAAQ,EAAE,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,OAAO,EAAE,WAAW,EAAE,CAAC;AACtE,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,aAAa,CAAC,IAAI;IAC9B,MAAM,OAAO,GAAG,mBAAmB,CAAC,IAAI,CAAC,CAAC;IAC1C,MAAM,cAAc,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC;IACjE,OAAO,EAAE,QAAQ,EAAE,cAAc,CAAC,MAAM,GAAG,CAAC,EAAE,OAAO,EAAE,cAAc,EAAE,CAAC;AAC5E,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,qBAAqB,CAAC,QAAQ;IAC1C,MAAM,GAAG,GAAG,kBAAkB,CAAC,QAAQ,CAAC,CAAC;IACzC,MAAM,WAAW,GAAG,CAAC,QAAQ,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,SAAS,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;IAC9D,MAAM,KAAK,GAAG,eAAe,CAAC,WAAW,CAAC,CAAC;IAC3C,MAAM,OAAO,GAAG,aAAa,CAAC,WAAW,CAAC,CAAC;IAE3C,IAAI,GAAG,CAAC,QAAQ;QAAE,OAAO,EAAE,IAAI,EAAE,WAAW,EAAE,GAAG,GAAG,EAAE,CAAC;IACvD,IAAI,KAAK,CAAC,QAAQ;QAAE,OAAO,EAAE,IAAI,EAAE,oBAAoB,CAAC,CAAC,eAAe,CAAC,EAAE,GAAG,KAAK,EAAE,CAAC;IACtF,IAAI,OAAO,CAAC,QAAQ;QAAE,OAAO,EAAE,IAAI,EAAE,oBAAoB,CAAC,CAAC,SAAS,CAAC,EAAE,GAAG,OAAO,EAAE,CAAC;IACpF,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;AAC1B,CAAC;AAED;;GAEG;AACH,SAAS,gBAAgB,CAAC,OAAO;IAC7B,MAAM,SAAS,GAAG,OAAO,CAAC,YAAY,CAAC,IAAI,OAAO,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC;IACvE,IAAI,CAAC,SAAS;QAAE,OAAO,EAAE,CAAC;IAC1B,OAAO,SAAS,CAAC,KAAK,CAAC,aAAa,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE;QAC1C,MAAM,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;QACpC,OAAO,IAAI,IAAI,EAAE,CAAC;IACtB,CAAC,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;AACvB,CAAC"}

package/dist/src/browser/adaptive-fetch/content-scorer.js

@@ -0,0 +1,145 @@
+// @ts-nocheck
+// Mirrored from agbrowse adaptive-fetch v2; keep runtime behavior aligned while cli-jaw mirror remains experimental.
+import { classifyBoundarySignals, findBoundaryMarkers } from './validators.js';
+const SOURCE_TRUST = {
+    public_endpoint: 20,
+    network_api: 16,
+    fetch: 12,
+    browser: 10,
+    metadata: 6,
+    third_party_reader: 5,
+    reader: 4,
+};
+/**
+ * @param {any} candidate
+ * @param {{ minStrongScore?: number, minWeakScore?: number }} [options]
+ */
+export function scoreReaderCandidate(candidate, options = {}) {
+    const text = String(candidate.text || '');
+    const title = String(candidate.title || '');
+    const metadata = candidate.metadata || {};
+    const markers = findBoundaryMarkers(`${title}\n${text}`);
+    const textLength = text.length;
+    const density = computeTextDensity(text, Number(candidate.rawTextLength || textLength));
+    const metadataEvidence = countMetadataEvidence(metadata);
+    const boundary = classifyBoundarySignals({
+        status: candidate.status,
+        text: `${title}\n${text}`,
+    });
+    let score = 0;
+    score += Math.min(45, Math.floor(textLength / 80));
+    score += Math.min(20, Math.floor(density * 20));
+    score += title.length >= 8 ? 8 : 0;
+    score += Math.min(15, metadataEvidence * 5);
+    score += SOURCE_TRUST[candidate.source] || 0;
+    if (candidate.ok === false || Number(candidate.status || 0) >= 400) {
+        score = 0;
+        const verdict = boundary.verdict || 'blocked';
+        return {
+            candidate,
+            score,
+            verdict,
+            markers: boundary.markers || markers,
+            textLength,
+            density,
+            evidence: buildScoreEvidence(candidate, {
+                score,
+                textLength,
+                density,
+                metadataEvidence,
+                markers: boundary.markers || markers,
+                extra: ['candidate-not-ok', candidate.status ? `http-${candidate.status}` : null, boundary.reason],
+            }),
+        };
+    }
+    for (const marker of markers) {
+        if (marker.kind === 'challenge')
+            score -= 25;
+        if (marker.kind === 'auth')
+            score -= 35;
+        if (marker.kind === 'paywall')
+            score -= 25;
+    }
+    score = Math.max(0, score);
+    const verdict = verdictFromScore({ score, markers, textLength }, options);
+    return {
+        candidate,
+        score,
+        verdict,
+        markers,
+        textLength,
+        density,
+        evidence: buildScoreEvidence(candidate, { score, textLength, density, metadataEvidence, markers }),
+    };
+}
+/**
+ * @param {any[]} candidates
+ * @param {Record<string, unknown>} [options]
+ */
+export function chooseBestReaderCandidate(candidates = [], options = {}) {
+    const scored = candidates.map(candidate => scoreReaderCandidate(candidate, options));
+    scored.sort((a, b) => b.score - a.score);
+    return scored[0] || null;
+}
+/**
+ * @param {{ score: number, markers?: any[], textLength?: number }} scored
+ * @param {{ minStrongScore?: number, minWeakScore?: number }} [options]
+ */
+export function verdictFromScore(scored, options = {}) {
+    const minStrongScore = Number(options.minStrongScore || 50);
+    const minWeakScore = Number(options.minWeakScore || 20);
+    const markers = scored.markers || [];
+    const textLength = Number(scored.textLength || 0);
+    if (markers.some(m => m.kind === 'auth') && textLength < 1500)
+        return 'auth_required';
+    if (markers.some(m => m.kind === 'paywall') && textLength < 1500)
+        return 'paywall';
+    if (markers.some(m => m.kind === 'challenge') && textLength < 1000)
+        return 'challenge';
+    if (scored.score >= minStrongScore)
+        return 'strong_ok';
+    if (scored.score >= minWeakScore)
+        return 'weak_ok';
+    return 'blocked';
+}
+/**
+ * @param {string} text
+ * @param {number} rawLength
+ */
+function computeTextDensity(text, rawLength) {
+    if (!rawLength || rawLength <= 0)
+        return text.length > 0 ? 1 : 0;
+    return Math.max(0, Math.min(1, text.length / rawLength));
+}
+/**
+ * @param {any} metadata
+ */
+function countMetadataEvidence(metadata) {
+    let count = 0;
+    if (metadata.canonicalUrl)
+        count += 1;
+    if (metadata.description)
+        count += 1;
+    if (metadata.openGraph && Object.keys(metadata.openGraph).length > 0)
+        count += 1;
+    if (Array.isArray(metadata.jsonLd) && metadata.jsonLd.length > 0)
+        count += 1;
+    return count;
+}
+/**
+ * @param {any} candidate
+ * @param {{ score: number, textLength: number, density: number, metadataEvidence: number, markers: any[], extra?: any[] }} scored
+ */
+function buildScoreEvidence(candidate, scored) {
+    return [
+        `score:${scored.score}`,
+        `source:${candidate.source}`,
+        `text:${scored.textLength}`,
+        `density:${scored.density.toFixed(2)}`,
+        scored.metadataEvidence ? `metadata:${scored.metadataEvidence}` : null,
+        ...scored.markers.map(marker => `marker:${marker.kind}`),
+        ...(scored.extra || []),
+        ...(candidate.evidence || []),
+    ].filter(Boolean);
+}
+//# sourceMappingURL=content-scorer.js.map

package/dist/src/browser/adaptive-fetch/content-scorer.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"content-scorer.js","sourceRoot":"","sources":["../../../../src/browser/adaptive-fetch/content-scorer.ts"],"names":[],"mappings":"AAAA,cAAc;AACd,qHAAqH;AAErH,OAAO,EAAE,uBAAuB,EAAE,mBAAmB,EAAE,MAAM,iBAAiB,CAAC;AAE/E,MAAM,YAAY,GAAG;IACjB,eAAe,EAAE,EAAE;IACnB,WAAW,EAAE,EAAE;IACf,KAAK,EAAE,EAAE;IACT,OAAO,EAAE,EAAE;IACX,QAAQ,EAAE,CAAC;IACX,kBAAkB,EAAE,CAAC;IACrB,MAAM,EAAE,CAAC;CACZ,CAAC;AAEF;;;GAGG;AACH,MAAM,UAAU,oBAAoB,CAAC,SAAS,EAAE,OAAO,GAAG,EAAE;IACxD,MAAM,IAAI,GAAG,MAAM,CAAC,SAAS,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC;IAC1C,MAAM,KAAK,GAAG,MAAM,CAAC,SAAS,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC;IAC5C,MAAM,QAAQ,GAAG,SAAS,CAAC,QAAQ,IAAI,EAAE,CAAC;IAC1C,MAAM,OAAO,GAAG,mBAAmB,CAAC,GAAG,KAAK,KAAK,IAAI,EAAE,CAAC,CAAC;IACzD,MAAM,UAAU,GAAG,IAAI,CAAC,MAAM,CAAC;IAC/B,MAAM,OAAO,GAAG,kBAAkB,CAAC,IAAI,EAAE,MAAM,CAAC,SAAS,CAAC,aAAa,IAAI,UAAU,CAAC,CAAC,CAAC;IACxF,MAAM,gBAAgB,GAAG,qBAAqB,CAAC,QAAQ,CAAC,CAAC;IACzD,MAAM,QAAQ,GAAG,uBAAuB,CAAC;QACrC,MAAM,EAAE,SAAS,CAAC,MAAM;QACxB,IAAI,EAAE,GAAG,KAAK,KAAK,IAAI,EAAE;KAC5B,CAAC,CAAC;IACH,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,IAAI,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,CAAC,KAAK,CAAC,UAAU,GAAG,EAAE,CAAC,CAAC,CAAC;IACnD,KAAK,IAAI,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,CAAC,KAAK,CAAC,OAAO,GAAG,EAAE,CAAC,CAAC,CAAC;IAChD,KAAK,IAAI,KAAK,CAAC,MAAM,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACnC,KAAK,IAAI,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,gBAAgB,GAAG,CAAC,CAAC,CAAC;IAC5C,KAAK,IAAI,YAAY,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;IAC7C,IAAI,SAAS,CAAC,EAAE,KAAK,KAAK,IAAI,MAAM,CAAC,SAAS,CAAC,MAAM,IAAI,CAAC,CAAC,IAAI,GAAG,EAAE,CAAC;QACjE,KAAK,GAAG,CAAC,CAAC;QACV,MAAM,OAAO,GAAG,QAAQ,CAAC,OAAO,IAAI,SAAS,CAAC;QAC9C,OAAO;YACH,SAAS;YACT,KAAK;YACL,OAAO;YACP,OAAO,EAAE,QAAQ,CAAC,OAAO,IAAI,OAAO;YACpC,UAAU;YACV,OAAO;YACP,QAAQ,EAAE,kBAAkB,CAAC,SAAS,EAAE;gBACpC,KAAK;gBACL,UAAU;gBACV,OAAO;gBACP,gBAAgB;gBAChB,OAAO,EAAE,QAAQ,CAAC,OAAO,IAAI,OAAO;gBACpC,KAAK,EAAE,CAAC,kBAAkB,EAAE,SAAS,CAAC,MAAM,CAAC,CAAC,CAAC,QAAQ,SAAS,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,QAAQ,CAAC,MAAM,CAAC;aACrG,CAAC;SACL,CAAC;IACN,CAAC;IACD,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;QAC3B,IAAI,MAAM,CAAC,IAAI,KAAK,WAAW;YAAE,KAAK,IAAI,EAAE,CAAC;QAC7C,IAAI,MAAM,CAAC,IAAI,KAAK,MAAM;YAAE,KAAK,IAAI,EAAE,CAAC;QACxC,IAAI,MAAM,CAAC,IAAI,KAAK,SAAS;YAAE,KAAK,IAAI,EAAE,CAAC;IAC/C,CAAC;IACD,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;IAC3B,MAAM,OAAO,GAAG,gBAAgB,CAAC,EAAE,KAAK,EAAE,OAAO,EAAE,UAAU,EAAE,EAAE,OAAO,CAAC,CAAC;IAC1E,OAAO;QACH,SAAS;QACT,KAAK;QACL,OAAO;QACP,OAAO;QACP,UAAU;QACV,OAAO;QACP,QAAQ,EAAE,kBAAkB,CAAC,SAAS,EAAE,EAAE,KAAK,EAAE,UAAU,EAAE,OAAO,EAAE,gBAAgB,EAAE,OAAO,EAAE,CAAC;KACrG,CAAC;AACN,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,yBAAyB,CAAC,UAAU,GAAG,EAAE,EAAE,OAAO,GAAG,EAAE;IACnE,MAAM,MAAM,GAAG,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC,oBAAoB,CAAC,SAAS,EAAE,OAAO,CAAC,CAAC,CAAC;IACrF,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC;IACzC,OAAO,MAAM,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC;AAC7B,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,gBAAgB,CAAC,MAAM,EAAE,OAAO,GAAG,EAAE;IACjD,MAAM,cAAc,GAAG,MAAM,CAAC,OAAO,CAAC,cAAc,IAAI,EAAE,CAAC,CAAC;IAC5D,MAAM,YAAY,GAAG,MAAM,CAAC,OAAO,CAAC,YAAY,IAAI,EAAE,CAAC,CAAC;IACxD,MAAM,OAAO,GAAG,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC;IACrC,MAAM,UAAU,GAAG,MAAM,CAAC,MAAM,CAAC,UAAU,IAAI,CAAC,CAAC,CAAC;IAClD,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,MAAM,CAAC,IAAI,UAAU,GAAG,IAAI;QAAE,OAAO,eAAe,CAAC;IACtF,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,SAAS,CAAC,IAAI,UAAU,GAAG,IAAI;QAAE,OAAO,SAAS,CAAC;IACnF,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,WAAW,CAAC,IAAI,UAAU,GAAG,IAAI;QAAE,OAAO,WAAW,CAAC;IACvF,IAAI,MAAM,CAAC,KAAK,IAAI,cAAc;QAAE,OAAO,WAAW,CAAC;IACvD,IAAI,MAAM,CAAC,KAAK,IAAI,YAAY;QAAE,OAAO,SAAS,CAAC;IACnD,OAAO,SAAS,CAAC;AACrB,CAAC;AAED;;;GAGG;AACH,SAAS,kBAAkB,CAAC,IAAI,EAAE,SAAS;IACvC,IAAI,CAAC,SAAS,IAAI,SAAS,IAAI,CAAC;QAAE,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACjE,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,SAAS,CAAC,CAAC,CAAC;AAC7D,CAAC;AAED;;GAEG;AACH,SAAS,qBAAqB,CAAC,QAAQ;IACnC,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,IAAI,QAAQ,CAAC,YAAY;QAAE,KAAK,IAAI,CAAC,CAAC;IACtC,IAAI,QAAQ,CAAC,WAAW;QAAE,KAAK,IAAI,CAAC,CAAC;IACrC,IAAI,QAAQ,CAAC,SAAS,IAAI,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC,MAAM,GAAG,CAAC;QAAE,KAAK,IAAI,CAAC,CAAC;IACjF,IAAI,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,QAAQ,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC;QAAE,KAAK,IAAI,CAAC,CAAC;IAC7E,OAAO,KAAK,CAAC;AACjB,CAAC;AAED;;;GAGG;AACH,SAAS,kBAAkB,CAAC,SAAS,EAAE,MAAM;IACzC,OAAO;QACH,SAAS,MAAM,CAAC,KAAK,EAAE;QACvB,UAAU,SAAS,CAAC,MAAM,EAAE;QAC5B,QAAQ,MAAM,CAAC,UAAU,EAAE;QAC3B,WAAW,MAAM,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE;QACtC,MAAM,CAAC,gBAAgB,CAAC,CAAC,CAAC,YAAY,MAAM,CAAC,gBAAgB,EAAE,CAAC,CAAC,CAAC,IAAI;QACtE,GAAG,MAAM,CAAC,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,UAAU,MAAM,CAAC,IAAI,EAAE,CAAC;QACxD,GAAG,CAAC,MAAM,CAAC,KAAK,IAAI,EAAE,CAAC;QACvB,GAAG,CAAC,SAAS,CAAC,QAAQ,IAAI,EAAE,CAAC;KAChC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;AACtB,CAAC"}