@pellux/goodvibes-sdk 0.25.21 → 0.26.0

package/dist/_internal/platform/knowledge/browser-history/readers.js

@@ -0,0 +1,374 @@
+import { readFile } from 'node:fs/promises';
+import { Database } from 'bun:sqlite';
+import bplistParserDefault from 'bplist-parser';
+import { copyLockedBrowserSqlite } from './locked-db.js';
+const bplistParser = bplistParserDefault;
+const CHROMIUM_EPOCH_OFFSET_MS = 11_644_473_600_000;
+const MAC_ABSOLUTE_TIME_OFFSET_S = 978_307_200;
+function chromiumMicrosToMs(micros) {
+    return Math.floor(micros / 1000) - CHROMIUM_EPOCH_OFFSET_MS;
+}
+function msToChromiumMicros(ms) {
+    return Math.floor((ms + CHROMIUM_EPOCH_OFFSET_MS) * 1000);
+}
+function geckoMicrosToMs(micros) {
+    return Math.floor(micros / 1000);
+}
+function macAbsoluteTimeToMs(seconds) {
+    return Math.floor((seconds + MAC_ABSOLUTE_TIME_OFFSET_S) * 1000);
+}
+function msToMacAbsoluteSeconds(ms) {
+    return Math.floor(ms / 1000) - MAC_ABSOLUTE_TIME_OFFSET_S;
+}
+function cleanOptionalText(value) {
+    if (typeof value !== 'string')
+        return undefined;
+    const trimmed = value.trim();
+    return trimmed ? trimmed : undefined;
+}
+function decodeChromiumTransition(raw) {
+    return {
+        0: 'LINK',
+        1: 'TYPED',
+        2: 'AUTO_BOOKMARK',
+        3: 'AUTO_SUBFRAME',
+        4: 'MANUAL_SUBFRAME',
+        5: 'GENERATED',
+        6: 'START_PAGE',
+        7: 'FORM_SUBMIT',
+        8: 'RELOAD',
+        9: 'KEYWORD',
+        10: 'KEYWORD_GENERATED',
+    }[raw & 0xff] ?? 'UNKNOWN';
+}
+function decodeGeckoVisitType(visitType) {
+    return {
+        1: 'LINK',
+        2: 'TYPED',
+        3: 'BOOKMARK',
+        4: 'EMBED',
+        5: 'PERMANENT_REDIRECT',
+        6: 'TEMPORARY_REDIRECT',
+        7: 'DOWNLOAD',
+        8: 'FRAMED_LINK',
+        9: 'RELOAD',
+    }[visitType] ?? 'UNKNOWN';
+}
+function normalizeLimit(limit, fallback = 1000) {
+    const candidate = typeof limit === 'number' && Number.isFinite(limit) ? limit : fallback;
+    return Math.max(1, Math.min(50_000, Math.floor(candidate)));
+}
+async function readChromiumHistory(profile, options) {
+    if (!profile.historyPath)
+        return [];
+    const limit = normalizeLimit(options.limit);
+    const since = typeof options.sinceMs === 'number' ? msToChromiumMicros(options.sinceMs) : 0;
+    const copy = await copyLockedBrowserSqlite(profile.historyPath);
+    try {
+        const db = new Database(copy.copiedDbPath, { readonly: true });
+        try {
+            const urlRows = db.query('SELECT id, url, title, visit_count FROM urls').all();
+            const urlMap = new Map(urlRows.map((row) => [row.id, row]));
+            const visitRows = db.query(`SELECT url, visit_time, transition
+         FROM visits
+         WHERE visit_time >= ?
+         ORDER BY visit_time DESC
+         LIMIT ?`).all(since, limit);
+            return visitRows.flatMap((visit) => {
+                const url = urlMap.get(visit.url);
+                if (!url?.url)
+                    return [];
+                return [{
+                        sourceKind: 'history',
+                        url: url.url,
+                        ...(cleanOptionalText(url.title) ? { title: cleanOptionalText(url.title) } : {}),
+                        browser: profile.browser,
+                        family: profile.family,
+                        profileName: profile.profileName,
+                        profilePath: profile.profilePath,
+                        visitedAtMs: chromiumMicrosToMs(visit.visit_time),
+                        visitCount: url.visit_count,
+                        transition: decodeChromiumTransition(visit.transition),
+                        rawId: url.id,
+                    }];
+            });
+        }
+        finally {
+            db.close();
+        }
+    }
+    finally {
+        await copy.cleanup();
+    }
+}
+function walkChromiumBookmarkNode(profile, node, folders, out, limit) {
+    if (out.length >= limit)
+        return;
+    if (node.type === 'url') {
+        const url = cleanOptionalText(node.url);
+        if (!url)
+            return;
+        const micros = Number(node.date_added ?? 0);
+        out.push({
+            sourceKind: 'bookmark',
+            url,
+            ...(cleanOptionalText(node.name) ? { title: cleanOptionalText(node.name) } : {}),
+            browser: profile.browser,
+            family: profile.family,
+            profileName: profile.profileName,
+            profilePath: profile.profilePath,
+            ...(folders.length ? { folderPath: `/${folders.join('/')}` } : {}),
+            ...(Number.isFinite(micros) && micros > 0 ? { addedAtMs: chromiumMicrosToMs(micros) } : {}),
+            ...(node.id ? { rawId: node.id } : {}),
+        });
+        return;
+    }
+    if (node.type === 'folder') {
+        const nextFolders = cleanOptionalText(node.name) ? [...folders, node.name.trim()] : [...folders];
+        for (const child of node.children ?? []) {
+            walkChromiumBookmarkNode(profile, child, nextFolders, out, limit);
+            if (out.length >= limit)
+                return;
+        }
+    }
+}
+async function readChromiumBookmarks(profile, options) {
+    if (!profile.bookmarksPath)
+        return [];
+    const limit = normalizeLimit(options.limit);
+    const parsed = JSON.parse(await readFile(profile.bookmarksPath, 'utf8'));
+    const out = [];
+    for (const key of ['bookmark_bar', 'other', 'synced']) {
+        const root = parsed.roots?.[key];
+        if (!root)
+            continue;
+        walkChromiumBookmarkNode(profile, root, [], out, limit);
+        if (out.length >= limit)
+            break;
+    }
+    return out;
+}
+async function readGeckoHistory(profile, options) {
+    if (!profile.historyPath)
+        return [];
+    const copy = await copyLockedBrowserSqlite(profile.historyPath);
+    try {
+        const db = new Database(copy.copiedDbPath, { readonly: true });
+        try {
+            const since = typeof options.sinceMs === 'number' ? Math.floor(options.sinceMs * 1000) : 0;
+            const rows = db.query(`SELECT p.id AS place_id, p.url, p.title, p.visit_count, h.visit_date, h.visit_type
+         FROM moz_historyvisits h
+         JOIN moz_places p ON p.id = h.place_id
+         WHERE p.hidden = 0 AND h.visit_date >= ?
+         ORDER BY h.visit_date DESC
+         LIMIT ?`).all(since, normalizeLimit(options.limit));
+            return rows.map((row) => ({
+                sourceKind: 'history',
+                url: row.url,
+                ...(cleanOptionalText(row.title) ? { title: cleanOptionalText(row.title) } : {}),
+                browser: profile.browser,
+                family: profile.family,
+                profileName: profile.profileName,
+                profilePath: profile.profilePath,
+                visitedAtMs: geckoMicrosToMs(row.visit_date),
+                visitCount: row.visit_count,
+                transition: decodeGeckoVisitType(row.visit_type),
+                rawId: row.place_id,
+            }));
+        }
+        finally {
+            db.close();
+        }
+    }
+    finally {
+        await copy.cleanup();
+    }
+}
+const GECKO_ROOT_LABELS = {
+    1: 'root',
+    2: 'Bookmarks Menu',
+    3: 'Bookmarks Toolbar',
+    4: 'Tags',
+    5: 'Other Bookmarks',
+    6: 'Mobile Bookmarks',
+};
+function geckoFolderPath(parent, rows) {
+    const segments = [];
+    const visited = new Set();
+    let current = parent;
+    while (current > 0 && !visited.has(current)) {
+        visited.add(current);
+        const row = rows.get(current);
+        if (!row)
+            break;
+        const root = GECKO_ROOT_LABELS[current];
+        if (root) {
+            segments.unshift(root);
+            break;
+        }
+        if (row.title)
+            segments.unshift(row.title);
+        current = row.parent;
+    }
+    return segments.length ? `/${segments.join('/')}` : undefined;
+}
+async function readGeckoBookmarks(profile, options) {
+    const path = profile.bookmarksPath ?? profile.historyPath;
+    if (!path)
+        return [];
+    const copy = await copyLockedBrowserSqlite(path);
+    try {
+        const db = new Database(copy.copiedDbPath, { readonly: true });
+        try {
+            const rows = db.query('SELECT id, type, parent, title, fk, dateAdded FROM moz_bookmarks ORDER BY id').all();
+            const places = db.query('SELECT id, url, title FROM moz_places').all();
+            const rowById = new Map(rows.map((row) => [row.id, row]));
+            const placeById = new Map(places.map((place) => [place.id, place]));
+            const out = [];
+            for (const row of rows) {
+                if (out.length >= normalizeLimit(options.limit))
+                    break;
+                if (row.type !== 1 || row.fk === null)
+                    continue;
+                const place = placeById.get(row.fk);
+                if (!place?.url)
+                    continue;
+                const addedAtMs = Math.max(0, Math.floor(row.dateAdded / 1000));
+                out.push({
+                    sourceKind: 'bookmark',
+                    url: place.url,
+                    ...(cleanOptionalText(row.title ?? place.title) ? { title: cleanOptionalText(row.title ?? place.title) } : {}),
+                    browser: profile.browser,
+                    family: profile.family,
+                    profileName: profile.profileName,
+                    profilePath: profile.profilePath,
+                    ...(geckoFolderPath(row.parent, rowById) ? { folderPath: geckoFolderPath(row.parent, rowById) } : {}),
+                    ...(addedAtMs > 0 ? { addedAtMs } : {}),
+                    rawId: row.id,
+                });
+            }
+            return out;
+        }
+        finally {
+            db.close();
+        }
+    }
+    finally {
+        await copy.cleanup();
+    }
+}
+async function readWebKitHistory(profile, options) {
+    if (!profile.historyPath)
+        return [];
+    const copy = await copyLockedBrowserSqlite(profile.historyPath);
+    try {
+        const db = new Database(copy.copiedDbPath, { readonly: true });
+        try {
+            if (profile.browser === 'epiphany') {
+                const since = typeof options.sinceMs === 'number' ? Math.floor(options.sinceMs / 1000) : 0;
+                const rows = db.query('SELECT id, url, title, visit_count, last_visit_time FROM urls WHERE last_visit_time >= ? ORDER BY last_visit_time DESC LIMIT ?').all(since, normalizeLimit(options.limit));
+                return rows.map((row) => ({
+                    sourceKind: 'history',
+                    url: row.url,
+                    ...(cleanOptionalText(row.title) ? { title: cleanOptionalText(row.title) } : {}),
+                    browser: profile.browser,
+                    family: profile.family,
+                    profileName: profile.profileName,
+                    profilePath: profile.profilePath,
+                    visitedAtMs: row.last_visit_time * 1000,
+                    visitCount: row.visit_count,
+                    rawId: row.id,
+                }));
+            }
+            const items = db.query('SELECT id, url, visit_count, title FROM history_items').all();
+            const itemById = new Map(items.map((item) => [item.id, item]));
+            const since = typeof options.sinceMs === 'number' ? msToMacAbsoluteSeconds(options.sinceMs) : 0;
+            const rows = db.query('SELECT history_item, visit_time, title FROM history_visits WHERE visit_time >= ? ORDER BY visit_time DESC LIMIT ?').all(since, normalizeLimit(options.limit));
+            return rows.flatMap((visit) => {
+                const item = itemById.get(visit.history_item);
+                if (!item?.url)
+                    return [];
+                return [{
+                        sourceKind: 'history',
+                        url: item.url,
+                        ...(cleanOptionalText(visit.title ?? item.title) ? { title: cleanOptionalText(visit.title ?? item.title) } : {}),
+                        browser: profile.browser,
+                        family: profile.family,
+                        profileName: profile.profileName,
+                        profilePath: profile.profilePath,
+                        visitedAtMs: macAbsoluteTimeToMs(visit.visit_time),
+                        visitCount: item.visit_count,
+                        rawId: item.id,
+                    }];
+            });
+        }
+        finally {
+            db.close();
+        }
+    }
+    finally {
+        await copy.cleanup();
+    }
+}
+function walkWebKitBookmarkNode(profile, node, folderPath, out, limit) {
+    if (out.length >= limit)
+        return;
+    if (node.WebBookmarkType === 'WebBookmarkTypeLeaf') {
+        const url = cleanOptionalText(node.URLString);
+        if (!url)
+            return;
+        out.push({
+            sourceKind: 'bookmark',
+            url,
+            ...(cleanOptionalText(node.URIDictionary?.title) ? { title: cleanOptionalText(node.URIDictionary?.title) } : {}),
+            browser: profile.browser,
+            family: profile.family,
+            profileName: profile.profileName,
+            profilePath: profile.profilePath,
+            ...(folderPath ? { folderPath } : {}),
+            ...(typeof node.WebBookmarkDateAdded === 'number' ? { addedAtMs: macAbsoluteTimeToMs(node.WebBookmarkDateAdded) } : {}),
+        });
+        return;
+    }
+    if (node.WebBookmarkType === 'WebBookmarkTypeList') {
+        const next = cleanOptionalText(node.Title) ? `${folderPath}/${node.Title.trim()}` : folderPath;
+        for (const child of node.Children ?? []) {
+            walkWebKitBookmarkNode(profile, child, next, out, limit);
+            if (out.length >= limit)
+                return;
+        }
+    }
+}
+async function readWebKitBookmarks(profile, options) {
+    if (!profile.bookmarksPath || profile.browser === 'epiphany')
+        return [];
+    const [root] = bplistParser.parseFileSync(profile.bookmarksPath);
+    const out = [];
+    for (const child of root.Children ?? []) {
+        walkWebKitBookmarkNode(profile, child, '', out, normalizeLimit(options.limit));
+        if (out.length >= normalizeLimit(options.limit))
+            break;
+    }
+    return out;
+}
+export async function readBrowserKnowledgeProfile(profile, options = {}) {
+    const kinds = new Set(options.sourceKinds ?? ['history', 'bookmark']);
+    const entries = [];
+    if (kinds.has('history')) {
+        if (profile.family === 'chromium')
+            entries.push(...await readChromiumHistory(profile, options));
+        if (profile.family === 'gecko')
+            entries.push(...await readGeckoHistory(profile, options));
+        if (profile.family === 'webkit')
+            entries.push(...await readWebKitHistory(profile, options));
+    }
+    if (kinds.has('bookmark')) {
+        if (profile.family === 'chromium')
+            entries.push(...await readChromiumBookmarks(profile, options));
+        if (profile.family === 'gecko')
+            entries.push(...await readGeckoBookmarks(profile, options));
+        if (profile.family === 'webkit')
+            entries.push(...await readWebKitBookmarks(profile, options));
+    }
+    return entries.slice(0, normalizeLimit(options.limit, 2000));
+}

package/dist/_internal/platform/knowledge/browser-history/types.d.ts

@@ -0,0 +1,50 @@
+export type BrowserKnowledgeFamily = 'chromium' | 'gecko' | 'webkit';
+export type BrowserKnowledgeKind = 'chrome' | 'chromium' | 'brave' | 'edge' | 'vivaldi' | 'arc' | 'opera' | 'firefox' | 'zen' | 'librewolf' | 'waterfox' | 'floorp' | 'safari' | 'orion' | 'epiphany';
+export type BrowserKnowledgeSourceKind = 'history' | 'bookmark';
+export interface BrowserKnowledgeProfile {
+    readonly family: BrowserKnowledgeFamily;
+    readonly browser: BrowserKnowledgeKind;
+    readonly profileName: string;
+    readonly profilePath: string;
+    readonly historyPath?: string;
+    readonly bookmarksPath?: string;
+}
+export interface BrowserHistoryEntry {
+    readonly sourceKind: 'history';
+    readonly url: string;
+    readonly title?: string;
+    readonly browser: BrowserKnowledgeKind;
+    readonly family: BrowserKnowledgeFamily;
+    readonly profileName: string;
+    readonly profilePath: string;
+    readonly visitedAtMs?: number;
+    readonly visitCount?: number;
+    readonly transition?: string;
+    readonly rawId?: string | number;
+}
+export interface BrowserBookmarkEntry {
+    readonly sourceKind: 'bookmark';
+    readonly url: string;
+    readonly title?: string;
+    readonly browser: BrowserKnowledgeKind;
+    readonly family: BrowserKnowledgeFamily;
+    readonly profileName: string;
+    readonly profilePath: string;
+    readonly folderPath?: string;
+    readonly addedAtMs?: number;
+    readonly rawId?: string | number;
+}
+export type BrowserKnowledgeEntry = BrowserHistoryEntry | BrowserBookmarkEntry;
+export interface BrowserKnowledgeFilter {
+    readonly browsers?: readonly BrowserKnowledgeKind[];
+    readonly sourceKinds?: readonly BrowserKnowledgeSourceKind[];
+    readonly homeOverride?: string;
+    readonly limit?: number;
+    readonly sinceMs?: number;
+}
+export interface BrowserKnowledgeCollectResult {
+    readonly profiles: readonly BrowserKnowledgeProfile[];
+    readonly entries: readonly BrowserKnowledgeEntry[];
+    readonly errors: readonly string[];
+}
+//# sourceMappingURL=types.d.ts.map

package/dist/_internal/platform/knowledge/browser-history/types.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../../../../../src/_internal/platform/knowledge/browser-history/types.ts"],"names":[],"mappings":"AAAA,MAAM,MAAM,sBAAsB,GAAG,UAAU,GAAG,OAAO,GAAG,QAAQ,CAAC;AAErE,MAAM,MAAM,oBAAoB,GAC5B,QAAQ,GACR,UAAU,GACV,OAAO,GACP,MAAM,GACN,SAAS,GACT,KAAK,GACL,OAAO,GACP,SAAS,GACT,KAAK,GACL,WAAW,GACX,UAAU,GACV,QAAQ,GACR,QAAQ,GACR,OAAO,GACP,UAAU,CAAC;AAEf,MAAM,MAAM,0BAA0B,GAAG,SAAS,GAAG,UAAU,CAAC;AAEhE,MAAM,WAAW,uBAAuB;IACtC,QAAQ,CAAC,MAAM,EAAE,sBAAsB,CAAC;IACxC,QAAQ,CAAC,OAAO,EAAE,oBAAoB,CAAC;IACvC,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,WAAW,CAAC,EAAE,MAAM,CAAC;IAC9B,QAAQ,CAAC,aAAa,CAAC,EAAE,MAAM,CAAC;CACjC;AAED,MAAM,WAAW,mBAAmB;IAClC,QAAQ,CAAC,UAAU,EAAE,SAAS,CAAC;IAC/B,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAC;IACrB,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,CAAC;IACxB,QAAQ,CAAC,OAAO,EAAE,oBAAoB,CAAC;IACvC,QAAQ,CAAC,MAAM,EAAE,sBAAsB,CAAC;IACxC,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,WAAW,CAAC,EAAE,MAAM,CAAC;IAC9B,QAAQ,CAAC,UAAU,CAAC,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,UAAU,CAAC,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;CAClC;AAED,MAAM,WAAW,oBAAoB;IACnC,QAAQ,CAAC,UAAU,EAAE,UAAU,CAAC;IAChC,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAC;IACrB,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,CAAC;IACxB,QAAQ,CAAC,OAAO,EAAE,oBAAoB,CAAC;IACvC,QAAQ,CAAC,MAAM,EAAE,sBAAsB,CAAC;IACxC,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,UAAU,CAAC,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,SAAS,CAAC,EAAE,MAAM,CAAC;IAC5B,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;CAClC;AAED,MAAM,MAAM,qBAAqB,GAAG,mBAAmB,GAAG,oBAAoB,CAAC;AAE/E,MAAM,WAAW,sBAAsB;IACrC,QAAQ,CAAC,QAAQ,CAAC,EAAE,SAAS,oBAAoB,EAAE,CAAC;IACpD,QAAQ,CAAC,WAAW,CAAC,EAAE,SAAS,0BAA0B,EAAE,CAAC;IAC7D,QAAQ,CAAC,YAAY,CAAC,EAAE,MAAM,CAAC;IAC/B,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,CAAC;IACxB,QAAQ,CAAC,OAAO,CAAC,EAAE,MAAM,CAAC;CAC3B;AAED,MAAM,WAAW,6BAA6B;IAC5C,QAAQ,CAAC,QAAQ,EAAE,SAAS,uBAAuB,EAAE,CAAC;IACtD,QAAQ,CAAC,OAAO,EAAE,SAAS,qBAAqB,EAAE,CAAC;IACnD,QAAQ,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC;CACpC"}

package/dist/_internal/platform/knowledge/browser-history/types.js

@@ -0,0 +1 @@
+export {};

package/dist/_internal/platform/knowledge/consolidation.js

@@ -245,6 +245,7 @@ function inferMemoryClassForCandidate(context, subjectKind, subjectId) {
         case 'bookmark':
         case 'url':
         case 'bookmark-list':
+        case 'history':
             return 'fact';
         case 'document':
             return 'runbook';
@@ -262,6 +263,7 @@ function getSourceRefreshWindowMs(source) {
     return {
         bookmark: 7 * 24 * 60 * 60 * 1000,
         'bookmark-list': 7 * 24 * 60 * 60 * 1000,
+        history: 14 * 24 * 60 * 60 * 1000,
         'url-list': 7 * 24 * 60 * 60 * 1000,
         url: 14 * 24 * 60 * 60 * 1000,
         repo: 14 * 24 * 60 * 60 * 1000,

package/dist/_internal/platform/knowledge/extractors.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"extractors.d.ts","sourceRoot":"","sources":["../../../../src/_internal/platform/knowledge/extractors.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAsB,cAAc,EAAE,MAAM,uBAAuB,CAAC;AAEhF,OAAO,KAAK,EAAE,yBAAyB,EAAE,MAAM,YAAY,CAAC;AAE5D,MAAM,WAAW,yBAAyB;IACxC,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,MAAM,EAAE,yBAAyB,CAAC;IAC3C,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,CAAC;IACxB,QAAQ,CAAC,OAAO,CAAC,EAAE,MAAM,CAAC;IAC1B,QAAQ,CAAC,OAAO,CAAC,EAAE,MAAM,CAAC;IAC1B,QAAQ,CAAC,QAAQ,EAAE,SAAS,MAAM,EAAE,CAAC;IACrC,QAAQ,CAAC,KAAK,EAAE,SAAS,MAAM,EAAE,CAAC;IAClC,QAAQ,CAAC,eAAe,EAAE,MAAM,CAAC;IACjC,QAAQ,CAAC,SAAS,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IAC5C,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;CAC5C;AAweD,wBAAsB,wBAAwB,CAC5C,QAAQ,EAAE,IAAI,CAAC,cAAc,EAAE,IAAI,GAAG,UAAU,GAAG,UAAU,CAAC,EAC9D,MAAM,EAAE,MAAM,GACb,OAAO,CAAC,yBAAyB,CAAC,CA4CpC"}
+{"version":3,"file":"extractors.d.ts","sourceRoot":"","sources":["../../../../src/_internal/platform/knowledge/extractors.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAsB,cAAc,EAAE,MAAM,uBAAuB,CAAC;AAGhF,OAAO,KAAK,EAAE,yBAAyB,EAAE,MAAM,YAAY,CAAC;AAE5D,MAAM,WAAW,yBAAyB;IACxC,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,MAAM,EAAE,yBAAyB,CAAC;IAC3C,QAAQ,CAAC,KAAK,CAAC,EAAE,MAAM,CAAC;IACxB,QAAQ,CAAC,OAAO,CAAC,EAAE,MAAM,CAAC;IAC1B,QAAQ,CAAC,OAAO,CAAC,EAAE,MAAM,CAAC;IAC1B,QAAQ,CAAC,QAAQ,EAAE,SAAS,MAAM,EAAE,CAAC;IACrC,QAAQ,CAAC,KAAK,EAAE,SAAS,MAAM,EAAE,CAAC;IAClC,QAAQ,CAAC,eAAe,EAAE,MAAM,CAAC;IACjC,QAAQ,CAAC,SAAS,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IAC5C,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;CAC5C;AAqgBD,wBAAsB,wBAAwB,CAC5C,QAAQ,EAAE,IAAI,CAAC,cAAc,EAAE,IAAI,GAAG,UAAU,GAAG,UAAU,CAAC,EAC9D,MAAM,EAAE,MAAM,GACb,OAAO,CAAC,yBAAyB,CAAC,CA4CpC"}

package/dist/_internal/platform/knowledge/extractors.js

@@ -1,6 +1,7 @@
 import JSZip from 'jszip';
 import { extname } from 'node:path';
 import { guessMimeType } from '../artifacts/types.js';
+import { extractReadableHtml } from './html-readability.js';
 function decodeHtmlEntities(value) {
     return value
         .replace(/ /g, ' ')
@@ -85,6 +86,36 @@ function extractLinksFromHtml(html) {
 }
 function extractHtml(buffer) {
     const html = buffer.toString('utf-8');
+    try {
+        const readable = extractReadableHtml(html);
+        if (readable) {
+            const summary = summarizeText([readable.excerpt, readable.paragraphSamples[0], readable.textContent].filter(Boolean).join(' '));
+            return {
+                extractorId: 'html-readability',
+                format: 'html',
+                ...(readable.title ? { title: readable.title } : {}),
+                ...(summary ? { summary } : {}),
+                ...(excerptText(readable.textContent) ? { excerpt: excerptText(readable.textContent) } : {}),
+                sections: readable.headings.length > 0 ? readable.headings.slice(0, 16) : uniqueStrings(readable.textContent.split(/\n+/), 8),
+                links: readable.links,
+                estimatedTokens: estimateTokens(readable.title, summary, readable.textContent),
+                structure: {
+                    headings: readable.headings,
+                    readableLength: readable.length,
+                    paragraphSampleCount: readable.paragraphSamples.length,
+                },
+                metadata: {
+                    paragraphSamples: readable.paragraphSamples.slice(0, 4),
+                    ...(readable.byline ? { byline: readable.byline } : {}),
+                    ...(readable.siteName ? { siteName: readable.siteName } : {}),
+                    extractionPath: 'readability',
+                },
+            };
+        }
+    }
+    catch {
+        // Fall back to the lightweight extractor below for malformed or hostile HTML.
+    }
     const title = cleanText(html.match(/<title[^>]*>([\s\S]*?)<\/title>/i)?.[1] ?? '')
         || cleanText(html.match(/<h1[^>]*>([\s\S]*?)<\/h1>/i)?.[1] ?? '');
     const headings = uniqueStrings(Array.from(html.matchAll(/<h([1-3])[^>]*>([\s\S]*?)<\/h\1>/gi), (match) => stripHtml(match[2] ?? '')), 16);

package/dist/_internal/platform/knowledge/graphql-schema.d.ts

@@ -1,6 +1,6 @@
 import { buildSchema, type DocumentNode, type ValueNode } from 'graphql';
 import type { KnowledgePacketDetail, KnowledgeProjectionTargetKind } from './types.js';
-export declare const KNOWLEDGE_GRAPHQL_SDL = "\n  scalar JSON\n\n  enum KnowledgeProjectionKind {\n    OVERVIEW\n    BUNDLE\n    SOURCE\n    NODE\n    ISSUE\n    DASHBOARD\n    ROLLUP\n  }\n\n  enum KnowledgePacketDetail {\n    COMPACT\n    STANDARD\n    DETAILED\n  }\n\n  enum KnowledgeJobMode {\n    INLINE\n    BACKGROUND\n  }\n\n  type KnowledgeStatus {\n    ready: Boolean!\n    storagePath: String!\n    sourceCount: Int!\n    nodeCount: Int!\n    edgeCount: Int!\n    issueCount: Int!\n    extractionCount: Int!\n    jobRunCount: Int!\n    usageCount: Int!\n    candidateCount: Int!\n    reportCount: Int!\n    scheduleCount: Int!\n    note: String!\n  }\n\n  type KnowledgeConnectorSetupField {\n    key: String!\n    label: String!\n    kind: String!\n    optional: Boolean\n    source: String\n    description: String\n  }\n\n  type KnowledgeConnectorSetup {\n    version: String!\n    summary: String!\n    transportHints: [String!]!\n    steps: [String!]!\n    fields: [KnowledgeConnectorSetupField!]!\n    metadata: JSON\n  }\n\n  type KnowledgeConnectorDoctorCheck {\n    id: String!\n    label: String!\n    status: String!\n    detail: String!\n    metadata: JSON\n  }\n\n  type KnowledgeConnectorDoctorReport {\n    connectorId: String!\n    ready: Boolean!\n    summary: String!\n    checks: [KnowledgeConnectorDoctorCheck!]!\n    hints: [String!]!\n    metadata: JSON\n  }\n\n  type KnowledgeConnector {\n    id: String!\n    displayName: String\n    version: String\n    description: String!\n    sourceType: String!\n    inputSchema: JSON\n    examples: [JSON!]!\n    capabilities: [String!]!\n    setup: KnowledgeConnectorSetup\n    metadata: JSON\n  }\n\n  type KnowledgeSource {\n    id: String!\n    connectorId: String!\n    sourceType: String!\n    title: String\n    sourceUri: String\n    canonicalUri: String\n    summary: String\n    description: String\n    tags: [String!]!\n    folderPath: String\n    status: String!\n    artifactId: String\n    contentHash: String\n    lastCrawledAt: Float\n    crawlError: String\n    sessionId: String\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeNode {\n    id: String!\n    kind: String!\n    slug: String!\n    title: String!\n    summary: String\n    aliases: [String!]!\n    status: String!\n    confidence: Int!\n    sourceId: String\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeEdge {\n    id: String!\n    fromKind: String!\n    fromId: String!\n    toKind: String!\n    toId: String!\n    relation: String!\n    weight: Float!\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeIssue {\n    id: String!\n    severity: String!\n    code: String!\n    message: String!\n    status: String!\n    sourceId: String\n    nodeId: String\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeExtraction {\n    id: String!\n    sourceId: String!\n    artifactId: String\n    extractorId: String!\n    format: String!\n    title: String\n    summary: String\n    excerpt: String\n    sections: [String!]!\n    links: [String!]!\n    estimatedTokens: Int!\n    structure: JSON\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeItemView {\n    source: KnowledgeSource\n    node: KnowledgeNode\n    issue: KnowledgeIssue\n    relatedEdges: [KnowledgeEdge!]!\n    linkedSources: [KnowledgeSource!]!\n    linkedNodes: [KnowledgeNode!]!\n  }\n\n  type KnowledgeSearchResult {\n    kind: String!\n    id: String!\n    score: Float!\n    reason: String!\n    source: KnowledgeSource\n    node: KnowledgeNode\n  }\n\n  type KnowledgePacketItem {\n    kind: String!\n    id: String!\n    title: String!\n    summary: String\n    uri: String\n    reason: String!\n    score: Float!\n    estimatedTokens: Int!\n    related: [String!]!\n    evidence: [String!]!\n    metadata: JSON\n  }\n\n  type KnowledgePacket {\n    task: String!\n    writeScope: [String!]!\n    generatedAt: Float!\n    detail: KnowledgePacketDetail!\n    strategy: String!\n    budgetLimit: Int!\n    estimatedTokens: Int!\n    items: [KnowledgePacketItem!]!\n  }\n\n  type KnowledgeProjectionTarget {\n    targetId: String!\n    kind: KnowledgeProjectionKind!\n    title: String!\n    description: String!\n    itemId: String\n    defaultPath: String!\n    defaultFilename: String!\n    metadata: JSON\n  }\n\n  type KnowledgeProjectionPage {\n    path: String!\n    title: String!\n    format: String!\n    content: String!\n    itemIds: [String!]!\n    metadata: JSON\n  }\n\n  type KnowledgeProjectionBundle {\n    id: String!\n    target: KnowledgeProjectionTarget!\n    generatedAt: Float!\n    pageCount: Int!\n    pages: [KnowledgeProjectionPage!]!\n    metadata: JSON\n  }\n\n  type ArtifactDescriptor {\n    id: String!\n    kind: String!\n    mimeType: String!\n    filename: String\n    sizeBytes: Int!\n    sourceUri: String\n    createdAt: Float!\n    metadata: JSON\n  }\n\n  type KnowledgeJob {\n    id: String!\n    label: String!\n    description: String!\n    defaultMode: KnowledgeJobMode!\n    triggerKinds: [String!]!\n    lastRunAt: Float\n    metadata: JSON\n  }\n\n  type KnowledgeJobRun {\n    id: String!\n    jobId: String!\n    status: String!\n    mode: KnowledgeJobMode!\n    requestedAt: Float!\n    startedAt: Float\n    completedAt: Float\n    error: String\n    result: JSON\n    metadata: JSON\n  }\n\n  type KnowledgeBatchIngestResult {\n    imported: Int!\n    failed: Int!\n    sources: [KnowledgeSource!]!\n    errors: [String!]!\n  }\n\n  type KnowledgeUsageRecord {\n    id: String!\n    targetKind: String!\n    targetId: String!\n    usageKind: String!\n    task: String\n    sessionId: String\n    score: Float\n    metadata: JSON\n    createdAt: Float!\n  }\n\n  type KnowledgeConsolidationCandidate {\n    id: String!\n    candidateType: String!\n    status: String!\n    subjectKind: String!\n    subjectId: String!\n    title: String!\n    summary: String\n    score: Float!\n    evidence: [String!]!\n    suggestedMemoryClass: String\n    suggestedScope: String\n    decidedAt: Float\n    decidedBy: String\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeConsolidationReport {\n    id: String!\n    kind: String!\n    title: String!\n    summary: String!\n    highlights: [String!]!\n    metrics: JSON\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeSchedule {\n    id: String!\n    jobId: String!\n    label: String!\n    enabled: Boolean!\n    schedule: JSON!\n    lastRunAt: Float\n    nextRunAt: Float\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  input KnowledgeJobRunInput {\n    mode: KnowledgeJobMode\n  }\n\n  input KnowledgePacketInput {\n    task: String!\n    writeScope: [String!]\n    limit: Int\n    detail: KnowledgePacketDetail\n    budgetLimit: Int\n  }\n\n  type Query {\n    status: KnowledgeStatus!\n    connectors: [KnowledgeConnector!]!\n    connectorDoctor(id: String!): KnowledgeConnectorDoctorReport\n    sources(limit: Int, offset: Int): [KnowledgeSource!]!\n    nodes(limit: Int, offset: Int): [KnowledgeNode!]!\n    issues(limit: Int, offset: Int): [KnowledgeIssue!]!\n    itemView(kind: String!, id: String!): KnowledgeItemView\n    extractions(limit: Int, sourceId: String): [KnowledgeExtraction!]!\n    neighbors(kind: String!, id: String!, relation: String, limit: Int): [KnowledgeEdge!]!\n    search(query: String!, limit: Int): [KnowledgeSearchResult!]!\n    packet(task: String!, writeScope: [String!], limit: Int, detail: KnowledgePacketDetail, budgetLimit: Int): KnowledgePacket!\n    projectionTargets(limit: Int): [KnowledgeProjectionTarget!]!\n    projection(kind: KnowledgeProjectionKind!, id: String, limit: Int): KnowledgeProjectionBundle!\n    jobs: [KnowledgeJob!]!\n    job(id: String!): KnowledgeJob\n    jobRuns(limit: Int, jobId: String): [KnowledgeJobRun!]!\n    usage(limit: Int, targetKind: String, targetId: String, usageKind: String): [KnowledgeUsageRecord!]!\n    usageRecords(limit: Int, targetKind: String, targetId: String, usageKind: String): [KnowledgeUsageRecord!]!\n    consolidationCandidates(limit: Int, status: String): [KnowledgeConsolidationCandidate!]!\n    consolidationReports(limit: Int): [KnowledgeConsolidationReport!]!\n    schedules(limit: Int): [KnowledgeSchedule!]!\n    sourceExtraction(sourceId: String!): KnowledgeExtraction\n  }\n\n  type Mutation {\n    ingestUrl(url: String!, sourceType: String, connectorId: String, folderPath: String, tags: [String!], sessionId: String, allowPrivateHosts: Boolean): KnowledgeSource!\n    ingestArtifact(path: String!, connectorId: String, tags: [String!], sessionId: String, allowPrivateHosts: Boolean): KnowledgeSource!\n    importBookmarksFromFile(path: String!, sessionId: String, allowPrivateHosts: Boolean): KnowledgeBatchIngestResult!\n    importUrlsFromFile(path: String!, sessionId: String, allowPrivateHosts: Boolean): KnowledgeBatchIngestResult!\n    ingestBookmarkSeeds(seeds: [String!]!, connectorId: String, sessionId: String): KnowledgeBatchIngestResult!\n    ingestWithConnector(connectorId: String!, input: String!, sessionId: String, allowPrivateHosts: Boolean): KnowledgeBatchIngestResult!\n    ingestConnectorInput(input: String!, connectorId: String, sessionId: String, allowPrivateHosts: Boolean): KnowledgeBatchIngestResult!\n    renderProjection(kind: KnowledgeProjectionKind!, id: String, limit: Int): KnowledgeProjectionBundle!\n    materializeProjection(kind: KnowledgeProjectionKind!, id: String, limit: Int): KnowledgeMaterializedProjection!\n    reindex: KnowledgeStatus!\n    runJob(id: String!, mode: KnowledgeJobMode): KnowledgeJobRun!\n    saveSchedule(id: String, jobId: String!, label: String!, enabled: Boolean!, schedule: JSON!, metadata: JSON): KnowledgeSchedule!\n    deleteSchedule(id: String!): Boolean!\n    setScheduleEnabled(id: String!, enabled: Boolean!): KnowledgeSchedule\n    decideConsolidationCandidate(id: String!, status: String!, decidedBy: String): KnowledgeConsolidationCandidate!\n    lint: [KnowledgeIssue!]!\n  }\n\n  type KnowledgeMaterializedProjection {\n    artifact: ArtifactDescriptor!\n    bundle: KnowledgeProjectionBundle!\n  }\n";
+export declare const KNOWLEDGE_GRAPHQL_SDL = "\n  scalar JSON\n\n  enum KnowledgeProjectionKind {\n    OVERVIEW\n    BUNDLE\n    SOURCE\n    NODE\n    ISSUE\n    DASHBOARD\n    ROLLUP\n  }\n\n  enum KnowledgePacketDetail {\n    COMPACT\n    STANDARD\n    DETAILED\n  }\n\n  enum KnowledgeJobMode {\n    INLINE\n    BACKGROUND\n  }\n\n  type KnowledgeStatus {\n    ready: Boolean!\n    storagePath: String!\n    sourceCount: Int!\n    nodeCount: Int!\n    edgeCount: Int!\n    issueCount: Int!\n    extractionCount: Int!\n    jobRunCount: Int!\n    usageCount: Int!\n    candidateCount: Int!\n    reportCount: Int!\n    scheduleCount: Int!\n    note: String!\n  }\n\n  type KnowledgeConnectorSetupField {\n    key: String!\n    label: String!\n    kind: String!\n    optional: Boolean\n    source: String\n    description: String\n  }\n\n  type KnowledgeConnectorSetup {\n    version: String!\n    summary: String!\n    transportHints: [String!]!\n    steps: [String!]!\n    fields: [KnowledgeConnectorSetupField!]!\n    metadata: JSON\n  }\n\n  type KnowledgeConnectorDoctorCheck {\n    id: String!\n    label: String!\n    status: String!\n    detail: String!\n    metadata: JSON\n  }\n\n  type KnowledgeConnectorDoctorReport {\n    connectorId: String!\n    ready: Boolean!\n    summary: String!\n    checks: [KnowledgeConnectorDoctorCheck!]!\n    hints: [String!]!\n    metadata: JSON\n  }\n\n  type KnowledgeConnector {\n    id: String!\n    displayName: String\n    version: String\n    description: String!\n    sourceType: String!\n    inputSchema: JSON\n    examples: [JSON!]!\n    capabilities: [String!]!\n    setup: KnowledgeConnectorSetup\n    metadata: JSON\n  }\n\n  type KnowledgeSource {\n    id: String!\n    connectorId: String!\n    sourceType: String!\n    title: String\n    sourceUri: String\n    canonicalUri: String\n    summary: String\n    description: String\n    tags: [String!]!\n    folderPath: String\n    status: String!\n    artifactId: String\n    contentHash: String\n    lastCrawledAt: Float\n    crawlError: String\n    sessionId: String\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeNode {\n    id: String!\n    kind: String!\n    slug: String!\n    title: String!\n    summary: String\n    aliases: [String!]!\n    status: String!\n    confidence: Int!\n    sourceId: String\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeEdge {\n    id: String!\n    fromKind: String!\n    fromId: String!\n    toKind: String!\n    toId: String!\n    relation: String!\n    weight: Float!\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeIssue {\n    id: String!\n    severity: String!\n    code: String!\n    message: String!\n    status: String!\n    sourceId: String\n    nodeId: String\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeExtraction {\n    id: String!\n    sourceId: String!\n    artifactId: String\n    extractorId: String!\n    format: String!\n    title: String\n    summary: String\n    excerpt: String\n    sections: [String!]!\n    links: [String!]!\n    estimatedTokens: Int!\n    structure: JSON\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeItemView {\n    source: KnowledgeSource\n    node: KnowledgeNode\n    issue: KnowledgeIssue\n    relatedEdges: [KnowledgeEdge!]!\n    linkedSources: [KnowledgeSource!]!\n    linkedNodes: [KnowledgeNode!]!\n  }\n\n  type KnowledgeSearchResult {\n    kind: String!\n    id: String!\n    score: Float!\n    reason: String!\n    source: KnowledgeSource\n    node: KnowledgeNode\n  }\n\n  type KnowledgePacketItem {\n    kind: String!\n    id: String!\n    title: String!\n    summary: String\n    uri: String\n    reason: String!\n    score: Float!\n    estimatedTokens: Int!\n    related: [String!]!\n    evidence: [String!]!\n    metadata: JSON\n  }\n\n  type KnowledgePacket {\n    task: String!\n    writeScope: [String!]!\n    generatedAt: Float!\n    detail: KnowledgePacketDetail!\n    strategy: String!\n    budgetLimit: Int!\n    estimatedTokens: Int!\n    items: [KnowledgePacketItem!]!\n  }\n\n  type KnowledgeProjectionTarget {\n    targetId: String!\n    kind: KnowledgeProjectionKind!\n    title: String!\n    description: String!\n    itemId: String\n    defaultPath: String!\n    defaultFilename: String!\n    metadata: JSON\n  }\n\n  type KnowledgeProjectionPage {\n    path: String!\n    title: String!\n    format: String!\n    content: String!\n    itemIds: [String!]!\n    metadata: JSON\n  }\n\n  type KnowledgeProjectionBundle {\n    id: String!\n    target: KnowledgeProjectionTarget!\n    generatedAt: Float!\n    pageCount: Int!\n    pages: [KnowledgeProjectionPage!]!\n    metadata: JSON\n  }\n\n  type ArtifactDescriptor {\n    id: String!\n    kind: String!\n    mimeType: String!\n    filename: String\n    sizeBytes: Int!\n    sourceUri: String\n    createdAt: Float!\n    metadata: JSON\n  }\n\n  type KnowledgeJob {\n    id: String!\n    label: String!\n    description: String!\n    defaultMode: KnowledgeJobMode!\n    triggerKinds: [String!]!\n    lastRunAt: Float\n    metadata: JSON\n  }\n\n  type KnowledgeJobRun {\n    id: String!\n    jobId: String!\n    status: String!\n    mode: KnowledgeJobMode!\n    requestedAt: Float!\n    startedAt: Float\n    completedAt: Float\n    error: String\n    result: JSON\n    metadata: JSON\n  }\n\n  type KnowledgeBatchIngestResult {\n    imported: Int!\n    failed: Int!\n    sources: [KnowledgeSource!]!\n    errors: [String!]!\n  }\n\n  type KnowledgeUsageRecord {\n    id: String!\n    targetKind: String!\n    targetId: String!\n    usageKind: String!\n    task: String\n    sessionId: String\n    score: Float\n    metadata: JSON\n    createdAt: Float!\n  }\n\n  type KnowledgeConsolidationCandidate {\n    id: String!\n    candidateType: String!\n    status: String!\n    subjectKind: String!\n    subjectId: String!\n    title: String!\n    summary: String\n    score: Float!\n    evidence: [String!]!\n    suggestedMemoryClass: String\n    suggestedScope: String\n    decidedAt: Float\n    decidedBy: String\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeConsolidationReport {\n    id: String!\n    kind: String!\n    title: String!\n    summary: String!\n    highlights: [String!]!\n    metrics: JSON\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  type KnowledgeSchedule {\n    id: String!\n    jobId: String!\n    label: String!\n    enabled: Boolean!\n    schedule: JSON!\n    lastRunAt: Float\n    nextRunAt: Float\n    metadata: JSON\n    createdAt: Float!\n    updatedAt: Float!\n  }\n\n  input KnowledgeJobRunInput {\n    mode: KnowledgeJobMode\n  }\n\n  input KnowledgePacketInput {\n    task: String!\n    writeScope: [String!]\n    limit: Int\n    detail: KnowledgePacketDetail\n    budgetLimit: Int\n  }\n\n  type Query {\n    status: KnowledgeStatus!\n    connectors: [KnowledgeConnector!]!\n    connectorDoctor(id: String!): KnowledgeConnectorDoctorReport\n    sources(limit: Int, offset: Int, knowledgeSpaceId: String): [KnowledgeSource!]!\n    nodes(limit: Int, offset: Int, knowledgeSpaceId: String): [KnowledgeNode!]!\n    issues(limit: Int, offset: Int, knowledgeSpaceId: String): [KnowledgeIssue!]!\n    itemView(kind: String!, id: String!): KnowledgeItemView\n    extractions(limit: Int, sourceId: String, knowledgeSpaceId: String): [KnowledgeExtraction!]!\n    neighbors(kind: String!, id: String!, relation: String, limit: Int): [KnowledgeEdge!]!\n    search(query: String!, limit: Int, knowledgeSpaceId: String): [KnowledgeSearchResult!]!\n    packet(task: String!, writeScope: [String!], limit: Int, detail: KnowledgePacketDetail, budgetLimit: Int): KnowledgePacket!\n    projectionTargets(limit: Int): [KnowledgeProjectionTarget!]!\n    projection(kind: KnowledgeProjectionKind!, id: String, limit: Int): KnowledgeProjectionBundle!\n    jobs: [KnowledgeJob!]!\n    job(id: String!): KnowledgeJob\n    jobRuns(limit: Int, jobId: String): [KnowledgeJobRun!]!\n    usage(limit: Int, targetKind: String, targetId: String, usageKind: String): [KnowledgeUsageRecord!]!\n    usageRecords(limit: Int, targetKind: String, targetId: String, usageKind: String): [KnowledgeUsageRecord!]!\n    consolidationCandidates(limit: Int, status: String): [KnowledgeConsolidationCandidate!]!\n    consolidationReports(limit: Int): [KnowledgeConsolidationReport!]!\n    schedules(limit: Int): [KnowledgeSchedule!]!\n    sourceExtraction(sourceId: String!): KnowledgeExtraction\n  }\n\n  type Mutation {\n    ingestUrl(url: String!, sourceType: String, connectorId: String, folderPath: String, tags: [String!], sessionId: String, allowPrivateHosts: Boolean): KnowledgeSource!\n    ingestArtifact(path: String!, connectorId: String, tags: [String!], sessionId: String, allowPrivateHosts: Boolean): KnowledgeSource!\n    importBookmarksFromFile(path: String!, sessionId: String, allowPrivateHosts: Boolean): KnowledgeBatchIngestResult!\n    importUrlsFromFile(path: String!, sessionId: String, allowPrivateHosts: Boolean): KnowledgeBatchIngestResult!\n    ingestBookmarkSeeds(seeds: [String!]!, connectorId: String, sessionId: String): KnowledgeBatchIngestResult!\n    ingestWithConnector(connectorId: String!, input: String!, sessionId: String, allowPrivateHosts: Boolean): KnowledgeBatchIngestResult!\n    ingestConnectorInput(input: String!, connectorId: String, sessionId: String, allowPrivateHosts: Boolean): KnowledgeBatchIngestResult!\n    renderProjection(kind: KnowledgeProjectionKind!, id: String, limit: Int): KnowledgeProjectionBundle!\n    materializeProjection(kind: KnowledgeProjectionKind!, id: String, limit: Int): KnowledgeMaterializedProjection!\n    reindex: KnowledgeStatus!\n    runJob(id: String!, mode: KnowledgeJobMode): KnowledgeJobRun!\n    saveSchedule(id: String, jobId: String!, label: String!, enabled: Boolean!, schedule: JSON!, metadata: JSON): KnowledgeSchedule!\n    deleteSchedule(id: String!): Boolean!\n    setScheduleEnabled(id: String!, enabled: Boolean!): KnowledgeSchedule\n    decideConsolidationCandidate(id: String!, status: String!, decidedBy: String): KnowledgeConsolidationCandidate!\n    lint: [KnowledgeIssue!]!\n  }\n\n  type KnowledgeMaterializedProjection {\n    artifact: ArtifactDescriptor!\n    bundle: KnowledgeProjectionBundle!\n  }\n";
 export declare function parseJsonAst(node: ValueNode): unknown;
 export declare function installJsonScalar(schema: ReturnType<typeof buildSchema>): void;
 export declare function toProjectionKind(value: string): KnowledgeProjectionTargetKind;

package/dist/_internal/platform/knowledge/graphql-schema.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"graphql-schema.d.ts","sourceRoot":"","sources":["../../../../src/_internal/platform/knowledge/graphql-schema.ts"],"names":[],"mappings":"AAAA,OAAO,EAIL,WAAW,EACX,KAAK,YAAY,EACjB,KAAK,SAAS,EACf,MAAM,SAAS,CAAC;AACjB,OAAO,KAAK,EAAE,qBAAqB,EAAE,6BAA6B,EAAE,MAAM,YAAY,CAAC;AAEvF,eAAO,MAAM,qBAAqB,siUA8YjC,CAAC;AAEF,wBAAgB,YAAY,CAAC,IAAI,EAAE,SAAS,GAAG,OAAO,CAoBrD;AAED,wBAAgB,iBAAiB,CAAC,MAAM,EAAE,UAAU,CAAC,OAAO,WAAW,CAAC,GAAG,IAAI,CAQ9E;AAED,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,MAAM,GAAG,6BAA6B,CAuB7E;AAED,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,MAAM,CAiBlE;AAED,wBAAgB,cAAc,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,qBAAqB,CAM/E;AAED,wBAAgB,kBAAkB,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,MAAM,CAEpE;AAED,wBAAgB,SAAS,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,QAAQ,GAAG,YAAY,CAE5E;AAED,wBAAgB,aAAa,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,MAAM,CAE/D;AAED,wBAAgB,mBAAmB,CAAC,CAAC,SAAS;IAAE,IAAI,CAAC,EAAE,MAAM,CAAA;CAAE,EAAE,MAAM,EAAE,CAAC,GAAG,CAAC,GAAG;IAAE,IAAI,EAAE,MAAM,CAAA;CAAE,CAKhG;AAED,wBAAgB,mBAAmB,CAAC,CAAC,SAAS;IAAE,MAAM,EAAE;QAAE,IAAI,CAAC,EAAE,MAAM,CAAA;KAAE,CAAC;IAAC,KAAK,EAAE,SAAS,OAAO,EAAE,CAAA;CAAE,EAAE,MAAM,EAAE,CAAC,GAAG,CAAC,GAAG;IAAE,MAAM,EAAE;QAAE,IAAI,EAAE,MAAM,CAAA;KAAE,CAAA;CAAE,CAKnJ;AAED,wBAAgB,SAAS,CAAC,CAAC,SAAS;IAAE,MAAM,CAAC,EAAE,MAAM,CAAA;CAAE,EAAE,MAAM,EAAE,CAAC,GAAG,CAAC,GAAG;IAAE,MAAM,EAAE,MAAM,CAAA;CAAE,CAK1F;AAED,wBAAgB,MAAM,CAAC,CAAC,SAAS;IAAE,WAAW,CAAC,EAAE,MAAM,CAAA;CAAE,EAAE,GAAG,EAAE,CAAC,GAAG,CAAC,GAAG;IAAE,WAAW,EAAE,MAAM,CAAA;CAAE,CAK9F;AAED,wBAAgB,SAAS,CAAC,CAAC,SAAS;IAAE,IAAI,CAAC,EAAE,MAAM,CAAA;CAAE,EAAE,GAAG,EAAE,CAAC,GAAG,CAAC,GAAG;IAAE,IAAI,EAAE,MAAM,CAAA;CAAE,CAKnF;AAED,wBAAgB,QAAQ,CAAC,KAAK,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,EAAE,QAAQ,EAAE,MAAM,GAAG,MAAM,CAEnF;AAED,wBAAgB,WAAW,CAAC,KAAK,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,GAAG,MAAM,CAEpE;AAED,wBAAgB,aAAa,CAAC,QAAQ,EAAE,YAAY,EAAE,aAAa,CAAC,EAAE,MAAM,GAAG,OAAO,GAAG,UAAU,CASlG"}
+{"version":3,"file":"graphql-schema.d.ts","sourceRoot":"","sources":["../../../../src/_internal/platform/knowledge/graphql-schema.ts"],"names":[],"mappings":"AAAA,OAAO,EAIL,WAAW,EACX,KAAK,YAAY,EACjB,KAAK,SAAS,EACf,MAAM,SAAS,CAAC;AACjB,OAAO,KAAK,EAAE,qBAAqB,EAAE,6BAA6B,EAAE,MAAM,YAAY,CAAC;AAEvF,eAAO,MAAM,qBAAqB,wqUA8YjC,CAAC;AAEF,wBAAgB,YAAY,CAAC,IAAI,EAAE,SAAS,GAAG,OAAO,CAoBrD;AAED,wBAAgB,iBAAiB,CAAC,MAAM,EAAE,UAAU,CAAC,OAAO,WAAW,CAAC,GAAG,IAAI,CAQ9E;AAED,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,MAAM,GAAG,6BAA6B,CAuB7E;AAED,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,MAAM,CAiBlE;AAED,wBAAgB,cAAc,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,qBAAqB,CAM/E;AAED,wBAAgB,kBAAkB,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,MAAM,CAEpE;AAED,wBAAgB,SAAS,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,QAAQ,GAAG,YAAY,CAE5E;AAED,wBAAgB,aAAa,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,MAAM,CAE/D;AAED,wBAAgB,mBAAmB,CAAC,CAAC,SAAS;IAAE,IAAI,CAAC,EAAE,MAAM,CAAA;CAAE,EAAE,MAAM,EAAE,CAAC,GAAG,CAAC,GAAG;IAAE,IAAI,EAAE,MAAM,CAAA;CAAE,CAKhG;AAED,wBAAgB,mBAAmB,CAAC,CAAC,SAAS;IAAE,MAAM,EAAE;QAAE,IAAI,CAAC,EAAE,MAAM,CAAA;KAAE,CAAC;IAAC,KAAK,EAAE,SAAS,OAAO,EAAE,CAAA;CAAE,EAAE,MAAM,EAAE,CAAC,GAAG,CAAC,GAAG;IAAE,MAAM,EAAE;QAAE,IAAI,EAAE,MAAM,CAAA;KAAE,CAAA;CAAE,CAKnJ;AAED,wBAAgB,SAAS,CAAC,CAAC,SAAS;IAAE,MAAM,CAAC,EAAE,MAAM,CAAA;CAAE,EAAE,MAAM,EAAE,CAAC,GAAG,CAAC,GAAG;IAAE,MAAM,EAAE,MAAM,CAAA;CAAE,CAK1F;AAED,wBAAgB,MAAM,CAAC,CAAC,SAAS;IAAE,WAAW,CAAC,EAAE,MAAM,CAAA;CAAE,EAAE,GAAG,EAAE,CAAC,GAAG,CAAC,GAAG;IAAE,WAAW,EAAE,MAAM,CAAA;CAAE,CAK9F;AAED,wBAAgB,SAAS,CAAC,CAAC,SAAS;IAAE,IAAI,CAAC,EAAE,MAAM,CAAA;CAAE,EAAE,GAAG,EAAE,CAAC,GAAG,CAAC,GAAG;IAAE,IAAI,EAAE,MAAM,CAAA;CAAE,CAKnF;AAED,wBAAgB,QAAQ,CAAC,KAAK,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,EAAE,QAAQ,EAAE,MAAM,GAAG,MAAM,CAEnF;AAED,wBAAgB,WAAW,CAAC,KAAK,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,GAAG,MAAM,CAEpE;AAED,wBAAgB,aAAa,CAAC,QAAQ,EAAE,YAAY,EAAE,aAAa,CAAC,EAAE,MAAM,GAAG,OAAO,GAAG,UAAU,CASlG"}

package/dist/_internal/platform/knowledge/graphql-schema.js

@@ -353,13 +353,13 @@ export const KNOWLEDGE_GRAPHQL_SDL = `
     status: KnowledgeStatus!
     connectors: [KnowledgeConnector!]!
     connectorDoctor(id: String!): KnowledgeConnectorDoctorReport
-    sources(limit: Int, offset: Int): [KnowledgeSource!]!
-    nodes(limit: Int, offset: Int): [KnowledgeNode!]!
-    issues(limit: Int, offset: Int): [KnowledgeIssue!]!
+    sources(limit: Int, offset: Int, knowledgeSpaceId: String): [KnowledgeSource!]!
+    nodes(limit: Int, offset: Int, knowledgeSpaceId: String): [KnowledgeNode!]!
+    issues(limit: Int, offset: Int, knowledgeSpaceId: String): [KnowledgeIssue!]!
     itemView(kind: String!, id: String!): KnowledgeItemView
-    extractions(limit: Int, sourceId: String): [KnowledgeExtraction!]!
+    extractions(limit: Int, sourceId: String, knowledgeSpaceId: String): [KnowledgeExtraction!]!
     neighbors(kind: String!, id: String!, relation: String, limit: Int): [KnowledgeEdge!]!
-    search(query: String!, limit: Int): [KnowledgeSearchResult!]!
+    search(query: String!, limit: Int, knowledgeSpaceId: String): [KnowledgeSearchResult!]!
     packet(task: String!, writeScope: [String!], limit: Int, detail: KnowledgePacketDetail, budgetLimit: Int): KnowledgePacket!
     projectionTargets(limit: Int): [KnowledgeProjectionTarget!]!
     projection(kind: KnowledgeProjectionKind!, id: String, limit: Int): KnowledgeProjectionBundle!

package/dist/_internal/platform/knowledge/graphql.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"graphql.d.ts","sourceRoot":"","sources":["../../../../src/_internal/platform/knowledge/graphql.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,cAAc,CAAC;AAiBrD,MAAM,WAAW,6BAA6B;IAC5C,QAAQ,CAAC,SAAS,EAAE,OAAO,GAAG,UAAU,CAAC;IACzC,QAAQ,CAAC,cAAc,EAAE,SAAS,MAAM,EAAE,CAAC;IAC3C,QAAQ,CAAC,aAAa,EAAE,OAAO,CAAC;CACjC;AAED,wBAAgB,6BAA6B,CAC3C,MAAM,EAAE,MAAM,EACd,aAAa,CAAC,EAAE,MAAM,GACrB,6BAA6B,CAM/B;AAiBD,MAAM,WAAW,4BAA4B;IAC3C,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;IACvB,QAAQ,CAAC,aAAa,CAAC,EAAE,MAAM,CAAC;IAChC,QAAQ,CAAC,SAAS,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IAC7C,QAAQ,CAAC,KAAK,EAAE,OAAO,CAAC;IACxB,QAAQ,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC;CACpC;AAED,qBAAa,uBAAuB;IAStB,OAAO,CAAC,QAAQ,CAAC,OAAO;IARpC,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,MAAM,CAIzB;IAEL,MAAM,CAAC,QAAQ,CAAC,SAAS,SAA+C;gBAE3C,OAAO,EAAE,gBAAgB;IAEtD,IAAI,UAAU,IAAI,MAAM,CAEvB;IAEK,OAAO,CAAC,KAAK,EAAE,4BAA4B;IAkBjD,OAAO,CAAC,eAAe;CAoOxB;AAED,wBAAgB,6BAA6B,IAAI,MAAM,CAEtD"}
+{"version":3,"file":"graphql.d.ts","sourceRoot":"","sources":["../../../../src/_internal/platform/knowledge/graphql.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,cAAc,CAAC;AAkBrD,MAAM,WAAW,6BAA6B;IAC5C,QAAQ,CAAC,SAAS,EAAE,OAAO,GAAG,UAAU,CAAC;IACzC,QAAQ,CAAC,cAAc,EAAE,SAAS,MAAM,EAAE,CAAC;IAC3C,QAAQ,CAAC,aAAa,EAAE,OAAO,CAAC;CACjC;AAED,wBAAgB,6BAA6B,CAC3C,MAAM,EAAE,MAAM,EACd,aAAa,CAAC,EAAE,MAAM,GACrB,6BAA6B,CAM/B;AAiBD,MAAM,WAAW,4BAA4B;IAC3C,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;IACvB,QAAQ,CAAC,aAAa,CAAC,EAAE,MAAM,CAAC;IAChC,QAAQ,CAAC,SAAS,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IAC7C,QAAQ,CAAC,KAAK,EAAE,OAAO,CAAC;IACxB,QAAQ,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC;CACpC;AAED,qBAAa,uBAAuB;IAStB,OAAO,CAAC,QAAQ,CAAC,OAAO;IARpC,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,MAAM,CAIzB;IAEL,MAAM,CAAC,QAAQ,CAAC,SAAS,SAA+C;gBAE3C,OAAO,EAAE,gBAAgB;IAEtD,IAAI,UAAU,IAAI,MAAM,CAEvB;IAEK,OAAO,CAAC,KAAK,EAAE,4BAA4B;IAkBjD,OAAO,CAAC,eAAe;CAwPxB;AAED,wBAAgB,6BAA6B,IAAI,MAAM,CAEtD"}

package/dist/_internal/platform/knowledge/graphql.js

@@ -1,5 +1,6 @@
 import { GraphQLError, buildSchema, graphql, parse, printSchema } from 'graphql';
 import { KNOWLEDGE_GRAPHQL_SDL, clampInt, clampOffset, installJsonScalar, mapJob, mapJobRun, mapPacket, mapProjectionBundle, mapProjectionTarget, pickOperation, toJobMode, toPacketDetail, toProjectionKind, } from './graphql-schema.js';
+import { isInKnowledgeSpace } from './spaces.js';
 export function inspectKnowledgeGraphqlAccess(source, operationName) {
     const document = parse(source);
     const operation = pickOperation(document, operationName);
@@ -49,9 +50,15 @@ export class KnowledgeGraphqlService {
     createRootValue() {
         return {
             status: async () => this.service.getStatus(),
-            sources: ({ limit }) => this.service.listSources(clampInt(limit, 100)),
-            nodes: ({ limit }) => this.service.listNodes(clampInt(limit, 100)),
-            issues: ({ limit }) => this.service.listIssues(clampInt(limit, 100)),
+            sources: ({ limit, knowledgeSpaceId }) => (this.service.listSources(knowledgeSpaceId ? 10_000 : clampInt(limit, 100))
+                .filter((source) => !knowledgeSpaceId || isInKnowledgeSpace(source, knowledgeSpaceId))
+                .slice(0, clampInt(limit, 100))),
+            nodes: ({ limit, knowledgeSpaceId }) => (this.service.listNodes(knowledgeSpaceId ? 10_000 : clampInt(limit, 100))
+                .filter((node) => !knowledgeSpaceId || isInKnowledgeSpace(node, knowledgeSpaceId))
+                .slice(0, clampInt(limit, 100))),
+            issues: ({ limit, knowledgeSpaceId }) => (this.service.listIssues(knowledgeSpaceId ? 10_000 : clampInt(limit, 100))
+                .filter((issue) => !knowledgeSpaceId || isInKnowledgeSpace(issue, knowledgeSpaceId))
+                .slice(0, clampInt(limit, 100))),
             source: ({ id }) => this.service.listSources(10_000).find((source) => source.id === id) ?? null,
             node: ({ id }) => this.service.listNodes(10_000).find((node) => node.id === id) ?? null,
             issue: ({ id }) => this.service.listIssues(10_000).find((issue) => issue.id === id) ?? null,
@@ -81,7 +88,9 @@ export class KnowledgeGraphqlService {
                 code: args.code,
                 query: args.query,
             }),
-            extractions: ({ limit, sourceId }) => this.service.listExtractions(clampInt(limit, 100), sourceId),
+            extractions: ({ limit, sourceId, knowledgeSpaceId }) => (this.service.listExtractions(clampInt(limit, 100), sourceId)
+                .filter((extraction) => !knowledgeSpaceId || isInKnowledgeSpace(extraction, knowledgeSpaceId))
+                .slice(0, clampInt(limit, 100))),
             sourceExtraction: ({ sourceId }) => this.service.getSourceExtraction(sourceId),
             neighbors: ({ kind, id, relation, limit }) => {
                 if (kind !== 'source' && kind !== 'node') {
@@ -89,7 +98,9 @@ export class KnowledgeGraphqlService {
                 }
                 return this.service.getNeighbors(kind, id, { relation, limit: clampInt(limit, 20) });
             },
-            search: ({ query, limit }) => this.service.search(query, clampInt(limit, 10)),
+            search: ({ query, limit, knowledgeSpaceId }) => (this.service.search(query, knowledgeSpaceId ? 10_000 : clampInt(limit, 10))
+                .filter((result) => !knowledgeSpaceId || isInKnowledgeSpace(result.source ?? result.node, knowledgeSpaceId))
+                .slice(0, clampInt(limit, 10))),
             packet: async ({ task, writeScope, limit, detail, budgetLimit }) => mapPacket(await this.service.buildPacket(task, writeScope ?? [], clampInt(limit, 6), {
                 detail: toPacketDetail(detail),
                 ...(typeof budgetLimit === 'number' ? { budgetLimit } : {}),