@moxn/kb-migrate 0.4.28 → 0.4.30

package/dist/sources/onenote/onenote-tree.js

@@ -0,0 +1,131 @@
+/**
+ * Discovery: walk Microsoft Graph's OneNote hierarchy and flatten into a
+ * list of pages with their computed KB path.
+ *
+ * The hierarchy is: Notebook → SectionGroup* → Section → Page.
+ * SectionGroups can nest; OneNote caps at 4 levels of nesting in practice
+ * and we cap recursion at 4 to avoid accidental cycles.
+ *
+ * Selection semantics: callers pass a set of selected Graph IDs (any mix of
+ * notebook / sectionGroup / section / page). A page is included iff its
+ * own ID or any ancestor's ID is in the set. If the set is empty, every
+ * page is included.
+ */
+const MAX_GROUP_DEPTH = 4;
+/**
+ * Normalize any human text into a ltree-safe slug segment.
+ * - lowercases
+ * - replaces spaces/underscores/periods with hyphens (ltree forbids _ and .)
+ * - drops anything outside [a-z0-9-]
+ * - collapses runs of hyphens; trims leading/trailing hyphens
+ * Falls back to `untitled` when the result is empty.
+ */
+export function slugifyPathSegment(raw) {
+    const ascii = raw
+        .normalize('NFKD')
+        .replace(/[̀-ͯ]/g, '') // strip combining marks
+        .toLowerCase()
+        .replace(/[\s_.]+/g, '-')
+        .replace(/[^a-z0-9-]/g, '-')
+        .replace(/-+/g, '-')
+        .replace(/^-+|-+$/g, '');
+    return ascii.length > 0 ? ascii : 'untitled';
+}
+export async function discoverOneNotePages(client, opts = {}) {
+    const pathPrefix = opts.pathPrefix
+        ? opts.pathPrefix.replace(/^\/+|\/+$/g, '')
+        : 'imported/onenote';
+    const sel = opts.selectedIds && opts.selectedIds.size > 0 ? opts.selectedIds : null;
+    const notebooks = await client.listNotebooks();
+    const out = [];
+    for (const notebook of notebooks) {
+        const notebookSelected = !sel || sel.has(notebook.id);
+        // When nothing at notebook-or-below is selected we can skip entirely.
+        // Cheap upper bound: if sel is set and notebook isn't in it, we still
+        // need to walk in case a descendant is selected.
+        const notebookSlug = slugifyPathSegment(notebook.displayName);
+        await walkNotebook(client, notebook, [pathPrefix, notebookSlug], notebookSelected, sel, opts, out);
+    }
+    out.sort((a, b) => a.kbPath.localeCompare(b.kbPath));
+    return out;
+}
+async function walkNotebook(client, notebook, pathSegs, ancestorSelected, sel, opts, out) {
+    const [sectionGroups, sections] = await Promise.all([
+        client.listSectionGroups(notebook.id),
+        client.listSectionsInNotebook(notebook.id),
+    ]);
+    const ctx = {
+        notebookId: notebook.id,
+        notebookName: notebook.displayName,
+        sectionGroups: [],
+    };
+    for (const section of sections) {
+        const sectionSelected = ancestorSelected || (!!sel && sel.has(section.id));
+        await walkSection(client, section, pathSegs, ctx, sectionSelected, sel, opts, out);
+    }
+    for (const group of sectionGroups) {
+        await walkSectionGroup(client, group, pathSegs, ctx, ancestorSelected, sel, opts, out, 1);
+    }
+}
+async function walkSectionGroup(client, group, pathSegs, ctx, ancestorSelected, sel, opts, out, depth) {
+    if (depth > MAX_GROUP_DEPTH)
+        return;
+    const selfSelected = ancestorSelected || (!!sel && sel.has(group.id));
+    const nextSegs = [...pathSegs, slugifyPathSegment(group.displayName)];
+    const nextCtx = {
+        ...ctx,
+        sectionGroups: [...ctx.sectionGroups, { id: group.id, name: group.displayName }],
+    };
+    const [nestedGroups, groupSections] = await Promise.all([
+        client.listNestedSectionGroups(group.id),
+        client.listSectionsInGroup(group.id),
+    ]);
+    for (const section of groupSections) {
+        const sectionSelected = selfSelected || (!!sel && sel.has(section.id));
+        await walkSection(client, section, nextSegs, nextCtx, sectionSelected, sel, opts, out);
+    }
+    for (const nested of nestedGroups) {
+        await walkSectionGroup(client, nested, nextSegs, nextCtx, selfSelected, sel, opts, out, depth + 1);
+    }
+}
+async function walkSection(client, section, pathSegs, ctx, ancestorSelected, sel, opts, out) {
+    const pages = await client.listPages(section.id, {
+        modifiedAfter: opts.modifiedAfter,
+        modifiedBefore: opts.modifiedBefore,
+    });
+    const sectionSeg = slugifyPathSegment(section.displayName);
+    for (const page of pages) {
+        const pageSelected = ancestorSelected || (!!sel && sel.has(page.id));
+        if (!pageSelected)
+            continue;
+        const pageSeg = slugifyPathSegment(page.title ?? 'untitled');
+        out.push(toDiscovered(page, section, [...pathSegs, sectionSeg, pageSeg], ctx));
+    }
+}
+function toDiscovered(page, section, segs, ctx) {
+    return {
+        onenotePageId: page.id,
+        title: page.title ?? 'Untitled',
+        kbPath: '/' + segs.join('/'),
+        notebookId: ctx.notebookId,
+        notebookName: ctx.notebookName,
+        sectionId: section.id,
+        sectionName: section.displayName,
+        sectionGroups: ctx.sectionGroups,
+        onenoteWebUrl: page.links?.oneNoteWebUrl?.href,
+        createdDateTime: page.createdDateTime,
+        lastModifiedDateTime: page.lastModifiedDateTime,
+        createdByEmail: page.createdBy?.user?.email ?? page.createdBy?.user?.userPrincipalName,
+        createdByDisplayName: page.createdBy?.user?.displayName,
+        lastModifiedByEmail: page.lastModifiedBy?.user?.email ??
+            page.lastModifiedBy?.user?.userPrincipalName,
+    };
+}
+/**
+ * Join kbPath segments into `/a/b/c` — public helper used by the orchestrator
+ * when it needs to derive a path outside the discovery walk.
+ */
+export function joinKbPath(...segments) {
+    const filtered = segments.filter((s) => Boolean(s && s.length));
+    return '/' + filtered.map((s) => s.replace(/^\/+|\/+$/g, '')).join('/');
+}

package/dist/sources/onenote/types.d.ts

@@ -0,0 +1,124 @@
+/**
+ * Shared types for the OneNote source (Microsoft Graph).
+ *
+ * These mirror the Graph API payloads we actually read. Graph's own TypeScript
+ * types are available via @microsoft/microsoft-graph-types but include every
+ * field of every shape; we trim to what the importer touches.
+ */
+export interface GraphIdentity {
+    user?: {
+        id?: string;
+        displayName?: string;
+        email?: string;
+        /** work/school accounts — full UPN */
+        userPrincipalName?: string;
+    };
+}
+export interface OneNoteNotebook {
+    id: string;
+    displayName: string;
+    isDefault?: boolean;
+    isShared?: boolean;
+    createdDateTime?: string;
+    lastModifiedDateTime?: string;
+    createdBy?: GraphIdentity;
+    lastModifiedBy?: GraphIdentity;
+    links?: {
+        oneNoteClientUrl?: {
+            href: string;
+        };
+        oneNoteWebUrl?: {
+            href: string;
+        };
+    };
+}
+export interface OneNoteSectionGroup {
+    id: string;
+    displayName: string;
+    /** ID of the notebook this group lives in */
+    parentNotebookId?: string;
+    /** ID of parent section group (for nested groups) */
+    parentSectionGroupId?: string;
+    createdDateTime?: string;
+    lastModifiedDateTime?: string;
+}
+export interface OneNoteSection {
+    id: string;
+    displayName: string;
+    parentNotebookId?: string;
+    parentSectionGroupId?: string;
+    createdDateTime?: string;
+    lastModifiedDateTime?: string;
+}
+export interface OneNotePage {
+    id: string;
+    title: string;
+    createdDateTime?: string;
+    lastModifiedDateTime?: string;
+    contentUrl?: string;
+    parentSection?: {
+        id: string;
+        displayName?: string;
+    };
+    parentNotebook?: {
+        id: string;
+        displayName?: string;
+    };
+    createdBy?: GraphIdentity;
+    lastModifiedBy?: GraphIdentity;
+    links?: {
+        oneNoteClientUrl?: {
+            href: string;
+        };
+        oneNoteWebUrl?: {
+            href: string;
+        };
+    };
+}
+/**
+ * A page flattened with its full derived kbPath, ready for the fan-out task.
+ */
+export interface DiscoveredOneNotePage {
+    onenotePageId: string;
+    title: string;
+    kbPath: string;
+    notebookId: string;
+    notebookName: string;
+    sectionId: string;
+    sectionName: string;
+    /** Zero or more nested section groups, outermost first */
+    sectionGroups: Array<{
+        id: string;
+        name: string;
+    }>;
+    onenoteWebUrl?: string;
+    createdDateTime?: string;
+    lastModifiedDateTime?: string;
+    createdByEmail?: string;
+    createdByDisplayName?: string;
+    lastModifiedByEmail?: string;
+}
+/**
+ * A cross-page reference captured during HTML conversion, to be resolved
+ * into a KB cross-link after all pages are imported.
+ */
+export interface OneNoteExtractedReference {
+    sectionIndex: number;
+    targetOnenotePageId: string;
+    displayText: string;
+}
+/**
+ * Output of the HTML converter: what the fan-out task hands to
+ * `create_document` along with per-page cross-references.
+ */
+export interface ConvertedOneNotePage {
+    title: string;
+    sections: import('../../types.js').SectionInput[];
+    extractedReferences: OneNoteExtractedReference[];
+    /** Count of images/attachments that were re-hosted */
+    mediaCount: number;
+    /** Count of content items skipped (unknown MIME, oversized, etc.) */
+    skippedItemCount: number;
+    /** Human-readable reasons for skipped items, logged back to the job. */
+    skippedReasons: string[];
+}

package/dist/sources/onenote/types.js

@@ -0,0 +1,8 @@
+/**
+ * Shared types for the OneNote source (Microsoft Graph).
+ *
+ * These mirror the Graph API payloads we actually read. Graph's own TypeScript
+ * types are available via @microsoft/microsoft-graph-types but include every
+ * field of every shape; we trim to what the importer touches.
+ */
+export {};

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@moxn/kb-migrate",
-  "version": "0.4.28",
+  "version": "0.4.30",
   "description": "Migration tool for importing documents into Moxn Knowledge Base from local files, Notion, Google Docs, and more",
   "type": "module",
   "main": "dist/index.js",
@@ -38,6 +38,26 @@
       "types": "./dist/sources/local.d.ts",
       "default": "./dist/sources/local.js"
     },
+    "./sources/onenote": {
+      "types": "./dist/sources/onenote/index.d.ts",
+      "default": "./dist/sources/onenote/index.js"
+    },
+    "./sources/onenote/onenote-api": {
+      "types": "./dist/sources/onenote/onenote-api.d.ts",
+      "default": "./dist/sources/onenote/onenote-api.js"
+    },
+    "./sources/onenote/onenote-auth": {
+      "types": "./dist/sources/onenote/onenote-auth.d.ts",
+      "default": "./dist/sources/onenote/onenote-auth.js"
+    },
+    "./sources/onenote/onenote-html": {
+      "types": "./dist/sources/onenote/onenote-html.d.ts",
+      "default": "./dist/sources/onenote/onenote-html.js"
+    },
+    "./sources/onenote/onenote-tree": {
+      "types": "./dist/sources/onenote/onenote-tree.d.ts",
+      "default": "./dist/sources/onenote/onenote-tree.js"
+    },
     "./sources/base": {
       "types": "./dist/sources/base.d.ts",
       "default": "./dist/sources/base.js"
@@ -82,11 +102,14 @@
     "prepublishOnly": "npm run build"
   },
   "dependencies": {
+    "@azure/msal-node": "^3.8.10",
+    "@microsoft/microsoft-graph-client": "^3.0.7",
     "@moxn/kb-migrate": "^0.4.14",
     "@notionhq/client": "^5.9.0",
     "@tryfabric/martian": "^1.2.4",
     "commander": "^12.0.0",
     "glob": "^10.0.0",
+    "node-html-parser": "^6.1.13",
     "remark-parse": "^11.0.0",
     "unified": "^11.0.0"
   },