From aecbe6ae8b3dbc7bcdcf33f1c8c086dafb77eb24 Mon Sep 17 00:00:00 2001 From: Mohamed Bassem Date: Sat, 30 Aug 2025 15:26:02 +0000 Subject: fix: handle list with slashes in their names and truncate long list names. fixes #1597 --- apps/web/app/api/bookmarks/export/route.tsx | 8 +- apps/web/lib/exportBookmarks.ts | 111 ----------- apps/web/lib/hooks/useBookmarkImport.ts | 272 ++++++++------------------ apps/web/lib/importBookmarkParser.ts | 286 ---------------------------- 4 files changed, 86 insertions(+), 591 deletions(-) delete mode 100644 apps/web/lib/exportBookmarks.ts delete mode 100644 apps/web/lib/importBookmarkParser.ts (limited to 'apps') diff --git a/apps/web/app/api/bookmarks/export/route.tsx b/apps/web/app/api/bookmarks/export/route.tsx index 47fdeebc..ad309877 100644 --- a/apps/web/app/api/bookmarks/export/route.tsx +++ b/apps/web/app/api/bookmarks/export/route.tsx @@ -1,12 +1,12 @@ import { NextRequest } from "next/server"; +import { api, createContextFromRequest } from "@/server/api/client"; +import { z } from "zod"; + import { toExportFormat, toNetscapeFormat, zExportSchema, -} from "@/lib/exportBookmarks"; -import { api, createContextFromRequest } from "@/server/api/client"; -import { z } from "zod"; - +} from "@karakeep/shared/import-export"; import { MAX_NUM_BOOKMARKS_PER_PAGE } from "@karakeep/shared/types/bookmarks"; export const dynamic = "force-dynamic"; diff --git a/apps/web/lib/exportBookmarks.ts b/apps/web/lib/exportBookmarks.ts deleted file mode 100644 index 5dc26e78..00000000 --- a/apps/web/lib/exportBookmarks.ts +++ /dev/null @@ -1,111 +0,0 @@ -import { z } from "zod"; - -import { BookmarkTypes, ZBookmark } from "@karakeep/shared/types/bookmarks"; - -export const zExportBookmarkSchema = z.object({ - createdAt: z.number(), - title: z.string().nullable(), - tags: z.array(z.string()), - content: z - .discriminatedUnion("type", [ - z.object({ - type: z.literal(BookmarkTypes.LINK), - url: z.string(), - }), - z.object({ - type: z.literal(BookmarkTypes.TEXT), - text: z.string(), - }), - ]) - .nullable(), - note: z.string().nullable(), - archived: z.boolean().optional().default(false), -}); - -export const zExportSchema = z.object({ - bookmarks: z.array(zExportBookmarkSchema), -}); - -export function toExportFormat( - bookmark: ZBookmark, -): z.infer { - let content = null; - switch (bookmark.content.type) { - case BookmarkTypes.LINK: { - content = { - type: bookmark.content.type, - url: bookmark.content.url, - }; - break; - } - case BookmarkTypes.TEXT: { - content = { - type: bookmark.content.type, - text: bookmark.content.text, - }; - break; - } - // Exclude asset types for now - } - return { - createdAt: Math.floor(bookmark.createdAt.getTime() / 1000), - title: - bookmark.title ?? - (bookmark.content.type === BookmarkTypes.LINK - ? (bookmark.content.title ?? null) - : null), - tags: bookmark.tags.map((t) => t.name), - content, - note: bookmark.note ?? null, - archived: bookmark.archived, - }; -} - -export function toNetscapeFormat(bookmarks: ZBookmark[]): string { - const header = ` - - -Bookmarks -

Bookmarks

-

`; - - const footer = `

`; - - const bookmarkEntries = bookmarks - .map((bookmark) => { - if (bookmark.content?.type !== BookmarkTypes.LINK) { - return ""; - } - const addDate = bookmark.createdAt - ? `ADD_DATE="${Math.floor(bookmark.createdAt.getTime() / 1000)}"` - : ""; - - const tagNames = bookmark.tags.map((t) => t.name).join(","); - const tags = tagNames.length > 0 ? `TAGS="${tagNames}"` : ""; - - const encodedUrl = encodeURI(bookmark.content.url); - const displayTitle = bookmark.title ?? bookmark.content.url; - const encodedTitle = escapeHtml(displayTitle); - - return `

${encodedTitle}`; - }) - .filter(Boolean) - .join("\n"); - - return `${header}\n${bookmarkEntries}\n${footer}`; -} - -function escapeHtml(input: string): string { - const escapeMap: Record = { - "&": "&", - "'": "'", - "`": "`", - '"': """, - "<": "<", - ">": ">", - }; - - return input.replace(/[&'`"<>]/g, (match) => escapeMap[match] || ""); -} diff --git a/apps/web/lib/hooks/useBookmarkImport.ts b/apps/web/lib/hooks/useBookmarkImport.ts index 7e5f6111..de515677 100644 --- a/apps/web/lib/hooks/useBookmarkImport.ts +++ b/apps/web/lib/hooks/useBookmarkImport.ts @@ -4,16 +4,6 @@ import { useState } from "react"; import { useRouter } from "next/navigation"; import { toast } from "@/components/ui/use-toast"; import { useTranslation } from "@/lib/i18n/client"; -import { - deduplicateBookmarks, - ParsedBookmark, - parseKarakeepBookmarkFile, - parseLinkwardenBookmarkFile, - parseNetscapeBookmarkFile, - parseOmnivoreBookmarkFile, - parsePocketBookmarkFile, - parseTabSessionManagerStateFile, -} from "@/lib/importBookmarkParser"; import { useMutation } from "@tanstack/react-query"; import { @@ -24,16 +14,15 @@ import { useAddBookmarkToList, useCreateBookmarkList, } from "@karakeep/shared-react/hooks/lists"; -import { limitConcurrency } from "@karakeep/shared/concurrency"; -import { BookmarkTypes } from "@karakeep/shared/types/bookmarks"; - -export type ImportSource = - | "html" - | "pocket" - | "omnivore" - | "karakeep" - | "linkwarden" - | "tab-session-manager"; +import { + importBookmarksFromFile, + ImportSource, + ParsedBookmark, +} from "@karakeep/shared/import-export"; +import { + BookmarkTypes, + MAX_BOOKMARK_TITLE_LENGTH, +} from "@karakeep/shared/types/bookmarks"; export interface ImportProgress { done: number; @@ -53,53 +42,6 @@ export function useBookmarkImport() { const { mutateAsync: addToList } = useAddBookmarkToList(); const { mutateAsync: updateTags } = useUpdateBookmarkTags(); - const { mutateAsync: parseAndCreateBookmark } = useMutation({ - mutationFn: async (toImport: { - bookmark: ParsedBookmark; - listIds: string[]; - }) => { - const bookmark = toImport.bookmark; - if (bookmark.content === undefined) { - throw new Error("Content is undefined"); - } - const created = await createBookmark({ - crawlPriority: "low", - title: bookmark.title, - createdAt: bookmark.addDate - ? new Date(bookmark.addDate * 1000) - : undefined, - note: bookmark.notes, - archived: bookmark.archived, - ...(bookmark.content.type === BookmarkTypes.LINK - ? { - type: BookmarkTypes.LINK, - url: bookmark.content.url, - } - : { - type: BookmarkTypes.TEXT, - text: bookmark.content.text, - }), - }); - - await Promise.all([ - ...toImport.listIds.map((listId) => - addToList({ - bookmarkId: created.id, - listId, - }), - ), - bookmark.tags.length > 0 - ? updateTags({ - bookmarkId: created.id, - attach: bookmark.tags.map((t) => ({ tagName: t })), - detach: [], - }) - : undefined, - ]); - return created; - }, - }); - const uploadBookmarkFileMutation = useMutation({ mutationFn: async ({ file, @@ -108,138 +50,87 @@ export function useBookmarkImport() { file: File; source: ImportSource; }) => { - if (source === "html") { - return await parseNetscapeBookmarkFile(file); - } else if (source === "pocket") { - return await parsePocketBookmarkFile(file); - } else if (source === "karakeep") { - return await parseKarakeepBookmarkFile(file); - } else if (source === "omnivore") { - return await parseOmnivoreBookmarkFile(file); - } else if (source === "linkwarden") { - return await parseLinkwardenBookmarkFile(file); - } else if (source === "tab-session-manager") { - return await parseTabSessionManagerStateFile(file); - } else { - throw new Error("Unknown source"); - } - }, - onSuccess: async (parsedBookmarks) => { - if (parsedBookmarks.length === 0) { - toast({ description: "No bookmarks found in the file." }); - return; - } - - const rootList = await createList({ - name: t("settings.import.imported_bookmarks"), - icon: "⬆️", - }); - - const finalBookmarksToImport = deduplicateBookmarks(parsedBookmarks); - - setImportProgress({ done: 0, total: finalBookmarksToImport.length }); - - const allRequiredPaths = new Set(); - for (const bookmark of finalBookmarksToImport) { - for (const path of bookmark.paths) { - if (path && path.length > 0) { - for (let i = 1; i <= path.length; i++) { - const subPath = path.slice(0, i); - const pathKey = subPath.join("/"); - allRequiredPaths.add(pathKey); - } - } - } - } - - const allRequiredPathsArray = Array.from(allRequiredPaths).sort( - (a, b) => a.split("/").length - b.split("/").length, - ); - - const pathMap: Record = {}; - pathMap[""] = rootList.id; - - for (const pathKey of allRequiredPathsArray) { - const parts = pathKey.split("/"); - const parentKey = parts.slice(0, -1).join("/"); - const parentId = pathMap[parentKey] || rootList.id; - - const folderName = parts[parts.length - 1]; - const folderList = await createList({ - name: folderName, - parentId: parentId, - icon: "📁", - }); - pathMap[pathKey] = folderList.id; - } - - const importPromises = finalBookmarksToImport.map( - (bookmark) => async () => { - const listIds = bookmark.paths.map( - (path) => pathMap[path.join("/")] || rootList.id, - ); - if (listIds.length === 0) { - listIds.push(rootList.id); - } - - try { - const created = await parseAndCreateBookmark({ - bookmark: bookmark, + const result = await importBookmarksFromFile( + { + file, + source, + rootListName: t("settings.import.imported_bookmarks"), + deps: { + createList: createList, + createBookmark: async (bookmark: ParsedBookmark) => { + if (bookmark.content === undefined) { + throw new Error("Content is undefined"); + } + const created = await createBookmark({ + crawlPriority: "low", + title: bookmark.title.substring(0, MAX_BOOKMARK_TITLE_LENGTH), + createdAt: bookmark.addDate + ? new Date(bookmark.addDate * 1000) + : undefined, + note: bookmark.notes, + archived: bookmark.archived, + ...(bookmark.content.type === BookmarkTypes.LINK + ? { + type: BookmarkTypes.LINK, + url: bookmark.content.url, + } + : { + type: BookmarkTypes.TEXT, + text: bookmark.content.text, + }), + }); + return created as { id: string; alreadyExists?: boolean }; + }, + addBookmarkToLists: async ({ + bookmarkId, listIds, - }); - - setImportProgress((prev) => { - const newDone = (prev?.done ?? 0) + 1; - return { - done: newDone, - total: finalBookmarksToImport.length, - }; - }); - return { status: "fulfilled" as const, value: created }; - } catch { - setImportProgress((prev) => { - const newDone = (prev?.done ?? 0) + 1; - return { - done: newDone, - total: finalBookmarksToImport.length, - }; - }); - return { status: "rejected" as const }; - } + }: { + bookmarkId: string; + listIds: string[]; + }) => { + await Promise.all( + listIds.map((listId) => + addToList({ + bookmarkId, + listId, + }), + ), + ); + }, + updateBookmarkTags: async ({ + bookmarkId, + tags, + }: { + bookmarkId: string; + tags: string[]; + }) => { + if (tags.length > 0) { + await updateTags({ + bookmarkId, + attach: tags.map((t) => ({ tagName: t })), + detach: [], + }); + } + }, + }, + onProgress: (done, total) => setImportProgress({ done, total }), }, + {}, ); - - const CONCURRENCY_LIMIT = 20; - const resultsPromises = limitConcurrency( - importPromises, - CONCURRENCY_LIMIT, - ); - - const results = await Promise.all(resultsPromises); - - let successes = 0; - let failures = 0; - let alreadyExisted = 0; - - for (const result of results) { - if (result.status === "fulfilled") { - if (result.value.alreadyExists) { - alreadyExisted++; - } else { - successes++; - } - } else { - failures++; - } + return result; + }, + onSuccess: async (result) => { + if (result.counts.total === 0) { + toast({ description: "No bookmarks found in the file." }); + return; } - + const { successes, failures, alreadyExisted } = result.counts; if (successes > 0 || alreadyExisted > 0) { toast({ description: `Imported ${successes} bookmarks and skipped ${alreadyExisted} bookmarks that already existed`, variant: "default", }); } - if (failures > 0) { toast({ description: `Failed to import ${failures} bookmarks. Check console for details.`, @@ -247,7 +138,8 @@ export function useBookmarkImport() { }); } - router.push(`/dashboard/lists/${rootList.id}`); + if (result.rootListId) + router.push(`/dashboard/lists/${result.rootListId}`); }, onError: (error) => { setImportProgress(null); diff --git a/apps/web/lib/importBookmarkParser.ts b/apps/web/lib/importBookmarkParser.ts deleted file mode 100644 index 44fe872c..00000000 --- a/apps/web/lib/importBookmarkParser.ts +++ /dev/null @@ -1,286 +0,0 @@ -// Copied from https://gist.github.com/devster31/4e8c6548fd16ffb75c02e6f24e27f9b9 -import * as cheerio from "cheerio"; -import { parse } from "csv-parse/sync"; -import { z } from "zod"; - -import { BookmarkTypes } from "@karakeep/shared/types/bookmarks"; - -import { zExportSchema } from "./exportBookmarks"; - -export interface ParsedBookmark { - title: string; - content?: - | { type: BookmarkTypes.LINK; url: string } - | { type: BookmarkTypes.TEXT; text: string }; - tags: string[]; - addDate?: number; - notes?: string; - archived?: boolean; - paths: string[][]; -} - -export async function parseNetscapeBookmarkFile( - file: File, -): Promise { - const textContent = await file.text(); - - if (!textContent.startsWith("")) { - throw Error("The uploaded html file does not seem to be a bookmark file"); - } - - const $ = cheerio.load(textContent); - - return $("a") - .map(function (_index, a) { - const $a = $(a); - const addDate = $a.attr("add_date"); - let tags: string[] = []; - - const tagsStr = $a.attr("tags"); - try { - tags = tagsStr && tagsStr.length > 0 ? tagsStr.split(",") : []; - } catch { - /* empty */ - } - const url = $a.attr("href"); - - // Build folder path by traversing up the hierarchy - const path: string[] = []; - let current = $a.parent(); - while (current && current.length > 0) { - const h3 = current.find("> h3").first(); - if (h3.length > 0) { - path.unshift(h3.text()); - } - current = current.parent(); - } - - return { - title: $a.text(), - content: url ? { type: BookmarkTypes.LINK as const, url } : undefined, - tags, - addDate: typeof addDate === "undefined" ? undefined : parseInt(addDate), - paths: [path], - }; - }) - .get(); -} - -export async function parsePocketBookmarkFile( - file: File, -): Promise { - const textContent = await file.text(); - - const records = parse(textContent, { - columns: true, - skip_empty_lines: true, - }) as { - title: string; - url: string; - time_added: string; - tags: string; - status?: string; - }[]; - - return records.map((record) => { - return { - title: record.title, - content: { type: BookmarkTypes.LINK as const, url: record.url }, - tags: record.tags.length > 0 ? record.tags.split("|") : [], - addDate: parseInt(record.time_added), - archived: record.status === "archive", - paths: [], // TODO - }; - }); -} - -export async function parseKarakeepBookmarkFile( - file: File, -): Promise { - const textContent = await file.text(); - - const parsed = zExportSchema.safeParse(JSON.parse(textContent)); - if (!parsed.success) { - throw new Error( - `The uploaded JSON file contains an invalid bookmark file: ${parsed.error.toString()}`, - ); - } - - return parsed.data.bookmarks.map((bookmark) => { - let content = undefined; - if (bookmark.content?.type == BookmarkTypes.LINK) { - content = { - type: BookmarkTypes.LINK as const, - url: bookmark.content.url, - }; - } else if (bookmark.content?.type == BookmarkTypes.TEXT) { - content = { - type: BookmarkTypes.TEXT as const, - text: bookmark.content.text, - }; - } - return { - title: bookmark.title ?? "", - content, - tags: bookmark.tags, - addDate: bookmark.createdAt, - notes: bookmark.note ?? undefined, - archived: bookmark.archived, - paths: [], // TODO - }; - }); -} - -export async function parseOmnivoreBookmarkFile( - file: File, -): Promise { - const textContent = await file.text(); - const zOmnivoreExportSchema = z.array( - z.object({ - title: z.string(), - url: z.string(), - labels: z.array(z.string()), - savedAt: z.coerce.date(), - state: z.string().optional(), - }), - ); - - const parsed = zOmnivoreExportSchema.safeParse(JSON.parse(textContent)); - if (!parsed.success) { - throw new Error( - `The uploaded JSON file contains an invalid omnivore bookmark file: ${parsed.error.toString()}`, - ); - } - - return parsed.data.map((bookmark) => { - return { - title: bookmark.title ?? "", - content: { type: BookmarkTypes.LINK as const, url: bookmark.url }, - tags: bookmark.labels, - addDate: bookmark.savedAt.getTime() / 1000, - archived: bookmark.state === "Archived", - paths: [], - }; - }); -} - -export async function parseLinkwardenBookmarkFile( - file: File, -): Promise { - const textContent = await file.text(); - const zLinkwardenExportSchema = z.object({ - collections: z.array( - z.object({ - links: z.array( - z.object({ - name: z.string(), - url: z.string(), - tags: z.array(z.object({ name: z.string() })), - createdAt: z.coerce.date(), - }), - ), - }), - ), - }); - - const parsed = zLinkwardenExportSchema.safeParse(JSON.parse(textContent)); - if (!parsed.success) { - throw new Error( - `The uploaded JSON file contains an invalid Linkwarden bookmark file: ${parsed.error.toString()}`, - ); - } - - return parsed.data.collections.flatMap((collection) => { - return collection.links.map((bookmark) => ({ - title: bookmark.name ?? "", - content: { type: BookmarkTypes.LINK as const, url: bookmark.url }, - tags: bookmark.tags.map((tag) => tag.name), - addDate: bookmark.createdAt.getTime() / 1000, - paths: [], // TODO - })); - }); -} - -export async function parseTabSessionManagerStateFile( - file: File, -): Promise { - const textContent = await file.text(); - - const zTab = z.object({ - url: z.string(), - title: z.string(), - lastAccessed: z.number(), - }); - - const zSession = z.object({ - windows: z.record(z.string(), z.record(z.string(), zTab)), - date: z.number(), - }); - - const zTabSessionManagerSchema = z.array(zSession); - - const parsed = zTabSessionManagerSchema.safeParse(JSON.parse(textContent)); - if (!parsed.success) { - throw new Error( - `The uploaded JSON file contains an invalid Tab Session Manager bookmark file: ${parsed.error.toString()}`, - ); - } - - // Get the object in data that has the most recent `date` - const { windows } = parsed.data.reduce((prev, curr) => - prev.date > curr.date ? prev : curr, - ); - - return Object.values(windows).flatMap((window) => - Object.values(window).map((tab) => ({ - title: tab.title, - content: { type: BookmarkTypes.LINK as const, url: tab.url }, - tags: [], - addDate: tab.lastAccessed, - paths: [], // Tab Session Manager doesn't have folders - })), - ); -} - -export function deduplicateBookmarks( - bookmarks: ParsedBookmark[], -): ParsedBookmark[] { - const deduplicatedBookmarksMap = new Map(); - const textBookmarks: ParsedBookmark[] = []; - - for (const bookmark of bookmarks) { - if (bookmark.content?.type === BookmarkTypes.LINK) { - const url = bookmark.content.url; - if (deduplicatedBookmarksMap.has(url)) { - const existing = deduplicatedBookmarksMap.get(url)!; - // Merge tags - existing.tags = [...new Set([...existing.tags, ...bookmark.tags])]; - // Merge paths - existing.paths = [...existing.paths, ...bookmark.paths]; - const existingDate = existing.addDate ?? Infinity; - const newDate = bookmark.addDate ?? Infinity; - if (newDate < existingDate) { - existing.addDate = bookmark.addDate; - } - // Append notes if both exist - if (existing.notes && bookmark.notes) { - existing.notes = `${existing.notes}\n---\n${bookmark.notes}`; - } else if (bookmark.notes) { - existing.notes = bookmark.notes; - } - // For archived status, prefer archived if either is archived - if (bookmark.archived === true) { - existing.archived = true; - } - // Title: keep existing one for simplicity - } else { - deduplicatedBookmarksMap.set(url, bookmark); - } - } else { - // Keep text bookmarks as they are (no URL to dedupe on) - textBookmarks.push(bookmark); - } - } - - return [...deduplicatedBookmarksMap.values(), ...textBookmarks]; -} -- cgit v1.2.3-70-g09d2