Files
sori.studio/server/repositories/post-import-repository.js

535 lines
14 KiB
JavaScript

import { mkdir, stat, writeFile } from 'node:fs/promises'
import { basename, extname, join } from 'node:path'
import { readZipBufferEntries } from '../utils/zip-reader'
import { upsertMediaMetadataCategory } from '../utils/media-library'
import { getPostgresClient } from './postgres-client'
import { createAdminPost } from './content-repository'
const UPLOAD_BASE_URL = '/uploads'
const MAX_IMPORT_POSTS = 1000
const MARKDOWN_EXTENSION_PATTERN = /\.md$/i
/**
* 파일명에 안전한 문자열로 정리한다.
* @param {string} value - 원본 문자열
* @returns {string} 정리된 문자열
*/
const sanitizeFilenameSegment = (value) => String(value || '')
.trim()
.replace(/[\\/:*?"<>|]+/g, '-')
.replace(/\s+/g, '-')
.replace(/-+/g, '-')
.replace(/^-|-$/g, '')
.slice(0, 80) || 'asset'
/**
* 슬러그에 안전한 문자열로 정리한다.
* @param {string} value - 원본 슬러그
* @returns {string} 정리된 슬러그
*/
const sanitizeSlug = (value) => String(value || '')
.trim()
.toLowerCase()
.normalize('NFC')
.replace(/[^a-z0-9가-힣]+/g, '-')
.replace(/-+/g, '-')
.replace(/^-|-$/g, '')
|| 'imported-post'
/**
* YAML 문자열 따옴표를 해제한다.
* @param {string} value - YAML 값
* @returns {string} 문자열 값
*/
const unquoteYamlString = (value) => {
const trimmed = String(value || '').trim()
if (!trimmed) {
return ''
}
if (trimmed.startsWith('"') && trimmed.endsWith('"')) {
return trimmed
.slice(1, -1)
.replace(/\\"/g, '"')
.replace(/\\\\/g, '\\')
}
return trimmed
}
/**
* YAML 배열 값을 파싱한다.
* @param {string} value - YAML 배열 문자열
* @returns {Array<string>} 문자열 배열
*/
const parseYamlArray = (value) => {
const trimmed = String(value || '').trim()
if (!trimmed || trimmed === '[]') {
return []
}
if (!trimmed.startsWith('[') || !trimmed.endsWith(']')) {
return []
}
const inner = trimmed.slice(1, -1).trim()
if (!inner) {
return []
}
const values = []
let current = ''
let inQuote = false
let escaped = false
for (const char of inner) {
if (escaped) {
current += char
escaped = false
continue
}
if (char === '\\') {
current += char
escaped = true
continue
}
if (char === '"') {
current += char
inQuote = !inQuote
continue
}
if (char === ',' && !inQuote) {
values.push(unquoteYamlString(current))
current = ''
continue
}
current += char
}
if (current.trim()) {
values.push(unquoteYamlString(current))
}
return values.map((item) => item.trim()).filter(Boolean)
}
/**
* YAML 단일 값을 파싱한다.
* @param {string} value - YAML 값
* @returns {unknown} 파싱된 값
*/
const parseYamlValue = (value) => {
const trimmed = String(value || '').trim()
if (trimmed === 'null') {
return null
}
if (trimmed === 'true') {
return true
}
if (trimmed === 'false') {
return false
}
if (trimmed.startsWith('[')) {
return parseYamlArray(trimmed)
}
return unquoteYamlString(trimmed)
}
/**
* Markdown frontmatter와 본문을 분리한다.
* @param {string} markdown - Markdown 문서
* @returns {{ frontmatter: Object, content: string }} 분리 결과
*/
const parseMarkdownDocument = (markdown) => {
const normalized = String(markdown || '').replace(/^\uFEFF/, '')
if (!normalized.startsWith('---\n')) {
return {
frontmatter: {},
content: normalized
}
}
const endIndex = normalized.indexOf('\n---', 4)
if (endIndex < 0) {
return {
frontmatter: {},
content: normalized
}
}
const frontmatterText = normalized.slice(4, endIndex)
const content = normalized.slice(endIndex + 4).replace(/^\n/, '')
const frontmatter = {}
for (const line of frontmatterText.split('\n')) {
const separatorIndex = line.indexOf(':')
if (separatorIndex < 0) {
continue
}
const key = line.slice(0, separatorIndex).trim()
const value = line.slice(separatorIndex + 1)
if (key) {
frontmatter[key] = parseYamlValue(value)
}
}
return {
frontmatter,
content
}
}
/**
* ZIP 엔트리를 경로 기준 Map으로 만든다.
* @param {Array<{ path: string, data: Buffer }>} entries - ZIP 엔트리
* @returns {Map<string, Buffer>} 엔트리 맵
*/
const createZipEntryMap = (entries) => new Map(entries.map((entry) => [entry.path, entry.data]))
/**
* Markdown 파일의 상위 폴더를 조회한다.
* @param {string} path - ZIP 내부 경로
* @returns {string} 상위 폴더
*/
const getPostFolder = (path) => {
const parts = String(path || '').split('/').filter(Boolean)
parts.pop()
return parts.join('/')
}
/**
* 자산 경로를 Markdown 파일 기준 ZIP 엔트리 경로로 해석한다.
* @param {string} postFolder - 게시물 폴더
* @param {string} assetPath - Markdown 안의 자산 경로
* @returns {string} ZIP 엔트리 경로
*/
const resolveAssetEntryPath = (postFolder, assetPath) => {
const cleaned = String(assetPath || '')
.split(/[?#]/)[0]
.replace(/^\.\/+/, '')
.replace(/^\/+/, '')
if (!cleaned) {
return ''
}
const base = postFolder ? `${postFolder}/${cleaned}` : cleaned
return base
.split('/')
.filter((part) => part && part !== '.' && part !== '..')
.join('/')
}
/**
* 저장할 고유 파일명을 고른다.
* @param {string} directoryPath - 저장 디렉터리
* @param {string} originalName - 원본 파일명
* @returns {Promise<{ fileName: string, filePath: string }>} 저장 파일명과 경로
*/
const pickUniqueDiskFileName = async (directoryPath, originalName) => {
const extension = extname(originalName || '') || '.bin'
const stem = sanitizeFilenameSegment(String(originalName || '').replace(/\.[^.]+$/g, '')) || 'asset'
let suffix = 1
while (suffix < 10000) {
const fileName = suffix === 1 ? `${stem}${extension}` : `${stem}-${suffix}${extension}`
const filePath = join(directoryPath, fileName)
try {
await stat(filePath)
suffix += 1
} catch {
return {
fileName,
filePath
}
}
}
throw new Error('IMPORT_ASSET_FILENAME_FAILED')
}
/**
* Import 자산을 업로드 폴더에 저장한다.
* @param {Object} input - 저장 입력
* @param {Map<string, Buffer>} input.entryMap - ZIP 엔트리 맵
* @param {string} input.postFolder - 게시물 폴더
* @param {Set<string>} input.assetPaths - 자산 경로 목록
* @returns {Promise<Map<string, string>>} 원본 경로별 새 URL
*/
const saveImportAssets = async ({ entryMap, postFolder, assetPaths }) => {
const now = new Date()
const year = String(now.getFullYear())
const month = String(now.getMonth() + 1).padStart(2, '0')
const directoryPath = join(process.cwd(), 'public', 'uploads', 'posts', year, month)
const replacements = new Map()
await mkdir(directoryPath, { recursive: true })
for (const assetPath of assetPaths) {
const entryPath = resolveAssetEntryPath(postFolder, assetPath)
const data = entryMap.get(entryPath)
if (!data) {
continue
}
const { fileName, filePath } = await pickUniqueDiskFileName(directoryPath, basename(entryPath))
await writeFile(filePath, data)
const publicUrl = `${UPLOAD_BASE_URL}/posts/${year}/${month}/${fileName}`
await upsertMediaMetadataCategory(publicUrl, '미분류')
replacements.set(assetPath, publicUrl)
replacements.set(assetPath.replace(/^\.\//, ''), publicUrl)
replacements.set(`./${assetPath.replace(/^\.\//, '')}`, publicUrl)
}
return replacements
}
/**
* Markdown 안의 로컬 자산 경로를 새 업로드 URL로 교체한다.
* @param {string} content - 원본 본문
* @param {Map<string, string>} replacements - 경로 교체 맵
* @returns {string} 교체된 본문
*/
const replaceAssetPaths = (content, replacements) => {
let next = String(content || '')
const entries = [...replacements.entries()]
.sort((a, b) => b[0].length - a[0].length)
for (const [source, target] of entries) {
next = next.split(source).join(target)
}
return next
}
/**
* Import 대상 자산 경로를 수집한다.
* @param {Object} input - 수집 입력
* @param {Object} input.frontmatter - frontmatter
* @param {string} input.content - 본문
* @returns {Set<string>} 자산 경로 목록
*/
const collectImportAssetPaths = ({ frontmatter, content }) => {
const paths = new Set()
const localAssetPattern = /(?:\.\/)?(?:images|files)\/[^\s"'<>)]*/g
for (const match of String(content || '').match(localAssetPattern) || []) {
paths.add(match)
}
for (const key of ['featured_image', 'og_image']) {
const value = frontmatter[key]
if (typeof value === 'string' && /^(?:\.\/)?(?:images|files)\//.test(value)) {
paths.add(value)
}
}
return paths
}
/**
* 게시물 슬러그 중복을 피한다.
* @param {string} baseSlug - 기준 슬러그
* @param {Set<string>} reservedSlugs - 이번 Import에서 예약된 슬러그
* @returns {Promise<string>} 고유 슬러그
*/
const createUniquePostSlug = async (baseSlug, reservedSlugs) => {
const sql = getPostgresClient()
const base = sanitizeSlug(baseSlug)
let next = base
let suffix = 2
while (reservedSlugs.has(next)) {
next = `${base}-${suffix}`
suffix += 1
}
if (!sql) {
reservedSlugs.add(next)
return next
}
while (suffix < 10000) {
const rows = await sql`
SELECT 1
FROM posts
WHERE slug = ${next}
LIMIT 1
`
if (!rows.length && !reservedSlugs.has(next)) {
reservedSlugs.add(next)
return next
}
next = `${base}-${suffix}`
suffix += 1
}
throw new Error('IMPORT_SLUG_FAILED')
}
/**
* 게시물 상태를 Import 가능한 값으로 정리한다.
* @param {unknown} value - 상태 값
* @returns {'published'|'draft'|'members'|'private'} 게시물 상태
*/
const normalizePostStatus = (value) => {
const status = String(value || '').trim()
if (['published', 'draft', 'members', 'private'].includes(status)) {
return status
}
return 'draft'
}
/**
* frontmatter 이미지 값을 Import 후 URL로 정리한다.
* @param {unknown} value - frontmatter 이미지 값
* @param {Map<string, string>} replacements - 자산 교체 맵
* @returns {string|null} 저장할 이미지 URL
*/
const resolveImportedImageUrl = (value, replacements) => {
if (typeof value !== 'string' || !value.trim()) {
return null
}
const trimmed = value.trim()
const replaced = replacements.get(trimmed)
if (replaced) {
return replaced
}
if (/^(?:\.\/)?(?:images|files)\//.test(trimmed)) {
return null
}
return trimmed
}
/**
* ISO 날짜 문자열을 정리한다.
* @param {unknown} value - 날짜 값
* @returns {string|null} ISO 문자열
*/
const normalizeIsoDate = (value) => {
if (!value) {
return null
}
const date = new Date(String(value))
if (Number.isNaN(date.getTime())) {
return null
}
return date.toISOString()
}
/**
* ZIP 엔트리에서 Markdown 게시물 목록을 만든다.
* @param {Array<{ path: string, data: Buffer }>} entries - ZIP 엔트리
* @returns {Array<Object>} Markdown 문서 목록
*/
const collectMarkdownPosts = (entries) => entries
.filter((entry) => MARKDOWN_EXTENSION_PATTERN.test(entry.path))
.map((entry) => ({
path: entry.path,
postFolder: getPostFolder(entry.path),
...parseMarkdownDocument(entry.data.toString('utf8'))
}))
/**
* Export ZIP을 게시물로 가져온다.
* @param {{ zipBuffer: Buffer, authorId: string }} input - Import 입력
* @returns {Promise<{ importedCount: number, assetCount: number, posts: Array<Object> }>} Import 결과
*/
export const importPostsFromExportZip = async ({ zipBuffer, authorId }) => {
const entries = readZipBufferEntries(zipBuffer)
const entryMap = createZipEntryMap(entries)
const markdownPosts = collectMarkdownPosts(entries)
if (!markdownPosts.length) {
throw new Error('IMPORT_MARKDOWN_NOT_FOUND')
}
if (markdownPosts.length > MAX_IMPORT_POSTS) {
throw new Error('IMPORT_POST_LIMIT_EXCEEDED')
}
const reservedSlugs = new Set()
const importedPosts = []
let importedAssetCount = 0
for (const markdownPost of markdownPosts) {
const { frontmatter, content, postFolder } = markdownPost
const assetPaths = collectImportAssetPaths({ frontmatter, content })
const replacements = await saveImportAssets({
entryMap,
postFolder,
assetPaths
})
importedAssetCount += replacements.size ? new Set(replacements.values()).size : 0
const title = String(frontmatter.title || basename(markdownPost.path).replace(MARKDOWN_EXTENSION_PATTERN, '') || 'Imported Post').trim()
const slug = await createUniquePostSlug(frontmatter.slug || title, reservedSlugs)
const featuredImage = resolveImportedImageUrl(frontmatter.featured_image, replacements)
const ogImage = resolveImportedImageUrl(frontmatter.og_image, replacements)
const status = normalizePostStatus(frontmatter.status)
const publishedAt = status === 'published' || status === 'members'
? normalizeIsoDate(frontmatter.published_at) || new Date().toISOString()
: normalizeIsoDate(frontmatter.published_at)
const post = await createAdminPost({
title,
slug,
content: replaceAssetPaths(content, replacements),
excerpt: String(frontmatter.excerpt || ''),
featuredImage,
isFeatured: false,
seoTitle: String(frontmatter.seo_title || ''),
seoDescription: String(frontmatter.seo_description || ''),
canonicalUrl: String(frontmatter.canonical_url || ''),
noindex: Boolean(frontmatter.noindex),
ogImage,
status,
publishedAt,
tags: Array.isArray(frontmatter.tags) ? frontmatter.tags : []
}, authorId)
importedPosts.push(post)
}
return {
importedCount: importedPosts.length,
assetCount: importedAssetCount,
posts: importedPosts.map((post) => ({
id: post.id,
title: post.title,
slug: post.slug
}))
}
}