#399

frontend/src/lib/postNewQueryState.ts

@@ -0,0 +1,410 @@
+import { resultRepairMode } from '@/lib/postImportRows'
+
+import type {
+    PostImportOrigin,
+    PostImportRepairMode,
+    PostImportRow,
+    PostImportSkipReason,
+} from '@/lib/postImportTypes'
+
+type QueryRowState = {
+  url:                     string
+  title:                   string
+  thumbnail_base:          string
+  original_created_from:   string
+  original_created_before: string
+  duration:                string
+  tags:                    string
+  parent_post_ids:         string
+  manual:                  string[]
+  skip:                    PostImportSkipReason | null
+  existing_post_id:        number | null
+  import_status:           PostImportRow['importStatus'] | null
+  created_post_id:         number | null
+  recoverable:             boolean | null }
+
+type MultiRowMetaState = {
+  v:    1
+  rows: QueryRowState[] }
+
+type FullState = {
+  v:    1
+  rows: Array<QueryRowState & { url: string }> }
+
+type ParsedPostNewState = {
+  rows:       PostImportRow[]
+  source:     string
+  repairMode: PostImportRepairMode
+  shortcut:   boolean }
+
+const BASIC_KEYS = [
+    'url',
+    'title',
+    'thumbnail_base',
+    'original_created_from',
+    'original_created_before',
+    'duration',
+    'tags',
+    'parent_post_ids'] as const
+const STATE_KEYS = [
+    'manual',
+    'skip',
+    'existing_post_id',
+    'import_status',
+    'created_post_id',
+    'recoverable'] as const
+const SINGLE_ROW_KEYS = [...BASIC_KEYS, ...STATE_KEYS] as const
+const MAX_REGULAR_URL_LENGTH = 768
+
+const textEncoder = new TextEncoder ()
+const textDecoder = new TextDecoder ()
+
+const isValidSkipReason = (value: unknown): value is PostImportSkipReason =>
+    value === 'existing' || value === 'manual'
+
+
+const isValidImportStatus = (
+    value: unknown,
+): value is PostImportRow['importStatus'] =>
+    value === 'pending'
+    || value === 'created'
+    || value === 'skipped'
+    || value === 'failed'
+
+
+const parsePositiveInt = (value: string | null): number | null => {
+  if (value == null || value === '')
+    return null
+
+  const parsed = Number.parseInt (value, 10)
+  return Number.isInteger (parsed) && parsed > 0 ? parsed : null
+}
+
+
+const decodeBase64Url = (value: string): string | null => {
+  try
+  {
+    const padded = value.replaceAll ('-', '+').replaceAll ('_', '/')
+    const remainder = padded.length % 4
+    const base64 =
+        remainder === 0
+        ? padded
+        : `${ padded }${ '='.repeat (4 - remainder) }`
+    const binary = window.atob (base64)
+    const bytes = Uint8Array.from (binary, char => char.charCodeAt (0))
+    return textDecoder.decode (bytes)
+  }
+  catch
+  {
+    return null
+  }
+}
+
+
+const encodeBase64Url = (value: string): string => {
+  const bytes = textEncoder.encode (value)
+  const binary = Array.from (bytes, byte => String.fromCharCode (byte)).join ('')
+  return window.btoa (binary).replaceAll ('+', '-').replaceAll ('/', '_').replaceAll ('=', '')
+}
+
+
+const manualFields = (row: PostImportRow): string[] =>
+    ['title',
+     'thumbnailBase',
+     'originalCreatedFrom',
+     'originalCreatedBefore',
+     'tags',
+     'parentPostIds']
+    .filter (field => row.provenance[field] === 'manual')
+
+
+const baseRowState = (row: PostImportRow): QueryRowState => ({
+    url: row.url,
+    title: String (row.attributes.title ?? ''),
+    thumbnail_base: String (row.attributes.thumbnailBase ?? ''),
+    original_created_from: String (row.attributes.originalCreatedFrom ?? ''),
+    original_created_before: String (row.attributes.originalCreatedBefore ?? ''),
+    duration: String (row.attributes.duration ?? ''),
+    tags: String (row.attributes.tags ?? ''),
+    parent_post_ids: String (row.attributes.parentPostIds ?? ''),
+    manual: manualFields (row),
+    skip: row.skipReason ?? null,
+    existing_post_id: row.existingPostId ?? null,
+    import_status: row.importStatus ?? null,
+    created_post_id: row.createdPostId ?? null,
+    recoverable: row.recoverable ?? null })
+
+
+const buildRow = (
+    sourceRow: number,
+    state:     QueryRowState & { url: string },
+): PostImportRow => {
+  const manual = new Set (state.manual)
+  const provenance: Record<string, PostImportOrigin> = {
+      url: 'manual',
+      title: manual.has ('title') ? 'manual' : 'automatic',
+      thumbnailBase: manual.has ('thumbnailBase') ? 'manual' : 'automatic',
+      originalCreatedFrom: manual.has ('originalCreatedFrom') ? 'manual' : 'automatic',
+      originalCreatedBefore: manual.has ('originalCreatedBefore') ? 'manual' : 'automatic',
+      duration: 'automatic',
+      tags: manual.has ('tags') ? 'manual' : 'automatic',
+      parentPostIds: manual.has ('parentPostIds') ? 'manual' : 'automatic' }
+  const tagSources = {
+      automatic: '',
+      manual: provenance.tags === 'manual' ? state.tags : '' }
+  const attributes = {
+      title: state.title,
+      thumbnailBase: state.thumbnail_base,
+      originalCreatedFrom: state.original_created_from,
+      originalCreatedBefore: state.original_created_before,
+      duration: state.duration,
+      tags: state.tags,
+      parentPostIds: state.parent_post_ids }
+
+  return {
+      sourceRow,
+      url: state.url,
+      attributes,
+      fieldWarnings: { },
+      baseWarnings: [],
+      validationErrors: { },
+      provenance,
+      tagSources,
+      status: 'ready',
+      skipReason: state.skip ?? undefined,
+      existingPostId: state.existing_post_id ?? undefined,
+      resetSnapshot: {
+	  url: state.url,
+	  attributes: { ...attributes },
+	  provenance: { ...provenance },
+	  tagSources: { ...tagSources },
+	  fieldWarnings: { },
+	  baseWarnings: [] },
+      createdPostId: state.created_post_id ?? undefined,
+      importStatus: state.import_status ?? undefined,
+      recoverable: state.recoverable ?? undefined }
+}
+
+
+const parseManual = (value: string | null): string[] =>
+    (value ?? '')
+    .split (',')
+    .map (entry => entry.trim ())
+    .filter (entry => entry !== '')
+
+
+const parseSingleRow = (params: URLSearchParams): ParsedPostNewState | null => {
+  const url = params.get ('url')
+  if (url == null || url === '')
+    return null
+
+  const hasOnlyUrl = Array.from (params.keys ()).every (key => key === 'url')
+  if (hasOnlyUrl)
+    {
+      const row = buildRow (1, {
+	  url,
+	  title: '',
+	  thumbnail_base: '',
+	  original_created_from: '',
+	  original_created_before: '',
+	  duration: '',
+	  tags: '',
+	  parent_post_ids: '',
+	  manual: [],
+	  skip: null,
+	  existing_post_id: null,
+	  import_status: null,
+	  created_post_id: null,
+	  recoverable: null })
+      return {
+	  rows: [row],
+	  source: url,
+	  repairMode: 'all',
+	  shortcut: true }
+    }
+
+  if (!(Array.from (params.keys ()).every (key => SINGLE_ROW_KEYS.includes (key as never))))
+    return null
+
+  const skipValue = params.get ('skip')
+  const importStatus = params.get ('import_status')
+  const recoverable =
+      params.get ('recoverable') == null
+      ? null
+      : params.get ('recoverable') === 'true'
+  if (skipValue != null && !(isValidSkipReason (skipValue)))
+    return null
+  if (importStatus != null && !(isValidImportStatus (importStatus)))
+    return null
+
+  const row = buildRow (1, {
+      url,
+      title: params.get ('title') ?? '',
+      thumbnail_base: params.get ('thumbnail_base') ?? '',
+      original_created_from: params.get ('original_created_from') ?? '',
+      original_created_before: params.get ('original_created_before') ?? '',
+      duration: params.get ('duration') ?? '',
+      tags: params.get ('tags') ?? '',
+      parent_post_ids: params.get ('parent_post_ids') ?? '',
+      manual: parseManual (params.get ('manual')),
+      skip: skipValue,
+      existing_post_id: parsePositiveInt (params.get ('existing_post_id')),
+      import_status: importStatus,
+      created_post_id: parsePositiveInt (params.get ('created_post_id')),
+      recoverable })
+
+  return {
+      rows: [row],
+      source: url,
+      repairMode: resultRepairMode ([row]),
+      shortcut: false }
+}
+
+
+const parseMetaRows = (
+    urlsValue: string,
+    metaValue: string,
+): ParsedPostNewState | null => {
+  const decoded = decodeBase64Url (metaValue)
+  if (decoded == null)
+    return null
+
+  let parsed: MultiRowMetaState
+  try
+  {
+    parsed = JSON.parse (decoded) as MultiRowMetaState
+  }
+  catch
+  {
+    return null
+  }
+
+  if (parsed.v !== 1 || !(Array.isArray (parsed.rows)))
+    return null
+
+  const urls = urlsValue.split (' ').filter (url => url !== '')
+  if (urls.length < 2 || urls.length !== parsed.rows.length)
+    return null
+
+  const rows = parsed.rows.map ((row, index) => buildRow (index + 1, {
+      ...row,
+      url: urls[index] ?? '' }))
+
+  return {
+      rows,
+      source: urls.join ('\n'),
+      repairMode: resultRepairMode (rows),
+      shortcut: false }
+}
+
+
+const parseFullState = (stateValue: string): ParsedPostNewState | null => {
+  const decoded = decodeBase64Url (stateValue)
+  if (decoded == null)
+    return null
+
+  let parsed: FullState
+  try
+  {
+    parsed = JSON.parse (decoded) as FullState
+  }
+  catch
+  {
+    return null
+  }
+
+  if (parsed.v !== 1 || !(Array.isArray (parsed.rows)))
+    return null
+
+  const rows = parsed.rows.map ((row, index) => buildRow (index + 1, row))
+  return {
+      rows,
+      source: rows.map (row => row.url).join ('\n'),
+      repairMode: resultRepairMode (rows),
+      shortcut: false }
+}
+
+
+const regularQuery = (rows: PostImportRow[]): URLSearchParams => {
+  if (rows.length === 1)
+    {
+      const row = rows[0]
+      const params = new URLSearchParams ()
+      params.set ('url', row?.url ?? '')
+      params.set ('title', String (row?.attributes.title ?? ''))
+      params.set ('thumbnail_base', String (row?.attributes.thumbnailBase ?? ''))
+      params.set ('original_created_from', String (row?.attributes.originalCreatedFrom ?? ''))
+      params.set ('original_created_before', String (row?.attributes.originalCreatedBefore ?? ''))
+      params.set ('duration', String (row?.attributes.duration ?? ''))
+      params.set ('tags', String (row?.attributes.tags ?? ''))
+      params.set ('parent_post_ids', String (row?.attributes.parentPostIds ?? ''))
+      const manual = manualFields (row).join (',')
+      if (manual !== '')
+	params.set ('manual', manual)
+      if (row.skipReason != null)
+	params.set ('skip', row.skipReason)
+      if (row.existingPostId != null)
+	params.set ('existing_post_id', String (row.existingPostId))
+      if (row.importStatus != null)
+	params.set ('import_status', row.importStatus)
+      if (row.createdPostId != null)
+	params.set ('created_post_id', String (row.createdPostId))
+      if (row.recoverable != null)
+	params.set ('recoverable', row.recoverable ? 'true' : 'false')
+      return params
+    }
+
+  const urls = rows.map (row => row.url).join (' ')
+  const metaRows: QueryRowState[] = rows.map (row => baseRowState (row))
+  const meta = encodeBase64Url (JSON.stringify ({
+      v: 1,
+      rows: metaRows }))
+  const params = new URLSearchParams ()
+  params.set ('urls', urls)
+  params.set ('meta', meta)
+  return params
+}
+
+
+export const hasPostNewReviewState = (search: string): boolean => {
+  const params = new URLSearchParams (search)
+  return params.has ('state')
+      || params.has ('urls')
+      || params.has ('meta')
+      || params.has ('url')
+}
+
+
+export const parsePostNewState = (search: string): ParsedPostNewState | null => {
+  const params = new URLSearchParams (search)
+  const stateValue = params.get ('state')
+  if (stateValue != null)
+    return parseFullState (stateValue)
+
+  const urlsValue = params.get ('urls')
+  const metaValue = params.get ('meta')
+  if (urlsValue != null || metaValue != null)
+    {
+      if (urlsValue == null || metaValue == null)
+	return null
+      return parseMetaRows (urlsValue, metaValue)
+    }
+
+  return parseSingleRow (params)
+}
+
+
+export const serialisePostNewState = (rows: PostImportRow[]): string => {
+  const regular = regularQuery (rows).toString ()
+  const regularPath = `/posts/new?${ regular }`
+  if (regularPath.length < MAX_REGULAR_URL_LENGTH)
+    return `/posts/new?${ regular }`
+
+  const stateRows: FullState['rows'] = rows.map (row => ({
+      ...baseRowState (row),
+      url: row.url }))
+  const state = encodeBase64Url (JSON.stringify ({
+      v: 1,
+      rows: stateRows }))
+  return `/posts/new?state=${ state }`
+}