@sanity/export 3.39.0 → 3.41.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sanity/export",
-  "version": "3.39.0",
+  "version": "3.41.0",
   "description": "Export Sanity documents and assets",
   "keywords": [
     "sanity",

package/src/constants.js

@@ -33,3 +33,11 @@ exports.DOCUMENT_STREAM_DEBUG_INTERVAL = 10000
  * @internal
  */
 exports.REQUEST_READ_TIMEOUT = 3 * 60 * 1000 // 3 minutes
+
+/**
+  What mode to use when exporting documents.
+  stream: Export all documents in the dataset in one request, this will be consistent but might be slow on large datasets.
+  cursor: Export documents using a cursor, this might lead to inconsistent results if a mutation is performed while exporting.
+*/
+exports.MODE_STREAM = 'stream'
+exports.MODE_CURSOR = 'cursor'

package/src/export.js

@@ -11,13 +11,14 @@ const filterDocumentTypes = require('./filterDocumentTypes')
 const filterDrafts = require('./filterDrafts')
 const filterSystemDocuments = require('./filterSystemDocuments')
 const getDocumentsStream = require('./getDocumentsStream')
+const getDocumentCursorStream = require('./getDocumentCursorStream')
 const logFirstChunk = require('./logFirstChunk')
 const rejectOnApiError = require('./rejectOnApiError')
 const stringifyStream = require('./stringifyStream')
 const tryParseJson = require('./tryParseJson')
 const rimraf = require('./util/rimraf')
 const validateOptions = require('./validateOptions')
-const {DOCUMENT_STREAM_DEBUG_INTERVAL} = require('./constants')
+const {DOCUMENT_STREAM_DEBUG_INTERVAL, MODE_CURSOR, MODE_STREAM} = require('./constants')
 
 const noop = () => null
 
@@ -90,7 +91,7 @@ async function exportDataset(opts) {
     debug('Archive finished')
   })
 
-  debug('Getting dataset export stream')
+  debug('Getting dataset export stream, mode: "%s"', options.mode)
   onProgress({step: 'Exporting documents...'})
 
   let documentCount = 0
@@ -118,9 +119,13 @@ async function exportDataset(opts) {
     cb(null, doc)
   }
 
-  const inputStream = await getDocumentsStream(options)
-  debug('Got HTTP %d', inputStream.statusCode)
-  debug('Response headers: %o', inputStream.headers)
+  const inputStream = await getDocumentInputStream(options)
+  if (inputStream.statusCode) {
+    debug('Got HTTP %d', inputStream.statusCode)
+  }
+  if (inputStream.headers) {
+    debug('Response headers: %o', inputStream.headers)
+  }
 
   let debugTimer = null
   function scheduleDebugTimer() {
@@ -250,6 +255,17 @@ async function exportDataset(opts) {
   return result
 }
 
+function getDocumentInputStream(options) {
+  if (options.mode === MODE_STREAM) {
+    return getDocumentsStream(options)
+  }
+  if (options.mode === MODE_CURSOR) {
+    return getDocumentCursorStream(options)
+  }
+
+  throw new Error(`Invalid mode: ${options.mode}`)
+}
+
 function isWritableStream(val) {
   return (
     val !== null &&

package/src/filterSystemDocuments.js

@@ -2,6 +2,7 @@ const miss = require('mississippi')
 const debug = require('./debug')
 
 const isSystemDocument = (doc) => doc && doc._id && doc._id.indexOf('_.') === 0
+const isCursor = (doc) => doc && !doc._id && doc.nextCursor !== undefined
 
 module.exports = () =>
   miss.through.obj((doc, enc, callback) => {
@@ -9,6 +10,10 @@ module.exports = () =>
       debug('%s is a system document, skipping', doc && doc._id)
       return callback()
     }
+    if (isCursor(doc)) {
+      debug('%o is a cursor, skipping', doc)
+      return callback()
+    }
 
     return callback(null, doc)
   })

package/src/getDocumentCursorStream.js

@@ -0,0 +1,85 @@
+const {Transform} = require('node:stream')
+
+const pkg = require('../package.json')
+const debug = require('./debug')
+const requestStream = require('./requestStream')
+
+// same regex as split2 is using by default: https://github.com/mcollina/split2/blob/53432f54bd5bf422bd55d91d38f898b6c9496fc1/index.js#L86
+const splitRegex = /\r?\n/
+
+module.exports = async (options) => {
+  let streamsInflight = 0
+  function decrementInflight(stream) {
+    streamsInflight--
+    if (streamsInflight === 0) {
+      stream.end()
+    }
+  }
+
+  const stream = new Transform({
+    async transform(chunk, encoding, callback) {
+      if (encoding !== 'buffer' && encoding !== 'string') {
+        callback(null, chunk)
+        return
+      }
+      this.push(chunk, encoding)
+
+      let parsedChunk = null
+      for (const chunkStr of chunk.toString().split(splitRegex)) {
+        if (chunkStr.trim() === '') {
+          continue
+        }
+
+        try {
+          parsedChunk = JSON.parse(chunkStr)
+        } catch (err) {
+          // Ignore JSON parse errors
+          // this can happen if the chunk is not a JSON object. We just pass it through and let the caller handle it.
+          debug('Failed to parse JSON chunk, ignoring', err, chunkStr)
+        }
+
+        if (
+          parsedChunk !== null &&
+          typeof parsedChunk === 'object' &&
+          'nextCursor' in parsedChunk &&
+          typeof parsedChunk.nextCursor === 'string' &&
+          !('_id' in parsedChunk)
+        ) {
+          debug('Got next cursor "%s", fetching next stream', parsedChunk.nextCursor)
+          streamsInflight++
+
+          const reqStream = await startStream(options, parsedChunk.nextCursor)
+          reqStream.on('end', () => decrementInflight(this))
+          reqStream.pipe(this, {end: false})
+        }
+      }
+
+      callback()
+    },
+  })
+
+  streamsInflight++
+  const reqStream = await startStream(options, '')
+  reqStream.on('end', () => decrementInflight(stream))
+  reqStream.pipe(stream, {end: false})
+  return stream
+}
+
+function startStream(options, nextCursor) {
+  const url = options.client.getUrl(
+    `/data/export/${options.dataset}?nextCursor=${encodeURIComponent(nextCursor)}`,
+  )
+  const token = options.client.config().token
+  const headers = {
+    'User-Agent': `${pkg.name}@${pkg.version}`,
+    ...(token ? {Authorization: `Bearer ${token}`} : {}),
+  }
+
+  debug('Starting stream with cursor "%s"', nextCursor)
+
+  return requestStream({url, headers, maxRetries: options.maxRetries}).then((res) => {
+    debug('Got stream with HTTP %d', res.statusCode)
+
+    return res
+  })
+}

package/src/validateOptions.js

@@ -3,6 +3,8 @@ const {
   DOCUMENT_STREAM_MAX_RETRIES,
   ASSET_DOWNLOAD_MAX_RETRIES,
   REQUEST_READ_TIMEOUT,
+  MODE_STREAM,
+  MODE_CURSOR,
 } = require('./constants')
 
 const clientMethods = ['getUrl', 'config']
@@ -13,6 +15,7 @@ const exportDefaults = {
   drafts: true,
   assets: true,
   raw: false,
+  mode: MODE_STREAM,
   maxRetries: DOCUMENT_STREAM_MAX_RETRIES,
   maxAssetRetries: ASSET_DOWNLOAD_MAX_RETRIES,
   readTimeout: REQUEST_READ_TIMEOUT,
@@ -25,6 +28,15 @@ function validateOptions(opts) {
     throw new Error(`options.dataset must be a valid dataset name`)
   }
 
+  if (
+    typeof options.mode !== 'string' ||
+    (options.mode !== MODE_STREAM && options.mode !== MODE_CURSOR)
+  ) {
+    throw new Error(
+      `options.mode must be either "${MODE_STREAM}" or "${MODE_CURSOR}", got "${options.mode}"`,
+    )
+  }
+
   if (options.onProgress && typeof options.onProgress !== 'function') {
     throw new Error(`options.onProgress must be a function`)
   }