@sanity/export 3.38.2 → 3.40.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sanity/export",
-  "version": "3.38.2",
+  "version": "3.40.0",
   "description": "Export Sanity documents and assets",
   "keywords": [
     "sanity",
@@ -34,7 +34,7 @@
     "@sanity/util": "3.37.2",
     "archiver": "^7.0.0",
     "debug": "^4.3.4",
-    "get-it": "^8.4.21",
+    "get-it": "^8.6.2",
     "lodash": "^4.17.21",
     "mississippi": "^4.0.0",
     "p-queue": "^2.3.0",

package/src/constants.js

@@ -26,3 +26,18 @@ exports.ASSET_DOWNLOAD_CONCURRENCY = 8
  * @internal
  */
 exports.DOCUMENT_STREAM_DEBUG_INTERVAL = 10000
+
+/**
+ * How long to wait before timing out the read of a request due to inactivity.
+ * User overridable as `options.readTimeout`.
+ * @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
 
@@ -118,7 +119,7 @@ async function exportDataset(opts) {
     cb(null, doc)
   }
 
-  const inputStream = await getDocumentsStream(options)
+  const inputStream = await getDocumentInputStream(options)
   debug('Got HTTP %d', inputStream.statusCode)
   debug('Response headers: %o', inputStream.headers)
 
@@ -250,6 +251,17 @@ 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/getDocumentCursorStream.js

@@ -0,0 +1,73 @@
+const {Transform} = require('node:stream')
+
+const pkg = require('../package.json')
+const requestStream = require('./requestStream')
+
+module.exports = async (options) => {
+  let streamsInflight = 0
+  const stream = new Transform({
+    async transform(chunk, encoding, callback) {
+      if (encoding !== 'buffer' && encoding !== 'string') {
+        callback(null, chunk)
+        return
+      }
+
+      let parsedChunk = null
+      try {
+        parsedChunk = JSON.parse(chunk.toString())
+      } 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.
+      }
+
+      if (
+        parsedChunk !== null &&
+        typeof parsedChunk === 'object' &&
+        'nextCursor' in parsedChunk &&
+        typeof parsedChunk.nextCursor === 'string' &&
+        !('_id' in parsedChunk)
+      ) {
+        streamsInflight++
+
+        const reqStream = await startStream(options, parsedChunk.nextCursor)
+        reqStream.on('end', () => {
+          streamsInflight--
+          if (streamsInflight === 0) {
+            stream.end()
+          }
+        })
+        reqStream.pipe(this, {end: false})
+
+        callback()
+        return
+      }
+
+      callback(null, chunk)
+    },
+  })
+
+  streamsInflight++
+  const reqStream = await startStream(options, '')
+  reqStream.on('end', () => {
+    streamsInflight--
+    if (streamsInflight === 0) {
+      stream.end()
+    }
+  })
+
+  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}`} : {}),
+  }
+
+  return requestStream({url, headers, maxRetries: options.maxRetries})
+}

package/src/getDocumentsStream.js

@@ -11,5 +11,10 @@ module.exports = (options) => {
     ...(token ? {Authorization: `Bearer ${token}`} : {}),
   }
 
-  return requestStream({url, headers, maxRetries: options.maxRetries})
+  return requestStream({
+    url,
+    headers,
+    maxRetries: options.maxRetries,
+    readTimeout: options.readTimeout,
+  })
 }

package/src/requestStream.js

@@ -2,13 +2,11 @@ const {getIt} = require('get-it')
 const {keepAlive, promise} = require('get-it/middleware')
 const debug = require('./debug')
 const {extractFirstError} = require('./util/extractFirstError')
-const {DOCUMENT_STREAM_MAX_RETRIES} = require('./constants')
+const {DOCUMENT_STREAM_MAX_RETRIES, REQUEST_READ_TIMEOUT} = require('./constants')
 
 const request = getIt([keepAlive(), promise({onlyBody: true})])
-const socketsWithTimeout = new WeakSet()
 
 const CONNECTION_TIMEOUT = 15 * 1000 // 15 seconds
-const READ_TIMEOUT = 3 * 60 * 1000 // 3 minutes
 const RETRY_DELAY_MS = 1500 // 1.5 seconds
 
 function delay(ms) {
@@ -20,30 +18,18 @@ module.exports = async (options) => {
   const maxRetries =
     typeof options.maxRetries === 'number' ? options.maxRetries : DOCUMENT_STREAM_MAX_RETRIES
 
+  const readTimeout =
+    typeof options.readTimeout === 'number' ? options.readTimeout : REQUEST_READ_TIMEOUT
+
   let error
   for (let i = 0; i < maxRetries; i++) {
     try {
-      const response = await request({
+      return await request({
         ...options,
         stream: true,
         maxRedirects: 0,
-        timeout: {connect: CONNECTION_TIMEOUT, socket: READ_TIMEOUT},
+        timeout: {connect: CONNECTION_TIMEOUT, socket: readTimeout},
       })
-
-      if (
-        response.connection &&
-        typeof response.connection.setTimeout === 'function' &&
-        !socketsWithTimeout.has(response.connection)
-      ) {
-        socketsWithTimeout.add(response.connection)
-        response.connection.setTimeout(READ_TIMEOUT, () => {
-          response.destroy(
-            new Error(`Export: Read timeout: No data received on socket for ${READ_TIMEOUT} ms`),
-          )
-        })
-      }
-
-      return response
     } catch (err) {
       error = extractFirstError(err)
 

package/src/validateOptions.js

@@ -1,16 +1,24 @@
 const defaults = require('lodash/defaults')
-const {DOCUMENT_STREAM_MAX_RETRIES, ASSET_DOWNLOAD_MAX_RETRIES} = require('./constants')
+const {
+  DOCUMENT_STREAM_MAX_RETRIES,
+  ASSET_DOWNLOAD_MAX_RETRIES,
+  REQUEST_READ_TIMEOUT,
+  MODE_STREAM,
+  MODE_CURSOR,
+} = require('./constants')
 
 const clientMethods = ['getUrl', 'config']
 const booleanFlags = ['assets', 'raw', 'compress', 'drafts']
-const numberFlags = ['maxAssetRetries', 'maxRetries', 'assetConcurrency']
+const numberFlags = ['maxAssetRetries', 'maxRetries', 'assetConcurrency', 'readTimeout']
 const exportDefaults = {
   compress: true,
   drafts: true,
   assets: true,
   raw: false,
+  mode: MODE_STREAM,
   maxRetries: DOCUMENT_STREAM_MAX_RETRIES,
   maxAssetRetries: ASSET_DOWNLOAD_MAX_RETRIES,
+  readTimeout: REQUEST_READ_TIMEOUT,
 }
 
 function validateOptions(opts) {
@@ -20,6 +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`)
   }