@sanity/export 3.38.1 → 3.39.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sanity/export",
-  "version": "3.38.1",
+  "version": "3.39.0",
   "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

@@ -20,3 +20,16 @@ exports.ASSET_DOWNLOAD_MAX_RETRIES = 10
  * @internal
  */
 exports.ASSET_DOWNLOAD_CONCURRENCY = 8
+
+/**
+ * How frequently we will `debug` log while streaming the documents.
+ * @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

package/src/export.js

@@ -17,6 +17,7 @@ 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 noop = () => null
 
@@ -93,11 +94,14 @@ async function exportDataset(opts) {
   onProgress({step: 'Exporting documents...'})
 
   let documentCount = 0
+  let lastDocumentID = null
   let lastReported = Date.now()
-  const reportDocumentCount = (chunk, enc, cb) => {
+  const reportDocumentCount = (doc, enc, cb) => {
     ++documentCount
 
     const now = Date.now()
+    // We report to the `onProgress` handler every 50 ms.
+    // It's up to the caller to not do too much expensive work.
     if (now - lastReported > 50) {
       onProgress({
         step: 'Exporting documents...',
@@ -109,13 +113,30 @@ async function exportDataset(opts) {
       lastReported = now
     }
 
-    cb(null, chunk)
+    lastDocumentID = doc._id
+
+    cb(null, doc)
   }
 
   const inputStream = await getDocumentsStream(options)
   debug('Got HTTP %d', inputStream.statusCode)
   debug('Response headers: %o', inputStream.headers)
 
+  let debugTimer = null
+  function scheduleDebugTimer() {
+    debugTimer = setTimeout(() => {
+      debug('Still streaming documents', {
+        documentCount,
+        lastDocumentID,
+      })
+
+      // Schedule another tick:
+      scheduleDebugTimer()
+    }, DOCUMENT_STREAM_DEBUG_INTERVAL)
+  }
+
+  scheduleDebugTimer()
+
   const jsonStream = miss.pipeline(
     inputStream,
     logFirstChunk(),
@@ -125,13 +146,15 @@ async function exportDataset(opts) {
     assetStreamHandler,
     filterDocumentTypes(options.types),
     options.drafts ? miss.through.obj() : filterDrafts(),
+    miss.through.obj(reportDocumentCount),
     stringifyStream(),
-    miss.through(reportDocumentCount),
   )
 
   miss.pipe(jsonStream, fs.createWriteStream(dataPath), async (err) => {
+    if (debugTimer !== null) clearTimeout(debugTimer)
+
     if (err) {
-      debug('Export stream error: ', err)
+      debug(`Export stream error @ ${lastDocumentID}/${documentCount}: `, err)
       reject(err)
       return
     }

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/rejectOnApiError.js

@@ -2,18 +2,26 @@ const miss = require('mississippi')
 
 module.exports = () =>
   miss.through.obj((doc, enc, callback) => {
-    if (doc.error && doc.statusCode) {
-      callback(
-        new Error(
-          ['Export', `HTTP ${doc.statusCode}`, doc.error, doc.message]
-            .filter((part) => typeof part === 'string')
-            .join(': '),
-        ),
-      )
+    // check if the document passed contains a document attribtue first, and return early.
+    if (doc._id) {
+      callback(null, doc)
       return
     }
 
-    if (!doc._id && doc.error) {
+    if (doc.error) {
+      // if we got a statusCode we can decorate the error with it
+      if (doc.statusCode) {
+        callback(
+          new Error(
+            ['Export', `HTTP ${doc.statusCode}`, doc.error, doc.message]
+              .filter((part) => typeof part === 'string')
+              .join(': '),
+          ),
+        )
+        return
+      }
+
+      // no statusCode, just serialize and return the error
       callback(new Error(doc.error.description || doc.error.message || JSON.stringify(doc)))
       return
     }

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

@@ -1,9 +1,13 @@
 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,
+} = 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,
@@ -11,6 +15,7 @@ const exportDefaults = {
   raw: false,
   maxRetries: DOCUMENT_STREAM_MAX_RETRIES,
   maxAssetRetries: ASSET_DOWNLOAD_MAX_RETRIES,
+  readTimeout: REQUEST_READ_TIMEOUT,
 }
 
 function validateOptions(opts) {