fix(gatsby-source-contentful): reduce memory usage (#37910)

* add touch nodes optout action * cleanup * start adding backreference cache * revert formatting * remove unused variable * remove unused var * switch from opting out types from stale nodes to opting out entire plugin * add no nodes warning * fix bug where double bound actions are ignored when there are no args * update progress mock * get owner from plugin * rename type * rename types * use redux types instead of pulling from lmdb * remove unused line * Update source-nodes.ts * use CREATE_NODE action instead of adding a new type owner action * add typeowners test * test touchNodes and enableStatefulSourceNodes() * fix contentful tests * snapshot updates * chore(changelogs): update changelogs (#37808) * fix(deps): update starters and examples - gatsby to ^5.8.1 (#37806) Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com> * fix(gatsby): Validate sub plugins options (#37804) * fix(gatsby): validate sub plugins options of gatsby-plugin-mdx * revert hardcoded path to subplugins * validate subplugins if they are not under 'options.plugins' field --------- Co-authored-by: Michal Piechowiak <[email protected]> * fix(create-gatsby): Use correct name in summary message (#37809) * actual fix * other misc changes * chore(release): Publish next - [email protected] - [email protected] - [email protected] - [email protected] - [email protected] * remove testing timeout * minimal docs * reword * add comment * reportOnce instead of throwing an error * consolidate typeOwners * use new typesToPlugins Map keys instead of pluginsToTypes Map values * consolidate remaining typeOwners object checks into new typeOwners reducer * fix missing owner error * fix type errors and incorrect plugin object reference * maybe fix unit tests * make SitePage nodes owned by internal-data-bridge * add missing fields on some TOUCH_NODE actions * skip owner checks when deleting child nodes * sp * Update yarn.lock * reduce memory usage by storing minimal backreference info and not blocking the event loop * remove coment * remove extra logs * add helper to explain what the setImmediate promise is for * remove unneeded entryNode condition * refactor createAssetNodes to use async/await * add cached node counts by assets vs content * refactor to fix linting failure * fix tests * ensure mutated existing node data is applied during updates * restore previous logging behaviour * fix test state * always reset memory node count when existingNodes is empty * revert more logging * update test to use new createAssetNodes changes * store count instead of huge set of id's to use less memory * remove debugger * make sure sys.type exists before checking * move enableStatefulSourceNodes call to main sourceNodes fn * use util * fix e2e test - ContentfulTag node doesn't have sys.type * always exclude ContentfulTag nodes from memory Map cache * fix linting (consistent return) --------- Co-authored-by: GatsbyJS Bot <[email protected]> Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com> Co-authored-by: Taozc <[email protected]> Co-authored-by: Michal Piechowiak <[email protected]> Co-authored-by: Lennart <[email protected]>
gatsbyjs · Apr 18, 2023 · 311b202 · 311b202
1 parent c1f0bbf
commit 311b202
Show file tree

Hide file tree

Showing 6 changed files with 667 additions and 408 deletions.
diff --git a/packages/gatsby-source-contentful/src/__tests__/download-contentful-assets.js b/packages/gatsby-source-contentful/src/__tests__/download-contentful-assets.js
@@ -66,17 +66,15 @@ describe(`downloadContentfulAssets`, () => {
     const assetNodes = []
     for (const assetItem of fixtures) {
       assetNodes.push(
-        ...(await Promise.all(
-          createAssetNodes({
-            assetItem,
-            createNode,
-            createNodeId,
-            defaultLocale,
-            locales,
-            space,
-            pluginConfig,
-          })
-        ))
+        ...(await createAssetNodes({
+          assetItem,
+          createNode,
+          createNodeId,
+          defaultLocale,
+          locales,
+          space,
+          pluginConfig,
+        }))
       )
     }
 

diff --git a/packages/gatsby-source-contentful/src/__tests__/gatsby-node.js b/packages/gatsby-source-contentful/src/__tests__/gatsby-node.js
@@ -6,6 +6,7 @@ import {
   sourceNodes,
   onPreInit,
 } from "../gatsby-node"
+import { existingNodes, is, memoryNodeCounts } from "../backreferences"
 import { fetchContent, fetchContentTypes } from "../fetch"
 import { makeId } from "../normalize"
 
@@ -59,7 +60,12 @@ describe(`gatsby-node`, () => {
 
   const actions = {
     createTypes: jest.fn(),
-    setPluginStatus: jest.fn(),
+    setPluginStatus: jest.fn(pluginStatusObject => {
+      pluginStatus = {
+        ...pluginStatus,
+        ...pluginStatusObject,
+      }
+    }),
     createNode: jest.fn(async node => {
       // similar checks as gatsby does
       if (!_.isPlainObject(node)) {
@@ -99,9 +105,20 @@ describe(`gatsby-node`, () => {
     }),
     buildInterfaceType: jest.fn(),
   }
+  let pluginStatus = {}
+  const resetPluginStatus = () => {
+    pluginStatus = {}
+  }
   const store = {
     getState: jest.fn(() => {
-      return { program: { directory: process.cwd() }, status: {} }
+      return {
+        program: { directory: process.cwd() },
+        status: {
+          plugins: {
+            [`gatsby-source-contentful`]: pluginStatus,
+          },
+        },
+      }
     }),
   }
   const cache = createMockCache()
@@ -404,7 +421,11 @@ describe(`gatsby-node`, () => {
     })
   }
 
-  beforeEach(() => {
+  beforeEach(async () => {
+    existingNodes.clear()
+    is.firstSourceNodesCallOfCurrentNodeProcess = true
+    resetPluginStatus()
+
     // @ts-ignore
     fetchContent.mockClear()
     // @ts-ignore

diff --git a/packages/gatsby-source-contentful/src/backreferences.js b/packages/gatsby-source-contentful/src/backreferences.js
@@ -0,0 +1,139 @@
+// @ts-check
+import { hasFeature } from "gatsby-plugin-utils/index"
+import { getDataStore } from "gatsby/dist/datastore"
+import { untilNextEventLoopTick } from "./utils"
+
+// Array of all existing Contentful nodes. Make it global and incrementally update it because it's hella slow to recreate this on every data update for large sites.
+export const existingNodes = new Map()
+
+let allNodesLoopCount = 0
+
+// "is" === object so it can be overridden by tests
+export const is = {
+  firstSourceNodesCallOfCurrentNodeProcess: true,
+}
+
+export async function getExistingCachedNodes({
+  actions,
+  getNode,
+  pluginConfig,
+}) {
+  const { touchNode } = actions
+
+  const needToTouchNodes =
+    !hasFeature(`stateful-source-nodes`) &&
+    is.firstSourceNodesCallOfCurrentNodeProcess
+
+  if (existingNodes.size === 0) {
+    memoryNodeCountsBySysType.Asset = 0
+    memoryNodeCountsBySysType.Entry = 0
+
+    const dataStore = getDataStore()
+    const allNodeTypeNames = Array.from(dataStore.getTypes())
+
+    for (const typeName of allNodeTypeNames) {
+      const typeNodes = dataStore.iterateNodesByType(typeName)
+
+      const firstNodeOfType = Array.from(typeNodes.slice(0, 1))[0]
+
+      if (
+        !firstNodeOfType ||
+        firstNodeOfType.internal.owner !== `gatsby-source-contentful`
+      ) {
+        continue
+      }
+
+      for (const node of typeNodes) {
+        if (needToTouchNodes) {
+          touchNode(node)
+
+          if (node?.fields?.includes(`localFile`)) {
+            // Prevent GraphQL type inference from crashing on this property
+            const fullNode = getNode(node.id)
+            const localFileNode = getNode(fullNode.fields.localFile)
+            touchNode(localFileNode)
+          }
+        }
+
+        if (++allNodesLoopCount % 5000 === 0) {
+          // dont block the event loop
+          await untilNextEventLoopTick()
+        }
+
+        addNodeToExistingNodesCache(node)
+      }
+
+      // dont block the event loop
+      await untilNextEventLoopTick()
+    }
+  }
+
+  is.firstSourceNodesCallOfCurrentNodeProcess = false
+
+  return {
+    existingNodes,
+    memoryNodeCountsBySysType,
+  }
+}
+
+const memoryNodeCountsBySysType = {
+  Asset: 0,
+  Entry: 0,
+}
+
+// store only the fields we need to compare to reduce memory usage. if a node is updated we'll use getNode to grab the whole node before updating it
+export function addNodeToExistingNodesCache(node) {
+  if (node.internal.type === `ContentfulTag`) {
+    return
+  }
+
+  if (
+    node.sys.type in memoryNodeCountsBySysType &&
+    !existingNodes.has(node.id)
+  ) {
+    memoryNodeCountsBySysType[node.sys.type] ||= 0
+    memoryNodeCountsBySysType[node.sys.type]++
+  }
+
+  const cacheNode = {
+    id: node.id,
+    contentful_id: node.contentful_id,
+    sys: {
+      type: node.sys.type,
+    },
+    node_locale: node.node_locale,
+    children: node.children,
+    internal: {
+      owner: node.internal.owner,
+    },
+    __memcache: true,
+  }
+
+  for (const key of Object.keys(node)) {
+    if (key.endsWith(`___NODE`)) {
+      cacheNode[key] = node[key]
+    }
+  }
+
+  existingNodes.set(node.id, cacheNode)
+}
+
+export function removeNodeFromExistingNodesCache(node) {
+  if (node.internal.type === `ContentfulTag`) {
+    return
+  }
+
+  if (
+    node.sys.type in memoryNodeCountsBySysType &&
+    existingNodes.has(node.id)
+  ) {
+    memoryNodeCountsBySysType[node.sys.type] ||= 0
+    memoryNodeCountsBySysType[node.sys.type]--
+
+    if (memoryNodeCountsBySysType[node.sys.type] < 0) {
+      memoryNodeCountsBySysType[node.sys.type] = 0
+    }
+  }
+
+  existingNodes.delete(node.id)
+}