Merge pull request #116 from sillsdev/DownloadVideo

fix: Download Notion-hosted videos so the links don't expire (#110)

Author: andrew-polk

Diff for: src/assets.ts

@@ -0,0 +1,23 @@
+import * as fs from "fs-extra";
+import * as Path from "path";
+import { verbose } from "./log";
+
+export enum AssetType {
+  Image = "image",
+  Video = "video",
+}
+
+export function writeAsset(path: string, buffer: Buffer): void {
+  // Note: it's tempting to not spend time writing this out if we already have
+  // it from a previous run. But we don't really know it's the same. A) it
+  // could just have the same name, B) it could have been previously
+  // unlocalized and thus filled with a copy of the primary language image
+  // while and now is localized.
+  if (fs.pathExistsSync(path)) {
+    verbose("Replacing asset " + path);
+  } else {
+    verbose("Adding asset " + path);
+    fs.mkdirsSync(Path.dirname(path));
+  }
+  fs.createWriteStream(path).write(buffer); // async but we're not waiting
+}

Diff for: src/images.ts

@@ -1,7 +1,6 @@
 import * as fs from "fs-extra";
 import FileType, { FileTypeResult } from "file-type";
 import axios from "axios";
-import * as Path from "path";
 import { makeImagePersistencePlan } from "./MakeImagePersistencePlan";
 import { warning, logDebug, verbose, info } from "./log";
 import { ListBlockChildrenResponseResult } from "notion-to-md/build/types";
@@ -10,6 +9,7 @@ import {
   IDocuNotionContextPageInfo,
   IPlugin,
 } from "./plugins/pluginTypes";
+import { writeAsset } from "./assets";
 
 // We handle several things here:
 // 1) copy images locally instead of leaving them in Notion
@@ -158,11 +158,13 @@ async function readPrimaryImage(imageSet: ImageSet) {
 }
 
 async function saveImage(imageSet: ImageSet): Promise<void> {
-  writeImageIfNew(imageSet.primaryFileOutputPath!, imageSet.primaryBuffer!);
+  const path = imageSet.primaryFileOutputPath!;
+  imageWasSeen(path);
+  writeAsset(path, imageSet.primaryBuffer!);
 
   for (const localizedImage of imageSet.localizedUrls) {
     let buffer = imageSet.primaryBuffer!;
-    // if we have a urls for the localized screenshot, download it
+    // if we have a url for the localized screenshot, download it
     if (localizedImage?.url.length > 0) {
       verbose(`Retrieving ${localizedImage.iso632Code} version...`);
       const response = await fetch(localizedImage.url);
@@ -180,30 +182,15 @@ async function saveImage(imageSet: ImageSet): Promise<void> {
       imageSet.pageInfo!.relativeFilePathToFolderContainingPage
     }`;
 
-    writeImageIfNew(
-      (directory + "/" + imageSet.outputFileName!).replaceAll("//", "/"),
-      buffer
+    const newPath = (directory + "/" + imageSet.outputFileName!).replaceAll(
+      "//",
+      "/"
     );
+    imageWasSeen(newPath);
+    writeAsset(newPath, buffer);
   }
 }
 
-function writeImageIfNew(path: string, buffer: Buffer) {
-  imageWasSeen(path);
-
-  // Note: it's tempting to not spend time writing this out if we already have
-  // it from a previous run. But we don't really know it's the same. A) it
-  // could just have the same name, B) it could have been previously
-  // unlocalized and thus filled with a copy of the primary language image
-  // while and now is localized.
-  if (fs.pathExistsSync(path)) {
-    verbose("Replacing image " + path);
-  } else {
-    verbose("Adding image " + path);
-    fs.mkdirsSync(Path.dirname(path));
-  }
-  fs.createWriteStream(path).write(buffer); // async but we're not waiting
-}
-
 export function parseImageBlock(image: any): ImageSet {
   if (!locales) throw Error("Did you call initImageHandling()?");
   const imageSet: ImageSet = {

Diff for: src/plugins/VideoTransformer.spec.ts

@@ -1,7 +1,28 @@
+import * as fs from "fs-extra";
 import { setLogLevel } from "../log";
 import { NotionBlock } from "../types";
 import { standardVideoTransformer } from "./VideoTransformer";
-import { blocksToMarkdown } from "./pluginTestRun";
+import { blocksToMarkdown, kTemporaryTestDirectory } from "./pluginTestRun";
+
+beforeAll(async () => {
+  try {
+    if (await fs.pathExists(kTemporaryTestDirectory)) {
+      await fs.emptyDir(kTemporaryTestDirectory);
+    } else {
+      await fs.mkdirp(kTemporaryTestDirectory);
+    }
+  } catch (err) {
+    console.error("Error in beforeAll:", err);
+  }
+});
+
+afterAll(async () => {
+  try {
+    await fs.remove(kTemporaryTestDirectory);
+  } catch (err) {
+    console.error("Error in afterAll:", err);
+  }
+});
 
 test("youtube embedded", async () => {
   const config = { plugins: [standardVideoTransformer] };
@@ -89,6 +110,9 @@ test("video link, not embedded", async () => {
 test("direct upload to to Notion (embedded)", async () => {
   setLogLevel("verbose");
   const config = { plugins: [standardVideoTransformer] };
+
+  const fileName1 = "first_video.mp4";
+  const fileName2 = "second_video.mp4";
   const result = await blocksToMarkdown(config, [
     {
       object: "block",
@@ -103,13 +127,39 @@ test("direct upload to to Notion (embedded)", async () => {
         caption: [],
         type: "file",
         file: {
-          url: "https://s3.us-west-2.amazonaws.com/secure.notion-static.com/f6bc4746-011e-2124-86ca-ed4337d70891/people_fre_motionAsset_p3.mp4?X-Blah-blah",
+          url: `https://s3.us-west-2.amazonaws.com/secure.notion-static.com/f6bc4746-011e-2124-86ca-ed4337d70891/${fileName1}?X-Blah-blah`,
+        },
+      },
+    } as unknown as NotionBlock,
+    {
+      object: "block",
+      id: "12f7db3b-4412-4be9-a3f7-6ac423fee94b",
+      parent: {
+        type: "page_id",
+        page_id: "edaffeb2-ece8-4d44-976f-351e6b5757bb",
+      },
+
+      type: "video",
+      video: {
+        caption: [],
+        type: "file",
+        file: {
+          url: `https://s3.us-west-2.amazonaws.com/secure.notion-static.com/f6bc4746-011e-2124-86ca-ed4337d70891/${fileName2}?X-Blah-blah`,
         },
       },
     } as unknown as NotionBlock,
   ]);
+
   expect(result).toContain(`import ReactPlayer from "react-player";`);
-  expect(result).toContain(
-    `<ReactPlayer controls url="https://s3.us-west-2.amazonaws.com/secure.notion-static.com/f6bc4746-011e-2124-86ca-ed4337d70891/people_fre_motionAsset_p3.mp4?X-Blah-blah" />`
-  );
+  expect(result).toContain(`import video1 from "./${fileName1}";`);
+  expect(result).toContain(`import video2 from "./${fileName2}";`);
+  expect(result).toContain(`<ReactPlayer controls url={video1} />`);
+  expect(result).toContain(`<ReactPlayer controls url={video2} />`);
+
+  // Wait half a second for the files to be written
+  await new Promise(resolve => setTimeout(resolve, 500));
+
+  // We should have actually created files in "tempTestFileDir/"
+  expect(await fs.pathExists("tempTestFileDir/" + fileName1)).toBe(true);
+  expect(await fs.pathExists("tempTestFileDir/" + fileName2)).toBe(true);
 });

Diff for: src/plugins/VideoTransformer.ts

@@ -1,42 +1,96 @@
+import * as Path from "path";
 import { VideoBlockObjectResponse } from "@notionhq/client/build/src/api-endpoints";
+import { ListBlockChildrenResponseResult } from "notion-to-md/build/types";
 import { IDocuNotionContext, IPlugin } from "./pluginTypes";
 import { warning } from "../log";
 import { NotionBlock } from "../types";
+import { writeAsset } from "../assets";
 
 export const standardVideoTransformer: IPlugin = {
   name: "video",
   notionToMarkdownTransforms: [
     {
       type: "video",
-      getStringFromBlock: (
-        context: IDocuNotionContext,
-        block: NotionBlock
-      ): string => {
-        const video = (block as VideoBlockObjectResponse).video;
-        let url = "";
-        switch (video.type) {
-          case "external":
-            url = video.external.url;
-            break;
-          case "file":
-            url = video.file.url;
-            break;
-          default:
-            // video.type can only be "external" or "file" as of the writing of this code, so typescript
-            // isn't happy trying to turn video.type into a string. But this default in our switch is
-            // just attempting some future-proofing. Thus the strange typing/stringifying below.
-            warning(
-              `[standardVideoTransformer] Found Notion "video" block with type ${JSON.stringify(
-                (video as any).type
-              )}. The best docu-notion can do for now is ignore it.`
-            );
-            return "";
-            break;
-        }
-
-        context.imports.push(`import ReactPlayer from "react-player";`);
-        return `<ReactPlayer controls url="${url}" />`;
-      },
+      getStringFromBlock: (context: IDocuNotionContext, block: NotionBlock) =>
+        markdownToMDVideoTransformer(block, context),
     },
   ],
 };
+
+async function markdownToMDVideoTransformer(
+  block: ListBlockChildrenResponseResult,
+  context: IDocuNotionContext
+): Promise<string> {
+  const videoBlock = block as VideoBlockObjectResponse;
+  const video = videoBlock.video;
+  let url = "";
+  switch (video.type) {
+    case "external":
+      url = `"${video.external.url}"`;
+      break;
+    case "file":
+      // The url we get for a Notion-hosted asset expires after an hour, so we have to download it locally.
+      url = await downloadVideoAndConvertUrl(
+        context,
+        video.file.url,
+        videoBlock.id
+      );
+      break;
+    default:
+      // video.type can only be "external" or "file" as of the writing of this code, so typescript
+      // isn't happy trying to turn video.type into a string. But this default in our switch is
+      // just attempting some future-proofing. Thus the strange typing/stringifying below.
+      warning(
+        `[standardVideoTransformer] Found Notion "video" block with type ${JSON.stringify(
+          (video as any).type
+        )}. The best docu-notion can do for now is ignore it.`
+      );
+      return "";
+  }
+
+  context.imports.push(`import ReactPlayer from "react-player";`);
+  return `<ReactPlayer controls url=${url} />`;
+}
+
+// ENHANCE: One day, we may want to allow for options of where to place the files, how
+// to name them, etc. Or we could at least follow the image options.
+// But for now, I'm just trying to fix the bug that Notion-hosted videos don't work at all.
+async function downloadVideoAndConvertUrl(
+  context: IDocuNotionContext,
+  notionVideoUrl: string,
+  blockId: string
+): Promise<string> {
+  // Get the file name from the url. Ignore query parameters and fragments.
+  let newFileName = notionVideoUrl.split("?")[0].split("#")[0].split("/").pop();
+
+  if (!newFileName) {
+    // If something went wrong, fall back to the block ID.
+    // But at least try to get the extension from the url.
+    const extension = notionVideoUrl
+      .split("?")[0]
+      .split("#")[0]
+      .split(".")
+      .pop();
+    newFileName = blockId + (extension ? "." + extension : "");
+  }
+
+  const newPath = Path.posix.join(
+    context.pageInfo.directoryContainingMarkdown,
+    newFileName
+  );
+
+  const response = await fetch(notionVideoUrl);
+  const arrayBuffer = await response.arrayBuffer();
+  const buffer = Buffer.from(arrayBuffer);
+  writeAsset(newPath, buffer);
+
+  // Add an import statement for the video file.
+  // Otherwise, the docusaurus build won't include the video file in the build.
+  const countVideoImports = context.imports.filter(i => {
+    return /import video\d+/.exec(i);
+  }).length;
+  const importName = `video${countVideoImports + 1}`;
+  context.imports.push(`import ${importName} from "./${newFileName}";`);
+
+  return `{${importName}}`;
+}

Diff for: src/plugins/pluginTestRun.ts

@@ -10,6 +10,8 @@ import { NotionBlock } from "../types";
 import { convertInternalUrl } from "./internalLinks";
 import { numberChildrenIfNumberedList } from "../pull";
 
+export const kTemporaryTestDirectory = "tempTestFileDir";
+
 export async function blocksToMarkdown(
   config: IDocuNotionConfig,
   blocks: NotionBlock[],
@@ -49,7 +51,7 @@ export async function blocksToMarkdown(
 
     //TODO might be needed for some tests, e.g. the image transformer...
     pageInfo: {
-      directoryContainingMarkdown: "not yet",
+      directoryContainingMarkdown: kTemporaryTestDirectory,
       relativeFilePathToFolderContainingPage: "not yet",
       slug: "not yet",
     },