Add new config options & flags (#8)

* feat: augmented feats

Added two features which are allowed by the Notion endpoint API and can be useful to further
customize the export:
- recursive, allows to export one block and alla the block children
- no-files, to avoid downloading files like pdf or images (as it is optional from the UI)

Also, addeda function `getMdFiles` to dump the files in a folder (this is useful if we
want to use a function from langchain, for example, to load all files in a folder).

* chore: Readme updated

* chore: remove comment

* feat: added new flags

* Clean-up

* Introduce config object.

* Adapt readme.

---------

Co-authored-by: Guille <guillermo.c.martinez@dcsl.com>

Author: yannbolliger

README.md

@@ -43,10 +43,11 @@ most accurate information.
 #### Constructor
 
 Provide the [required Cookies](#needed-cookies) as authentification to create a
-new exporter client.
+new exporter client. For configuration options,
+[refer to the definition](./src/config.ts).
 
 ```ts
-const exporter = new NotionExporter(tokenV2: string, fileToken: string)
+const exporter = new NotionExporter(tokenV2: string, fileToken: string, config?: Config)
 ```
 
 #### Methods
@@ -130,4 +131,5 @@ use-case? Please submit issues and PRs on Github.
 
 ### Contributors
 
-- Yann Bolliger, [@yannbolliger](https://github.com/yannbolliger).
+- Yann Bolliger, [@yannbolliger](https://github.com/yannbolliger)
+- Guillermo C. Martínez, [@telekosmos](https://github.com/telekosmos)

src/NotionExporter.ts

@@ -2,6 +2,7 @@ import axios, { AxiosInstance } from "axios"
 import AdmZip from "adm-zip"
 
 import { blockIdFromUrl, validateUuid } from "./blockId"
+import { Config, defaultConfig } from "./config"
 
 interface Task {
   id: string
@@ -12,6 +13,7 @@ interface Task {
 /** Lightweight client to export ZIP, Markdown or CSV files from a Notion block/page. */
 export class NotionExporter {
   protected readonly client: AxiosInstance
+  private readonly config: Config
 
   /**
    * Create a new NotionExporter client. To export any blocks/pages from
@@ -21,13 +23,14 @@ export class NotionExporter {
    * @param tokenV2 – the Notion `token_v2` Cookie value
    * @param fileToken – the Notion `file_token` Cookie value
    */
-  constructor(tokenV2: string, fileToken: string) {
+  constructor(tokenV2: string, fileToken: string, config?: Config) {
     this.client = axios.create({
       baseURL: "https://www.notion.so/api/v3/",
       headers: {
         Cookie: `token_v2=${tokenV2};file_token=${fileToken}`,
       },
     })
+    this.config = Object.assign(defaultConfig, config)
   }
 
   /**
@@ -40,17 +43,17 @@ export class NotionExporter {
     const id = validateUuid(blockIdFromUrl(idOrUrl))
     if (!id) return Promise.reject(`Invalid URL or blockId: ${idOrUrl}`)
 
+    const { recursive, includeContents, ...config } = this.config
     const res = await this.client.post("enqueueTask", {
       task: {
         eventName: "exportBlock",
         request: {
           block: { id },
-          recursive: false,
+          recursive,
           exportOptions: {
             exportType: "markdown",
-            timeZone: "Europe/Zurich",
-            locale: "en",
-            collectionViewExportType: "all",
+            includeContents: !includeContents ? "no_files" : undefined,
+            ...config,
           },
         },
       },

src/action.ts

@@ -2,6 +2,7 @@ import rl from "readline"
 import { AxiosError } from "axios"
 
 import { NotionExporter } from "./NotionExporter"
+import { Config } from "./config"
 
 export const FileType = ["md", "csv"] as const
 type FileType = (typeof FileType)[number]
@@ -25,15 +26,15 @@ const askToken = (tokenName: string): Promise<string> => {
 const envOrAskToken = async (tokenName: string) =>
   process.env[tokenName] || (await askToken(tokenName))
 
-const action = async (blockId: string, fileType: string) => {
+const action = async (blockId: string, fileType: string, config?: Config) => {
   if (!isFileType(fileType)) {
     console.log(`File type (-t, --type) has to be one of: ${FileType}`)
     process.exit(1)
   }
 
   const tokenV2 = await envOrAskToken("NOTION_TOKEN")
   const fileToken = await envOrAskToken("NOTION_FILE_TOKEN")
-  const exporter = new NotionExporter(tokenV2, fileToken)
+  const exporter = new NotionExporter(tokenV2, fileToken, config)
 
   const outputStr =
     fileType === "csv"

src/config.ts

@@ -0,0 +1,19 @@
+/** Configuration options that are passed to the Notion API. */
+export interface Config {
+  /** Export embedded image and pdf files, not only the (text) content. Default: false */
+  includeContents?: boolean
+  /** Export children subpages recursively. Default: false */
+  recursive?: boolean
+  /** Default: UTC */
+  timeZone?: string
+  /** Default: en */
+  locale?: string
+  /** Export all blocks of the DB/page or just the ones in the current view. Default: "all" */
+  collectionViewExportType?: "currentView" | "all"
+}
+
+export const defaultConfig: Config = {
+  timeZone: "UTC",
+  locale: "en",
+  collectionViewExportType: "all",
+}

src/index.ts

@@ -4,6 +4,7 @@ import { NotionExporter } from "./NotionExporter"
 import action, { FileType } from "./action"
 
 export default NotionExporter
+export { Config, defaultConfig } from "./config"
 
 export const cli = (args: string[]) => {
   const pkg = require("../package.json")
@@ -21,14 +22,24 @@ export const cli = (args: string[]) => {
 
     © ${pkg.author}, 2022.`
     )
+
     .option("-t, --type", `File type to be exported: ${FileType}`, "md")
-    //  .option("-o, --output", "Output path of the exported file, stdin if empty")
+    .option("-r, --recursive", "Export children subpages", false)
+    .option(
+      "-a, --all-files",
+      "Export image and pdf files, not only content",
+      false
+    )
     .example(
       "https://www.notion.so/Notion-Official-83715d7703ee4b8699b5e659a4712dd8"
     )
     .example("83715d7703ee4b8699b5e659a4712dd8 -t md")
     .example("3af0a1e347dd40c5ba0a2c91e234b2a5 -t csv > list.csv")
-    //    .example("83715d7703ee4b8699b5e659a4712dd8 -t md -o blog.md")
-    .action((blockId, opts) => action(blockId, opts.type))
+    .action((blockId, opts) =>
+      action(blockId, opts.type, {
+        includeContents: opts["all-files"],
+        recursive: opts.recursive,
+      })
+    )
     .parse(args)
 }