Add more page content

Signed-off-by: Daishan Peng <[email protected]>

src/index.ts

@@ -2,7 +2,9 @@ import { Client } from "@notionhq/client";
 import dotenv from "dotenv";
 import { writeFile, mkdir } from "fs/promises";
 import path from "path";
-import { SearchResponse } from "@notionhq/client/build/src/api-endpoints";
+import { PageObjectResponse, SearchResponse } from "@notionhq/client/build/src/api-endpoints";
+import { getPageContent } from "./page";
+import * as fs from "node:fs";
 
 dotenv.config();
 
@@ -12,18 +14,25 @@ async function main() {
   });
 
   // Function to write a page to a file
-  async function writePageToFile(page: any, directory: string) {
-    const pageId = page.id.replace(/-/g, '');
-    const filePath = path.join(directory, `${pageId}.data`);
-    await writeFile(filePath, JSON.stringify(page, null, 2));
-    console.log(`Wrote page ${pageId} to ${filePath}`);
+  async function writePageToFile(page: PageObjectResponse, directory: string) {
+    const pageId = page.id;
+    const pageContent = await getPageContent(notion, pageId);
+    const fileDir = path.join(directory, pageId.toString());
+    await mkdir(fileDir, { recursive: true });
+    let title = ((page.properties?.title ?? page.properties?.Name) as any)?.title[0]?.plain_text?.trim().replaceAll(/\//g, "-");
+    if (!title) {
+      title = pageId.toString();
+    }
+    const filePath = path.join(fileDir, title + ".md");
+    fs.writeFileSync(filePath, pageContent, "utf8");
   }
 
   // Function to fetch all pages
   async function fetchAllPages() {
     let pages: any[] = [];
     let cursor: string | undefined = undefined;
 
+
     while (true) {
       const response: SearchResponse = await notion.search({
         filter: {
@@ -47,15 +56,22 @@ async function main() {
 
   // Fetch all pages
   const pages = await fetchAllPages();
+  let metadata: Map<string, {
+    url: string;
+  }> = new Map();
 
-  // Define the output directory
   const outputDir = path.join(process.env.WORKSPACE_DIR!!, 'knowledge', 'integrations', 'notion');
-  await mkdir(outputDir, { recursive: true }); // Ensure the directory exists
+  await mkdir(outputDir, { recursive: true });
+
+  for (const page of pages) {
+    await writePageToFile(page, outputDir);
+    metadata.set(page.id, {
+      url: page.url,
+    })
+  }
 
-  // Write all pages to files
-  await Promise.all(
-    pages.map((page) => writePageToFile(page, outputDir))
-  );
+  const metadataPath = path.join(outputDir, 'metadata.json');
+  await writeFile(metadataPath, JSON.stringify(Object.fromEntries(metadata)), 'utf8');
 
   console.log(`Finished writing ${pages.length} pages to ${outputDir}`);
 }

src/page.ts

@@ -0,0 +1,178 @@
+import { Client } from "@notionhq/client";
+import { BlockObjectResponse } from "@notionhq/client/build/src/api-endpoints";
+
+
+export async function getPageContent(client: Client, id: string, indentation = 0): Promise<string> {
+  const blocks = await client.blocks.children.list({block_id: id})
+  let result: string = '';
+  for (let b of blocks.results) {
+    let block = b as BlockObjectResponse;
+    // Tables are complicated, so we handle them completely separately
+    if (block.type === "table") {
+      result += await printTable(client, b)
+      continue
+    }
+
+    result += await printBlock(client, b as BlockObjectResponse, indentation)
+    if (block.has_children && block.type !== "child_page" && block.type !== "synced_block") {
+      result += await getPageContent(client, b.id, indentation + 2)
+    }
+  }
+  return result
+}
+
+async function printBlock(client: Client, b: BlockObjectResponse, indentation: number): Promise<string> {
+  let result: string = ""
+  if (indentation > 0) {
+    result += " ".repeat(indentation)
+  }
+  switch (b.type) {
+    case "bookmark":
+      if (b.bookmark.caption !== null && richTextArrayToString(b.bookmark.caption) !== "") {
+        result += `Bookmark: ${b.bookmark.url} (${richTextArrayToString(b.bookmark.caption)})`
+      } else {
+        result += `Bookmark: ${b.bookmark.url}`
+      }
+      break
+    case "bulleted_list_item":
+      result += `- ${richTextArrayToString(b.bulleted_list_item.rich_text)}`
+      break
+    case "callout":
+      result += `> ${richTextArrayToString(b.callout.rich_text)}`
+      break
+    case "child_database":
+      result += `Child Database: ${b.child_database.title}`
+      break
+    case "child_page":
+      result += `Child Page: ${b.child_page.title}`
+      break
+    case "code":
+      if (b.code.language !== null) {
+        result += "```" + b.code.language + "\n"
+      } else {
+        result += "```\n"
+      }
+      result += richTextArrayToString(b.code.rich_text)
+      result += "\n```"
+      if (b.code.caption !== null && richTextArrayToString(b.code.caption) !== "") {
+        result += `\n(${richTextArrayToString(b.code.caption)})`
+      }
+      break
+    case "divider":
+      result += "-------------------------------------"
+      break
+    case "embed":
+      result += `Embed: ${b.embed.url}`
+      break
+    case "equation":
+      result += `Equation: ${b.equation.expression}`
+      break
+    case "file":
+      result += fileToString("File", b.file)
+      break
+    case "heading_1":
+      result += `# ${richTextArrayToString(b.heading_1.rich_text)}`
+      break
+    case "heading_2":
+      result += `## ${richTextArrayToString(b.heading_2.rich_text)}`
+      break
+    case "heading_3":
+      result += `### ${richTextArrayToString(b.heading_3.rich_text)}`
+      break
+    case "image":
+      result += fileToString("Image", b.image)
+      break
+    case "link_preview":
+      result += b.link_preview.url
+      break
+    case "numbered_list_item":
+      result += `1. ${richTextArrayToString(b.numbered_list_item.rich_text)}`
+      break
+    case "paragraph":
+      result += richTextArrayToString(b.paragraph.rich_text)
+      break
+    case "pdf":
+      result += fileToString("PDF", b.pdf)
+      break
+    case "quote":
+      result += "\"\"\"\n"
+      result += richTextArrayToString(b.quote.rich_text)
+      result += "\n\"\"\""
+      break
+    case "synced_block":
+      if (b.synced_block.synced_from !== null) {
+        await getPageContent(client, b.synced_block.synced_from.block_id, indentation)
+      }
+      break
+    case "to_do":
+      if (b.to_do.checked) {
+        result += `[x] ${richTextArrayToString(b.to_do.rich_text)}`
+      } else {
+        result += `[ ] ${richTextArrayToString(b.to_do.rich_text)}`
+      }
+      break
+    case "toggle":
+      result += `> ${richTextArrayToString(b.toggle.rich_text)}`
+      break
+    case "video":
+      result += fileToString("Video", b.video)
+      break
+  }
+  return result.replace("\n", "\n" + " ".repeat(indentation))
+}
+
+export function richTextArrayToString(richTextArray: any[]) {
+  let result = ""
+  for (let r of richTextArray) {
+    result += r.plain_text + " "
+  }
+  return result
+}
+
+function fileToString(prefix: any, file: any) {
+  let result = ""
+  if (file.type === "file") {
+    result = `${prefix}: ${file.file.url} (expires ${file.file.expiry_time})`
+  } else if (file.type === "external") {
+    result = `External ${prefix}: ${file.external.url}`
+  }
+  if (file.caption !== null && richTextArrayToString(file.caption) !== "") {
+    result += ` (${richTextArrayToString(file.caption)})`
+  }
+  return result
+}
+
+async function printTable(client: Client, table: any) {
+  const children = await client.blocks.children.list({block_id: table.id})
+  if (table.table.has_column_header && children.results.length > 0) {
+    printTableRow((children.results[0] as any).table_row, table.table.has_row_header, true)
+    for (let i = 1; i < children.results.length; i++) {
+      printTableRow((children.results[i] as any).table_row, table.table.has_row_header, false)
+    }
+  } else {
+    for (let r of children.results) {
+      printTableRow((r as any).table_row, table.table.has_row_header, false)
+    }
+  }
+}
+
+function printTableRow(row: any, boldFirst: any, boldAll: any): string {
+  let result = "|"
+  if (boldAll) {
+    for (let c of row.cells) {
+      result += ` **${richTextArrayToString(c)}** |`
+    }
+    let len = result.length
+    result += "\n|" + "-".repeat(len - 2) + "|"
+  } else if (boldFirst && row.cells.length > 0) {
+    result += ` **${richTextArrayToString(row.cells[0])}** |`
+    for (let i = 1; i < row.cells.length; i++) {
+      result += ` ${richTextArrayToString(row.cells[i])} |`
+    }
+  } else {
+    for (let c of row.cells) {
+      result += ` ${richTextArrayToString(c)} |`
+    }
+  }
+  return result
+}

tool.gpt

@@ -1,10 +1,7 @@
 Name: Sync Notion Pages
 Description: Provides access to the Notion API (read-only)
-Credential: github.com/gptscript-ai/gateway-oauth2 as notion.read with NOTION_TOKEN as env and notion as integration
-Context: syncPages
+Credential: github.com/gptscript-ai/gateway-oauth2 as notion.sync-file with NOTION_TOKEN as env and notion as integration
+
+#!/usr/bin/env npm --prefix ${GPTSCRIPT_TOOL_DIR} run start
 
----
-Name: syncPages
-Description: Syncs Notion Pages
 
-#!/usr/bin/env npm run start