Merge pull request #11 from JohnThomson/newlineInTable

Handle newlines in notion table cells

Author: hatton

.gitignore

@@ -4,3 +4,4 @@ sample_img/
 i18n/
 node_modules/
 version.json
+docs/

package.json

@@ -52,13 +52,15 @@
     "file-type": "16.5.1",
     "fs-extra": "^10.1.0",
     "limiter": "^2.1.0",
+    "markdown-table": "^2.0.0",
     "node-fetch": "2.6.6",
     "notion-to-md": "^2.5.2",
     "path": "^0.12.7",
     "postinstall-postinstall": "^2.1.0",
     "sanitize-filename": "^1.6.3"
   },
   "devDependencies": {
+    "@types/markdown-table": "^2.0.0",
     "@types/fs-extra": "^9.0.13",
     "@types/jest": "^28.1.6",
     "@types/node": "^12.20.11",

src/CustomTranformers.ts

@@ -1,6 +1,7 @@
 import { Client } from "@notionhq/client";
 import { ListBlockChildrenResponse } from "@notionhq/client/build/src/api-endpoints";
 import { NotionToMarkdown } from "notion-to-md";
+import markdownTable from "markdown-table";
 import {
   ListBlockChildrenResponseResult,
   ListBlockChildrenResponseResults,
@@ -22,6 +23,59 @@ export function setupCustomTransformers(
       notionColumnToMarkdown(notionToMarkdown, notionClient, block)
   );
 
+  // This is mostly a copy of the table handler from notion-to-md. The change is to handle newlines in the
+  // notion cell content.
+  notionToMarkdown.setCustomTransformer(
+    "table",
+    async (block: ListBlockChildrenResponseResult) => {
+      const { id, has_children } = block as any;
+      const tableArr: string[][] = [];
+      if (has_children) {
+        const tableRows = await getBlockChildren(notionClient, id, 100);
+        // console.log(">>", tableRows);
+        const rowsPromise = tableRows?.map(async row => {
+          const { type } = row as any;
+          const cells = (row as any)[type]["cells"];
+
+          /**
+           * this is more like a hack since matching the type text was
+           * difficult. So converting each cell to paragraph type to
+           * reuse the blockToMarkdown function
+           */
+          const cellStringPromise = cells.map(
+            async (cell: any) =>
+              await notionToMarkdown.blockToMarkdown({
+                type: "paragraph",
+                paragraph: { rich_text: cell },
+              } as ListBlockChildrenResponseResult)
+          );
+
+          const cellStringArrRaw: string[] = await Promise.all(
+            cellStringPromise
+          );
+          // This is our patch to the original notion-to-md code.
+          const cellStringArr = cellStringArrRaw.map(c =>
+            c
+              // Trailing newlines are almost certainly not wanted, and converting to br's gives weird results
+              .replace(/[\r\n]+$/, "")
+              // Preserving line breaks within cells can't be done in stock markdown. Since we're producing
+              // mdx, which supports embedded HTML, we can handle it with <br/>.
+              // I'm not sure exactly what line breaks might occur in the input, depending on platform,
+              // so handle all the common cases.
+              .replaceAll("\r\n", "<br/>")
+              .replaceAll("\n", "<br/>")
+              .replaceAll("\r", "<br/>")
+          );
+          // console.log("~~", cellStringArr);
+          tableArr.push(cellStringArr);
+          // console.log(tableArr);
+        });
+        await Promise.all(rowsPromise || []);
+      }
+      return markdownTable(tableArr);
+    }
+  );
+
   // Note: Pull.ts also adds an image transformer, but has to do that for each
   // page so we don't do it here.
 }

yarn.lock

@@ -1272,6 +1272,11 @@
   resolved "https://registry.npmjs.org/@types/json-schema/-/json-schema-7.0.11.tgz"
   integrity sha512-wOuvG1SN4Us4rez+tylwwwCV1psiNVOkJeM3AUWUNWg/jDQY2+HE/444y5gc+jBmRqASOm2Oeh5c1axHobwRKQ==
 
+"@types/markdown-table@^2.0.0":
+  version "2.0.0"
+  resolved "https://registry.yarnpkg.com/@types/markdown-table/-/markdown-table-2.0.0.tgz#d2a3458c61ee71c8ee2b40b76c199b85b8dbd70c"
+  integrity sha512-fVZN/DRjZvjuk+lo7ovlI/ZycS51gpYU5vw5EcFeqkcX6lucQ+UWgEOH2O4KJHkSck4DHAY7D7CkVLD0wzc5qw==
+
 "@types/minimist@^1.2.0":
   version "1.2.2"
   resolved "https://registry.npmjs.org/@types/minimist/-/minimist-1.2.2.tgz"