improve table parsing (Rebilly#38)

npentrel · Aug 2, 2023 · 21825dc · 21825dc
1 parent 831c383
commit 21825dc
Show file tree

Hide file tree

Showing 4 changed files with 55 additions and 4 deletions.
diff --git a/dist/index.js b/dist/index.js
diff --git a/dist/index.js.map b/dist/index.js.map
diff --git a/src/readability.test.ts b/src/readability.test.ts
@@ -178,8 +178,34 @@ More text after the list.
  );
 
  expect(stripped).toMatchInlineSnapshot(`
- "Chance of processing a transaction.
  "
+ Chance of processing a transaction.
+ "
+ `);
+ });
+
+ it('should convert tables without compressing surrounding text', () => {
+ const stripped = preprocessMarkdown(
+ `
+### Some heading text here
+
+Some content text.
+
+|A table header here|A table header here|
+|---|---|
+|Cell contents goes here|Cell contents goes here|
+
+This content comes after the table
+`
+ );
+
+ expect(stripped).toMatchInlineSnapshot(`
+ "Some content text.
+ A table header here.
+ A table header here.
+ Cell contents goes here.
+ Cell contents goes here.
+ This content comes after the table "
  `);
  });
 

diff --git a/src/readability.ts b/src/readability.ts
@@ -74,7 +74,6 @@ const replaceNodesWithTheirTextContent: Plugin = () => (tree) => {
  'linkReference',
  'table',
  'tableRow',
- 'tableCell',
  ];
 
  visit(tree, nodeTypesToReplace, (node, index, parent) => {
@@ -193,6 +192,21 @@ const convertTableToText: Plugin = () => (tree) => {
  tableCellNode.children = [];
  }
  });
+
+ // Encapsulate with a paragraph, replacing the table cell
+ visit(tree, 'tableCell', (node, index, parent) => {
+ const newNode = {
+ type: 'paragraph',
+ // @ts-ignore
+ children: node.children,
+ };
+
+ // @ts-ignore
+ parent?.children.splice(index, 1, newNode);
+
+ // Do not traverse `node`, continue at the node *now* at `index`.
+ return [SKIP, index];
+ });
 };
 
 // Remark plugin to remove list items that have less than 4 words.