support html past (e.g., jupyter notebook, QUIP)

package.json

@@ -24,6 +24,7 @@
 		"esbuild": "0.13.12",
 		"obsidian": "^0.15.1",
 		"tslib": "2.3.1",
-		"typescript": "4.4.4"
+		"typescript": "4.4.4",
+		"cheerio": "^1.0.0-rc.10"
 	}
 }

src/excel-markdown-tables.ts

@@ -30,5 +30,14 @@ export function excelRowsToMarkdown(rows: string[][]): string {
 }
 
 export function isExcelData(rows: string[][]): boolean {
-    return rows && rows[0] && rows[0].length > 1 ? true : false
+    if (!(rows && rows[0] && rows[0].length > 1)) {
+        return false;
+    }
+    // If we have multiple entries per row, each if this is the case for all rows.
+    const width: number = rows[0].length;
+    for (var row of rows) {
+        console.log("lenght: " + row.length);
+        if (row.length != width) { return false; }
+    }
+    return true;
 }

src/html-markdown-helpers.ts

@@ -0,0 +1,32 @@
+import * as cheerio from 'cheerio';
+
+/**
+ * Parse HTML content into text with \t and \n
+ * @param table_html The raw html from clipboard
+ * @returns Text string similiar to the row content from excel
+ */
+export function parseHtml(table_html: string): string {
+    const $ = cheerio.load(table_html);
+    let trs : cheerio.Cheerio<cheerio.Element> = $('tbody').find('tr');
+
+    let table_str: string = "";
+    trs.each((r_idx: number, row: cheerio.Element) => {
+        const cells: cheerio.Cheerio<cheerio.Element> = $(row).find('td,th');
+        const row_text: string[] = [];
+        cells.each((c_idx: number, cell: cheerio.Element) => {
+            // ignore hidden rows
+            const style: string | undefined = $(row).attr('style');
+            if (style) {
+                const m = style.match(/.*display.*:.*none.*/g);
+                if (m && m.length > 0) return;
+            }
+
+            // only consider cell text and ignore html
+            const content: string = $(cell).text().trim();
+            row_text.push(content);
+
+        });
+        table_str = table_str + row_text.join("\t") + "\n";
+    });
+    return table_str;
+}

src/main.ts

@@ -1,5 +1,6 @@
 import { Editor, MarkdownView, Plugin } from 'obsidian';
 import { excelToMarkdown, getExcelRows, isExcelData, excelRowsToMarkdown } from './excel-markdown-tables';
+import {parseHtml} from './html-markdown-helpers';
 
 export default class ExcelToMarkdownTablePlugin extends Plugin {
 	pasteHandler = (evt: ClipboardEvent, editor: Editor) => {
@@ -9,17 +10,29 @@ export default class ExcelToMarkdownTablePlugin extends Plugin {
 
 		// Check for `Shift + Mod + V` triggered event.
 		// Do not handle `Shift + Mod + V` events.
-		if (evt.clipboardData.types.length === 1 && evt.clipboardData.types[0] === 'text/plain') {	
+		if (evt.clipboardData.types.length === 1 && evt.clipboardData.types[0] === 'text/plain') {
 			return;
 		}
 
 		const rawData = evt.clipboardData.getData("text");
 		const rows = getExcelRows(rawData);
-		
+
 		if (isExcelData(rows)) {
 			const markdownData = excelRowsToMarkdown(rows);
 			editor.replaceSelection(markdownData + '\n');
 			evt.preventDefault();
+			return;
+		}
+
+		// Try parsing as HTML
+		const rawHtml = evt.clipboardData.getData("text/html");
+		const htmlRows = getExcelRows(parseHtml(rawHtml));
+
+		if (isExcelData(htmlRows)) {
+			const markdownData = excelRowsToMarkdown(htmlRows);
+			editor.replaceSelection(markdownData + '\n');
+			evt.preventDefault();
+			return;
 		}
 	}