From 145e26698791221b007c7dd460fb506cb0237235 Mon Sep 17 00:00:00 2001
From: wxiaoguang <wxiaoguang@gmail.com>
Date: Thu, 7 Nov 2024 11:57:07 +0800
Subject: Support quote selected comments to reply (#32431)

Many existing tests were quite hacky, these could be improved later.

<details>

![image](https://github.com/user-attachments/assets/93aebb4f-9de5-4cb8-910b-50c64cbcd25a)

</details>
---
 web_src/js/markup/html2markdown.test.ts |  24 +++++++
 web_src/js/markup/html2markdown.ts      | 119 ++++++++++++++++++++++++++++++++
 2 files changed, 143 insertions(+)
 create mode 100644 web_src/js/markup/html2markdown.test.ts
 create mode 100644 web_src/js/markup/html2markdown.ts

(limited to 'web_src/js/markup')

diff --git a/web_src/js/markup/html2markdown.test.ts b/web_src/js/markup/html2markdown.test.ts
new file mode 100644
index 0000000000..99a63956a0
--- /dev/null
+++ b/web_src/js/markup/html2markdown.test.ts
@@ -0,0 +1,24 @@
+import {convertHtmlToMarkdown} from './html2markdown.ts';
+import {createElementFromHTML} from '../utils/dom.ts';
+
+const h = createElementFromHTML;
+
+test('convertHtmlToMarkdown', () => {
+  expect(convertHtmlToMarkdown(h(`<h1>h</h1>`))).toBe('# h');
+  expect(convertHtmlToMarkdown(h(`<strong>txt</strong>`))).toBe('**txt**');
+  expect(convertHtmlToMarkdown(h(`<em>txt</em>`))).toBe('_txt_');
+  expect(convertHtmlToMarkdown(h(`<del>txt</del>`))).toBe('~~txt~~');
+
+  expect(convertHtmlToMarkdown(h(`<a href="link">txt</a>`))).toBe('[txt](link)');
+  expect(convertHtmlToMarkdown(h(`<a href="https://link">https://link</a>`))).toBe('https://link');
+
+  expect(convertHtmlToMarkdown(h(`<img src="link">`))).toBe('![image](link)');
+  expect(convertHtmlToMarkdown(h(`<img src="link" alt="name">`))).toBe('![name](link)');
+  expect(convertHtmlToMarkdown(h(`<img src="link" width="1" height="1">`))).toBe('<img alt="image" width="1" height="1" src="link">');
+
+  expect(convertHtmlToMarkdown(h(`<p>txt</p>`))).toBe('txt\n');
+  expect(convertHtmlToMarkdown(h(`<blockquote>a\nb</blockquote>`))).toBe('> a\n> b\n');
+
+  expect(convertHtmlToMarkdown(h(`<ol><li>a<ul><li>b</li></ul></li></ol>`))).toBe('1. a\n    * b\n\n');
+  expect(convertHtmlToMarkdown(h(`<ol><li><input checked>a</li></ol>`))).toBe('1. [x] a\n');
+});
diff --git a/web_src/js/markup/html2markdown.ts b/web_src/js/markup/html2markdown.ts
new file mode 100644
index 0000000000..c690e0c8b1
--- /dev/null
+++ b/web_src/js/markup/html2markdown.ts
@@ -0,0 +1,119 @@
+import {htmlEscape} from 'escape-goat';
+
+type Processors = {
+  [tagName: string]: (el: HTMLElement) => string | HTMLElement | void;
+}
+
+type ProcessorContext = {
+  elementIsFirst: boolean;
+  elementIsLast: boolean;
+  listNestingLevel: number;
+}
+
+function prepareProcessors(ctx:ProcessorContext): Processors {
+  const processors = {
+    H1(el) {
+      const level = parseInt(el.tagName.slice(1));
+      el.textContent = `${'#'.repeat(level)} ${el.textContent.trim()}`;
+    },
+    STRONG(el) {
+      return `**${el.textContent}**`;
+    },
+    EM(el) {
+      return `_${el.textContent}_`;
+    },
+    DEL(el) {
+      return `~~${el.textContent}~~`;
+    },
+
+    A(el) {
+      const text = el.textContent || 'link';
+      const href = el.getAttribute('href');
+      if (/^https?:/.test(text) && text === href) {
+        return text;
+      }
+      return href ? `[${text}](${href})` : text;
+    },
+    IMG(el) {
+      const alt = el.getAttribute('alt') || 'image';
+      const src = el.getAttribute('src');
+      const widthAttr = el.hasAttribute('width') ? ` width="${htmlEscape(el.getAttribute('width') || '')}"` : '';
+      const heightAttr = el.hasAttribute('height') ? ` height="${htmlEscape(el.getAttribute('height') || '')}"` : '';
+      if (widthAttr || heightAttr) {
+        return `<img alt="${htmlEscape(alt)}"${widthAttr}${heightAttr} src="${htmlEscape(src)}">`;
+      }
+      return `![${alt}](${src})`;
+    },
+
+    P(el) {
+      el.textContent = `${el.textContent}\n`;
+    },
+    BLOCKQUOTE(el) {
+      el.textContent = `${el.textContent.replace(/^/mg, '> ')}\n`;
+    },
+
+    OL(el) {
+      const preNewLine = ctx.listNestingLevel ? '\n' : '';
+      el.textContent = `${preNewLine}${el.textContent}\n`;
+    },
+    LI(el) {
+      const parent = el.parentNode;
+      const bullet = parent.tagName === 'OL' ? `1. ` : '* ';
+      const nestingIdentLevel = Math.max(0, ctx.listNestingLevel - 1);
+      el.textContent = `${' '.repeat(nestingIdentLevel * 4)}${bullet}${el.textContent}${ctx.elementIsLast ? '' : '\n'}`;
+      return el;
+    },
+    INPUT(el) {
+      return el.checked ? '[x] ' : '[ ] ';
+    },
+
+    CODE(el) {
+      const text = el.textContent;
+      if (el.parentNode && el.parentNode.tagName === 'PRE') {
+        el.textContent = `\`\`\`\n${text}\n\`\`\`\n`;
+        return el;
+      }
+      if (text.includes('`')) {
+        return `\`\` ${text} \`\``;
+      }
+      return `\`${text}\``;
+    },
+  };
+  processors['UL'] = processors.OL;
+  for (let level = 2; level <= 6; level++) {
+    processors[`H${level}`] = processors.H1;
+  }
+  return processors;
+}
+
+function processElement(ctx :ProcessorContext, processors: Processors, el: HTMLElement) {
+  if (el.hasAttribute('data-markdown-generated-content')) return el.textContent;
+  if (el.tagName === 'A' && el.children.length === 1 && el.children[0].tagName === 'IMG') {
+    return processElement(ctx, processors, el.children[0] as HTMLElement);
+  }
+
+  const isListContainer = el.tagName === 'OL' || el.tagName === 'UL';
+  if (isListContainer) ctx.listNestingLevel++;
+  for (let i = 0; i < el.children.length; i++) {
+    ctx.elementIsFirst = i === 0;
+    ctx.elementIsLast = i === el.children.length - 1;
+    processElement(ctx, processors, el.children[i] as HTMLElement);
+  }
+  if (isListContainer) ctx.listNestingLevel--;
+
+  if (processors[el.tagName]) {
+    const ret = processors[el.tagName](el);
+    if (ret && ret !== el) {
+      el.replaceWith(typeof ret === 'string' ? document.createTextNode(ret) : ret);
+    }
+  }
+}
+
+export function convertHtmlToMarkdown(el: HTMLElement): string {
+  const div = document.createElement('div');
+  div.append(el);
+  const ctx = {} as ProcessorContext;
+  ctx.listNestingLevel = 0;
+  processElement(ctx, prepareProcessors(ctx), el);
+  return div.textContent;
+}
-- 
cgit v1.2.3