feat(convert): add unsupportedTagsStrategy option that allows to es…

…cape unsupported tags Closes #5
skoropadas · Mar 15, 2023 · 4e78930 · 4e78930
1 parent f10c703
commit 4e78930
Show file tree

Hide file tree

Showing 5 changed files with 148 additions and 59 deletions.
diff --git a/README.md b/README.md
@@ -46,4 +46,39 @@ And simple text with \+ some \- symbols\.
 */
 ```
 
+## Possible options
+
+You can also add unsupported tags strategy as a second argument, which can be one of the following:
+
+- `escape` - escape unsupported symbols for unsupported tags
+- `remove` - remove unsupported tags
+- `keep` - ignore unsupported tags (default)
+
+```js
+const telegramifyMarkdown = require('telegramify-markdown');
+const markdown = `
+# Header
+
+> Blockquote
+
+<div>Text in div</div>
+`;
+
+telegramifyMarkdown(markdown, 'escape');
+/*
+*Header*
+
+\> Blockquote
+
+<div\>Text in div</div\>
+*/
+
+telegramifyMarkdown(markdown, 'remove');
+/*
+*Header*
+ */
+```
+
+sec
+
 [MIT Licence](LICENSE)
diff --git a/lib/convert.js b/lib/convert.js
@@ -7,13 +7,13 @@ const unified = require('unified');
 const { collectDefinitions, removeDefinitions } = require('./definitions');
 const createTelegramifyOptions = require('./telegramify');
 
-module.exports = (markdown, options) => {
+module.exports = (markdown, unsupportedTagsStrategy) => {
 	const definitions = {};
 
-	const telegramifyOptions = createTelegramifyOptions(definitions);
+	const telegramifyOptions = createTelegramifyOptions(definitions, unsupportedTagsStrategy);
 
 	return unified()
-		.use(parse, options)
+		.use(parse)
 		.use(gfm)
 		.use(removeComments)
 		.use(collectDefinitions, definitions)

diff --git a/lib/telegramify.js b/lib/telegramify.js
@@ -1,7 +1,7 @@
 const defaultHandlers = require('mdast-util-to-markdown/lib/handle');
 const phrasing = require('mdast-util-to-markdown/lib/util/container-phrasing');
 
-const {wrap, isURL, escapeSymbols} = require('./utils');
+const {wrap, isURL, escapeSymbols, processUnsupportedTags} = require('./utils');
 
 /**
  * Creates custom `mdast-util-to-markdown` handlers that tailor the output for
@@ -12,7 +12,7 @@ const {wrap, isURL, escapeSymbols} = require('./utils');
  *
  * @returns {import('mdast-util-to-markdown').Handlers}
  */
-const createHandlers = definitions => ({
+const createHandlers = (definitions, unsupportedTagsStrategy) => ({
 	heading: (node, _parent, context) => {
 		// make headers to be just *strong*
 		const marker = '*';
@@ -54,8 +54,7 @@ const createHandlers = definitions => ({
 		return wrap(value, marker);
 	},
 
-	list: (...args) => defaultHandlers.list(...args).replace(/^(\d+)./gm, '$1\\.')
-	,
+	list: (...args) => defaultHandlers.list(...args).replace(/^(\d+)./gm, '$1\\.'),
 
 	listItem: (...args) => defaultHandlers.listItem(...args).replace(/^\*/, '•'),
 
@@ -127,6 +126,11 @@ const createHandlers = definitions => ({
 
 		return escapeSymbols(text);
 	},
+
+	blockquote: (node, _parent, context) =>
+		processUnsupportedTags(defaultHandlers.blockquote(node, _parent, context), unsupportedTagsStrategy),
+	html: (node, _parent, context) =>
+		processUnsupportedTags(defaultHandlers.html(node, _parent, context), unsupportedTagsStrategy),
 });
 
 /**
@@ -138,9 +142,10 @@ const createHandlers = definitions => ({
  *
  * @returns {import('remark-stringify').RemarkStringifyOptions}
  */
-const createOptions = definitions => ({
+const createOptions = (definitions, unsupportedTagsStrategy) => ({
 	bullet: '*',
-	handlers: createHandlers(definitions),
+	tightDefinitions: true,
+	handlers: createHandlers(definitions, unsupportedTagsStrategy),
 });
 
 module.exports = createOptions;
diff --git a/lib/utils.js b/lib/utils.js
@@ -1,57 +1,74 @@
 const { URL } = require('url');
 
-module.exports = {
-	wrap(string, ...wrappers) {
-		return [
-			...wrappers,
-			string,
-			...wrappers.reverse(),
-		].join('');
-	},
+function wrap(string, ...wrappers) {
+	return [
+		...wrappers,
+		string,
+		...wrappers.reverse(),
+	].join('');
+}
 
-	isURL(string) {
-		try {
-			return Boolean(new URL(string));
-		} catch (error) {
-			return false;
-		}
-	},
+function isURL(string) {
+	try {
+		return Boolean(new URL(string));
+	} catch (error) {
+		return false;
+	}
+}
 
-	escapeSymbols(text, textType = 'text') {
-		if (!text) {
-			return text;
-		}
-		switch (textType) {
-			case 'code':
-				return text
-					.replace(/`/g, '\\`')
-					.replace(/\\/g, '\\\\')
-			case 'link':
-				return text
-					.replace(/\(/g, '\\(')
-					.replace(/\)/g, '\\)')
-					.replace(/\\/g, '\\\\')
-			default:
-				return text
-					.replace(/_/g, '\\_')
-					.replace(/\*/g, '\\*')
-					.replace(/\[/g, '\\[')
-					.replace(/]/g, '\\]')
-					.replace(/\(/g, '\\(')
-					.replace(/\)/g, '\\)')
-					.replace(/~/g, '\\~')
-					.replace(/`/g, '\\`')
-					.replace(/>/g, '\\>')
-					.replace(/#/g, '\\#')
-					.replace(/\+/g, '\\+')
-					.replace(/-/g, '\\-')
-					.replace(/=/g, '\\=')
-					.replace(/\|/g, '\\|')
-					.replace(/{/g, '\\{')
-					.replace(/}/g, '\\}')
-					.replace(/\./g, '\\.')
-					.replace(/!/g, '\\!');
+function escapeSymbols(text, textType = 'text') {
+	if (!text) {
+		return text;
+	}
+	switch (textType) {
+		case 'code':
+			return text
+				.replace(/`/g, '\\`')
+				.replace(/\\/g, '\\\\')
+		case 'link':
+			return text
+				.replace(/\(/g, '\\(')
+				.replace(/\)/g, '\\)')
+				.replace(/\\/g, '\\\\')
+		default:
+			return text
+				.replace(/_/g, '\\_')
+				.replace(/\*/g, '\\*')
+				.replace(/\[/g, '\\[')
+				.replace(/]/g, '\\]')
+				.replace(/\(/g, '\\(')
+				.replace(/\)/g, '\\)')
+				.replace(/~/g, '\\~')
+				.replace(/`/g, '\\`')
+				.replace(/>/g, '\\>')
+				.replace(/#/g, '\\#')
+				.replace(/\+/g, '\\+')
+				.replace(/-/g, '\\-')
+				.replace(/=/g, '\\=')
+				.replace(/\|/g, '\\|')
+				.replace(/{/g, '\\{')
+				.replace(/}/g, '\\}')
+				.replace(/\./g, '\\.')
+				.replace(/!/g, '\\!');
 
-		}
 	}
+}
+
+function processUnsupportedTags(content, strategy) {
+	switch (strategy) {
+		case 'escape':
+			return escapeSymbols(content);
+		case 'remove':
+			return '';
+		case 'keep':
+		default:
+			return content;
+	}
+}
+
+module.exports = {
+	wrap,
+	isURL,
+	escapeSymbols,
+	processUnsupportedTags,
 };
diff --git a/tests/convert.spec.js b/tests/convert.spec.js
@@ -230,4 +230,36 @@ describe('Test convert method', () => {
 
 		expect(convert(markdown)).toBe(tgMarkdown);
 	})
+
+	describe('escape unsupported tags', () => {
+		it('should escape blockquote', () => {
+			const markdown = '> test';
+			const tgMarkdown = '\\> test\n';
+
+			expect(convert(markdown, 'escape')).toBe(tgMarkdown);
+		});
+
+		it('should escape html', () => {
+			const markdown = '<div></div>';
+			const tgMarkdown = '<div\\></div\\>\n';
+
+			expect(convert(markdown, 'escape')).toBe(tgMarkdown);
+		});
+	})
+
+	describe('remove unsupported tags', () => {
+		it('should remove blockquote', () => {
+			const markdown = '> test';
+			const tgMarkdown = '';
+
+			expect(convert(markdown, 'remove')).toBe(tgMarkdown);
+		});
+
+		it('should remove html', () => {
+			const markdown = '<div></div>';
+			const tgMarkdown = '';
+
+			expect(convert(markdown, 'remove')).toBe(tgMarkdown);
+		});
+	})
 });