remove-markdown - npm Package Compare versions

Comparing version 0.3.0 to 0.5.0

index.js

@@ -7,2 +7,5 @@ module.exports = function(md, options) {
   options.useImgAltText = options.hasOwnProperty('useImgAltText') ? options.useImgAltText : true;
+  options.abbr = options.hasOwnProperty('abbr') ? options.abbr : false;
+  options.replaceLinksWithURL = options.hasOwnProperty('replaceLinksWithURL') ? options.replaceLinksWithURL : false;
+  options.htmlTagsToSkip = options.hasOwnProperty('htmlTagsToSkip') ? options.htmlTagsToSkip : [];
 
@@ -12,3 +15,3 @@ var output = md || '';
   // Remove horizontal rules (stripListHeaders conflict with this rule, which is why it has been moved to the top)
-  output = output.replace(/^(-\s*?|\*\s*?|_\s*?){3,}\s*$/gm, '');
+  output = output.replace(/^(-\s*?|\*\s*?|_\s*?){3,}\s*/gm, '');
 
@@ -24,3 +27,3 @@ try {
       output = output
-        // Header
+      // Header
         .replace(/\n={2,}/g, '\n')
@@ -34,5 +37,27 @@ // Fenced codeblocks
     }
+    if (options.abbr) {
+      // Remove abbreviations
+      output = output.replace(/\*\[.*\]:.*\n/, '');
+    }
     output = output
+    // Remove HTML tags
+      .replace(/<[^>]*>/g, '')
+
+    var htmlReplaceRegex = new RegExp('<[^>]*>', 'g');
+    if (options.htmlTagsToSkip.length > 0) {
+      // Using negative lookahead. Eg. (?!sup|sub) will not match 'sup' and 'sub' tags.
+      var joinedHtmlTagsToSkip = '(?!' + options.htmlTagsToSkip.join("|") + ')';
+
+      // Adding the lookahead literal with the default regex for html. Eg./<(?!sup|sub)[^>]*>/ig
+      htmlReplaceRegex = new RegExp(
+          '<' +
+          joinedHtmlTagsToSkip +
+          '[^>]*>', 
+          'ig'
+      );
+    }
+
+    output = output
       // Remove HTML tags
-      .replace(/<[^>]*>/g, '')
+      .replace(htmlReplaceRegex, '')
       // Remove setext-style headers
@@ -46,12 +71,16 @@ .replace(/^[=\-]{2,}\s*$/g, '')
       // Remove inline links
-      .replace(/\[(.*?)\][\[\(].*?[\]\)]/g, '$1')
+      .replace(/\[([^\]]*?)\][\[\(].*?[\]\)]/g, options.replaceLinksWithURL ? '$2' : '$1')
       // Remove blockquotes
-      .replace(/^\s{0,3}>\s?/g, '')
+      .replace(/^\s{0,3}>\s?/gm, '')
+      // .replace(/(^|\n)\s{0,3}>\s?/g, '\n\n')
       // Remove reference-style links?
       .replace(/^\s{1,2}\[(.*?)\]: (\S+)( ".*?")?\s*$/g, '')
       // Remove atx-style headers
-      .replace(/^(\n)?\s{0,}#{1,6}\s+| {0,}(\n)?\s{0,}#{0,} {0,}(\n)?\s{0,}$/gm, '$1$2$3')
-      // Remove emphasis (repeat the line to remove double emphasis)
-      .replace(/([\*_]{1,3})(\S.*?\S{0,1})\1/g, '$2')
-      .replace(/([\*_]{1,3})(\S.*?\S{0,1})\1/g, '$2')
+      .replace(/^(\n)?\s{0,}#{1,6}\s+| {0,}(\n)?\s{0,}#{0,} #{0,}(\n)?\s{0,}$/gm, '$1$2$3')
+      // Remove * emphasis
+      .replace(/([\*]+)(\S)(.*?\S)??\1/g, '$2$3')
+      // Remove _ emphasis. Unlike *, _ emphasis gets rendered only if 
+      //   1. Either there is a whitespace character before opening _ and after closing _.
+      //   2. Or _ is at the start/end of the string.
+      .replace(/(^|\W)([_]+)(\S)(.*?\S)??\2($|\W)/g, '$1$3$4$5')
       // Remove code blocks
@@ -61,4 +90,8 @@ .replace(/(`{3,})(.*?)\1/gm, '$2')
       .replace(/`(.+?)`/g, '$1')
-      // Replace two or more newlines with exactly two? Not entirely sure this belongs here...
-      .replace(/\n{2,}/g, '\n\n');
+      // // Replace two or more newlines with exactly two? Not entirely sure this belongs here...
+      // .replace(/\n{2,}/g, '\n\n')
+      // // Remove newlines in a paragraph
+      // .replace(/(\S+)\n\s*(\S+)/g, '$1 $2')
+      // Replace strike through
+      .replace(/~(.*?)~/g, '$1');
   } catch(e) {
@@ -65,0 +98,0 @@ console.error(e);

package.json

 {
   "name": "remove-markdown",
-  "version": "0.3.0",
+  "version": "0.5.0",
   "description": "Remove Markdown formatting from text",
@@ -5,0 +5,0 @@ "main": "index.js",

test/remove-markdown.js

@@ -72,2 +72,14 @@ 'use strict';
 
+    it('should remove emphasis only if there is no space between word and emphasis characters.', function () {
+      const string = 'There should be no _space_, *before* *closing * _ephasis character _.';
+      const expected = 'There should be no space, before *closing * _ephasis character _.';
+      expect(removeMd(string)).to.equal(expected);
+    });
+
+    it('should remove "_" emphasis only if there is space before opening and after closing emphasis characters.', function () {
+      const string = '._Spaces_ _ before_ and _after _ emphasised character results in no emphasis.';
+      const expected = '.Spaces _ before_ and _after _ emphasised character results in no emphasis.';
+      expect(removeMd(string)).to.equal(expected);
+    });
+
     it('should remove double emphasis', function () {
@@ -113,14 +125,27 @@ const string = '**this sentence has __double styling__**';
     });
+    
+    it('should remove blockquotes over multiple lines', function () {
+      const string = '> I am a blockquote firstline  \n>I am a blockquote secondline';
+      const expected = 'I am a blockquote firstline\nI am a blockquote secondline';
+      expect(removeMd(string)).to.equal(expected);
+    });
 
+    // it('should remove blockquotes following other content', function () {
+    //   const string = '## A headline\n\nA paragraph of text\n\n> I am a blockquote';
+    //   const expected = 'A headline\n\nA paragraph of text\n\nI am a blockquote';
+
+    //   expect(removeMd(string)).to.equal(expected);
+    // });
+
     it('should not remove greater than signs', function () {
       var tests = [
-          { string: '100 > 0', expected: '100 > 0' },
-          { string: '100 >= 0', expected: '100 >= 0' },
-          { string: '100>0', expected: '100>0' },
-          { string: '> 100 > 0', expected: '100 > 0' },
-          { string: '1 < 100', expected: '1 < 100' },
-          { string: '1 <= 100', expected: '1 <= 100' },
+        { string: '100 > 0', expected: '100 > 0' },
+        { string: '100 >= 0', expected: '100 >= 0' },
+        { string: '100>0', expected: '100>0' },
+        { string: '> 100 > 0', expected: '100 > 0' },
+        { string: '1 < 100', expected: '1 < 100' },
+        { string: '1 <= 100', expected: '1 <= 100' },
       ];
       tests.forEach(function (test) {
-          expect(removeMd(test.string)).to.equal(test.expected);
+        expect(removeMd(test.string)).to.equal(test.expected);
       });
@@ -146,3 +171,20 @@ });
     });
+
+    it('should not strip paragraphs without content', function() {
+      const paragraph = '\n#This paragraph\n##This paragraph#';
+      const expected = paragraph;
+      expect(removeMd(paragraph)).to.equal(expected);
+    });
+
+    it('should not trigger ReDoS with atx-headers', function () {
+      const start = Date.now();
+
+      const paragraph = '\n## This is a long "'+' '.repeat(200)+'" heading ##\n';
+      const expected = /\nThis is a long " {200}" heading\n/;
+      expect(removeMd(paragraph)).to.match(expected);
+
+      const duration = Date.now()-start;
+      expect(duration).to.be.lt(500);
+    });
   });
 });

No alert changes

Improved metrics

Total package byte prevSize

20675

increased by19.5%

Lines of code

247

increased by34.24%

Worsened metrics

Number of package files

6

decreased by-14.29%

No dependency changes