turndown - npm Package Compare versions

Comparing version 6.0.0 to 7.0.0

dist/turndown.js

@@ -19,28 +19,53 @@ var TurndownService = (function () {
   var blockElements = [
-    'address', 'article', 'aside', 'audio', 'blockquote', 'body', 'canvas',
-    'center', 'dd', 'dir', 'div', 'dl', 'dt', 'fieldset', 'figcaption',
-    'figure', 'footer', 'form', 'frameset', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
-    'header', 'hgroup', 'hr', 'html', 'isindex', 'li', 'main', 'menu', 'nav',
-    'noframes', 'noscript', 'ol', 'output', 'p', 'pre', 'section', 'table',
-    'tbody', 'td', 'tfoot', 'th', 'thead', 'tr', 'ul'
+    'ADDRESS', 'ARTICLE', 'ASIDE', 'AUDIO', 'BLOCKQUOTE', 'BODY', 'CANVAS',
+    'CENTER', 'DD', 'DIR', 'DIV', 'DL', 'DT', 'FIELDSET', 'FIGCAPTION', 'FIGURE',
+    'FOOTER', 'FORM', 'FRAMESET', 'H1', 'H2', 'H3', 'H4', 'H5', 'H6', 'HEADER',
+    'HGROUP', 'HR', 'HTML', 'ISINDEX', 'LI', 'MAIN', 'MENU', 'NAV', 'NOFRAMES',
+    'NOSCRIPT', 'OL', 'OUTPUT', 'P', 'PRE', 'SECTION', 'TABLE', 'TBODY', 'TD',
+    'TFOOT', 'TH', 'THEAD', 'TR', 'UL'
   ];
 
   function isBlock (node) {
-    return blockElements.indexOf(node.nodeName.toLowerCase()) !== -1
+    return is(node, blockElements)
   }
 
   var voidElements = [
-    'area', 'base', 'br', 'col', 'command', 'embed', 'hr', 'img', 'input',
-    'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr'
+    'AREA', 'BASE', 'BR', 'COL', 'COMMAND', 'EMBED', 'HR', 'IMG', 'INPUT',
+    'KEYGEN', 'LINK', 'META', 'PARAM', 'SOURCE', 'TRACK', 'WBR'
   ];
 
   function isVoid (node) {
-    return voidElements.indexOf(node.nodeName.toLowerCase()) !== -1
+    return is(node, voidElements)
   }
 
-  var voidSelector = voidElements.join();
   function hasVoid (node) {
-    return node.querySelector && node.querySelector(voidSelector)
+    return has(node, voidElements)
   }
 
+  var meaningfulWhenBlankElements = [
+    'A', 'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TH', 'TD', 'IFRAME', 'SCRIPT',
+    'AUDIO', 'VIDEO'
+  ];
+
+  function isMeaningfulWhenBlank (node) {
+    return is(node, meaningfulWhenBlankElements)
+  }
+
+  function hasMeaningfulWhenBlank (node) {
+    return has(node, meaningfulWhenBlankElements)
+  }
+
+  function is (node, tagNames) {
+    return tagNames.indexOf(node.nodeName) >= 0
+  }
+
+  function has (node, tagNames) {
+    return (
+      node.getElementsByTagName &&
+      tagNames.some(function (tagName) {
+        return node.getElementsByTagName(tagName).length
+      })
+    )
+  }
+
   var rules = {};
@@ -155,3 +180,3 @@
     replacement: function (content, node, options) {
-      var className = node.firstChild.className || '';
+      var className = node.firstChild.getAttribute('class') || '';
       var language = (className.match(/language-(\S+)/) || [null, ''])[1];
@@ -200,3 +225,4 @@ var code = node.firstChild.textContent;
       var href = node.getAttribute('href');
-      var title = node.title ? ' "' + node.title + '"' : '';
+      var title = cleanAttribute(node.getAttribute('title'));
+      if (title) title = ' "' + title + '"';
       return '[' + content + '](' + href + title + ')'
@@ -217,3 +243,4 @@ }
       var href = node.getAttribute('href');
-      var title = node.title ? ' "' + node.title + '"' : '';
+      var title = cleanAttribute(node.getAttribute('title'));
+      if (title) title = ' "' + title + '"';
       var replacement;
@@ -300,5 +327,5 @@ var reference;
     replacement: function (content, node) {
-      var alt = node.alt || '';
+      var alt = cleanAttribute(node.getAttribute('alt'));
       var src = node.getAttribute('src') || '';
-      var title = node.title || '';
+      var title = cleanAttribute(node.getAttribute('title'));
       var titlePart = title ? ' "' + title + '"' : '';
@@ -309,2 +336,6 @@ return src ? '![' + alt + ']' + '(' + src + titlePart + ')' : ''
 
+  function cleanAttribute (attribute) {
+    return attribute ? attribute.replace(/(\n+\s*)+/g, '\n') : ''
+  }
+
   /**
@@ -623,6 +654,7 @@ * Manages a collection of rules used to convert HTML to Markdown
     return (
-      ['A', 'TH', 'TD', 'IFRAME', 'SCRIPT', 'AUDIO', 'VIDEO'].indexOf(node.nodeName) === -1 &&
+      !isVoid(node) &&
+      !isMeaningfulWhenBlank(node) &&
       /^\s*$/i.test(node.textContent) &&
-      !isVoid(node) &&
-      !hasVoid(node)
+      !hasVoid(node) &&
+      !hasMeaningfulWhenBlank(node)
     )
@@ -629,0 +661,0 @@ }

lib/turndown.browser.cjs.js

@@ -18,28 +18,53 @@ 'use strict';
 var blockElements = [
-  'address', 'article', 'aside', 'audio', 'blockquote', 'body', 'canvas',
-  'center', 'dd', 'dir', 'div', 'dl', 'dt', 'fieldset', 'figcaption',
-  'figure', 'footer', 'form', 'frameset', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
-  'header', 'hgroup', 'hr', 'html', 'isindex', 'li', 'main', 'menu', 'nav',
-  'noframes', 'noscript', 'ol', 'output', 'p', 'pre', 'section', 'table',
-  'tbody', 'td', 'tfoot', 'th', 'thead', 'tr', 'ul'
+  'ADDRESS', 'ARTICLE', 'ASIDE', 'AUDIO', 'BLOCKQUOTE', 'BODY', 'CANVAS',
+  'CENTER', 'DD', 'DIR', 'DIV', 'DL', 'DT', 'FIELDSET', 'FIGCAPTION', 'FIGURE',
+  'FOOTER', 'FORM', 'FRAMESET', 'H1', 'H2', 'H3', 'H4', 'H5', 'H6', 'HEADER',
+  'HGROUP', 'HR', 'HTML', 'ISINDEX', 'LI', 'MAIN', 'MENU', 'NAV', 'NOFRAMES',
+  'NOSCRIPT', 'OL', 'OUTPUT', 'P', 'PRE', 'SECTION', 'TABLE', 'TBODY', 'TD',
+  'TFOOT', 'TH', 'THEAD', 'TR', 'UL'
 ];
 
 function isBlock (node) {
-  return blockElements.indexOf(node.nodeName.toLowerCase()) !== -1
+  return is(node, blockElements)
 }
 
 var voidElements = [
-  'area', 'base', 'br', 'col', 'command', 'embed', 'hr', 'img', 'input',
-  'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr'
+  'AREA', 'BASE', 'BR', 'COL', 'COMMAND', 'EMBED', 'HR', 'IMG', 'INPUT',
+  'KEYGEN', 'LINK', 'META', 'PARAM', 'SOURCE', 'TRACK', 'WBR'
 ];
 
 function isVoid (node) {
-  return voidElements.indexOf(node.nodeName.toLowerCase()) !== -1
+  return is(node, voidElements)
 }
 
-var voidSelector = voidElements.join();
 function hasVoid (node) {
-  return node.querySelector && node.querySelector(voidSelector)
+  return has(node, voidElements)
 }
 
+var meaningfulWhenBlankElements = [
+  'A', 'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TH', 'TD', 'IFRAME', 'SCRIPT',
+  'AUDIO', 'VIDEO'
+];
+
+function isMeaningfulWhenBlank (node) {
+  return is(node, meaningfulWhenBlankElements)
+}
+
+function hasMeaningfulWhenBlank (node) {
+  return has(node, meaningfulWhenBlankElements)
+}
+
+function is (node, tagNames) {
+  return tagNames.indexOf(node.nodeName) >= 0
+}
+
+function has (node, tagNames) {
+  return (
+    node.getElementsByTagName &&
+    tagNames.some(function (tagName) {
+      return node.getElementsByTagName(tagName).length
+    })
+  )
+}
+
 var rules = {};
@@ -154,3 +179,3 @@
   replacement: function (content, node, options) {
-    var className = node.firstChild.className || '';
+    var className = node.firstChild.getAttribute('class') || '';
     var language = (className.match(/language-(\S+)/) || [null, ''])[1];
@@ -199,3 +224,4 @@ var code = node.firstChild.textContent;
     var href = node.getAttribute('href');
-    var title = node.title ? ' "' + node.title + '"' : '';
+    var title = cleanAttribute(node.getAttribute('title'));
+    if (title) title = ' "' + title + '"';
     return '[' + content + '](' + href + title + ')'
@@ -216,3 +242,4 @@ }
     var href = node.getAttribute('href');
-    var title = node.title ? ' "' + node.title + '"' : '';
+    var title = cleanAttribute(node.getAttribute('title'));
+    if (title) title = ' "' + title + '"';
     var replacement;
@@ -299,5 +326,5 @@ var reference;
   replacement: function (content, node) {
-    var alt = node.alt || '';
+    var alt = cleanAttribute(node.getAttribute('alt'));
     var src = node.getAttribute('src') || '';
-    var title = node.title || '';
+    var title = cleanAttribute(node.getAttribute('title'));
     var titlePart = title ? ' "' + title + '"' : '';
@@ -308,2 +335,6 @@ return src ? '![' + alt + ']' + '(' + src + titlePart + ')' : ''
 
+function cleanAttribute (attribute) {
+  return attribute ? attribute.replace(/(\n+\s*)+/g, '\n') : ''
+}
+
 /**
@@ -622,6 +653,7 @@ * Manages a collection of rules used to convert HTML to Markdown
   return (
-    ['A', 'TH', 'TD', 'IFRAME', 'SCRIPT', 'AUDIO', 'VIDEO'].indexOf(node.nodeName) === -1 &&
+    !isVoid(node) &&
+    !isMeaningfulWhenBlank(node) &&
     /^\s*$/i.test(node.textContent) &&
-    !isVoid(node) &&
-    !hasVoid(node)
+    !hasVoid(node) &&
+    !hasMeaningfulWhenBlank(node)
   )
@@ -628,0 +660,0 @@ }

lib/turndown.browser.es.js

@@ -16,28 +16,53 @@ function extend (destination) {
 var blockElements = [
-  'address', 'article', 'aside', 'audio', 'blockquote', 'body', 'canvas',
-  'center', 'dd', 'dir', 'div', 'dl', 'dt', 'fieldset', 'figcaption',
-  'figure', 'footer', 'form', 'frameset', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
-  'header', 'hgroup', 'hr', 'html', 'isindex', 'li', 'main', 'menu', 'nav',
-  'noframes', 'noscript', 'ol', 'output', 'p', 'pre', 'section', 'table',
-  'tbody', 'td', 'tfoot', 'th', 'thead', 'tr', 'ul'
+  'ADDRESS', 'ARTICLE', 'ASIDE', 'AUDIO', 'BLOCKQUOTE', 'BODY', 'CANVAS',
+  'CENTER', 'DD', 'DIR', 'DIV', 'DL', 'DT', 'FIELDSET', 'FIGCAPTION', 'FIGURE',
+  'FOOTER', 'FORM', 'FRAMESET', 'H1', 'H2', 'H3', 'H4', 'H5', 'H6', 'HEADER',
+  'HGROUP', 'HR', 'HTML', 'ISINDEX', 'LI', 'MAIN', 'MENU', 'NAV', 'NOFRAMES',
+  'NOSCRIPT', 'OL', 'OUTPUT', 'P', 'PRE', 'SECTION', 'TABLE', 'TBODY', 'TD',
+  'TFOOT', 'TH', 'THEAD', 'TR', 'UL'
 ];
 
 function isBlock (node) {
-  return blockElements.indexOf(node.nodeName.toLowerCase()) !== -1
+  return is(node, blockElements)
 }
 
 var voidElements = [
-  'area', 'base', 'br', 'col', 'command', 'embed', 'hr', 'img', 'input',
-  'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr'
+  'AREA', 'BASE', 'BR', 'COL', 'COMMAND', 'EMBED', 'HR', 'IMG', 'INPUT',
+  'KEYGEN', 'LINK', 'META', 'PARAM', 'SOURCE', 'TRACK', 'WBR'
 ];
 
 function isVoid (node) {
-  return voidElements.indexOf(node.nodeName.toLowerCase()) !== -1
+  return is(node, voidElements)
 }
 
-var voidSelector = voidElements.join();
 function hasVoid (node) {
-  return node.querySelector && node.querySelector(voidSelector)
+  return has(node, voidElements)
 }
 
+var meaningfulWhenBlankElements = [
+  'A', 'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TH', 'TD', 'IFRAME', 'SCRIPT',
+  'AUDIO', 'VIDEO'
+];
+
+function isMeaningfulWhenBlank (node) {
+  return is(node, meaningfulWhenBlankElements)
+}
+
+function hasMeaningfulWhenBlank (node) {
+  return has(node, meaningfulWhenBlankElements)
+}
+
+function is (node, tagNames) {
+  return tagNames.indexOf(node.nodeName) >= 0
+}
+
+function has (node, tagNames) {
+  return (
+    node.getElementsByTagName &&
+    tagNames.some(function (tagName) {
+      return node.getElementsByTagName(tagName).length
+    })
+  )
+}
+
 var rules = {};
@@ -152,3 +177,3 @@
   replacement: function (content, node, options) {
-    var className = node.firstChild.className || '';
+    var className = node.firstChild.getAttribute('class') || '';
     var language = (className.match(/language-(\S+)/) || [null, ''])[1];
@@ -197,3 +222,4 @@ var code = node.firstChild.textContent;
     var href = node.getAttribute('href');
-    var title = node.title ? ' "' + node.title + '"' : '';
+    var title = cleanAttribute(node.getAttribute('title'));
+    if (title) title = ' "' + title + '"';
     return '[' + content + '](' + href + title + ')'
@@ -214,3 +240,4 @@ }
     var href = node.getAttribute('href');
-    var title = node.title ? ' "' + node.title + '"' : '';
+    var title = cleanAttribute(node.getAttribute('title'));
+    if (title) title = ' "' + title + '"';
     var replacement;
@@ -297,5 +324,5 @@ var reference;
   replacement: function (content, node) {
-    var alt = node.alt || '';
+    var alt = cleanAttribute(node.getAttribute('alt'));
     var src = node.getAttribute('src') || '';
-    var title = node.title || '';
+    var title = cleanAttribute(node.getAttribute('title'));
     var titlePart = title ? ' "' + title + '"' : '';
@@ -306,2 +333,6 @@ return src ? '![' + alt + ']' + '(' + src + titlePart + ')' : ''
 
+function cleanAttribute (attribute) {
+  return attribute ? attribute.replace(/(\n+\s*)+/g, '\n') : ''
+}
+
 /**
@@ -620,6 +651,7 @@ * Manages a collection of rules used to convert HTML to Markdown
   return (
-    ['A', 'TH', 'TD', 'IFRAME', 'SCRIPT', 'AUDIO', 'VIDEO'].indexOf(node.nodeName) === -1 &&
+    !isVoid(node) &&
+    !isMeaningfulWhenBlank(node) &&
     /^\s*$/i.test(node.textContent) &&
-    !isVoid(node) &&
-    !hasVoid(node)
+    !hasVoid(node) &&
+    !hasMeaningfulWhenBlank(node)
   )
@@ -626,0 +658,0 @@ }

lib/turndown.browser.umd.js

@@ -22,28 +22,53 @@ (function (global, factory) {
   var blockElements = [
-    'address', 'article', 'aside', 'audio', 'blockquote', 'body', 'canvas',
-    'center', 'dd', 'dir', 'div', 'dl', 'dt', 'fieldset', 'figcaption',
-    'figure', 'footer', 'form', 'frameset', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
-    'header', 'hgroup', 'hr', 'html', 'isindex', 'li', 'main', 'menu', 'nav',
-    'noframes', 'noscript', 'ol', 'output', 'p', 'pre', 'section', 'table',
-    'tbody', 'td', 'tfoot', 'th', 'thead', 'tr', 'ul'
+    'ADDRESS', 'ARTICLE', 'ASIDE', 'AUDIO', 'BLOCKQUOTE', 'BODY', 'CANVAS',
+    'CENTER', 'DD', 'DIR', 'DIV', 'DL', 'DT', 'FIELDSET', 'FIGCAPTION', 'FIGURE',
+    'FOOTER', 'FORM', 'FRAMESET', 'H1', 'H2', 'H3', 'H4', 'H5', 'H6', 'HEADER',
+    'HGROUP', 'HR', 'HTML', 'ISINDEX', 'LI', 'MAIN', 'MENU', 'NAV', 'NOFRAMES',
+    'NOSCRIPT', 'OL', 'OUTPUT', 'P', 'PRE', 'SECTION', 'TABLE', 'TBODY', 'TD',
+    'TFOOT', 'TH', 'THEAD', 'TR', 'UL'
   ];
 
   function isBlock (node) {
-    return blockElements.indexOf(node.nodeName.toLowerCase()) !== -1
+    return is(node, blockElements)
   }
 
   var voidElements = [
-    'area', 'base', 'br', 'col', 'command', 'embed', 'hr', 'img', 'input',
-    'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr'
+    'AREA', 'BASE', 'BR', 'COL', 'COMMAND', 'EMBED', 'HR', 'IMG', 'INPUT',
+    'KEYGEN', 'LINK', 'META', 'PARAM', 'SOURCE', 'TRACK', 'WBR'
   ];
 
   function isVoid (node) {
-    return voidElements.indexOf(node.nodeName.toLowerCase()) !== -1
+    return is(node, voidElements)
   }
 
-  var voidSelector = voidElements.join();
   function hasVoid (node) {
-    return node.querySelector && node.querySelector(voidSelector)
+    return has(node, voidElements)
   }
 
+  var meaningfulWhenBlankElements = [
+    'A', 'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TH', 'TD', 'IFRAME', 'SCRIPT',
+    'AUDIO', 'VIDEO'
+  ];
+
+  function isMeaningfulWhenBlank (node) {
+    return is(node, meaningfulWhenBlankElements)
+  }
+
+  function hasMeaningfulWhenBlank (node) {
+    return has(node, meaningfulWhenBlankElements)
+  }
+
+  function is (node, tagNames) {
+    return tagNames.indexOf(node.nodeName) >= 0
+  }
+
+  function has (node, tagNames) {
+    return (
+      node.getElementsByTagName &&
+      tagNames.some(function (tagName) {
+        return node.getElementsByTagName(tagName).length
+      })
+    )
+  }
+
   var rules = {};
@@ -158,3 +183,3 @@
     replacement: function (content, node, options) {
-      var className = node.firstChild.className || '';
+      var className = node.firstChild.getAttribute('class') || '';
       var language = (className.match(/language-(\S+)/) || [null, ''])[1];
@@ -203,3 +228,4 @@ var code = node.firstChild.textContent;
       var href = node.getAttribute('href');
-      var title = node.title ? ' "' + node.title + '"' : '';
+      var title = cleanAttribute(node.getAttribute('title'));
+      if (title) title = ' "' + title + '"';
       return '[' + content + '](' + href + title + ')'
@@ -220,3 +246,4 @@ }
       var href = node.getAttribute('href');
-      var title = node.title ? ' "' + node.title + '"' : '';
+      var title = cleanAttribute(node.getAttribute('title'));
+      if (title) title = ' "' + title + '"';
       var replacement;
@@ -303,5 +330,5 @@ var reference;
     replacement: function (content, node) {
-      var alt = node.alt || '';
+      var alt = cleanAttribute(node.getAttribute('alt'));
       var src = node.getAttribute('src') || '';
-      var title = node.title || '';
+      var title = cleanAttribute(node.getAttribute('title'));
       var titlePart = title ? ' "' + title + '"' : '';
@@ -312,2 +339,6 @@ return src ? '![' + alt + ']' + '(' + src + titlePart + ')' : ''
 
+  function cleanAttribute (attribute) {
+    return attribute ? attribute.replace(/(\n+\s*)+/g, '\n') : ''
+  }
+
   /**
@@ -626,6 +657,7 @@ * Manages a collection of rules used to convert HTML to Markdown
     return (
-      ['A', 'TH', 'TD', 'IFRAME', 'SCRIPT', 'AUDIO', 'VIDEO'].indexOf(node.nodeName) === -1 &&
+      !isVoid(node) &&
+      !isMeaningfulWhenBlank(node) &&
       /^\s*$/i.test(node.textContent) &&
-      !isVoid(node) &&
-      !hasVoid(node)
+      !hasVoid(node) &&
+      !hasMeaningfulWhenBlank(node)
     )
@@ -632,0 +664,0 @@ }

lib/turndown.cjs.js

@@ -18,28 +18,53 @@ 'use strict';
 var blockElements = [
-  'address', 'article', 'aside', 'audio', 'blockquote', 'body', 'canvas',
-  'center', 'dd', 'dir', 'div', 'dl', 'dt', 'fieldset', 'figcaption',
-  'figure', 'footer', 'form', 'frameset', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
-  'header', 'hgroup', 'hr', 'html', 'isindex', 'li', 'main', 'menu', 'nav',
-  'noframes', 'noscript', 'ol', 'output', 'p', 'pre', 'section', 'table',
-  'tbody', 'td', 'tfoot', 'th', 'thead', 'tr', 'ul'
+  'ADDRESS', 'ARTICLE', 'ASIDE', 'AUDIO', 'BLOCKQUOTE', 'BODY', 'CANVAS',
+  'CENTER', 'DD', 'DIR', 'DIV', 'DL', 'DT', 'FIELDSET', 'FIGCAPTION', 'FIGURE',
+  'FOOTER', 'FORM', 'FRAMESET', 'H1', 'H2', 'H3', 'H4', 'H5', 'H6', 'HEADER',
+  'HGROUP', 'HR', 'HTML', 'ISINDEX', 'LI', 'MAIN', 'MENU', 'NAV', 'NOFRAMES',
+  'NOSCRIPT', 'OL', 'OUTPUT', 'P', 'PRE', 'SECTION', 'TABLE', 'TBODY', 'TD',
+  'TFOOT', 'TH', 'THEAD', 'TR', 'UL'
 ];
 
 function isBlock (node) {
-  return blockElements.indexOf(node.nodeName.toLowerCase()) !== -1
+  return is(node, blockElements)
 }
 
 var voidElements = [
-  'area', 'base', 'br', 'col', 'command', 'embed', 'hr', 'img', 'input',
-  'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr'
+  'AREA', 'BASE', 'BR', 'COL', 'COMMAND', 'EMBED', 'HR', 'IMG', 'INPUT',
+  'KEYGEN', 'LINK', 'META', 'PARAM', 'SOURCE', 'TRACK', 'WBR'
 ];
 
 function isVoid (node) {
-  return voidElements.indexOf(node.nodeName.toLowerCase()) !== -1
+  return is(node, voidElements)
 }
 
-var voidSelector = voidElements.join();
 function hasVoid (node) {
-  return node.querySelector && node.querySelector(voidSelector)
+  return has(node, voidElements)
 }
 
+var meaningfulWhenBlankElements = [
+  'A', 'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TH', 'TD', 'IFRAME', 'SCRIPT',
+  'AUDIO', 'VIDEO'
+];
+
+function isMeaningfulWhenBlank (node) {
+  return is(node, meaningfulWhenBlankElements)
+}
+
+function hasMeaningfulWhenBlank (node) {
+  return has(node, meaningfulWhenBlankElements)
+}
+
+function is (node, tagNames) {
+  return tagNames.indexOf(node.nodeName) >= 0
+}
+
+function has (node, tagNames) {
+  return (
+    node.getElementsByTagName &&
+    tagNames.some(function (tagName) {
+      return node.getElementsByTagName(tagName).length
+    })
+  )
+}
+
 var rules = {};
@@ -154,3 +179,3 @@
   replacement: function (content, node, options) {
-    var className = node.firstChild.className || '';
+    var className = node.firstChild.getAttribute('class') || '';
     var language = (className.match(/language-(\S+)/) || [null, ''])[1];
@@ -199,3 +224,4 @@ var code = node.firstChild.textContent;
     var href = node.getAttribute('href');
-    var title = node.title ? ' "' + node.title + '"' : '';
+    var title = cleanAttribute(node.getAttribute('title'));
+    if (title) title = ' "' + title + '"';
     return '[' + content + '](' + href + title + ')'
@@ -216,3 +242,4 @@ }
     var href = node.getAttribute('href');
-    var title = node.title ? ' "' + node.title + '"' : '';
+    var title = cleanAttribute(node.getAttribute('title'));
+    if (title) title = ' "' + title + '"';
     var replacement;
@@ -299,5 +326,5 @@ var reference;
   replacement: function (content, node) {
-    var alt = node.alt || '';
+    var alt = cleanAttribute(node.getAttribute('alt'));
     var src = node.getAttribute('src') || '';
-    var title = node.title || '';
+    var title = cleanAttribute(node.getAttribute('title'));
     var titlePart = title ? ' "' + title + '"' : '';
@@ -308,2 +335,6 @@ return src ? '![' + alt + ']' + '(' + src + titlePart + ')' : ''
 
+function cleanAttribute (attribute) {
+  return attribute ? attribute.replace(/(\n+\s*)+/g, '\n') : ''
+}
+
 /**
@@ -549,5 +580,5 @@ * Manages a collection of rules used to convert HTML to Markdown
   {
-    var JSDOM = require('jsdom').JSDOM;
+    var domino = require('domino');
     Parser.prototype.parseFromString = function (string) {
-      return new JSDOM(string).window.document
+      return domino.createDocument(string)
     };
@@ -599,6 +630,7 @@ }
   return (
-    ['A', 'TH', 'TD', 'IFRAME', 'SCRIPT', 'AUDIO', 'VIDEO'].indexOf(node.nodeName) === -1 &&
+    !isVoid(node) &&
+    !isMeaningfulWhenBlank(node) &&
     /^\s*$/i.test(node.textContent) &&
-    !isVoid(node) &&
-    !hasVoid(node)
+    !hasVoid(node) &&
+    !hasMeaningfulWhenBlank(node)
   )
@@ -605,0 +637,0 @@ }

lib/turndown.es.js

@@ -16,28 +16,53 @@ function extend (destination) {
 var blockElements = [
-  'address', 'article', 'aside', 'audio', 'blockquote', 'body', 'canvas',
-  'center', 'dd', 'dir', 'div', 'dl', 'dt', 'fieldset', 'figcaption',
-  'figure', 'footer', 'form', 'frameset', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
-  'header', 'hgroup', 'hr', 'html', 'isindex', 'li', 'main', 'menu', 'nav',
-  'noframes', 'noscript', 'ol', 'output', 'p', 'pre', 'section', 'table',
-  'tbody', 'td', 'tfoot', 'th', 'thead', 'tr', 'ul'
+  'ADDRESS', 'ARTICLE', 'ASIDE', 'AUDIO', 'BLOCKQUOTE', 'BODY', 'CANVAS',
+  'CENTER', 'DD', 'DIR', 'DIV', 'DL', 'DT', 'FIELDSET', 'FIGCAPTION', 'FIGURE',
+  'FOOTER', 'FORM', 'FRAMESET', 'H1', 'H2', 'H3', 'H4', 'H5', 'H6', 'HEADER',
+  'HGROUP', 'HR', 'HTML', 'ISINDEX', 'LI', 'MAIN', 'MENU', 'NAV', 'NOFRAMES',
+  'NOSCRIPT', 'OL', 'OUTPUT', 'P', 'PRE', 'SECTION', 'TABLE', 'TBODY', 'TD',
+  'TFOOT', 'TH', 'THEAD', 'TR', 'UL'
 ];
 
 function isBlock (node) {
-  return blockElements.indexOf(node.nodeName.toLowerCase()) !== -1
+  return is(node, blockElements)
 }
 
 var voidElements = [
-  'area', 'base', 'br', 'col', 'command', 'embed', 'hr', 'img', 'input',
-  'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr'
+  'AREA', 'BASE', 'BR', 'COL', 'COMMAND', 'EMBED', 'HR', 'IMG', 'INPUT',
+  'KEYGEN', 'LINK', 'META', 'PARAM', 'SOURCE', 'TRACK', 'WBR'
 ];
 
 function isVoid (node) {
-  return voidElements.indexOf(node.nodeName.toLowerCase()) !== -1
+  return is(node, voidElements)
 }
 
-var voidSelector = voidElements.join();
 function hasVoid (node) {
-  return node.querySelector && node.querySelector(voidSelector)
+  return has(node, voidElements)
 }
 
+var meaningfulWhenBlankElements = [
+  'A', 'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TH', 'TD', 'IFRAME', 'SCRIPT',
+  'AUDIO', 'VIDEO'
+];
+
+function isMeaningfulWhenBlank (node) {
+  return is(node, meaningfulWhenBlankElements)
+}
+
+function hasMeaningfulWhenBlank (node) {
+  return has(node, meaningfulWhenBlankElements)
+}
+
+function is (node, tagNames) {
+  return tagNames.indexOf(node.nodeName) >= 0
+}
+
+function has (node, tagNames) {
+  return (
+    node.getElementsByTagName &&
+    tagNames.some(function (tagName) {
+      return node.getElementsByTagName(tagName).length
+    })
+  )
+}
+
 var rules = {};
@@ -152,3 +177,3 @@
   replacement: function (content, node, options) {
-    var className = node.firstChild.className || '';
+    var className = node.firstChild.getAttribute('class') || '';
     var language = (className.match(/language-(\S+)/) || [null, ''])[1];
@@ -197,3 +222,4 @@ var code = node.firstChild.textContent;
     var href = node.getAttribute('href');
-    var title = node.title ? ' "' + node.title + '"' : '';
+    var title = cleanAttribute(node.getAttribute('title'));
+    if (title) title = ' "' + title + '"';
     return '[' + content + '](' + href + title + ')'
@@ -214,3 +240,4 @@ }
     var href = node.getAttribute('href');
-    var title = node.title ? ' "' + node.title + '"' : '';
+    var title = cleanAttribute(node.getAttribute('title'));
+    if (title) title = ' "' + title + '"';
     var replacement;
@@ -297,5 +324,5 @@ var reference;
   replacement: function (content, node) {
-    var alt = node.alt || '';
+    var alt = cleanAttribute(node.getAttribute('alt'));
     var src = node.getAttribute('src') || '';
-    var title = node.title || '';
+    var title = cleanAttribute(node.getAttribute('title'));
     var titlePart = title ? ' "' + title + '"' : '';
@@ -306,2 +333,6 @@ return src ? '![' + alt + ']' + '(' + src + titlePart + ')' : ''
 
+function cleanAttribute (attribute) {
+  return attribute ? attribute.replace(/(\n+\s*)+/g, '\n') : ''
+}
+
 /**
@@ -547,5 +578,5 @@ * Manages a collection of rules used to convert HTML to Markdown
   {
-    var JSDOM = require('jsdom').JSDOM;
+    var domino = require('domino');
     Parser.prototype.parseFromString = function (string) {
-      return new JSDOM(string).window.document
+      return domino.createDocument(string)
     };
@@ -597,6 +628,7 @@ }
   return (
-    ['A', 'TH', 'TD', 'IFRAME', 'SCRIPT', 'AUDIO', 'VIDEO'].indexOf(node.nodeName) === -1 &&
+    !isVoid(node) &&
+    !isMeaningfulWhenBlank(node) &&
     /^\s*$/i.test(node.textContent) &&
-    !isVoid(node) &&
-    !hasVoid(node)
+    !hasVoid(node) &&
+    !hasMeaningfulWhenBlank(node)
   )
@@ -603,0 +635,0 @@ }

lib/turndown.umd.js

@@ -22,28 +22,53 @@ (function (global, factory) {
   var blockElements = [
-    'address', 'article', 'aside', 'audio', 'blockquote', 'body', 'canvas',
-    'center', 'dd', 'dir', 'div', 'dl', 'dt', 'fieldset', 'figcaption',
-    'figure', 'footer', 'form', 'frameset', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
-    'header', 'hgroup', 'hr', 'html', 'isindex', 'li', 'main', 'menu', 'nav',
-    'noframes', 'noscript', 'ol', 'output', 'p', 'pre', 'section', 'table',
-    'tbody', 'td', 'tfoot', 'th', 'thead', 'tr', 'ul'
+    'ADDRESS', 'ARTICLE', 'ASIDE', 'AUDIO', 'BLOCKQUOTE', 'BODY', 'CANVAS',
+    'CENTER', 'DD', 'DIR', 'DIV', 'DL', 'DT', 'FIELDSET', 'FIGCAPTION', 'FIGURE',
+    'FOOTER', 'FORM', 'FRAMESET', 'H1', 'H2', 'H3', 'H4', 'H5', 'H6', 'HEADER',
+    'HGROUP', 'HR', 'HTML', 'ISINDEX', 'LI', 'MAIN', 'MENU', 'NAV', 'NOFRAMES',
+    'NOSCRIPT', 'OL', 'OUTPUT', 'P', 'PRE', 'SECTION', 'TABLE', 'TBODY', 'TD',
+    'TFOOT', 'TH', 'THEAD', 'TR', 'UL'
   ];
 
   function isBlock (node) {
-    return blockElements.indexOf(node.nodeName.toLowerCase()) !== -1
+    return is(node, blockElements)
   }
 
   var voidElements = [
-    'area', 'base', 'br', 'col', 'command', 'embed', 'hr', 'img', 'input',
-    'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr'
+    'AREA', 'BASE', 'BR', 'COL', 'COMMAND', 'EMBED', 'HR', 'IMG', 'INPUT',
+    'KEYGEN', 'LINK', 'META', 'PARAM', 'SOURCE', 'TRACK', 'WBR'
   ];
 
   function isVoid (node) {
-    return voidElements.indexOf(node.nodeName.toLowerCase()) !== -1
+    return is(node, voidElements)
   }
 
-  var voidSelector = voidElements.join();
   function hasVoid (node) {
-    return node.querySelector && node.querySelector(voidSelector)
+    return has(node, voidElements)
   }
 
+  var meaningfulWhenBlankElements = [
+    'A', 'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TH', 'TD', 'IFRAME', 'SCRIPT',
+    'AUDIO', 'VIDEO'
+  ];
+
+  function isMeaningfulWhenBlank (node) {
+    return is(node, meaningfulWhenBlankElements)
+  }
+
+  function hasMeaningfulWhenBlank (node) {
+    return has(node, meaningfulWhenBlankElements)
+  }
+
+  function is (node, tagNames) {
+    return tagNames.indexOf(node.nodeName) >= 0
+  }
+
+  function has (node, tagNames) {
+    return (
+      node.getElementsByTagName &&
+      tagNames.some(function (tagName) {
+        return node.getElementsByTagName(tagName).length
+      })
+    )
+  }
+
   var rules = {};
@@ -158,3 +183,3 @@
     replacement: function (content, node, options) {
-      var className = node.firstChild.className || '';
+      var className = node.firstChild.getAttribute('class') || '';
       var language = (className.match(/language-(\S+)/) || [null, ''])[1];
@@ -203,3 +228,4 @@ var code = node.firstChild.textContent;
       var href = node.getAttribute('href');
-      var title = node.title ? ' "' + node.title + '"' : '';
+      var title = cleanAttribute(node.getAttribute('title'));
+      if (title) title = ' "' + title + '"';
       return '[' + content + '](' + href + title + ')'
@@ -220,3 +246,4 @@ }
       var href = node.getAttribute('href');
-      var title = node.title ? ' "' + node.title + '"' : '';
+      var title = cleanAttribute(node.getAttribute('title'));
+      if (title) title = ' "' + title + '"';
       var replacement;
@@ -303,5 +330,5 @@ var reference;
     replacement: function (content, node) {
-      var alt = node.alt || '';
+      var alt = cleanAttribute(node.getAttribute('alt'));
       var src = node.getAttribute('src') || '';
-      var title = node.title || '';
+      var title = cleanAttribute(node.getAttribute('title'));
       var titlePart = title ? ' "' + title + '"' : '';
@@ -312,2 +339,6 @@ return src ? '![' + alt + ']' + '(' + src + titlePart + ')' : ''
 
+  function cleanAttribute (attribute) {
+    return attribute ? attribute.replace(/(\n+\s*)+/g, '\n') : ''
+  }
+
   /**
@@ -553,5 +584,5 @@ * Manages a collection of rules used to convert HTML to Markdown
     {
-      var JSDOM = require('jsdom').JSDOM;
+      var domino = require('domino');
       Parser.prototype.parseFromString = function (string) {
-        return new JSDOM(string).window.document
+        return domino.createDocument(string)
       };
@@ -603,6 +634,7 @@ }
     return (
-      ['A', 'TH', 'TD', 'IFRAME', 'SCRIPT', 'AUDIO', 'VIDEO'].indexOf(node.nodeName) === -1 &&
+      !isVoid(node) &&
+      !isMeaningfulWhenBlank(node) &&
       /^\s*$/i.test(node.textContent) &&
-      !isVoid(node) &&
-      !hasVoid(node)
+      !hasVoid(node) &&
+      !hasMeaningfulWhenBlank(node)
     )
@@ -609,0 +641,0 @@ }

package.json

 {
   "name": "turndown",
   "description": "A library that converts HTML to Markdown",
-  "version": "6.0.0",
+  "version": "7.0.0",
   "author": "Dom Christie",
@@ -10,6 +10,6 @@ "main": "lib/turndown.cjs.js",
   "browser": {
-    "jsdom": false
+    "domino": false
   },
   "dependencies": {
-    "jsdom": "^16.2.0"
+    "domino": "^2.1.6"
   },
@@ -16,0 +16,0 @@ "devDependencies": {

No alert changes

Improved metrics

Total package byte prevSize

347046

increased by92.68%

Number of package files

17

increased by70%

Lines of code

10956

increased by102.59%

Dependency changes

• + Addeddomino@^2.1.6

• + Addeddomino@2.1.6(transitive)

• - Removedjsdom@^16.2.0

• - Removed@tootallnate/once@1.1.2(transitive)

• - Removedabab@2.0.6(transitive)

• - Removedacorn@7.4.18.11.3(transitive)

• - Removedacorn-globals@6.0.0(transitive)

• - Removedacorn-walk@7.2.0(transitive)

• - Removedagent-base@6.0.2(transitive)

• - Removedasynckit@0.4.0(transitive)

• - Removedbrowser-process-hrtime@1.0.0(transitive)

• - Removedcombined-stream@1.0.8(transitive)

• - Removedcssom@0.3.80.4.4(transitive)

• - Removedcssstyle@2.3.0(transitive)

• - Removeddata-urls@2.0.0(transitive)

• - Removeddebug@4.3.5(transitive)

• - Removeddecimal.js@10.4.3(transitive)

• - Removeddelayed-stream@1.0.0(transitive)

• - Removeddomexception@2.0.1(transitive)

• - Removedescodegen@2.1.0(transitive)

• - Removedesprima@4.0.1(transitive)

• - Removedestraverse@5.3.0(transitive)

• - Removedesutils@2.0.3(transitive)

• - Removedform-data@3.0.1(transitive)

• - Removedhtml-encoding-sniffer@2.0.1(transitive)

• - Removedhttp-proxy-agent@4.0.1(transitive)

• - Removedhttps-proxy-agent@5.0.1(transitive)

• - Removediconv-lite@0.4.24(transitive)

• - Removedis-potential-custom-element-name@1.0.1(transitive)

• - Removedjsdom@16.7.0(transitive)

• - Removedlodash@4.17.21(transitive)

• - Removedmime-db@1.52.0(transitive)

• - Removedmime-types@2.1.35(transitive)

• - Removedms@2.1.2(transitive)

• - Removednwsapi@2.2.10(transitive)

• - Removedparse5@6.0.1(transitive)

• - Removedpsl@1.9.0(transitive)

• - Removedpunycode@2.3.1(transitive)

• - Removedquerystringify@2.2.0(transitive)

• - Removedrequires-port@1.0.0(transitive)

• - Removedsafer-buffer@2.1.2(transitive)

• - Removedsaxes@5.0.1(transitive)

• - Removedsource-map@0.6.1(transitive)

• - Removedsymbol-tree@3.2.4(transitive)

• - Removedtough-cookie@4.1.4(transitive)

• - Removedtr46@2.1.0(transitive)

• - Removeduniversalify@0.2.0(transitive)

• - Removedurl-parse@1.5.10(transitive)

• - Removedw3c-hr-time@1.0.2(transitive)

• - Removedw3c-xmlserializer@2.0.0(transitive)

• - Removedwebidl-conversions@5.0.06.1.0(transitive)

• - Removedwhatwg-encoding@1.0.5(transitive)

• - Removedwhatwg-mimetype@2.3.0(transitive)

• - Removedwhatwg-url@8.7.0(transitive)

• - Removedws@7.5.9(transitive)

• - Removedxml-name-validator@3.0.0(transitive)

• - Removedxmlchars@2.2.0(transitive)