Allow extra spaces and linebreaks in DFN autolinking

Fixes #93.

lib/autolinker.js

@@ -35,10 +35,10 @@ function linkClause(spec, clause, replacerCache) {
     autolinkmap = {};
 
     spec.biblio.inScopeByType(clause.namespace, 'term')
-      .forEach(entry => autolinkmap[entry.key.toLowerCase()] = entry);
+      .forEach(entry => autolinkmap[narrowSpace(entry.key.toLowerCase())] = entry);
 
     spec.biblio.inScopeByType(clause.namespace, 'op')
-      .forEach(entry => autolinkmap[entry.key.toLowerCase()] = entry);
+      .forEach(entry => autolinkmap[narrowSpace(entry.key.toLowerCase())] = entry);
 
     clauseReplacer = new RegExp(Object.keys(autolinkmap)
       .sort(function (a, b) { return b.length - a.length; })
@@ -48,9 +48,13 @@ function linkClause(spec, clause, replacerCache) {
 
         if (entry.type === 'term') {
           if (isCommonTerm(key)) {
-            return '\\b' + key + '\\b(?!\\.\\w|%%|\\]\\])';
+            return '\\b' +
+                   widenSpace(key) +
+                   '\\b(?!\\.\\w|%%|\\]\\])';
           } else if (key[0].match(/[A-Za-z0-9]/)) {
-            return '\\b' + caseInsensitiveRegExp(key) + '\\b(?!\\.\\w|%%|\\]\\])';
+            return '\\b' +
+                   widenSpace(caseInsensitiveRegExp(key)) +
+                   '\\b(?!\\.\\w|%%|\\]\\])';
           } else {
             return key;
           }
@@ -86,7 +90,7 @@ function autolink(replacer, autolinkmap, spec, node, parentId, allowSameId) {
     const template = spec.doc.createElement('template');
     const content = escape(node.textContent);
     const autolinked = content.replace(replacer, match => {
-      const entry = autolinkmap[match.toLowerCase()];
+      const entry = autolinkmap[narrowSpace(match.toLowerCase())];
       if (!entry) {
         return match;
       }
@@ -132,3 +136,16 @@ function caseInsensitiveRegExp(str) {
 
   return `[${lower}${upper}]${str.slice(1)}`;
 }
+
+// given a regexp string, returns a regexp string where each space can be
+// many spaces or line breaks.
+function widenSpace(str) {
+  return str.replace(/\s+/g, '[\\s\\r\\n]+');
+}
+
+// given a regexp string, returns a regexp string where multiple spaces
+// or linebreaks can only be a single space
+function narrowSpace(str) {
+  return str.replace(/[\s\r\n]+/g, ' ');
+}
+

test/autolinking.html

@@ -10,11 +10,14 @@ <h1>Autolinking</h1>
   <p><dfn>Lowercase</dfn></p>
   <p><dfn>strict mode</dfn></p>
   <p><dfn>%Percent%</dfn></p>
+  <p><dfn>extra      spaces</dfn></p>
 </emu-clause>
 <emu-clause id="sec-bar">
   <h1>Autolinking 2</h1>
   <p>lowercase should not autolink. But Lowercase should. But not LowerCase.</p>
-  <p>Strict mode shoud link. But Strict Mode should not.</p>
+  <p>Strict mode shoud link. But Strict Mode should not. Also, strict
+  mode can be wrapped across lines and contain extra whitespace.</p>
+  <p>extra spaces in a dfn should be narrowed to one space.</p>
   <p>%Percent% should autolink.</p>
   <p>Vars to dfns should be vars not dfns: _Lowercase_.</p>
   <p>Also, no autolinks in <a href="#">anchors: Lowercase</a>.</p>

test/autolinking.html.baseline

@@ -8,11 +8,14 @@
   <p><dfn>Lowercase</dfn></p>
   <p><dfn>strict mode</dfn></p>
   <p><dfn>%Percent%</dfn></p>
+  <p><dfn>extra      spaces</dfn></p>
 </emu-clause>
 <emu-clause id="sec-bar">
   <h1><span class="secnum">2</span>Autolinking 2<span class="utils"><span class="anchor"><a href="#sec-bar">#</a></span></span></h1>
   <p>lowercase should not autolink. But <emu-xref href="#sec-foo"><a href="#sec-foo">Lowercase</a></emu-xref> should. But not LowerCase.</p>
-  <p><emu-xref href="#sec-foo"><a href="#sec-foo">Strict mode</a></emu-xref> shoud link. But Strict Mode should not.</p>
+  <p><emu-xref href="#sec-foo"><a href="#sec-foo">Strict mode</a></emu-xref> shoud link. But Strict Mode should not. Also, <emu-xref href="#sec-foo"><a href="#sec-foo">strict
+  mode</a></emu-xref> can be wrapped across lines and contain extra whitespace.</p>
+  <p><emu-xref href="#sec-foo"><a href="#sec-foo">extra spaces</a></emu-xref> in a dfn should be narrowed to one space.</p>
   <p><emu-xref href="#sec-foo"><a href="#sec-foo">%Percent%</a></emu-xref> should autolink.</p>
   <p>Vars to dfns should be vars not dfns: <var>Lowercase</var>.</p>
   <p>Also, no autolinks in  <a href="#">anchors: Lowercase</a>.</p>