adding URI.iso8859() and URI.unicode() to switch base charsets (Issue #…

…10, mortenn)

adding .iso8859() and .unicode() to convert an URI's escape encoding

README.md

@@ -165,6 +165,9 @@ URI.js is published under the [MIT license](http://www.opensource.org/licenses/m
 
 ## Changelog ##
 
+* added URI.iso8859() and URI.unicode() to switch base charsets (Issue #10, mortenn)
+* added .iso8859() and .unicode() to convert an URI's escape encoding
+
 ### 1.3.1 ###
 
 * Updated Punycode.js to version 0.3.0

docs.html

@@ -85,6 +85,13 @@ <h1><a href="https://github.com/medialize/URI.js">URI.js</a></h1>
                 <li><a href="#normalize-hash">normalizeHash(), normalizeFragment()</a></li>
             </ul>
         </li>
+        <li>
+            Charsets / Encodings
+            <ul>
+                <li><a href="#iso8859">iso8859()</a></li>
+                <li><a href="#unicode">unicode()</a></li>
+            </ul>
+        </li>
         <li><a href="#readable">readable()</a></li>
         <li>
             Relative and Absolute URLs
@@ -110,14 +117,12 @@ <h1><a href="https://github.com/medialize/URI.js">URI.js</a></h1>
                 <li><a href="#static-buildAuthority">URI.buildAuthority()</a></li>
                 <li><a href="#static-buildHost">URI.buildHost()</a></li>
                 <li><a href="#static-buildQuery">URI.buildQuery()</a></li>
-
                 <li><a href="#static-addQuery">URI.addQuery()</a></li>
                 <li><a href="#static-removeQuery">URI.removeQuery()</a></li>
-
                 <li><a href="#static-commonPath">URI.commonPath()</a></li>
                 <li><a href="#static-withinString">URI.withinString()</a></li>
-    		<li><a href="#static-iso8859">URI.iso8859()</a></li>
-    		<li><a href="#static-unicode">URI.unicode()</a></li>
+                <li><a href="#static-iso8859">URI.iso8859()</a></li>
+                <li><a href="#static-unicode">URI.unicode()</a></li>
             </ul>
         </li>
     </ul>
@@ -501,6 +506,22 @@ <h3 id="normalize-hash">normalizeHash(), normalizeFragment()</h3>
 uri.normalizeHash(); // returns the URI instance for chaining
 // uri == "http://example.org/bar/world.xml"</pre>
 
+
+    <h2 id="charsets">Charsets / Encodings</h2>
+
+    <h3 id="#iso8859">iso8859()</h3>
+    <p>.iso8859() converts unicode-encoded escape sequences to ISO8859-encoded escape sequences. It does this by calling <a href="#normalize">.normalize()</a> internally.</p>
+    <pre class="prettyprint lang-js">var uri = new URI("/%C3%A4.html");
+uri.iso8859(); // returns the URI instance for chaining
+// uri == "/%E4.html"</pre>
+    <p>NOTE: You can make URI work with ISO8859 encoding by default by calling <a href="#static-iso8859">URI.iso8859()</a>.
+
+    <h3 id="#unicode">unicode()</h3>
+    <p>.unicode() converts ISO8859-encoded escape sequences to unicode-encoded escape sequences. It does this by calling <a href="#normalize">.normalize()</a> internally.</p>
+    <pre class="prettyprint lang-js">var uri = new URI("/%E4.html");
+uri.unicode(); // returns the URI instance for chaining
+// uri == "/%C3%A4.html"</pre>
+
 
     <h2 id="formatting">Formatting URLs</h2>
 
@@ -530,8 +551,7 @@ <h3 id="absoluteto">absoluteTo()</h3>
 var relUri = uri.absoluteTo("/relative/sub/foo/sub/file"); // returns a new URI instance
 // relUri == "/relative/path"</pre>
     <p>.relativeTo() and .absoluteTo() reverse each other.</p>
-
-
+
 
     <h2 id="comparison">Comparing URLs</h2>
 
@@ -760,7 +780,7 @@ <h3 id="static-withinString">URI.withinString()</h3>
 </pre>
 
     <h3 id="static-iso8859">URI.iso8859()</h3>
-    <p>URI.iso8859() tells URI.js to use the older escape/unescape methods, for backwards compatibility with older platforms.</p>
+    <p>URI.iso8859() tells URI.js to use the older escape/unescape methods, for backwards compatibility with non-unicode platforms.</p>
     <pre class="prettyprint lang-js">URI.iso8859();
 
 var uri = new URI("http://example.org/foo/æ.html");

src/URI.js

@@ -68,16 +68,6 @@ var URI = function(url, base) {
     };
 var p = URI.prototype;
 
-URI.iso8859 = function() {
-	URI.encode = escape;
-	URI.decode = unescape;
-}
-
-URI.unicode = function() {
-	URI.encode = encodeURIComponent;
-	URI.decode = decodeURIComponent;
-}
-
 // static properties
 URI.idn_expression = /[^a-z0-9\.-]/i;
 URI.punycode_expression = /(xn--)/i;
@@ -103,6 +93,14 @@ URI.invalid_hostname_characters = /[^a-zA-Z0-9\.-]/;
 // encoding / decoding according to RFC3986
 URI.encode = encodeURIComponent;
 URI.decode = decodeURIComponent;
+URI.iso8859 = function() {
+    URI.encode = escape;
+    URI.decode = unescape;
+};
+URI.unicode = function() {
+    URI.encode = encodeURIComponent;
+    URI.decode = decodeURIComponent;
+};
 URI.characters = {
     pathname: {
         encode: {
@@ -160,7 +158,7 @@ var _parts = {'encode':'encode', 'decode':'decode'},
 for (_part in _parts) {
     URI[_part + "PathSegment"] = (function(_part){
         return function(string) {
-            return window[_part + 'URIComponent'](string + "").replace(URI.characters.pathname[_part].expression, function(c) {
+            return URI[_part](string + "").replace(URI.characters.pathname[_part].expression, function(c) {
                 return URI.characters.pathname[_part].map[c];
             });
         };
@@ -1079,6 +1077,32 @@ p.normalizeFragment = function(build) {
 p.normalizeSearch = p.normalizeQuery;
 p.normalizeHash = p.normalizeFragment;
 
+p.iso8859 = function() {
+    // expect unicode input, iso8859 output
+    var e = URI.encode,
+        d = URI.decode;
+
+    URI.encode = escape;
+    URI.decode = decodeURIComponent;
+    this.normalize();
+    URI.encode = e;
+    URI.decode = d;
+    return this;
+};
+
+p.unicode = function() {
+    // expect iso8859 input, unicode output
+    var e = URI.encode,
+        d = URI.decode;
+
+    URI.encode = encodeURIComponent;
+    URI.decode = unescape;
+    this.normalize();
+    URI.encode = e;
+    URI.decode = d;
+    return this;
+};
+
 p.readable = function() {
     var uri = new URI(this);
     // removing username, password, because they shouldn't be displayed according to RFC 3986

test/test.js

@@ -737,3 +737,30 @@ test("equals", function() {
     }
 });
 
+module("Charset");
+test("iso8859", function() {
+    var u = new URI("/ä.html");
+    u.normalizePath();
+    equal(u.path(), "/%C3%A4.html", 'Unicode');
+
+    URI.iso8859();
+    u = new URI("/ä.html");
+    u.normalizePath();
+    equal(u.path(), "/%E4.html", 'ISO8859');
+    u.path('/ö.html');
+    equal(u.path(), "/%F6.html", 'ISO8859');
+
+    URI.unicode();
+    u = new URI("/ä.html");
+    u.normalizePath();
+    equal(u.path(), "/%C3%A4.html", 'Unicode again');
+
+    u = new URI("/ä.html");
+    u.normalizePath();
+    equal(u.path(), "/%C3%A4.html", 'convert unicode start');
+    u.iso8859();
+    equal(u.path(), "/%E4.html", 'convert iso8859');
+    u.unicode();
+    equal(u.path(), "/%C3%A4.html", 'convert unicode');
+});
+