add fuzzy mode

1 files changed, 49 insertions, 18 deletions

diff --git a/lib/utils.js b/lib/utils.js
index 8aada3a1..922e7db3 100644
--- a/lib/utils.js
+++ b/lib/utils.js
@@ -21,17 +21,26 @@ var utils = {
   },
 
   /**
-   * Creates a search URL from the given :query.
+   * Generates a unique ID
    */
-  createSearchUrl: function(query) {
-    return "http://www.google.com/search?q=" + query;
+  createUniqueId: (function() {
+    id = 0;
+    return function() { return ++id; };
+  })(),
+
+  /**
+   * Completes a partial URL (without scheme)
+   */
+  createFullUrl: function(partialUrl) {
+    if (!/^[a-z]{3,}:\/\//.test(partialUrl))
+      partialUrl = 'http://' + partialUrl;
+    return partialUrl
   },
 
   /**
-   * Tries to convert :str into a valid URL.
-   * We don't bother with escaping characters, however, as Chrome will do that for us.
+   * Tries to detect, whether :str is a valid URL.
    */
-  ensureUrl: function(str) {
+  isUrl: function(str) {
     // more or less RFC compliant URL host part parsing. This should be sufficient
     // for our needs
     var urlRegex = new RegExp(
@@ -55,38 +64,60 @@ var utils = {
 
     // it starts with a scheme, so it's definitely an URL
     if (/^[a-z]{3,}:\/\//.test(str))
-      return str;
-    var strWithScheme = 'http://' + str;
+      return true;
 
-    // definitely not a valid URL; treat as search query
+    // spaces => definitely not a valid URL
     if (str.indexOf(' ') >= 0)
-      return utils.createSearchUrl(str);
+      return false;
 
     // assuming that this is an URL, try to parse it into its meaningful parts. If matching fails, we're
     // pretty sure that we don't have some kind of URL here.
     var match = urlRegex.exec(str.split('/')[0]);
     if (!match)
-      return utils.createSearchUrl(str);
+      return false;
     var hostname = match[3];
 
     // allow known special host names
     if (specialHostNames.indexOf(hostname) >= 0)
-      return strWithScheme;
+      return true;
 
     // allow IPv6 addresses (need to be wrapped in brackets, as required by RFC).  It is sufficient to check
     // for a colon here, as the regex wouldn't match colons in the host name unless it's an v6 address
     if (hostname.indexOf(':') >= 0)
-      return strWithScheme;
+      return true;
 
     // at this point we have to make a decision. As a heuristic, we check if the input has dots in it. If
-    // yes, and if the last part could be a TLD, treat it as an URL
+    // yes, and if the last part could be a TLD, treat it as an URL.
     var dottedParts = hostname.split('.');
     var lastPart = dottedParts[dottedParts.length-1];
-    if (dottedParts.length > 1 && (lastPart.length <= 3 || longTlds.indexOf(lastPart) >= 0))
-      return strWithScheme;
+    if (dottedParts.length > 1 && ((lastPart.length >= 2 && lastPart.length <= 3)
+                                   || longTlds.indexOf(lastPart) >= 0))
+      return true;
+
+    // also allow IPv4 addresses
+    if (/^(\d{1,3}\.){3}\d{1,3}$/.test(hostname))
+      return true;
+
+    // fallback: no URL
+    return false
+  },
 
-    // fallback: use as search query
-    return utils.createSearchUrl(str);
+  /**
+   * Creates a search URL from the given :query.
+   */
+  createSearchUrl: function(query) {
+    return "http://www.google.com/search?q=" + query;
+  },
+
+  /**
+   * Tries to convert :str into a valid URL.
+   * We don't bother with escaping characters, however, as Chrome will do that for us.
+   */
+  ensureUrl: function(str) {
+    if (utils.isUrl(str))
+      return utils.createFullUrl(str);
+    else
+      return utils.createSearchUrl(str);
   },
 
 };