Fixed support for character class regular expressions in [attr=~regex…

…] selector
rorygibson · Jul 31, 2010 · 98f3cce · 98f3cce
1 parent ef1bbcb
commit 98f3cce
Show file tree

Hide file tree

Showing 3 changed files with 28 additions and 18 deletions.
diff --git a/CHANGES b/CHANGES
@@ -20,11 +20,11 @@ jsoup changelog
  * Added support for tag names with - and _ (<abc_foo>, <abc-foo>)
 
  * Improved HTML output format for empty elements and auto-detected self closing tags
- 
+
  * Implemented Node.ownerDocument DOM method
 
  * Fixed support for character class regular expressions in [attr=~regex] selector
- 
+
 *** Release 1.2.2 [2010-Jul-11]
 
  * Performance optimisation:

diff --git a/src/main/java/org/jsoup/select/Selector.java b/src/main/java/org/jsoup/select/Selector.java
@@ -153,7 +153,7 @@ private Elements findElements() {
             return byClass();
         } else if (tq.matchesWord()) {
             return byTag();
-        } else if (tq.matchChomp("[")) {
+        } else if (tq.matches("[")) {
             return byAttribute();
         } else if (tq.matchChomp("*")) {
             return allElements();
@@ -211,32 +211,34 @@ private Elements byTag() {
     }
 
     private Elements byAttribute() {
-        String key = tq.consumeToAny("=", "!=", "^=", "$=", "*=", "~=", "]"); // eq, not, start, end, contain, match, (no val)
+        TokenQueue cq = new TokenQueue(tq.chompBalanced('[', ']')); // content queue
+        String key = cq.consumeToAny("=", "!=", "^=", "$=", "*=", "~="); // eq, not, start, end, contain, match, (no val)
         Validate.notEmpty(key);
+        cq.consumeWhitespace();
 
-        if (tq.matchChomp("]")) {
+        if (cq.isEmpty()) {
             return key.startsWith("^") ? root.getElementsByAttributeStarting(key.substring(1)) : root.getElementsByAttribute(key);
         } else {
-            if (tq.matchChomp("="))
-                return root.getElementsByAttributeValue(key, tq.chompTo("]"));
+            if (cq.matchChomp("="))
+                return root.getElementsByAttributeValue(key, cq.remainder());
 
-            else if (tq.matchChomp("!="))
-                return root.getElementsByAttributeValueNot(key, tq.chompTo("]"));
+            else if (cq.matchChomp("!="))
+                return root.getElementsByAttributeValueNot(key, cq.remainder());
 
-            else if (tq.matchChomp("^="))
-                return root.getElementsByAttributeValueStarting(key, tq.chompTo("]"));
+            else if (cq.matchChomp("^="))
+                return root.getElementsByAttributeValueStarting(key, cq.remainder());
 
-            else if (tq.matchChomp("$="))
-                return root.getElementsByAttributeValueEnding(key, tq.chompTo("]"));
+            else if (cq.matchChomp("$="))
+                return root.getElementsByAttributeValueEnding(key, cq.remainder());
 
-            else if (tq.matchChomp("*="))
-                return root.getElementsByAttributeValueContaining(key, tq.chompTo("]"));
+            else if (cq.matchChomp("*="))
+                return root.getElementsByAttributeValueContaining(key, cq.remainder());
 
-            else if (tq.matchChomp("~="))
-                return root.getElementsByAttributeValueMatching(key, tq.chompTo("]"));
+            else if (cq.matchChomp("~="))
+                return root.getElementsByAttributeValueMatching(key, cq.remainder());
 
             else
-                throw new SelectorParseException("Could not parse attribute query '%s': unexpected token at '%s'", query, tq.remainder());
+                throw new SelectorParseException("Could not parse attribute query '%s': unexpected token at '%s'", query, cq.remainder());
         }
     }
 

diff --git a/src/test/java/org/jsoup/select/SelectorTest.java b/src/test/java/org/jsoup/select/SelectorTest.java
@@ -116,6 +116,14 @@ public class SelectorTest {
         assertEquals("3", imgs.get(2).id());
     }
 
+    @Test public void testByAttributeRegexCharacterClass() {
+        Document doc = Jsoup.parse("<p><img src=foo.png id=1><img src=bar.jpg id=2><img src=qux.JPEG id=3><img src=old.gif id=4></p>");
+        Elements imgs = doc.select("img[src~=[o]]");
+        assertEquals(2, imgs.size());
+        assertEquals("1", imgs.get(0).id());
+        assertEquals("4", imgs.get(1).id());
+    }
+
     @Test public void testAllElements() {
         String h = "<div><p>Hello</p><p><b>there</b></p></div>";
         Document doc = Jsoup.parse(h);