implement validations of pagination URLs

javatests/org/chromium/distiller/PagePatternTest.java

+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+package org.chromium.distiller;
+
+public class PagePatternTest extends DomDistillerJsTestCase {
+
+    public void testIsPagingUrl() {
+        assertTrue(isPagingUrl("http://www.foo.com/a/b?queryA=v1&queryB=4&queryC=v3",
+                               "http://www.foo.com/a/b?queryA=v1&queryB=[*!]&queryC=v3"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/b?queryA=v1&queryC=v3",
+                               "http://www.foo.com/a/b?queryA=v1&queryB=[*!]&queryC=v3"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/b?queryB=2&queryC=v3",
+                               "http://www.foo.com/a/b?queryB=[*!]&queryC=v3"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/b?queryC=v3",
+                               "http://www.foo.com/a/b?queryB=[*!]&queryC=v3"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/b", "http://www.foo.com/a/b?page=[*!]"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/b?page=3",
+                               "http://www.foo.com/a/b?page=[*!]"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/b/", "http://www.foo.com/a/b?page=[*!]"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/b.htm", "http://www.foo.com/a/b?page=[*!]"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/b.html", "http://www.foo.com/a/b?page=[*!]"));
+        assertFalse(isPagingUrl("http://www.foo.com/a/b?queryA=v1&queryC=v3",
+                                "http://www.foo.com/a/b?queryB=[*!]&queryC=v3"));
+        assertFalse(isPagingUrl("http://www.foo.com/a/b?queryB=bar&queryC=v3",
+                                "http://www.foo.com/a/b?queryB=[*!]&queryC=v3"));
+        assertFalse(isPagingUrl("http://www.foo.com/a/b?queryC=v3&queryB=3",
+                                "http://www.foo.com/a/b?queryB=[*!]&queryC=v3"));
+        assertFalse(isPagingUrl("http://www.foo.com/a/b?queryA=v1",
+                                "http://www.foo.com/a/b?queryA=v1&queryB=[*!]&queryC=v3"));
+
+        assertTrue(isPagingUrl("http://www.foo.com/a/abc-2.html",
+                               "http://www.foo.com/a/abc-[*!].html"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/abc.html",
+                               "http://www.foo.com/a/abc-[*!].html"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/abc", "http://www.foo.com/a/abc-[*!]"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/abc-2", "http://www.foo.com/a/abc-[*!]"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/b-c-3", "http://www.foo.com/a/b-[*!]-c-3"));
+        assertTrue(isPagingUrl("http://www.foo.com/a-c-3", "http://www.foo.com/a-[*!]-c-3"));
+        assertTrue(isPagingUrl("http://www.foo.com/a-p-1-c-3", "http://www.foo.com/a-p-[*!]-c-3"));
+        assertFalse(isPagingUrl("http://www.foo.com/a/abc-page", "http://www.foo.com/a/abc-[*!]"));
+        assertFalse(isPagingUrl("http://www.foo.com/a/2", "http://www.foo.com/a/abc-[*!]"));
+        assertFalse(isPagingUrl("http://www.foo.com/a", "http://www.foo.com/a/abc-[*!]"));
+
+        assertTrue(isPagingUrl("http://www.foo.com/a/page/2", "http://www.foo.com/a/page/[*!]"));
+        assertTrue(isPagingUrl("http://www.foo.com/a", "http://www.foo.com/a/page/[*!]"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/page/2/abc.html",
+                               "http://www.foo.com/a/page/[*!]/abc.html"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/abc.html",
+                               "http://www.foo.com/a/page/[*!]/abc.html"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/abc.html",
+                               "http://www.foo.com/a/[*!]/abc.html"));
+        assertTrue(isPagingUrl("http://www.foo.com/a/2/abc.html",
+                               "http://www.foo.com/a/[*!]/abc.html"));
+        assertTrue(isPagingUrl("http://www.foo.com/abc.html",
+                               "http://www.foo.com/a/[*!]/abc.html"));
+        assertFalse(isPagingUrl("http://www.foo.com/a/page/b", "http://www.foo.com/a/page/[*!]"));
+        assertFalse(isPagingUrl("http://www.foo.com/m/page/2", "http://www.foo.com/p/page/[*!]"));
+    }
+
+    public void testIsPagePatternValid() {
+        assertTrue(isPagePatternValid("http://www.google.com/forum-12",
+                "http://www.google.com/forum-12/page/[*!]"));
+        assertTrue(isPagePatternValid("http://www.google.com/forum-12",
+                "http://www.google.com/forum-12/[*!]"));
+        assertTrue(isPagePatternValid("http://www.google.com/forum-12",
+                "http://www.google.com/forum-12/page-[*!]"));
+        assertTrue(isPagePatternValid("http://www.google.com/forum-12",
+                "http://www.google.com/forum-12?page=[*!]"));
+        assertTrue(isPagePatternValid("http://www.google.com/forum-12?sid=12345",
+                "http://www.google.com/forum-12?page=[*!]&sort=d"));
+        assertTrue(isPagePatternValid("http://www.google.com/forum-12/food",
+                "http://www.google.com/forum-12/food/for/bar/[*!]"));
+        assertTrue(isPagePatternValid("http://www.google.com/forum-12-food",
+                "http://www.google.com/forum-12-food-[*!]"));
+
+        assertFalse(isPagePatternValid("http://www.google.com/forum-12/food",
+                "http://www.google.com/forum-12/food/2012/01/[*!]"));
+        assertFalse(isPagePatternValid("http://www.google.com/forum-12/food/2012/01/01",
+                "http://www.google.com/forum-12/food/2012/01/[*!]"));
+        assertFalse(isPagePatternValid("http://www.google.com/forum-12/food",
+                "http://www.google.com/forum-12?page=[*!]"));
+
+        assertTrue(isPagePatternValid("http://www.google.com/thread/12",
+                "http://www.google.com/thread/12/page/[*!]"));
+        assertFalse(isPagePatternValid("http://www.google.com/thread/12/foo",
+                "http://www.google.com/thread/12/page/[*!]/foo"));
+        assertTrue(isPagePatternValid("http://www.google.com/thread/12/foo",
+                "http://www.google.com/thread/12/[*!]/foo"));
+    }
+
+    private static boolean isPagingUrl(String urlStr, String patternStr) {
+        PagePattern pattern = PagePattern.create(patternStr);
+        assertTrue(pattern != null);
+        return pattern.isPagingUrl(urlStr);
+    }
+
+    private static boolean isPagePatternValid(String urlStr, String patternStr) {
+        ParsedUrl parsedUrl = ParsedUrl.create(urlStr);
+        assertTrue(parsedUrl != null);
+        PagePattern pattern = PagePattern.create(patternStr);
+        assertTrue(pattern != null);
+        return pattern.isValidFor(parsedUrl);
+    }
+
+}