DomFramework/Tests/robotstxtTest.php

<?php
/** Test the mrobotstxt.txt file
  */
class test_model extends PHPUnit_Framework_TestCase
{
  // Empty Robots
  public function test_Construct_1 ()
  {
    $robotstxt = new robotstxt ("");
    $res = $robotstxt->allow ();
    $this->assertSame ($res, ["/"]);
  }
  public function test_Construct_2 ()
  {
    $robotstxt = new robotstxt ("");
    $res = $robotstxt->disallow ();
    $this->assertSame ($res, array ());
  }
  public function test_Construct_3 ()
  {
    $robotstxt = new robotstxt ("");
    $res = $robotstxt->sitemaps ();
    $this->assertSame ($res, array ());
  }
  public function test_Construct_4 ()
  {
    $robotstxt = new robotstxt ("");
    $res = $robotstxt->crawldelay ();
    $this->assertSame ($res, 3);
  }

  // Allow
  public function test_allow_1 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow:\n");
    $res = $robotstxt->allow ();
    $this->assertSame ($res, ["/"]);
  }
  public function test_allow_2 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow:\n\nUser-Agent: DomSearch\nDisallow:\n");
    $res = $robotstxt->allow ();
    $this->assertSame ($res, ["/"]);
  }
  public function test_allow_3 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: DomSearch\nDisallow:\n\nUser-Agent: *\nDisallow:\n");
    $res = $robotstxt->allow ();
    $this->assertSame ($res, ["/"]);
  }
  public function test_allow_4 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: DomSearch\n".
      "User-Agent: User1\n".
      "User-Agent: User2\n".
      "Disallow:\n\n".
      "User-Agent: *\n".
      "Disallow: /\n");
    $res = $robotstxt->allow ();
    $this->assertSame ($res, ["/"]);
  }

  // Disallow
  public function test_disallow_1 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow: /\n");
    $res = $robotstxt->disallow ();
    $this->assertSame ($res, ["/"]);
  }
  public function test_disallow_2 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow: /\n\nUser-Agent: DomSearch\nDisallow: /\n");
    $res = $robotstxt->disallow ();
    $this->assertSame ($res, ["/"]);
  }
  public function test_disallow_3 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: DomSearch\nDisallow: /\n\nUser-Agent: *\nDisallow: /\n");
    $res = $robotstxt->disallow ();
    $this->assertSame ($res, ["/"]);
  }

  // Sitemaps
  public function test_sitemaps_1 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: DomSearch\nDisallow: /\n\nUser-Agent: *\nDisallow: /\n");
    $res = $robotstxt->sitemaps ();
    $this->assertSame ($res, []);
  }
  public function test_sitemaps_2 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow: /\nSitemap: http://example.com/sitemap.xml");
    $res = $robotstxt->sitemaps ();
    $this->assertSame ($res, ["http://example.com/sitemap.xml"]);
  }
  public function test_sitemaps_3 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow: /\n".
      "Sitemap: http://example.com/sitemap.xml\n".
      "Sitemap: http://example.com/SITEMAP.XML");
    $res = $robotstxt->sitemaps ();
    $this->assertSame ($res,
      ["http://example.com/sitemap.xml", "http://example.com/SITEMAP.XML"]);
  }

  // Host
  public function test_host_1 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow: /\n");
    $res = $robotstxt->host ();
    $this->assertSame ($res, null);
  }
  public function test_host_2 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow: /\n\nHost: localhost");
    $res = $robotstxt->host ();
    $this->assertSame ($res, "localhost");
  }

  // URLAllow
  public function test_urlallow_1 ()
  {
    $robotstxt = new robotstxt ("");
    $res = $robotstxt->URLAllow ("/");
    $this->assertSame ($res, true);
  }
  public function test_urlallow_2 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow: /");
    $res = $robotstxt->URLAllow ("/");
    $this->assertSame ($res, false);
  }
  public function test_urlallow_3 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow: /\nAllow: /allow/");
    $res = $robotstxt->URLAllow ("/");
    $this->assertSame ($res, false);
  }
  public function test_urlallow_4 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow: /\nAllow: /allow/");
    $res = $robotstxt->URLAllow ("/allow/file");
    $this->assertSame ($res, true);
  }
  public function test_urlallow_5 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow: /\nAllow: /allow/*.gif$");
    $res = $robotstxt->URLAllow ("/allow/file.gif");
    $this->assertSame ($res, true);
  }
  public function test_urlallow_6 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow: /\nAllow: /allow/*.gif$");
    $res = $robotstxt->URLAllow ("/allow/.gif");
    $this->assertSame ($res, false);
  }
  public function test_urlallow_7 ()
  {
    $robotstxt = new robotstxt (
      "User-Agent: *\nDisallow: /\nAllow: /allow/*.gif\$");
    $res = $robotstxt->URLAllow ("/allow/file.png");
    $this->assertSame ($res, false);
  }
}