You are here

class SearchExcerptTest in Drupal 9

Same name and namespace in other branches
  1. 8 core/modules/search/tests/src/Kernel/SearchExcerptTest.php \Drupal\Tests\search\Kernel\SearchExcerptTest

Tests the search_excerpt() function.

@group search

Hierarchy

Expanded class hierarchy of SearchExcerptTest

File

core/modules/search/tests/src/Kernel/SearchExcerptTest.php, line 12

Namespace

Drupal\Tests\search\Kernel
View source
class SearchExcerptTest extends KernelTestBase {

  /**
   * Modules to enable.
   *
   * @var array
   */
  protected static $modules = [
    'search',
    'search_langcode_test',
  ];

  /**
   * Tests search_excerpt() with several simulated search keywords.
   *
   * Passes keywords and a sample marked up string, "The quick
   * brown fox jumps over the lazy dog", and compares it to the
   * correctly marked up string. The correctly marked up string
   * contains either highlighted keywords or the original marked
   * up string if no keywords matched the string.
   */
  public function testSearchExcerpt() {

    // Make some text with entities and tags.
    $text = 'The <strong>quick</strong> <a href="#">brown</a> fox &amp; jumps <h2>over</h2> the lazy dog';
    $expected = 'The quick brown fox &amp; jumps over the lazy dog';
    $result = $this
      ->doSearchExcerpt('nothing', $text);
    $this
      ->assertEquals($expected, preg_replace('| +|', ' ', $result), 'Entire string, stripped of HTML tags, is returned when keyword is not found in short string');
    $result = $this
      ->doSearchExcerpt('fox', $text);
    $this
      ->assertEquals('The quick brown <strong>fox</strong> &amp; jumps over the lazy dog', $result, 'Found keyword is highlighted');
    $expected = '<strong>The</strong> quick brown fox &amp; jumps over <strong>the</strong> lazy dog';
    $result = $this
      ->doSearchExcerpt('The', $text);
    $this
      ->assertEquals($expected, preg_replace('| +|', ' ', $result), 'Keyword is highlighted at beginning of short string');
    $expected = 'The quick brown fox &amp; jumps over the lazy <strong>dog</strong>';
    $result = $this
      ->doSearchExcerpt('dog', $text);
    $this
      ->assertEquals($expected, preg_replace('| +|', ' ', $result), 'Keyword is highlighted at end of short string');
    $longtext = str_repeat(str_replace('brown', 'silver', $text) . ' ', 10) . $text . str_repeat(' ' . str_replace('brown', 'pink', $text), 10);
    $result = $this
      ->doSearchExcerpt('brown', $longtext);
    $expected = '… silver fox &amp; jumps over the lazy dog The quick <strong>brown</strong> fox &amp; jumps over the lazy dog The quick …';
    $this
      ->assertEquals($expected, $result, 'Snippet around keyword in long text is correctly capped');
    $longtext = str_repeat($text . ' ', 10);
    $result = $this
      ->doSearchExcerpt('nothing', $longtext);
    $expected = 'The quick brown fox &amp; jumps over the lazy dog';
    $this
      ->assertStringStartsWith($expected, $result, 'When keyword is not found in long string, return value starts as expected');
    $entities = str_repeat('k&eacute;sz&iacute;t&eacute;se ', 20);
    $result = $this
      ->doSearchExcerpt('nothing', $entities);
    $this
      ->assertStringNotContainsString('&', $result, 'Entities are not present in excerpt');
    $this
      ->assertStringContainsString('í', $result, 'Entities are converted in excerpt');

    // The node body that will produce this rendered $text is:
    // 123456789 HTMLTest +123456789+&lsquo;  +&lsquo;  +&lsquo;  +&lsquo;  +12345678  &nbsp;&nbsp;  +&lsquo;  +&lsquo;  +&lsquo;   &lsquo;
    $text = "<div class=\"field field--name-body field--type-text-with-summary field--label-hidden\"><div class=\"field__items\"><div class=\"field__item even\" property=\"content:encoded\"><p>123456789 HTMLTest +123456789+‘  +‘  +‘  +‘  +12345678      +‘  +‘  +‘   ‘</p>\n</div></div></div> ";
    $result = $this
      ->doSearchExcerpt('HTMLTest', $text);
    $this
      ->assertFalse(empty($result), 'Rendered Multi-byte HTML encodings are not corrupted in search excerpts');
  }

  /**
   * Tests search_excerpt() with search keywords matching simplified words.
   *
   * Excerpting should handle keywords that are matched only after going through
   * text analysis. This test passes keywords that match simplified words
   * and compares them with strings that contain the original unsimplified word.
   */
  public function testSearchExcerptSimplified() {
    $start_time = microtime(TRUE);

    // cSpell:disable
    $lorem1 = 'Lorem ipsum dolor sit amet, consectetur adipiscing elit. Etiam vitae arcu at leo cursus laoreet. Curabitur dui tortor, adipiscing malesuada tempor in, bibendum ac diam. Cras non tellus a libero pellentesque condimentum. What is a Drupalism? Suspendisse ac lacus libero. Ut non est vel nisl faucibus interdum nec sed leo. Pellentesque sem risus, vulputate eu semper eget, auctor in libero.';
    $lorem2 = 'Ut fermentum est vitae metus convallis scelerisque. Phasellus pellentesque rhoncus tellus, eu dignissim purus posuere id. Quisque eu fringilla ligula. Morbi ullamcorper, lorem et mattis egestas, tortor neque pretium velit, eget eleifend odio turpis eu purus. Donec vitae metus quis leo pretium tincidunt a pulvinar sem. Morbi adipiscing laoreet mauris vel placerat. Nullam elementum, nisl sit amet scelerisque malesuada, dolor nunc hendrerit quam, eu ultrices erat est in orci.';

    // cSpell:enable
    // Make some text with some keywords that will get simplified.
    $text = $lorem1 . ' Number: 123456.7890 Hyphenated: one-two abc,def ' . $lorem2;

    // Note: The search_excerpt() function adds some extra spaces -- not
    // important for HTML formatting. Remove these for comparison.
    $result = $this
      ->doSearchExcerpt('123456.7890', $text);
    $this
      ->assertStringContainsString('Number: <strong>123456.7890</strong>', $result, 'Numeric keyword is highlighted with exact match');
    $result = $this
      ->doSearchExcerpt('1234567890', $text);
    $this
      ->assertStringContainsString('Number: <strong>123456.7890</strong>', $result, 'Numeric keyword is highlighted with simplified match');
    $result = $this
      ->doSearchExcerpt('Number 1234567890', $text);
    $this
      ->assertStringContainsString('<strong>Number</strong>: <strong>123456.7890</strong>', $result, 'Punctuated and numeric keyword is highlighted with simplified match');
    $result = $this
      ->doSearchExcerpt('"Number 1234567890"', $text);
    $this
      ->assertStringContainsString('<strong>Number: 123456.7890</strong>', $result, 'Phrase with punctuated and numeric keyword is highlighted with simplified match');
    $result = $this
      ->doSearchExcerpt('"Hyphenated onetwo"', $text);
    $this
      ->assertStringContainsString('<strong>Hyphenated: one-two</strong>', $result, 'Phrase with punctuated and hyphenated keyword is highlighted with simplified match');
    $result = $this
      ->doSearchExcerpt('"abc def"', $text);
    $this
      ->assertStringContainsString('<strong>abc,def</strong>', $result, 'Phrase with keyword simplified into two separate words is highlighted with simplified match');

    // Test phrases with characters which are being truncated.
    $result = $this
      ->doSearchExcerpt('"ipsum _"', $text);
    $this
      ->assertStringContainsString('<strong>ipsum</strong>', $result, 'Only valid part of the phrase is highlighted and invalid part containing "_" is ignored.');
    $result = $this
      ->doSearchExcerpt('"ipsum 0000"', $text);
    $this
      ->assertStringContainsString('<strong>ipsum</strong>', $result, 'Only valid part of the phrase is highlighted and invalid part "0000" is ignored.');

    // Test combination of the valid keyword and keyword containing only
    // characters which are being truncated during simplification.
    $result = $this
      ->doSearchExcerpt('ipsum _', $text);
    $this
      ->assertStringContainsString('<strong>ipsum</strong>', $result, 'Only valid keyword is highlighted and invalid keyword "_" is ignored.');
    $result = $this
      ->doSearchExcerpt('ipsum 0000', $text);
    $this
      ->assertStringContainsString('<strong>ipsum</strong>', $result, 'Only valid keyword is highlighted and invalid keyword "0000" is ignored.');

    // Test using the hook_search_preprocess() from the test module.
    // The hook replaces "finding" or "finds" with "find".
    // So, if we search for "find" or "finds" or "finding", we should
    // highlight "finding".
    $text = "this tests finding a string";
    $result = $this
      ->doSearchExcerpt('finds', $text, 'ex');
    $this
      ->assertStringContainsString('<strong>finding</strong>', $result, 'Search excerpt works with preprocess hook, search for finds');
    $result = $this
      ->doSearchExcerpt('find', $text, 'ex');
    $this
      ->assertStringContainsString('<strong>finding</strong>', $result, 'Search excerpt works with preprocess hook, search for find');

    // Just to be sure, test with the replacement at the beginning and end.
    $text = "finding at the beginning";
    $result = $this
      ->doSearchExcerpt('finds', $text, 'ex');
    $this
      ->assertStringContainsString('<strong>finding</strong>', $result, 'Search excerpt works with preprocess hook, text at start');
    $text = "at the end finding";
    $result = $this
      ->doSearchExcerpt('finds', $text, 'ex');
    $this
      ->assertStringContainsString('<strong>finding</strong>', $result, 'Search excerpt works with preprocess hook, text at end');

    // Testing with a one-to-many replacement: the test module replaces DIC
    // with Dependency Injection Container.
    $text = "something about the DIC is happening";
    $result = $this
      ->doSearchExcerpt('Dependency', $text, 'ex');
    $this
      ->assertStringContainsString('<strong>DIC</strong>', $result, 'Search excerpt works with preprocess hook, acronym first word');
    $result = $this
      ->doSearchExcerpt('Injection', $text, 'ex');
    $this
      ->assertStringContainsString('<strong>DIC</strong>', $result, 'Search excerpt works with preprocess hook, acronym second word');
    $result = $this
      ->doSearchExcerpt('Container', $text, 'ex');
    $this
      ->assertStringContainsString('<strong>DIC</strong>', $result, 'Search excerpt works with preprocess hook, acronym third word');

    // Testing with a many-to-one replacement: the test module replaces
    // hypertext markup language with HTML.
    $text = "we always use hypertext markup language to describe things";
    $result = $this
      ->doSearchExcerpt('html', $text, 'ex');
    $this
      ->assertStringContainsString('<strong>hypertext markup language</strong>', $result, 'Search excerpt works with preprocess hook, acronym many to one');

    // Test with accents and caps in a longer piece of text with the target
    // near the end.
    $text = str_repeat($lorem2, 20) . ' ' . $lorem1;
    $result = $this
      ->doSearchExcerpt('Lìbêró', $text);
    $this
      ->assertStringContainsString('<strong>libero</strong>', $result, 'Search excerpt works with caps and accents in longer text');

    // Test with an acronym provided by the hook, with the target text in the
    // middle of a long string.
    $text = str_repeat($lorem2, 10) . ' DIC ' . str_repeat($lorem2, 10);
    $result = $this
      ->doSearchExcerpt('Dependency', $text, 'ex');
    $this
      ->assertStringContainsString('<strong>DIC</strong>', $result, 'Search excerpt works with acronym in longer text');

    // Test a long string with a lot of whitespace in it.
    $lorem3 = str_replace(' ', str_repeat(" \n", 20), $lorem2);
    $text = str_repeat($lorem3, 20) . ' ' . $lorem1;
    $result = $this
      ->doSearchExcerpt('Lìbêró', $text);
    $this
      ->assertStringContainsString('<strong>libero</strong>', $result, 'Search excerpt works with caps and accents in longer text with whitespace');
  }

  /**
   * Calls search_excerpt() and renders output.
   *
   * @param string $keys
   *   A string containing a search query.
   * @param string $render_array
   *   The text to extract fragments from.
   * @param string|null $langcode
   *   Language code for the language of $text, if known.
   *
   * @return string
   *   A string containing HTML for the excerpt.
   */
  protected function doSearchExcerpt($keys, $render_array, $langcode = NULL) {
    $render_array = search_excerpt($keys, $render_array, $langcode);
    $text = \Drupal::service('renderer')
      ->renderPlain($render_array);

    // The search_excerpt() function adds some extra spaces -- not
    // important for HTML formatting or this test. Remove these for comparison.
    return preg_replace('| +|', ' ', $text);
  }

}

Members

Namesort descending Modifiers Type Description Overrides
AssertContentTrait::$content protected property The current raw content.
AssertContentTrait::$drupalSettings protected property The drupalSettings value from the current raw $content.
AssertContentTrait::$elements protected property The XML structure parsed from the current raw $content. 1
AssertContentTrait::$plainTextContent protected property The plain-text content of raw $content (text nodes).
AssertContentTrait::assertEscaped protected function Passes if the raw text IS found escaped on the loaded page, fail otherwise.
AssertContentTrait::assertField protected function Asserts that a field exists with the given name or ID.
AssertContentTrait::assertFieldById protected function Asserts that a field exists with the given ID and value.
AssertContentTrait::assertFieldByName protected function Asserts that a field exists with the given name and value.
AssertContentTrait::assertFieldByXPath protected function Asserts that a field exists in the current page by the given XPath.
AssertContentTrait::assertFieldChecked protected function Asserts that a checkbox field in the current page is checked.
AssertContentTrait::assertFieldsByValue protected function Asserts that a field exists in the current page with a given Xpath result.
AssertContentTrait::assertLink protected function Passes if a link with the specified label is found.
AssertContentTrait::assertLinkByHref protected function Passes if a link containing a given href (part) is found.
AssertContentTrait::assertNoDuplicateIds protected function Asserts that each HTML ID is used for just a single element.
AssertContentTrait::assertNoEscaped protected function Passes if the raw text IS NOT found escaped on the loaded page, fail otherwise.
AssertContentTrait::assertNoField protected function Asserts that a field does not exist with the given name or ID.
AssertContentTrait::assertNoFieldById protected function Asserts that a field does not exist with the given ID and value.
AssertContentTrait::assertNoFieldByName protected function Asserts that a field does not exist with the given name and value.
AssertContentTrait::assertNoFieldByXPath protected function Asserts that a field does not exist or its value does not match, by XPath.
AssertContentTrait::assertNoFieldChecked protected function Asserts that a checkbox field in the current page is not checked.
AssertContentTrait::assertNoLink protected function Passes if a link with the specified label is not found.
AssertContentTrait::assertNoLinkByHref protected function Passes if a link containing a given href (part) is not found.
AssertContentTrait::assertNoLinkByHrefInMainRegion protected function Passes if a link containing a given href is not found in the main region.
AssertContentTrait::assertNoOption protected function Asserts that a select option in the current page does not exist.
AssertContentTrait::assertNoOptionSelected protected function Asserts that a select option in the current page is not checked.
AssertContentTrait::assertNoPattern protected function Triggers a pass if the perl regex pattern is not found in raw content.
AssertContentTrait::assertNoRaw protected function Passes if the raw text is NOT found on the loaded page, fail otherwise.
AssertContentTrait::assertNoText protected function Passes if the page (with HTML stripped) does not contains the text.
AssertContentTrait::assertNoTitle protected function Pass if the page title is not the given string.
AssertContentTrait::assertNoUniqueText protected function Passes if the text is found MORE THAN ONCE on the text version of the page.
AssertContentTrait::assertOption protected function Asserts that a select option in the current page exists.
AssertContentTrait::assertOptionByText protected function Asserts that a select option with the visible text exists.
AssertContentTrait::assertOptionSelected protected function Asserts that a select option in the current page is checked.
AssertContentTrait::assertOptionSelectedWithDrupalSelector protected function Asserts that a select option in the current page is checked.
AssertContentTrait::assertOptionWithDrupalSelector protected function Asserts that a select option in the current page exists.
AssertContentTrait::assertPattern protected function Triggers a pass if the Perl regex pattern is found in the raw content.
AssertContentTrait::assertRaw protected function Passes if the raw text IS found on the loaded page, fail otherwise.
AssertContentTrait::assertText protected function Passes if the page (with HTML stripped) contains the text.
AssertContentTrait::assertTextHelper protected function Helper for assertText and assertNoText.
AssertContentTrait::assertTextPattern protected function Asserts that a Perl regex pattern is found in the plain-text content.
AssertContentTrait::assertThemeOutput protected function Asserts themed output.
AssertContentTrait::assertTitle protected function Pass if the page title is the given string.
AssertContentTrait::assertUniqueText protected function Passes if the text is found ONLY ONCE on the text version of the page.
AssertContentTrait::assertUniqueTextHelper protected function Helper for assertUniqueText and assertNoUniqueText.
AssertContentTrait::buildXPathQuery protected function Builds an XPath query.
AssertContentTrait::constructFieldXpath protected function Helper: Constructs an XPath for the given set of attributes and value.
AssertContentTrait::cssSelect protected function Searches elements using a CSS selector in the raw content.
AssertContentTrait::getAllOptions protected function Get all option elements, including nested options, in a select.
AssertContentTrait::getDrupalSettings protected function Gets the value of drupalSettings for the currently-loaded page.
AssertContentTrait::getRawContent protected function Gets the current raw content.
AssertContentTrait::getSelectedItem protected function Get the selected value from a select field.
AssertContentTrait::getTextContent protected function Retrieves the plain-text content from the current raw content.
AssertContentTrait::getUrl protected function Get the current URL from the cURL handler. 1
AssertContentTrait::parse protected function Parse content returned from curlExec using DOM and SimpleXML.
AssertContentTrait::removeWhiteSpace protected function Removes all white-space between HTML tags from the raw content.
AssertContentTrait::setDrupalSettings protected function Sets the value of drupalSettings for the currently-loaded page.
AssertContentTrait::setRawContent protected function Sets the raw content (e.g. HTML).
AssertContentTrait::xpath protected function Performs an xpath search on the contents of the internal browser.
AssertLegacyTrait::assert Deprecated protected function
AssertLegacyTrait::assertEqual Deprecated protected function
AssertLegacyTrait::assertIdentical Deprecated protected function
AssertLegacyTrait::assertIdenticalObject Deprecated protected function
AssertLegacyTrait::assertNotEqual Deprecated protected function
AssertLegacyTrait::assertNotIdentical Deprecated protected function
AssertLegacyTrait::pass Deprecated protected function
AssertLegacyTrait::verbose Deprecated protected function
ConfigTestTrait::configImporter protected function Returns a ConfigImporter object to import test configuration.
ConfigTestTrait::copyConfig protected function Copies configuration objects from source storage to target storage.
ExtensionListTestTrait::getModulePath protected function Gets the path for the specified module.
ExtensionListTestTrait::getThemePath protected function Gets the path for the specified theme.
KernelTestBase::$backupGlobals protected property Back up and restore any global variables that may be changed by tests.
KernelTestBase::$backupStaticAttributes protected property Back up and restore static class properties that may be changed by tests.
KernelTestBase::$backupStaticAttributesBlacklist protected property Contains a few static class properties for performance.
KernelTestBase::$classLoader protected property
KernelTestBase::$configImporter protected property @todo Move into Config test base class. 7
KernelTestBase::$configSchemaCheckerExclusions protected static property An array of config object names that are excluded from schema checking.
KernelTestBase::$container protected property
KernelTestBase::$databasePrefix protected property
KernelTestBase::$preserveGlobalState protected property Do not forward any global state from the parent process to the processes that run the actual tests.
KernelTestBase::$root protected property The app root.
KernelTestBase::$runTestInSeparateProcess protected property Kernel tests are run in separate processes because they allow autoloading of code from extensions. Running the test in a separate process isolates this behavior from other tests. Subclasses should not override this property.
KernelTestBase::$siteDirectory protected property
KernelTestBase::$strictConfigSchema protected property Set to TRUE to strict check all configuration saved. 6
KernelTestBase::$vfsRoot protected property The virtual filesystem root directory.
KernelTestBase::assertPostConditions protected function 1
KernelTestBase::bootEnvironment protected function Bootstraps a basic test environment.
KernelTestBase::bootKernel private function Bootstraps a kernel for a test.
KernelTestBase::config protected function Configuration accessor for tests. Returns non-overridden configuration.
KernelTestBase::disableModules protected function Disables modules for this test.
KernelTestBase::enableModules protected function Enables modules for this test.
KernelTestBase::getConfigSchemaExclusions protected function Gets the config schema exclusions for this test.
KernelTestBase::getDatabaseConnectionInfo protected function Returns the Database connection info to be used for this test. 3
KernelTestBase::getDatabasePrefix public function
KernelTestBase::getExtensionsForModules private function Returns Extension objects for $modules to enable.
KernelTestBase::getModulesToEnable private static function Returns the modules to enable for this test.
KernelTestBase::initFileCache protected function Initializes the FileCache component.
KernelTestBase::installConfig protected function Installs default configuration for a given list of modules.
KernelTestBase::installEntitySchema protected function Installs the storage schema for a specific entity type.
KernelTestBase::installSchema protected function Installs database tables from a module schema definition.
KernelTestBase::prepareTemplate protected function
KernelTestBase::register public function Registers test-specific services. Overrides ServiceProviderInterface::register 24
KernelTestBase::render protected function Renders a render array. 1
KernelTestBase::setInstallProfile protected function Sets the install profile and rebuilds the container to update it.
KernelTestBase::setSetting protected function Sets an in-memory Settings variable.
KernelTestBase::setUp protected function 334
KernelTestBase::setUpBeforeClass public static function 1
KernelTestBase::setUpFilesystem protected function Sets up the filesystem, so things like the file directory. 2
KernelTestBase::stop protected function Stops test execution.
KernelTestBase::tearDown protected function 4
KernelTestBase::tearDownCloseDatabaseConnection public function @after
KernelTestBase::vfsDump protected function Dumps the current state of the virtual filesystem to STDOUT.
KernelTestBase::__sleep public function Prevents serializing any properties.
PhpUnitWarnings::$deprecationWarnings private static property Deprecation warnings from PHPUnit to raise with @trigger_error().
PhpUnitWarnings::addWarning public function Converts PHPUnit deprecation warnings to E_USER_DEPRECATED.
RandomGeneratorTrait::$randomGenerator protected property The random generator.
RandomGeneratorTrait::getRandomGenerator protected function Gets the random generator for the utility methods.
RandomGeneratorTrait::randomMachineName protected function Generates a unique random string containing letters and numbers. 1
RandomGeneratorTrait::randomObject public function Generates a random PHP object.
RandomGeneratorTrait::randomString public function Generates a pseudo-random string of ASCII characters of codes 32 to 126.
RandomGeneratorTrait::randomStringValidate public function Callback for random string validation.
SearchExcerptTest::$modules protected static property Modules to enable. Overrides KernelTestBase::$modules
SearchExcerptTest::doSearchExcerpt protected function Calls search_excerpt() and renders output.
SearchExcerptTest::testSearchExcerpt public function Tests search_excerpt() with several simulated search keywords.
SearchExcerptTest::testSearchExcerptSimplified public function Tests search_excerpt() with search keywords matching simplified words.
StorageCopyTrait::replaceStorageContents protected static function Copy the configuration from one storage to another and remove stale items.
TestRequirementsTrait::checkModuleRequirements private function Checks missing module requirements.
TestRequirementsTrait::checkRequirements protected function Check module requirements for the Drupal use case. 1
TestRequirementsTrait::getDrupalRoot protected static function Returns the Drupal root directory.