Add test for searching content across files with different format with search text highlight (#7236)

tests/acceptance/features/apiSearch/contentSearch.feature

@@ -1,3 +1,4 @@
+@tikaServiceNeeded
 Feature: content search
   As a user
   I want to do search resources by content
@@ -172,3 +173,32 @@ Feature: content search
       | content:hel*                                | 2            | /technical task.txt | /task comments.txt |
       | content:hel* AND tag:test                   | 1            | /technical task.txt |                    |
       | (name:*task* AND content:hel*) NOT tag:test | 1            | /task comments.txt  |                    |
+
+
+  Scenario Outline: search across files with different format with search text highlight
+    Given using <dav-path-version> DAV path
+    And the administrator has assigned the role "Space Admin" to user "Alice" using the Graph API
+    And user "Alice" has created a space "project-space" with the default quota using the GraphApi
+    And user "Alice" has uploaded file with content "this is a simple text file" to "test-text-file.txt"
+    And user "Alice" has uploaded file with content "this is a simple pdf file" to "test-pdf-file.pdf"
+    And user "Alice" has uploaded file with content "this is a simple cpp file" to "test-cpp-file.cpp"
+    And user "Alice" has uploaded file with content "this is another text file" to "testfile.txt"
+    And user "Alice" has uploaded file "filesForUpload/testavatar.png" to "/testavatar.png"
+    And user "Alice" has uploaded a file inside space "project-space" with content "this is a simple markdown file" to "test-md-file.md"
+    And user "Alice" has uploaded a file inside space "project-space" with content "this is a simple odt file" to "test-odt-file.odt"
+    When user "Alice" searches for "Content:simple" using the WebDAV API
+    Then the HTTP status code should be "207"
+    And the search result should contain these entries with highlight on keyword "simple"
+      | test-text-file.txt |
+      | test-pdf-file.pdf  |
+      | test-cpp-file.cpp  |
+      | test-md-file.md    |
+      | test-odt-file.odt  |
+    But the search result of user "Alice" should not contain these entries:
+      | testavatar.png |
+      | testfile.txt   |
+    Examples:
+      | dav-path-version |
+      | old              |
+      | new              |
+      | spaces           |

tests/acceptance/features/bootstrap/SearchContext.php

@@ -52,13 +52,13 @@ class SearchContext implements Context {
 	 * @return void
 	 */
 	public function userSearchesUsingWebDavAPI(
-		string $user,
-		string $pattern,
-		?string	$limit = null,
-		?string $scope = null,
-		?string $spaceName = null,
+		string    $user,
+		string    $pattern,
+		?string   $limit = null,
+		?string   $scope = null,
+		?string   $spaceName = null,
 		TableNode $properties = null
-	):void {
+	): void {
 		// Because indexing of newly uploaded files or directories with ocis is decoupled and occurs asynchronously, a short wait is necessary before searching files or folders.
 		sleep(4);
 		$user = $this->featureContext->getActualUsername($user);
@@ -113,10 +113,10 @@ class SearchContext implements Context {
 	 * @throws Exception
 	 */
 	public function fileOrFolderInTheSearchResultShouldContainProperties(
-		string $path,
-		string $user,
+		string    $path,
+		string    $user,
 		TableNode $properties
-	):void {
+	): void {
 		$user = $this->featureContext->getActualUsername($user);
 		$this->featureContext->verifyTableNodeColumns($properties, ['name', 'value']);
 		$properties = $properties->getHash();
@@ -163,10 +163,47 @@ class SearchContext implements Context {
 	 *
 	 * @return void
 	 */
-	public function before(BeforeScenarioScope $scope):void {
+	public function before(BeforeScenarioScope $scope): void {
 		// Get the environment
 		$environment = $scope->getEnvironment();
 		// Get all the contexts you need in this context
 		$this->featureContext = $environment->getContext('FeatureContext');
 	}
+
+	/**
+	 * @Then /^the search result should contain these (?:files|entries) with highlight on keyword "([^"]*)"/
+	 *
+	 * @param TableNode $expectedFiles
+	 * @param string $expectedContent
+	 *
+	 * @return void
+	 *
+	 * @throws Exception
+	 */
+	public function theSearchResultShouldContainEntriesWithHighlight(
+		TableNode $expectedFiles,
+		string    $expectedContent
+	): void {
+		$this->featureContext->verifyTableNodeColumnsCount($expectedFiles, 1);
+		$elementRows = $expectedFiles->getRows();
+		$foundEntries = $this->featureContext->findEntryFromSearchResponse(
+			null,
+			true
+		);
+		foreach ($elementRows as $index => $expectedFile) {
+			$filename = $expectedFile[0];
+			$content = $foundEntries[$filename];
+			// Extract the content between the <mark> tags
+			preg_match('/<mark>(.*?)<\/mark>/s', $content, $matches);
+			$actualContent = isset($matches[1]) ? $matches[1] : '';
+
+			// Remove any leading/trailing whitespace for comparison
+			$actualContent = trim($actualContent);
+			Assert::assertEquals(
+				$expectedContent,
+				$actualContent,
+				"Expected text highlight to be '$expectedContent' but found '$actualContent'"
+			);
+		}
+	}
 }

tests/acceptance/features/bootstrap/WebDav.php

@@ -5355,6 +5355,7 @@ trait WebDav {
 	 * and returns found search results if found else returns false
 	 *
 	 * @param string|null $entryNameToSearch
+	 * @param bool|null $searchForHighlightString
 	 *
 	 * @return string|array|boolean
 	 *
@@ -5362,10 +5363,11 @@ trait WebDav {
 	 * array if $entryNameToSearch is not given
 	 * boolean false if $entryNameToSearch is given and is not found
 	 *
-	 * @throws GuzzleException
+	 * @throws Exception
 	 */
 	public function findEntryFromSearchResponse(
-		?string $entryNameToSearch = null
+		?string $entryNameToSearch = null,
+		?bool $searchForHighlightString = false
 	) {
 		// trim any leading "/" passed by the caller, we can just match the "raw" name
 		if ($entryNameToSearch !== null) {
@@ -5387,7 +5389,12 @@ trait WebDav {
 			if ($entryNameToSearch === $resourcePath) {
 				return $resourcePath;
 			}
-			$results[] = $resourcePath;
+			if ($searchForHighlightString) {
+				$actualHighlightString =  $item->xpath("d:propstat//oc:highlights");
+				$results[$resourcePath] = (string)$actualHighlightString[0];
+			} else {
+				$results[] = $resourcePath;
+			}
 		}
 		if ($entryNameToSearch === null) {
 			return $results;