Optimize dist-archive by preventing descent into ignored directories

.gitignore

@@ -6,3 +6,5 @@ vendor/
 phpunit.xml
 phpcs.xml
 .phpcs.xml
+.phpunit.result.cache
+.phpunit.cache

composer.json

@@ -13,7 +13,7 @@
     ],
     "require": {
         "php": ">=7.2",
-        "wp-cli/wp-cli": "^2",
+        "wp-cli/wp-cli": "^2.13",
         "inmarelibero/gitignore-checker": "^1.0.4"
     },
     "require-dev": {

features/distignore.feature

@@ -379,3 +379,50 @@ Feature: Generate a distribution archive of a project with .distignore
       """
       Error: Broken symlink at /symlink. Target missing at
       """
+
+  Scenario: Efficiently ignores directories with many files
+    # Performance test: ensure ignored directories are not scanned
+    # @see https://github.com/wp-cli/dist-archive-command/issues/115
+    Given an empty directory
+    And a foo/.distignore file:
+      """
+      node_modules
+      .git
+      """
+    And a foo/plugin.php file:
+      """
+      <?php
+      /**
+       * Plugin Name: Test Plugin
+       * Version: 1.0.0
+       */
+      """
+    And a foo/readme.txt file:
+      """
+      === Test Plugin ===
+      """
+
+    When I run `mkdir -p foo/node_modules/package1 foo/node_modules/package2 foo/node_modules/package3`
+    Then STDERR should be empty
+
+    When I run `sh -c 'i=1; while [ $i -le 50 ]; do touch foo/node_modules/package1/file$i.js; i=$((i+1)); done'`
+    Then STDERR should be empty
+
+    When I run `sh -c 'i=1; while [ $i -le 50 ]; do touch foo/node_modules/package2/file$i.js; i=$((i+1)); done'`
+    Then STDERR should be empty
+
+    When I run `sh -c 'i=1; while [ $i -le 50 ]; do touch foo/node_modules/package3/file$i.js; i=$((i+1)); done'`
+    Then STDERR should be empty
+
+    When I run `wp dist-archive foo`
+    Then STDOUT should match /^Success: Created foo\.[^ ]+ \(Size: \d+(?:\.\d*)? [a-zA-Z]{1,3}\)$/
+    And STDERR should be empty
+
+    When I run `rm -rf foo`
+    Then the foo directory should not exist
+
+    When I try `unzip foo.*.zip`
+    Then the foo directory should exist
+    And the foo/plugin.php file should exist
+    And the foo/readme.txt file should exist
+    And the foo/node_modules directory should not exist

phpunit.xml.dist

@@ -0,0 +1,26 @@
+<phpunit xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+		 xsi:noNamespaceSchemaLocation="https://schema.phpunit.de/4.8/phpunit.xsd"
+		 bootstrap="vendor/autoload.php"
+		 backupGlobals="false"
+		 beStrictAboutCoversAnnotation="true"
+		 beStrictAboutOutputDuringTests="true"
+		 beStrictAboutTestsThatDoNotTestAnything="true"
+		 beStrictAboutTodoAnnotatedTests="true"
+		 convertErrorsToExceptions="true"
+		 convertWarningsToExceptions="true"
+		 convertNoticesToExceptions="true"
+		 convertDeprecationsToExceptions="true"
+		 colors="true"
+		 verbose="true">
+	<testsuites>
+		<testsuite name="wp-cli/dist-archive-command tests">
+			<directory suffix="Test.php">tests</directory>
+		</testsuite>
+	</testsuites>
+
+	<coverage processUncoveredFiles="false">
+		<include>
+			<directory suffix=".php">src</directory>
+		</include>
+	</coverage>
+</phpunit>

src/Dist_Archive_Command.php

@@ -493,46 +493,61 @@ protected function is_path_contains_symlink( $source_dir_path ) {
 	private function get_file_list( $source_dir_path, $excluded = false ) {
 
 		$included_files = [];
-		$excluded_files = [];
 
-		$iterator = new RecursiveIteratorIterator(
-			new RecursiveDirectoryIterator( $source_dir_path, RecursiveDirectoryIterator::SKIP_DOTS ),
+		$directory_iterator = new RecursiveDirectoryIterator( $source_dir_path, RecursiveDirectoryIterator::SKIP_DOTS );
+		$filter_iterator    = new Distignore_Filter_Iterator( $directory_iterator, $this->checker, $source_dir_path );
+		$iterator           = new RecursiveIteratorIterator(
+			$filter_iterator,
 			RecursiveIteratorIterator::SELF_FIRST
 		);
 
 		/**
 		 * @var SplFileInfo $item
 		 */
 		foreach ( $iterator as $item ) {
-			$relative_filepath = str_replace( $source_dir_path, '', $item->getPathname() );
-			try {
-				if ( $this->checker->isPathIgnored( $relative_filepath ) ) {
-					$excluded_files[] = $relative_filepath;
-				} else {
-					$included_files[] = $relative_filepath;
-				}
-			} catch ( \Inmarelibero\GitIgnoreChecker\Exception\InvalidArgumentException $exception ) {
+			$pathname = $item->getPathname();
+			if ( 0 === strpos( $pathname, $source_dir_path ) ) {
+				$relative_filepath = substr( $pathname, strlen( $source_dir_path ) );
+			} else {
+				$relative_filepath = $pathname;
+			}
+
+			// Check if this item had an error during filtering.
+			$error = $filter_iterator->getErrorForItem( $relative_filepath );
+			if ( $error ) {
 				if ( $item->isLink() && ! file_exists( (string) readlink( $item->getPathname() ) ) ) {
 					WP_CLI::error( "Broken symlink at {$relative_filepath}. Target missing at {$item->getLinkTarget()}." );
 				} else {
-					WP_CLI::error( $exception->getMessage() );
+					WP_CLI::error( $error->getMessage() );
 				}
 			}
-		}
 
-		// Check all excluded directories and remove them from the excluded list if they contain included files.
-		foreach ( $excluded_files as $excluded_file_index => $excluded_relative_path ) {
-			if ( ! is_dir( $source_dir_path . $excluded_relative_path ) ) {
-				continue;
+			// Check if this item is ignored (directories may still be yielded even if ignored).
+			if ( ! $filter_iterator->isPathIgnoredCached( $relative_filepath ) ) {
+				$included_files[] = $relative_filepath;
 			}
-			foreach ( $included_files as $included_relative_path ) {
-				if ( 0 === strpos( $included_relative_path, $excluded_relative_path ) ) {
-					unset( $excluded_files[ $excluded_file_index ] );
+		}
+
+		if ( $excluded ) {
+			// Get excluded files from the filter iterator.
+			$excluded_files = $filter_iterator->getExcludedFiles();
+
+			// Check all excluded directories and remove them from the excluded list if they contain included files.
+			foreach ( $excluded_files as $excluded_file_index => $excluded_relative_path ) {
+				if ( ! is_dir( $source_dir_path . $excluded_relative_path ) ) {
+					continue;
+				}
+				foreach ( $included_files as $included_relative_path ) {
+					if ( 0 === strpos( $included_relative_path, $excluded_relative_path ) ) {
+						unset( $excluded_files[ $excluded_file_index ] );
+					}
 				}
 			}
+
+			return $excluded_files;
 		}
 
-		return $excluded ? $excluded_files : $included_files;
+		return $included_files;
 	}
 
 	/**
@@ -561,6 +576,12 @@ private function get_size_format( $bytes, $decimals = 0 ) {
 		$size_key = floor( log( $bytes ) / log( 1000 ) );
 		$sizes    = [ 'B', 'KB', 'MB', 'GB', 'TB' ];
 
+		if ( is_infinite( $size_key ) ) {
+			$size_key = 0;
+		}
+
+		$size_key = (int) $size_key;
+
 		$size_format = isset( $sizes[ $size_key ] ) ? $sizes[ $size_key ] : $sizes[0];
 
 		// Display the size as a number.