Fix breaking multi-line CSV values on reading (#5939)

Takshil-Kunadia · web-flow · commit d96b792a972f · 2024-04-26T13:50:29.000-07:00
* Refactor CSV iterator to use fgetcsv() to account for multiple line comma valued CSVs

* Add Unit tests for CSV Iterator
diff --git a/php/WP_CLI/Iterators/CSV.php b/php/WP_CLI/Iterators/CSV.php
@@ -54,14 +54,12 @@ public function next() {
 		$this->current_element = false;
 
 		while ( true ) {
-			$str = fgets( $this->file_pointer );
+			$row = fgetcsv( $this->file_pointer, self::ROW_SIZE, $this->delimiter );
 
-			if ( false === $str ) {
+			if ( false === $row ) {
 				break;
 			}
 
-			$row = str_getcsv( $str, $this->delimiter );
-
 			$element = [];
 			foreach ( $this->columns as $i => $key ) {
 				if ( isset( $row[ $i ] ) ) {
diff --git a/tests/WP_CLI/Iterators/CSVTest.php b/tests/WP_CLI/Iterators/CSVTest.php
@@ -0,0 +1,109 @@
+<?php
+
+namespace WP_CLI\Tests\CSV;
+
+use WP_CLI\Tests\TestCase;
+use WP_CLI\Iterators\CSV;
+
+class CSVTest extends TestCase {
+
+	public function test_it_can_iterate_over_a_csv_file() {
+		$filename = $this->create_csv_file(
+			array(
+				array( 'foo', 'bar' ),
+				array( 'baz', 'qux' ),
+			)
+		);
+
+		$expected = array(
+			0 => array(
+				'foo' => 'baz',
+				'bar' => 'qux',
+			),
+		);
+
+		foreach ( new CSV( $filename ) as $index => $row ) {
+			$this->assertEquals( $expected[ $index ], $row );
+		}
+	}
+
+	public function test_it_can_iterate_over_a_csv_file_with_custom_delimiter() {
+		$filename = $this->create_csv_file(
+			array(
+				array( 'foo|bar' ),
+				array( 'baz|qux' ),
+			),
+			'|'
+		);
+
+		$expected = array(
+			0 => array(
+				'foo|bar' => 'baz|qux',
+			),
+		);
+
+		foreach ( new CSV( $filename, '|' ) as $index => $row ) {
+			$this->assertEquals( $expected[ $index ], $row );
+		}
+	}
+
+	public function test_it_can_iterate_over_a_csv_file_with_multiple_lines_in_a_value() {
+		$filename = $this->create_csv_file(
+			array(
+				array( 'foo', "bar\nbaz" ),
+				array( 'qux', "quux\nquuz" ),
+			)
+		);
+
+		$expected = array(
+			0 => array(
+				'foo'      => 'qux',
+				"bar\nbaz" => "quux\nquuz",
+			),
+		);
+
+		foreach ( new CSV( $filename ) as $index => $row ) {
+			$this->assertEquals( $expected[ $index ], $row );
+		}
+	}
+
+	public function test_it_can_iterate_over_a_csv_file_with_multiple_lines_and_comma_in_a_value() {
+		$filename = $this->create_csv_file(
+			array(
+				array( 'foo', "bar\nbaz,qux" ),
+				array( 'quux', "quuz\ncorge,grault" ),
+			)
+		);
+
+		$expected = array(
+			0 => array(
+				'foo'          => 'quux',
+				"bar\nbaz,qux" => "quuz\ncorge,grault",
+			),
+		);
+
+		foreach ( new CSV( $filename ) as $index => $row ) {
+			$this->assertEquals( $expected[ $index ], $row );
+		}
+	}
+
+	private function create_csv_file( $data, $delimiter = ',' ) {
+		$filename = tempnam( sys_get_temp_dir(), 'wp-cli-tests-' );
+
+		$fp = fopen( $filename, 'wb' );
+
+		foreach ( $data as $row ) {
+			fputcsv( $fp, $row, $delimiter );
+		}
+
+		fclose( $fp );
+
+		register_shutdown_function(
+			function () use ( $filename ) {
+				unlink( $filename );
+			}
+		);
+
+		return $filename;
+	}
+}
