From: Brian Wolff <bawolff@users.mediawiki.org>
Date: Thu, 18 Aug 2011 05:45:04 +0000 (+0000)
Subject: follow-up r86169: unit tests for extraction of JPEG comment (COM) segments.
X-Git-Tag: 1.31.0-rc.0~28192
X-Git-Url: http://git.cyclocoop.org/%24self?a=commitdiff_plain;h=c0e419ce5e847fc5adc4a4e336a54f69d535ac93;p=lhc%2Fweb%2Fwiklou.git

follow-up r86169: unit tests for extraction of JPEG comment (COM) segments.
---

diff --git a/tests/phpunit/data/media/README b/tests/phpunit/data/media/README
index 224db9acc4..864e9ff48b 100644
--- a/tests/phpunit/data/media/README
+++ b/tests/phpunit/data/media/README
@@ -21,7 +21,8 @@ TastyCakes on English Wikipedia
 
 greyscale-na-png.png, rgb-png.png, Xmp-exif-multilingual_test.jpg
 greyscale-png.png, 1bit-png.png, Png-native-test.png, rgb-na-png.png,
-test.tiff, test.jpg
+test.tiff, test.jpg, jpeg-comment-multiple.jpg, jpeg-comment-utf.jpg,
+jpeg-comment-iso8859-1.jpg, jpeg-comment-binary.jpg
 Are all by Bawolff. I don't think they contain enough originality to
 claim copyright, but on the off chance they do, feel free to use them
 however you feel fit, without restriction.
diff --git a/tests/phpunit/data/media/jpeg-comment-binary.jpg b/tests/phpunit/data/media/jpeg-comment-binary.jpg
new file mode 100644
index 0000000000..b467fe43c6
Binary files /dev/null and b/tests/phpunit/data/media/jpeg-comment-binary.jpg differ
diff --git a/tests/phpunit/data/media/jpeg-comment-iso8859-1.jpg b/tests/phpunit/data/media/jpeg-comment-iso8859-1.jpg
new file mode 100644
index 0000000000..d9ffbac1db
Binary files /dev/null and b/tests/phpunit/data/media/jpeg-comment-iso8859-1.jpg differ
diff --git a/tests/phpunit/data/media/jpeg-comment-multiple.jpg b/tests/phpunit/data/media/jpeg-comment-multiple.jpg
new file mode 100644
index 0000000000..363c738513
Binary files /dev/null and b/tests/phpunit/data/media/jpeg-comment-multiple.jpg differ
diff --git a/tests/phpunit/data/media/jpeg-comment-utf.jpg b/tests/phpunit/data/media/jpeg-comment-utf.jpg
new file mode 100644
index 0000000000..d6d35b4bee
Binary files /dev/null and b/tests/phpunit/data/media/jpeg-comment-utf.jpg differ
diff --git a/tests/phpunit/includes/media/BitmapMetadataHandlerTest.php b/tests/phpunit/includes/media/BitmapMetadataHandlerTest.php
index a587bbe54b..148b90ac09 100644
--- a/tests/phpunit/includes/media/BitmapMetadataHandlerTest.php
+++ b/tests/phpunit/includes/media/BitmapMetadataHandlerTest.php
@@ -32,4 +32,19 @@ class BitmapMetadataHandlerTest extends MediaWikiTestCase {
 
 		$this->assertEquals( $expected, $meta['ImageDescription'] );
 	}
+
+	/**
+	 * Test for jpeg comments are being handled by
+	 * BitmapMetadataHandler correctly.
+	 *
+	 * There's more extensive tests of comment extraction in
+	 * JpegMetadataExtractorTests.php
+	 */
+	public function testJpegComment() {
+		$meta = BitmapMetadataHandler::Jpeg( $this->filePath .
+			'jpeg-comment-utf.jpg' );
+
+		$this->assertEquals( 'UTF-8 JPEG Comment — ¼',
+			$meta['JPEGFileComment'][0] );
+	}
 }
diff --git a/tests/phpunit/includes/media/JpegMetadataExtractorTest.php b/tests/phpunit/includes/media/JpegMetadataExtractorTest.php
new file mode 100644
index 0000000000..cc57925622
--- /dev/null
+++ b/tests/phpunit/includes/media/JpegMetadataExtractorTest.php
@@ -0,0 +1,32 @@
+<?php
+class JpegMetadataExtractorTest extends MediaWikiTestCase {
+
+	public function setUp() {
+		$this->filePath = dirname( __FILE__ ) . '/../../data/media/';
+	}
+
+	public function testUtf8Comment() {
+		$res = JpegMetadataExtractor::segmentSplitter( $this->filePath . 'jpeg-comment-utf.jpg' );
+		$this->assertEquals( array( 'UTF-8 JPEG Comment — ¼' ), $res['COM'] );
+	}
+	/** The file is iso-8859-1, but it should get auto converted */
+	public function testIso88591Comment() {
+		$res = JpegMetadataExtractor::segmentSplitter( $this->filePath . 'jpeg-comment-iso8859-1.jpg' );
+		$this->assertEquals( array( 'ISO-8859-1 JPEG Comment - ¼' ), $res['COM'] );
+	}
+	/** Comment values that are non-textual (random binary junk) should not be shown.
+	 * The example test file has a comment with a 0x5 byte in it which is a control character
+	 * and considered binary junk for our purposes.
+	 */
+	public function testBinaryCommentStripped() {
+		$res = JpegMetadataExtractor::segmentSplitter( $this->filePath . 'jpeg-comment-binary.jpg' );
+		$this->assertEmpty( $res['COM'] );
+	}
+	/* Very rarely a file can have multiple comments.
+	 *   Order of comments is based on order inside the file.
+	 */
+	public function testMultipleComment() {
+		$res = JpegMetadataExtractor::segmentSplitter( $this->filePath . 'jpeg-comment-multiple.jpg' );
+		$this->assertEquals( array( 'foo', 'bar' ), $res['COM'] );
+	}
+}