Merge "Fix mime detection of easily-confused-with text/plain formats"

includes/MimeMagic.php

@@ -485,14 +485,6 @@ class MimeMagic {
 	 * by looking at the file extension. Typically, this method would be called on the
 	 * result of guessMimeType().
 	 *
-	 * Currently, this method does the following:
-	 *
-	 * If $mime is "unknown/unknown" and isRecognizableExtension( $ext ) returns false,
-	 * return the result of guessTypesForExtension($ext).
-	 *
-	 * If $mime is "application/x-opc+zip" and isMatchingExtension( $ext, $mime )
-	 * gives true, return the result of guessTypesForExtension($ext).
-	 *
 	 * @param string $mime The mime type, typically guessed from a file's content.
 	 * @param string $ext The file extension, as taken from the file name
 	 *
@@ -518,6 +510,12 @@ class MimeMagic {
 					".$ext is not a known OPC extension.\n" );
 				$mime = 'application/zip';
 			}
+		} elseif ( $mime === 'text/plain' && $this->findMediaType( ".$ext" ) === MEDIATYPE_TEXT ) {
+			// Textual types are sometimes not recognized properly.
+			// If detected as text/plain, and has an extension which is textual
+			// improve to the extension's type. For example, csv and json are often
+			// misdetected as text/plain.
+			$mime = $this->guessTypesForExtension( $ext );
 		}
 
 		if ( isset( $this->mMimeTypeAliases[$mime] ) ) {

includes/mime.info

@@ -65,6 +65,9 @@ text/plain	[TEXT]
 text/html application/xhtml+xml	[TEXT]
 application/xml text/xml	[TEXT]
 text	[TEXT]
+application/json	[TEXT]
+text/csv	[TEXT]
+text/tab-separated-values	[TEXT]
 
 application/zip application/x-zip	[ARCHIVE]
 application/x-gzip	[ARCHIVE]

includes/mime.types

@@ -35,6 +35,7 @@ application/x-gzip gz
 application/x-hdf hdf
 application/x-jar jar
 application/x-javascript js
+application/json json
 application/x-koan skp skd skt skm
 application/x-latex latex
 application/x-netcdf nc cdf
@@ -109,6 +110,7 @@ model/mesh msh mesh silo
 model/vrml wrl vrml
 text/calendar ics ifb
 text/css css
+text/csv csv
 text/html html htm
 text/plain txt
 text/richtext rtx

includes/upload/UploadBase.php

@@ -435,7 +435,7 @@ abstract class UploadBase {
 		}
 
 		$this->mFileProps = FSFile::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
-		$mime = $this->mFileProps['file-mime'];
+		$mime = $this->mFileProps['mime'];
 
 		if ( $wgVerifyMimeType ) {
 			# XXX: Missing extension will be caught by validateName() via getTitle()

tests/phpunit/includes/MimeMagicTest.php

@@ -0,0 +1,39 @@
+<?php
+class MimeMagicTest extends MediaWikiTestCase {
+
+	/** @var MimeMagic */
+	private $mimeMagic;
+
+	function setUp() {
+		$this->mimeMagic = MimeMagic::singleton();
+		parent::setUp();
+	}
+
+	/**
+	 * @dataProvider providerImproveTypeFromExtension
+	 * @param $ext String File extension (no leading dot)
+	 * @param $oldMime String Initially detected mime
+	 * @param $expectedMime String Mime type after taking extension into account
+	 */
+	function testImproveTypeFromExtension( $ext, $oldMime, $expectedMime ) {
+		$actualMime = $this->mimeMagic->improveTypeFromExtension( $oldMime, $ext );
+		$this->assertEquals( $expectedMime, $actualMime );
+	}
+
+	function providerImproveTypeFromExtension() {
+		return array(
+			array( 'gif', 'image/gif', 'image/gif' ),
+			array( 'gif', 'unknown/unknown', 'unknown/unknown' ),
+			array( 'wrl', 'unknown/unknown', 'model/vrml' ),
+			array( 'txt', 'text/plain', 'text/plain' ),
+			array( 'csv', 'text/plain', 'text/csv' ),
+			array( 'tsv', 'text/plain', 'text/tab-separated-values' ),
+			array( 'json', 'text/plain', 'application/json' ),
+			array( 'foo', 'application/x-opc+zip', 'application/zip' ),
+			array( 'docx', 'application/x-opc+zip', 'application/vnd.openxmlformats-officedocument.wordprocessingml.document' ),
+			array( 'djvu', 'image/x-djvu', 'image/vnd.djvu' ),
+			array( 'wav', 'audio/wav', 'audio/wav' ),
+		);
+	}
+
+}