summaryrefslogtreecommitdiff
path: root/tests/phpunit/includes/StringUtilsTest.php
diff options
context:
space:
mode:
authorLuke Shumaker <LukeShu@sbcglobal.net>2014-01-28 09:50:25 -0500
committerLuke Shumaker <LukeShu@sbcglobal.net>2014-01-28 09:50:25 -0500
commit5744df39e15f85c6cc8a9faf8924d77e76d2b216 (patch)
treea8c8dd40a94d1fa0d5377566aa5548ae55a163da /tests/phpunit/includes/StringUtilsTest.php
parent4bb2aeca1d198391ca856aa16c40b8559c68daec (diff)
parent224b22a051051f6c2e494c3a2fb4adb42898e2d1 (diff)
Merge branch 'archwiki'
Conflicts: extensions/FluxBBAuthPlugin.php extensions/SyntaxHighlight_GeSHi/README extensions/SyntaxHighlight_GeSHi/SyntaxHighlight_GeSHi.class.php extensions/SyntaxHighlight_GeSHi/SyntaxHighlight_GeSHi.i18n.php extensions/SyntaxHighlight_GeSHi/SyntaxHighlight_GeSHi.php extensions/SyntaxHighlight_GeSHi/geshi/docs/CHANGES extensions/SyntaxHighlight_GeSHi/geshi/docs/THANKS extensions/SyntaxHighlight_GeSHi/geshi/docs/TODO extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/AbstractClass.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/AbstractClass_logo.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/AbstractMethod.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/AbstractPrivateClass.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/AbstractPrivateClass_logo.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/AbstractPrivateMethod.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Class.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Class_logo.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Constant.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Constructor.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Destructor.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Function.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Global.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/I.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Index.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Interface.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Interface_logo.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/L.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Lminus.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Lplus.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Method.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Page.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Page_logo.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/PrivateClass.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/PrivateClass_logo.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/PrivateMethod.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/PrivateVariable.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/StaticMethod.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/StaticVariable.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/T.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Tminus.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Tplus.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/Variable.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/blank.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/class_folder.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/file.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/folder.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/function_folder.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/next_button.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/next_button_disabled.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/package.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/package_folder.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/previous_button.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/previous_button_disabled.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/private_class_logo.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/tutorial.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/tutorial_folder.png extensions/SyntaxHighlight_GeSHi/geshi/docs/api/media/images/up_button.png extensions/SyntaxHighlight_GeSHi/geshi/docs/geshi-doc.html extensions/SyntaxHighlight_GeSHi/geshi/docs/geshi-doc.txt extensions/SyntaxHighlight_GeSHi/geshi/geshi.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/4cs.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/6502acme.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/6502kickass.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/6502tasm.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/68000devpac.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/abap.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/actionscript.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/actionscript3.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/ada.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/algol68.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/apache.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/applescript.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/apt_sources.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/asm.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/asp.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/autoconf.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/autohotkey.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/autoit.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/avisynth.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/awk.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/bascomavr.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/bash.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/basic4gl.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/bf.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/bibtex.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/blitzbasic.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/bnf.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/boo.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/c.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/c_loadrunner.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/c_mac.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/caddcl.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/cadlisp.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/cfdg.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/cfm.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/chaiscript.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/cil.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/clojure.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/cmake.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/cobol.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/coffeescript.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/cpp-qt.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/cpp.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/csharp.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/css.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/cuesheet.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/d.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/dcs.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/delphi.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/diff.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/div.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/dos.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/dot.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/e.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/ecmascript.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/eiffel.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/email.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/epc.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/erlang.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/euphoria.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/f1.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/falcon.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/fo.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/fortran.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/freebasic.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/fsharp.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/gambas.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/gdb.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/genero.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/genie.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/gettext.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/glsl.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/gml.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/gnuplot.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/go.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/groovy.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/gwbasic.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/haskell.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/hicest.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/hq9plus.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/html4strict.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/html5.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/icon.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/idl.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/ini.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/inno.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/intercal.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/io.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/j.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/java.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/java5.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/javascript.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/jquery.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/kixtart.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/klonec.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/klonecpp.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/latex.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/lb.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/lisp.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/llvm.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/locobasic.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/logtalk.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/lolcode.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/lotusformulas.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/lotusscript.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/lscript.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/lsl2.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/lua.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/m68k.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/magiksf.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/make.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/mapbasic.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/matlab.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/mirc.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/mmix.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/modula2.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/modula3.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/mpasm.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/mxml.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/mysql.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/newlisp.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/nsis.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/oberon2.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/objc.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/objeck.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/ocaml-brief.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/ocaml.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/oobas.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/oracle11.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/oracle8.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/oxygene.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/oz.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/pascal.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/pcre.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/per.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/perl.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/perl6.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/pf.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/php-brief.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/php.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/pic16.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/pike.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/pixelbender.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/pli.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/plsql.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/postgresql.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/povray.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/powerbuilder.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/powershell.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/proftpd.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/progress.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/prolog.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/properties.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/providex.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/purebasic.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/pycon.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/python.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/q.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/qbasic.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/rails.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/rebol.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/reg.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/robots.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/rpmspec.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/rsplus.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/ruby.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/sas.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/scala.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/scheme.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/scilab.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/sdlbasic.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/smalltalk.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/smarty.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/sql.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/systemverilog.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/tcl.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/teraterm.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/text.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/thinbasic.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/tsql.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/typoscript.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/unicon.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/uscript.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/vala.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/vb.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/vbnet.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/verilog.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/vhdl.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/vim.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/visualfoxpro.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/visualprolog.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/whitespace.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/whois.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/winbatch.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/xbasic.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/xml.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/xorg_conf.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/xpp.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/yaml.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/z80.php extensions/SyntaxHighlight_GeSHi/geshi/geshi/zxbasic.php
Diffstat (limited to 'tests/phpunit/includes/StringUtilsTest.php')
-rw-r--r--tests/phpunit/includes/StringUtilsTest.php147
1 files changed, 147 insertions, 0 deletions
diff --git a/tests/phpunit/includes/StringUtilsTest.php b/tests/phpunit/includes/StringUtilsTest.php
new file mode 100644
index 00000000..89759e5c
--- /dev/null
+++ b/tests/phpunit/includes/StringUtilsTest.php
@@ -0,0 +1,147 @@
+<?php
+
+class StringUtilsTest extends MediaWikiTestCase {
+
+ /**
+ * This tests StringUtils::isUtf8 whenever we have the mbstring extension
+ * loaded.
+ *
+ * @covers StringUtils::isUtf8
+ * @dataProvider provideStringsForIsUtf8Check
+ */
+ public function testIsUtf8WithMbstring( $expected, $string ) {
+ if ( !function_exists( 'mb_check_encoding' ) ) {
+ $this->markTestSkipped( 'Test requires the mbstring PHP extension' );
+ }
+ $this->assertEquals( $expected,
+ StringUtils::isUtf8( $string ),
+ 'Testing string "' . $this->escaped( $string ) . '" with mb_check_encoding'
+ );
+ }
+
+ /**
+ * This tests StringUtils::isUtf8 making sure we use the pure PHP
+ * implementation used as a fallback when mb_check_encoding() is
+ * not available.
+ *
+ * @covers StringUtils::isUtf8
+ * @dataProvider provideStringsForIsUtf8Check
+ */
+ public function testIsUtf8WithPhpFallbackImplementation( $expected, $string ) {
+ $this->assertEquals( $expected,
+ StringUtils::isUtf8( $string, /** disable mbstring: */true ),
+ 'Testing string "' . $this->escaped( $string ) . '" with pure PHP implementation'
+ );
+ }
+
+ /**
+ * Print high range characters as an hexadecimal
+ */
+ function escaped( $string ) {
+ $escaped = '';
+ $length = strlen( $string );
+ for ( $i = 0; $i < $length; $i++ ) {
+ $char = $string[$i];
+ $val = ord( $char );
+ if ( $val > 127 ) {
+ $escaped .= '\x' . dechex( $val );
+ } else {
+ $escaped .= $char;
+ }
+ }
+
+ return $escaped;
+ }
+
+ /**
+ * See also "UTF-8 decoder capability and stress test" by
+ * Markus Kuhn:
+ * http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt
+ */
+ public static function provideStringsForIsUtf8Check() {
+ // Expected return values for StringUtils::isUtf8()
+ $PASS = true;
+ $FAIL = false;
+
+ return array(
+ 'some ASCII' => array( $PASS, 'Some ASCII' ),
+ 'euro sign' => array( $PASS, "Euro sign €" ),
+
+ 'first possible sequence 1 byte' => array( $PASS, "\x00" ),
+ 'first possible sequence 2 bytes' => array( $PASS, "\xc2\x80" ),
+ 'first possible sequence 3 bytes' => array( $PASS, "\xe0\xa0\x80" ),
+ 'first possible sequence 4 bytes' => array( $PASS, "\xf0\x90\x80\x80" ),
+ 'first possible sequence 5 bytes' => array( $FAIL, "\xf8\x88\x80\x80\x80" ),
+ 'first possible sequence 6 bytes' => array( $FAIL, "\xfc\x84\x80\x80\x80\x80" ),
+
+ 'last possible sequence 1 byte' => array( $PASS, "\x7f" ),
+ 'last possible sequence 2 bytes' => array( $PASS, "\xdf\xbf" ),
+ 'last possible sequence 3 bytes' => array( $PASS, "\xef\xbf\xbf" ),
+ 'last possible sequence 4 bytes (U+1FFFFF)' => array( $FAIL, "\xf7\xbf\xbf\xbf" ),
+ 'last possible sequence 5 bytes' => array( $FAIL, "\xfb\xbf\xbf\xbf\xbf" ),
+ 'last possible sequence 6 bytes' => array( $FAIL, "\xfd\xbf\xbf\xbf\xbf\xbf" ),
+
+ 'boundary 1' => array( $PASS, "\xed\x9f\xbf" ),
+ 'boundary 2' => array( $PASS, "\xee\x80\x80" ),
+ 'boundary 3' => array( $PASS, "\xef\xbf\xbd" ),
+ 'boundary 4' => array( $PASS, "\xf2\x80\x80\x80" ),
+ 'boundary 5 (U+FFFFF)' => array( $PASS, "\xf3\xbf\xbf\xbf" ),
+ 'boundary 6 (U+100000)' => array( $PASS, "\xf4\x80\x80\x80" ),
+ 'boundary 7 (U+10FFFF)' => array( $PASS, "\xf4\x8f\xbf\xbf" ),
+ 'boundary 8 (U+110000)' => array( $FAIL, "\xf4\x90\x80\x80" ),
+
+ 'malformed 1' => array( $FAIL, "\x80" ),
+ 'malformed 2' => array( $FAIL, "\xbf" ),
+ 'malformed 3' => array( $FAIL, "\x80\xbf" ),
+ 'malformed 4' => array( $FAIL, "\x80\xbf\x80" ),
+ 'malformed 5' => array( $FAIL, "\x80\xbf\x80\xbf" ),
+ 'malformed 6' => array( $FAIL, "\x80\xbf\x80\xbf\x80" ),
+ 'malformed 7' => array( $FAIL, "\x80\xbf\x80\xbf\x80\xbf" ),
+ 'malformed 8' => array( $FAIL, "\x80\xbf\x80\xbf\x80\xbf\x80" ),
+
+ 'last byte missing 1' => array( $FAIL, "\xc0" ),
+ 'last byte missing 2' => array( $FAIL, "\xe0\x80" ),
+ 'last byte missing 3' => array( $FAIL, "\xf0\x80\x80" ),
+ 'last byte missing 4' => array( $FAIL, "\xf8\x80\x80\x80" ),
+ 'last byte missing 5' => array( $FAIL, "\xfc\x80\x80\x80\x80" ),
+ 'last byte missing 6' => array( $FAIL, "\xdf" ),
+ 'last byte missing 7' => array( $FAIL, "\xef\xbf" ),
+ 'last byte missing 8' => array( $FAIL, "\xf7\xbf\xbf" ),
+ 'last byte missing 9' => array( $FAIL, "\xfb\xbf\xbf\xbf" ),
+ 'last byte missing 10' => array( $FAIL, "\xfd\xbf\xbf\xbf\xbf" ),
+
+ 'extra continuation byte 1' => array( $FAIL, "e\xaf" ),
+ 'extra continuation byte 2' => array( $FAIL, "\xc3\x89\xaf" ),
+ 'extra continuation byte 3' => array( $FAIL, "\xef\xbc\xa5\xaf" ),
+ 'extra continuation byte 4' => array( $FAIL, "\xf0\x9d\x99\xb4\xaf" ),
+
+ 'impossible bytes 1' => array( $FAIL, "\xfe" ),
+ 'impossible bytes 2' => array( $FAIL, "\xff" ),
+ 'impossible bytes 3' => array( $FAIL, "\xfe\xfe\xff\xff" ),
+
+ 'overlong sequences 1' => array( $FAIL, "\xc0\xaf" ),
+ 'overlong sequences 2' => array( $FAIL, "\xc1\xaf" ),
+ 'overlong sequences 3' => array( $FAIL, "\xe0\x80\xaf" ),
+ 'overlong sequences 4' => array( $FAIL, "\xf0\x80\x80\xaf" ),
+ 'overlong sequences 5' => array( $FAIL, "\xf8\x80\x80\x80\xaf" ),
+ 'overlong sequences 6' => array( $FAIL, "\xfc\x80\x80\x80\x80\xaf" ),
+
+ 'maximum overlong sequences 1' => array( $FAIL, "\xc1\xbf" ),
+ 'maximum overlong sequences 2' => array( $FAIL, "\xe0\x9f\xbf" ),
+ 'maximum overlong sequences 3' => array( $FAIL, "\xf0\x8f\xbf\xbf" ),
+ 'maximum overlong sequences 4' => array( $FAIL, "\xf8\x87\xbf\xbf" ),
+ 'maximum overlong sequences 5' => array( $FAIL, "\xfc\x83\xbf\xbf\xbf\xbf" ),
+
+ 'surrogates 1 (U+D799)' => array( $PASS, "\xed\x9f\xbf" ),
+ 'surrogates 2 (U+E000)' => array( $PASS, "\xee\x80\x80" ),
+ 'surrogates 3 (U+D800)' => array( $FAIL, "\xed\xa0\x80" ),
+ 'surrogates 4 (U+DBFF)' => array( $FAIL, "\xed\xaf\xbf" ),
+ 'surrogates 5 (U+DC00)' => array( $FAIL, "\xed\xb0\x80" ),
+ 'surrogates 6 (U+DFFF)' => array( $FAIL, "\xed\xbf\xbf" ),
+ 'surrogates 7 (U+D800 U+DC00)' => array( $FAIL, "\xed\xa0\x80\xed\xb0\x80" ),
+
+ 'noncharacters 1' => array( $PASS, "\xef\xbf\xbe" ),
+ 'noncharacters 2' => array( $PASS, "\xef\xbf\xbf" ),
+ );
+ }
+}