From 4574035ec43cabf050a7ab036bbba1cd2447a18f Mon Sep 17 00:00:00 2001 From: Urban Müller Date: Thu, 22 Mar 2012 18:48:52 +0000 Subject: check encoding in Q() and _tag(), adapt tests to utf-8 --- tests/it_html.t | 42 +++++++++++++++++++++++++++--------------- 1 file changed, 27 insertions(+), 15 deletions(-) (limited to 'tests/it_html.t') diff --git a/tests/it_html.t b/tests/it_html.t index 0def431..c955359 100755 --- a/tests/it_html.t +++ b/tests/it_html.t @@ -4,6 +4,7 @@ # Tests for html.class # Traditional html generation +ini_set('default_charset', "utf-8"); new it_html(array('htmltype' => "html")); is( @@ -95,30 +96,22 @@ is(

Wolken

'), - ' swisspics posted < < & yesterday a photo tag missmatch:

', + ' swisspics posted < < ä & yesterday a photo tag missmatch:

', 'it_html::sanitize tag soup' ); is( it_html::sanitize('q←x'), - "q←x", + "q←x", 'it_html::sanitize preserve numeric entities' ); -it_html::configure(array('charset' => "utf-8")); is( it_html::sanitize('qüx'), "q\xc3\xbcx", 'it_html::sanitize with utf-8' ); -it_html::configure(array('charset' => "iso-8859-1")); -is( - it_html::sanitize('qüx'), - "q\xfcx", - 'it_html::sanitize with latin1' -); - is( it_html::sanitize('a
b'), "a
b", @@ -126,8 +119,8 @@ is( ); is( - U("/foo.html", array('bar' => array('gna' => 42, 'qux' => array('quux' => "", 'gnp' => "fasel")))), - '/foo.html?bar[gna]=42&bar[qux][quux]=%3CZ%FCrich%3E&bar[qux][gn%F6p]=fasel', + U("/foo.html", array('bar' => array('gna' => 42, 'qux' => array('quux' => "", 'gnöp' => "fasel")))), + '/foo.html?bar[gna]=42&bar[qux][quux]=%3CZ%C3%BCrich%3E&bar[qux][gn%C3%B6p]=fasel', 'U() with nested arrays' ); @@ -149,11 +142,30 @@ is( 'U() converting of \ to /' ); -is(it_html::entity_decode("ä"), ""); -is(it_html::entity_decode("’"), "'"); +is(it_html::entity_decode("ä"), "ä"); is(it_html::entity_decode("J"), "J"); is(it_html::entity_decode("J"), "J"); -is(it_html::entity_decode("࿿"), " "); is(it_html::entity_decode("A"), "A"); + + +# +# check transliterations in iso-8859-1 +# + +it_html::configure(array('charset' => "iso-8859-1")); + +is( + it_html::sanitize('qüx'), + "q\xfcx", + 'it_html::sanitize with latin1' +); + +is( + it_html::sanitize('q←x'), + "q←x", + 'it_html::sanitize preserve non-decodable numeric entities' +); +is(it_html::entity_decode("’"), "'"); +is(it_html::entity_decode("࿿"), " "); is(it_html::entity_decode("ϧ"), " "); ?> -- cgit v1.2.3 From b7200b739ff651a7647d2d666e3674a7fe3cb6e2 Mon Sep 17 00:00:00 2001 From: Urban Müller Date: Mon, 26 Mar 2012 15:11:39 +0000 Subject: fixed it_html::fix_encoding --- tests/it_html.t | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'tests/it_html.t') diff --git a/tests/it_html.t b/tests/it_html.t index c955359..a576b47 100755 --- a/tests/it_html.t +++ b/tests/it_html.t @@ -168,4 +168,16 @@ is( is(it_html::entity_decode("’"), "'"); is(it_html::entity_decode("࿿"), " "); is(it_html::entity_decode("ϧ"), " "); + +is(it_html::fix_encoding("Meier"), "Meier"); +is(it_html::fix_encoding("Müller"), "Müller"); +is(it_html::fix_encoding("Aslı"), "Aslı"); +is(it_html::fix_encoding("é»"), "é»"); + +is(it_html::fix_encoding(utf8_encode("Müller"), true), "Müller", "double encoded latin1"); # Double encoded latin1 +is(it_html::fix_encoding(utf8_encode("Aslı"), true), "Aslı"); # Double encoded non-latin1 +is(it_html::fix_encoding(utf8_encode("é»"), true), "é»"); # Double encoded special combination + +is(it_html::fix_encoding(utf8_decode("Müller"), true), "Müller"); # Incorrectly decoded latin1 + ?> -- cgit v1.2.3 From bbcc6615dc5316a73f07b262950ac18d50c80497 Mon Sep 17 00:00:00 2001 From: Christian Schneider Date: Mon, 26 Mar 2012 15:16:01 +0000 Subject: Improve it_html tests --- tests/it_html.t | 22 +++++++++++----------- 1 file changed, 11 insertions(+), 11 deletions(-) (limited to 'tests/it_html.t') diff --git a/tests/it_html.t b/tests/it_html.t index a576b47..b1d271e 100755 --- a/tests/it_html.t +++ b/tests/it_html.t @@ -165,19 +165,19 @@ is( "q←x", 'it_html::sanitize preserve non-decodable numeric entities' ); -is(it_html::entity_decode("’"), "'"); -is(it_html::entity_decode("࿿"), " "); -is(it_html::entity_decode("ϧ"), " "); +is(it_html::entity_decode("’"), "'", "it_html::entity_decode numeric decimal entity"); +is(it_html::entity_decode("࿿"), " ", "it_html::entity_decode invalid numeric hex entity"); +is(it_html::entity_decode("ϧ"), " ", "it_html::entity_decode invalid numeric decimal entity"); -is(it_html::fix_encoding("Meier"), "Meier"); -is(it_html::fix_encoding("Müller"), "Müller"); -is(it_html::fix_encoding("Aslı"), "Aslı"); -is(it_html::fix_encoding("é»"), "é»"); +is(it_html::fix_encoding("Meier"), "Meier", "it_html::fix_encoding ascii"); +is(it_html::fix_encoding("Müller"), "Müller", "it_html::fix_encoding utf-8 latin1"); +is(it_html::fix_encoding("Aslı"), "Aslı", "it_html::fix_encoding utf-8 non-latin1"); +is(it_html::fix_encoding("é»"), "é»", "it_html::fix_encoding utf-8 latin1 special combination"); -is(it_html::fix_encoding(utf8_encode("Müller"), true), "Müller", "double encoded latin1"); # Double encoded latin1 -is(it_html::fix_encoding(utf8_encode("Aslı"), true), "Aslı"); # Double encoded non-latin1 -is(it_html::fix_encoding(utf8_encode("é»"), true), "é»"); # Double encoded special combination +is(it_html::fix_encoding(utf8_encode("Müller"), true), "Müller", "it_html::fix_encoding double encoded latin1"); +is(it_html::fix_encoding(utf8_encode("Aslı"), true), "Aslı", "it_html::fix_encoding double encoded non-latin1"); +is(it_html::fix_encoding(utf8_encode("é»"), true), "é»", "it_html::fix_encoding double encoded latin1 special combination"); -is(it_html::fix_encoding(utf8_decode("Müller"), true), "Müller"); # Incorrectly decoded latin1 +is(it_html::fix_encoding(utf8_decode("Müller"), true), "Müller", "it_html::fix_encoding incorrectly encoded latin1"); ?> -- cgit v1.2.3 From b1c0b4946572027c8de564730a89ec584c830bf3 Mon Sep 17 00:00:00 2001 From: Christian Schneider Date: Wed, 28 Mar 2012 13:00:39 +0000 Subject: Added it::any2utf8, fixed it::replace fast path to add u modified, added error reporting for invalid utf-8 input to it::match and it::replace --- tests/it_html.t | 1 + 1 file changed, 1 insertion(+) (limited to 'tests/it_html.t') diff --git a/tests/it_html.t b/tests/it_html.t index b1d271e..770d11a 100755 --- a/tests/it_html.t +++ b/tests/it_html.t @@ -153,6 +153,7 @@ is(it_html::entity_decode("A"), "A"); # it_html::configure(array('charset' => "iso-8859-1")); +ini_set('default_charset', "iso-8859-1"); is( it_html::sanitize('qüx'), -- cgit v1.2.3 From a60fb69ff8e7755ab969298ab85a2b63bbcc8b14 Mon Sep 17 00:00:00 2001 From: Christian Weber Date: Fri, 30 Mar 2012 16:13:15 +0000 Subject: add GREEN HEART unit test for it_html::fix_encoding() --- tests/it_html.t | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tests/it_html.t') diff --git a/tests/it_html.t b/tests/it_html.t index 770d11a..2323e5e 100755 --- a/tests/it_html.t +++ b/tests/it_html.t @@ -181,4 +181,6 @@ is(it_html::fix_encoding(utf8_encode("é»"), true), "é»", "it_html::fix_encod is(it_html::fix_encoding(utf8_decode("Müller"), true), "Müller", "it_html::fix_encoding incorrectly encoded latin1"); +is(it_html::fix_encoding("a💚b"), "a💚b", "it_html::fix_encoding incorrectly encoded GREEN HEART"); + ?> -- cgit v1.2.3 From 3242f516a6c3e89fa898d96927746447e15d7750 Mon Sep 17 00:00:00 2001 From: Christian Weber Date: Fri, 30 Mar 2012 16:17:08 +0000 Subject: correct GREEN HEART test description --- tests/it_html.t | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tests/it_html.t') diff --git a/tests/it_html.t b/tests/it_html.t index 2323e5e..174c487 100755 --- a/tests/it_html.t +++ b/tests/it_html.t @@ -181,6 +181,6 @@ is(it_html::fix_encoding(utf8_encode("é»"), true), "é»", "it_html::fix_encod is(it_html::fix_encoding(utf8_decode("Müller"), true), "Müller", "it_html::fix_encoding incorrectly encoded latin1"); -is(it_html::fix_encoding("a💚b"), "a💚b", "it_html::fix_encoding incorrectly encoded GREEN HEART"); +is(it_html::fix_encoding("a💚b"), "a💚b", "it_html::fix_encoding correctly handles 4-byte utf-8 character GREEN HEART"); ?> -- cgit v1.2.3