symfony / polyfill Goto Github PK

Maximum execution time of 60 seconds exceeded in vendor/symfony/polyfill-mbstring/Mbstring.php on line 183

String of death (urlencoded): d?d???d?d%A6%CC??????d%A1%C0d?d?d%A1%E3

My guess - infinite while cycles

Our listener doesn't support phpunit 7.2 because of incompatible type-hints.
See #131 for inspiration (and scope.)

What do you think about adding a polyfill for intl-idn?

The project fails the PHP lint check. Here are the reproducible steps in the empty folder:

$ composer require symfony/polyfill-php54
...
  - Installing symfony/polyfill-php54 (v1.6.0)
...
$ find vendor -name "*.php" -exec php -l {} ';'
...
No syntax errors detected in vendor/symfony/polyfill-php54/Php54.php
Fatal error: Cannot declare class CallbackFilterIterator, because the name is already in use in vendor/symfony/polyfill-php54/Resources/stubs/CallbackFilterIterator.php on line 28
Errors parsing vendor/symfony/polyfill-php54/Resources/stubs/CallbackFilterIterator.php

I'm using these checks as part of the build for all Composer's libraries to verify whether the files are compatible with my PHP version, however I've got this fatal error instead.

This library was requested by symfony/http-foundation v2.8.32 which was requested by symfony/http-kernel and vipsoft/code-coverage-extension which is used for code coverage tests, so I believe v3.x is not supported yet.

I would be good to support normalizer_get_raw_decomposition function that appeared in PHP 7.3. I've tried to implement it but seems that existing decomposition data is already optimized to get the final decomposition. Example test case is available here.
Pinging @nicolas-grekas as he is the author of Normalizer polyfill.

I dont understand why symfony/polyfill-intl-icu requires symfony/intl and vice versa.

Shouldnt symfony/polyfill-intl-icu provide all the stubs, making symfony/polyfill-intl-messageformatter obsolete and remove symfony/intl as a polyfill dep.

Effectively symony/intl provides all stubs, except MessageFormatter.

It can be as simple as symfony/polyfill-intl + symfony/intl (specialized lib) no?

Hello,

There is the Error class in polyfill-php70. But why isn't there any Throwable interface ?

http://php.net/manual/fr/language.errors.php7.php

Thanks

Zend Server tries to implement compatibility with apc by registering compatibility functions with the same name as the APC functions.

http://files.zend.com/help/Zend-Server-6/content/zend_data_cache_-_configuration_directives.htm

Specifically, the option is "zend_datacache.apc_compatibility". It looks like it gets aliased to zend_shm_cache_store, which mostly works. Unfortunately, it doesn't support the array form of apc_store(), so every call results in errors:

"apc_store() expects parameter 1 to be string, array given"

It looks like this is a problem that other projects have dealt with, and it looks like it's one that's been going on for 8 years or so, so I'm not expecting Zend to fix it anytime soon.

My inclination would be to add a check to see if the parameter to apcu_add() or apcu_store() is an array, and if so to call the apc function individually for each item in the array. It would maintain compatibility with APC, and fix the issues with Zend Server. It's a little bit of a hack, though, and I figured I'd ask before submitting a PR.

Alternatively, I could add support for Zend Data Cache to the Symfony Cache Component.

hex2bin() returns false if NUL bytes are in decoded string. No error or exception is thrown which breaks code written for PHP 5.4+ that does not explicitly test for this condition, or which uses another test such as preg_match('/^([a-f0-9]{2})*$/', $str).

This breaks binary-safety and is inconsistent with PHP's behavior.

Otherwise, the subtree split repos provide a bad experience to people seeing them

I think it would be useful to have a polyfill for Apc, for the cases that apcu_* functions are available put apc_* is not.

This is the default situation when using Apcu > v5.

The polyfill would mostly be helpful in using older code that relies on apc_* and avoid installing the apcu_bc extension.

It there is agreement that this could be useful I could write it. I imagine it to be basically a complete copy of the Apcu polyfill, but with the u in the opposite place.

I would even be tempted to put it in the same polyfill making it simple to have code work in either situation.

Add in PHP 7.3 as Normalizer::normalize() argument for NFKC_Casefold normalization.

The current release of symfony/polyfill-mbstring has an invalid composer.json scheme. This has already been fixed in the master branch, however this change has not yet been released.

This break the composer validation of dependencies for any projects that has this polyfill included:

symfony/polyfill-mbstring is valid for simple usage with composer but has
strict errors that make it unable to be published as a package:
The property - suggests - is not defined and the definition does not allow additional properties

Could you create a new release including this change.

APCu Polyfill currently returns early if the legacy APC extension is missing, even though the APCUIterator wrapper is still useful with APCu 4.x (which is the only APCu that works with PHP 5.6).

Hello

I have the same error #63

[2016-08-24 15:03:51] lumen.ERROR: ErrorException: iconv_substr(): Detected an illegal character in input string in /var/www/app-storage/vendor/symfony/polyfill/src/Mbstring/Mbstring.php:417
Stack trace:
#0 [internal function]: Laravel\Lumen\Application->Laravel\Lumen\Concerns\{closure}(8, 'iconv_substr():...', '/var/www/heaven...', 417, Array)
#1 /var/www/app-storage/vendor/symfony/polyfill/src/Mbstring/Mbstring.php(417): iconv_substr('\x01\x00\x00\x00\x00\x01\xE5', 0, 1, 'ASCII')
#2 /var/www/app-storage/vendor/symfony/polyfill/src/Mbstring/bootstrap.php(36): Symfony\Polyfill\Mbstring\Mbstring::mb_substr('\x01\x00\x00\x00\x00\x01\xE5', 0, 1, 'ASCII')
#3 /var/www/app-storage/vendor/php-amqplib/php-amqplib/PhpAmqpLib/Wire/AMQPReader.php(163): mb_substr('\x01\x00\x00\x00\x00\x01\xE5', 0, 1, 'ASCII')

Using PHP 7.3 on Debian unstable, I see the following test failures:

There were 18 failures:

1) Symfony\Polyfill\Tests\Apcu\ApcuTest::testApcu
Failed asserting that false is true.

/home/km/projects/symfony-polyfill/tests/Apcu/ApcuTest.php:27

2) Symfony\Polyfill\Tests\Apcu\ApcuTest::testArrayCompatibility
Failed asserting that false matches expected 'value1'.

/home/km/projects/symfony-polyfill/tests/Apcu/ApcuTest.php:52

3) Symfony\Polyfill\Tests\Apcu\ApcuTest::testAPCUIterator
Failed asserting that false is true.

/home/km/projects/symfony-polyfill/tests/Apcu/ApcuTest.php:71

4) Symfony\Polyfill\Tests\Intl\Idn\IdnTest::testDecodeInvalid2003 with data set #0 ('xn--zcaccffbljjkknnoorrssuuxx...g9g.de', 'xn--zcaccffbljjkknnoorrssuuxx...g9g.de')
Failed asserting that two strings are identical.
--- Expected
+++ Actual
@@ @@
-'xn--zcaccffbljjkknnoorrssuuxxd5e0a0a3ae9c6a4a9bzdzdxdudwdxd2d2d8d0dse7d6dwe9dxeueweye4eyewe9e5ewkkewc9ftfpfplwexfwf4infvf2f6f6f7f8fpg8fmgngrgrgvgzgygxg3gyg1g3g5gykqg9g.de'
+'äöüßáàăâåãąāæćĉčċçďđéèĕêěëėęēğĝġģĥħíìĭîïĩįīıĵķĺľļłńňñņŋóòŏôőõøōœĸŕřŗśŝšşťţŧúùŭûůűũųūŵýŷÿźžżðþ.de'

/home/km/projects/symfony-polyfill/tests/Intl/Idn/IdnTest.php:83

5) Symfony\Polyfill\Tests\Intl\Idn\IdnTest::testDecodeInvalid2003 with data set #1 ('xn--zcaccffbljjkknnoorrssuuxx...vda.de', 'xn--zcaccffbljjkknnoorrssuuxx...g.þ.de')
Failed asserting that two strings are identical.
--- Expected
+++ Actual
@@ @@
-'xn--zcaccffbljjkknnoorrssuuxxd5e0a0a3ae9c8c1b0dxdvdvdxdvd3d0d6dyd8d5d4due7dveseuewe2eweue7e3esk9dxc7frf9e7kuevfuf1ilftf5f4f4f5f6fng6f8f9fpgpgtgxgwgvg1g2gzg1g3gvkog7g.þ.de'
+'äöüßáàăâåãąāæćĉčċçďđéèĕêěëėęēğĝġģĥħíìĭîïĩįīıĵķĺľļłńňñņŋóòŏôőõøōœĸŕřŗśŝšşťţŧúùŭûůűũųūŵýŷÿźžżð.þ.de'

/home/km/projects/symfony-polyfill/tests/Intl/Idn/IdnTest.php:83

6) Symfony\Polyfill\Tests\Intl\Idn\IdnTest::testUppercase2003 with data set #0 ('рф.RU', 'xn--p1ai.RU', 'рф.RU')
Failed asserting that two strings are identical.
--- Expected
+++ Actual
@@ @@
-'xn--p1ai.RU'
+'xn--p1ai.ru'

/home/km/projects/symfony-polyfill/tests/Intl/Idn/IdnTest.php:93

7) Symfony\Polyfill\Tests\Intl\Idn\IdnTest::testUppercase2003 with data set #1 ('GUANGDONG.广东', 'GUANGDONG.xn--xhq521b', 'GUANGDONG.广东')
Failed asserting that two strings are identical.
--- Expected
+++ Actual
@@ @@
-'GUANGDONG.xn--xhq521b'
+'guangdong.xn--xhq521b'

/home/km/projects/symfony-polyfill/tests/Intl/Idn/IdnTest.php:93

8) Symfony\Polyfill\Tests\Intl\Idn\IdnTest::testUppercase2003 with data set #2 ('renanGonçalves.COM', 'xn--renangonalves-pgb.COM', 'renangonçalves.COM')
Failed asserting that two strings are identical.
--- Expected
+++ Actual
@@ @@
-'xn--renangonalves-pgb.COM'
+'xn--renangonalves-pgb.com'

/home/km/projects/symfony-polyfill/tests/Intl/Idn/IdnTest.php:93

9) Symfony\Polyfill\Tests\Intl\Normalizer\NormalizerTest::testConstants
Failed asserting that Array &0 (
    'FORM_C' => 4
    'FORM_D' => 2
    'FORM_KC' => 5
    'FORM_KD' => 3
    'NFC' => 4
    'NFD' => 2
    'NFKC' => 5
    'NFKD' => 3
    'NONE' => 1
) is identical to Array &0 (
    'FORM_C' => 16
    'FORM_D' => 4
    'FORM_KC' => 32
    'FORM_KC_CF' => 48
    'FORM_KD' => 8
    'NFC' => 16
    'NFD' => 4
    'NFKC' => 32
    'NFKC_CF' => 48
    'NFKD' => 8
    'NONE' => 2
).

/home/km/projects/symfony-polyfill/tests/Intl/Normalizer/NormalizerTest.php:37

10) Symfony\Polyfill\Tests\Intl\Normalizer\NormalizerTest::testIsNormalized
Failed asserting that false is true.

/home/km/projects/symfony-polyfill/tests/Intl/Normalizer/NormalizerTest.php:51

11) Symfony\Polyfill\Tests\Intl\Normalizer\NormalizerTest::testNormalize
Failed asserting that false is identical to 'déjà훈쇼™'.

/home/km/projects/symfony-polyfill/tests/Intl/Normalizer/NormalizerTest.php:68

12) Symfony\Polyfill\Tests\Intl\Normalizer\NormalizerTest::testNormalizeConformance
Failed asserting that two strings are identical.
--- Expected
+++ Actual
@@ @@
-'Ḋ'
+'Ḋ'

/home/km/projects/symfony-polyfill/tests/Intl/Normalizer/NormalizerTest.php:109

13) Symfony\Polyfill\Tests\Apcu\ApcuTest::testApcu
Polyfills enabled, Failed asserting that false is true.

/home/km/projects/symfony-polyfill/tests/Apcu/ApcuTest.php:27

14) Symfony\Polyfill\Tests\Apcu\ApcuTest::testArrayCompatibility
Polyfills enabled, Failed asserting that false matches expected 'value1'.

/home/km/projects/symfony-polyfill/tests/Apcu/ApcuTest.php:52

15) Symfony\Polyfill\Tests\Apcu\ApcuTest::testAPCUIterator
Polyfills enabled, Failed asserting that false is true.

/home/km/projects/symfony-polyfill/tests/Apcu/ApcuTest.php:71

16) Symfony\Polyfill\Tests\Intl\Normalizer\NormalizerTest::testConstants
Polyfills enabled, Failed asserting that Array &0 (
    'FORM_C' => 4
    'FORM_D' => 2
    'FORM_KC' => 5
    'FORM_KD' => 3
    'NFC' => 4
    'NFD' => 2
    'NFKC' => 5
    'NFKD' => 3
    'NONE' => 1
) is identical to Array &0 (
    'FORM_C' => 16
    'FORM_D' => 4
    'FORM_KC' => 32
    'FORM_KC_CF' => 48
    'FORM_KD' => 8
    'NFC' => 16
    'NFD' => 4
    'NFKC' => 32
    'NFKC_CF' => 48
    'NFKD' => 8
    'NONE' => 2
).

/home/km/projects/symfony-polyfill/tests/Intl/Normalizer/NormalizerTest.php:37

17) Symfony\Polyfill\Tests\Intl\Normalizer\NormalizerTest::testIsNormalized
Polyfills enabled, Failed asserting that true is false.

/home/km/projects/symfony-polyfill/tests/Intl/Normalizer/NormalizerTest.php:53

18) Symfony\Polyfill\Tests\Intl\Normalizer\NormalizerTest::testNormalize
Polyfills enabled, Failed asserting that two strings are identical.
--- Expected
+++ Actual
@@ @@
-'déjà 훈쇼™'
+'déjà 훈쇼™'

/home/km/projects/symfony-polyfill/tests/Intl/Normalizer/NormalizerTest.php:78

FAILURES!
Tests: 1100, Assertions: 364410, Failures: 18, Skipped: 12.

Unsilenced deprecation notices (5)

  5x: Normalizer::NONE is obsolete with ICU 56 and above and will be removed in later PHP versions
    3x in NormalizerTest::testNormalize from Symfony\Polyfill\Tests\Intl\Normalizer
    2x in NormalizerTest::testIsNormalized from Symfony\Polyfill\Tests\Intl\Normalizer

Legacy deprecation notices (136)

There's also a Debian bug report at https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=911832

We were happy #94 was closed and 1.5.0 released.
However, the additional file Apcu.php is neither in the autoload.files property in composer.json nor is it required in bootstrap.php.

Maybe its best to simply require the file in
https://github.com/symfony/polyfill/blob/master/src/Apcu/bootstrap.php#L20
so it is not loaded when Zend Server Data Cache is not enabled?

This library is now used as a dependency in Twig 1.23 (and other projects are starting to use it too).
Not having a stable release means that Twig 1.23 is not installable by default in a composer project (you have to change the minimum stability or to require dev versions of symfony/polyfill-mbstring in your project to whitelist it)

The PHP function iconv() of the iconv extension allows both options to be specified at the same time. However, symfony/polyfill-iconv does not support that.

Example code:

$conv = function ($item) {
    return chr($item);
};

$isoString= implode(array_map($conv, [0x54,0xe4,0x73,0x74,0xdf]));

var_dump(iconv('ISO-8859-1', 'ASCII//IGNORE', $isoString));
var_dump(iconv('ISO-8859-1', 'ASCII//TRANSLIT', $isoString));
var_dump(iconv('ISO-8859-1', 'ASCII//TRANSLIT//IGNORE', $isoString));

Expected output (as with glibc iconv 2.19):

string(3) "Tst"
string(6) "Tastss"
string(6) "Tastss"

Actual output:

string(3) "Tst"
string(6) "Tastss"
PHP Notice:  iconv(): Wrong charset, conversion from `iso-8859-1' to `ascii//translit' is not allowed in /home/user/pdfparser/vendor/symfony/polyfill-iconv/Iconv.php on line 176
bool(false)

Tested with:

• symfony/polyfill-iconv (v1.5.0)
• symfony/polyfill-iconv (dev-master 1ea0e08)

This can be seen using the string "\xcc\x83\xc3\x92\xd5\x9b", which gets decomposed into "\xcc\x83\x4f\xcc\x80\xd5\x9b". On recompose, $lastUcls isn't reset on the ASCII "\x4f" so the string gets left in this decomposed form instead of the expected NFC normalized "\xcc\x83\xc3\x92\xd5\x9b".

The fix is reset the $lastUcls variable to zero on ASCII, see for instance the version of "Normalizer.php" I'm using in a fork of the WordPress plugin "tl-normalizer" https://github.com/gitlost/tl-normalizer/blob/master/Symfony/Normalizer.php#L184

This package could polyfill the NumberFormatter. This is something symfony/intl never tried to do, but what we've mostly accomplished in commerceguys/intl.
We don't try to replicate the rounding behavior, but that could be done, at least partially.

Happy to discuss it via other channels as well.

Hi all,

This is just a "service message".

For those people who use this library and use PHPCompatibility in their CI process, there is now a set of custom rulesets available which can be used to prevent false positives being thrown by PHPCompatibility for the native PHP functionality being polyfilled by the various Symfony polyfill-php#.# packages.

You can find the repo for the Symfony polyfill rulesets here on Github as well as on Packagist.

• https://github.com/PHPCompatibility/PHPCompatibilitySymfony
• https://packagist.org/packages/phpcompatibility/phpcompatibility-symfony

Hope someone will find it useful 😄

P.S.: If anyone is interested in helping us maintaining the rulesets and/or adding rulesets for the other Symfony polyfill packages, please open an issue in the repo.

Could paragonie/random_compat require be updated to "^1.0|^2.0"?

For compatibility with the intl extension. See https://github.com/gitlost/tl-normalizer/blob/master/Symfony/Normalizer.php#L52

Currently, for plural rules, the MessageFormatter polyfills uses the English rules for all locales (it ignores the locale).
To be consistent with what we do in symfony/intl (used by symfony/polyfill-intl-icu to implement NumberFormatter and DateFormatter), we should rather fail explicitly here (using the English rules for a different locale would not give the right result anyway)

Me again. The functions assume strlen() and substr() haven't been overloaded by mbstring. See https://github.com/gitlost/tl-normalizer/blob/master/Symfony/Normalizer.php#L57 and https://github.com/gitlost/tl-normalizer/blob/master/Symfony/Normalizer.php#L115

Is anyone aware of a polyfill for thr IMAP extension? Specifically the imap_mime_header_decode function, which is critical to properly decoding email headers.

In APCu 4.x, attempting to increment or decrement a key that doesn't exist fails. The workaround is storing a value of 1 or -1 to initialize the key after failing on inc/dec. I'm not sure if it would be part of the mission of this project to support things like that, but it's one of the APCu 4.x vs. 5.x challenges I deal with in LCache (in addition to the iterator).

Hi. I trying to use this amazing library in my work but still not understand what happens. I manually disable mbstring module in my php-7 setup for testing purposes. But all of this functions from this library still not defined.

I debug into the bootstrap.php and see that function_exists('mb_strlen') returns true. I hope that I'm doing something wrong but I can't explain that happening.

Why mb_strlen() defined if mbstring module is disabled?

$exists = function_exists('mb_strlen');
(bool)true

$phpVersion = phpversion();
7.0.12-1+deb.sury.org~trusty+1

$extensions = get_loaded_extensions();
array (
  0 => 'Core',
  1 => 'date',
  2 => 'libxml',
  3 => 'openssl',
  4 => 'pcre',
  5 => 'zlib',
  6 => 'filter',
  7 => 'hash',
  8 => 'Reflection',
  9 => 'SPL',
  10 => 'session',
  11 => 'standard',
  12 => 'cgi-fcgi',
  13 => 'mysqlnd',
  14 => 'PDO',
  15 => 'xml',
  16 => 'bcmath',
  17 => 'calendar',
  18 => 'ctype',
  19 => 'curl',
  20 => 'dom',
  21 => 'exif',
  22 => 'fileinfo',
  23 => 'ftp',
  24 => 'gd',
  25 => 'gettext',
  26 => 'iconv',
  27 => 'imagick',
  28 => 'imap',
  29 => 'json',
  30 => 'mcrypt',
  31 => 'memcache',
  32 => 'mysqli',
  33 => 'pdo_mysql',
  34 => 'Phar',
  35 => 'posix',
  36 => 'readline',
  37 => 'shmop',
  38 => 'SimpleXML',
  39 => 'soap',
  40 => 'sockets',
  41 => 'ssh2',
  42 => 'sysvmsg',
  43 => 'sysvsem',
  44 => 'sysvshm',
  45 => 'tokenizer',
  46 => 'wddx',
  47 => 'xmlreader',
  48 => 'xmlwriter',
  49 => 'xsl',
  50 => 'zip',
  51 => 'Zend OPcache',
  52 => 'xdebug',
)

missing in the PHP 7.3 polyfill

@carusogabriel maybe? :)

The polyfill-mbstring library depends on iconv, but does not require the polyfill-iconv library or ext-iconv. This can be an issue for things such as minimal PHP docker installs that include --configure = --disable-all before listing specific extensions to build.

Note: This is really for the subtree repo as installing the entire polyfill repo would not have this issue. symphony/console only installs polyfill-mbstring.

As suggested by @stof in php-amqplib/php-amqplib#72 (comment), I'm opening this ticket.

I can't really tell how to replicate this - I ran into it when installing RabbitMQ and not having mbstring extension installed.

Seems like theres being passed an "illegal" character in the string and its not being considered. This triggers a Notice, and in RabbitMQ case, it throws an exception.

PHP Notice:  iconv_strlen(): Detected an illegal character in input string in …/vendor/symfony/polyfill-mbstring/Mbstring.php on line 338

    public static function mb_strlen($s, $encoding = null)
    {
        $encoding = self::getEncoding($encoding);
        var_dump($s);

        return iconv_strlen($s, $encoding);
    }

string(14) "\000\000\000\000�\000
\000
\000    \000"

Installing the mbstring module prevents RabbitMQ from throwing the Exception, which I guess means this polyfill method is not really replacing the module (?)

They were added in PHP 7.3 and return the Decomposition_Mapping property of a character.

In the test-class (https://github.com/voku/swiftmailer/blob/5.x/tests/unit/Swift/Transport/Esmtp/Auth/NTLMAuthenticatorTest.php) from swiftmailer, there is already a simple polyfill for php 5.3. But when I use the "symfony-polyfill"-package, some of this tests will break... Here the test results via Travis-CI: https://travis-ci.org/voku/swiftmailer/jobs/117393133

Is it possible to add all the replaced PHP extensions in "replace"? In our case it is the 'mbstring' extension. If a package requires the mbstring extension, and I am installing this package, it should work.

Ref composer/composer#6525

phar:///usr/local/Cellar/composer/1.0.0-beta1/libexec/composer.phar/vendor/symfony/polyfill-mbstring/Mbstring.php undefined function Symfony\Polyfill\Mbstring\iconv() when called by composer cmd on oxs 10.10.5,I had checked the phar that it had add mbstring/iconv check, How to solve this problem?
Fatal error: Call to undefined function Symfony\Polyfill\Mbstring\iconv() in phar:///usr/local/Cellar/composer/1.0.0-beta1/libexec/composer.phar/vendor/symfony/polyfill-mbstring/Mbstring.php on line 513

https://github.com/symfony/polyfill/blob/master/src/Mbstring/Mbstring.php#L244

is it this way because it's not done yet? otherwise, is there any reason why it just returns array('UTF-8')?

thank you!

The current homebrew icu4c package is on version 58.1, while the Symfony intl polyfill package supports 57.1. This results in the following recommendation while running ./bin/symfony_requirements: intl ICU version installed on your system (58.1) does not match the ICU data bundled with Symfony (57.1).

Is it possible to support version 58.1 from icu4c?

Hello, I see a possible endianess issue on is_countable testsuite and latest polyfill release 1.8.0
https://autopkgtest.ubuntu.com/packages/php-symfony-polyfill/cosmic/s390x

Setting up autopkgtest-satdep (0) ...
(Reading database ... 73688 files and directories currently installed.)
Removing autopkgtest-satdep (0) ...
autopkgtest [16:09:03]: test command1: phpunit --bootstrap /usr/share/php/Symfony/Polyfill/autoload.php
autopkgtest [16:09:03]: test command1: [-----------------------
PHPUnit 7.2.7 by Sebastian Bergmann and contributors.

SSS............................................................  63 / 766 (  8%)
............................................................... 126 / 766 ( 16%)
............................................................... 189 / 766 ( 24%)
............................................................... 252 / 766 ( 32%)
............................................................... 315 / 766 ( 41%)
..............................................................S 378 / 766 ( 49%)
SS..F.SSS...................................................... 441 / 766 ( 57%)
............................................................... 504 / 766 ( 65%)
............................................................... 567 / 766 ( 74%)
............................................................... 630 / 766 ( 82%)
............................................................... 693 / 766 ( 90%)
............................................................... 756 / 766 ( 98%)
.....SSS..                                                      766 / 766 (100%)

Time: 2.22 seconds, Memory: 18.22MB

There was 1 failure:

1) Symfony\Polyfill\Tests\Php73\Php73Test::testIsCountable
Failed asserting that false is true.

/tmp/autopkgtest.Lf6G0W/build.nrL/src/tests/Php73/Php73Test.php:27

FAILURES!
Tests: 778, Assertions: 726412, Failures: 1, Skipped: 12.

I'm tempted to disable this part of the testsuite, since php 7.3 is not yet a real thing, and nothing is using is_countable in the archive, and this is a new test... but can you please have a look, or tell me if this is not a real issue?

the architecture is s390x

Natively, mb_convert_case('my_underscored_string', MB_CASE_TITLE, "UTF-8") would return My_Underscored_String, but this polyfill returns My_underscored_string instead.

This introduces a degree of uncertainty when using this polyfill for projects on systems without the native extension.

See #149 (comment) and following comments

Not sure if it's me or everyone, but when running composer update with minimum-stability:dev I now receive this error:

Script php aritsan clear-compiled handling the post-update-cmd event returned with an error

[RuntimeException]
  Error Output: PHP Fatal error:  Class 'APCIterator' not found in /path/to/app/vendor/symfony/polyfill-apcu/bootstrap.php on line 32

I'm guessing this has something to do with this recent commit

Increasing the minimum stability bypasses the issue. However, I would like to work in dev stability, is there any workaround or fix?

Your requirements could not be resolved to an installable set of packages.

Problem 1
- symfony/symfony 2.8.x-dev requires symfony/polyfill-intl-icu ~1.0 -> no matching package found.

Shouldn't it be 1.0.x-dev ?

If I try to transliterate the U+0096 character from UTF-8 (0xC2 0x96) to e.g. ASCII, the Iconv::iconv() function hangs due to an infinite loop.

Minimal example:

<?php
require_once('vendor/autoload.php');
use Symfony\Polyfill\Iconv\Iconv;
$text = "the \xC2\x96 character causes problems";
$converted = Iconv::iconv("UTF-8", "ASCII//TRANSLIT", $text);

I have identified that the problem is due to this line.

I tried the following patch, which makes it work and doesn't break any tests, but I don't fully understand how this function works, so it may have unintended consequences.

From 2e8ac7c206166a38526d10bb4df8eaa9c8bdc568 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Koszta=20D=C3=A1niel?= <[email protected]>
Date: Wed, 1 Aug 2018 16:44:38 +0200
Subject: [PATCH] Avoid infinite loop

---
 src/Iconv/Iconv.php | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/src/Iconv/Iconv.php b/src/Iconv/Iconv.php
index b9abecb..4d0ebba 100644
--- a/src/Iconv/Iconv.php
+++ b/src/Iconv/Iconv.php
@@ -695,6 +695,8 @@ final class Iconv
                     } else {
                         return false;
                     }
+                } else {
+                    continue;
                 }
 
                 $str = $uchr.substr($str, $i);
-- 
2.18.0

When encountering nested tokens, they are parsed only during formatting, not when compiling the pattern.
And the type of tokens are validated only during the rendering as well. But the native implementation detects such error during the instantiation: https://3v4l.org/Sf69D
This means that the error handling required for the polyfill is not the same than for the native implementation.

Perhaps following http://keepachangelog.com/

On 32bit builds of Ubuntu 14.04LTS (PHP 5.5.9), three functions are created with the wrong names.
See one of many bug reports.
There seems to be no plans to fix this in Ubuntu, so users of this LTS release cannot use the GZIP library.

Is there a place in this package for the necessary polyfills?

if (!function_exists('gzopen') && function_exists('gzopen64')) {
    function gzopen($filename , $mode = 'r', $use_include_path = 0 ) {
        return gzopen64($filename, $mode, $use_include_path);
    }
}
if (!function_exists('gzseek') && function_exists('gzseek64')) {
    function gzseek($zp , $offset, $whence = SEEK_SET) {
        return gzseek64($zp, $offset, $whence);
    }
}
if (!function_exists('gztell') && function_exists('gztell64')) {
    function gztell($zp) {
        return gztell64($zp);
    }
}

The \APCUIterator stubs is register to be autoloadable but it extend a class that may very well not exist.
Therfore calling :
class_exists('\APCUIterator');
will trigger a Fatal Error when \APCIterator isnt available.

For exemple, \Stash\Driver\Apc::isAvailable() rely on such a call.

I think, the stubs should only be know by Composer's autoloader if it's instanciable.

If feels like it's something that could be provided by the polyfill as well. I can try to do a PR if that's ok.

Note that there is also the exception JsonException missing.

Also the doc for the PHP 7.3 polyfill is outdated, given the entry of the docs for the polyfills, maybe those could be automated somehow to keep them in sync.

This can be seen using the string "\xe0\xbd\xb6\xe0\xbe\x81", which currently gets decomposed into "\xe0\xbe\xb2\xe0\xbe\x80\xe0\xbd\xb1\xe0\xbe\x80", but the reordered sequences "\xe0\xbe\x80" and "\xe0\xbd\xb1" should themselves be reordered, to give the expected NFD form "\xe0\xbe\xb2\xe0\xbd\xb1\xe0\xbe\x80\xe0\xbe\x80".

I've put in a hacky fix for this on the version of "Normalizer.php" I'm using in the "tl-normalizer" WP plugin fork, which recurses when a reordered sequence has a reordering combining class, see https://github.com/gitlost/tl-normalizer/blob/master/Symfony/Normalizer.php#L237 passim, but I'm sure there's a better way of doing it.