krakjoe / ustring Goto Github PK

What do you think about making strings immutable? I.e. methods would always return the result without modifying the original object.

It would make much more sense IMO.

For example:

if ($str->toUpper() === 'HELLO') {
    // ...
}

Here $str would be modified, which is confusing.

Because this would be really useful.

Hi,

The indentation style of this project is starting to look a bit funky, so can we agree on a crude coding standard we should be following?

I'm used to php-src style (tabs in most places), but I'm willing to compromise on a sane alternative.

C:\php-sdk\php70dev>nmake php_ustring.dll

Microsoft (R) Program Maintenance Utility Version 11.00.61030.0
Copyright (C) Microsoft Corporation. All rights reserved.
...
ustring.obj : error LNK2001: unresolved external symbol _zend_new_interned_string
ustring.obj : error LNK2001: unresolved external symbol _std_object_handlers
ustring.obj : error LNK2001: unresolved external symbol _compiler_globals
C:\php-sdk\php70dev\Release\php_ustring.dll : fatal error LNK1120: 3 unresolved externals
NMAKE : fatal error U1077: '"C:\Program Files (x86)\Microsoft Visual Studio 11.0\VC\BIN\link.exe"' : return code '0x460'
Stop.

I can fix it here later but just in case you are on it:

in php_ustring_cast:

Z_STR_P(zwrite) = STR_ALLOC(length+1, 0); already sets the length correctly and it should not be overridden afterwards. It leads to crash (wrong buffer size given to ICU > crash).

One important thing needing doing is making sure indexOf, split, startsWith and so on work properly with empty strings, i.e. pretending there's one between each actual codepoint. Otherwise it's harder to write userland string-handling functions, because you need to work around incorrectly-implemented boundary cases :(

Hello :-),

The Hoa\String library provides a class, called Hoa\String\String, that helps to manipulate UTF-8 strings. Its API has been designed during several months. It could be a source of inspiration for your work.

If you want some clarifications, ping me.

Since this doesn't operate on grapheme clusters, we should refer to dealing with codepoints, not characters.

In particular, documentation comments need changing, and charAt should be codepointAt.

ICU supports this in UnicodeString, but our API doesn't. We need to allow specifying the locale, because toUpper and toLower will behave differently for different locales, e.g. in Turkish, I becomes ı in lower case, not i.

C'mon, strings gotta have one. Either split or explode, I'd go with the former as it's a more common name in other languages, and it's the verb the manual uses to describe what explode does.

I'd suggest this signature (pseudo-code):

UString::split(UString $delimeter, int $limit = NULL): Array<UString>

The optional $limit parameter is a number. If specified, then the string is only split $limit times. This is a really useful feature.

Examples:

(new UString("1,2,3,4"))->split(new UString(","));
// => ["1", "2", "3", "4"]

(new UString("1,2,3,4"))->split(new UString(","), 1);
// => ["1", "2,3,4"]

(new UString("1,2,3,4"))->split(new UString(","), 2);
// => ["1", "2", "3,4"]

(new UString(",1,2,,3,4,"))->split(new UString(","));
// => ["", "1", "2", "", "3", "4", ""]

One way to implement it (in Game Maker Language, it's C-like but strings indexed from 1): https://github.com/Medo42/Gang-Garrison-2/blob/master/Source/gg2/Scripts/split.gml

I would like to see the API more feature safe to other encodings/charsets.

• rename [get|set][Default]Codepage to [get|set][Default]Encoding
• remove the U from UString (or use a different name)

$german = new UString("T\xD4st", 'ISO-8859-15');

Thoughts?

Currently, string reversal works on code points, it doesn't care what kind. So it won't reverse strings containing combining characters properly.

We could quite simply implement the Missy Elliot algorithm that @mathiasbynens came up with, and then it'd do what people expect.

Thoughts?

stuff already exists in the class Normalizer but would be handy to have it in this class as well. could look like JS's normalize()
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/String/normalize

echo u($str)->normalize(UString::NFKC)->subString(0, 4);

if it's kept, it should pad to the "visible width", ergo should take count of combining glyphs

which characters will trim() trim?

whitespace is very subjective and different in every language, including unicode

i would use C#'s whitespace definition as default (https://msdn.microsoft.com/en-us/library/t809ektx(v=vs.110).aspx) plus the null byte as in php's trim()
but i would also add an optional parameter "$charset" so that the behavior would be completely customizable

UString::trim(UString $charset = null)

they are redundant with substring(). there are better ways to avoid unnecessary copies [stringbuilder is one, or interning], provided that we care of this kind of optimizations