Patch add str begins and ends

[wkhudgins92]

This is the patch for RFC
https://wiki.php.net/rfc/add_str_begin_and_end_functions

This PR contains the implementations of str_begins() and str_ends() and also the phpt tests.

[wkhudgins92]

Not sure what the build failure is about. I was able to compile this on my local machine using make -j2 and I was able to run the compiled PHP and also execute make test.

[nikic]

The int i should go to the top of the function. That should fix the build issue.

[laruence]

how does str_begins different from strncmp/strncasecmp?

<?php
$str = "bEginningMiddleEnd";

var_dump(strcmp($str, "bEg", 3));
var_dump(strncasecmp($str, "beg", 3));

[laruence]

and how does str_begins and str_ends better than substr_compare?

<?php
$str = "bEginningMiddleEnd";

var_dump(substr_compare($str, "End", -3));
var_dump(substr_compare($str, "end", -3, 3, true));
?>

[ju1ius]

how does str_begins different from strncmp/strncasecmp?
and how does str_begins and str_ends better than substr_compare?

Without getting into micro-optimization,

str_ends($str, $end);

is easier to write but more importantly a lot easier to read than

substr_compare($str, $end, -strlen($end), strlen($end), true);

Just from the name of it anybody should understand what it does without having to check the documentation.

@wkhudgins92 It would have been even better to name them str_endswith() and str_startswith(), which read like: «Does this string starts with this other string?», whereas str_ends sounds more like «Does this string ends?».
My $0.2

[staabm]

IMO this can easily be achieved in userland with a simple function.

[ju1ius]

IMO this can easily be achieved in userland with a simple function.

@staabm Of course !

But this is a so common task that most programming languages have their native implementations:

• JS: String.prototype.startsWith
• Python str.startswith
• Ruby String.start_with?
• Go HasPrefix
• Swift String.hasPrefix
• Haskell startswith
• Why not PHP? 😉

[cmb69]

It would have been even better to name them str_endswith() and str_startswith(), […]

I would fully agree, if these were methods of a string object ($str->startsWith('foo') reads so fine), but it appears to be confusing for functions: str_startswith($str, 'foo') or str_startwith('foo', $str) :-/

[nikic]

Not sure I see the "functions" issue. "startswith" still imposes an ordering, that the first arg startswith the second arg. Unlike just "starts" which would imply the reverse argument order (first arg starts second arg).

[cmb69]

@nikic Haskell appears to differ. :-)

[nikic]

Heh. It kinda does make sense to invert the argument order in Haskell in particular, because this makes partial application more useful...

[cmb69]

Yes, that makes particular sense for Haskell (and other functional languages), but they could have called it startsthen.

[brzuchal]

There are already such implementations based on @nikic extension in here https://github.com/rossriley/php-scalar-objects#startswith scalar objects in PHP would be awesome feature, would love to have it in core. Great job on this extension from @nikic

[ju1ius]

but they could have called it starts then.

But they didn't. 😉

Wrt the order of arguments, as @nikic said, you're either in an OO-like language (or library, like glib/gobject) and you expect $this to be the first argument, or a functional-like language (or library) and you expect the data to be operated on to be the last argument. Both make perfect sense and it's not confusing as long as the api is consistent throughout the library.

Which raises the infamous issue:

function array_filter(array $array, callable $callback): array;
function array_map(callable $callback, array $array): array;

deserving a RFC on it's own ! 😆

[wkhudgins92]

Edited the code and modified the RFC (https://wiki.php.net/rfc/add_str_begin_and_end_functions) to reflect the comments brought up here and in the mailing list.

I split the functions into separate functions based on case sensitivity and switched the order of $str and $seach_value to be consistent with the other str_* functions. I also switched from _begins() and _ends() to the singular _begin() and _end(). In addition, the requested changes now pass the Travis CI tests.

[ChadSikorra]

Will these have mb_* equivalents?

[bepsvpt]

When the length of search_value is greater than str, maybe it will access out of bounds?

[blar]

can you change the the interface to bool str_begin(string $haytstack, $needle) and bool str_ends(string $haystack, string $needle) for consistency with other string function like stristr, strpos, strrchr?

[eddy8]

support~

[glensc]

Suggested change

	Checks if haystack begins with needle */
	Checks if haystack begins with a needle */

[azjezz]

Thanks for working on this 👍

It would have been even better to name them str_endswith() and str_startswith()

i quite agree with this ( or str_starts_with / str_ends_with instead ? ), but maybe that's just because I'm more familiar with hack standard library. ( HH\Lib\Str\starts_with(_ci) / HH\Lib\Str\ends_with(_ci) )

[theodorejb]

Is this RFC going to vote before the deadline for PHP 7.4 in a few days?

[wkhudgins92]

@theodorejb Yesterday marked 2 weeks of discussion, so I plan to move the RFC to a vote.

[wkhudgins92]

The RFC related to this has gone to a vote: https://wiki.php.net/rfc/add_str_begin_and_end_functions

[ElvenSpellmaker]

Does anyone know why the mb_* functions are being voted against in the poll? 🤔

[nikic]

The RFC was unfortunately :( :( :( declined, so I'm closing this PR.

[ElvenSpellmaker]

That's mad... All thanks to the silly 2/3 thing is seems. >50% is a majority, when will people realise that 😂 =|

[glensc]

the plain string vs mbstring vote difference seems stupid. probably shouldn't even have the opportunity to vote separately.

however, I'm personally looking for Symfony 5 string component:

• https://symfony.com/blog/new-in-symfony-5-0-string-component

[simivar]

So... is there any possibility to redo the voting and add those functions together with str_contains? It’s the same family. :/

[wkhudgins92]

So... is there any possibility to redo the voting and add those functions together with str_contains? It’s the same family. :/

I am going to re-raise the RFC+PR for only str_starts_with() and str_ends_with() @simivar

[glensc]

otoh, maybe instead of adding two sets of functions, add support for substring operators on the language level, like in Python (and Ruby?) do:

• $string[-2] = substring last two elements
• $string[:2] = substring first two elements
• $string[2:4] = substring from offsets 2 to 4

but this can't be done until the $string can be declared as ByteArray or UnicodeString, or maybe add also object support to have custom slicers support like ArrayAcess interface exists to emulate arrays on onjects.

[guilliamxavier]

@glensc: The "slice" syntactic sugar idea (which I have already seen proposed elsewhere BTW) is IMHO orthogonal to thoses functions (which are not implemented as $needle === substr($haystack, ...))