Incorrect regex for floating point numbers

[diesirae68]

From manual page: https://www.php.net/manual/en/language.types.float.php
Hi! I think the regex for floating point numbers is wrong.

Please, replace

DNUM ([0-9]*(_[0-9]+)*[\.]{LNUM}) | ({LNUM}[\.][0-9]*(_[0-9]+)*)

with

DNUM (([0-9]+(_[0-9]+)*)?\.{LNUM}) | ({LNUM}\.([0-9]+(_[0-9]+)*)?)

The regex specified in the documentation accepts such numbers as correct, but they are incorrect:
_123.456
123._456

And need to fix this article in all languages. Thanx!

[nathansalter]

I actually think the REGEX can be simplified further to this:

LNUM = [0-9](_?[0-9]+)*
DNUM = [+-]?{LNUM}(\.{LNUM})?
EXPONENT_DNUM = {DNUM}[Ee]{DNUM}

Which might make it a bit easier to read

[Girgias]

I actually think the REGEX can be simplified further to this:

LNUM = [0-9](_?[0-9]+)*
DNUM = [+-]?{LNUM}(\.{LNUM})?
EXPONENT_DNUM = {DNUM}[Ee]{DNUM}

Which might make it a bit easier to read

This doesn't seem to accept -.5 as a valid float.

[diesirae68]

I actually think the REGEX can be simplified further to this:

LNUM = [0-9](_?[0-9]+)*
DNUM = [+-]?{LNUM}(\.{LNUM})?
EXPONENT_DNUM = {DNUM}[Ee]{DNUM}

Which might make it a bit easier to read

This doesn't seem to accept -.5 as a valid float.

You're right. But [+-]?(({LNUM})?\.{LNUM})|({LNUM}\.({LNUM})?) works correctly (this is a shorter expression than I originally wrote).

You can check this code:

$LNUM = '[0-9]+(_[0-9]+)*';
$DNUM = "[+-]?(({$LNUM})?\.{$LNUM})|({$LNUM}\.({$LNUM})?)";
var_dump(preg_match("/^{$DNUM}$/", '-.5'));

[Girgias]

This looks correct, can you make a PR?

[thegallagher]

Sorry to jump in here at the last minute, but I'd like to offer an alternative opinion on this.

I believe the purpose of this section of the documentation is to describe the structure of a PHP float literal. The description doesn't need to be useful for other purposes (eg. input validation) and it doesn't need to be a valid PCRE regular expression.

The grammar used by the PHP lexer is:

LNUM            [0-9]+(_[0-9]+)*
DNUM            ({LNUM}?"."{LNUM})|({LNUM}"."{LNUM}?)
EXPONENT_DNUM   (({LNUM}|{DNUM})[eE][+-]?{LNUM})

The above grammar or something derived from it should be used to correct the documentation.

While obviously correct, it is unclear:

• that both DNUM and EXPONENT_DNUM define complete float literals
• that LNUM defines only part of a float literal
• what LNUM and DNUM are

To clarify the above grammar and for consistency with the integer documentation, I offer the following alternative:

digits     : [0-9]+(_[0-9]+)*

decimal    : (digits? "." digits)
           | (digits "." digits?)

scientific : (digits | decimal) [eE][+-]? digits

float      : decimal
           | scientific

For me, the above grammar is easier to read and therefore better achieves the purpose I've described for the documentation, but I'd like to hear other thoughts.

[diesirae68]

Sorry to jump in here at the last minute, but I'd like to offer an alternative opinion on this.

I believe the purpose of this section of the documentation is to describe the structure of a PHP float literal. The description doesn't need to be useful for other purposes (eg. input validation) and it doesn't need to be a valid PCRE regular expression.

The grammar used by the PHP lexer is:

LNUM            [0-9]+(_[0-9]+)*
DNUM            ({LNUM}?"."{LNUM})|({LNUM}"."{LNUM}?)
EXPONENT_DNUM   (({LNUM}|{DNUM})[eE][+-]?{LNUM})

The above grammar or something derived from it should be used to correct the documentation.

While obviously correct, it is unclear:

• that both DNUM and EXPONENT_DNUM define complete float literals
• that LNUM defines only part of a float literal
• what LNUM and DNUM are

To clarify the above grammar and for consistency with the integer documentation, I offer the following alternative:

digits     : [0-9]+(_[0-9]+)*

decimal    : (digits? "." digits)
           | (digits "." digits?)

scientific : (digits | decimal) [eE][+-]? digits

float      : decimal
           | scientific

For me, the above grammar is easier to read and therefore better achieves the purpose I've described for the documentation, but I'd like to hear other thoughts.

Your description of the floating point number format would be ideal for documentation. I would choose this description.

[Girgias]

People, if you have suggestions, you can just send a PR.

[diesirae68]

People, if you have suggestions, you can just send a PR.

Done! :)

[diesirae68]

Do I need to submit a PR for documentation in other languages?

[Girgias]

Do I need to submit a PR for documentation in other languages?

You can, but this change is tracked via our RevCheck for all the languages via: http://doc.php.net

[Girgias]

Fixed via #2967