Methods for converting mixed values to specific data types #17295

jozefgrencik · 2023-09-20T20:58:21Z

This PR introduces a set of methods for converting mixed values to string/int/bool data types, ensuring a type-safe approach to data manipulation. This utility is useful for safely narrowing down the data types.

Look at the discussion:
#17177 (comment)

https://en.wikipedia.org/wiki/IEEE_754
https://www.h-schmidt.net/FloatConverter/IEEE754.html https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Number/MAX_SAFE_INTEGER

This PR introduces a set of methods for converting mixed values to string/int/bool data types, ensuring a type-safe approach to data manipulation. This utility is useful for safely narrowing down the data types. cakephp#17177 (comment) https://en.wikipedia.org/wiki/IEEE_754 https://www.h-schmidt.net/FloatConverter/IEEE754.html https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Number/MAX_SAFE_INTEGER

jozefgrencik · 2023-09-20T20:58:50Z

The class name is derived from the native PHP function filter_var().
It hasn't been tested how the functions will behave on 32-bit systems. Does CakePhp support 32-bit systems?

Notes on Unconventional Code:

Method: toString

bool -> string:

Handling of true is same as (string)true === '1'
Handling of false differs from standard PHP typecasting: (string)false === '', but toString(false) === '0' ('0' as opposed to '1').

float -> string:

Handling of NaN differs from standard PHP typecasting: (string)NaN === 'NAN', but toString(NaN) === null
Handling of (+/-)infinity differs from standard PHP typecasting: (string)INF === 'INF', but toString(INF) === null.
For other floats, attempts are made to convert via json_encode(), and if unsuccessful, then through sprintf(). For sprintf, PHP_FLOAT_DIG + 3 is used as the number of decimal places. Further experimentation may be needed to determine the optimal value.

object -> string:

For objects, is used similar approach like is in h() function in functions.php.

Method: toInt

float -> int:

Values are rounded as per default PHP behavior, e.g., (int)5.6 === 5
Outside the range -(2^53)+1 to 2^53-1, null is returned. These numbers represent the maximum safe integers that can be represented without losing precision. Beyond this numerical limit, the equality (int)9007199254740993.0 === 9007199254740992 returns true
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Number/MAX_SAFE_INTEGER

Failed asserting that two strings are identical. --- Expected +++ Actual @@ @@ -'10:10 AM' +'10:10 AM'

markstory

I like the direction this is headed. Nice work on the tests.

src/Utility/Filter.php

ADmad · 2023-09-21T03:24:51Z

src/Utility/Filter.php

+ * Redistributions of files must retain the above copyright notice.
+ *
+ * @copyright     Copyright (c) Cake Software Foundation, Inc. (https://cakefoundation.org)
+ * @since         5.0.1


Suggested change

* @since 5.0.1

* @since 5.1.0

Patch releases only contain bug fixes, new features need to go in the next minor release.

thx. Fixed in 01ec52d

…s Cake\Core\toString.

ADmad · 2023-09-22T02:52:02Z

src/Core/functions.php

@@ -325,3 +326,119 @@ function deprecationWarning(string $version, string $message, int $stackFrame =
        trigger_error($message, E_USER_DEPRECATED);
    }
 }
+
+if (!function_exists('Cake\Core\toString')) {


This check is not necessary, no one else should be defining this namespaced function. Such checks are necessary only when defining global functions.

thx. Fixed in 112b6d8

are these checks necessary for the remaining functions in functions.php?

markstory · 2023-09-25T01:57:20Z

src/Core/functions.php

+        try {
+            $return = json_encode($value, JSON_THROW_ON_ERROR);
+        } catch (JsonException) {
+            $return = null;


What are the advantages of using json_encode for a float over a simpler method like printf()?

The new function toString() combines the advantages of json_encode() and sprintf().

raw input (string)$value json_encode() sprintf() toString()

5.0 5 5 5 5

5.00000003 5.00000003 5.00000003 5.000000029999999818 5.00000003

55666666666.56 55666666666.56 55666666666.56 55666666666.55999755859375 55666666666.56

9223372036778.2233 9223372036778.2 9223372036778.223 9223372036778.22265625 9223372036778.223

88888888888888888888888 8.8888888888889E+22 8.888888888888889e+22 88888888888888890753024 88888888888888890753024

Test:

public function testJsonEncodeVsSprintf(): void { $a = [ 5.0, 5.00000003, 55666666666.56, 9223372036778.2233, 88888888888888888888888, ]; foreach ($a as $value) { $php = (string)$value; $json = json_encode($value); $sprint = rtrim(sprintf('%.' . (PHP_FLOAT_DIG + 3) . 'F', $value), '.0'); $toString = toString($value); echo '| ' . implode(' | ', [$php, $json, $sprint, $toString]) . ' |' . PHP_EOL; } $this->markTestSkipped(); }

markstory · 2023-09-25T02:15:01Z

src/Core/functions.php

+         * @link https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Number/MAX_SAFE_INTEGER
+         * 9007199254740991 = 2^53-1 = the maximum safe integer that can be represented without losing precision.


I'm not sure this same constraint applies to PHP. We should be able to go to 2^64-1 or PHP_INT_MAX

Yes, this constant also has the same value in PHP. I'm attaching a test that will find the first values that are no longer correctly typecasted.

Test:

public function testMargin(): void { $margin = 9007199254740991; $errors = 0; for ($i = $margin - 10_000_000; $i <= $margin + 10_000; $i++) { if ((int)(float)$i !== $i) { ++$errors; echo $i . PHP_EOL; if ($errors > 20) { break; } } } $this->markTestSkipped(); }

Results:

9007199254740993 9007199254740995 9007199254740997 9007199254740999 9007199254741001 9007199254741003 9007199254741005 9007199254741007 9007199254741009 9007199254741011 9007199254741013 9007199254741015 9007199254741017 9007199254741019 9007199254741021 9007199254741023 9007199254741025 9007199254741027 9007199254741029 9007199254741031 9007199254741033

I found in PHP documentation that it might not always be the same constant. It might be something else on exotic systems.

Floating point numbers have limited precision. Although it depends on the system, PHP typically uses the IEEE 754 double precision format..
https://www.php.net/manual/en/language.types.float.php

So I changed af822bb the logic to use (int)$value, but around PHP_INT_MIN/PHP_INT_MAX, it returns very strange values.
Beyond 2^53-1, there are type errors like 9007199254740993.0 ==> 9007199254740992

But I'm okay with this solution. Such large numbers are not used often. And programmers rarely take care of them.

markstory · 2023-09-25T02:18:45Z

src/Core/functions_global.php

+     * @return ?string Returns the string representation of the value, or null if the value is not a string.
+     * @since 5.1.0
+     */
+    function toString(mixed $value): ?string


I'm not sure we should be adding more global functions. When we introduced namespace functions in 5.0 and 4.5, we retained global function aliases as a way to make upgrading easier. With new functions, there is no concern around compatibility, but there could be conflicts with other libraries that define global functions.

oki. removed in 24c80f7

I personally prefer having these as global functions too. They are not autoloaded when using individual packages and for the app too users have the option to opt out.

markstory

Sorry for the delay in coming back to this. I think this is ready to merge once the few remaining redundant imports are cleaned up.

markstory · 2023-12-31T03:29:54Z

src/Core/functions_global.php

@@ -23,6 +23,9 @@
 use function Cake\Core\pj as cakePj;
 use function Cake\Core\pluginSplit as cakePluginSplit;
 use function Cake\Core\pr as cakePr;
+use function Cake\Core\toBool as cakeToBool;


Could we get these imports cleaned up?

Refs cakephp/cakephp#17295

othercorey · 2024-01-07T05:23:11Z

src/Core/functions.php

+ * @return ?string Returns the string representation of the value, or null if the value is not a string.
+ * @since 5.1.0
+ */
+function toString(mixed $value): ?string


This docblock needs to describe the nuances of each conversion such as bool to a 1 and 0 including failure cases.

I can rewrite the docs when this is merged.

src/Core/functions.php

Co-authored-by: othercorey <corey.taylor.fl@gmail.com>

markstory · 2024-01-09T01:57:52Z

Thank you 🎉

other Stringable object

2bd348f

Failed asserting that two strings are identical. --- Expected +++ Actual @@ @@ -'10:10 AM' +'10:10 AM'

dereuromark added this to the 5.0.1 milestone Sep 20, 2023

dereuromark added the enhancement label Sep 20, 2023

ADmad changed the base branch from 5.x to 5.next September 21, 2023 02:49

markstory reviewed Sep 21, 2023

View reviewed changes

src/Utility/Filter.php Outdated Show resolved Hide resolved

src/Utility/Filter.php Outdated Show resolved Hide resolved

ADmad reviewed Sep 21, 2023

View reviewed changes

dereuromark modified the milestones: 5.0.1, 5.1.0 Sep 21, 2023

jozefgrencik added 4 commits September 21, 2023 21:18

fixed version in DocBlock 5.0.1 >> 5.1.0

01ec52d

fix elseif with return

7f0d028

methods moved to functions.php

a2f6684

Use statements should be sorted alphabetically. The first wrong one i…

b473a4d

…s Cake\Core\toString.

ADmad reviewed Sep 22, 2023

View reviewed changes

Unnecessary checks removed

112b6d8

markstory reviewed Sep 25, 2023

View reviewed changes

jozefgrencik mentioned this pull request Dec 15, 2023

[5.x] Functions to access Request parameters with specific data types #17177

Open

jozefgrencik added 5 commits December 15, 2023 21:27

Merge branch 'cakephp:5.x' into 5.x-narrowing-down-data-types

b67b968

removed functions from functions_global.php

24c80f7

removed MAX_SAFE_INTEGER

af822bb

Merge branch '5.next' into 5.x-narrowing-down-data-types

7de9338

Merge branch '5.next' into 5.x-narrowing-down-data-types

c4878e1

markstory reviewed Dec 31, 2023

View reviewed changes

jozefgrencik added 2 commits January 4, 2024 22:53

remove unused imports

4f6ea96

PhpStan: Expected at least 1 space after "-"

8dcc58c

markstory added a commit to cakephp/docs that referenced this pull request Jan 7, 2024

Add docs for casting functions

58ccfad

Refs cakephp/cakephp#17295

markstory mentioned this pull request Jan 7, 2024

Add docs for casting functions cakephp/docs#7793

Merged

othercorey reviewed Jan 7, 2024

View reviewed changes

src/Core/functions.php Outdated Show resolved Hide resolved

Update src/Core/functions.php

244fdb7

Co-authored-by: othercorey <corey.taylor.fl@gmail.com>

markstory merged commit a0029a3 into cakephp:5.next Jan 9, 2024
6 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Methods for converting mixed values to specific data types #17295

Methods for converting mixed values to specific data types #17295

jozefgrencik commented Sep 20, 2023

jozefgrencik commented Sep 20, 2023

markstory left a comment

ADmad Sep 21, 2023

jozefgrencik Sep 21, 2023

ADmad Sep 22, 2023 •

edited

jozefgrencik Sep 23, 2023

jozefgrencik Sep 23, 2023

markstory Sep 25, 2023

jozefgrencik Dec 16, 2023

markstory Sep 25, 2023

jozefgrencik Dec 16, 2023

jozefgrencik Dec 16, 2023

markstory Sep 25, 2023

jozefgrencik Dec 16, 2023

ADmad Jan 7, 2024

markstory left a comment

markstory Dec 31, 2023

jozefgrencik Jan 4, 2024

othercorey Jan 7, 2024

othercorey Jan 8, 2024

markstory commented Jan 9, 2024

raw input	(string)$value	json_encode()	sprintf()	toString()
5.0	5	5	5	5
5.00000003	5.00000003	5.00000003	5.000000029999999818	5.00000003
55666666666.56	55666666666.56	55666666666.56	55666666666.55999755859375	55666666666.56
9223372036778.2233	9223372036778.2	9223372036778.223	9223372036778.22265625	9223372036778.223
88888888888888888888888	8.8888888888889E+22	8.888888888888889e+22	88888888888888890753024	88888888888888890753024

		* @link https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Number/MAX_SAFE_INTEGER
		* 9007199254740991 = 2^53-1 = the maximum safe integer that can be represented without losing precision.

Methods for converting mixed values to specific data types #17295

Methods for converting mixed values to specific data types #17295

Conversation

jozefgrencik commented Sep 20, 2023

jozefgrencik commented Sep 20, 2023

Method: toString

Method: toInt

markstory left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ADmad Sep 22, 2023 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

markstory left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

markstory commented Jan 9, 2024

ADmad Sep 22, 2023 •

edited