php-value-mask

Purpose

Google in their Performance Tips for APIs, suggest to limit required bandwidth by filtering out unused fields in response. Their APIs support additional URL parameter fields which asks API to include only specific fields in response.

fields parameter follows a simple syntax, which allows to query for nested keys, multiple keys or use wildcard to include all fields (see Syntax and Grammar sections below).

This library implements parsing of fields parameter and filtering of arrays / objects.

See also: PSR-15 compatible middleware based on this library: Partial Response PSR-15 Middleware

Usage example

<?php
require_once 'vendor/autoload.php';

use messere\phpValueMask\Parser\Parser;
use messere\phpValueMask\Parser\ParserException;

$parser = new Parser();

$input = [
    'id' => 1,
    'resource' => 'book',
    'title' => 'Good Omens',
    'identifiers' => (object)[
        'isbn' => 'ISBN 83-85100-63-6​',
        'amazon' => '0060853980',  
    ],
    'authors' => [
        [
            'firstName' => 'Terry',
            'lastName' => 'Pratchett'
        ],
        [
            'firstName' => 'Neil',
            'lastName' => 'Gaiman'
        ],
    ],
    'year' => [
        'us' => 1990,
        'uk' => 1990,
        'pl' => 1992,
    ],
    'publisher' => [
        'us' => 'Workman',
        'uk' => 'Gollancz',
        'pl' => 'CIA-Books-SVARO',
    ],
];

$filter = 'title,identifiers/isbn,authors/firstName,*(us,uk),keywords';

try {
    $filteredInput = $parser->parse($filter)->filter($input);
    print_r($filteredInput);
} catch (ParserException $e) {
    echo 'Parser error: ' . $e->getMessage();
} 

Let's analyze elements of used filter:

• title matches top level element with key title ('Good Omens')
• identifiers/isbn matches top level element identifiers and then includes isbn element from matched object ('ISBN 83-85100-63-6')
• authors/firstName finds an array of elements (list) under the key authors and examines all elements, extracting firstName from each. ('Terry' and 'Neil')
• *(us,uk) examines all properties and extracts fields us and uk. ('1990' and '1990' from year element, 'Workman', 'Gollancz' from publisher)
• keywords does not match anything and is silently ignored.

As a result we expect the following output:

Array
(
    [title] => Good Omens
    [identifiers] => Array
        (
            [isbn] => ISBN 83-85100-63-6​
        )

    [authors] => Array
        (
            [0] => Array
                (
                    [firstName] => Terry
                )

            [1] => Array
                (
                    [firstName] => Neil
                )

        )

    [year] => Array
        (
            [us] => 1990
            [uk] => 1999
        )

    [publisher] => Array
        (
            [us] => Workman
            [uk] => Gollancz
        )

)

ready to serialize to JSON, etc.

Note that library does preserve the structure/nesting of values, but not necessarily types of values - all objects are converted to associative arrays with object's public properties as keys.

Syntax

• a selects key a from input
• a,b,c comma separated list of elements: selects keys a and b and c
• a/b/c nested elements: selects key c from parent element b which in turn has parent element a
• a(b,c) multiple elements: selects elements b and c from parent element a
• a(b,c/d) multiple elements: from parent a select element b and element d nested in c
• a/*/c wildcard: selects element c from all children of element a
• *(b,c) wildcard: selects elements b and c from any parent

Etc. See tests for more examples as well as examples of invalid filters.

Grammar

Since Google does not provide detailed grammar for their "fields" language, this package uses the following arbitrarily selected rules, that in author's opinion closely resamble intent of original authors.

In EBNF notation:

Mask         = MaskElement | MaskElement , "," , Mask ;
MaskElement  = ArrayOfMasks | NestedKeys ;
ArrayOfMasks = Key , "(" , Mask , ")" ;
NestedKeys   = Key , [ "/" , NestedKeys ] ;
Key          = Wildcard | Identifier ;
Identifier   = Letter , { Letter | Digit }
Wildcard     = "*" ;
Letter       = "A" | "B" | "C" | "D" | "E" | "F" | "G" | "H" | "I" | "J" | "K" | "L" | "M" |
               "N" | "O" | "P" | "Q" | "R" | "S" | "T" | "U" | "V" | "W" | "X" | "Y" | "Z" |
               "a" | "b" | "c" | "d" | "e" | "f" | "g" | "h" | "i" | "j" | "k" | "l" | "m" |
               "n" | "o" | "p" | "q" | "r" | "s" | "t" | "u" | "v" | "w" | "x" | "y" | "z" |
               "_";
Digit        = "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9" | "0" ;

Acknowledgements

Library is inspired by:

• Google's API performance tips.
• Similar JavaScript library: nemtsov/json-mask.

License

MIT