Skip to content

Latest commit

 

History

History
166 lines (134 loc) · 5.47 KB

README.md

File metadata and controls

166 lines (134 loc) · 5.47 KB
Raw

Query tokens extractor

Extract tokens from a query using regex defined tokens. The library is written in PHP 8.2.

Installation

The package is available on Packagist and can be installed using Composer:

composer req codeinc/query-token-extractor

Usage

use CodeInc\QueryTokenExtractor\QueryTokenExtractor;
use CodeInc\QueryTokensExtractor\Type\RegexType;
use CodeInc\QueryTokensExtractor\Type\WordType;
use CodeInc\QueryTokensExtractor\Type\FrenchPhoneNumberType;
use CodeInc\QueryTokensExtractor\Type\FrenchPostalCodeType;
use CodeInc\QueryTokensExtractor\Type\YearType;
use CodeInc\QueryTokensExtractor\Dto\QueryToken;

$tokensExtractor = new QueryTokensExtractor([
    new FrenchPhoneNumberType(),
    new FrenchPostalCodeType(),
    new YearType(),
    new RegexType('my_custom_token', '/^this a custom token/ui'),
    new WordType(),
]);

$tokens = $tokensExtractor->extract('paris (75001) these are words 01.00.00.00.00 this a custom token 2023');

/** @var QueryToken $token */
foreach ($tokens as $token) {
    echo "Position: " . $token->position . "\n"
        ."Class: " . get_class($token->type) . "\n"
        ."Name: " . $token->type->name . "\n"
        ."Value: " . $token->value . "\n";
}

The above exemple will generate the following output:

Position: 0
Class: CodeInc\QueryTokensExtractor\Type\WordType
Name: word
Value: paris

Position: 1
Class: CodeInc\QueryTokensExtractor\Type\FrenchPostalCodeType
Name: french_postal_code
Value: 75001

Position: 2
Class: CodeInc\QueryTokensExtractor\Type\WordType
Name: word
Value: these

Position: 3
Class: CodeInc\QueryTokensExtractor\Type\WordType
Name: word
Value: are

Position: 4
Class: CodeInc\QueryTokensExtractor\Type\WordType
Name: word
Value: words

Position: 5
Class: CodeInc\QueryTokensExtractor\Type\FrenchPhoneNumberType
Name: french_phone_number
Value: 01 00 00 00 00 (the original value without punctuation)

Position: 6
Class: CodeInc\QueryTokensExtractor\Type\CustomTokenType
Name: my_custom_token
Value: this a custom token

Position: 7
Class: CodeInc\QueryTokensExtractor\Type\YearType
Name: year
Value: 2023

Token types

Available token types

  • WordType: extract words from the query
  • YearType: extract years from the query
  • FrenchPhoneNumberType: extract French phone numbers from the query
  • FrenchPostalCodeType: extract French postal codes from the query
  • HashtagType: extract hashtags from the query
  • RegexTokenType: extract tokens from the query using a regex

Token type priority

The token type priority is determined by the order in which the token types are passed to the QueryTokensExtractor constructor.

The priority is used to determine the order in which the tokens are extracted. The higher the priority, the sooner the token will be extracted.

⚠️ The WordType should always be used last as it will match any string.

use CodeInc\QueryTokenExtractor\QueryTokenExtractor;
use CodeInc\QueryTokensExtractor\Type\WordType;
use CodeInc\QueryTokensExtractor\Type\FrenchPhoneNumberType;
use CodeInc\QueryTokensExtractor\Type\FrenchPostalCodeType;
use CodeInc\QueryTokensExtractor\Type\YearType;

$tokensExtractor = new QueryTokensExtractor([
    new FrenchPhoneNumberType(), // highest priority
    new FrenchPostalCodeType(),
    new YearType(),
    new WordType(), // lowest priority
]);

Creating custom token types

Custom token types can be created by instantiating or extending RegexTokenType. The constructor of RegexTokenType takes four arguments:

  • string $name: the name of the token type
  • string $regex: the regex used to extract the token
  • \Closure $valueFormatter: a closure used to format the extracted value (optional)

The regexp value capturing group is used as the extracted value (for instance the HashtagType type uses the regex '/^#(?<value>.[a-z0-9_]+)/ui'). If no group named value is defined, the whole match is used as the token value.

The regexp should always start with ^ and do not constrain the end of the string with $ as the query is split into tokens using the preg_replace_callback() function.

use CodeInc\QueryTokensExtractor\Type\RegexType;

class MyCustomTokenType extends RegexType
{
    public function __construct()
    {
        parent::__construct(
            name: 'my_custom_token',
            regexp: '/^this a custom token/ui'
        );
    }
}

// alternatively tokens can be defined directly using the RegexType class
$myCustomToken2 = new RegexType(
    name: 'my_custom_token',
    regexp: '/^this a custom token/ui'
);

Token value formatting

The extracted token value can be formatted using the valueFormatter closure. The closure takes the extracted value as argument and must return the formatted value.

use CodeInc\QueryTokensExtractor\Type\RegexType;

$tokensExtractor = new QueryTokensExtractor([
    new RegexType(
        name: 'my_custom_token',
        regexp: '/^this a custom token/ui',
        // a simple closure called by QueryToken::getFormattedValue()
        valueFormatter: fn($value) => strtoupper($value)
    )
]);
$tokens = $tokensExtractor->extract('this a custom token');
$tokens->getByPosition(0)->getFormattedValue(); // THIS A CUSTOM TOKEN

License

This library is published under the MIT license (see the LICENSE file).