CLDRParser.php

Go to the documentation of this file.

<?php
 
class CLDRParser {
    public function parse( $inputFile, $outputFile ) {
        // Open the input file for reading
 
        $contents = file_get_contents( $inputFile );
        $doc = new SimpleXMLElement( $contents );
 
        $data = [
            'languageNames' => [],
            'currencyNames' => [],
            'currencySymbols' => [],
            'countryNames' => [],
            'timeUnits' => [],
        ];
 
        foreach ( $doc->xpath( '//languages/language' ) as $elem ) {
            if ( (string)$elem['alt'] !== '' ) {
                continue;
            }
 
            if ( (string)$elem['type'] === 'root' ) {
                continue;
            }
 
            $key = str_replace( '_', '-', strtolower( $elem['type'] ) );
 
            $data['languageNames'][$key] = (string)$elem;
        }
 
        foreach ( $doc->xpath( '//currencies/currency' ) as $elem ) {
            if ( (string)$elem->displayName[0] === '' ) {
                continue;
            }
 
            $data['currencyNames'][(string)$elem['type']] = (string)$elem->displayName[0];
            if ( (string)$elem->symbol[0] !== '' ) {
                $data['currencySymbols'][(string)$elem['type']] = (string)$elem->symbol[0];
            }
        }
 
        foreach ( $doc->xpath( '//territories/territory' ) as $elem ) {
            if ( (string)$elem['alt'] !== '' && (string)$elem['alt'] !== 'short' ) {
                continue;
            }
 
            if ( (string)$elem['type'] === 'ZZ' ||
                !preg_match( '/^[A-Z][A-Z]$/', $elem['type'] )
            ) {
                continue;
            }
 
            $data['countryNames'][(string)$elem['type']] = (string)$elem;
        }
        foreach ( $doc->xpath( '//units/unitLength' ) as $unitLength ) {
            if ( (string)$unitLength['type'] !== 'long' ) {
                continue;
            }
            foreach ( $unitLength->unit as $elem ) {
                $type = (string)$elem['type'];
                $pos = strpos( $type, 'duration' );
                if ( $pos === false ) {
                    continue;
                }
                $type = substr( $type, strlen( 'duration-' ) );
                foreach ( $elem->unitPattern as $pattern ) {
                    $data['timeUnits'][$type . '-' . (string)$pattern['count']] = (string)$pattern;
                }
            }
        }
        foreach ( $doc->xpath( '//fields/field' ) as $field ) {
            $fieldType = (string)$field['type'];
 
            foreach ( $field->relativeTime as $relative ) {
                $type = (string)$relative['type'];
                foreach ( $relative->relativeTimePattern as $pattern ) {
                    $data['timeUnits'][$fieldType . '-' . $type
                    . '-' . (string)$pattern['count']] = (string)$pattern;
                }
            }
        }
 
        ksort( $data['timeUnits'] );
 
        $this->savephp( $data, $outputFile );
    }
 
    public function parse_supplemental( $inputFile, $outputFile ) {
        // Open the input file for reading
 
        $contents = file_get_contents( $inputFile );
        $doc = new SimpleXMLElement( $contents );
 
        $data = [
            'currencyFractions' => [],
            'localeCurrencies' => [],
        ];
 
        // Pull currency attributes - digits, rounding, and cashRounding.
        // This will tell us how many decmal places make sense to use with any currency,
        // or if the currency is totally non-fractional
        foreach ( $doc->xpath( '//currencyData/fractions/info' ) as $elem ) {
            if ( (string)$elem['iso4217'] === '' ) {
                continue;
            }
 
            $attributes = [ 'digits', 'rounding', 'cashDigits', 'cashRounding' ];
            foreach ( $attributes as $att ) {
                if ( (string)$elem[$att] !== '' ) {
                    $data['currencyFractions'][(string)$elem['iso4217']][$att] = (string)$elem[$att];
                }
            }
        }
 
        // Pull a map of regions to currencies in order of preference.
        foreach ( $doc->xpath( '//currencyData/region' ) as $elem ) {
            if ( (string)$elem['iso3166'] === '' ) {
                continue;
            }
 
            $region = (string)$elem['iso3166'];
 
            foreach ( $elem->currency as $currencynode ) {
                if ( (string)$currencynode['to'] === '' && (string)$currencynode['tender'] !== 'false' ) {
                    $data['localeCurrencies'][$region][] = (string)$currencynode['iso4217'];
                }
            }
        }
 
        $this->savephp( $data, $outputFile );
    }
 
    public function parse_currency_symbols( $inputDir, $outputFile ) {
        if ( !file_exists( $inputDir ) ) {
            return;
        }
        $files = scandir( $inputDir );
 
        $data = [
            'currencySymbols' => [],
        ];
 
        // Foreach files!
        foreach ( $files as $inputFile ) {
            if ( strpos( $inputFile, '.xml' ) < 1 ) {
                continue;
            }
 
            $contents = file_get_contents( $inputDir . '/' . $inputFile );
            $doc = new SimpleXMLElement( $contents );
 
            // Tags in the <identity> section are guaranteed to appear once
            $languages = $doc->xpath( '//identity/language/@type' );
            $language = $languages
                ? (string)$languages[0]
                : pathinfo( $inputFile, PATHINFO_FILENAME );
 
            // The <territory> element is optional
            $territories = $doc->xpath( '//identity/territory/@type' );
            $territory = $territories ? (string)$territories[0] : 'DEFAULT';
 
            foreach ( $doc->xpath( '//currencies/currency' ) as $elem ) {
                if ( (string)$elem->symbol[0] !== '' ) {
                    $data['currencySymbols'][(string)$elem['type']][$language][$territory] =
                        (string)$elem->symbol[0];
                }
            }
        }
 
        // now massage the data somewhat. It's pretty blown up at this point.
 
        foreach ( $data['currencySymbols'] as $currency => $language ) {
            // get the currency default symbol. This will either be defined in the
            // 'root' language file, or taken from the ISO code.
            $default = $language['root']['DEFAULT'] ?? $currency;
 
            foreach ( $language as $lang => $territories ) {
                // Collapse a language (no locality) array if it's just the default. One value will do fine.
                if ( is_array( $territories ) ) {
                    if ( count( $territories ) === 1 && array_key_exists( 'DEFAULT', $territories ) ) {
                        $data['currencySymbols'][$currency][$lang] = $territories['DEFAULT'];
                        if ( $territories['DEFAULT'] === $default && $lang !== 'root' ) {
                            unset( $data['currencySymbols'][$currency][$lang] );
                        }
                    } else {
                        ksort( $data['currencySymbols'][$currency][$lang] );
                    }
                }
            }
 
            ksort( $data['currencySymbols'][$currency] );
        }
 
        ksort( $data['currencySymbols'] );
 
        $this->savephp( $data, $outputFile );
    }
 
    protected function savephp( $data, $location ) {
        $hasData = false;
        foreach ( $data as $v ) {
            if ( count( $v ) ) {
                $hasData = true;
                break;
            }
        }
 
        if ( !$hasData ) {
            return;
        }
 
        // Yes, I am aware I could have simply used var_export.
        // ...the spacing was ugly.
        $output = "<?php\n";
        foreach ( $data as $varname => $values ) {
            if ( !count( $values ) ) {
                // Don't output empty arrays
                continue;
            }
            $output .= "\n\$$varname = [\n";
            if ( $this->isAssoc( $values ) ) {
                foreach ( $values as $key => $value ) {
                    if ( is_array( $value ) ) {
                        $output .= $this->makePrettyArrayOuts( $key, $value, 1 );
                    } else {
                        $key = addcslashes( $key, "'" );
                        $value = addcslashes( $value, "'" );
                        if ( !is_numeric( $key ) ) {
                            $key = "'$key'";
                        }
                        $output .= "\t$key => '$value',\n";
                    }
                }
            } else {
                foreach ( $values as $value ) {
                    if ( is_array( $value ) ) {
                        $output .= $this->makePrettyArrayOuts( null, $value, 1 );
                    } else {
                        $value = addcslashes( $value, "'" );
                        $output .= "\t'$value',\n";
                    }
                }
            }
            $output .= "];\n";
        }
 
        file_put_contents( $location, $output );
    }
 
    protected function makePrettyArrayOuts( $key, $value, $level = 1 ) {
        $subKeys = '';
        $isAssoc = $this->isAssoc( $value );
        $tabs = str_repeat( "\t", $level );
 
        foreach ( $value as $subkey => $subvalue ) {
            $subkey = $isAssoc ? $subkey : null;
 
            if ( is_array( $subvalue ) ) {
                $subKeys .= $this->makePrettyArrayOuts( $subkey, $subvalue, $level + 1 );
            } else {
                $subkey = $isAssoc ? $this->formatKey( $subkey ) : '';
                $subvalue = addcslashes( $subvalue, "'" );
                $subKeys .= "$tabs\t$subkey'$subvalue',\n";
            }
        }
 
        if ( $subKeys === '' ) {
            return '';
        }
 
        $key = $key !== null ? $this->formatKey( $key ) : '';
        return "$tabs$key" . "[\n$subKeys$tabs],\n";
    }
 
    protected function formatKey( $key ) {
        $key = addcslashes( $key, "'" );
        if ( !is_numeric( $key ) ) {
            $key = "'$key'";
        }
 
        return "$key => ";
    }
 
    protected function isAssoc( array $arr ) {
        return array_keys( $arr ) !== range( 0, count( $arr ) - 1 );
    }
}