Add codelist to enum generator (#49)

bin/generate-codelists.php

@@ -0,0 +1,9 @@
+#!/usr/bin/env php
+<?php
+
+require __DIR__ . '/../vendor/autoload.php'; // Ensure autoload is included
+
+use Tiime\EN16931\Codelist\Generator\CodelistGenerator;
+
+// Call the generateCodelists method
+CodelistGenerator::generateCodelists(__DIR__ . DIRECTORY_SEPARATOR . '..' . DIRECTORY_SEPARATOR . 'tests/EN16931_code_lists_values_v14-used_from_2024-11-15.xlsx');

composer.json

@@ -19,14 +19,16 @@
     ],
     "minimum-stability": "stable",
     "require": {
-        "php": "^8.1",
-        "ext-bcmath": "*"
+        "php": ">=8.3",
+        "ext-bcmath": "*",
+        "ext-iconv": "*"
     },
     "require-dev": {
         "phpunit/phpunit": "^10.0",
         "squizlabs/php_codesniffer": "^3.7",
         "phpstan/phpstan": "^1.10",
-        "staabm/annotate-pull-request-from-checkstyle": "^1.8"
+        "staabm/annotate-pull-request-from-checkstyle": "^1.8",
+        "phpoffice/phpspreadsheet": "^3.3"
     },
     "autoload": {
         "psr-4": {
@@ -41,6 +43,7 @@
     "scripts": {
         "test": "vendor/bin/phpunit tests",
         "code_sniffer": "vendor/bin/phpcs -q --report=checkstyle --standard=PSR12 src/",
-        "phpstan": "vendor/bin/phpstan analyse -l 9 src tests"
+        "phpstan": "vendor/bin/phpstan analyse -l 9 src tests",
+        "generate-codelists": "php bin/generate-codelists.php"
     }
 }

src/Codelist/Generator/CodelistGenerator.php

@@ -0,0 +1,50 @@
+<?php
+
+namespace Tiime\EN16931\Codelist\Generator;
+
+final readonly class CodelistGenerator
+{
+    public static function generateCodelists(
+        string $fileToRead
+    ): void {
+        $generator = new Generator();
+
+        $countryList = XlsxReader::read(filename: $fileToRead, valueColumn: 'B', nameColumn: 'A', sheetName: 'Country');
+        $currencyList = XlsxReader::read(filename: $fileToRead, valueColumn: 'B', nameColumn: 'A', sheetName: 'Currency');
+        $internationalCodeDesignatorList = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'B', sheetName: 'ICD');
+        $code1001List = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'B', sheetName: '1001');
+        $code1153List = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'B', sheetName: '1153');
+        $code2005List = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'B', sheetName: 'Time', startLine: 3);
+        $code2475List = XlsxReader::read(filename: $fileToRead, valueColumn: 'C', nameColumn: 'D', sheetName: 'Time', startLine: 3);
+        $code4451List = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'B', sheetName: 'Text',);
+        $code4461List = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'B', sheetName: 'Payment',);
+        $code5305List = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'B', sheetName: '5305',);
+        $code5189List = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'B', sheetName: 'Allowance',);
+        $code7143List = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'B', sheetName: 'Item',);
+        $code7161List = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'B', sheetName: 'Charge',);
+        $mimeList = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'A', sheetName: 'MIME',);
+        $electronicAddressSchemeList = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'B', sheetName: 'EAS',);
+        $vatExemptionReasonCodeList = XlsxReader::read(filename: $fileToRead, valueColumn: 'A', nameColumn: 'B', sheetName: 'VATEX',);
+        $unitOfMeasure = XlsxReader::read(filename: $fileToRead, valueColumn: 'B', nameColumn: 'C', sheetName: 'Unit',);
+
+
+        // @todo Greece : EL for France Spec
+        $generator->generateCodelist(className: 'CountryAlpha2Code', cases: $countryList);
+        $generator->generateCodelist(className: 'CurrencyCodeISO4217', cases: $currencyList);
+        $generator->generateCodelist(className: 'InternationalCodeDesignator', cases: $internationalCodeDesignatorList);
+        $generator->generateCodelist(className: 'InvoiceTypeCodeUNTDID1001', cases: $code1001List);
+        $generator->generateCodelist(className: 'ReferenceQualifierCodeUNTDID1153', cases: $code1153List);
+        $generator->generateCodelist(className: 'TimeReferencingCodeUNTDID2005', cases: $code2005List);
+        $generator->generateCodelist(className: 'TimeReferencingCodeUNTDID2475', cases: $code2475List);
+        $generator->generateCodelist(className: 'TextSubjectCodeUNTDID4451', cases: $code4451List);
+        $generator->generateCodelist(className: 'PaymentMeansCodeUNTDID4461', cases: $code4461List);
+        $generator->generateCodelist(className: 'DutyTaxFeeCategoryCodeUNTDID5305', cases: $code5305List);
+        $generator->generateCodelist(className: 'AllowanceReasonCodeUNTDID5189', cases: $code5189List);
+        $generator->generateCodelist(className: 'ItemTypeCodeUNTDID7143', cases: $code7143List);
+        $generator->generateCodelist(className: 'ChargeReasonCodeUNTDID7161', cases: $code7161List);
+        $generator->generateCodelist(className: 'MimeCode', cases: $mimeList);
+        $generator->generateCodelist(className: 'ElectronicAddressSchemeCode', cases: $electronicAddressSchemeList);
+        $generator->generateCodelist(className: 'VatExemptionReasonCodeList', cases: $vatExemptionReasonCodeList);
+        $generator->generateCodelist(className: 'UnitOfMeasureCode', cases: $unitOfMeasure);
+    }
+}

src/Codelist/Generator/Generator.php

@@ -0,0 +1,46 @@
+<?php
+
+namespace Tiime\EN16931\Codelist\Generator;
+
+/**
+ * @internal
+ */
+final readonly class Generator
+{
+    private const string ENUM_TEMPLATE = <<<'TEMPLATE'
+<?php
+
+declare(strict_types=1);
+
+namespace <namespace>;
+
+enum <className> : string
+{
+<cases>
+}
+TEMPLATE;
+
+    public function generateCodelist(
+        string $className,
+        XlsxReaderResult $cases
+    ): void {
+        $code = [];
+
+        /** @var XlsxReaderResultItem $case */
+        foreach ($cases->getHarmonized() as $case) {
+            $code[] = (string) $case;
+        }
+
+        $replacements = [
+            '<cases>' => '        ' . implode("\n        ", $code),
+            '<namespace>' => 'Tiime\EN16931\Codelist',
+            '<className>' => $className
+        ];
+
+        $code = strtr(self::ENUM_TEMPLATE, $replacements);
+        $code = preg_replace('/^ +$/m', '', $code);
+        $path            = __DIR__ . DIRECTORY_SEPARATOR . '..' . DIRECTORY_SEPARATOR . $className . '.php';
+
+        file_put_contents($path, $code);
+    }
+}

src/Codelist/Generator/XlsxReader.php

@@ -0,0 +1,45 @@
+<?php
+
+namespace Tiime\EN16931\Codelist\Generator;
+
+use PhpOffice\PhpSpreadsheet\Reader\Xlsx;
+use PhpOffice\PhpSpreadsheet\Worksheet\Worksheet;
+
+final readonly class XlsxReader
+{
+    public static function read(
+        string $filename,
+        string $valueColumn,
+        string $nameColumn,
+        string $sheetName,
+        int $startLine = 2,
+    ): XlsxReaderResult {
+        $reader = (new Xlsx())
+            ->setLoadSheetsOnly($sheetName)
+            ->setReadEmptyCells(false)
+            ->setIgnoreRowsWithNoCells(true)
+        ;
+        $spreadsheet = $reader->load($filename);
+
+        $worksheet = $spreadsheet->getSheetByNameOrThrow($sheetName);
+        $highestRow = $worksheet->getHighestRow();
+
+        $result = new XlsxReaderResult();
+
+        for ($row = $startLine; $row <= $highestRow; $row++) {
+            $nameFromCell = $worksheet->getCell($nameColumn . $row)->getValue();
+            assert(is_string($nameFromCell));
+            $valueFromCell = $worksheet->getCell($valueColumn . $row)->getValue();
+            assert(is_string($valueFromCell));
+            $name = trim($nameFromCell);
+            $value = trim($valueFromCell);
+
+            $result->add(new XlsxReaderResultItem(
+                name: $name,
+                value: $value,
+            ));
+        }
+
+        return $result;
+    }
+}

src/Codelist/Generator/XlsxReaderResult.php

@@ -0,0 +1,110 @@
+<?php
+
+namespace Tiime\EN16931\Codelist\Generator;
+
+use ArrayObject;
+
+final class XlsxReaderResult
+{
+    /**
+     * @var XlsxReaderResultItem[]
+     */
+    private array $result;
+
+    public function add(XlsxReaderResultItem $item): void
+    {
+        $this->result[] = $item;
+    }
+
+    /**
+     * @return XlsxReaderResultItem[]
+     */
+    public function get(): array
+    {
+        return $this->result;
+    }
+
+    /**
+     * @return XlsxReaderResultItem[]
+     */
+    public function getHarmonized(): array
+    {
+        $nameCount = [];
+        $harmonizedResult = [];
+
+        foreach ($this->result as $item) {
+            $name = $item->name;
+
+            $name = preg_replace('/§\s*27/', '_', $name); // Specific ICD sheet
+            assert(is_string($name));
+            $name = preg_replace('/®/', '_', $name);
+            assert(is_string($name));
+            $name = preg_replace('/@/', 'A', $name);
+            assert(is_string($name));
+
+            $replacements = [
+                '–' => '_',
+                '>' => 'GREATER',  // Replace > with GREATER
+                '%' => 'PERCENT',  // Replace % with PERCENT
+                '+' => 'PLUS',     // Replace + with PLUS
+                '°' => 'DEGREE'    // Replace ° with DEGREE
+            ];
+
+            $name = strtr($name, $replacements);
+
+            $name = preg_replace('/[()]/', '', $name); // Remove parentheses
+            assert(is_string($name));
+            $name = preg_replace('/\s*\[.*?\]\s*/', '', $name); // Remove brackets and content inside
+            assert(is_string($name));
+            $name = preg_replace('/-/', '_', $name); // Replace - with underscores
+            assert(is_string($name));
+            $name = preg_replace('/,/', '_', $name); // Replace , with underscores
+            assert(is_string($name));
+            $name = preg_replace('/&/', '_', $name); // Replace & with underscores
+            assert(is_string($name));
+            $name = preg_replace('/\//', '_', $name); // Replace / with underscores
+            assert(is_string($name));
+            $name = preg_replace('/[.\'“”’:=*]/u', '', $name);
+            assert(is_string($name));
+
+            $name = trim($name, '_'); // Ensure the result does not start or end with an underscore
+
+            $name = preg_replace('/^(\d)/', '_$1', $name); // Add underscore at the beginning if starting by a number
+            assert(is_string($name));
+
+            // Convert accented characters to non-accented
+            $name = iconv('UTF-8', 'ASCII//TRANSLIT//IGNORE', $name);
+            assert(is_string($name));
+            $name = mb_strtoupper($name);
+            assert(is_string($name));
+            $name = preg_replace('/\s+/', '_', $name); // Replace spaces with underscores
+            assert(is_string($name));
+            $name = preg_replace('/_+/', '_', $name); // Replace multiple underscores with a single underscore
+            assert(is_string($name));
+
+            // Initialize the count for this name if it doesn't exist
+            if (!isset($nameCount[$name])) {
+                $nameCount[$name] = 0;
+            }
+
+            $nameCount[$name]++;
+
+            if ($nameCount[$name] > 1) {
+                $name .= '_' . $this->getSuffix($nameCount[$name]);
+            }
+
+            $harmonizedResult[] = new XlsxReaderResultItem(name: $name, value: $item->value);
+        }
+
+        return $harmonizedResult;
+    }
+
+    private function getSuffix(int $count): string
+    {
+        return match ($count) {
+            2 => 'SECOND',
+            3 => 'THIRD',
+            default => $count . 'TH',
+        };
+    }
+}

src/Codelist/Generator/XlsxReaderResultItem.php

@@ -0,0 +1,17 @@
+<?php
+
+namespace Tiime\EN16931\Codelist\Generator;
+
+final readonly class XlsxReaderResultItem
+{
+    public function __construct(
+        public string $name,
+        public string $value,
+    ) {
+    }
+
+    public function __toString(): string
+    {
+        return sprintf("case %s = '%s';", $this->name, $this->value);
+    }
+}