Merge pull request #7 from eclipxe13/catalogs-cfdi_40

Catalogs CFDI 4.0

README.md

@@ -52,6 +52,10 @@ php bin/sat-catalogos-update update-origins catalogs/
 php bin/sat-catalogos-update update-database catalogs/ catalogs/catalogos.sqlite3
 ```
 
+## About the catalogs included
+
+Read the file [*Catálogos SAT*](docs/Catalogos.md) for more information about the catalogs included (in spanish).
+
 ## PHP Support
 
 This tool is compatible with last [PHP supported version](https://www.php.net/supported-versions.php).

docs/CHANGELOG.md

@@ -1,5 +1,15 @@
 # phpcfdi/sat-catalogos-populate Changelog
 
+## Version 2.3.0 2022-03-05
+
+Add CFDI 4.0 catalogs.
+
+- Add *Origin* as a scrap from SAT website, saving the file as `cfdi_40.xsl`.
+- Add *Source importer* from `cfdi_40.xsl`. It includes 25 catalogs.
+- Importers create tables with prefix `cfdi_40_*`.
+- Update `docs/Catalogos.md` with CFDI 4.0 catalogs.
+- Refactor how to know what lines to skip when join two CVS files.
+
 ## Version 2.2.1 2022-03-04
 
 Include `Nóminas - Estados` origin to `dump-origins` command.

docs/Pruebas.md

@@ -19,6 +19,7 @@ estos archivos y correr los tests. De esta forma podremos notar un cambio import
 
 - `tests/_files/cce/*.csv`: Archivos que salen de los catálogos de Comercio Exterior
 - `tests/_files/cfdi/*.csv`: Archivos que salen de los catálogos de CFDI 3.3
+- `tests/_files/cfdi40/*.csv`: Archivos que salen de los catálogos de CFDI 4.0
 - `tests/_files/nomina/*.csv`: Archivos que salen de los catálogos de Nómina 1.2
 - `tests/_files/rep/*.csv`: Archivos que salen de los catálogos de recibo electrónico de pagos
 
@@ -42,19 +43,22 @@ php bin/sat-catalogos-update update-origins build/temp/origins.xml
 
 rm -rf tests/_files/sources
 rm -rf tests/_files/cfdi
+rm -rf tests/_files/cfdi40
 rm -rf tests/_files/nomina
 rm -rf tests/_files/rep
 rm -rf tests/_files/cce
 
 mkdir -p tests/_files/sources
 mkdir -p tests/_files/cfdi
+mkdir -p tests/_files/cfdi40
 mkdir -p tests/_files/nomina
 mkdir -p tests/_files/rep
 mkdir -p tests/_files/cce
 
 cp build/temp/*.xls tests/_files/sources
 
 php tests/convert-xls-to-csv-folder.php tests/_files/sources/catCFDI.xls tests/_files/cfdi
+php tests/convert-xls-to-csv-folder.php tests/_files/sources/cfdi_40.xls tests/_files/cfdi40
 php tests/convert-xls-to-csv-folder.php tests/_files/sources/catNomina.xls tests/_files/nomina
 php tests/convert-xls-to-csv-folder.php tests/_files/sources/catPagos.xls tests/_files/rep
 php tests/convert-xls-to-csv-folder.php tests/_files/sources/c_ClavePedimento.xls tests/_files/cce

src/Commands/DumpOrigins.php

@@ -20,7 +20,13 @@ public function run(): int
                 'CFDI',
                 'http://omawww.sat.gob.mx/tramitesyservicios/Paginas/anexo_20_version3-3.htm',
                 'catCFDI.xls',
-                'Catálogos CFDI Versión 3.3(xls)',
+                'Catálogos CFDI Versión 3.3*',
+            ),
+            new ScrapingOrigin(
+                'CFDI',
+                'http://omawww.sat.gob.mx/tramitesyservicios/Paginas/anexo_20_version3-3.htm',
+                'cfdi_40.xls',
+                'Catálogos CFDI Versión 4.0*',
             ),
             new ConstantOrigin('Nóminas', "{$common}/catNomina.xls"),
             new ConstantOrigin('Nóminas - Estados', "{$common}/C_Estado.xls", null, 'nominas_estados.xls'),

src/Converters/CsvFolderJoinFiles.php

@@ -78,6 +78,7 @@ function ($path): array {
     /**
      * @param array{destination: string, index: int} $first
      * @param array{destination: string, index: int} $second
+     * @return int
      */
     private function compareFiles(array $first, array $second): int
     {
@@ -97,42 +98,46 @@ public function writeLines(string $source, string $destination, int $skipFirstLi
 
     public function findLinesToSkip(string $firstPath, string $secondPath): int
     {
-        $lines = 0;
         $first = new SplFileObject($firstPath, 'r');
         $second = new SplFileObject($secondPath, 'r');
-        while ($this->splCurrentLinesAreEqual($first, $second)) {
-            $first->next();
-            $second->next();
-            $lines = $lines + 1;
+
+        $firstTenLines = $this->splReadTenLines($first);
+        $secondTenLines = $this->splReadTenLines($second);
+
+        for ($i = 9; $i > 0; $i--) {
+            $firstValue = $firstTenLines[$i] ?? null;
+            $secondValue = $secondTenLines[$i] ?? null;
+            if (null !== $firstValue && null !== $secondValue && $firstValue === $secondValue) {
+                return $i + 1;
+            }
         }
-        return $lines;
+        return 0;
     }
 
     /**
-     * @param Iterator<mixed> $first
-     * @param Iterator<mixed> $second
+     * @param Iterator<string> $iterator
+     * @return array<int, string>
      */
-    private function splCurrentLinesAreEqual(Iterator $first, Iterator $second): bool
+    private function splReadTenLines(Iterator $iterator): array
     {
-        if (! $first->valid() || ! $second->valid()) {
-            return false;
+        $result = [];
+        $counter = 0;
+        foreach ($iterator as $line) {
+            $result[] = $this->splCurrentLinesNormalizeValue($line);
+            $counter = $counter + 1;
+            if (10 === $counter) {
+                break;
+            }
         }
-        $firstValue = $this->splCurrentLinesAreEqualNormalizeValue($first->current());
-        $secondValue = $this->splCurrentLinesAreEqualNormalizeValue($second->current());
-        return ($firstValue === $secondValue);
+        return $result;
     }
 
-    private function splCurrentLinesAreEqualNormalizeValue(mixed $current): mixed
+    private function splCurrentLinesNormalizeValue(string $current): string
     {
-        if (! is_string($current)) {
-            return $current;
-        }
         return trim(implode(',', array_map('trim', explode(',', rtrim($current, ',')))));
     }
 
-    /**
-     * @param string[] $searchterms
-     */
+    /** @param string[] $searchterms */
     public function lastLineContains(string $filename, array $searchterms): bool
     {
         $lastline = $this->obtainFileLastLine($filename);

src/Importers/Cfdi40/Injectors/Aduanas.php

@@ -0,0 +1,43 @@
+<?php
+
+declare(strict_types=1);
+
+namespace PhpCfdi\SatCatalogosPopulate\Importers\Cfdi40\Injectors;
+
+use PhpCfdi\SatCatalogosPopulate\AbstractCsvInjector;
+use PhpCfdi\SatCatalogosPopulate\Database\DataFields;
+use PhpCfdi\SatCatalogosPopulate\Database\DataTable;
+use PhpCfdi\SatCatalogosPopulate\Database\DateDataField;
+use PhpCfdi\SatCatalogosPopulate\Database\PaddingDataField;
+use PhpCfdi\SatCatalogosPopulate\Database\TextDataField;
+use PhpCfdi\SatCatalogosPopulate\InjectorInterface;
+use PhpCfdi\SatCatalogosPopulate\Utils\CsvFile;
+use RuntimeException;
+
+use function PhpCfdi\SatCatalogosPopulate\Utils\array_rtrim;
+
+class Aduanas extends AbstractCsvInjector implements InjectorInterface
+{
+    public function checkHeaders(CsvFile $csv): void
+    {
+        $csv->move(3);
+        $expected = ['c_Aduana', 'Descripción', 'Fecha inicio de vigencia', 'Fecha fin de vigencia'];
+        $headers = array_rtrim($csv->readLine());
+
+        if ($expected !== $headers) {
+            throw new RuntimeException("The headers did not match on file {$this->sourceFile()}");
+        }
+
+        $csv->next();
+    }
+
+    public function dataTable(): DataTable
+    {
+        return new DataTable('cfdi_40_aduanas', new DataFields([
+            new PaddingDataField('id', '0', 2),
+            new TextDataField('texto'),
+            new DateDataField('vigencia_desde'),
+            new DateDataField('vigencia_hasta'),
+        ]));
+    }
+}

src/Importers/Cfdi40/Injectors/ClavesUnidades.php

@@ -0,0 +1,54 @@
+<?php
+
+declare(strict_types=1);
+
+namespace PhpCfdi\SatCatalogosPopulate\Importers\Cfdi40\Injectors;
+
+use PhpCfdi\SatCatalogosPopulate\AbstractCsvInjector;
+use PhpCfdi\SatCatalogosPopulate\Database\DataFields;
+use PhpCfdi\SatCatalogosPopulate\Database\DataTable;
+use PhpCfdi\SatCatalogosPopulate\Database\DateDataField;
+use PhpCfdi\SatCatalogosPopulate\Database\PaddingDataField;
+use PhpCfdi\SatCatalogosPopulate\Database\TextDataField;
+use PhpCfdi\SatCatalogosPopulate\InjectorInterface;
+use PhpCfdi\SatCatalogosPopulate\Utils\CsvFile;
+use RuntimeException;
+
+use function PhpCfdi\SatCatalogosPopulate\Utils\array_rtrim;
+
+class ClavesUnidades extends AbstractCsvInjector implements InjectorInterface
+{
+    public function checkHeaders(CsvFile $csv): void
+    {
+        $csv->move(3);
+        $expected = [
+            'c_ClaveUnidad',
+            'Nombre',
+            'Descripción',
+            'Nota',
+            'Fecha de inicio de vigencia',
+            'Fecha de fin de vigencia',
+            'Símbolo',
+        ];
+        $headers = array_rtrim($csv->readLine());
+
+        if ($expected !== $headers) {
+            throw new RuntimeException("The headers did not match on file {$this->sourceFile()}");
+        }
+
+        $csv->next();
+    }
+
+    public function dataTable(): DataTable
+    {
+        return new DataTable('cfdi_40_claves_unidades', new DataFields([
+            new PaddingDataField('id', '0', 2),
+            new TextDataField('texto'),
+            new TextDataField('descripcion'),
+            new TextDataField('notas'),
+            new DateDataField('vigencia_desde'),
+            new DateDataField('vigencia_hasta'),
+            new TextDataField('simbolo'),
+        ]));
+    }
+}

src/Importers/Cfdi40/Injectors/CodigosPostales.php

@@ -0,0 +1,95 @@
+<?php
+
+declare(strict_types=1);
+
+namespace PhpCfdi\SatCatalogosPopulate\Importers\Cfdi40\Injectors;
+
+use Generator;
+use PhpCfdi\SatCatalogosPopulate\AbstractCsvInjector;
+use PhpCfdi\SatCatalogosPopulate\Database\BoolDataField;
+use PhpCfdi\SatCatalogosPopulate\Database\DataFields;
+use PhpCfdi\SatCatalogosPopulate\Database\DataTable;
+use PhpCfdi\SatCatalogosPopulate\Database\DateDataField;
+use PhpCfdi\SatCatalogosPopulate\Database\TextDataField;
+use PhpCfdi\SatCatalogosPopulate\Utils\CsvFile;
+use RuntimeException;
+
+class CodigosPostales extends AbstractCsvInjector
+{
+    public function checkHeaders(CsvFile $csv): void
+    {
+        $csv->move(3);
+        $expectedLines = [
+            [
+                'c_CodigoPostal',
+                'c_Estado',
+                'c_Municipio',
+                'c_Localidad',
+                'Estímulo Franja Fronteriza',
+                'Fecha inicio de vigencia',
+                'Fecha fin de vigencia',
+                'Referencias del Huso Horario',
+            ],
+            [
+                '',
+                '',
+                '',
+                '',
+                '',
+                '',
+                '',
+                'Descripción del Huso Horario',
+                'Mes_Inicio_Horario_Verano',
+                'Día_Inicio_Horario_Verano',
+                'Día_Inicio_Horario_Verano',
+                'Diferencia_Horaria_Verano',
+                'Mes_Inicio_Horario_Invierno',
+                'Día_Inicio_Horario_Invierno',
+                'Día_Inicio_Horario_Invierno',
+                'Diferencia_Horaria_Invierno',
+            ],
+        ];
+        foreach ($expectedLines as $line => $expected) {
+            $line = $line + 1;
+            $headers = array_map(fn ($line): string => trim((string) $line), $csv->readLine());
+            if ($expected !== $headers) {
+                throw new RuntimeException("The headers did not match on file {$this->sourceFile()} line {$line}");
+            }
+            $csv->next();
+        }
+    }
+
+    public function dataTable(): DataTable
+    {
+        return new DataTable('cfdi_40_codigos_postales', new DataFields([
+            new TextDataField('id'),
+            new TextDataField('estado'),
+            new TextDataField('municipio'),
+            new TextDataField('localidad'),
+            new BoolDataField('estimulo_frontera', ['1']),
+            new DateDataField('vigencia_desde'),
+            new DateDataField('vigencia_hasta'),
+            new TextDataField('huso_descripcion'),
+            new TextDataField('huso_verano_mes_inicio'),
+            new TextDataField('huso_verano_dia_inicio'),
+            new TextDataField('huso_verano_hora_inicio'),
+            new TextDataField('huso_verano_diferencia'),
+            new TextDataField('huso_invierno_mes_inicio'),
+            new TextDataField('huso_invierno_dia_inicio'),
+            new TextDataField('huso_invierno_hora_inicio'),
+            new TextDataField('huso_invierno_diferencia'),
+        ]));
+    }
+
+    /** @inheritdoc */
+    protected function readLinesFromCsv(CsvFile $csv): Generator
+    {
+        foreach ($csv->readLines() as $line) {
+            if ('00000' === $line[0]) {
+                continue;
+            }
+
+            yield $line;
+        }
+    }
+}

src/Importers/Cfdi40/Injectors/Colonias.php

@@ -0,0 +1,38 @@
+<?php
+
+declare(strict_types=1);
+
+namespace PhpCfdi\SatCatalogosPopulate\Importers\Cfdi40\Injectors;
+
+use PhpCfdi\SatCatalogosPopulate\AbstractCsvInjector;
+use PhpCfdi\SatCatalogosPopulate\Database\DataFields;
+use PhpCfdi\SatCatalogosPopulate\Database\DataTable;
+use PhpCfdi\SatCatalogosPopulate\Database\PaddingDataField;
+use PhpCfdi\SatCatalogosPopulate\Database\TextDataField;
+use PhpCfdi\SatCatalogosPopulate\Utils\CsvFile;
+use RuntimeException;
+
+class Colonias extends AbstractCsvInjector
+{
+    public function checkHeaders(CsvFile $csv): void
+    {
+        $csv->move(3);
+        $expected = ['c_Colonia', 'c_CodigoPostal', 'Nombre del asentamiento'];
+        $headers = $csv->readLine();
+
+        if ($expected !== $headers) {
+            throw new RuntimeException("The headers did not match on file {$this->sourceFile()}");
+        }
+
+        $csv->next();
+    }
+
+    public function dataTable(): DataTable
+    {
+        return new DataTable('cfdi_40_colonias', new DataFields([
+            new PaddingDataField('colonia', '0', 4),
+            new PaddingDataField('codigo_postal', '0', 5),
+            new TextDataField('texto'),
+        ]), ['colonia', 'codigo_postal']);
+    }
+}