Update dataset + handle multiple cities and cantons per zipcode (#47)

* Fix UpdateZipcodeDatasetCommand.php * Change methods in order to retrieve list of cities based of a zipcode * New findOnBy method in CantonManager.php * Exit loop when csv file has been found in UpdateZipcodeDatasetCommand * Add Docs for new findOneBy Method * Rename Method * Use "Ortschaftsname" as city name --------- Co-authored-by: Stefan Zweifel <[email protected]>
stefanzweifel · May 26, 2024 · 9ea3da5 · 9ea3da5
1 parent f895d8e
commit 9ea3da5
Show file tree

Hide file tree

Showing 12 changed files with 18,052 additions and 15,660 deletions.
diff --git a/README.md b/README.md
@@ -53,7 +53,7 @@ $canton = $cantonManager->getByAbbreviation('GR');
 
 ### `getByName()`
 
-Search for a Canton by it's name. The name must exactly match one of the translations of the Canton (German, French, Italian, Romansh or English).
+Search for a Canton by its name. The name must exactly match one of the translations of the Canton (German, French, Italian, Romansh or English).
 
 ```php
 $cantonManager = new Wnx\SwissCantons\CantonManager();
@@ -64,13 +64,25 @@ $canton = $cantonManager->getByName('Zürich');
 
 ### `getByZipcode()`
 
-Search for a Canton by a zipcode.
+Returns an array of possible Cantons for a given Zipcode. (Some zipcodes are shared between multiple Cantons).
+
+```php
+$cantonManager = new Wnx\SwissCantons\CantonManager();
+
+/** @var \Wnx\SwissCantons\Canton[] $cantons */
+$cantons = $cantonManager->getByZipcode(3005);
+```
+
+### `getByZipcodeAndCity()`
+
+Find Canton by a given zipcode and optionally by a city name.
 
 ```php
 $cantonManager = new Wnx\SwissCantons\CantonManager();
 
 /** @var \Wnx\SwissCantons\Canton $canton */
-$canton = $cantonManager->getByZipcode(3005);
+$canton = $cantonManager->getByZipcodeAndCity(1003);
+$canton = $cantonManager->getByZipcodeAndCity(1290, 'Lausanne');
 ```
 
 ## `Canton`

diff --git a/composer.json b/composer.json
@@ -18,7 +18,9 @@
     ],
     "require": {
         "php": "^8.2",
-        "ext-json": "*"
+        "ext-json": "*",
+        "ext-zip": "*",
+        "symfony/http-client": "^7.0"
     },
     "require-dev": {
         "phpunit/phpunit": "^10.1",

diff --git a/src/CantonManager.php b/src/CantonManager.php
@@ -2,31 +2,30 @@
 
 namespace Wnx\SwissCantons;
 
-use Wnx\SwissCantons\Exceptions\CantonException;
+use Wnx\SwissCantons\Exceptions\CantonNotFoundException;
 
 class CantonManager
 {
     protected CantonSearch $search;
-
-    protected ZipcodeSearch $zipcodeSearch;
+    protected CitySearch $citySearch;
 
     public function __construct()
     {
         $this->search = new CantonSearch();
-        $this->zipcodeSearch = new ZipcodeSearch();
+        $this->citySearch = new CitySearch();
     }
 
     /**
      * Get Canton by abbreviation.
      *
-     * @throws CantonException
+     * @throws CantonNotFoundException
      */
     public function getByAbbreviation(string $abbreviation): Canton
     {
         $result = $this->search->findByAbbreviation($abbreviation);
 
         if (is_null($result)) {
-            throw CantonException::notFoundForAbbreviation($abbreviation);
+            throw CantonNotFoundException::notFoundForAbbreviation($abbreviation);
         }
 
         return $result;
@@ -35,32 +34,76 @@ public function getByAbbreviation(string $abbreviation): Canton
     /**
      * Get Canton by Name.
      *
-     * @throws CantonException
+     * @throws CantonNotFoundException
      */
     public function getByName(string $name): Canton
     {
         $result = $this->search->findByName($name);
 
         if (is_null($result)) {
-            throw CantonException::notFoundForName($name);
+            throw CantonNotFoundException::notFoundForName($name);
         }
 
         return $result;
     }
 
     /**
-     * Get Canton by Zipcode.
+     * Get possible Cantons with a Zipcode.
      *
-     * @throws CantonException
+     * @param int $zipcode
+     * @return Canton[]
+     * @throws CantonNotFoundException
      */
-    public function getByZipcode(int $zipcode): Canton
+    public function getByZipcode(int $zipcode): array
     {
-        $result = $this->zipcodeSearch->findByZipcode($zipcode);
+        $cities = $this->citySearch->findByZipcode($zipcode);
 
-        if (is_null($result)) {
-            throw CantonException::notFoundForZipcode($zipcode);
+        // Get cantons abbreviations
+        $cantonAbbreviations = array_column($cities, 'canton');
+
+        // Remove duplicates
+        $cantonAbbreviations = array_unique($cantonAbbreviations);
+
+        // Search cantons by abbreviation
+        $cantons = array_map(fn (string $abbreviation) => $this->search->findByAbbreviation($abbreviation), $cantonAbbreviations);
+
+        // Call 'array_filter' without callback to remove null values
+        $cantons = array_filter($cantons);
+
+        if (empty($cantons)) {
+            throw CantonNotFoundException::notFoundForZipcode($zipcode);
+        }
+
+        return $cantons;
+    }
+
+    /**
+     * @param int $zipcode
+     * @param ?string $cityName
+     * @return Canton
+     * @throws CantonNotFoundException
+     */
+    public function getByZipcodeAndCity(int $zipcode, ?string $cityName = null): Canton
+    {
+        $cities = $this->citySearch->findByZipcode($zipcode);
+
+        if (1 === count($cities)) {
+            return $this->search->findByAbbreviation($cities[0]['canton'])
+                ?? throw CantonNotFoundException::notFoundForZipcode($zipcode);
         }
 
-        return $this->getByAbbreviation($result['canton']);
+        if (null !== $cityName) {
+            foreach ($cities as $city) {
+                if ($city['city'] === $cityName) {
+                    return $this->search->findByAbbreviation($city['canton'])
+                        ?? throw CantonNotFoundException::notFoundForZipcodeAndCity($zipcode, $cityName);
+                }
+            }
+
+            throw CantonNotFoundException::notFoundForZipcodeAndCity($zipcode, $cityName);
+        }
+
+        throw CantonNotFoundException::notFoundForZipcode($zipcode);
     }
+
 }
diff --git a/src/CantonSearch.php b/src/CantonSearch.php
@@ -4,16 +4,16 @@
 
 class CantonSearch
 {
-    protected array $dataSet;
+    protected array $dataset;
 
     public function __construct()
     {
-        $this->dataSet = (new Cantons())->getAll();
+        $this->dataset = (new Cantons())->getAll();
     }
 
     public function findByAbbreviation(string $abbreviation): ?Canton
     {
-        $result = array_filter($this->dataSet, fn (Canton $canton) => $canton->getAbbreviation() === strtoupper($abbreviation));
+        $result = array_filter($this->dataset, fn (Canton $canton) => $canton->getAbbreviation() === strtoupper($abbreviation));
 
         if (count($result) === 0) {
             return null;
@@ -24,7 +24,7 @@ public function findByAbbreviation(string $abbreviation): ?Canton
 
     public function findByName(string $name): ?Canton
     {
-        $result = array_filter($this->dataSet, fn (Canton $canton) => in_array($name, $canton->getNamesArray()));
+        $result = array_filter($this->dataset, fn (Canton $canton) => in_array($name, $canton->getNamesArray()));
 
         if (count($result) === 0) {
             return null;

diff --git a/src/CitySearch.php b/src/CitySearch.php
@@ -0,0 +1,44 @@
+<?php declare(strict_types=1);
+
+namespace Wnx\SwissCantons;
+
+/**
+ * @phpstan-type City array{canton: string, city: string, zipcode: int}
+ */
+class CitySearch
+{
+    /** @var City[] */
+    protected array $dataset;
+
+    public function __construct()
+    {
+        $this->dataset = $this->loadDataset();
+    }
+
+    /**
+     * Find Data Set for a City by Zipcode.
+     *
+     * @return City[]
+     */
+    public function findByZipcode(int $zipcode): array
+    {
+        return array_values(array_filter($this->dataset, fn (array $city) => $city['zipcode'] === $zipcode));
+    }
+
+    /**
+     * @return City[]
+     * @throws \JsonException
+     */
+    private function loadDataset(): array
+    {
+        return json_decode(file_get_contents(__DIR__.'/data/cities.json'), true, 512, JSON_THROW_ON_ERROR);
+    }
+
+    /**
+     * @return City[]
+     */
+    public function getDataSet(): array
+    {
+        return $this->dataset;
+    }
+}
diff --git a/src/Console/UpdateZipcodeDatasetCommand.php b/src/Console/UpdateZipcodeDatasetCommand.php
@@ -7,20 +7,34 @@
 use League\Csv\Reader;
 use League\Csv\Statement;
 use League\Csv\TabularDataReader;
+use RecursiveDirectoryIterator;
+use RecursiveIteratorIterator;
+use RuntimeException;
 use Symfony\Component\Console\Command\Command;
 use Symfony\Component\Console\Input\InputInterface;
 use Symfony\Component\Console\Output\OutputInterface;
+use Symfony\Component\HttpClient\HttpClient;
+use Symfony\Contracts\HttpClient\HttpClientInterface;
+use ZipArchive;
 
 class UpdateZipcodeDatasetCommand extends Command
 {
-    final public const PATH_TO_CSV = __DIR__ . '/../data/zipcodes.csv';
-    final public const PATH_TO_JSON = __DIR__ . '/../data/zipcodes.json';
+    final public const PATH_TO_CSV = __DIR__ . '/../data/cities.csv';
+    final public const PATH_TO_JSON = __DIR__ . '/../data/cities.json';
+
+    private HttpClientInterface $httpClient;
+
+    public function __construct(?HttpClientInterface $httpClient = null)
+    {
+        parent::__construct();
+        $this->httpClient = $httpClient ?? HttpClient::create();
+    }
 
     protected function configure(): void
     {
         $this
-            ->setName('update-zipcode-dataset')
-            ->setDescription('Fetch dataset from Swiss Post and create zipcodes.json file');
+            ->setName('update-cities-dataset')
+            ->setDescription('Fetch dataset from Swiss Post and create cities.json file');
     }
 
     /**
@@ -32,7 +46,7 @@ protected function execute(InputInterface $input, OutputInterface $output): int
         $output->writeln('🚧  Fetch dataset');
         $this->fetchDataset();
 
-        $output->writeln('🔮  Create zipcodes.json');
+        $output->writeln('🔮  Create cities.json');
         $records = $this->parseCsvDataset();
         $this->generateZipcodesFiles($records);
 
@@ -44,13 +58,56 @@ protected function execute(InputInterface $input, OutputInterface $output): int
         return 0;
     }
 
+    /**
+     * @return void
+     * @throws \Throwable
+     */
     protected function fetchDataset(): void
     {
-        $urlToDataset = "https://swisspost.opendatasoft.com/explore/dataset/plz_verzeichnis_v2/download/?format=csv&timezone=Europe/Berlin&lang=de&use_labels_for_header=true&csv_separator=%3B";
+        $urlToDataset = "https://data.geo.admin.ch/ch.swisstopo-vd.ortschaftenverzeichnis_plz/ortschaftenverzeichnis_plz/ortschaftenverzeichnis_plz_2056.csv.zip";
+
+        $response = $this->httpClient->request('GET', $urlToDataset);
 
-        $response = file_get_contents($urlToDataset);
+        // Check for successful response (200 OK)
+        if ($response->getStatusCode() !== 200) {
+            throw new RuntimeException("Failed to download file. Status code: " . $response->getStatusCode());
+        }
+
+        // Open the destination file for writing in binary mode
+        $fileHandler = fopen('/tmp/dataset.zip', 'wb');
+        if (!$fileHandler) {
+            throw new RuntimeException("Failed to open file for writing: '/tmp/dataset.zip'");
+        }
 
-        file_put_contents(self::PATH_TO_CSV, $response);
+        foreach ($this->httpClient->stream($response) as $chunk) {
+            fwrite($fileHandler, $chunk->getContent());
+        }
+
+        fclose($fileHandler);
+
+        $zip = new ZipArchive();
+        $res = $zip->open('/tmp/dataset.zip');
+        if ($res) {
+            $zip->extractTo('/tmp/extracted_dataset');
+            $zip->close();
+        }
+
+        $iterator = new RecursiveIteratorIterator(
+            new RecursiveDirectoryIterator('/tmp/extracted_dataset'),
+            RecursiveIteratorIterator::SELF_FIRST
+        );
+
+        foreach ($iterator as $file) {
+            if ($file->isFile() && pathinfo($file->getPathname(), PATHINFO_EXTENSION) === 'csv') {
+                $destinationFile = self::PATH_TO_CSV;
+                if (!rename($file->getPathname(), $destinationFile)) {
+                    // Handle potential errors during move operation
+                    throw new \Exception("Error moving file: " . $file->getPathname());
+                }
+
+                return;
+            }
+        }
     }
 
     /**
@@ -64,13 +121,7 @@ protected function parseCsvDataset(): TabularDataReader
         $csv->setHeaderOffset(0);
 
         return Statement::create()
-            ->where(fn ($record) => $record['KANTON'] !== 'FL')
-            ->orderBy(function (array $recordA, array $recordB): int {
-                if ($recordA['POSTLEITZAHL'] === $recordB["POSTLEITZAHL"]) {
-                    return $recordA['ORTBEZ27'] <=> $recordB['ORTBEZ27'];
-                }
-                return $recordA['POSTLEITZAHL'] <=> $recordB['POSTLEITZAHL'];
-            })
+            ->where(fn ($record) => $record['Kantonskürzel'] !== '')
             ->process($csv);
     }
 
@@ -80,9 +131,9 @@ protected function generateZipcodesFiles(TabularDataReader $records): void
 
         foreach ($records as $zipcodeRecord) {
             $data[] = [
-                'city' => $zipcodeRecord['ORTBEZ27'],
-                'zipcode' => (int) $zipcodeRecord['POSTLEITZAHL'],
-                'canton' => $zipcodeRecord['KANTON'],
+                'city' => $zipcodeRecord['Ortschaftsname'],
+                'zipcode' => (int) $zipcodeRecord['PLZ'],
+                'canton' => $zipcodeRecord['Kantonskürzel'],
             ];
         }
 
@@ -91,6 +142,7 @@ protected function generateZipcodesFiles(TabularDataReader $records): void
 
     protected function cleanup(): void
     {
+        unlink('/tmp/dataset.zip');
         unlink(self::PATH_TO_CSV);
     }
 }