implement json source

Author: recca0120

resources/Zip32_utf8_10501_1.zip

@@ -2,21 +2,21 @@
 
 include __DIR__.'/../vendor/autoload.php';
 
-use Recca0120\Twzipcode\Sources\CSV;
+use Recca0120\Twzipcode\Sources\Csv;
+use Recca0120\Twzipcode\Sources\Json;
 use Recca0120\Twzipcode\Storages\File;
 
+// https://data.gov.tw/dataset/5948
+$downloadUrl = 'https://quality.data.gov.tw/dq_download_json.php?nid=5948&md5_url=e1f6004ad33eb3ff3a824fb992a4b01a';
+$extension = 'json';
+$file = __DIR__.'/Zip32_11208.zip';
+
 set_error_handler(static function ($severity, $message, $file, $line) {
     throw new ErrorException($message, $severity, $severity, $file, $line);
 });
 
-$start = microtime(true);
-$file = __DIR__.'/Zip32_utf8_10501_1.zip';
-
-// https://data.gov.tw/dataset/5948
-$url = 'https://quality.data.gov.tw/dq_download_csv.php?nid=5948&md5_url=e1f6004ad33eb3ff3a824fb992a4b01a';
-
-if (file_exists($file) === false) {
-    touch($file);
+function csv($url)
+{
     $contents = file_get_contents($url);
 
     $encoding = mb_detect_encoding($contents, ['UCS-2LE', 'BIG5', 'UTF-8']);
@@ -30,11 +30,32 @@
         throw new RuntimeException($contents);
     }
 
+    return $contents;
+}
+
+function json($url)
+{
+    return file_get_contents($url);
+}
+
+$start = microtime(true);
+if (file_exists($file) === false) {
+    $contents = $extension($downloadUrl);
+
+    touch($file);
     $zip = new ZipArchive;
     $zip->open($file, ZipArchive::OVERWRITE);
-    $zip->addFromString(pathinfo($file, PATHINFO_FILENAME).'.csv', $contents);
+    $zip->addFromString(pathinfo($file, PATHINFO_FILENAME).'.'.$extension, $contents);
     $zip->close();
 }
 
-(new File)->load(new CSV($file));
+$lookup = [
+    'csv' => Csv::class,
+    'json' => Json::class,
+];
+$class = $lookup[$extension];
+
+$source = new $class($file);
+
+(new File)->load($source);
 echo 'benchmark: '.(microtime(true) - $start)."\n";

resources/converter.php

@@ -2,7 +2,7 @@
 
 namespace Recca0120\Twzipcode\Sources;
 
-class CSV extends Source
+class Csv extends Source
 {
     /** @var string */
     protected $file;

resources/data/zip5.rules

@@ -0,0 +1,21 @@
+<?php
+
+namespace Recca0120\Twzipcode\Sources;
+
+class Json extends Csv
+{
+    /**
+     * @return array{array{zipcode: string, county: string, district: string, text: string}} $rows
+     */
+    protected function rows()
+    {
+        return array_map(static function ($data) {
+            return [
+                'zipcode' => $data['郵遞區號'],
+                'county' => $data['縣市名稱'],
+                'district' => $data['鄉鎮市區'],
+                'rule' => implode(',', $data),
+            ];
+        }, json_decode($this->contents(), true));
+    }
+}

src/Sources/CSV.php src/Sources/Csv.php

@@ -33,10 +33,15 @@ protected function rows()
             return ! empty(trim($line));
         });
 
-        return array_map(static function ($line) {
-            $data = explode(',', $line);
+        return array_map(static function ($rule) {
+            $data = explode(',', $rule);
 
-            return ['zipcode' => $data[0], 'county' => $data[1], 'district' => $data[2], 'text' => $line];
+            return [
+                'zipcode' => $data[0],
+                'county' => $data[1],
+                'district' => $data[2],
+                'rule' => $rule,
+            ];
         }, $lines);
     }
 
@@ -51,7 +56,7 @@ protected static function prepare($rows)
                 ? self::$tricks[$row['county'].$row['district']]
                 : substr($row['zipcode'], 0, 3);
 
-            $results[$row['county']][$row['district']][$zip3][] = $row['text'];
+            $results[$row['county']][$row['district']][$zip3][] = $row['rule'];
 
             return $results;
         }, []);

src/Sources/Json.php

@@ -0,0 +1,42 @@
+<?php
+
+namespace Recca0120\Twzipcode\Tests\Sources;
+
+use PHPUnit\Framework\TestCase;
+use Recca0120\Twzipcode\Sources\Csv;
+
+class CsvTest extends TestCase
+{
+    public function test_csv_contents()
+    {
+        $source = new StubCsv('test.csv');
+        $source->setContents('10058,臺北市,中正區,八德路１段,全
+10079,臺北市,中正區,三元街,單全
+');
+
+        $source->each(function ($zipcode, $county, $district, $rules) {
+            self::assertEquals(100, $zipcode);
+            self::assertEquals('臺北市', $county);
+            self::assertEquals('中正區', $district);
+            self::assertEquals([
+                '10058,臺北市,中正區,八德路１段,全',
+                '10079,臺北市,中正區,三元街,單全',
+            ], $rules);
+        });
+    }
+}
+
+class StubCsv extends Csv
+{
+    private $contents = '';
+
+    public function setContents($contents)
+    {
+        $this->contents = $contents;
+    }
+
+    public function contents()
+    {
+        return $this->contents;
+    }
+}

src/Sources/Source.php

@@ -0,0 +1,40 @@
+<?php
+
+namespace Recca0120\Twzipcode\Tests\Sources;
+
+use PHPUnit\Framework\TestCase;
+use Recca0120\Twzipcode\Sources\Json;
+
+class JsonTest extends TestCase
+{
+    public function test_csv_contents()
+    {
+        $source = new StubJson('test.json');
+        $source->setContents('[{"郵遞區號":"10058","縣市名稱":"臺北市","鄉鎮市區":"中正區","原始路名":"八德路１段","投遞範圍":"全"},{"郵遞區號":"10079","縣市名稱":"臺北市","鄉鎮市區":"中正區","原始路名":"三元街","投遞範圍":"單全"}]');
+
+        $source->each(function ($zipcode, $county, $district, $rules) {
+            self::assertEquals(100, $zipcode);
+            self::assertEquals('臺北市', $county);
+            self::assertEquals('中正區', $district);
+            self::assertEquals([
+                '10058,臺北市,中正區,八德路１段,全',
+                '10079,臺北市,中正區,三元街,單全',
+            ], $rules);
+        });
+    }
+}
+
+class StubJson extends Json
+{
+    private $contents = '';
+
+    public function setContents($contents)
+    {
+        $this->contents = $contents;
+    }
+
+    public function contents()
+    {
+        return $this->contents;
+    }
+}

tests/Sources/CsvTest.php

@@ -7,7 +7,7 @@
 use org\bovigo\vfs\vfsStream;
 use PHPUnit\Framework\TestCase;
 use Recca0120\Twzipcode\Address;
-use Recca0120\Twzipcode\Sources\CSV;
+use Recca0120\Twzipcode\Sources\Csv;
 use Recca0120\Twzipcode\Sources\Text;
 use Recca0120\Twzipcode\Storages\File as Storage;
 
@@ -125,7 +125,7 @@ public function testLoadResources()
         Storage::$cached = ['zip3' => null, 'zip5' => null];
         $root = vfsStream::setup();
         $storage = new Storage($root->url());
-        $storage->flush()->load(new CSV(__DIR__.'/../../resources/Zip32_utf8_10501_1.zip'));
+        $storage->flush()->load(new Csv(__DIR__.'/../../resources/Zip32_utf8_10501_1.zip'));
 
         $address = m::mock(Address::class);