implement json source

.gitattributes

@@ -9,3 +9,5 @@
 .php_cs export-ignore
 .travis.yml export-ignore
 phpunit.xml.dist export-ignore
+pint.json export-ignore
+

resources/converter.php

@@ -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';
+$file = __DIR__.'/Zip32_11208.'.$extension.'.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 +30,29 @@
         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";

src/Address.php

@@ -21,9 +21,6 @@ class Address
     /** @var Normalizer */
     public $normalizer;
 
-    /** @var Tricky */
-    public $tricky;
-
     /** @var JArray */
     public $tokens = [];
 

src/Rule.php

@@ -138,7 +138,8 @@ private function tokenize($rule, Closure $addressResolver)
             '[連至單雙全](?=[\d全]|$)',
         ];
 
-        $addressResolver($this->normalize($rule)->replace('/'.implode('|', $pattern).'/u',
+        $addressResolver($this->normalize($rule)->replace(
+            '/'.implode('|', $pattern).'/u',
             function ($m) use (&$tokens) {
                 $token = &$m[0];
                 if ($token === '連') {
@@ -148,7 +149,8 @@ function ($m) use (&$tokens) {
                 $tokens[] = $token;
 
                 return $token === '附號全' ? '號' : '';
-            }));
+            }
+        ));
 
         return $tokens;
     }
@@ -187,8 +189,11 @@ private function normalizeAddress(Address $address, JArray $ruleAddressTokens)
 
         return new Address(
             new JArray($address->tokens()->filter(function ($token) use ($removeUnits) {
-                return isset($token[Address::UNIT]) === true && in_array($token[Address::UNIT], $removeUnits,
-                    true) === false;
+                return isset($token[Address::UNIT]) === true && in_array(
+                    $token[Address::UNIT],
+                    $removeUnits,
+                    true
+                ) === false;
             })->map(function ($token) {
                 return implode('', $token);
             }))

src/Rules.php

@@ -14,9 +14,6 @@ class Rules
      */
     private $storage;
 
-    /**
-     * @param  Storage|null  $storage
-     */
     public function __construct(Storage $storage = null)
     {
         $this->storage = $storage ?: new File;

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

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

src/Sources/Json.php

@@ -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));
+    }
+}