Number formatter update.

Changelog excerpt: - Added some limited support to the number formatter class for unformatting a formatted number.
Maikuolan · Feb 5, 2025 · 16f390e · 16f390e
1 parent bd743fa
commit 16f390e
Show file tree

Hide file tree

Showing 6 changed files with 187 additions and 7 deletions.
diff --git a/.github/workflows/php-cs-fixer.yml b/.github/workflows/php-cs-fixer.yml
@@ -24,5 +24,5 @@ jobs:
 
   - name: PHP-CS-Fixer Check
     run: |
-     php-cs-fixer fix -v --dry-run ./src --rules=@PSR12
-     php-cs-fixer fix -v --dry-run ./.tests --rules=@PSR12
+     php-cs-fixer fix -v --dry-run ./src --rules=@PSR12,-blank_line_after_opening_tag
+     php-cs-fixer fix -v --dry-run ./.tests --rules=@PSR12,-blank_line_after_opening_tag
diff --git a/.tests/NumberFormatterTest.php b/.tests/NumberFormatterTest.php
@@ -227,3 +227,48 @@
     echo 'Test failed: ' . $Case . ':L' . __LINE__ . '().' . PHP_EOL;
     exit($ExitCode);
 }
+
+$Formats = [
+    'Arabic-1',
+    'Arabic-2',
+    'Arabic-3',
+    'Arabic-4',
+    'Bengali-1',
+    'Burmese-1',
+    'Fullwidth',
+    'Geez',
+    'India-1',
+    'India-2',
+    'India-3',
+    'India-4',
+    'India-5',
+    'India-6',
+    'Javanese',
+    'Kaktovik',
+    'Khmer-1',
+    'Lao-1',
+    'Latin-1',
+    'Latin-2',
+    'Latin-3',
+    'Latin-4',
+    'Latin-5',
+    'Mayan',
+    'Mongolian',
+    'Odia',
+    'Thai-1',
+    'Thai-2',
+    'Tibetan'
+];
+
+$ExitCode++;
+foreach ($Formats as $Format) {
+    $Obj = new \Maikuolan\Common\NumberFormatter($Format);
+    foreach (['123456789', '987654321', '102030405060708090', '100', '1000', '10000'] as $Number) {
+        $Try = $Obj->format($Number);
+        $Compare = $Obj->unformat($Try);
+        if ($Compare !== $Number) {
+            echo 'Test failed: ' . $Case . ':L' . __LINE__ . '(). ' . $Compare . ' !== ' . $Number . ' (' . $Format . ')!' . PHP_EOL;
+            exit($ExitCode);
+        }
+    }
+}
diff --git a/.tests/OperationTest.php b/.tests/OperationTest.php
@@ -280,8 +280,6 @@
     }
 }
 
-echo $Out;
-
 $ExitCode++;
 if ($Out !== $Expected) {
     echo 'Test failed: ' . $Case . ':L' . __LINE__ . '().' . PHP_EOL;

diff --git a/Changelog.txt b/Changelog.txt
@@ -11,7 +11,8 @@ found at:
 
 === Changes made since last versioned release ===
 
-(none)
+- [2025.05.02; Maikuolan]: Added some limited support to the number formatter
+  class for unformatting a formatted number.
 
 === Version/Release 2.12.3 ===
 PATCH RELEASE.

diff --git a/_docs/NumberFormatter.md b/_docs/NumberFormatter.md
@@ -10,6 +10,7 @@
 - [NumberFormatter constructor.](#numberformatter-constructor)
 - [format method.](#format-method)
 - [getSetJSON method.](#getsetjson-method)
+- [unformat method.](#unformat-method)
 - [ConversionSet member.](#conversionset-member)
 - [GroupSeparator member.](#groupseparator-member)
 - [GroupSize member.](#groupsize-member)
@@ -226,6 +227,52 @@ Output:
 ["\u06f0","\u06f1","\u06f2","\u06f3","\u06f4","\u06f5","\u06f6","\u06f7","\u06f8","\u06f9"]
 ```
 
+
+#### unformat method.
+
+Unformats the formatted number according to predefined patterns and lookup tables.
+
+*Warning: Doesn't work for ALL formats (..yet), won't work for fractions (only intended for whole numbers), and other data (e.g., decimal separators, thousands separators) will be disregarded entirely.*
+
+```PHP
+$Obj = new \Maikuolan\Common\NumberFormatter('Arabic-2');
+$Try = $Obj->format('12345678987654321');
+echo $Try . PHP_EOL;
+$Try = $Obj->unformat($Try);
+echo $Try . PHP_EOL;
+
+$Obj = new \Maikuolan\Common\NumberFormatter('India-2');
+$Try = $Obj->format('12345678987654321');
+echo $Try . PHP_EOL;
+$Try = $Obj->unformat($Try);
+echo $Try . PHP_EOL;
+
+$Obj = new \Maikuolan\Common\NumberFormatter('Geez');
+$Try = $Obj->format('12345678987654321');
+echo $Try . PHP_EOL;
+$Try = $Obj->unformat($Try);
+echo $Try . PHP_EOL;
+
+$Obj = new \Maikuolan\Common\NumberFormatter('Kaktovik');
+$Try = $Obj->format('12345678987654321');
+echo $Try . PHP_EOL;
+$Try = $Obj->unformat($Try);
+echo $Try . PHP_EOL;
+```
+
+Output:
+
+```
+١٢٬٣٤٥٬٦٧٨٬٩٨٧٬٦٥٤٬٣٢١
+12345678987654321
+१२,३४,५६,७८,९८,७६,५४,३२१
+12345678987654321
+፼፳፫፻፵፭፼፷፯፻፹፱፼፹፯፻፷፭፼፵፫፻፳፩
+12345678987654321
+�������������
+12345678987654321
+```
+
 #### ConversionSet member.
 
 ```PHP
@@ -547,4 +594,4 @@ Base 36: a.i000 ~ 74.9000
 ---
 
 
-Last Updated: 26 June 2024 (2024.06.26).
+Last Updated: 5 February 2025 (2025.02.05).
diff --git a/src/NumberFormatter.php b/src/NumberFormatter.php
@@ -1,6 +1,6 @@
 <?php
 /**
- * Number formatter (last modified: 2024.06.26).
+ * Number formatter (last modified: 2025.02.05).
  *
  * This file is a part of the "common classes package", utilised by a number of
  * packages and projects, including CIDRAM and phpMussel.
@@ -1058,6 +1058,63 @@ class NumberFormatter extends CommonAbstract
         'z' => '35'
     ];
 
+    /**
+     * @var array Lookup table for unformatting a number.
+     */
+    private $UnformatTable = [
+        '0' => ['٠', '۰', '০', '०', '૦', '੦', '೦', '౦', '၀', '០', '๐', '໐', '꧐', '୦', '༠', '᠐', '０', '᱐', '〇', '零', 'Z', '௰'],
+        '1' => ['١', '۱', '১', '१', '૧', '੧', '೧', '౧', '၁', '១', '๑', '໑', '꧑', '୧', '༡', '᠑', '１', '᱑', '一', '壹', '፩', '፲', '௧'],
+        '2' => ['٢', '۲', '২', '२', '૨', '੨', '೨', '౨', '၂', '២', '๒', '໒', '꧒', '୨', '༢', '᠒', '２', '᱒', '二', '贰', '貳', '፪', '፳', '௨'],
+        '3' => ['٣', '۳', '৩', '३', '૩', '੩', '೩', '౩', '၃', '៣', '๓', '໓', '꧓', '୩', '༣', '᠓', '３', '᱓', '三', '叁', '叄', '፫', '፴', '௩'],
+        '4' => ['٤', '۴', '৪', '४', '૪', '੪', '೪', '౪', '၄', '៤', '๔', '໔', '꧔', '୪', '༤', '᠔', '４', '᱔', '四', '肆', '፬', '፵', '௪'],
+        '5' => ['٥', '۵', '৫', '५', '૫', '੫', '೫', '౫', '၅', '៥', '๕', '໕', '꧕', '୫', '༥', '᠕', '５', '᱕', '五', '伍', '፭', '፶', '௫'],
+        '6' => ['٦', '۶', '৬', '६', '૬', '੬', '೬', '౬', '၆', '៦', '๖', '໖', '꧖', '୬', '༦', '᠖', '６', '᱖', '六', '陆', '陸', '፮', '፷', '௬'],
+        '7' => ['٧', '۷', '৭', '७', '૭', '੭', '೭', '౭', '၇', '៧', '๗', '໗', '꧗', '୭', '༧', '᠗', '７', '᱗', '七', '柒', '፯', '፸', '௭'],
+        '8' => ['٨', '۸', '৮', '८', '૮', '੮', '೮', '౮', '၈', '៨', '๘', '໘', '꧘', '୮', '༨', '᠘', '８', '᱘', '八', '捌', '፰', '፹', '௮'],
+        '9' => ['٩', '۹', '৯', '९', '૯', '੯', '೯', '౯', '၉', '៩', '๙', '໙', '꧙', '୯', '༩', '᠙', '９', '᱙', '九', '玖', '፱', '፺', '௯']
+    ];
+
+    /**
+     * @var array Patterns for unformatting a number.
+     */
+    private $UnformatPattern = [
+        '~(?<!一|二|三|四|五|六|七|八|九|十|百|千)(十|百|千|拾|万|億|兆|京|垓)~' => '1\1',
+        '~^(፻|፼|十|百|千|拾|万|億|兆|京|垓|௰|௱|௲)~' => '1\1',
+        '~(፻|፼)(?!፲|፳|፴|፵|፶|፷|፸|፹|፺|\d)~' => '\1Z',
+        '~(፻[\dZ]|፼[\dZ])(?!፩|፪|፫|፬|፭|፮|፯|፰|፱|\d)~' => '\1Z',
+        '~(፲|፳|፴|፵|፶|፷|፸|፹|፺)(?!፩|፪|፫|፬|፭|፮|፯|፰|፱)~' => '\1Z',
+        '~(十|拾)$~' => '0',
+        '~(፻|百)$~' => '00',
+        '~千$~' => '000',
+        '~፼$~' => '0000'
+    ];
+
+    /**
+     * @var array Lookup table for unformatting a base-20 number.
+     */
+    private $UnformatTableKakMay = [
+        '0' => ['𝋀', '𝋠'],
+        '1' => ['𝋁', '𝋡'],
+        '2' => ['𝋂', '𝋢'],
+        '3' => ['𝋃', '𝋣'],
+        '4' => ['𝋄', '𝋤'],
+        '5' => ['𝋅', '𝋥'],
+        '6' => ['𝋆', '𝋦'],
+        '7' => ['𝋇', '𝋧'],
+        '8' => ['𝋈', '𝋨'],
+        '9' => ['𝋉', '𝋩'],
+        'a' => ['𝋊', '𝋪'],
+        'b' => ['𝋋', '𝋫'],
+        'c' => ['𝋌', '𝋬'],
+        'd' => ['𝋍', '𝋭'],
+        'e' => ['𝋎', '𝋮'],
+        'f' => ['𝋏', '𝋯'],
+        'g' => ['𝋐', '𝋰'],
+        'h' => ['𝋑', '𝋱'],
+        'i' => ['𝋒', '𝋲'],
+        'j' => ['𝋓', '𝋳'],
+    ];
+
     /**
      * Constructor.
      *
@@ -1384,6 +1441,38 @@ public function getSetJSON(string $Set = ''): string
         return isset($this->{$Set}) ? json_encode($this->{$Set}) : '[]';
     }
 
+    /**
+     * Unformats the formatted number according to predefined patterns and lookup
+     * tables. Warning: Doesn't work for all formats (..yet), won't work for
+     * fractions (only intended for whole numbers), and other data (e.g., decimal
+     * separators, thousands separators) will be disregarded entirely.
+     *
+     * @param string $Number The number to unformat.
+     * @return string The unformatted number (returned as string rather than as an
+     *      integer or a float in order to retain decimal precision).
+     */
+    public function unformat(string $Number): string
+    {
+        if (preg_match('~\D~', $Number)) {
+            foreach ($this->UnformatPattern as $Pattern => $Replacement) {
+                $Number = preg_replace($Pattern, $Replacement, $Number);
+            }
+            foreach ($this->UnformatTable as $Replacement => $Lookup) {
+                $Number = str_replace($Lookup, $Replacement, $Number);
+            }
+            $KakMay = $Number;
+            foreach ($this->UnformatTableKakMay as $Replacement => $Lookup) {
+                $KakMay = str_replace($Lookup, $Replacement, $KakMay);
+            }
+            if ($KakMay !== $Number) {
+                $Number = base_convert(preg_replace('~[^\da-j]~', '', $KakMay), 20, 10);
+            } else {
+                $Number = preg_replace('~\D~', '', $Number);
+            }
+        }
+        return preg_replace('~^0+~', '', $Number);
+    }
+
     /**
      * Prepare to convert a fraction.
      *