Merge pull request #1 from Moxio/multi-letter

Add (optional) support for multi-letter list markers
Moxio · Oct 25, 2022 · 36972d0 · 36972d0
2 parents 4849c16 + 18bfc6e
commit 36972d0
Show file tree

Hide file tree

Showing 8 changed files with 218 additions and 49 deletions.
diff --git a/README.md b/README.md
@@ -151,6 +151,24 @@ Supported configuration options:
 
   Because the ordinal indicator is commonly confused with other characters like the degree symbol, these
   characters are tolerated and considered equivalent to the ordinal indicator.
+* `allow_multi_letter` - Whether to allow multi-letter alphabetic numerals, to number lists beyond 26
+  (default: `false`). If this option is enabled,
+  input like
+  ```markdown
+  AA. foo
+  AB. bar
+  AC. baz
+  ```
+  will be converted to
+  ```html
+  <ol type="A" start="27">
+    <li>foo</li>
+    <li>bar</li>
+    <li>baz</li>
+  </ol>
+  ```
+  Multi-letter alphabetic numerals can consist of at most 3 characters, which should be enough for a
+  typical list.
 
 Versioning
 ----------

diff --git a/src/ListParser.php b/src/ListParser.php
@@ -78,7 +78,7 @@ public function parse(ContextInterface $context, Cursor $cursor): bool
             $number = null;
             $numberingType = null;
             $hasOrdinalIndicator = false;
-        } elseif (($matches = RegexHelper::matchAll('/^(\d{1,9}|[a-z]|[A-Z]|[ivxlcdm]+|[IVXLCDM]+|#)([\x{00BA}\x{00B0}\x{02DA}\x{1D52}]?)([.)])/u', $rest)) && (!($context->getContainer() instanceof Paragraph) || in_array($matches[1], ['1', 'a', 'A', 'i', 'I', '#'], true))) {
+        } elseif (($matches = RegexHelper::matchAll('/^(\d{1,9}|[a-z]{1,3}|[A-Z]{1,3}|[ivxlcdm]+|[IVXLCDM]+|#)([\x{00BA}\x{00B0}\x{02DA}\x{1D52}]?)([.)])/u', $rest)) && (!($context->getContainer() instanceof Paragraph) || in_array($matches[1], ['1', 'a', 'A', 'i', 'I', '#'], true))) {
             $data = new ListData();
             $data->markerOffset = $indent;
             $data->type = ListBlock::TYPE_ORDERED;
@@ -102,14 +102,14 @@ public function parse(ContextInterface $context, Cursor $cursor): bool
                 } catch (RomansLexerException|RomansParserException $e) {
                     $isValidRoman = false;
                 }
-                $isValidAlpha = strlen($matches[1]) === 1;
+                $isValidAlpha = strlen($matches[1]) === 1 || $this->config->get('allow_multi_letter', false);
                 $preferRomanOverAlpha = $withinRomanList || (!$withinAlphaList && $matches[1] === 'I');
 
                 if ($isValidRoman && (!$isValidAlpha || $preferRomanOverAlpha)) {
                     $data->start = $parsedRomanNumber;
                     $numberingType = 'I';
                 } else if ($isValidAlpha) {
-                    $data->start = ord($matches[1]) - ord('A') + 1;
+                    $data->start = NumberingUtils::convertAlphaMarkerToOrdinalNumber($matches[1]);
                     $numberingType = 'A';
                 } else {
                     return false;
@@ -123,14 +123,14 @@ public function parse(ContextInterface $context, Cursor $cursor): bool
                 } catch (RomansLexerException|RomansParserException $e) {
                     $isValidRoman = false;
                 }
-                $isValidAlpha = strlen($matches[1]) === 1;
+                $isValidAlpha = strlen($matches[1]) === 1 || $this->config->get('allow_multi_letter', false);
                 $preferRomanOverAlpha = $withinRomanList || (!$withinAlphaList && $matches[1] === 'i');
 
                 if ($isValidRoman && (!$isValidAlpha || $preferRomanOverAlpha)) {
                     $data->start = $parsedRomanNumber;
                     $numberingType = 'i';
                 } else if ($isValidAlpha) {
-                    $data->start = ord($matches[1]) - ord('a') + 1;
+                    $data->start = NumberingUtils::convertAlphaMarkerToOrdinalNumber($matches[1]);
                     $numberingType = 'a';
                 } else {
                     return false;

diff --git a/src/NumberingUtils.php b/src/NumberingUtils.php
@@ -0,0 +1,16 @@
+<?php
+namespace Moxio\CommonMark\Extension\FancyLists;
+
+class NumberingUtils
+{
+    public static function convertAlphaMarkerToOrdinalNumber(string $alphaMarker): int
+    {
+        $lastLetterValue = ord(strtolower($alphaMarker[-1])) - ord('a') + 1;
+        if (strlen($alphaMarker) > 1) {
+            $prefixValue = self::convertAlphaMarkerToOrdinalNumber(substr($alphaMarker, 0, -1));
+            return $prefixValue * 26 + $lastLetterValue;
+        } else {
+            return $lastLetterValue;
+        }
+    }
+}
diff --git a/test/AbstractIntegrationTest.php b/test/AbstractIntegrationTest.php
@@ -0,0 +1,26 @@
+<?php
+namespace Moxio\CommonMark\Extension\FancyLists\Test;
+
+use League\CommonMark\DocParser;
+use League\CommonMark\Environment;
+use League\CommonMark\HtmlRenderer;
+use Moxio\CommonMark\Extension\FancyLists\FancyListsExtension;
+use PHPUnit\Framework\TestCase;
+
+abstract class AbstractIntegrationTest extends TestCase
+{
+    protected function assertMarkdownIsConvertedTo(string $expectedHtml, string $markdown, ?array $config = null): void
+    {
+        $environment = Environment::createCommonMarkEnvironment();
+        $environment->addExtension(new FancyListsExtension());
+        if ($config !== null) {
+            $environment->setConfig($config);
+        }
+
+        $parser = new DocParser($environment);
+        $renderer = new HtmlRenderer($environment);
+        $actualOutput = $renderer->renderBlock($parser->parse($markdown));
+
+        $this->assertXmlStringEqualsXmlString("<html>$expectedHtml</html>", "<html>$actualOutput</html>");
+    }
+}
diff --git a/test/IntegrationTest.php b/test/IntegrationTest.php
@@ -1,13 +1,7 @@
 <?php
 namespace Moxio\CommonMark\Extension\FancyLists\Test;
 
-use League\CommonMark\DocParser;
-use League\CommonMark\Environment;
-use League\CommonMark\HtmlRenderer;
-use Moxio\CommonMark\Extension\FancyLists\FancyListsExtension;
-use PHPUnit\Framework\TestCase;
-
-class IntegrationTest extends TestCase
+class IntegrationTest extends AbstractIntegrationTest
 {
     // Testcase from https://spec.commonmark.org/0.29/#example-272
     public function testDoesNotAlterOrdinaryOrderedListSyntax(): void
@@ -471,19 +465,4 @@ public function testRequiresTwoSpacesAfterACapitalLetterAndAPeriod(): void
 
         $this->assertMarkdownIsConvertedTo($expectedHtml, $markdown);
     }
-
-    public function assertMarkdownIsConvertedTo(string $expectedHtml, string $markdown, ?array $config = null): void
-    {
-        $environment = Environment::createCommonMarkEnvironment();
-        $environment->addExtension(new FancyListsExtension());
-        if ($config !== null) {
-            $environment->setConfig($config);
-        }
-
-        $parser = new DocParser($environment);
-        $renderer = new HtmlRenderer($environment);
-        $actualOutput = $renderer->renderBlock($parser->parse($markdown));
-
-        $this->assertXmlStringEqualsXmlString("<html>$expectedHtml</html>", "<html>$actualOutput</html>");
-    }
 }
diff --git a/test/MultiLetterIntegrationTest.php b/test/MultiLetterIntegrationTest.php
@@ -0,0 +1,117 @@
+<?php
+namespace Moxio\CommonMark\Extension\FancyLists\Test;
+
+class MultiLetterIntegrationTest extends AbstractIntegrationTest
+{
+    public function testDoesNotSupportMultiLetterListMarkersByDefault(): void
+    {
+        $markdown = <<<MD
+AA) foo
+AB) bar
+AC) baz
+MD;
+        $expectedHtml = <<<HTML
+<p>AA) foo
+AB) bar
+AC) baz</p>
+HTML;
+
+        $this->assertMarkdownIsConvertedTo($expectedHtml, $markdown);
+    }
+
+    public function testSupportsMultiLetterListMarkersIfEnabledInConfiguration(): void
+    {
+        $markdown = <<<MD
+AA) foo
+AB) bar
+AC) baz
+MD;
+        $expectedHtml = <<<HTML
+<ol type="A" start="27">
+  <li>foo</li>
+  <li>bar</li>
+  <li>baz</li>
+</ol>
+HTML;
+
+        $this->assertMarkdownIsConvertedTo($expectedHtml, $markdown, [
+            'allow_multi_letter' => true,
+        ]);
+    }
+
+    public function testSupportsContinuingASingleLetterListWithMultiLetterListMarkers(): void
+    {
+        $markdown = <<<MD
+Z) foo
+AA) bar
+AB) baz
+MD;
+        $expectedHtml = <<<HTML
+<ol type="A" start="26">
+  <li>foo</li>
+  <li>bar</li>
+  <li>baz</li>
+</ol>
+HTML;
+
+        $this->assertMarkdownIsConvertedTo($expectedHtml, $markdown, [
+            'allow_multi_letter' => true,
+        ]);
+    }
+
+    public function testSupportsLowercaseMultiLetterListMarkers(): void
+    {
+        $markdown = <<<MD
+aa) foo
+ab) bar
+ac) baz
+MD;
+        $expectedHtml = <<<HTML
+<ol type="a" start="27">
+  <li>foo</li>
+  <li>bar</li>
+  <li>baz</li>
+</ol>
+HTML;
+
+        $this->assertMarkdownIsConvertedTo($expectedHtml, $markdown, [
+            'allow_multi_letter' => true,
+        ]);
+    }
+
+    public function testAllowsAtMost3CharactersForMultiLetterListMarkers(): void
+    {
+        $markdown = <<<MD
+AAAA) foo
+AAAB) bar
+AAAC) baz
+MD;
+        $expectedHtml = <<<HTML
+<p>AAAA) foo
+AAAB) bar
+AAAC) baz</p>
+HTML;
+
+        $this->assertMarkdownIsConvertedTo($expectedHtml, $markdown, [
+            'allow_multi_letter' => true,
+        ]);
+    }
+
+    public function testDoesNotSupportMixingUppercaseAndLowercaseLetters(): void
+    {
+        $markdown = <<<MD
+Aa) foo
+Ab) bar
+Ac) baz
+MD;
+        $expectedHtml = <<<HTML
+<p>Aa) foo
+Ab) bar
+Ac) baz</p>
+HTML;
+
+        $this->assertMarkdownIsConvertedTo($expectedHtml, $markdown, [
+            'allow_multi_letter' => true,
+        ]);
+    }
+}
diff --git a/test/NumberingUtilsTest.php b/test/NumberingUtilsTest.php
@@ -0,0 +1,34 @@
+<?php
+namespace Moxio\CommonMark\Extension\FancyLists\Test;
+
+use Moxio\CommonMark\Extension\FancyLists\NumberingUtils;
+use PHPUnit\Framework\TestCase;
+
+class NumberingUtilsTest extends TestCase
+{
+    public function testCanConvertAlphaListMarkersToOrdinalNumber(): void
+    {
+        $this->assertSame(1, NumberingUtils::convertAlphaMarkerToOrdinalNumber("A"));
+        $this->assertSame(2, NumberingUtils::convertAlphaMarkerToOrdinalNumber("B"));
+        $this->assertSame(26, NumberingUtils::convertAlphaMarkerToOrdinalNumber("Z"));
+        $this->assertSame(27, NumberingUtils::convertAlphaMarkerToOrdinalNumber("AA"));
+        $this->assertSame(52, NumberingUtils::convertAlphaMarkerToOrdinalNumber("AZ"));
+        $this->assertSame(53, NumberingUtils::convertAlphaMarkerToOrdinalNumber("BA"));
+        $this->assertSame(677, NumberingUtils::convertAlphaMarkerToOrdinalNumber("ZA"));
+        $this->assertSame(702, NumberingUtils::convertAlphaMarkerToOrdinalNumber("ZZ"));
+        $this->assertSame(703, NumberingUtils::convertAlphaMarkerToOrdinalNumber("AAA"));
+    }
+
+    public function testCanAlsoConvertLowercaseAlphaListMarkers(): void
+    {
+        $this->assertSame(1, NumberingUtils::convertAlphaMarkerToOrdinalNumber("a"));
+        $this->assertSame(2, NumberingUtils::convertAlphaMarkerToOrdinalNumber("b"));
+        $this->assertSame(26, NumberingUtils::convertAlphaMarkerToOrdinalNumber("z"));
+        $this->assertSame(27, NumberingUtils::convertAlphaMarkerToOrdinalNumber("aa"));
+        $this->assertSame(52, NumberingUtils::convertAlphaMarkerToOrdinalNumber("az"));
+        $this->assertSame(53, NumberingUtils::convertAlphaMarkerToOrdinalNumber("ba"));
+        $this->assertSame(677, NumberingUtils::convertAlphaMarkerToOrdinalNumber("za"));
+        $this->assertSame(702, NumberingUtils::convertAlphaMarkerToOrdinalNumber("zz"));
+        $this->assertSame(703, NumberingUtils::convertAlphaMarkerToOrdinalNumber("aaa"));
+    }
+}
diff --git a/test/OrdinalIntegrationTest.php b/test/OrdinalIntegrationTest.php
@@ -1,13 +1,7 @@
 <?php
 namespace Moxio\CommonMark\Extension\FancyLists\Test;
 
-use League\CommonMark\DocParser;
-use League\CommonMark\Environment;
-use League\CommonMark\HtmlRenderer;
-use Moxio\CommonMark\Extension\FancyLists\FancyListsExtension;
-use PHPUnit\Framework\TestCase;
-
-class OrdinalIntegrationTest extends TestCase
+class OrdinalIntegrationTest extends AbstractIntegrationTest
 {
     public function testDoesNotSupportAnOrdinalIndicatorByDefault(): void
     {
@@ -112,19 +106,4 @@ public function testToleratesCharactersCommonlyMistakenForOrdinalIndicators(): v
             'allow_ordinal' => true,
         ]);
     }
-
-    public function assertMarkdownIsConvertedTo(string $expectedHtml, string $markdown, ?array $config = null): void
-    {
-        $environment = Environment::createCommonMarkEnvironment();
-        $environment->addExtension(new FancyListsExtension());
-        if ($config !== null) {
-            $environment->setConfig($config);
-        }
-
-        $parser = new DocParser($environment);
-        $renderer = new HtmlRenderer($environment);
-        $actualOutput = $renderer->renderBlock($parser->parse($markdown));
-
-        $this->assertXmlStringEqualsXmlString("<html>$expectedHtml</html>", "<html>$actualOutput</html>");
-    }
 }