Add processors for table rowspan and colspan issues

src/Converter/DokuwikiConverter.php

@@ -7,6 +7,9 @@
 use HalloWelt\MediaWiki\Lib\Migration\Workspace;
 use HalloWelt\MigrateDokuwiki\Converter\PostProcessors\Image as ImagePostProcessor;
 use HalloWelt\MigrateDokuwiki\Converter\PostProcessors\Link as PostProcessorsLink;
+use HalloWelt\MigrateDokuwiki\Converter\PostProcessors\Table\Colspan as ColspanPostProcessor;
+use HalloWelt\MigrateDokuwiki\Converter\PostProcessors\Table\Rowspan as RowspanPostProcessor;
+use HalloWelt\MigrateDokuwiki\Converter\PreProcessors\Table\Colspan as ColspanPreProcessor;
 use HalloWelt\MigrateDokuwiki\Converter\Processors\Image as ImageProcessor;
 use HalloWelt\MigrateDokuwiki\Converter\Processors\Link;
 use HalloWelt\MigrateDokuwiki\IProcessor;
@@ -25,7 +28,9 @@ class DokuwikiConverter extends PandocDokuwiki implements IOutputAwareInterface
 	 * @return array
 	 */
 	private function getPreProcessors(): array {
-		return [];
+		return [
+			new ColspanPreProcessor()
+		];
 	}
 
 	/**
@@ -44,7 +49,9 @@ private function getProcessors(): array {
 	private function getPostProcessors(): array {
 		return [
 			new ImagePostProcessor(),
-			new PostProcessorsLink()
+			new PostProcessorsLink(),
+			new ColspanPostProcessor(),
+			new RowspanPostProcessor()
 		];
 	}
 

src/Converter/PostProcessors/Table/Colspan.php

@@ -0,0 +1,77 @@
+<?php
+
+namespace HalloWelt\MigrateDokuwiki\Converter\PostProcessors\Table;
+
+use HalloWelt\MigrateDokuwiki\IProcessor;
+
+class Colspan implements IProcessor {
+
+	/**
+	 * @param string $text
+	 * @return string
+	 */
+	public function process( string $text ): string {
+		$lines = explode( "\n", $text );
+
+		foreach ( $lines as $lineIndex => &$line ) {
+			$line = trim( $line );
+
+			$colspanPos = strpos( $line, "###COLSPAN_" );
+			if ( $colspanPos !== false ) {
+				// In wikitext one line = one cell
+				// One cell may have few blocks separated by "|"
+				$cellBlocks = explode( '|', $line );
+
+				$colspanCount = 0;
+
+				foreach ( $cellBlocks as $cellBlockIndex => $cellBlock ) {
+					$matches = [];
+
+					preg_match( "/###COLSPAN_(.*?)###/", $cellBlock, $matches );
+
+					// Current cell block contains "colspan"
+					if ( isset( $matches[1] ) ) {
+						$colspanCount = (int)$matches[1];
+
+						// If cell already contains block with HTML attributes (like "style" or "colspan")
+						// Then just append "colspan" there
+						if ( count( $cellBlocks ) > 2 ) {
+							$cellBlocks[$cellBlockIndex - 1] = $cellBlocks[$cellBlockIndex - 1] . " colspan=\"$colspanCount\"";
+						} else {
+							// Otherwise add such block
+							$cellBlocks = array_merge(
+								[
+									$cellBlocks[0]
+								],
+								[
+									"colspan=\"$colspanCount\""
+								],
+								array_slice( $cellBlocks, 1 )
+							);
+						}
+
+						// Remove "###COLSPAN_<N>###" string from current cell block
+						$cellBlocks[$cellBlockIndex] = str_replace(
+							"###COLSPAN_$colspanCount###", "", $cellBlocks[$cellBlockIndex]
+						);
+					}
+				}
+
+				$line = implode( "|", $cellBlocks );
+
+				// If we found "colspan" on current line -
+				// then remove corresponding amount of next redundant empty lines (produced by Pandoc)
+				if ( $colspanCount > 0 ) {
+					for ( $i = $lineIndex + 1; $i < $lineIndex + $colspanCount; $i++ ) {
+						unset( $lines[$i] );
+					}
+				}
+			}
+		}
+		unset( $line );
+
+		$text = implode( "\n", $lines );
+
+		return $text;
+	}
+}

src/Converter/PostProcessors/Table/Rowspan.php

@@ -0,0 +1,149 @@
+<?php
+
+namespace HalloWelt\MigrateDokuwiki\Converter\PostProcessors\Table;
+
+use HalloWelt\MigrateDokuwiki\IProcessor;
+
+class Rowspan implements IProcessor {
+
+	/**
+	 * @param string $text
+	 * @return string
+	 */
+	public function process( string $text ): string {
+		$lines = explode( "\n", $text );
+
+		$isTable = false;
+
+		$rowIndex = 0;
+		$colIndex = 0;
+
+		$linesToDelete = [];
+		$rowspanCounts = [];
+
+		// We'll use this mapping later to change only lines which we need,
+		// Which contain "rowspan" cells
+		// Without need to process all table cells once more
+		$cellCoordsToLineMap = [];
+
+		foreach ( $lines as $lineIndex => &$line ) {
+			$line = trim( $line );
+
+			if ( strpos( $line, '{|' ) === 0 ) {
+				// If we found "{|" - then we start processing the table
+				$isTable = true;
+
+				$rowIndex = 0;
+				$colIndex = 0;
+
+				$rowspanCounts = [];
+
+				// We use separate "cell -> line" mapping for each table we process
+				$cellCoordsToLineMap = [];
+
+				continue;
+			}
+
+			if ( strpos( $line, '|}' ) === 0 ) {
+				$isTable = false;
+
+				// If we got finished with current table - set proper "rowspan" attributes
+				// for specific cells in this table
+				foreach ( $rowspanCounts as $rowIndex => $rowspanCells ) {
+					foreach ( $rowspanCells as $colIndex => $rowspanCount ) {
+						$tableCellLineIndex = $cellCoordsToLineMap[$rowIndex][$colIndex];
+
+						// We found necessary line in the wikitext
+						// Now parse this table line and add necessary "rowspan" attribute value
+						$rowspanLine = $lines[$tableCellLineIndex];
+
+						$cellBlocks = explode( '|', $rowspanLine );
+
+						// If cell already contains block with HTML attributes (like "style" or "colspan")
+						// Then just append "colspan" there
+						if ( count( $cellBlocks ) > 2 ) {
+							$cellBlocks[1] = $cellBlocks[1] . " rowspan=\"$rowspanCount\"";
+						} else {
+							// Otherwise add such block
+							$cellBlocks = array_merge(
+								[
+									$cellBlocks[0]
+								],
+								[
+									"rowspan=\"$rowspanCount\""
+								],
+								array_slice( $cellBlocks, 1 )
+							);
+						}
+
+						$lines[$tableCellLineIndex] = implode( '|', $cellBlocks );
+					}
+				}
+
+				continue;
+			}
+
+			if ( $isTable ) {
+				// Row separator
+				if ( strpos( $line, '|-' ) === 0 ) {
+					$rowIndex++;
+					$colIndex = 0;
+
+					continue;
+				}
+
+				$colIndex++;
+
+				// It makes sense to record lines indexes only for "cell content lines"
+				$cellCoordsToLineMap[$rowIndex][$colIndex] = $lineIndex;
+
+				// Check if ":::" is the only content of current cell
+				$cellBlocks = explode( '|', $line );
+				foreach ( $cellBlocks as $cellBlock ) {
+					$cellBlockTrimmed = trim( $cellBlock );
+
+					if ( $cellBlockTrimmed === ':::' ) {
+						$linesToDelete[] = $lineIndex;
+
+						// Mark that cell for removal
+						$rowspanCounts[$rowIndex][$colIndex] = -1;
+
+						$rowspanRow = $rowIndex - 1;
+
+						// There can be any amount of cells united in one vertically using "rowspan"
+						// And in that case in DokuWiki there will be corresponding amount
+						// of vertical aligned cells filled with ":::"
+						while ( true ) {
+							// If that cell above also contains ":::" - go one more row above
+							if (
+								isset( $rowspanCounts[$rowspanRow][$colIndex] ) &&
+								$rowspanCounts[$rowspanRow][$colIndex] === -1
+							) {
+								$rowspanRow--;
+							} else {
+								break;
+							}
+						}
+
+						if ( !isset( $rowspanCounts[$rowspanRow][$colIndex] ) ) {
+							// Minimum amount for "rowspan" attribute is 2
+							$rowspanCounts[$rowspanRow][$colIndex] = 2;
+						} else {
+							$rowspanCounts[$rowspanRow][$colIndex]++;
+						}
+					}
+				}
+			}
+		}
+		unset( $line );
+
+		// Remove all cells containing only ":::"
+		foreach ( $linesToDelete as $lineToDelete ) {
+			unset( $lines[$lineToDelete] );
+		}
+
+		$text = implode( "\n", $lines );
+
+		return $text;
+	}
+}

src/Converter/PreProcessors/Table/Colspan.php

@@ -0,0 +1,46 @@
+<?php
+
+namespace HalloWelt\MigrateDokuwiki\Converter\PreProcessors\Table;
+
+use HalloWelt\MigrateDokuwiki\IProcessor;
+
+class Colspan implements IProcessor {
+
+	/**
+	 * @param string $text
+	 * @return string
+	 */
+	public function process( string $text ): string {
+		$lines = explode( "\n", $text );
+
+		foreach ( $lines as $index => &$line ) {
+			$line = trim( $line );
+
+			// Each table has either "|" or "^" at the line start
+			if (
+				strpos( $line, "|" ) === 0 ||
+				strpos( $line, "^" ) === 0
+			) {
+				$isTable = true;
+			} else {
+				$isTable = false;
+			}
+
+			if ( !$isTable ) {
+				continue;
+			}
+
+			$regex = '/(.*?)(\|\|+)/';
+			$line = preg_replace_callback( $regex, static function( $matches ) {
+				$colspanCount = strlen( $matches[2] );
+
+				return $matches[1] . '###COLSPAN_' . $colspanCount . '###|';
+			}, $line );
+		}
+		unset( $line );
+
+		$text = implode( "\n", $lines );
+
+		return $text;
+	}
+}

tests/phpunit/Converter/PostProcessor/Table/ColspanTest.php

@@ -0,0 +1,22 @@
+<?php
+
+namespace HalloWelt\MigrateDokuwiki\Tests\Converter\PostProcessor\Table;
+
+use HalloWelt\MigrateDokuwiki\Converter\PostProcessors\Table\Colspan;
+use PHPUnit\Framework\TestCase;
+
+class ColspanTest extends TestCase {
+
+	/**
+	 * @covers \HalloWelt\MigrateDokuwiki\Converter\PostProcessors\Table\Colspan::process()
+	 */
+	public function testProcess() {
+		$processor = new Colspan();
+
+		$dataDir = dirname( __DIR__, 3 ) . '/data/Converter/PostProcessor/Table';
+		$input = file_get_contents( "$dataDir/colspan-input.txt" );
+		$expected = file_get_contents( "$dataDir/colspan-output.txt" );
+		$actual = $processor->process( $input );
+		$this->assertEquals( $expected, $actual );
+	}
+}

tests/phpunit/Converter/PostProcessor/Table/RowspanTest.php

@@ -0,0 +1,22 @@
+<?php
+
+namespace HalloWelt\MigrateDokuwiki\Tests\Converter\PostProcessor\Table;
+
+use HalloWelt\MigrateDokuwiki\Converter\PostProcessors\Table\Rowspan;
+use PHPUnit\Framework\TestCase;
+
+class RowspanTest extends TestCase {
+
+	/**
+	 * @covers \HalloWelt\MigrateDokuwiki\Converter\PostProcessors\Table\Rowspan::process()
+	 */
+	public function testProcess() {
+		$processor = new Rowspan();
+
+		$dataDir = dirname( __DIR__, 3 ) . '/data/Converter/PostProcessor/Table';
+		$input = file_get_contents( "$dataDir/rowspan-input.txt" );
+		$expected = file_get_contents( "$dataDir/rowspan-output.txt" );
+		$actual = $processor->process( $input );
+		$this->assertEquals( $expected, $actual );
+	}
+}

tests/phpunit/Converter/PreProcessor/Table/ColspanTest.php

@@ -0,0 +1,22 @@
+<?php
+
+namespace HalloWelt\MigrateDokuwiki\Tests\Converter\PreProcessor\Table;
+
+use HalloWelt\MigrateDokuwiki\Converter\PreProcessors\Table\Colspan;
+use PHPUnit\Framework\TestCase;
+
+class ColspanTest extends TestCase {
+
+	/**
+	 * @covers \HalloWelt\MigrateDokuwiki\Converter\PreProcessors\Table\Colspan::process()
+	 */
+	public function testProcess() {
+		$processor = new Colspan();
+
+		$dataDir = dirname( __DIR__, 3 ) . '/data/Converter/PreProcessor/Table';
+		$input = file_get_contents( "$dataDir/colspan-input.txt" );
+		$expected = file_get_contents( "$dataDir/colspan-output.txt" );
+		$actual = $processor->process( $input );
+		$this->assertEquals( $expected, $actual );
+	}
+}