ODText Reader: Basic ODText reader

CHANGELOG.md

@@ -38,6 +38,7 @@ This release marked heavy refactorings on internal code structure with the creat
 - DOCX Writer: Create `word/settings.xml` and `word/webSettings.xml` dynamically - @ivanlanin
 - ODT Writer: Basic image writing - @ivanlanin
 - ODT Writer: Link writing - @ivanlanin
+- ODT Reader: Basic ODText Reader - @ivanlanin
 
 ### Bugfixes
 

samples/Sample_24_ReadODText.php

@@ -0,0 +1,14 @@
+<?php
+include_once 'Sample_Header.php';
+
+// Read contents
+$name = basename(__FILE__, '.php');
+$source = "resources/{$name}.odt";
+echo date('H:i:s'), " Reading contents from `{$source}`", EOL;
+$phpWord = \PhpOffice\PhpWord\IOFactory::load($source, 'ODText');
+
+// Save file
+echo write($phpWord, basename(__FILE__, '.php'), $writers);
+if (!CLI) {
+    include_once 'Sample_Footer.php';
+}

src/PhpWord/IOFactory.php

@@ -43,7 +43,7 @@ abstract class IOFactory
      */
     public static function createReader($name = 'Word2007')
     {
-        if (!in_array($name, array('ReaderInterface', 'Word2007'))) {
+        if (!in_array($name, array('ReaderInterface', 'Word2007', 'ODText'))) {
             throw new Exception("\"{$name}\" is not a valid reader.");
         }
 

src/PhpWord/Reader/ODText.php

@@ -0,0 +1,86 @@
+<?php
+/**
+ * PHPWord
+ *
+ * @link        https://github.com/PHPOffice/PHPWord
+ * @copyright   2014 PHPWord
+ * @license     http://www.gnu.org/licenses/old-licenses/lgpl-2.1.txt LGPL
+ */
+
+namespace PhpOffice\PhpWord\Reader;
+
+use PhpOffice\PhpWord\PhpWord;
+use PhpOffice\PhpWord\Settings;
+use PhpOffice\PhpWord\Shared\XMLReader;
+
+/**
+ * Reader for ODText
+ *
+ * @since 0.10.0
+ */
+class ODText extends AbstractReader implements ReaderInterface
+{
+    /**
+     * Loads PhpWord from file
+     *
+     * @param string $docFile
+     * @return \PhpOffice\PhpWord\PhpWord
+     */
+    public function load($docFile)
+    {
+        $phpWord = new PhpWord();
+        $relationships = $this->readRelationships($docFile);
+
+        $readerParts = array(
+            'content.xml' => 'Content',
+        );
+
+        foreach ($readerParts as $xmlFile => $partName) {
+            $this->readPart($phpWord, $relationships, $partName, $docFile, $xmlFile);
+        }
+
+        return $phpWord;
+    }
+
+    /**
+     * Read document part
+     *
+     * @param \PhpOffice\PhpWord\PhpWord $phpWord
+     * @param array $relationships
+     * @param string $partName
+     * @param string $docFile
+     * @param string $xmlFile
+     */
+    private function readPart(PhpWord &$phpWord, $relationships, $partName, $docFile, $xmlFile)
+    {
+        $partClass = "PhpOffice\\PhpWord\\Reader\\ODText\\{$partName}";
+        if (class_exists($partClass)) {
+            $part = new $partClass($docFile, $xmlFile);
+            $part->setRels($relationships);
+            $part->read($phpWord);
+        }
+
+    }
+
+    /**
+     * Read all relationship files
+     *
+     * @param string $docFile
+     * @return array
+     */
+    private function readRelationships($docFile)
+    {
+        $rels = array();
+        $xmlFile = 'META-INF/manifest.xml';
+        $xmlReader = new XMLReader();
+        $xmlReader->getDomFromZip($docFile, $xmlFile);
+        $nodes = $xmlReader->getElements('manifest:file-entry');
+        foreach ($nodes as $node) {
+            $type = $xmlReader->getAttribute('manifest:media-type', $node);
+            $target = $xmlReader->getAttribute('manifest:full-path', $node);
+            $rels[] = array('type' => $type, 'target' => $target);
+        }
+
+        return $rels;
+    }
+}

src/PhpWord/Reader/ODText/AbstractPart.php

@@ -0,0 +1,67 @@
+<?php
+/**
+ * PHPWord
+ *
+ * @link        https://github.com/PHPOffice/PHPWord
+ * @copyright   2014 PHPWord
+ * @license     http://www.gnu.org/licenses/old-licenses/lgpl-2.1.txt LGPL
+ */
+
+namespace PhpOffice\PhpWord\Reader\ODText;
+
+use PhpOffice\PhpWord\PhpWord;
+use PhpOffice\PhpWord\Shared\XMLReader;
+
+/**
+ * Abstract part reader
+ */
+abstract class AbstractPart
+{
+    /**
+     * Document file
+     *
+     * @var string
+     */
+    protected $docFile;
+
+    /**
+     * XML file
+     *
+     * @var string
+     */
+    protected $xmlFile;
+
+    /**
+     * Part relationships
+     *
+     * @var array
+     */
+    protected $rels = array();
+
+    /**
+     * Read part
+     */
+    abstract public function read(PhpWord &$phpWord);
+
+    /**
+     * Create new instance
+     *
+     * @param string $docFile
+     * @param string $xmlFile
+     */
+    public function __construct($docFile, $xmlFile)
+    {
+        $this->docFile = $docFile;
+        $this->xmlFile = $xmlFile;
+    }
+
+    /**
+     * Set relationships
+     *
+     * @param array $value
+     */
+    public function setRels($value)
+    {
+        $this->rels = $value;
+    }
+}

src/PhpWord/Reader/ODText/Content.php

@@ -0,0 +1,57 @@
+<?php
+/**
+ * PHPWord
+ *
+ * @link        https://github.com/PHPOffice/PHPWord
+ * @copyright   2014 PHPWord
+ * @license     http://www.gnu.org/licenses/old-licenses/lgpl-2.1.txt LGPL
+ */
+
+namespace PhpOffice\PhpWord\Reader\ODText;
+
+use PhpOffice\PhpWord\PhpWord;
+use PhpOffice\PhpWord\Shared\XMLReader;
+
+/**
+ * Content reader
+ */
+class Content extends AbstractPart
+{
+    /**
+     * Read content.xml
+     *
+     * @param \PhpOffice\PhpWord\PhpWord $phpWord
+     */
+    public function read(PhpWord &$phpWord)
+    {
+        $xmlReader = new XMLReader();
+        $xmlReader->getDomFromZip($this->docFile, $this->xmlFile);
+
+        $nodes = $xmlReader->getElements('office:body/office:text/*');
+        if ($nodes->length > 0) {
+            $section = $phpWord->addSection();
+            foreach ($nodes as $node) {
+                // $styleName = $xmlReader->getAttribute('text:style-name', $node);
+                switch ($node->nodeName) {
+
+                    case 'text:h': // Heading
+                        $depth = $xmlReader->getAttribute('text:outline-level', $node);
+                        $section->addTitle($node->nodeValue, $depth);
+                        break;
+
+                    case 'text:p': // Paragraph
+                        $section->addText($node->nodeValue);
+                        break;
+
+                    case 'text:list': // List
+                        $listItems = $xmlReader->getElements('text:list-item/text:p', $node);
+                        foreach ($listItems as $listItem) {
+                            // $listStyleName = $xmlReader->getAttribute('text:style-name', $listItem);
+                            $section->addListItem($listItem->nodeValue);
+                        }
+                        break;
+                }
+            }
+        }
+    }
+}

src/PhpWord/Style/Font.php

@@ -203,7 +203,7 @@ class Font extends AbstractStyle
     /**
      * Get font name
      *
-     * @return bool
+     * @return string
      */
     public function getName()
     {

tests/PhpWord/Tests/Reader/ODTextTest.php

@@ -0,0 +1,31 @@
+<?php
+/**
+ * PHPWord
+ *
+ * @link        https://github.com/PHPOffice/PHPWord
+ * @copyright   2014 PHPWord
+ * @license     http://www.gnu.org/licenses/old-licenses/lgpl-2.1.txt LGPL
+ */
+
+namespace PhpOffice\PhpWord\Tests\Reader;
+
+use PhpOffice\PhpWord\IOFactory;
+
+/**
+ * Test class for PhpOffice\PhpWord\Reader\ODText
+ *
+ * @coversDefaultClass \PhpOffice\PhpWord\Reader\ODText
+ * @runTestsInSeparateProcesses
+ */
+class ODTextTest extends \PHPUnit_Framework_TestCase
+{
+    /**
+     * Load
+     */
+    public function testLoad()
+    {
+        $filename = __DIR__ . '/../_files/documents/reader.odt';
+        $object = IOFactory::load($filename, 'ODText');
+        $this->assertInstanceOf('PhpOffice\\PhpWord\\PhpWord', $object);
+    }
+}