From 87f071d1b63183ffaf8801840386c65b4b9978e5 Mon Sep 17 00:00:00 2001
From: Ivan Lanin <ivanlanin@gmail.com>
Date: Tue, 27 May 2014 23:41:15 +0700
Subject: [PATCH] #72: Basic RTF Reader

---
 samples/Sample_27_ReadRTF.php          |  17 ++
 samples/resources/rtf.rtf              |  21 ++
 src/PhpWord/IOFactory.php              |   2 +-
 src/PhpWord/Reader/AbstractReader.php  |   2 +-
 src/PhpWord/Reader/RTF.php             |  50 ++++
 src/PhpWord/Reader/RTF/Document.php    | 351 +++++++++++++++++++++++++
 src/PhpWord/Shared/XMLReader.php       |  22 +-
 src/PhpWord/Writer/RTF/Part/Header.php |   2 +-
 8 files changed, 459 insertions(+), 8 deletions(-)
 create mode 100644 samples/Sample_27_ReadRTF.php
 create mode 100644 samples/resources/rtf.rtf
 create mode 100644 src/PhpWord/Reader/RTF.php
 create mode 100644 src/PhpWord/Reader/RTF/Document.php

diff --git a/samples/Sample_27_ReadRTF.php b/samples/Sample_27_ReadRTF.php
new file mode 100644
index 00000000..a28ba40b
--- /dev/null
+++ b/samples/Sample_27_ReadRTF.php
@@ -0,0 +1,17 @@
+<?php
+include_once 'Sample_Header.php';
+
+// Read contents
+$name = basename(__FILE__, '.php');
+$source = "results/Sample_01_SimpleText.rtf";
+$source = "resources/rtf.rtf";
+$source = "results/Sample_11_ReadWord2007.rtf";
+
+echo date('H:i:s'), " Reading contents from `{$source}`", EOL;
+$phpWord = \PhpOffice\PhpWord\IOFactory::load($source, 'RTF');
+
+// Save file
+echo write($phpWord, basename(__FILE__, '.php'), $writers);
+if (!CLI) {
+    include_once 'Sample_Footer.php';
+}
diff --git a/samples/resources/rtf.rtf b/samples/resources/rtf.rtf
new file mode 100644
index 00000000..6f9ac0f8
--- /dev/null
+++ b/samples/resources/rtf.rtf
@@ -0,0 +1,21 @@
+{\rtf1
+\ansi\ansicpg1252
+\deff0
+{\fonttbl{\f0\fnil\fcharset0 Arial;}{\f1\fnil\fcharset0 Times New Roman;}}
+{\colortbl;\red255\green0\blue0;\red14\green0\blue0}
+{\*\generator PhpWord;}
+
+{\info{\title }{\subject }{\category }{\keywords }{\comment }{\author }{\operator }{\creatim \yr2014\mo05\dy27\hr23\min36\sec45}{\revtim \yr2014\mo05\dy27\hr23\min36\sec45}{\company }{\manager }}
+\deftab720\viewkind1\uc1\pard\nowidctlpar\lang1036\kerning1\fs20
+{Welcome to PhpWord}\par
+\pard\nowidctlpar{\cf0\f0 Hello World!}\par
+\par
+\par
+\pard\nowidctlpar{\cf0\f0\fs32\b\i I am styled by a font style definition.}\par
+\pard\nowidctlpar{\cf0\f0 I am styled by a paragraph style definition.}\par
+\pard\nowidctlpar\qc\sa100{\cf0\f0\fs32\b\i I am styled by both font and paragraph style.}\par
+\pard\nowidctlpar{\cf1\f1\fs40\b\i\ul\strike\super I am inline styled.}\par
+\par
+{\field {\*\fldinst {HYPERLINK "http://www.google.com"}}{\fldrslt {Google}}}\par
+\par
+}
\ No newline at end of file
diff --git a/src/PhpWord/IOFactory.php b/src/PhpWord/IOFactory.php
index 83076fcc..0d5fe689 100644
--- a/src/PhpWord/IOFactory.php
+++ b/src/PhpWord/IOFactory.php
@@ -51,7 +51,7 @@ abstract class IOFactory
      */
     public static function createReader($name = 'Word2007')
     {
-        if (!in_array($name, array('ReaderInterface', 'Word2007', 'ODText'))) {
+        if (!in_array($name, array('ReaderInterface', 'Word2007', 'ODText', 'RTF'))) {
             throw new Exception("\"{$name}\" is not a valid reader.");
         }
 
diff --git a/src/PhpWord/Reader/AbstractReader.php b/src/PhpWord/Reader/AbstractReader.php
index 3cc2d490..c08914e5 100644
--- a/src/PhpWord/Reader/AbstractReader.php
+++ b/src/PhpWord/Reader/AbstractReader.php
@@ -39,7 +39,7 @@ abstract class AbstractReader implements ReaderInterface
      *
      * @var bool|resource
      */
-    protected $fileHandle = true;
+    protected $fileHandle;
 
     /**
      * Read data only?
diff --git a/src/PhpWord/Reader/RTF.php b/src/PhpWord/Reader/RTF.php
new file mode 100644
index 00000000..1856116c
--- /dev/null
+++ b/src/PhpWord/Reader/RTF.php
@@ -0,0 +1,50 @@
+<?php
+/**
+ * This file is part of PHPWord - A pure PHP library for reading and writing
+ * word processing documents.
+ *
+ * PHPWord is free software distributed under the terms of the GNU Lesser
+ * General Public License version 3 as published by the Free Software Foundation.
+ *
+ * For the full copyright and license information, please read the LICENSE
+ * file that was distributed with this source code. For the full list of
+ * contributors, visit https://github.com/PHPOffice/PHPWord/contributors.
+ *
+ * @link        https://github.com/PHPOffice/PHPWord
+ * @copyright   2010-2014 PHPWord contributors
+ * @license     http://www.gnu.org/licenses/lgpl.txt LGPL version 3
+ */
+
+namespace PhpOffice\PhpWord\Reader;
+
+use PhpOffice\PhpWord\PhpWord;
+use PhpOffice\PhpWord\Reader\RTF\Document;
+
+/**
+ * RTF Reader class
+ *
+ * @since 0.11.0
+ */
+class RTF extends AbstractReader implements ReaderInterface
+{
+    /**
+     * Loads PhpWord from file
+     *
+     * @param string $docFile
+     * @return \PhpOffice\PhpWord\PhpWord
+     */
+    public function load($docFile)
+    {
+        $phpWord = new PhpWord();
+
+        if ($this->canRead($docFile)) {
+            $doc = new Document();
+            $doc->rtf = file_get_contents($docFile);
+            $doc->read($phpWord);
+        } else {
+            throw new \Exception("Cannot read {$docFile}.");
+        }
+
+        return $phpWord;
+    }
+}
diff --git a/src/PhpWord/Reader/RTF/Document.php b/src/PhpWord/Reader/RTF/Document.php
new file mode 100644
index 00000000..ccb9b595
--- /dev/null
+++ b/src/PhpWord/Reader/RTF/Document.php
@@ -0,0 +1,351 @@
+<?php
+/**
+ * This file is part of PHPWord - A pure PHP library for reading and writing
+ * word processing documents.
+ *
+ * PHPWord is free software distributed under the terms of the GNU Lesser
+ * General Public License version 3 as published by the Free Software Foundation.
+ *
+ * For the full copyright and license information, please read the LICENSE
+ * file that was distributed with this source code. For the full list of
+ * contributors, visit https://github.com/PHPOffice/PHPWord/contributors.
+ *
+ * @link        https://github.com/PHPOffice/PHPWord
+ * @copyright   2010-2014 PHPWord contributors
+ * @license     http://www.gnu.org/licenses/lgpl.txt LGPL version 3
+ */
+
+namespace PhpOffice\PhpWord\Reader\RTF;
+
+use PhpOffice\PhpWord\PhpWord;
+use PhpOffice\PhpWord\Element\Section;
+use PhpOffice\PhpWord\Element\TextRun;
+
+/**
+ * RTF document reader
+ *
+ * References:
+ * - How to Write an RTF Reader http://latex2rtf.sourceforge.net/rtfspec_45.html
+ * - PHP rtfclass by Markus Fischer https://github.com/mfn/rtfclass
+ * - JavaScript RTF-parser by LazyGyu https://github.com/lazygyu/RTF-parser
+ *
+ * @since 0.11.0
+ * @SuppressWarnings(PHPMD.UnusedPrivateMethod)
+ */
+class Document
+{
+    /** @const int */
+    const PARA = 0;
+    const STYL = 1;
+    const SKIP = 2;
+
+    /**
+     * PhpWord object
+     *
+     * @var \PhpOffice\PhpWord\PhpWord
+     */
+    private $phpWord;
+
+    /**
+     * Section object
+     *
+     * @var \PhpOffice\PhpWord\Element\Section
+     */
+    private $section;
+
+    /**
+     * Textrun object
+     *
+     * @var \PhpOffice\PhpWord\Element\TextRun
+     */
+    private $textrun;
+
+    /**
+     * RTF content
+     *
+     * @var string
+     */
+    public $rtf;
+
+    /**
+     * Content length
+     *
+     * @var int
+     */
+    private $length = 0;
+
+    /**
+     * Character index
+     *
+     * @var int
+     */
+    private $offset = 0;
+
+    /**
+     * Current control word
+     *
+     * @var string
+     */
+    private $control = '';
+
+    /**
+     * Text content
+     *
+     * @var string
+     */
+    private $text = '';
+
+    /**
+     * Parsing a control word flag
+     *
+     * @var bool
+     */
+    private $isControl = false;
+
+    /**
+     * First character flag: watch out for control symbols
+     *
+     * @var bool
+     */
+    private $isFirst = false;
+
+    /**
+     * Group groups
+     *
+     * @var array
+     */
+    private $groups = array();
+
+    /**
+     * Parser flags; not used
+     *
+     * @var array
+     */
+    private $flags = array();
+
+    /**
+     * Parse RTF content
+     *
+     * - Marks controlling characters `{`, `}`, and `\`
+     * - Removes line endings
+     * - Builds control words and control symbols
+     * - Pushes every other character into the text queue
+     *
+     * @param \PhpOffice\PhpWord\PhpWord $phpWord
+     * @todo Use `fread` stream for scalability
+     */
+    public function read(PhpWord &$phpWord)
+    {
+        $markers = array(
+            123 => 'markOpening',   // {
+            125 => 'markClosing',   // }
+            92  => 'markBackslash', // \
+            10  => 'markNewline',   // LF
+            13  => 'markNewline'    // CR
+        );
+
+        $this->phpWord = $phpWord;
+        $this->section = $phpWord->addSection();
+        $this->textrun = $this->section->addTextRun();
+        $this->length = strlen($this->rtf);
+
+        $this->flags['paragraph'] = true; // Set paragraph flag from the beginning
+
+        // Walk each characters
+        while ($this->offset < $this->length) {
+            $char  = $this->rtf[$this->offset];
+            $ascii = ord($char);
+
+            if (array_key_exists($ascii, $markers)) { // Marker found: {, }, \, LF, or CR
+                $markerFunction = $markers[$ascii];
+                $this->$markerFunction();
+            } else {
+                if ($this->isControl === false) { // Non control word: Push character
+                    $this->pushText($char);
+                } else {
+                    if (preg_match("/^[a-zA-Z0-9-]?$/", $char)) { // No delimiter: Buffer control
+                        $this->control .= $char;
+                        $this->isFirst = false;
+                    } else { // Delimiter found: Parse buffered control
+                        if ($this->isFirst) {
+                            $this->isFirst = false;
+                        } else {
+                            if ($char == ' ') { // Discard space as a control word delimiter
+                                $this->flushControl(true);
+                            }
+                        }
+                    }
+                }
+            }
+            $this->offset++;
+        }
+        $this->flushText();
+    }
+
+    /**
+     * Mark opening braket `{` character
+     */
+    private function markOpening()
+    {
+        $this->flush(true);
+        array_push($this->groups, $this->flags);
+    }
+
+    /**
+     * Mark closing braket `}` character
+     */
+    private function markClosing()
+    {
+        $this->flush(true);
+        $this->flags = array_pop($this->groups);
+    }
+
+    /**
+     * Mark backslash `\` character
+     */
+    private function markBackslash()
+    {
+        if ($this->isFirst) {
+            $this->setControl(false);
+            $this->text .= '\\';
+        } else {
+            $this->flush();
+            $this->setControl(true);
+            $this->control = '';
+        }
+    }
+
+    /**
+     * Mark newline character: Flush control word because it's not possible to span multiline
+     */
+    private function markNewline()
+    {
+        if ($this->isControl) {
+            $this->flushControl(true);
+        }
+    }
+
+    /**
+     * Flush control word or text
+     *
+     * @param bool $isControl
+     */
+    private function flush($isControl = false)
+    {
+        if ($this->isControl) {
+            $this->flushControl($isControl);
+        } else {
+            $this->flushText();
+        }
+    }
+
+    /**
+     * Flush control word
+     *
+     * @param bool $isControl
+     */
+    private function flushControl($isControl = false)
+    {
+        if (preg_match("/^([A-Za-z]+)(-?[0-9]*) ?$/", $this->control, $match) === 1) {
+            list(, $control, $parameter) = $match;
+            $this->parseControl($control, $parameter);
+        }
+
+        if ($isControl === true) {
+            $this->setControl(false);
+        }
+    }
+
+    /*
+     * Flush text in queue
+     */
+    private function flushText()
+    {
+        if ($this->text != '') {
+            if (isset($this->flags['property'])) {
+                $this->flags['value'] = $this->text;
+                var_dump($this->flags);
+            } else {
+                if ($this->flags['paragraph'] === true) {
+                    $this->flags['paragraph'] = false;
+                    $this->flags['text'] = $this->text;
+                }
+            }
+            if (!isset($this->flags['skipped'])) {
+                $this->textrun->addText($this->text);
+            }
+
+            $this->text = '';
+        }
+    }
+
+    /**
+     * Reset control word and first char state
+     *
+     * @param bool $state
+     */
+    private function setControl($value)
+    {
+        $this->isControl = $value;
+        $this->isFirst = $value;
+    }
+
+    /**
+     * Push text into queue
+     *
+     * @param string $char
+     */
+    private function pushText($char)
+    {
+        if ($char == '<') {
+            $this->text .= "&lt;";
+        } elseif ($char == '>') {
+            $this->text .= "&gt;";
+        } else {
+            $this->text .= $char;
+        }
+    }
+
+    /**
+     * Parse control
+     *
+     * @param string $control
+     * @param string $parameter
+     */
+    private function parseControl($control, $parameter)
+    {
+        $controls = array(
+            'par'       => array(self::PARA, 'paragraph',    true),
+            'b'         => array(self::STYL, 'bold',         true),
+            'i'         => array(self::STYL, 'italic',       true),
+            'u'         => array(self::STYL, 'underline',    true),
+            'fonttbl'   => array(self::SKIP, 'fonttbl',      null),
+            'colortbl'  => array(self::SKIP, 'colortbl',     null),
+            'info'      => array(self::SKIP, 'info',         null),
+            'generator' => array(self::SKIP, 'generator',    null),
+            'title'     => array(self::SKIP, 'title',        null),
+            'subject'   => array(self::SKIP, 'subject',      null),
+            'category'  => array(self::SKIP, 'category',     null),
+            'keywords'  => array(self::SKIP, 'keywords',     null),
+            'comment'   => array(self::SKIP, 'comment',      null),
+            'shppict'   => array(self::SKIP, 'pic',          null),
+            'fldinst'   => array(self::SKIP, 'link',         null),
+        );
+
+        if (array_key_exists($control, $controls)) {
+            list($mode, $property, $value) = $controls[$control];
+            switch ($mode) {
+                case self::PARA: // Paragraph
+                    $this->textrun = $this->section->addTextRun();
+                    $this->flags[$property] = $value;
+                    break;
+                case self::STYL: // Style
+                    $this->flags[$property] = $value;
+                    break;
+                case self::SKIP: // Destination
+                    $this->flags['property'] = $property;
+                    $this->flags['skipped'] = true;
+            }
+        }
+    }
+}
diff --git a/src/PhpWord/Shared/XMLReader.php b/src/PhpWord/Shared/XMLReader.php
index 60474ce9..153152ee 100644
--- a/src/PhpWord/Shared/XMLReader.php
+++ b/src/PhpWord/Shared/XMLReader.php
@@ -56,18 +56,30 @@ class XMLReader
 
         $zip = new ZipArchive();
         $zip->open($zipFile);
-        $contents = $zip->getFromName($xmlFile);
+        $content = $zip->getFromName($xmlFile);
         $zip->close();
 
-        if ($contents === false) {
+        if ($content === false) {
             return false;
         } else {
-            $this->dom = new \DOMDocument();
-            $this->dom->loadXML($contents);
-            return $this->dom;
+            return $this->getDomFromString($content);
         }
     }
 
+    /**
+     * Get DOMDocument from content string
+     *
+     * @param string $content
+     * @return \DOMDocument
+     */
+    public function getDomFromString($content)
+    {
+        $this->dom = new \DOMDocument();
+        $this->dom->loadXML($content);
+
+        return $this->dom;
+    }
+
     /**
      * Get elements
      *
diff --git a/src/PhpWord/Writer/RTF/Part/Header.php b/src/PhpWord/Writer/RTF/Part/Header.php
index c401b500..15a0c303 100644
--- a/src/PhpWord/Writer/RTF/Part/Header.php
+++ b/src/PhpWord/Writer/RTF/Part/Header.php
@@ -123,7 +123,7 @@ class Header extends AbstractPart
         $content .= '{';
         $content .= '\fonttbl';
         foreach ($this->fontTable as $index => $font) {
-            $content .= "{\\f{$index}\\fnil\\fcharset0{$font};}";
+            $content .= "{\\f{$index}\\fnil\\fcharset0 {$font};}";
         }
         $content .= '}';
         $content .= PHP_EOL;