MWCSVImportMergeClass.php

<?php
/**
 * This is NOT WORKING CODE but is provided for reference.  This is the portion of
 * code that was used on mwcsvimport.pronique.com to take a csv dataset and template
 * file and generate a mediawiki xml import file.
 * 
 * mwcsvimport.pronique.com was shutdown on 12/15/2018 but this code
 * is provided for inspection by anyone looking to recreate it's
 * functionality.
 * 
 * - joncutrer
 * 
 */

class MWCSVImportMergeClass {

    
    function doMerge( $id ) {
        if ( !$dataset = $this->Dataset->read(null, $id) ) {
            $this->Session->setFlash(__('Invalid dataset', true));
            $this->redirect(array('action' => 'index'));   
        } 

        //Record Activity
        $this->recordActivity( 'XML Merge on \'' . $dataset['Dataset']['name'] . '\' ' . $dataset['Dataset']['id'] . ' ' ); 
        
        if ( !$merge_file['data'] = $this->__mwmerge( $dataset['Dataset']['fields'], $dataset['Dataset']['data'], $dataset['Template'] ) ) {
            $this->log( 'id:' . $id . ' Error During Merge, No XML Generated', 'debug' );
        }
        
        $merge_file['dataset_id'] = $dataset['Dataset']['id'];
        $merge_file['name'] = $dataset['Dataset']['name'] . '.xml';
        $merge_file['size'] = strlen( $merge_file['data'] );
        $merge_file['user_id'] = $this->Auth->user('id');
        if ( !$this->Dataset->MergeFile->save( $merge_file ) ) {
            $this->log( 'id: ' . $id . ' Error Saving Merge File', 'error' );
            //echo "Error Saving Merge File";
        }
        $this->redirect(array('controller'=>'datasets','action'=>'view', $id ));
        echo "XML File Built";
        exit; 
    }

    
    /****************************************************************************************
    *  Private Methods below
    * **************************************************************************************/    
    
    /**
    * Parse CSV Data call and build page from template
    * Return the XML created by calling buildXML();
    * 
    * @param mixed $csvdata
    * @param mixed $template
    */
    function __mwmerge( $fields, $data, $template ) {
        $this->log( 'Merging ' .  count( $data) . ' pages', 'debug' );
        if ( $this->Auth->user('plan') == 'free' ) {
            $data = array_slice( $data, 0, Configure::read('App.Accounts.Free.ExportLimit') );
        }
        
        foreach ( $data as $key=>$row ) {
            //Old Method of Compiling Templates
            //$pages[$key] = $this->__buildPage( $fields, $row, $template );
            //New Methods of Compiling Templates
            $pages[$key] = $this->__compileTemplate( $fields, $row, $template );
        }
        return $this->__buildXML( $pages ); 
    }

    /**
    * Replace ${Variables} with $data in $template
    * Returns a Mediawiki markup page array
    * $array['title']
    * $array['body']
    * 
    */
    function __buildPage( $fields, $data, $template ) {
        $ntitle = $template['title'];
        $nbody = $template['body'];
        foreach ( $fields as $field ) {
            //Depreciated
            $ntitle = str_replace( '${' . $field . '}', trim($data[$field]), $ntitle);
            $ntitle = str_replace( '{$' . $field . '}', trim($data[$field]), $ntitle);
            //Depreciated
            $nbody = str_replace( '${' . $field . '}', trim($data[$field]), $nbody);
            $nbody = str_replace( '{$' . $field . '}', trim($data[$field]), $nbody);
        }
        $page['title'] = trim( $this->__sanatizeTitle( $ntitle ) ); $page['body'] = $nbody;
        return $page;
    }

    /**
    * Replace ${Variables} with $data in $template
    * Returns an array like
    * $array['title']
    * $array['body']
    * 
    */
    function __compileTemplate( $fields, $data, $template ) {
        //Mapping of modifiers to functions
        $modifier_map = array(
            //String modifiers
            'upper'=>'strtoupper',
            'lower'=>'strtolower',
            'upperfirst'=>'ucfirst',
            'camelize'=>array('Inflector', 'camelize'), 
            'capitalize'=>'ucwords',
            'sha1'=>'sha1',
            'md5'=>'md5',
            'trim'=>'trim',
            'rtrim'=>'rtrim',
            'ltrim'=>'ltrim',
            'reverse'=>'strrev',
            'strip_tags'=>'strip_tags',
            //Number modifiers
            'number_format'=>array( $this, '__numberFormatModifier' ),
            //'money_format'=>array( $this, '__moneyFormatModifier'),
            'uuid'=>array('String', 'uuid'),
            'ordinal'=>array( $this, '__ordinalModifier'),
            //Date modifiers
            'date'=>array( $this, '__dateModifier'),
            'time'=>'time',
            
        );
        
        $ntitle = $template['title'];
        $nbody = $template['body'];
        
        //Replace Variables
        foreach ( $fields as $field ) {
            //TODO Explore replacing with CakePHP's String::insert() method
            $ntitle = str_replace( '{$' . $field . '}', trim($data[$field]), $ntitle);
            $nbody = str_replace( '{$' . $field . '}', trim($data[$field]), $nbody);
        }
        
        //Process modifiers in body 
        preg_match_all('/\{(\w+)(.*?)\}(.*?)\{\/\\1\}/si', $nbody, $matches );

        foreach( $matches[1] as $key=>$func ) {
            if ( array_key_exists( $func, $modifier_map ) ) {
                //echo 'Debug: Calling ' . $modifier_map[$func] . ' on ' . $matches[3][$key] . "<br />";
                $args[] = $matches[3][$key];
                if( !empty( $matches[2][$key] ) ) { $args[] = $matches[2][$key]; }
                $imodified = call_user_func_array( $modifier_map[$func], $args );
                $nbody = str_replace( $matches[0][$key], $imodified, $nbody );        
                $imodified = '';
                unset( $args );
            }
        }
        
        //Sanatize Title
        $page['title'] = trim( $this->__sanatizeTitle( $ntitle ) ); $page['body'] = $nbody;
        return $page;
    }
    
  
    /**
    * Builds Mediawiki XML Import file from $pages array
    * 
    */
    function __buildXML( $pages ) {
        $xml = '<mediawiki xmlns="http://www.mediawiki.org/xml/export-0.4/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.mediawiki.org/xml/export-0.4/ http://www.mediawiki.org/xml/export-0.4.xsd" version="0.4" xml:lang="en">';
        foreach ( $pages as $page ) {
            $xml .= '<page>
    <title>' . $page['title'] . '</title>
    <revision>';
            $xml .= '<timestamp>'.date( 'Y-m-d\TH:i:s\Z' ,time() ).'</timestamp>';
            $xml .= '<text xml:space="preserve">' . htmlspecialchars($page['body']) . '</text>
    </revision>
  </page>
';
        }
        $xml .= '</mediawiki>';
        return $xml;
    }
    
    function __sanatizeTitle( $title ) {
        $newtitle = str_replace( Configure::read('App.ForbiddenTitleCharacters'), '' , $title );
        return $newtitle;    
    }
    
    
    /**
    * Called by __compileTemplate()
    * 
    * @param int $time
    * @param mixed $args
    * @return string
    */
    function __dateModifier( $time, $args) {
        $argKeys = $this->__extractModifierArguments( $args );

        if ( preg_match('/^[0-9]{10}$/', $time, $matches) == 0 ) {
            $time =  strtotime($time);            
        } 
        
        $date_formatted = date( $argKeys['format'], $time );
        return $date_formatted;
    } 

    function __numberFormatModifier( $number, $args) {
        $argKeys = $this->__extractModifierArguments( $args );
        if ( isset( $argKeys['decimals'] ) ) { $decimals = $argKeys['decimals']; } else { $decimals = 0; }
        if ( isset( $argKeys['dec_point'] ) ) { $dec_point = $argKeys['dec_point']; } else { $dec_point = '.'; }
        if ( isset( $argKeys['thousands_sep'] ) ) { $thousands_sep = $argKeys['thousands_sep']; } else { $thousands_sep = ','; }
        return number_format( $number, $decimals, $dec_point, $thousands_sep );    
    }
    
    /**
    * Called by __compileTemplate()
    * Add the suffix onto numbers, ie 2nd 3rd 15th
    * @param int $time
    * @param mixed $args
    * @return string
    */    
    function __ordinalModifier($cdnl){ 
        $test_c = abs($cdnl) % 10; 
        $ext = ((abs($cdnl) %100 < 21 && abs($cdnl) %100 > 4) ? 'th' 
                : (($test_c < 4) ? ($test_c < 3) ? ($test_c < 2) ? ($test_c < 1) 
                ? 'th' : 'st' : 'nd' : 'rd' : 'th')); 
        return $cdnl.$ext; 
    } 
    
    /**
    * Called by Modifier functions to extract the list parameters
    * ie format="M/d/Y"
    * returns array( 'format'=>'M/d/Y' ) 
    * ie pattern="/foo/i" match="/$1/i"
    * returns array( 'pattern'=>'/foo/i', 'match'=>'/$1/i' )
    * 
    */
    function __extractModifierArguments( $args ) {
        $argsArr = explode(' ', trim($args) );
        foreach( $argsArr as  $arg ) {
            $tmp = explode( '=', $arg );
            $argKeys[$tmp[0]] = str_replace('"', '', $tmp[1]);
        }
        return $argKeys;
    }
}
?>