1 <?php
  2 /**
  3  * Class to handle parsing of CSV files, where the column headers are in the first row.
  4  * The idea is that you pass it another object to handle the actual procesing of the data in the CSV file.
  5  * 
  6  * Usage:
  7  * <code>
  8  * $parser = new CSVParser('myfile.csv');
  9  * $parser->mapColumns(
 10  *    'first name' => 'FirstName'
 11  *    'lastname' => 'Surname',
 12  *    'last name' => 'Surname'
 13  * ));
 14  * foreach($parser as $row) {
 15  *   // $row is a map of column name => column value
 16  *   $obj = new MyDataObject();
 17  *   $obj->update($row);
 18  *   $obj->write();
 19  * }
 20  * </code>
 21  * 
 22  * @package sapphire
 23  * @subpackage bulkloading
 24  */
 25 class CSVParser extends Object implements Iterator {
 26     protected $filename;
 27     protected $fileHandle;
 28     
 29     /**
 30      * Map of source columns to output columns
 31      * Once they get into this variable, all of the source columns are in lowercase
 32      */
 33     protected $columnMap = array();
 34     
 35     /**
 36      * The header row used to map data in the CSV file
 37      * To begin with, this is null.  Once it has been set, data will get returned from the CSV file
 38      */
 39     protected $headerRow = null;
 40     
 41     /**
 42      * A custom header row provided by the caller
 43      */
 44     protected $providedHeaderRow = null;
 45     
 46     /**
 47      * The data of the current row
 48      */
 49     protected $currentRow = null;
 50     
 51     /**
 52      * The current row number
 53      * 1 is the first data row in the CSV file; the header row, if it exists, is ignored
 54      */
 55     protected $rowNum = 0;
 56     
 57     /**
 58      * The character for separating columns
 59      */
 60     protected $delimiter = ",";
 61     
 62     /**
 63      * The character for quoting colums
 64      */
 65     protected $enclosure = '"';
 66     
 67     /**
 68      * Open a CSV file for parsing.
 69      * You can use the object returned in a foreach loop to extract the data
 70      * @param $filename The name of the file.  If relative, it will be relative to the site's base dir
 71      * @param $delimiter The character for seperating columns
 72      * @param $enclosure The character for quoting or enclosing columns
 73      */
 74     function __construct($filename, $delimiter = ",", $enclosure = '"') {
 75         $filename = Director::getAbsFile($filename);
 76         $this->filename = $filename;
 77         $this->delimiter = $delimiter;
 78         $this->enclosure = $enclosure;
 79         parent::__construct();
 80     }
 81     
 82     /**
 83      * Re-map columns in the CSV file.
 84      * This can be useful for identifying synonyms in the file
 85      * For example:
 86      * <code>
 87      * $csv->mapColumns(array(
 88      *   'firstname' => 'FirstName',
 89      *   'last name' => 'Surname',
 90      * ));
 91      * </code>
 92      */
 93     function mapColumns($columnMap) {
 94         if($columnMap) {
 95             $lowerColumnMap = array();
 96             foreach($columnMap as $k => $v) {
 97                 $lowerColumnMap[strtolower($k)] = $v;
 98             }
 99             $this->columnMap = array_merge($this->columnMap, $lowerColumnMap);
100         }
101     }
102     
103     /**
104      * If your CSV file doesn't have a header row, then you can call this function to provide one.
105      * If you call this function, then the first row of the CSV will be included in the data returned.
106      */
107     function provideHeaderRow($headerRow) {
108         $this->providedHeaderRow = $headerRow;
109     }
110 
111     /**
112      * Open the CSV file for reading
113      */
114     protected function openFile() {
115         ini_set('auto_detect_line_endings',1);
116         $this->fileHandle = fopen($this->filename,'r');
117 
118         
119         if($this->providedHeaderRow) {
120             $this->headerRow = $this->remapHeader($this->providedHeaderRow);
121         }
122     }
123 
124     /**
125      * Close the CSV file and re-set all of the internal variables
126      */
127     protected function closeFile() {
128         if($this->fileHandle) fclose($this->fileHandle);
129         $this->fileHandle = null;
130 
131         $this->rowNum = 0;
132         $this->currentRow = null;
133         $this->headerRow = null;
134     }
135     
136 
137     /**
138      * Get a header row from the CSV file
139      */
140     protected function fetchCSVHeader() {
141         $srcRow = fgetcsv($this->fileHandle, 0, $this->delimiter, $this->enclosure);
142         $this->headerRow = $this->remapHeader($srcRow);
143     }
144     
145     /**
146      * Map the contents of a header array using $this->mappedColumns
147      */
148     protected function remapHeader($header) {
149         $mappedHeader = array();
150         foreach($header as $item) {
151             if(isset($this->columnMap[strtolower($item)])) $item = $this->columnMap[strtolower($item)];
152             $mappedHeader[] = $item;
153         }
154         return $mappedHeader;
155     }
156     
157     /**
158      * Get a row from the CSV file and update $this->currentRow;
159      */
160     protected function fetchCSVRow() {
161         if(!$this->fileHandle) $this->openFile();
162         if(!$this->headerRow) $this->fetchCSVHeader();
163 
164         $this->rowNum++;
165         
166         $srcRow = fgetcsv($this->fileHandle, 0, $this->delimiter, $this->enclosure);
167         if($srcRow) {
168             $row = array();
169             foreach($srcRow as $i => $value) {
170                 // Allow escaping of quotes and commas in the data
171                 $value = str_replace(
172                         array('\\'.$this->enclosure,'\\'.$this->delimiter),
173                         array($this->enclosure,$this->delimiter),$value);
174                 if(array_key_exists($i, $this->headerRow)) {
175                     if($this->headerRow[$i]) $row[$this->headerRow[$i]] = $value;
176                 } else {
177                     user_error("No heading for column $i on row $this->rowNum", E_USER_WARNING);
178                 }
179             }
180         
181             $this->currentRow = $row;
182         } else {
183             $this->closeFile();
184         }
185         return $this->currentRow;       
186     }
187     
188     /**
189      * @ignore
190      */
191     function __destruct() {
192         $this->closeFile();
193     }
194 
195     //// ITERATOR FUNCTIONS
196 
197     /**
198      * @ignore 
199      */
200     function rewind() {
201         $this->closeFile();
202         $this->fetchCSVRow();
203     }
204     
205     /**
206      * @ignore 
207      */
208     function current() {
209         return $this->currentRow;
210     }
211     
212     /**
213      * @ignore 
214      */
215     function key() {
216         return $this->rowNum;
217     }
218     
219     /**
220      * @ignore 
221      */
222     function next() {
223         $this->fetchCSVRow();
224         return $this->currentRow;
225     }
226     
227     /**
228      * @ignore 
229      */
230     function valid() {
231         return $this->currentRow ? true : false;
232     }
233     
234 
235 }
236 
237 ?>