<?php

/**
 * CSV file parser
 *
 * Currently the string matching doesn't work
 * if the output encoding is not ASCII or UTF-8
 *
 * Based on the excellent work by Ming Hong Ng:
 * http://minghong.blogspot.com/2006/07/csv-parser-for-php.html
 *
 * Edits by Fabrizio Angius (http://hippydream.42cows.org/):
 *  Added two additional output modes: hash (associative array) and object.
 *  Removing empty lines.
 *
 * License: http://creativecommons.org/licenses/by-nc/2.0/
 */

// Default mode - returns a 2D array with standard integer indexes
define('CSV_ARRAY_MODE',                     0);
// Assumes that the first line of the file contains the column names and returns an associative array
define('CSV_HASH_MODE',                      1);
// Assumes that the first line of the file contains the column names and returns an object with member variables named like the columns. 
// Warning: characters different than [a-zA-Z_0-9] are removed. The first char must be a non-digit or an underscore ("_") will be prepended. 
// Empty column names are named "colN" where N is the column index.
define('CSV_OBJECT_MODE',                    2);

class CsvItemWrapper {
	var $fields;              // Container for actual data

	/**
	 * Constructor
	 */
	function CsvItemWrapper() {
		$this->fields = array();
	}
	
	/**
	 * PHP5 magic method ;-)
	 */
	function __get( $key ) {
		return $this->fields[$key];
	}
	
	/**
	 * PHP5 magic method ;-)
	 */
	function __set( $key, $value ) {
		set( $key, $value );
	}
	
	/**
	 * Convenience method.
	 */
	function set( $key, $value ) {
		$this->fields[$key] = $value;
	}
}

class CsvFileParser {
	var $delimiter;         // Field delimiter
	var $enclosure;         // Field enclosure character
	var $inputEncoding;     // Input character encoding
	var $outputEncoding;    // Output character encoding
	var $data;              // CSV data as 2D array
	var $columns;           // Holds column names
	var $mode;              // See the CSV_*_MODE defines

	/**
	 * Constructor
	 */
	function CsvFileParser() {
		$this->delimiter = ",";
		$this->enclosure = '"';
		$this->inputEncoding = "ISO-8859-1";
		$this->outputEncoding = "ISO-8859-1";
		$this->data = array();
		$this->columns = array();
		$this->mode = CSV_ARRAY_MODE;
	}
	
	/**
	 * Cleans column names according to the selected output mode.
	 */
	function SetColumnNames($row) {
		for ($i = 0; $i < count($row); $i++) {
			
			$label = $row[$i];
			if ($this->mode == CSV_OBJECT_MODE) {
				$label = preg_replace('/[^a-z0-9_]/i', '', $label);
				if (preg_match('/^[0-9]/', $label))
					$label = '_'.$label;
			}
			if ($label == '')
				$label = 'col'.$i;
			$this->columns[$i] = $label;
		}
	}
	
	/**
	 * Creates a hash from a row of values.
	 */
	function CreateHash($row) {
		$item = array();
		for ($i = 0; $i < count($row); $i++) {
			if (!isset($this->columns[$i]))
				continue;
			$item[$this->columns[$i]] = $row[$i];
		}
		return $item;
	}
	
	/**
	 * Creates a CsvItemWrapper object from a row of values.
	 */
	function CreateObject($row) {
		$item = new CsvItemWrapper();
		for ($i = 0; $i < count($row); $i++) {
			if (!isset($this->columns[$i]))
				continue;
			$item->set($this->columns[$i], $row[$i]);
		}
		return $item;
	}

	/**
	 * Parse CSV from file
	 * @param   content     The CSV filename
	 * @param   hasBOM      Using BOM or not
	 * @return Success or not
	 */
	function ParseFromFile( $filename, $hasBOM = false ) {
		if ( !is_readable($filename) ) {
			// echo "not readable";
			return false;
		}
		return $this->ParseFromString( file_get_contents($filename), $hasBOM );
	}

	/**
	 * Parse CSV from string
	 * @param   content     The CSV string
	 * @param   hasBOM      Using BOM or not
	 * @return Success or not
	 */
	function ParseFromString( $content, $hasBOM = false ) {
		$content = iconv($this->inputEncoding, $this->outputEncoding, $content );
		$content = str_replace( "\r\n", "\n", $content );
		$content = str_replace( "\r", "\n", $content );
		$content = trim($content, "\n"); // Remove empty lines
		
		// Remove the BOM (first 3 bytes)
		if ( $hasBOM ) {
			$content = substr( $content, 3 );
		}
		// Make sure it always end with a newline
		if ( $content[strlen($content)-1] != "\n" ) {
			$content .= "\n";
		}

		// Parse the content character by character
		$row = array( "" );
		$idx = 0;
		$quoted = false;
		$headerFound = false;
		for ( $i = 0; $i < strlen($content); $i++ ) {
			$ch = $content[$i];
			if ( $ch == $this->enclosure ) {
				$quoted = !$quoted;
			}

			// End of line
			if ( $ch == "\n" && !$quoted ) {
				// Remove enclosure delimiters
				for ( $k = 0; $k < count($row); $k++ ) {
					if ( $row[$k] != "" && $row[$k][0] == $this->enclosure ) {
						$row[$k] = substr( $row[$k], 1, strlen($row[$k]) - 2 );
					}
					$row[$k] = str_replace( str_repeat($this->enclosure, 2), $this->enclosure, $row[$k] );
				}

				if ($this->mode != CSV_ARRAY_MODE && !$headerFound) {
					// Setup column names
					$this->SetColumnNames($row);
					$headerFound = true;
				} else {
					// Append row into table
					if ($this->mode == CSV_ARRAY_MODE) {
						$this->data[] = $row;
					} else if ($this->mode == CSV_HASH_MODE) {
						$this->data[] = $this->CreateHash($row);
					} else if ($this->mode == CSV_OBJECT_MODE) {
						$this->data[] = $this->CreateObject($row);
					}
				}
				
				$row = array( "" );
				$idx = 0;
			}

			// End of field
			else if ( $ch == $this->delimiter && !$quoted ) {
				$row[++$idx] = "";
			}

			// Inside the field
			else {
				$row[$idx] .= $ch;
			}
		}

		return true;
	}
}



//////// TEST

/* 
$filename = 'data.csv';

$parser = new CsvFileParser();
$parser->delimiter = ";";
$parser->enclosure = '';
$parser->inputEncoding = 'UTF-8';
$parser->outputEncoding = 'UTF-8';

echo "<h1>PHP CvsFileParser v. 1.0</h1><br><br>Array Mode (default):<br>";
$parser->ParseFromFile($filename, true);
print_r($parser->data);

echo "<br><br>Object Mode:<br>";
$parser->mode = CSV_OBJECT_MODE;
$parser->ParseFromFile('../osdab_data/data.csv', true);
print_r($parser->data);

echo "<br><br>Hash Mode:<br>";
$parser->mode = CSV_HASH_MODE;
$parser->ParseFromFile('../osdab_data/data.csv', true);
print_r($parser->data);
*/

//////////////


?>