class-readcsv.php

<?php
/**
 * @package uc_stock_update
 * @version 6.x-2.0.2
 * @licence GNU GPL v2
 */
// $Id: ReadCSV.inc 3 2011-09-19 02:32:14Z david $
/**
 * Use this to read CSV files. PHP's fgetcsv() does not conform to RFC
 * 4180. In particular, it doesn't handle the correct quote escaping
 * syntax. See http://tools.ietf.org/html/rfc4180
 *
 * David Houlder May 2010
 * http://davidhoulder.com
 */

class ReadCSV {
	const field_start = 0;
	const unquoted_field = 1;
	const quoted_field = 2;
	const found_quote = 3;
	const found_cr_q = 4;
	const found_cr =  5;

	private $file;
	private $sep;
	// If $eof is TRUE, the next next_char() will return FALSE.
	// Note that this is different to feof(), which is TRUE
	// _after_ EOF is encountered.
	private $eof;
	private $nc;

	/**
	 * @param $file_handle
	 *  open file to read from
	 * @param $sep
	 *  column separator character
	 * @param $skip
	 *  initial character sequence to skip if found. e.g. UTF-8 byte-order mark
	 */
	public function __construct($file_handle, $sep, $skip="") {
		$this->file = $file_handle;
		$this->sep = $sep;
		$this->nc = fgetc($this->file);
		// skip junk at start
		for ($i=0; $i < strlen($skip); $i++) {
			if ($this->nc !== $skip[$i])
				break;
			$this->nc = fgetc($this->file);
		}
		$this->eof = ($this->nc === FALSE);
	}

	private function next_char() {
		$c = $this->nc;
		$this->nc = fgetc($this->file);
		$this->eof = ($this->nc === FALSE);
		return $c;
	}


	/**
	 * Get next record from CSV file.
	 *
	 * @return
	 *  array of strings from the next record in the CSV file, or NULL if
	 *  there are no more records.
	 */
	public function get_row() {
		if ($this->eof)
			return NULL;

		$row=array();
		$field="";
		$state=self::field_start;

		while (1) {
			$char = $this->next_char();

			if ($state == self::quoted_field) {
				if ($char === FALSE) {
					// EOF. (TODO: error case - no closing quote)
					$row[]=$field;
					return $row;
				}
				// Fall through to accumulate quoted chars in switch() {...}
			} elseif ($char === FALSE || $char == "\n") {
				// End of record.
				// (TODO: error case if $state==self::field_start here - trailing comma)
				$row[] = $field;
				return $row;
			} elseif ($char == "\r") {
				// Possible start of \r\n line end, but might be just part of foo\rbar
				$state = ($state == self::found_quote)? self::found_cr_q: self::found_cr;
				continue;
			} elseif ($char == $this->sep && 
				($state == self::field_start ||
				$state == self::found_quote ||
				$state == self::unquoted_field)) {
					// End of current field, start of next field
					$row[]=$field;
					$field="";
					$state=self::field_start;
					continue;
				}

			switch ($state) {

			case self::field_start:
				if ($char == '"')
					$state = self::quoted_field;
				else {
					$state = self::unquoted_field;
					$field .= $char;
				}
				break;

			case self::quoted_field:
				if ($char == '"')
					$state = self::found_quote;
				else
					$field .= $char;
				break;

			case self::unquoted_field:
				$field .= $char;
				// (TODO: error case if '"' in middle of unquoted field)
				break;

			case self::found_quote:
				// Found '"' escape sequence
				$field .= $char;
				$state = self::quoted_field;
				// (TODO: error case if $char!='"' - non-separator char after single quote)
				break;

			case self::found_cr:
				// Lone \rX instead of \r\n. Treat as literal \rX. (TODO: error case?)
				$field .= "\r".$char;
				$state = self::unquoted_field;
				break;

			case self::found_cr_q:
				// (TODO: error case: "foo"\rX instead of "foo"\r\n or "foo"\n)
				$field .= "\r".$char;
				$state = self::quoted_field;
				break;
			}
		}
	}
}