import-users-from-csv / class-readcsv.php

class-readcsv.php in Import Users from CSV 1.2, at class-readcsv.php

152 lines 3.6 KB

Raw Download Zip

1	<?php
2	/**
3	* @package uc_stock_update
4	* @version 6.x-2.0.2
5	* @licence GNU GPL v2
6	*/
7	// $Id: ReadCSV.inc 3 2011-09-19 02:32:14Z david $
8	/**
9	* Use this to read CSV files. PHP's fgetcsv() does not conform to RFC
10	* 4180. In particular, it doesn't handle the correct quote escaping
11	* syntax. See http://tools.ietf.org/html/rfc4180
12	*
13	* David Houlder May 2010
14	* http://davidhoulder.com
15	*/
16
17	class ReadCSV {
18	const field_start = 0;
19	const unquoted_field = 1;
20	const quoted_field = 2;
21	const found_quote = 3;
22	const found_cr_q = 4;
23	const found_cr = 5;
24
25	private $file;
26	private $sep;
27	// If $eof is TRUE, the next next_char() will return FALSE.
28	// Note that this is different to feof(), which is TRUE
29	// _after_ EOF is encountered.
30	private $eof;
31	private $nc;
32
33	/**
34	* @param $file_handle
35	* open file to read from
36	* @param $sep
37	* column separator character
38	* @param $skip
39	* initial character sequence to skip if found. e.g. UTF-8 byte-order mark
40	*/
41	public function __construct($file_handle, $sep, $skip="") {
42	$this->file = $file_handle;
43	$this->sep = $sep;
44	$this->nc = fgetc($this->file);
45	// skip junk at start
46	for ($i=0; $i < strlen($skip); $i++) {
47	if ($this->nc !== $skip[$i])
48	break;
49	$this->nc = fgetc($this->file);
50	}
51	$this->eof = ($this->nc === FALSE);
52	}
53
54	private function next_char() {
55	$c = $this->nc;
56	$this->nc = fgetc($this->file);
57	$this->eof = ($this->nc === FALSE);
58	return $c;
59	}
60
61
62	/**
63	* Get next record from CSV file.
64	*
65	* @return
66	* array of strings from the next record in the CSV file, or NULL if
67	* there are no more records.
68	*/
69	public function get_row() {
70	if ($this->eof)
71	return NULL;
72
73	$row=array();
74	$field="";
75	$state=self::field_start;
76
77	while (1) {
78	$char = $this->next_char();
79
80	if ($state == self::quoted_field) {
81	if ($char === FALSE) {
82	// EOF. (TODO: error case - no closing quote)
83	$row[]=$field;
84	return $row;
85	}
86	// Fall through to accumulate quoted chars in switch() {...}
87	} elseif ($char === FALSE \|\| $char == "\n") {
88	// End of record.
89	// (TODO: error case if $state==self::field_start here - trailing comma)
90	$row[] = $field;
91	return $row;
92	} elseif ($char == "\r") {
93	// Possible start of \r\n line end, but might be just part of foo\rbar
94	$state = ($state == self::found_quote)? self::found_cr_q: self::found_cr;
95	continue;
96	} elseif ($char == $this->sep &&
97	($state == self::field_start \|\|
98	$state == self::found_quote \|\|
99	$state == self::unquoted_field)) {
100	// End of current field, start of next field
101	$row[]=$field;
102	$field="";
103	$state=self::field_start;
104	continue;
105	}
106
107	switch ($state) {
108
109	case self::field_start:
110	if ($char == '"')
111	$state = self::quoted_field;
112	else {
113	$state = self::unquoted_field;
114	$field .= $char;
115	}
116	break;
117
118	case self::quoted_field:
119	if ($char == '"')
120	$state = self::found_quote;
121	else
122	$field .= $char;
123	break;
124
125	case self::unquoted_field:
126	$field .= $char;
127	// (TODO: error case if '"' in middle of unquoted field)
128	break;
129
130	case self::found_quote:
131	// Found '"' escape sequence
132	$field .= $char;
133	$state = self::quoted_field;
134	// (TODO: error case if $char!='"' - non-separator char after single quote)
135	break;
136
137	case self::found_cr:
138	// Lone \rX instead of \r\n. Treat as literal \rX. (TODO: error case?)
139	$field .= "\r".$char;
140	$state = self::unquoted_field;
141	break;
142
143	case self::found_cr_q:
144	// (TODO: error case: "foo"\rX instead of "foo"\r\n or "foo"\n)
145	$field .= "\r".$char;
146	$state = self::quoted_field;
147	break;
148	}
149	}
150	}
151	}
152