← All changes | libraries/csv-parser.class.php +40 -34 1.12 → 3.4 View file →

		@@ -7,13 +7,16 @@
7	7	* @author Tobias Bäthge
8	8	* @since 1.0.0
9	9	*/
10	10
	11	+declare(strict_types=1);
	12	+
11	13	// Prohibit direct script loading.
12	14	defined( 'ABSPATH' ) || die( 'No direct script access allowed!' );
13	15
14	16	/**
15	17	* CSV Parsing class
	18	+ *
16	19	* @package TablePress
17	20	* @subpackage Import
18	21	* @author Tobias Bäthge
19	22	* @since 1.0.0
		@@ -23,43 +26,39 @@
23	26	/**
24	27	* The used character for the enclosure of a cell. Defaults to quotation mark ".
25	28	*
26	29	* @since 1.0.0
27		- * @var string
28	30	*/
29		- protected $enclosure = '"';
	31	+ protected string $enclosure = '"';
30	32
31	33	/**
32	34	* Number of rows to analyze when attempting to auto-detect the CSV delimiter.
33	35	*
34	36	* @since 1.0.0
35		- * @var int
36	37	*/
37		- protected $delimiter_search_max_lines = 15;
	38	+ protected int $delimiter_search_max_lines = 15;
38	39
39	40	/**
40	41	* Characters to ignore when attempting to auto-detect delimiter.
41	42	*
42	43	* @since 1.0.0
43		- * @var string
44	44	*/
45		- protected $non_delimiter_chars = "a-zA-Z0-9\n\r";
	45	+ protected string $non_delimiter_chars = "a-zA-Z0-9\n\r";
46	46
47	47	/**
48	48	* The preferred delimiter characters, only used when all filtering method return multiple possible delimiters (happens very rarely).
	49	+ * There must not be more than 9 characters in the preferred delimiter character list, see `_check_delimiter_count()`.
49	50	*
50	51	* @since 1.0.0
51		- * @var string
52	52	*/
53		- protected $preferred_delimiter_chars = ";,\t";
	53	+ protected string $preferred_delimiter_chars = ";,\t";
54	54
55	55	/**
56	56	* The CSV data string that shall be parsed to an array.
57	57	*
58	58	* @since 1.0.0
59		- * @var string
60	59	*/
61		- protected $import_data;
	60	+ protected string $import_data = '';
62	61
63	62	/**
64	63	* The error state while parsing input data.
65	64	*
		@@ -68,19 +67,18 @@
68	67	* 2 = The enclosure character was found in a non-enclosed field. This means the file is either corrupt,
69	68	* or does not follow the common CSV standard. Please validate the parsed data manually.
70	69	*
71	70	* @since 1.0.0
72		- * @var int
73	71	*/
74		- public $error = 0;
	72	+ public int $error = 0;
75	73
76	74	/**
77	75	* Detailed error information.
78	76	*
79	77	* @since 1.0.0
80		- * @var array
	78	+ * @var array<string, array<string, int|string>>
81	79	*/
82		- public $error_info = array();
	80	+ public array $error_info = array();
83	81
84	82	/**
85	83	* Class Constructor.
86	84	*
		@@ -96,11 +94,11 @@
96	94	* @since 1.0.0
97	95	*
98	96	* @param string $data Data to be parsed.
99	97	*/
100		- public function load_data( $data ) {
	98	+ public function load_data( string $data ): void {
101	99	// Check for mandatory trailing line break.
102		- if ( "\n" !== substr( $data, -1 ) ) {
	100	+ if ( ! str_ends_with( $data, "\n" ) ) {
103	101	$data .= "\n";
104	102	}
105	103	$this->import_data = $data;
106	104	}
		@@ -111,9 +109,9 @@
111	109	* @since 1.0.0
112	110	*
113	111	* @return string Most probable delimiter character.
114	112	*/
115		- public function find_delimiter() {
	113	+ public function find_delimiter(): string {
116	114	$data = &$this->import_data;
117	115
118	116	$delimiter_count = array();
119	117	$enclosed = false;
		@@ -130,24 +128,24 @@
130	128	// Open and closing quotes.
131	129	if ( ! $enclosed || $next_char !== $this->enclosure ) {
132	130	$enclosed = ! $enclosed; // Flip bool.
133	131	} elseif ( $enclosed ) {
134		- $i++; // Skip next character.
	132	+ ++$i; // Skip next character.
135	133	}
136		- } elseif ( ( "\n" === $curr_char && "\r" !== $prev_char || "\r" === $curr_char ) && ! $enclosed ) {
	134	+ } elseif ( ( ( "\n" === $curr_char && "\r" !== $prev_char ) || "\r" === $curr_char ) && ! $enclosed ) {
137	135	// Reached end of a line.
138		- $current_line++;
	136	+ ++$current_line;
139	137	if ( $current_line >= $this->delimiter_search_max_lines ) {
140	138	break;
141	139	}
142	140	} elseif ( ! $enclosed ) {
143	141	// At this point, $curr_char seems to be used as a delimiter, as it is not enclosed.
144		- // Count $curr_char if it is not in the $this->non_delimiter_chars list
	142	+ // Count $curr_char if it is not in the $this->non_delimiter_chars list.
145	143	if ( 0 === preg_match( '#[' . $this->non_delimiter_chars . ']#i', $curr_char ) ) {
146	144	if ( ! isset( $delimiter_count[ $curr_char ][ $current_line ] ) ) {
147		- $delimiter_count[ $curr_char ][ $current_line ] = 0; // Initialize empty
	145	+ $delimiter_count[ $curr_char ][ $current_line ] = 0; // Initialize empty.
148	146	}
149		- $delimiter_count[ $curr_char ][ $current_line ]++;
	147	+ ++$delimiter_count[ $curr_char ][ $current_line ];
150	148	}
151	149	}
152	150	}
153	151
		@@ -167,8 +165,13 @@
167	165	arsort( $delimiter_counts, SORT_NUMERIC );
168	166	$potential_delimiters = array_keys( $delimiter_counts );
169	167	}
170	168
	169	+ // If still no delimiter was found, fall back to a comma.
	170	+ if ( empty( $potential_delimiters ) ) {
	171	+ $potential_delimiters = array( ',' );
	172	+ }
	173	+
171	174	// Return first array element, as that has the highest count.
172	175	return array_shift( $potential_delimiters );
173	176	}
174	177
		@@ -177,13 +180,13 @@
177	180	*
178	181	* @since 1.0.0
179	182	*
180	183	* @param string $char Character to check.
181		- * @param array $line_counts Counts for the characters in the lines.
	184	+ * @param int[] $line_counts Counts for the characters in the lines.
182	185	* @param int $number_lines Number of lines.
183	186	* @return bool|string False if delimiter is not possible, string to be used as a sort key if character could be a delimiter.
184	187	*/
185		- protected function _check_delimiter_count( $char, array $line_counts, $number_lines ) {
	188	+ protected function _check_delimiter_count( string $char, array $line_counts, int $number_lines ) /* : bool|string */ {
186	189	// Was the potential delimiter found in every line?
187	190	if ( count( $line_counts ) !== $number_lines ) {
188	191	return false;
189	192	}
		@@ -191,9 +194,9 @@
191	194	// Check if the count in every line is the same (or one higher for an "almost").
192	195	$first = null;
193	196	$equal = null;
194	197	$almost = false;
195		- foreach ( $line_counts as $line => $count ) {
	198	+ foreach ( $line_counts as $count ) {
196	199	if ( is_null( $first ) ) {
197	200	$first = $count;
198	201	} elseif ( $count === $first && false !== $equal ) {
199	202	$equal = true;
		@@ -210,11 +213,14 @@
210	213	}
211	214
212	215	// At this point, count is equal in all lines, so determine a string to sort priority.
213	216	$match = ( $almost ) ? 2 : 1;
	217	+ // There must not be more than 9 characters in the preferred delimiter character list.
214	218	$pref = strpos( $this->preferred_delimiter_chars, $char );
215		- $pref = ( false !== $pref ) ? str_pad( $pref, 3, '0', STR_PAD_LEFT ) : '999';
216		- return $pref . $match . '.' . ( 99999 - str_pad( $first, 5, '0', STR_PAD_LEFT ) );
	219	+ if ( false === $pref ) {
	220	+ $pref = 9;
	221	+ }
	222	+ return $pref . $match . '.' . ( 99999 - $first );
217	223	}
218	224
219	225	/**
220	226	* Parse CSV string into a two-dimensional array.
		@@ -221,11 +227,11 @@
221	227	*
222	228	* @since 1.0.0
223	229	*
224	230	* @param string $delimiter Delimiter character for the CSV parsing.
225		- * @return array Two-dimensional array with the data from the CSV string.
	231	+ * @return array<int, array<int, string>> Two-dimensional array with the data from the CSV string.
226	232	*/
227		- public function parse( $delimiter ) {
	233	+ public function parse( string $delimiter ): array {
228	234	$data = &$this->import_data;
229	235
230	236	// Filter delimiter from the list, if it is a whitespace character.
231	237	$white_spaces = str_replace( $delimiter, '', " \t\x0B\0" );
		@@ -265,12 +271,12 @@
265	271	}
266	272	} elseif ( $next_char === $this->enclosure ) {
267	273	// Enclosure character within enclosed cell (" encoded as "").
268	274	$cell_content .= $curr_char;
269		- $i++; // Skip next character
	275	+ ++$i; // Skip next character.
270	276	} elseif ( $next_char !== $delimiter && "\r" !== $next_char && "\n" !== $next_char ) {
271	277	// for-loop (instead of while-loop) that skips whitespace.
272		- for ( $x = ( $i + 1 ); isset( $data[ $x ] ) && '' === ltrim( $data[ $x ], $white_spaces ); $x++ ) {
	278	+ for ( $x = ( $i + 1 ); isset( $data[ $x ] ) && '' === ltrim( $data[ $x ], $white_spaces ); $x++ ) { // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed,Generic.CodeAnalysis.EmptyStatement.DetectedFor
273	279	// Action is in iterator check.
274	280	}
275	281	if ( $data[ $x ] === $delimiter ) {
276	282	$enclosed = false;
		@@ -301,9 +307,9 @@
301	307
302	308	$row[ $column ] = ( $was_enclosed ) ? $cell_content : trim( $cell_content );
303	309	$cell_content = '';
304	310	$was_enclosed = false;
305		- $column++;
	311	+ ++$column;
306	312
307	313	// End of line.
308	314	if ( "\n" === $curr_char || "\r" === $curr_char ) {
309	315	// Append completed row.
		@@ -311,9 +317,9 @@
311	317	$row = array();
312	318	$column = 0;
313	319	if ( "\r" === $curr_char && "\n" === $next_char ) {
314	320	// Skip next character in \r\n line breaks.
315		- $i++;
	321	+ ++$i;
316	322	}
317	323	}
318	324	} else {
319	325	// Append character to current cell.