← All changes | libraries/csv-parser.class.php +38 -34 1.12 → 3.0 View file →

		@@ -12,8 +12,9 @@
12	12	defined( 'ABSPATH' ) || die( 'No direct script access allowed!' );
13	13
14	14	/**
15	15	* CSV Parsing class
	16	+ *
16	17	* @package TablePress
17	18	* @subpackage Import
18	19	* @author Tobias Bäthge
19	20	* @since 1.0.0
		@@ -23,43 +24,39 @@
23	24	/**
24	25	* The used character for the enclosure of a cell. Defaults to quotation mark ".
25	26	*
26	27	* @since 1.0.0
27		- * @var string
28	28	*/
29		- protected $enclosure = '"';
	29	+ protected string $enclosure = '"';
30	30
31	31	/**
32	32	* Number of rows to analyze when attempting to auto-detect the CSV delimiter.
33	33	*
34	34	* @since 1.0.0
35		- * @var int
36	35	*/
37		- protected $delimiter_search_max_lines = 15;
	36	+ protected int $delimiter_search_max_lines = 15;
38	37
39	38	/**
40	39	* Characters to ignore when attempting to auto-detect delimiter.
41	40	*
42	41	* @since 1.0.0
43		- * @var string
44	42	*/
45		- protected $non_delimiter_chars = "a-zA-Z0-9\n\r";
	43	+ protected string $non_delimiter_chars = "a-zA-Z0-9\n\r";
46	44
47	45	/**
48	46	* The preferred delimiter characters, only used when all filtering method return multiple possible delimiters (happens very rarely).
	47	+ * There must not be more than 9 characters in the preferred delimiter character list, see `_check_delimiter_count()`.
49	48	*
50	49	* @since 1.0.0
51		- * @var string
52	50	*/
53		- protected $preferred_delimiter_chars = ";,\t";
	51	+ protected string $preferred_delimiter_chars = ";,\t";
54	52
55	53	/**
56	54	* The CSV data string that shall be parsed to an array.
57	55	*
58	56	* @since 1.0.0
59		- * @var string
60	57	*/
61		- protected $import_data;
	58	+ protected string $import_data = '';
62	59
63	60	/**
64	61	* The error state while parsing input data.
65	62	*
		@@ -68,19 +65,18 @@
68	65	* 2 = The enclosure character was found in a non-enclosed field. This means the file is either corrupt,
69	66	* or does not follow the common CSV standard. Please validate the parsed data manually.
70	67	*
71	68	* @since 1.0.0
72		- * @var int
73	69	*/
74		- public $error = 0;
	70	+ public int $error = 0;
75	71
76	72	/**
77	73	* Detailed error information.
78	74	*
79	75	* @since 1.0.0
80		- * @var array
	76	+ * @var array<string, array<string, int|string>>
81	77	*/
82		- public $error_info = array();
	78	+ public array $error_info = array();
83	79
84	80	/**
85	81	* Class Constructor.
86	82	*
		@@ -96,11 +92,11 @@
96	92	* @since 1.0.0
97	93	*
98	94	* @param string $data Data to be parsed.
99	95	*/
100		- public function load_data( $data ) {
	96	+ public function load_data( string $data ): void {
101	97	// Check for mandatory trailing line break.
102		- if ( "\n" !== substr( $data, -1 ) ) {
	98	+ if ( ! str_ends_with( $data, "\n" ) ) {
103	99	$data .= "\n";
104	100	}
105	101	$this->import_data = $data;
106	102	}
		@@ -111,9 +107,9 @@
111	107	* @since 1.0.0
112	108	*
113	109	* @return string Most probable delimiter character.
114	110	*/
115		- public function find_delimiter() {
	111	+ public function find_delimiter(): string {
116	112	$data = &$this->import_data;
117	113
118	114	$delimiter_count = array();
119	115	$enclosed = false;
		@@ -130,24 +126,24 @@
130	126	// Open and closing quotes.
131	127	if ( ! $enclosed || $next_char !== $this->enclosure ) {
132	128	$enclosed = ! $enclosed; // Flip bool.
133	129	} elseif ( $enclosed ) {
134		- $i++; // Skip next character.
	130	+ ++$i; // Skip next character.
135	131	}
136		- } elseif ( ( "\n" === $curr_char && "\r" !== $prev_char || "\r" === $curr_char ) && ! $enclosed ) {
	132	+ } elseif ( ( ( "\n" === $curr_char && "\r" !== $prev_char ) || "\r" === $curr_char ) && ! $enclosed ) {
137	133	// Reached end of a line.
138		- $current_line++;
	134	+ ++$current_line;
139	135	if ( $current_line >= $this->delimiter_search_max_lines ) {
140	136	break;
141	137	}
142	138	} elseif ( ! $enclosed ) {
143	139	// 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
	140	+ // Count $curr_char if it is not in the $this->non_delimiter_chars list.
145	141	if ( 0 === preg_match( '#[' . $this->non_delimiter_chars . ']#i', $curr_char ) ) {
146	142	if ( ! isset( $delimiter_count[ $curr_char ][ $current_line ] ) ) {
147		- $delimiter_count[ $curr_char ][ $current_line ] = 0; // Initialize empty
	143	+ $delimiter_count[ $curr_char ][ $current_line ] = 0; // Initialize empty.
148	144	}
149		- $delimiter_count[ $curr_char ][ $current_line ]++;
	145	+ ++$delimiter_count[ $curr_char ][ $current_line ];
150	146	}
151	147	}
152	148	}
153	149
		@@ -167,8 +163,13 @@
167	163	arsort( $delimiter_counts, SORT_NUMERIC );
168	164	$potential_delimiters = array_keys( $delimiter_counts );
169	165	}
170	166
	167	+ // If still no delimiter was found, fall back to a comma.
	168	+ if ( empty( $potential_delimiters ) ) {
	169	+ $potential_delimiters = array( ',' );
	170	+ }
	171	+
171	172	// Return first array element, as that has the highest count.
172	173	return array_shift( $potential_delimiters );
173	174	}
174	175
		@@ -177,13 +178,13 @@
177	178	*
178	179	* @since 1.0.0
179	180	*
180	181	* @param string $char Character to check.
181		- * @param array $line_counts Counts for the characters in the lines.
	182	+ * @param int[] $line_counts Counts for the characters in the lines.
182	183	* @param int $number_lines Number of lines.
183	184	* @return bool|string False if delimiter is not possible, string to be used as a sort key if character could be a delimiter.
184	185	*/
185		- protected function _check_delimiter_count( $char, array $line_counts, $number_lines ) {
	186	+ protected function _check_delimiter_count( string $char, array $line_counts, int $number_lines ) /* : bool|string */ {
186	187	// Was the potential delimiter found in every line?
187	188	if ( count( $line_counts ) !== $number_lines ) {
188	189	return false;
189	190	}
		@@ -191,9 +192,9 @@
191	192	// Check if the count in every line is the same (or one higher for an "almost").
192	193	$first = null;
193	194	$equal = null;
194	195	$almost = false;
195		- foreach ( $line_counts as $line => $count ) {
	196	+ foreach ( $line_counts as $count ) {
196	197	if ( is_null( $first ) ) {
197	198	$first = $count;
198	199	} elseif ( $count === $first && false !== $equal ) {
199	200	$equal = true;
		@@ -210,11 +211,14 @@
210	211	}
211	212
212	213	// At this point, count is equal in all lines, so determine a string to sort priority.
213	214	$match = ( $almost ) ? 2 : 1;
	215	+ // There must not be more than 9 characters in the preferred delimiter character list.
214	216	$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 ) );
	217	+ if ( false === $pref ) {
	218	+ $pref = 9;
	219	+ }
	220	+ return $pref . $match . '.' . ( 99999 - $first );
217	221	}
218	222
219	223	/**
220	224	* Parse CSV string into a two-dimensional array.
		@@ -221,11 +225,11 @@
221	225	*
222	226	* @since 1.0.0
223	227	*
224	228	* @param string $delimiter Delimiter character for the CSV parsing.
225		- * @return array Two-dimensional array with the data from the CSV string.
	229	+ * @return array<int, array<int, string>> Two-dimensional array with the data from the CSV string.
226	230	*/
227		- public function parse( $delimiter ) {
	231	+ public function parse( string $delimiter ): array {
228	232	$data = &$this->import_data;
229	233
230	234	// Filter delimiter from the list, if it is a whitespace character.
231	235	$white_spaces = str_replace( $delimiter, '', " \t\x0B\0" );
		@@ -265,12 +269,12 @@
265	269	}
266	270	} elseif ( $next_char === $this->enclosure ) {
267	271	// Enclosure character within enclosed cell (" encoded as "").
268	272	$cell_content .= $curr_char;
269		- $i++; // Skip next character
	273	+ ++$i; // Skip next character.
270	274	} elseif ( $next_char !== $delimiter && "\r" !== $next_char && "\n" !== $next_char ) {
271	275	// for-loop (instead of while-loop) that skips whitespace.
272		- for ( $x = ( $i + 1 ); isset( $data[ $x ] ) && '' === ltrim( $data[ $x ], $white_spaces ); $x++ ) {
	276	+ for ( $x = ( $i + 1 ); isset( $data[ $x ] ) && '' === ltrim( $data[ $x ], $white_spaces ); $x++ ) { // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed,Generic.CodeAnalysis.EmptyStatement.DetectedFor
273	277	// Action is in iterator check.
274	278	}
275	279	if ( $data[ $x ] === $delimiter ) {
276	280	$enclosed = false;
		@@ -301,9 +305,9 @@
301	305
302	306	$row[ $column ] = ( $was_enclosed ) ? $cell_content : trim( $cell_content );
303	307	$cell_content = '';
304	308	$was_enclosed = false;
305		- $column++;
	309	+ ++$column;
306	310
307	311	// End of line.
308	312	if ( "\n" === $curr_char || "\r" === $curr_char ) {
309	313	// Append completed row.
		@@ -311,9 +315,9 @@
311	315	$row = array();
312	316	$column = 0;
313	317	if ( "\r" === $curr_char && "\n" === $next_char ) {
314	318	// Skip next character in \r\n line breaks.
315		- $i++;
	319	+ ++$i;
316	320	}
317	321	}
318	322	} else {
319	323	// Append character to current cell.