tablepress / libraries / vendor / PhpSpreadsheet / Reader / Csv / Delimiter.php

Delimiter.php in TablePress – Tables in WordPress made easy 3.4, at libraries/vendor/PhpSpreadsheet/Reader/Csv/Delimiter.php

145 lines 3.8 KB

Raw Download Zip

1	<?php
2
3	namespace TablePress\PhpOffice\PhpSpreadsheet\Reader\Csv;
4
5	class Delimiter
6	{
7	protected const POTENTIAL_DELIMITERS = [',', ';', "\t", '\|', ':', ' ', '~'];
8
9	/** @var resource */
10	protected $fileHandle;
11
12	protected string $escapeCharacter;
13
14	protected string $enclosure;
15
16	/** @var array<string, int[]> */
17	protected array $counts = [];
18
19	protected int $numberLines = 0;
20
21	protected ?string $delimiter = null;
22
23	/**
24	* @param resource $fileHandle
25	*/
26	public function __construct($fileHandle, string $escapeCharacter, string $enclosure)
27	{
28	$this->fileHandle = $fileHandle;
29	$this->escapeCharacter = $escapeCharacter;
30	$this->enclosure = $enclosure;
31
32	$this->countPotentialDelimiters();
33	}
34
35	public function getDefaultDelimiter(): string
36	{
37	return self::POTENTIAL_DELIMITERS[0];
38	}
39
40	public function linesCounted(): int
41	{
42	return $this->numberLines;
43	}
44
45	protected function countPotentialDelimiters(): void
46	{
47	$this->counts = array_fill_keys(self::POTENTIAL_DELIMITERS, []);
48	$delimiterKeys = array_flip(self::POTENTIAL_DELIMITERS);
49
50	// Count how many times each of the potential delimiters appears in each line
51	$this->numberLines = 0;
52	while (($line = $this->getNextLine()) !== false && (++$this->numberLines < 1000)) {
53	$this->countDelimiterValues($line, $delimiterKeys);
54	}
55	}
56
57	/** @param array<string, int> $delimiterKeys */
58	protected function countDelimiterValues(string $line, array $delimiterKeys): void
59	{
60	$splitString = mb_str_split($line, 1, 'UTF-8');
61	$distribution = array_count_values($splitString);
62	$countLine = array_intersect_key($distribution, $delimiterKeys);
63
64	foreach (self::POTENTIAL_DELIMITERS as $delimiter) {
65	$this->counts[$delimiter][] = $countLine[$delimiter] ?? 0;
66	}
67	}
68
69	public function infer(): ?string
70	{
71	// Calculate the mean square deviations for each delimiter
72	// (ignoring delimiters that haven't been found consistently)
73	$meanSquareDeviations = [];
74	$middleIdx = (int) floor(($this->numberLines - 1) / 2);
75
76	foreach (self::POTENTIAL_DELIMITERS as $delimiter) {
77	$series = $this->counts[$delimiter];
78	sort($series);
79
80	$median = ($this->numberLines % 2)
81	? $series[$middleIdx]
82	: ($series[$middleIdx] + $series[$middleIdx + 1]) / 2;
83
84	if ($median === 0) {
85	continue;
86	}
87
88	$meanSquareDeviations[$delimiter] = array_reduce(
89	$series,
90	fn ($sum, $value) => $sum + ($value - $median) ** 2
91	) / count($series);
92	}
93
94	// ... and pick the delimiter with the smallest mean square deviation
95	// (in case of ties, the order in potentialDelimiters is respected)
96	$min = INF;
97	foreach (self::POTENTIAL_DELIMITERS as $delimiter) {
98	if (!isset($meanSquareDeviations[$delimiter])) {
99	continue;
100	}
101
102	if ($meanSquareDeviations[$delimiter] < $min) {
103	$min = $meanSquareDeviations[$delimiter];
104	$this->delimiter = $delimiter;
105	}
106	}
107
108	return $this->delimiter;
109	}
110
111	/**
112	* Get the next full line from the file.
113	*
114	* @return false\|string
115	*/
116	public function getNextLine()
117	{
118	$line = '';
119	$enclosure = ($this->escapeCharacter === '' ? ''
120	: ('(?<!' . preg_quote($this->escapeCharacter, '/') . ')'))
121	. preg_quote($this->enclosure, '/');
122
123	do {
124	// Get the next line in the file
125	$newLine = fgets($this->fileHandle);
126
127	// Return false if there is no next line
128	if ($newLine === false) {
129	return false;
130	}
131
132	// Add the new line to the line passed in
133	$line = $line . $newLine;
134
135	// Drop everything that is enclosed to avoid counting false positives in enclosures
136	$line = (string) preg_replace('/(' . $enclosure . '.*' . $enclosure . ')/Us', '', $line);
137
138	// See if we have any enclosures left in the line
139	// if we still have an enclosure then we need to read the next line as well
140	} while (preg_match('/(' . $enclosure . ')/', $line) > 0);
141
142	return ($line !== '') ? $line : false;
143	}
144	}
145