404-solution / includes / services / ImportCsvParser.php

ImportCsvParser.php in 404 Solution trunk, at includes/services/ImportCsvParser.php

233 lines 8.1 KB

Raw Download Zip

1	<?php
2
3	if (!defined('ABSPATH')) {
4	exit;
5	}
6
7	/**
8	* Parses supported redirect-import CSV shapes into canonical row arrays.
9	*/
10	class ABJ_404_Solution_ImportCsvParser {
11
12	/**
13	* @param mixed $line
14	* @return array<string, string>
15	*/
16	function splitCsvLine($line) {
17	if (!is_string($line)) {
18	$line = is_scalar($line) ? (string)$line : '';
19	}
20
21	$data = array_map(function($v) {
22	return trim((string)$v);
23	}, str_getcsv($line, ',', '"', '\\'));
24
25	if (count($data) === 5) {
26	return array(
27	'from_url' => $data[0],
28	'status' => $data[1],
29	'type' => $data[2],
30	'to_url' => $data[3],
31	'wp_type' => $data[4]
32	);
33	} else if (count($data) === 2) {
34	return array(
35	'from_url' => $data[0],
36	'to_url' => $data[1]
37	);
38	}
39
40	return array('error' => sprintf(__('Invalid CSV format. %d columns found but 2 or 5 expected.', '404-solution'), count($data)));
41	}
42
43	/**
44	* @param array<int, string> $columns
45	* @return bool
46	*/
47	function isCompatibleImportHeaderRow($columns) {
48	$normalized = $this->normalizeImportHeaders($columns);
49	$fromIndex = $this->findImportHeaderIndex($normalized, array('from_url', 'request', 'source', 'url', 'match_url'));
50	$toIndex = $this->findImportHeaderIndex($normalized, array('to_url', 'target', 'destination', 'action_data', 'redirect_to', 'url_to'));
51	return ($fromIndex !== -1 && $toIndex !== -1);
52	}
53
54	/**
55	* @param array<int, string> $columns
56	* @return array<int, string\|null>
57	*/
58	function normalizeImportHeaders($columns) {
59	return array_map(function($value) {
60	$value = preg_replace('/^\xEF\xBB\xBF/', '', (string)$value);
61	$value = trim(strtolower((string)$value));
62	return preg_replace('/[^a-z0-9_]/', '', str_replace(' ', '_', $value));
63	}, $columns);
64	}
65
66	/**
67	* @param array<int, string> $columns
68	* @return string
69	*/
70	function detectImportFormatFromHeaders($columns) {
71	$normalized = $this->normalizeImportHeaders($columns);
72
73	if (in_array('source', $normalized, true) &&
74	in_array('target', $normalized, true) &&
75	in_array('regex', $normalized, true)) {
76	return 'redirection';
77	}
78
79	if (in_array('redirect_from', $normalized, true) &&
80	in_array('redirect_to', $normalized, true)) {
81	return 'safe_redirect_manager';
82	}
83
84	if (in_array('request', $normalized, true) &&
85	in_array('destination', $normalized, true)) {
86	return 'simple_301';
87	}
88
89	if ((in_array('from_url', $normalized, true) &&
90	in_array('to_url', $normalized, true)) \|\|
91	(in_array('from_url', $normalized, true) &&
92	in_array('status', $normalized, true) &&
93	in_array('type', $normalized, true) &&
94	in_array('to_url', $normalized, true))) {
95	return 'native';
96	}
97
98	return 'unknown';
99	}
100
101	/**
102	* @param array<int, string> $row
103	* @param array<int, string\|null> $normalizedHeaders
104	* @return array<string, string>
105	*/
106	function mapImportRowByHeaders($row, $normalizedHeaders) {
107	$fromIndex = $this->findImportHeaderIndex($normalizedHeaders, array('from_url', 'request', 'source', 'url', 'match_url'));
108	$toIndex = $this->findImportHeaderIndex($normalizedHeaders, array('to_url', 'target', 'destination', 'action_data', 'redirect_to', 'url_to'));
109
110	if ($fromIndex === -1 \|\| $toIndex === -1) {
111	return array('error' => __('Invalid CSV format. Could not map source/destination columns.', '404-solution'));
112	}
113
114	$from = array_key_exists($fromIndex, $row) ? trim((string)$row[$fromIndex]) : '';
115	$to = array_key_exists($toIndex, $row) ? trim((string)$row[$toIndex]) : '';
116
117	if ($from === '' && $to === '') {
118	return array('from_url' => '', 'to_url' => '');
119	}
120
121	$result = array(
122	'from_url' => $from,
123	'to_url' => $to,
124	);
125
126	$engineIndex = $this->findImportHeaderIndex($normalizedHeaders, array('engine'));
127	if ($engineIndex !== -1 && array_key_exists($engineIndex, $row)) {
128	$result['engine'] = trim((string)$row[$engineIndex]);
129	}
130
131	$codeIndex = $this->findImportHeaderIndex($normalizedHeaders, array('code', 'redirect_code', 'http_code'));
132	if ($codeIndex !== -1 && array_key_exists($codeIndex, $row)) {
133	$result['code'] = trim((string)$row[$codeIndex]);
134	}
135
136	$statusIndex = $this->findImportHeaderIndex($normalizedHeaders, array('status', 'redirect_status'));
137	if ($statusIndex !== -1 && array_key_exists($statusIndex, $row)) {
138	$result['status'] = trim((string)$row[$statusIndex]);
139	}
140
141	$regexIndex = $this->findImportHeaderIndex($normalizedHeaders, array('regex', 'is_regex'));
142	if ($regexIndex !== -1 && array_key_exists($regexIndex, $row)) {
143	$result['regex'] = trim((string)$row[$regexIndex]);
144	}
145
146	return $result;
147	}
148
149	/**
150	* @param array<int, string\|null> $headers
151	* @param array<int, string> $candidates
152	* @return int
153	*/
154	private function findImportHeaderIndex($headers, $candidates) {
155	foreach ($candidates as $candidate) {
156	$idx = array_search($candidate, $headers, true);
157	if ($idx !== false) {
158	return (int)$idx;
159	}
160	}
161	return -1;
162	}
163
164	/**
165	* @param array<int, string> $columns
166	* @return array<string, string>
167	*/
168	function mapImportRowWithoutHeaders($columns) {
169	$columns = array_values($columns);
170	if (count($columns) === 7) {
171	return array(
172	'from_url' => trim((string)$columns[0]),
173	'status' => trim((string)$columns[1]),
174	'type' => trim((string)$columns[2]),
175	'to_url' => trim((string)$columns[3]),
176	'wp_type' => trim((string)$columns[4]),
177	'engine' => trim((string)$columns[5]),
178	'code' => trim((string)$columns[6]),
179	);
180	}
181	if (count($columns) === 6) {
182	return array(
183	'from_url' => trim((string)$columns[0]),
184	'status' => trim((string)$columns[1]),
185	'type' => trim((string)$columns[2]),
186	'to_url' => trim((string)$columns[3]),
187	'wp_type' => trim((string)$columns[4]),
188	'engine' => trim((string)$columns[5]),
189	);
190	}
191	if (count($columns) === 5) {
192	return array(
193	'from_url' => trim((string)$columns[0]),
194	'status' => trim((string)$columns[1]),
195	'type' => trim((string)$columns[2]),
196	'to_url' => trim((string)$columns[3]),
197	'wp_type' => trim((string)$columns[4]),
198	);
199	}
200	if (count($columns) === 2) {
201	return array(
202	'from_url' => trim((string)$columns[0]),
203	'to_url' => trim((string)$columns[1]),
204	);
205	}
206	return array('error' => sprintf(__('Invalid CSV format. %d columns found but 2, 5, 6, or 7 expected.', '404-solution'), count($columns)));
207	}
208
209	/**
210	* @param resource $fileHandle
211	* @return string
212	*/
213	function detectCsvDelimiterFromFile($fileHandle) {
214	while (($line = fgets($fileHandle)) !== false) {
215	if (trim($line) === '') {
216	continue;
217	}
218	$comma = count(str_getcsv($line, ',', '"', '\\'));
219	$semicolon = count(str_getcsv($line, ';', '"', '\\'));
220	$tab = count(str_getcsv($line, "\t", '"', '\\'));
221
222	if ($semicolon > $comma && $semicolon >= $tab) {
223	return ';';
224	}
225	if ($tab > $comma && $tab > $semicolon) {
226	return "\t";
227	}
228	return ',';
229	}
230	return ',';
231	}
232	}
233