top-10 / includes / util / class-csv-helper.php

class-csv-helper.php in WebberZone Top 10 — Popular Posts 4.4.1, at includes/util/class-csv-helper.php

228 lines 7.8 KB

Raw Download Zip

1	<?php
2	/**
3	* CSV helper for import and export of count data.
4	*
5	* @package WebberZone\Top_Ten\Util
6	* @since 4.3.0
7	*/
8
9	namespace WebberZone\Top_Ten\Util;
10
11	use WebberZone\Top_Ten\Database;
12
13	// If this file is called directly, abort.
14	if ( ! defined( 'WPINC' ) ) {
15	die;
16	}
17
18	/**
19	* Shared CSV read/write logic used by both the admin Import/Export page
20	* and the WP-CLI `wp top10 counts export\|import` commands.
21	*
22	* @since 4.3.0
23	*/
24	class Csv_Helper {
25
26	/**
27	* Write count rows to an open file handle in the Top 10 CSV format.
28	*
29	* Writes UTF-8 BOM (for Excel compatibility), a header row, then one
30	* data row per result. The positional column layout is:
31	*
32	* Overall: Post ID, Visits, Blog ID[, URL]
33	* Daily: Post ID, Visits, Date, Blog ID[, URL]
34	*
35	* @since 4.3.0
36	*
37	* @param resource $fh Open file handle (writable).
38	* @param array $rows Result rows from the database. Each row
39	* must contain at least postnumber,
40	* cntaccess, and blog_id. Daily rows must
41	* also contain dp_date.
42	* @param bool $daily Whether rows are from the daily table.
43	* @param bool $include_urls Whether to append a "URL" column.
44	* @param bool $network_wide Whether this is a network-wide export
45	* (used for correct permalink resolution).
46	*/
47	public static function write_export_csv( $fh, array $rows, bool $daily, bool $include_urls = false, bool $network_wide = false ): void {
48	// UTF-8 BOM for Excel / spreadsheet compatibility.
49	fprintf( $fh, chr( 0xEF ) . chr( 0xBB ) . chr( 0xBF ) );
50
51	$header = array( 'Post ID', 'Visits' );
52	if ( $daily ) {
53	$header[] = 'Date';
54	}
55	$header[] = 'Blog ID';
56	if ( $include_urls ) {
57	$header[] = 'URL';
58	}
59
60	fputcsv( $fh, $header, ',', '"', '\\' );
61
62	$url_cache = array();
63	foreach ( $rows as $row ) {
64	$line = array( $row['postnumber'], $row['cntaccess'] );
65	if ( $daily ) {
66	$line[] = $row['dp_date'];
67	}
68	$line[] = $row['blog_id'];
69
70	if ( $include_urls ) {
71	$pid = (int) $row['postnumber'];
72	$bid = (int) $row['blog_id'];
73	if ( ! isset( $url_cache[ $bid ][ $pid ] ) ) {
74	$url_cache[ $bid ][ $pid ] = is_multisite() && $network_wide
75	? get_blog_permalink( $bid, $pid )
76	: get_permalink( $pid );
77	}
78	$line[] = $url_cache[ $bid ][ $pid ];
79	}
80
81	fputcsv( $fh, $line, ',', '"', '\\' );
82	}
83	}
84
85	/**
86	* Parse an import CSV file and return structured rows.
87	*
88	* Strips any UTF-8 BOM from the beginning of the file, reads the header
89	* to auto-detect whether the file targets the overall or daily table
90	* (by the presence of a "Date" column), then returns every data row in
91	* a canonical associative format.
92	*
93	* The caller is responsible for URL resolution, date normalisation,
94	* blog filtering, and database writes.
95	*
96	* @since 4.3.0
97	*
98	* @param string $file_path Path to the CSV file.
99	* @return array An associative array with the keys:
100	* - daily: bool Whether the file targets the daily table.
101	* - rows: array[] Parsed data rows. Each element is an
102	* associative array with the keys
103	* postnumber (int), cntaccess (int),
104	* blog_id (int), plus dp_date (string)
105	* when daily is true, and url (string)
106	* when an optional URL column is present.
107	* - total: int Total number of data rows.
108	*/
109	public static function parse_import_file( string $file_path ): array {
110	$handle = fopen( $file_path, 'r' ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fopen
111	if ( false === $handle ) {
112	return array(
113	'daily' => false,
114	'rows' => array(),
115	'total' => 0,
116	);
117	}
118
119	// Strip BOM if present.
120	$bom = fread( $handle, 3 ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fread
121	if ( ( chr( 0xEF ) . chr( 0xBB ) . chr( 0xBF ) ) !== $bom ) {
122	rewind( $handle );
123	}
124
125	$headers = fgetcsv( $handle, 0, ',', '"', '\\' );
126	if ( false === $headers ) {
127	fclose( $handle ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fclose
128	return array(
129	'daily' => false,
130	'rows' => array(),
131	'total' => 0,
132	);
133	}
134
135	$headers = array_map( 'trim', $headers );
136
137	// Auto-detect daily vs overall from presence of a "Date" column.
138	$daily = in_array( 'Date', $headers, true );
139	$col_post_id = 0;
140	$col_count = 1;
141	$col_date = $daily ? 2 : -1;
142	$col_blog_id = $daily ? 3 : 2;
143	$col_url = $daily ? 4 : 3;
144	$has_url = isset( $headers[ $col_url ] ) && 'URL' === $headers[ $col_url ];
145
146	$rows = array();
147	$total = 0;
148
149	while ( false !== ( $line = fgetcsv( $handle, 0, ',', '"', '\\' ) ) ) { // phpcs:ignore Generic.CodeAnalysis.AssignmentInCondition.FoundInWhileCondition
150	if ( 1 === count( $line ) && '' === $line[0] ) {
151	continue;
152	}
153
154	$row = array(
155	'postnumber' => isset( $line[ $col_post_id ] ) ? absint( $line[ $col_post_id ] ) : 0,
156	'cntaccess' => isset( $line[ $col_count ] ) ? absint( $line[ $col_count ] ) : 0,
157	'blog_id' => isset( $line[ $col_blog_id ] ) ? absint( $line[ $col_blog_id ] ) : get_current_blog_id(),
158	);
159
160	if ( $daily ) {
161	$row['dp_date'] = isset( $line[ $col_date ] ) ? trim( $line[ $col_date ] ) : '';
162	}
163	if ( $has_url && isset( $line[ $col_url ] ) ) {
164	$row['url'] = trim( $line[ $col_url ] );
165	}
166
167	$rows[] = $row;
168	++$total;
169	}
170
171	fclose( $handle ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fclose
172
173	return array(
174	'daily' => $daily,
175	'rows' => $rows,
176	'total' => $total,
177	);
178	}
179
180	/**
181	* Fetch raw count rows from the database for CSV export.
182	*
183	* Called by both the admin Import/Export page and the WP-CLI
184	* `wp top10 counts export` command.
185	*
186	* @since 4.3.0
187	*
188	* @param bool $daily Whether to query the daily table.
189	* @param bool $network_wide Whether to collect rows from all sites (multisite only).
190	* @param int $blog_id Specific blog ID (0 = current site). Only used when
191	* $network_wide is false.
192	* @param int $limit Max rows to return (0 = all).
193	* @return array Raw result rows (ARRAY_A).
194	*/
195	public static function fetch_export_data( bool $daily, bool $network_wide = false, int $blog_id = 0, int $limit = 0 ): array {
196	global $wpdb;
197
198	$table = Database::get_table( $daily );
199
200	if ( $network_wide && is_multisite() ) {
201	$results = array();
202	$sites = get_sites( array( 'number' => 1000 ) );
203	foreach ( $sites as $site ) {
204	switch_to_blog( (int) $site->blog_id );
205	// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
206	$sql = $wpdb->prepare( "SELECT * FROM `{$table}` WHERE blog_id = %d", (int) $site->blog_id );
207	$rows = $wpdb->get_results( $sql, ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.NotPrepared
208	restore_current_blog();
209	$results = array_merge( $results, is_array( $rows ) ? $rows : array() );
210	if ( $limit > 0 && count( $results ) >= $limit ) {
211	break;
212	}
213	}
214	return $limit > 0 ? array_slice( $results, 0, $limit ) : $results;
215	}
216
217	$bid = $blog_id > 0 ? $blog_id : get_current_blog_id();
218	// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
219	$sql = $wpdb->prepare( "SELECT * FROM `{$table}` WHERE blog_id = %d", $bid );
220	if ( $limit > 0 ) {
221	$sql .= $wpdb->prepare( ' LIMIT %d', $limit );
222	}
223
224	$rows = $wpdb->get_results( $sql, ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.NotPrepared
225	return is_array( $rows ) ? $rows : array();
226	}
227	}
228