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

class-csv-helper.php in WebberZone Top 10 — Popular Posts 4.5.1, at includes/util/class-csv-helper.php

220 lines 7.5 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	* Column layout: Overall = Post ID, Visits, Blog ID[, URL]. Daily = Post ID, Visits,
30	* Date, Blog ID[, URL].
31	*
32	* @since 4.3.0
33	*
34	* @param resource $fh Open file handle (writable).
35	* @param array $rows Result rows from the database. Each row
36	* must contain at least postnumber,
37	* cntaccess, and blog_id. Daily rows must
38	* also contain dp_date.
39	* @param bool $daily Whether rows are from the daily table.
40	* @param bool $include_urls Whether to append a "URL" column.
41	* @param bool $network_wide Whether this is a network-wide export
42	* (used for correct permalink resolution).
43	*/
44	public static function write_export_csv( $fh, array $rows, bool $daily, bool $include_urls = false, bool $network_wide = false ): void {
45	// UTF-8 BOM for Excel / spreadsheet compatibility.
46	fprintf( $fh, chr( 0xEF ) . chr( 0xBB ) . chr( 0xBF ) );
47
48	$header = array( 'Post ID', 'Visits' );
49	if ( $daily ) {
50	$header[] = 'Date';
51	}
52	$header[] = 'Blog ID';
53	if ( $include_urls ) {
54	$header[] = 'URL';
55	}
56
57	fputcsv( $fh, $header, ',', '"', '\\' );
58
59	$url_cache = array();
60	foreach ( $rows as $row ) {
61	$line = array( $row['postnumber'], $row['cntaccess'] );
62	if ( $daily ) {
63	$line[] = $row['dp_date'];
64	}
65	$line[] = $row['blog_id'];
66
67	if ( $include_urls ) {
68	$pid = (int) $row['postnumber'];
69	$bid = (int) $row['blog_id'];
70	if ( ! isset( $url_cache[ $bid ][ $pid ] ) ) {
71	$url_cache[ $bid ][ $pid ] = is_multisite() && $network_wide
72	? get_blog_permalink( $bid, $pid )
73	: get_permalink( $pid );
74	}
75	$line[] = $url_cache[ $bid ][ $pid ];
76	}
77
78	fputcsv( $fh, $line, ',', '"', '\\' );
79	}
80	}
81
82	/**
83	* Parse an import CSV file and return structured rows.
84	*
85	* Auto-detects overall vs. daily table by the presence of a "Date" column. Caller
86	* handles URL resolution, date normalisation, blog filtering, and database writes.
87	*
88	* @since 4.3.0
89	*
90	* @param string $file_path Path to the CSV file.
91	* @return array An associative array with the keys:
92	* - daily: bool Whether the file targets the daily table.
93	* - rows: array[] Parsed data rows. Each element is an
94	* associative array with the keys
95	* postnumber (int), cntaccess (int),
96	* blog_id (int), plus dp_date (string)
97	* when daily is true, and url (string)
98	* when an optional URL column is present.
99	* - total: int Total number of data rows.
100	*/
101	public static function parse_import_file( string $file_path ): array {
102	$handle = fopen( $file_path, 'r' ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fopen
103	if ( false === $handle ) {
104	return array(
105	'daily' => false,
106	'rows' => array(),
107	'total' => 0,
108	);
109	}
110
111	// Strip BOM if present.
112	$bom = fread( $handle, 3 ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fread
113	if ( ( chr( 0xEF ) . chr( 0xBB ) . chr( 0xBF ) ) !== $bom ) {
114	rewind( $handle );
115	}
116
117	$headers = fgetcsv( $handle, 0, ',', '"', '\\' );
118	if ( false === $headers ) {
119	fclose( $handle ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fclose
120	return array(
121	'daily' => false,
122	'rows' => array(),
123	'total' => 0,
124	);
125	}
126
127	$headers = array_map( 'trim', $headers );
128
129	// Auto-detect daily vs overall from presence of a "Date" column.
130	$daily = in_array( 'Date', $headers, true );
131	$col_post_id = 0;
132	$col_count = 1;
133	$col_date = $daily ? 2 : -1;
134	$col_blog_id = $daily ? 3 : 2;
135	$col_url = $daily ? 4 : 3;
136	$has_url = isset( $headers[ $col_url ] ) && 'URL' === $headers[ $col_url ];
137
138	$rows = array();
139	$total = 0;
140
141	while ( false !== ( $line = fgetcsv( $handle, 0, ',', '"', '\\' ) ) ) { // phpcs:ignore Generic.CodeAnalysis.AssignmentInCondition.FoundInWhileCondition
142	if ( 1 === count( $line ) && '' === $line[0] ) {
143	continue;
144	}
145
146	$row = array(
147	'postnumber' => isset( $line[ $col_post_id ] ) ? absint( $line[ $col_post_id ] ) : 0,
148	'cntaccess' => isset( $line[ $col_count ] ) ? absint( $line[ $col_count ] ) : 0,
149	'blog_id' => isset( $line[ $col_blog_id ] ) ? absint( $line[ $col_blog_id ] ) : get_current_blog_id(),
150	);
151
152	if ( $daily ) {
153	$row['dp_date'] = isset( $line[ $col_date ] ) ? trim( $line[ $col_date ], " \t\n\r\0\x0B" ) : '';
154	}
155	if ( $has_url && isset( $line[ $col_url ] ) ) {
156	$row['url'] = trim( $line[ $col_url ], " \t\n\r\0\x0B" );
157	}
158
159	$rows[] = $row;
160	++$total;
161	}
162
163	fclose( $handle ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fclose
164
165	return array(
166	'daily' => $daily,
167	'rows' => $rows,
168	'total' => $total,
169	);
170	}
171
172	/**
173	* Fetch raw count rows from the database for CSV export.
174	*
175	* Called by both the admin Import/Export page and the WP-CLI
176	* `wp top10 counts export` command.
177	*
178	* @since 4.3.0
179	*
180	* @param bool $daily Whether to query the daily table.
181	* @param bool $network_wide Whether to collect rows from all sites (multisite only).
182	* @param int $blog_id Specific blog ID (0 = current site). Only used when
183	* $network_wide is false.
184	* @param int $limit Max rows to return (0 = all).
185	* @return array Raw result rows (ARRAY_A).
186	*/
187	public static function fetch_export_data( bool $daily, bool $network_wide = false, int $blog_id = 0, int $limit = 0 ): array {
188	global $wpdb;
189
190	$table = Database::get_table( $daily );
191
192	if ( $network_wide && is_multisite() ) {
193	$results = array();
194	$sites = get_sites( array( 'number' => 1000 ) );
195	foreach ( $sites as $site ) {
196	switch_to_blog( (int) $site->blog_id );
197	// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
198	$sql = $wpdb->prepare( "SELECT * FROM `{$table}` WHERE blog_id = %d", (int) $site->blog_id );
199	$rows = $wpdb->get_results( $sql, ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.NotPrepared
200	restore_current_blog();
201	$results = array_merge( $results, is_array( $rows ) ? $rows : array() );
202	if ( $limit > 0 && count( $results ) >= $limit ) {
203	break;
204	}
205	}
206	return $limit > 0 ? array_slice( $results, 0, $limit ) : $results;
207	}
208
209	$bid = $blog_id > 0 ? $blog_id : get_current_blog_id();
210	// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
211	$sql = $wpdb->prepare( "SELECT * FROM `{$table}` WHERE blog_id = %d", $bid );
212	if ( $limit > 0 ) {
213	$sql .= $wpdb->prepare( ' LIMIT %d', $limit );
214	}
215
216	$rows = $wpdb->get_results( $sql, ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.NotPrepared
217	return is_array( $rows ) ? $rows : array();
218	}
219	}
220