404-solution / includes / database / CreateTableIndexParser.php

CreateTableIndexParser.php in 404 Solution 4.3.3, at includes/database/CreateTableIndexParser.php

139 lines 5.7 KB

Raw Download Zip

1	<?php
2
3	if (!defined('ABSPATH')) {
4	exit;
5	}
6
7	/**
8	* Parses the KEY / UNIQUE KEY declarations out of the plugin's own
9	* create*Table.sql templates into structured index specs.
10	*
11	* This is the DDL-source half of the index picture: pure text in, structure
12	* out, with no database connection and no knowledge of what any engine
13	* currently reports. The live-metadata half, and the comparison that decides
14	* whether the two agree, belong to
15	* {@see ABJ_404_Solution_TableIndexDefinitions}, which is where they can be
16	* compared as one operation.
17	*
18	* The parser's governing rule, and the reason it refuses rather than guesses:
19	* a fragment it only PARTLY understands must yield nothing at all. Scraping the
20	* recognisable names out of an unrecognised fragment produces a SHORTER column
21	* list that looks like a complete definition, and the repair path would then
22	* rebuild a real index to that shorter shape -- turning a parse gap into
23	* deliberate data-structure damage. Every caller already reads an empty result
24	* as "cannot describe this index", never as "this index has no columns".
25	*/
26	class ABJ_404_Solution_CreateTableIndexParser {
27
28	/**
29	* Extract index specs from a CREATE TABLE statement (plugin SQL templates),
30	* keyed by index name exactly as the DDL spells it.
31	*
32	* Only plain KEY / UNIQUE KEY definitions are recognised. FULLTEXT and
33	* SPATIAL keys are deliberately not matched: the plugin ships none, and
34	* silently mis-parsing one into a plain key would let the repair path
35	* rebuild it as the wrong kind of index.
36	*
37	* @param string $createTableSql
38	* @return array<string, array{name: string, columns: string, unique: bool}>
39	*/
40	public static function fromCreateTableSql($createTableSql): array {
41	if (!is_string($createTableSql) \|\| $createTableSql === '') {
42	return array();
43	}
44
45	$matches = array();
46	preg_match_all('/^\\s(?:unique\\s+)?key\\s+.+?\\s$/im', $createTableSql, $matches);
47
48	$specsByName = array();
49	foreach ($matches[0] as $line) {
50	$spec = self::parseIndexDdlLine($line);
51	if (empty($spec) \|\| empty($spec['name'])) {
52	continue;
53	}
54	$specsByName[$spec['name']] = $spec;
55	}
56
57	return $specsByName;
58	}
59
60	/**
61	* Parse one index DDL line from our CREATE TABLE SQL into a structured spec.
62	*
63	* Accepts forms like:
64	* - KEY `name` (`col`(190), `other`)
65	* - UNIQUE KEY `name` (`col`)
66	* - KEY `name` (`col`) USING BTREE
67	*
68	* Returns null if the line doesn't look like a KEY/UNIQUE KEY definition.
69	*
70	* @param string $indexDDL
71	* @return array{name: string, columns: string, unique: bool}\|null
72	*/
73	public static function parseIndexDdlLine($indexDDL) {
74	$indexDDL = trim((string)$indexDDL);
75	// Tolerate a trailing comma -- the line-extracting regex pulls each
76	// KEY definition out as-is from the surrounding CREATE TABLE list,
77	// and any KEY that isn't the LAST one will end with a comma. Same
78	// canonical form either way.
79	$indexDDL = rtrim($indexDDL, ',');
80	$matches = array();
81	if (!preg_match('/^(unique\\s+)?key\\s+`?([^`\\s]+)`?\\s(\$.+\$)\\s(?:using\\s+\\w+)?\\s*$/i', $indexDDL, $matches)) {
82	return null;
83	}
84
85	return array(
86	'name' => $matches[2],
87	'columns' => $matches[3],
88	'unique' => !empty($matches[1]),
89	);
90	}
91
92	/**
93	* The ordered (column, prefix) list a DDL column fragment describes.
94	*
95	* Input is the parenthesised fragment a spec carries, e.g.
96	* "(`status`, `disabled`, `logshits`, `id`)" or "(`url`(190), `disabled`)".
97	* Backticks are required (every shipped create*Table.sql uses them, and
98	* DDLColumnParsingRobustnessTest enforces it), so a fragment written some
99	* other way yields an empty list -- which the repair path treats as
100	* "cannot describe this index", never as "this index has no columns".
101	*
102	* @param string $columnsSql
103	* @return array<int, array{column: string, prefix: int\|null}>
104	*/
105	public static function ddlColumnList($columnsSql): array {
106	$fragment = trim((string)$columnsSql);
107	$columns = array();
108	$matches = array();
109	preg_match_all('/`([^`]+)`\\s(?:\$\\s(\\d+)\\s*\$)?/', $fragment, $matches,
110	PREG_SET_ORDER);
111	foreach ($matches as $match) {
112	// A (0) prefix and no prefix at all are the same physical index, and
113	// the live reader already reports a Sub_part of 0 as "no prefix".
114	// Spelling it `url(0)` here while the live side spells it `url`
115	// makes two descriptions of one index compare as drift, and the
116	// repair path answers drift by rewriting the table. MySQL rejects a
117	// zero-length key part, so no create*Table.sql the plugin ships can
118	// reach this -- but the two sides of a comparison agreeing about
119	// what a value MEANS should not rest on the value never occurring.
120	$prefix = isset($match[2]) ? (int)$match[2] : null;
121	$columns[] = array(
122	'column' => strtolower($match[1]),
123	'prefix' => ($prefix === null \|\| $prefix <= 0) ? null : $prefix,
124	);
125	}
126
127	// Verify the whole fragment was accounted for, not just the parts that
128	// happened to match. See the class comment: a partial parse that looks
129	// complete is worse than no parse at all.
130	$remainder = preg_replace('/`[^`]+`\\s(?:\$\\s\\d+\\s*\$)?/', '', $fragment);
131	$remainder = trim((string)$remainder, " \t\n\r\0\x0B(),");
132	if ($remainder !== '') {
133	return array();
134	}
135
136	return $columns;
137	}
138	}
139