Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
87.60% |
113 / 129 |
|
72.73% |
8 / 11 |
CRAP | |
0.00% |
0 / 1 |
| ReferenceStack | |
87.60% |
113 / 129 |
|
72.73% |
8 / 11 |
57.16 | |
0.00% |
0 / 1 |
| pushInvalidRef | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| pushRef | |
96.72% |
59 / 61 |
|
0.00% |
0 / 1 |
19 | |||
| rollbackRefs | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
4 | |||
| rollbackRef | |
81.58% |
31 / 38 |
|
0.00% |
0 / 1 |
18.81 | |||
| popGroup | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| hasGroup | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getGroups | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
3 | |||
| getGroupRefs | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| resolveFollow | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| listDefinedRef | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| nextRefSequence | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace Cite; |
| 4 | |
| 5 | use LogicException; |
| 6 | use MediaWiki\Parser\StripState; |
| 7 | |
| 8 | /** |
| 9 | * Encapsulates most of Cite state during parsing. This includes metadata about each ref tag, |
| 10 | * and a rollback stack to correct confusion caused by lost context when `{{#tag` is used. |
| 11 | * |
| 12 | * @license GPL-2.0-or-later |
| 13 | */ |
| 14 | class ReferenceStack { |
| 15 | |
| 16 | /** |
| 17 | * Data structure representing all <ref> tags parsed so far, indexed by group name (an empty |
| 18 | * string for the default group) and reference name. |
| 19 | * |
| 20 | * References without a name get a numeric index, starting from 0. Conflicts are avoided by |
| 21 | * disallowing numeric names (e.g. <ref name="1">) in {@see Validator::validateRef}. |
| 22 | * |
| 23 | * @var array<string,array<string|int,ReferenceStackItem>> |
| 24 | */ |
| 25 | private array $refs = []; |
| 26 | |
| 27 | /** |
| 28 | * Auto-incrementing sequence number for all <ref>, no matter which group |
| 29 | */ |
| 30 | private int $refSequence = 0; |
| 31 | |
| 32 | /** @var int[] Counter for the number of refs in each group */ |
| 33 | private array $groupRefSequence = []; |
| 34 | |
| 35 | /** |
| 36 | * <ref> call stack |
| 37 | * Used to cleanup out of sequence ref calls created by #tag |
| 38 | * See description of function rollbackRef. |
| 39 | * |
| 40 | * @var (array|false)[] |
| 41 | * @phan-var array<array{0:string,1:int,2:string,3:?string,4:?string,5:?string,6:array}|false> |
| 42 | */ |
| 43 | private array $refCallStack = []; |
| 44 | |
| 45 | private const ACTION_ASSIGN = 'assign'; |
| 46 | private const ACTION_INCREMENT = 'increment'; |
| 47 | private const ACTION_NEW_FROM_PLACEHOLDER = 'new-from-placeholder'; |
| 48 | private const ACTION_NEW = 'new'; |
| 49 | |
| 50 | /** |
| 51 | * Leave a mark in the stack which matches an invalid ref tag. |
| 52 | */ |
| 53 | public function pushInvalidRef(): void { |
| 54 | $this->refCallStack[] = false; |
| 55 | } |
| 56 | |
| 57 | /** |
| 58 | * Populate $this->refs and $this->refCallStack based on input and arguments to <ref> |
| 59 | * |
| 60 | * @param StripState $stripState |
| 61 | * @param ?string $text Content from the <ref> tag |
| 62 | * @param string[] $argv |
| 63 | * @param string $group |
| 64 | * @param ?string $name |
| 65 | * @param ?string $extends |
| 66 | * @param ?string $follow Guaranteed to not be a numeric string |
| 67 | * @param ?string $dir ref direction |
| 68 | * |
| 69 | * @return ?ReferenceStackItem ref structure, or null if no footnote marker should be rendered |
| 70 | */ |
| 71 | public function pushRef( |
| 72 | StripState $stripState, |
| 73 | ?string $text, |
| 74 | array $argv, |
| 75 | string $group, |
| 76 | ?string $name, |
| 77 | ?string $extends, |
| 78 | ?string $follow, |
| 79 | ?string $dir |
| 80 | ): ?ReferenceStackItem { |
| 81 | $this->refs[$group] ??= []; |
| 82 | $this->groupRefSequence[$group] ??= 0; |
| 83 | |
| 84 | $ref = new ReferenceStackItem(); |
| 85 | $ref->count = 1; |
| 86 | $ref->dir = $dir; |
| 87 | // TODO: Read from this group field or deprecate it. |
| 88 | $ref->group = $group; |
| 89 | $ref->name = $name; |
| 90 | $ref->text = $text; |
| 91 | |
| 92 | if ( $follow ) { |
| 93 | if ( !isset( $this->refs[$group][$follow] ) ) { |
| 94 | // Mark an incomplete follow="…" as such. This is valid e.g. in the Page:… namespace |
| 95 | // on Wikisource. |
| 96 | $ref->follow = $follow; |
| 97 | $ref->key = $this->nextRefSequence(); |
| 98 | $this->refs[$group][] = $ref; |
| 99 | $this->refCallStack[] = [ self::ACTION_NEW, $ref->key, $group, $name, $text, $argv ]; |
| 100 | } elseif ( $text !== null ) { |
| 101 | // We know the parent already, so just perform the follow="…" and bail out |
| 102 | $this->resolveFollow( $group, $follow, $text ); |
| 103 | } |
| 104 | // A follow="…" never gets its own footnote marker |
| 105 | return null; |
| 106 | } |
| 107 | |
| 108 | if ( !$name ) { |
| 109 | // This is an anonymous reference, which will be given a numeric index. |
| 110 | $this->refs[$group][] = &$ref; |
| 111 | $ref->key = $this->nextRefSequence(); |
| 112 | $action = self::ACTION_NEW; |
| 113 | } elseif ( !isset( $this->refs[$group][$name] ) ) { |
| 114 | // Valid key with first occurrence |
| 115 | $this->refs[$group][$name] = &$ref; |
| 116 | $ref->key = $this->nextRefSequence(); |
| 117 | $action = self::ACTION_NEW; |
| 118 | } elseif ( $this->refs[$group][$name]->placeholder ) { |
| 119 | // Populate a placeholder. |
| 120 | $ref->extendsCount = $this->refs[$group][$name]->extendsCount; |
| 121 | $ref->key = $this->nextRefSequence(); |
| 122 | $ref->number = $this->refs[$group][$name]->number; |
| 123 | $this->refs[$group][$name] =& $ref; |
| 124 | $action = self::ACTION_NEW_FROM_PLACEHOLDER; |
| 125 | } else { |
| 126 | // Change an existing entry. |
| 127 | $ref = &$this->refs[$group][$name]; |
| 128 | $ref->count++; |
| 129 | |
| 130 | if ( $ref->dir && $dir && $ref->dir !== $dir ) { |
| 131 | $ref->warnings[] = [ 'cite_error_ref_conflicting_dir', $name ]; |
| 132 | } |
| 133 | |
| 134 | if ( $ref->text === null && $text !== null ) { |
| 135 | // If no text was set before, use this text |
| 136 | $ref->text = $text; |
| 137 | // Use the dir parameter only from the full definition of a named ref tag |
| 138 | $ref->dir = $dir; |
| 139 | $action = self::ACTION_ASSIGN; |
| 140 | } else { |
| 141 | if ( $text !== null |
| 142 | // T205803 different strip markers might hide the same text |
| 143 | && $stripState->unstripBoth( $text ) |
| 144 | !== $stripState->unstripBoth( $ref->text ) |
| 145 | ) { |
| 146 | // two refs with same name and different text |
| 147 | $ref->warnings[] = [ 'cite_error_references_duplicate_key', $name ]; |
| 148 | } |
| 149 | $action = self::ACTION_INCREMENT; |
| 150 | } |
| 151 | } |
| 152 | |
| 153 | $ref->number ??= ++$this->groupRefSequence[$group]; |
| 154 | |
| 155 | // Do not mess with a known parent a second time |
| 156 | if ( $extends && $ref->extendsIndex === null ) { |
| 157 | $parentRef =& $this->refs[$group][$extends]; |
| 158 | if ( $parentRef === null ) { |
| 159 | // Create a new placeholder and give it the current sequence number. |
| 160 | $parentRef = new ReferenceStackItem(); |
| 161 | $parentRef->name = $extends; |
| 162 | $parentRef->number = $ref->number; |
| 163 | $parentRef->placeholder = true; |
| 164 | } else { |
| 165 | $ref->number = $parentRef->number; |
| 166 | // Roll back the group sequence number. |
| 167 | --$this->groupRefSequence[$group]; |
| 168 | } |
| 169 | $parentRef->extendsCount ??= 0; |
| 170 | $ref->extends = $extends; |
| 171 | $ref->extendsIndex = ++$parentRef->extendsCount; |
| 172 | } elseif ( $extends && $ref->extends !== $extends ) { |
| 173 | // TODO: Change the error message to talk about "conflicting content or parent"? |
| 174 | $ref->warnings[] = [ 'cite_error_references_duplicate_key', $name ]; |
| 175 | } |
| 176 | |
| 177 | $this->refCallStack[] = [ $action, $ref->key, $group, $name, $text, $argv ]; |
| 178 | return $ref; |
| 179 | } |
| 180 | |
| 181 | /** |
| 182 | * Undo the changes made by the last $count ref tags. This is used when we discover that the |
| 183 | * last few tags were actually inside of a references tag. |
| 184 | * |
| 185 | * @param int $count |
| 186 | * |
| 187 | * @return array[] Refs to restore under the correct context, as a list of [ $text, $argv ] |
| 188 | * @phan-return array<array{0:?string,1:array}> |
| 189 | */ |
| 190 | public function rollbackRefs( int $count ): array { |
| 191 | $redoStack = []; |
| 192 | while ( $count-- && $this->refCallStack ) { |
| 193 | $call = array_pop( $this->refCallStack ); |
| 194 | if ( $call ) { |
| 195 | // @phan-suppress-next-line PhanParamTooFewUnpack |
| 196 | $redoStack[] = $this->rollbackRef( ...$call ); |
| 197 | } |
| 198 | } |
| 199 | |
| 200 | // Drop unused rollbacks, this group is finished. |
| 201 | $this->refCallStack = []; |
| 202 | |
| 203 | return array_reverse( $redoStack ); |
| 204 | } |
| 205 | |
| 206 | /** |
| 207 | * Partially undoes the effect of calls to stack() |
| 208 | * |
| 209 | * The option to define <ref> within <references> makes the |
| 210 | * behavior of <ref> context dependent. This is normally fine |
| 211 | * but certain operations (especially #tag) lead to out-of-order |
| 212 | * parser evaluation with the <ref> tags being processed before |
| 213 | * their containing <reference> element is read. This leads to |
| 214 | * stack corruption that this function works to fix. |
| 215 | * |
| 216 | * This function is not a total rollback since some internal |
| 217 | * counters remain incremented. Doing so prevents accidentally |
| 218 | * corrupting certain links. |
| 219 | * |
| 220 | * @param string $action |
| 221 | * @param int $key Autoincrement counter for this ref. |
| 222 | * @param string $group |
| 223 | * @param ?string $name The name attribute passed in the ref tag. |
| 224 | * @param ?string $text |
| 225 | * @param array $argv |
| 226 | * |
| 227 | * @return array [ $text, $argv ] Ref redo item. |
| 228 | */ |
| 229 | private function rollbackRef( |
| 230 | string $action, |
| 231 | int $key, |
| 232 | string $group, |
| 233 | ?string $name, |
| 234 | ?string $text, |
| 235 | array $argv |
| 236 | ): array { |
| 237 | if ( !$this->hasGroup( $group ) ) { |
| 238 | throw new LogicException( "Cannot roll back ref with unknown group \"$group\"." ); |
| 239 | } |
| 240 | |
| 241 | $lookup = $name ?: null; |
| 242 | if ( $lookup === null ) { |
| 243 | // Find anonymous ref by key. |
| 244 | foreach ( $this->refs[$group] as $k => $v ) { |
| 245 | if ( $v->key === $key ) { |
| 246 | $lookup = $k; |
| 247 | break; |
| 248 | } |
| 249 | } |
| 250 | } |
| 251 | |
| 252 | // Obsessive sanity checks that the specified element exists. |
| 253 | if ( $lookup === null ) { |
| 254 | throw new LogicException( "Cannot roll back unknown ref by key $key." ); |
| 255 | } elseif ( !isset( $this->refs[$group][$lookup] ) ) { |
| 256 | throw new LogicException( "Cannot roll back missing named ref \"$lookup\"." ); |
| 257 | } elseif ( $this->refs[$group][$lookup]->key !== $key ) { |
| 258 | throw new LogicException( |
| 259 | "Cannot roll back corrupt named ref \"$lookup\" which should have had key $key." ); |
| 260 | } |
| 261 | $ref =& $this->refs[$group][$lookup]; |
| 262 | |
| 263 | switch ( $action ) { |
| 264 | case self::ACTION_NEW: |
| 265 | // Rollback the addition of new elements to the stack |
| 266 | unset( $this->refs[$group][$lookup] ); |
| 267 | if ( !$this->refs[$group] ) { |
| 268 | $this->popGroup( $group ); |
| 269 | } elseif ( isset( $this->groupRefSequence[$group] ) ) { |
| 270 | $this->groupRefSequence[$group]--; |
| 271 | } |
| 272 | if ( $ref->extends ) { |
| 273 | $this->refs[$group][$ref->extends]->extendsCount--; |
| 274 | } |
| 275 | break; |
| 276 | case self::ACTION_NEW_FROM_PLACEHOLDER: |
| 277 | $ref->placeholder = true; |
| 278 | $ref->count = 0; |
| 279 | break; |
| 280 | case self::ACTION_ASSIGN: |
| 281 | // Rollback assignment of text to pre-existing elements |
| 282 | $ref->text = null; |
| 283 | $ref->count--; |
| 284 | break; |
| 285 | case self::ACTION_INCREMENT: |
| 286 | // Rollback increase in named ref occurrences |
| 287 | $ref->count--; |
| 288 | break; |
| 289 | default: |
| 290 | throw new LogicException( "Unknown call stack action \"$action\"" ); |
| 291 | } |
| 292 | return [ $text, $argv ]; |
| 293 | } |
| 294 | |
| 295 | /** |
| 296 | * Clear state for a single group. |
| 297 | * |
| 298 | * @param string $group |
| 299 | * |
| 300 | * @return array<string|int,ReferenceStackItem> The references from the removed group |
| 301 | */ |
| 302 | public function popGroup( string $group ): array { |
| 303 | $refs = $this->getGroupRefs( $group ); |
| 304 | unset( $this->refs[$group] ); |
| 305 | unset( $this->groupRefSequence[$group] ); |
| 306 | return $refs; |
| 307 | } |
| 308 | |
| 309 | /** |
| 310 | * Returns true if the group exists and contains references. |
| 311 | */ |
| 312 | public function hasGroup( string $group ): bool { |
| 313 | return (bool)( $this->refs[$group] ?? false ); |
| 314 | } |
| 315 | |
| 316 | /** |
| 317 | * @return string[] List of group names that contain at least one reference |
| 318 | */ |
| 319 | public function getGroups(): array { |
| 320 | $groups = []; |
| 321 | foreach ( $this->refs as $group => $refs ) { |
| 322 | if ( $refs ) { |
| 323 | $groups[] = $group; |
| 324 | } |
| 325 | } |
| 326 | return $groups; |
| 327 | } |
| 328 | |
| 329 | /** |
| 330 | * Return all references for a group. |
| 331 | * |
| 332 | * @param string $group |
| 333 | * |
| 334 | * @return array<string|int,ReferenceStackItem> |
| 335 | */ |
| 336 | public function getGroupRefs( string $group ): array { |
| 337 | return $this->refs[$group] ?? []; |
| 338 | } |
| 339 | |
| 340 | private function resolveFollow( string $group, string $follow, string $text ): void { |
| 341 | $previousRef =& $this->refs[$group][$follow]; |
| 342 | $previousRef->text ??= ''; |
| 343 | $previousRef->text .= " $text"; |
| 344 | } |
| 345 | |
| 346 | public function listDefinedRef( string $group, string $name, string $text ): void { |
| 347 | $ref =& $this->refs[$group][$name]; |
| 348 | $ref ??= new ReferenceStackItem(); |
| 349 | $ref->placeholder = false; |
| 350 | if ( $ref->text === null ) { |
| 351 | $ref->text = $text; |
| 352 | } elseif ( $ref->text !== $text ) { |
| 353 | // two refs with same key and different content |
| 354 | $ref->warnings[] = [ 'cite_error_references_duplicate_key', $name ]; |
| 355 | } |
| 356 | } |
| 357 | |
| 358 | private function nextRefSequence(): int { |
| 359 | return ++$this->refSequence; |
| 360 | } |
| 361 | |
| 362 | } |