Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 90 |
|
0.00% |
0 / 9 |
CRAP | |
0.00% |
0 / 1 |
| ParserPipelineFactory | |
0.00% |
0 / 90 |
|
0.00% |
0 / 9 |
650 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| defaultOptions | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
6 | |||
| procNamesToProcs | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
12 | |||
| makePipeline | |
0.00% |
0 / 33 |
|
0.00% |
0 / 1 |
42 | |||
| getCacheKey | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
56 | |||
| parse | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| selectiveDOMUpdate | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| getPipeline | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
6 | |||
| returnPipeline | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Wt2Html; |
| 5 | |
| 6 | use Wikimedia\Assert\Assert; |
| 7 | use Wikimedia\Parsoid\Config\Env; |
| 8 | use Wikimedia\Parsoid\Core\InternalException; |
| 9 | use Wikimedia\Parsoid\Core\SelectiveUpdateData; |
| 10 | use Wikimedia\Parsoid\DOM\Document; |
| 11 | use Wikimedia\Parsoid\Utils\PHPUtils; |
| 12 | use Wikimedia\Parsoid\Utils\Utils; |
| 13 | use Wikimedia\Parsoid\Wt2Html\DOM\Handlers\AddAnnotationIds; |
| 14 | use Wikimedia\Parsoid\Wt2Html\DOM\Handlers\AddLinkAttributes; |
| 15 | use Wikimedia\Parsoid\Wt2Html\DOM\Handlers\CleanUp; |
| 16 | use Wikimedia\Parsoid\Wt2Html\DOM\Handlers\DedupeStyles; |
| 17 | use Wikimedia\Parsoid\Wt2Html\DOM\Handlers\DisplaySpace; |
| 18 | use Wikimedia\Parsoid\Wt2Html\DOM\Handlers\HandleLinkNeighbours; |
| 19 | use Wikimedia\Parsoid\Wt2Html\DOM\Handlers\Headings; |
| 20 | use Wikimedia\Parsoid\Wt2Html\DOM\Handlers\LiFixups; |
| 21 | use Wikimedia\Parsoid\Wt2Html\DOM\Handlers\TableFixups; |
| 22 | use Wikimedia\Parsoid\Wt2Html\DOM\Handlers\UnpackDOMFragments; |
| 23 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\AddMediaInfo; |
| 24 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\AddMetaData; |
| 25 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\AddRedLinks; |
| 26 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\ComputeDSR; |
| 27 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\ConvertOffsets; |
| 28 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\LangConverter; |
| 29 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\Linter; |
| 30 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\MarkFosteredContent; |
| 31 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\MigrateTemplateMarkerMetas; |
| 32 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\MigrateTrailingNLs; |
| 33 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\Normalize; |
| 34 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\ProcessEmbeddedDocs; |
| 35 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\ProcessTreeBuilderFixups; |
| 36 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\PWrap; |
| 37 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\RunExtensionProcessors; |
| 38 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\UpdateTemplateOutput; |
| 39 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\WrapAnnotations; |
| 40 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\WrapSections; |
| 41 | use Wikimedia\Parsoid\Wt2Html\DOM\Processors\WrapTemplates; |
| 42 | use Wikimedia\Parsoid\Wt2Html\TreeBuilder\TreeBuilderStage; |
| 43 | use Wikimedia\Parsoid\Wt2Html\TT\AttributeExpander; |
| 44 | use Wikimedia\Parsoid\Wt2Html\TT\BehaviorSwitchHandler; |
| 45 | use Wikimedia\Parsoid\Wt2Html\TT\DOMFragmentBuilder; |
| 46 | use Wikimedia\Parsoid\Wt2Html\TT\ExtensionHandler; |
| 47 | use Wikimedia\Parsoid\Wt2Html\TT\ExternalLinkHandler; |
| 48 | use Wikimedia\Parsoid\Wt2Html\TT\LanguageVariantHandler; |
| 49 | use Wikimedia\Parsoid\Wt2Html\TT\ListHandler; |
| 50 | use Wikimedia\Parsoid\Wt2Html\TT\OnlyInclude; |
| 51 | use Wikimedia\Parsoid\Wt2Html\TT\ParagraphWrapper; |
| 52 | use Wikimedia\Parsoid\Wt2Html\TT\PreHandler; |
| 53 | use Wikimedia\Parsoid\Wt2Html\TT\QuoteTransformer; |
| 54 | use Wikimedia\Parsoid\Wt2Html\TT\SanitizerHandler; |
| 55 | use Wikimedia\Parsoid\Wt2Html\TT\TemplateHandler; |
| 56 | use Wikimedia\Parsoid\Wt2Html\TT\TokenStreamPatcher; |
| 57 | use Wikimedia\Parsoid\Wt2Html\TT\WikiLinkHandler; |
| 58 | |
| 59 | /** |
| 60 | * This class assembles parser pipelines from parser stages |
| 61 | */ |
| 62 | class ParserPipelineFactory { |
| 63 | private static $globalPipelineId = 0; |
| 64 | |
| 65 | private const DOM_PROCESSOR_CONFIG = [ |
| 66 | 'addmetadata' => AddMetaData::class, |
| 67 | 'annwrap' => WrapAnnotations::class, |
| 68 | 'convertoffsets' => ConvertOffsets::class, |
| 69 | 'dsr' => ComputeDSR::class, |
| 70 | 'embedded-docs' => ProcessEmbeddedDocs::class, |
| 71 | 'extpp' => RunExtensionProcessors::class, |
| 72 | 'fostered' => MarkFosteredContent::class, |
| 73 | 'linter' => Linter::class, |
| 74 | 'lang-converter' => LangConverter::class, |
| 75 | 'media' => AddMediaInfo::class, |
| 76 | 'migrate-metas' => MigrateTemplateMarkerMetas::class, |
| 77 | 'migrate-nls' => MigrateTrailingNLs::class, |
| 78 | 'normalize' => Normalize::class, |
| 79 | 'process-fixups' => ProcessTreeBuilderFixups::class, |
| 80 | 'pwrap' => PWrap::class, |
| 81 | 'redlinks' => AddRedLinks::class, |
| 82 | 'sections' => WrapSections::class, // Don't process HTML in embedded attributes |
| 83 | 'tplwrap' => WrapTemplates::class, |
| 84 | 'update-template' => UpdateTemplateOutput::class, |
| 85 | 'ann-ids' => [ |
| 86 | 'name' => 'AddAnnotationIds', |
| 87 | 'handlers' => [ |
| 88 | [ 'nodeName' => 'meta', 'action' => [ AddAnnotationIds::class, 'handler' ] ] |
| 89 | ], |
| 90 | 'withAnnotations' => true |
| 91 | ], |
| 92 | 'linkneighbours+dom-unpack' => [ |
| 93 | 'name' => 'HandleLinkNeighbours,UnpackDOMFragments', |
| 94 | 'handlers' => [ |
| 95 | // Link prefixes and suffixes |
| 96 | [ 'nodeName' => 'a', 'action' => [ HandleLinkNeighbours::class, 'handler' ] ], |
| 97 | [ 'nodeName' => null, 'action' => [ UnpackDOMFragments::class, 'handler' ] ] |
| 98 | ] |
| 99 | ], |
| 100 | 'fixups' => [ |
| 101 | 'name' => 'MigrateTrailingCategories,TableFixups', |
| 102 | 'tplInfo' => true, |
| 103 | 'handlers' => [ |
| 104 | // 1. Move trailing categories in <li>s out of the list |
| 105 | [ 'nodeName' => 'li', 'action' => [ LiFixups::class, 'migrateTrailingSolTransparentLinks' ] ], |
| 106 | [ 'nodeName' => 'dt', 'action' => [ LiFixups::class, 'migrateTrailingSolTransparentLinks' ] ], |
| 107 | [ 'nodeName' => 'dd', 'action' => [ LiFixups::class, 'migrateTrailingSolTransparentLinks' ] ], |
| 108 | // 2. Fix up issues from templated table cells and table cell attributes |
| 109 | [ 'nodeName' => 'td', 'action' => [ TableFixups::class, 'handleTableCellTemplates' ] ], |
| 110 | [ 'nodeName' => 'th', 'action' => [ TableFixups::class, 'handleTableCellTemplates' ] ], |
| 111 | ] |
| 112 | ], |
| 113 | 'fixups+dedupe-styles' => [ |
| 114 | 'name' => 'MigrateTrailingCategories,TableFixups,DedupeStyles', |
| 115 | 'tplInfo' => true, |
| 116 | 'handlers' => [ |
| 117 | // 1. Move trailing categories in <li>s out of the list |
| 118 | [ 'nodeName' => 'li', 'action' => [ LiFixups::class, 'migrateTrailingSolTransparentLinks' ] ], |
| 119 | [ 'nodeName' => 'dt', 'action' => [ LiFixups::class, 'migrateTrailingSolTransparentLinks' ] ], |
| 120 | [ 'nodeName' => 'dd', 'action' => [ LiFixups::class, 'migrateTrailingSolTransparentLinks' ] ], |
| 121 | // 2. Fix up issues from templated table cells and table cell attributes |
| 122 | [ 'nodeName' => 'td', 'action' => [ TableFixups::class, 'handleTableCellTemplates' ] ], |
| 123 | [ 'nodeName' => 'th', 'action' => [ TableFixups::class, 'handleTableCellTemplates' ] ], |
| 124 | // 3. Deduplicate template styles |
| 125 | // (should run after dom-fragment expansion + after extension post-processors) |
| 126 | [ 'nodeName' => 'style', 'action' => [ DedupeStyles::class, 'dedupe' ] ] |
| 127 | ] |
| 128 | ], |
| 129 | // Strip marker metas -- removes left over marker metas (ex: metas |
| 130 | // nested in expanded tpl/extension output). |
| 131 | 'strip-metas' => [ |
| 132 | 'name' => 'CleanUp-stripMarkerMetas', |
| 133 | 'handlers' => [ |
| 134 | [ 'nodeName' => 'meta', 'action' => [ CleanUp::class, 'stripMarkerMetas' ] ] |
| 135 | ] |
| 136 | ], |
| 137 | 'displayspace+linkclasses' => [ |
| 138 | 'name' => 'DisplaySpace+AddLinkAttributes', |
| 139 | 'handlers' => [ |
| 140 | [ 'nodeName' => null, 'action' => [ DisplaySpace::class, 'leftHandler' ] ], |
| 141 | [ 'nodeName' => null, 'action' => [ DisplaySpace::class, 'rightHandler' ] ], |
| 142 | [ 'nodeName' => 'a', 'action' => [ AddLinkAttributes::class, 'handler' ] ] |
| 143 | ] |
| 144 | ], |
| 145 | 'gen-anchors' => [ |
| 146 | 'name' => 'Headings-genAnchors', |
| 147 | 'handlers' => [ |
| 148 | [ 'nodeName' => null, 'action' => [ Headings::class, 'genAnchors' ] ], |
| 149 | ] |
| 150 | ], |
| 151 | 'dedupe-heading-ids' => [ |
| 152 | 'name' => 'Headings-dedupeIds', |
| 153 | 'handlers' => [ |
| 154 | [ 'nodeName' => null, 'action' => [ Headings::class, 'dedupeHeadingIds' ] ] |
| 155 | ] |
| 156 | ], |
| 157 | 'heading-ids' => [ |
| 158 | 'name' => 'Headings-genAnchors', |
| 159 | 'handlers' => [ |
| 160 | [ 'nodeName' => null, 'action' => [ Headings::class, 'genAnchors' ] ], |
| 161 | [ 'nodeName' => null, 'action' => [ Headings::class, 'dedupeHeadingIds' ] ] |
| 162 | ] |
| 163 | ], |
| 164 | 'cleanup' => [ |
| 165 | 'name' => 'CleanUp-handleEmptyElts,CleanUp-cleanup', |
| 166 | 'tplInfo' => true, |
| 167 | 'handlers' => [ |
| 168 | // Strip empty elements from template content |
| 169 | [ 'nodeName' => null, 'action' => [ CleanUp::class, 'handleEmptyElements' ] ], |
| 170 | // Additional cleanup |
| 171 | [ 'nodeName' => null, 'action' => [ CleanUp::class, 'finalCleanup' ] ] |
| 172 | ] |
| 173 | ], |
| 174 | 'saveDP' => [ |
| 175 | 'name' => 'CleanUp-saveDataParsoid', |
| 176 | 'tplInfo' => true, |
| 177 | 'handlers' => [ |
| 178 | // Mark which data.parsoid's should be serialized into |
| 179 | // data-parsoid html attributes. |
| 180 | // Make this its own thing so that any changes to the DOM |
| 181 | // don't affect other handlers that run alongside it. |
| 182 | [ 'nodeName' => null, 'action' => [ CleanUp::class, 'saveDataParsoid' ] ] |
| 183 | ] |
| 184 | ] |
| 185 | ]; |
| 186 | |
| 187 | // NOTES about ordering / inclusion: |
| 188 | // |
| 189 | // media: |
| 190 | // This is run at all levels for now - gallery extension's "packed" mode |
| 191 | // would otherwise need a post-processing pass to scale media after it |
| 192 | // has been fetched. That introduces an ordering dependency that may |
| 193 | // or may not complicate things. |
| 194 | // migrate-metas: |
| 195 | // - Run this after 'pwrap' because it can add additional opportunities for |
| 196 | // meta migration which we will miss if we run this before p-wrapping. |
| 197 | // - We could potentially move this just before 'tplwrap' by seeing this |
| 198 | // as a preprocessing pass for that. But, we will have to update the pass |
| 199 | // to update DSR properties where required. |
| 200 | // - In summary, this can at most be moved before 'media' or after |
| 201 | // 'migrate-nls' without needing any other changes. |
| 202 | // dsr, tplwrap: |
| 203 | // DSR computation and template wrapping cannot be skipped for top-level content |
| 204 | // even if they are part of nested level pipelines, because such content might be |
| 205 | // embedded in attributes and they may need to be processed independently. |
| 206 | // |
| 207 | // Nested (non-top-level) pipelines can never include the following: |
| 208 | // - lang-converter, convertoffsets, dedupe-styles, cleanup, saveDP |
| 209 | // |
| 210 | // FIXME: Perhaps introduce a config flag in the processor config that |
| 211 | // verifies this property against a pipeline's 'toplevel' state. |
| 212 | public const NESTED_PIPELINE_DOM_TRANSFORMS = [ |
| 213 | 'fostered', 'process-fixups', 'normalize', 'pwrap', |
| 214 | 'media', 'migrate-metas', 'migrate-nls', 'dsr', 'tplwrap', |
| 215 | 'ann-ids', 'annwrap', 'linkneighbours+dom-unpack' |
| 216 | ]; |
| 217 | |
| 218 | // NOTES about ordering: |
| 219 | // lang-converter, redlinks: |
| 220 | // Language conversion and redlink marking are done here |
| 221 | // *before* we cleanup and save data-parsoid because they |
| 222 | // are also used in pb2pb/html2html passes, and we want to |
| 223 | // keep their input/output formats consistent. |
| 224 | public const FULL_PARSE_GLOBAL_DOM_TRANSFORMS = [ |
| 225 | // FIXME: It should be documented in the spec that an extension's |
| 226 | // wtDOMProcess handler is run once on the top level document. |
| 227 | 'extpp', |
| 228 | 'fixups+dedupe-styles', 'linter', 'strip-metas', |
| 229 | 'lang-converter', 'redlinks', 'displayspace+linkclasses', |
| 230 | // Benefits from running after determining which media are redlinks |
| 231 | 'heading-ids', |
| 232 | 'sections', 'convertoffsets', 'cleanup', |
| 233 | 'embedded-docs', |
| 234 | 'saveDP', 'addmetadata' |
| 235 | ]; |
| 236 | |
| 237 | // Skipping sections, addmetadata from the above pipeline |
| 238 | // |
| 239 | // FIXME: Skip extpp, linter, lang-converter, redlinks, heading-ids, convertoffsets, saveDP for now. |
| 240 | // This replicates behavior prior to this refactor. |
| 241 | public const FULL_PARSE_EMBEDDED_DOC_DOM_TRANSFORMS = [ |
| 242 | 'fixups+dedupe-styles', 'strip-metas', |
| 243 | 'displayspace+linkclasses', |
| 244 | 'cleanup', |
| 245 | // Need to run this recursively |
| 246 | 'embedded-docs', |
| 247 | // FIXME This means the data-* from embedded HTML fragments won't end up |
| 248 | // in the pagebundle. But, if we try to call this on those fragments, |
| 249 | // we get multiple calls to store embedded docs. So, we may need to |
| 250 | // write a custom traverser if we want these embedded data* objects |
| 251 | // in the pagebundle (this is not a regression since they weren't part |
| 252 | // of the pagebundle all this while anyway.) |
| 253 | /* 'saveDP' */ |
| 254 | ]; |
| 255 | |
| 256 | public const SELECTIVE_UPDATE_FRAGMENT_GLOBAL_DOM_TRANSFORMS = [ |
| 257 | 'extpp', // FIXME: this should be a different processor |
| 258 | 'fixups', 'strip-metas', 'redlinks', 'displayspace+linkclasses', |
| 259 | 'gen-anchors', 'convertoffsets', 'cleanup', |
| 260 | // FIXME: This will probably need some special-case code to first |
| 261 | // strip old metadata before adding fresh metadata. |
| 262 | 'addmetadata' |
| 263 | ]; |
| 264 | |
| 265 | public const SELECTIVE_UPDATE_GLOBAL_DOM_TRANSFORMS = [ |
| 266 | 'update-template', 'linter', 'lang-converter', /* FIXME: Are lang converters idempotent? */ |
| 267 | 'heading-ids', 'sections', 'saveDP' |
| 268 | ]; |
| 269 | |
| 270 | private static $stages = [ |
| 271 | "Tokenizer" => [ |
| 272 | "class" => PegTokenizer::class, |
| 273 | ], |
| 274 | "TokenTransform2" => [ |
| 275 | "class" => TokenTransformManager::class, |
| 276 | "transformers" => [ |
| 277 | OnlyInclude::class, |
| 278 | |
| 279 | TemplateHandler::class, |
| 280 | ExtensionHandler::class, |
| 281 | |
| 282 | // Expand attributes after templates to avoid expanding unused branches. |
| 283 | // No expansion of quotes, paragraphs etc in attributes, |
| 284 | // as with the legacy parser - up to end of TokenTransform2. |
| 285 | AttributeExpander::class, |
| 286 | |
| 287 | // now all attributes expanded to tokens or string |
| 288 | // more convenient after attribute expansion |
| 289 | WikiLinkHandler::class, |
| 290 | ExternalLinkHandler::class, |
| 291 | LanguageVariantHandler::class, |
| 292 | |
| 293 | // This converts dom-fragment-token tokens all the way to DOM |
| 294 | // and wraps them in DOMFragment wrapper tokens which will then |
| 295 | // get unpacked into the DOM by a dom-fragment unpacker. |
| 296 | DOMFragmentBuilder::class |
| 297 | ], |
| 298 | ], |
| 299 | "TokenTransform3" => [ |
| 300 | "class" => TokenTransformManager::class, |
| 301 | "transformers" => [ |
| 302 | TokenStreamPatcher::class, |
| 303 | // add <pre>s |
| 304 | PreHandler::class, |
| 305 | QuoteTransformer::class, |
| 306 | // add before transforms that depend on behavior switches |
| 307 | // examples: toc generation, edit sections |
| 308 | BehaviorSwitchHandler::class, |
| 309 | |
| 310 | ListHandler::class, |
| 311 | SanitizerHandler::class, |
| 312 | // Wrap tokens into paragraphs post-sanitization so that |
| 313 | // tags that converted to text by the sanitizer have a chance |
| 314 | // of getting wrapped into paragraphs. The sanitizer does not |
| 315 | // require the existence of p-tags for its functioning. |
| 316 | ParagraphWrapper::class |
| 317 | ], |
| 318 | ], |
| 319 | // Build a tree out of the fully processed token stream |
| 320 | "TreeBuilder" => [ |
| 321 | "class" => TreeBuilderStage::class, |
| 322 | ], |
| 323 | // DOM transformer for top-level documents. |
| 324 | // This performs a lot of post-processing of the DOM |
| 325 | // (Template wrapping, broken wikitext/html detection, etc.) |
| 326 | "FullParseDOMTransform" => [ |
| 327 | "class" => DOMPostProcessor::class, |
| 328 | "processors" => [ |
| 329 | self::NESTED_PIPELINE_DOM_TRANSFORMS, |
| 330 | self::FULL_PARSE_GLOBAL_DOM_TRANSFORMS |
| 331 | ], |
| 332 | ], |
| 333 | // DOM transformer for fragments of a top-level document |
| 334 | "NestedFragmentDOMTransform" => [ |
| 335 | "class" => DOMPostProcessor::class, |
| 336 | "processors" => self::NESTED_PIPELINE_DOM_TRANSFORMS |
| 337 | ], |
| 338 | // DOM transformations to run on attribute-embedded docs of the top level doc |
| 339 | "FullParseEmbeddedDocsDOMTransform" => [ |
| 340 | "class" => DOMPostProcessor::class, |
| 341 | "processors" => self::FULL_PARSE_EMBEDDED_DOC_DOM_TRANSFORMS |
| 342 | ], |
| 343 | // DOM transformer for fragments during selective updates. |
| 344 | // This may eventually become identical to NestedFrgmentDOMTransform, |
| 345 | // but at this time, it is unclear if that will materialize. |
| 346 | "SelectiveUpdateFragmentDOMTransform" => [ |
| 347 | "class" => DOMPostProcessor::class, |
| 348 | "processors" => [ |
| 349 | self::NESTED_PIPELINE_DOM_TRANSFORMS, |
| 350 | self::SELECTIVE_UPDATE_FRAGMENT_GLOBAL_DOM_TRANSFORMS |
| 351 | ], |
| 352 | ], |
| 353 | // DOM transformer for the top-level page during selective updates. |
| 354 | "SelectiveUpdateDOMTransform" => [ |
| 355 | // For use in the top-level of the selective-update pipeline |
| 356 | "class" => DOMPostProcessor::class, |
| 357 | "processors" => self::SELECTIVE_UPDATE_GLOBAL_DOM_TRANSFORMS |
| 358 | ] |
| 359 | ]; |
| 360 | |
| 361 | private static $pipelineRecipes = [ |
| 362 | // This pipeline takes wikitext as input and emits a fully |
| 363 | // processed DOM as output. This is the pipeline used for |
| 364 | // all top-level documents. |
| 365 | "fullparse-wikitext-to-dom" => [ |
| 366 | "alwaysToplevel" => true, |
| 367 | "outType" => "DOM", |
| 368 | "stages" => [ |
| 369 | "Tokenizer", "TokenTransform2", "TokenTransform3", "TreeBuilder", "FullParseDOMTransform" |
| 370 | ] |
| 371 | ], |
| 372 | |
| 373 | "fullparse-embedded-docs-dom-to-dom" => [ |
| 374 | "alwaysToplevel" => true, |
| 375 | "outType" => "DOM", |
| 376 | "stages" => [ "FullParseEmbeddedDocsDOMTransform" ] |
| 377 | ], |
| 378 | |
| 379 | // This pipeline takes a DOM and emits a fully processed DOM as output. |
| 380 | "selective-update-dom-to-dom" => [ |
| 381 | "alwaysToplevel" => true, |
| 382 | "outType" => "DOM", |
| 383 | "stages" => [ "SelectiveUpdateDOMTransform" ] |
| 384 | ], |
| 385 | |
| 386 | // This pipeline takes wikitext as input and emits a partially |
| 387 | // processed DOM as output. This is the pipeline used for processing |
| 388 | // page fragments to DOM in a selective page update context |
| 389 | // This is always toplevel because the wikitext being updated |
| 390 | // is found at the toplevel of the page. |
| 391 | "selective-update-fragment-wikitext-to-dom" => [ |
| 392 | "alwaysToplevel" => true, |
| 393 | "outType" => "DOM", |
| 394 | "stages" => [ |
| 395 | "Tokenizer", "TokenTransform2", "TokenTransform3", "TreeBuilder", "SelectiveUpdateFragmentDOMTransform" |
| 396 | ] |
| 397 | ], |
| 398 | |
| 399 | // This pipeline takes wikitext as input and emits a fully |
| 400 | // processed DOM as output. This is the pipeline used for |
| 401 | // wikitext fragments of a top-level document that should be |
| 402 | // processed to a DOM fragment. This pipeline doesn't run all |
| 403 | // of the DOM transformations in the DOMTransform pipeline. |
| 404 | // We will like use a specialized DOMTransform stage here. |
| 405 | "wikitext-to-fragment" => [ |
| 406 | // FIXME: This is known to be always *not* top-level |
| 407 | // We could use a different flag to lock these pipelines too. |
| 408 | "outType" => "DOM", |
| 409 | "stages" => [ |
| 410 | "Tokenizer", "TokenTransform2", "TokenTransform3", "TreeBuilder", "NestedFragmentDOMTransform" |
| 411 | ] |
| 412 | ], |
| 413 | |
| 414 | // This pipeline takes tokens from stage 2 and emits a DOM fragment |
| 415 | // as output - this runs the same DOM transforms as the 'wikitext-to-fragment' |
| 416 | // pipeline and will get a spcialized DOMTransform stage as above. |
| 417 | "expanded-tokens-to-fragment" => [ |
| 418 | "outType" => "DOM", |
| 419 | "stages" => [ "TokenTransform3", "TreeBuilder", "NestedFragmentDOMTransform" ] |
| 420 | ], |
| 421 | |
| 422 | // This pipeline takes wikitext as input and emits tokens that |
| 423 | // have had all templates, extensions, links, images processed |
| 424 | "wikitext-to-expanded-tokens" => [ |
| 425 | "outType" => "Tokens", |
| 426 | "stages" => [ "Tokenizer", "TokenTransform2" ] |
| 427 | ], |
| 428 | |
| 429 | // This pipeline takes tokens from the PEG tokenizer and emits |
| 430 | // tokens that have had all templates and extensions processed. |
| 431 | "peg-tokens-to-expanded-tokens" => [ |
| 432 | "outType" => "Tokens", |
| 433 | "stages" => [ "TokenTransform2" ] |
| 434 | ] |
| 435 | ]; |
| 436 | |
| 437 | private static $supportedOptions = [ |
| 438 | // If true, templates found in content will have its contents expanded |
| 439 | 'expandTemplates', |
| 440 | |
| 441 | // If true, indicates pipeline is processing the expanded content of a |
| 442 | // template or its arguments |
| 443 | 'inTemplate', |
| 444 | |
| 445 | // The extension tag that is being processed (Ex: ref, references) |
| 446 | // (in current usage, only used for native tag implementation) |
| 447 | 'extTag', |
| 448 | |
| 449 | // Extension-specific options |
| 450 | 'extTagOpts', |
| 451 | |
| 452 | // Content being parsed is used in an inline context |
| 453 | 'inlineContext', |
| 454 | |
| 455 | // Are we processing content of attributes? |
| 456 | // (in current usage, used for transcluded attr. keys/values) |
| 457 | 'attrExpansion', |
| 458 | ]; |
| 459 | |
| 460 | private array $pipelineCache = []; |
| 461 | |
| 462 | private Env $env; |
| 463 | |
| 464 | public function __construct( Env $env ) { |
| 465 | $this->env = $env; |
| 466 | } |
| 467 | |
| 468 | /** |
| 469 | * Default options processing |
| 470 | * |
| 471 | * @param array $options |
| 472 | * @return array |
| 473 | */ |
| 474 | private function defaultOptions( array $options ): array { |
| 475 | // default: not in a template |
| 476 | $options['inTemplate'] ??= false; |
| 477 | |
| 478 | // default: wrap templates |
| 479 | $options['expandTemplates'] ??= true; |
| 480 | |
| 481 | // Catch pipeline option typos |
| 482 | foreach ( $options as $k => $v ) { |
| 483 | Assert::invariant( |
| 484 | in_array( $k, self::$supportedOptions, true ), |
| 485 | 'Invalid cacheKey option: ' . $k |
| 486 | ); |
| 487 | } |
| 488 | |
| 489 | return $options; |
| 490 | } |
| 491 | |
| 492 | public static function procNamesToProcs( array $procNames ): array { |
| 493 | $processors = []; |
| 494 | foreach ( $procNames as $name ) { |
| 495 | $proc = self::DOM_PROCESSOR_CONFIG[$name]; |
| 496 | if ( !is_array( $proc ) ) { |
| 497 | $proc = [ |
| 498 | 'name' => Utils::stripNamespace( $proc ), |
| 499 | 'Processor' => $proc, |
| 500 | ]; |
| 501 | } |
| 502 | $proc['shortcut'] = $name; |
| 503 | $processors[] = $proc; |
| 504 | } |
| 505 | return $processors; |
| 506 | } |
| 507 | |
| 508 | /** |
| 509 | * Generic pipeline creation from the above recipes. |
| 510 | * |
| 511 | * @param string $type |
| 512 | * @param string $cacheKey |
| 513 | * @param array $options |
| 514 | * @return ParserPipeline |
| 515 | */ |
| 516 | private function makePipeline( |
| 517 | string $type, string $cacheKey, array $options |
| 518 | ): ParserPipeline { |
| 519 | if ( !isset( self::$pipelineRecipes[$type] ) ) { |
| 520 | throw new InternalException( 'Unsupported Pipeline: ' . $type ); |
| 521 | } |
| 522 | $recipe = self::$pipelineRecipes[$type]; |
| 523 | $pipeStages = []; |
| 524 | $prevStage = null; |
| 525 | $recipeStages = $recipe["stages"]; |
| 526 | |
| 527 | foreach ( $recipeStages as $stageId ) { |
| 528 | $stageData = self::$stages[$stageId]; |
| 529 | $stage = new $stageData["class"]( $this->env, $options, $stageId, $prevStage ); |
| 530 | if ( isset( $stageData["transformers"] ) ) { |
| 531 | foreach ( $stageData["transformers"] as $tName ) { |
| 532 | $stage->addTransformer( new $tName( $stage, $options ) ); |
| 533 | } |
| 534 | } elseif ( isset( $stageData["processors"] ) ) { |
| 535 | $processors = []; |
| 536 | array_walk_recursive( |
| 537 | $stageData["processors"], |
| 538 | static function ( $p ) use ( &$processors ) { |
| 539 | $processors[] = $p; |
| 540 | } |
| 541 | ); |
| 542 | $stage->registerProcessors( |
| 543 | self::procNamesToProcs( $processors ) |
| 544 | ); |
| 545 | } |
| 546 | $prevStage = $stage; |
| 547 | $pipeStages[] = $stage; |
| 548 | } |
| 549 | |
| 550 | return new ParserPipeline( |
| 551 | $recipe['alwaysToplevel'] ?? false, |
| 552 | $type, |
| 553 | $recipe["outType"], |
| 554 | $cacheKey, |
| 555 | $pipeStages, |
| 556 | $this->env |
| 557 | ); |
| 558 | } |
| 559 | |
| 560 | private function getCacheKey( string $cacheKey, array $options ): string { |
| 561 | if ( empty( $options['expandTemplates'] ) ) { |
| 562 | $cacheKey .= '::noExpand'; |
| 563 | } |
| 564 | if ( !empty( $options['inlineContext'] ) ) { |
| 565 | $cacheKey .= '::inlineContext'; |
| 566 | } |
| 567 | if ( !empty( $options['inTemplate'] ) ) { |
| 568 | $cacheKey .= '::inTemplate'; |
| 569 | } |
| 570 | if ( !empty( $options['attrExpansion'] ) ) { |
| 571 | $cacheKey .= '::attrExpansion'; |
| 572 | } |
| 573 | if ( isset( $options['extTag'] ) ) { |
| 574 | $cacheKey .= '::' . $options['extTag']; |
| 575 | // FIXME: This is not the best strategy. But, instead of |
| 576 | // premature complexity, let us see how extensions want to |
| 577 | // use this and then figure out what constraints are needed. |
| 578 | if ( isset( $options['extTagOpts'] ) ) { |
| 579 | $cacheKey .= '::' . PHPUtils::jsonEncode( $options['extTagOpts'] ); |
| 580 | } |
| 581 | } |
| 582 | return $cacheKey; |
| 583 | } |
| 584 | |
| 585 | public function parse( string $src ): Document { |
| 586 | $pipe = $this->getPipeline( 'fullparse-wikitext-to-dom' ); |
| 587 | $pipe->init( [ |
| 588 | 'frame' => $this->env->topFrame, |
| 589 | 'toFragment' => false, |
| 590 | ] ); |
| 591 | // Top-level doc parsing always start in SOL state |
| 592 | return $pipe->parseChunkily( $src, [ 'sol' => true ] )->ownerDocument; |
| 593 | } |
| 594 | |
| 595 | /** |
| 596 | * @param SelectiveUpdateData $selparData |
| 597 | * @param array $options Options for selective DOM update |
| 598 | * - mode: (string) One of "template", "section", "generic" |
| 599 | * For now, defaults to 'template', if absent |
| 600 | */ |
| 601 | public function selectiveDOMUpdate( SelectiveUpdateData $selparData, array $options = [] ): Document { |
| 602 | $pipe = $this->getPipeline( 'selective-update-dom-to-dom' ); |
| 603 | $pipe->init( [ |
| 604 | 'frame' => $this->env->topFrame, |
| 605 | 'toFragment' => false, |
| 606 | ] ); |
| 607 | return $pipe->selectiveParse( $selparData, $options ); |
| 608 | } |
| 609 | |
| 610 | /** |
| 611 | * Get a pipeline of a given type. Pipelines are cached as they are |
| 612 | * frequently created. |
| 613 | * |
| 614 | * @param string $type |
| 615 | * @param array $options These also determine the key under which the |
| 616 | * pipeline is cached for reuse. |
| 617 | * @return ParserPipeline |
| 618 | */ |
| 619 | public function getPipeline( |
| 620 | string $type, array $options = [] |
| 621 | ): ParserPipeline { |
| 622 | $options = $this->defaultOptions( $options ); |
| 623 | $cacheKey = $this->getCacheKey( $type, $options ); |
| 624 | |
| 625 | $this->pipelineCache[$cacheKey] ??= []; |
| 626 | |
| 627 | if ( $this->pipelineCache[$cacheKey] ) { |
| 628 | $pipe = array_pop( $this->pipelineCache[$cacheKey] ); |
| 629 | } else { |
| 630 | $pipe = $this->makePipeline( $type, $cacheKey, $options ); |
| 631 | } |
| 632 | |
| 633 | // Debugging aid: Assign unique id to the pipeline |
| 634 | $pipe->setPipelineId( self::$globalPipelineId++ ); |
| 635 | |
| 636 | return $pipe; |
| 637 | } |
| 638 | |
| 639 | /** |
| 640 | * Callback called by a pipeline at the end of its processing. Returns the |
| 641 | * pipeline to the cache. |
| 642 | * |
| 643 | * @param ParserPipeline $pipe |
| 644 | */ |
| 645 | public function returnPipeline( ParserPipeline $pipe ): void { |
| 646 | $cacheKey = $pipe->getCacheKey(); |
| 647 | $this->pipelineCache[$cacheKey] ??= []; |
| 648 | if ( count( $this->pipelineCache[$cacheKey] ) < 100 ) { |
| 649 | $this->pipelineCache[$cacheKey][] = $pipe; |
| 650 | } |
| 651 | } |
| 652 | } |