HTMLParser.h 7.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204
  1. /*
  2. * Copyright (c) 2020-2022, Andreas Kling <kling@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #pragma once
  7. #include <LibJS/Heap/Cell.h>
  8. #include <LibWeb/DOM/Node.h>
  9. #include <LibWeb/HTML/Parser/HTMLTokenizer.h>
  10. #include <LibWeb/HTML/Parser/ListOfActiveFormattingElements.h>
  11. #include <LibWeb/HTML/Parser/StackOfOpenElements.h>
  12. namespace Web::HTML {
  13. #define ENUMERATE_INSERTION_MODES \
  14. __ENUMERATE_INSERTION_MODE(Initial) \
  15. __ENUMERATE_INSERTION_MODE(BeforeHTML) \
  16. __ENUMERATE_INSERTION_MODE(BeforeHead) \
  17. __ENUMERATE_INSERTION_MODE(InHead) \
  18. __ENUMERATE_INSERTION_MODE(InHeadNoscript) \
  19. __ENUMERATE_INSERTION_MODE(AfterHead) \
  20. __ENUMERATE_INSERTION_MODE(InBody) \
  21. __ENUMERATE_INSERTION_MODE(Text) \
  22. __ENUMERATE_INSERTION_MODE(InTable) \
  23. __ENUMERATE_INSERTION_MODE(InTableText) \
  24. __ENUMERATE_INSERTION_MODE(InCaption) \
  25. __ENUMERATE_INSERTION_MODE(InColumnGroup) \
  26. __ENUMERATE_INSERTION_MODE(InTableBody) \
  27. __ENUMERATE_INSERTION_MODE(InRow) \
  28. __ENUMERATE_INSERTION_MODE(InCell) \
  29. __ENUMERATE_INSERTION_MODE(InSelect) \
  30. __ENUMERATE_INSERTION_MODE(InSelectInTable) \
  31. __ENUMERATE_INSERTION_MODE(InTemplate) \
  32. __ENUMERATE_INSERTION_MODE(AfterBody) \
  33. __ENUMERATE_INSERTION_MODE(InFrameset) \
  34. __ENUMERATE_INSERTION_MODE(AfterFrameset) \
  35. __ENUMERATE_INSERTION_MODE(AfterAfterBody) \
  36. __ENUMERATE_INSERTION_MODE(AfterAfterFrameset)
  37. class HTMLParser final : public JS::Cell {
  38. JS_CELL(HTMLParser, JS::Cell);
  39. friend class HTMLTokenizer;
  40. public:
  41. ~HTMLParser();
  42. static JS::NonnullGCPtr<HTMLParser> create_for_scripting(DOM::Document&);
  43. static JS::NonnullGCPtr<HTMLParser> create_with_uncertain_encoding(DOM::Document&, ByteBuffer const& input);
  44. static JS::NonnullGCPtr<HTMLParser> create(DOM::Document&, StringView input, DeprecatedString const& encoding);
  45. void run();
  46. void run(const AK::URL&);
  47. DOM::Document& document();
  48. static Vector<JS::Handle<DOM::Node>> parse_html_fragment(DOM::Element& context_element, StringView);
  49. static DeprecatedString serialize_html_fragment(DOM::Node const& node);
  50. enum class InsertionMode {
  51. #define __ENUMERATE_INSERTION_MODE(mode) mode,
  52. ENUMERATE_INSERTION_MODES
  53. #undef __ENUMERATE_INSERTION_MODE
  54. };
  55. InsertionMode insertion_mode() const { return m_insertion_mode; }
  56. static bool is_special_tag(DeprecatedFlyString const& tag_name, DeprecatedFlyString const& namespace_);
  57. HTMLTokenizer& tokenizer() { return m_tokenizer; }
  58. // https://html.spec.whatwg.org/multipage/parsing.html#abort-a-parser
  59. void abort();
  60. bool aborted() const { return m_aborted; }
  61. bool stopped() const { return m_stop_parsing; }
  62. size_t script_nesting_level() const { return m_script_nesting_level; }
  63. private:
  64. HTMLParser(DOM::Document&, StringView input, DeprecatedString const& encoding);
  65. HTMLParser(DOM::Document&);
  66. virtual void visit_edges(Cell::Visitor&) override;
  67. char const* insertion_mode_name() const;
  68. DOM::QuirksMode which_quirks_mode(HTMLToken const&) const;
  69. void handle_initial(HTMLToken&);
  70. void handle_before_html(HTMLToken&);
  71. void handle_before_head(HTMLToken&);
  72. void handle_in_head(HTMLToken&);
  73. void handle_in_head_noscript(HTMLToken&);
  74. void handle_after_head(HTMLToken&);
  75. void handle_in_body(HTMLToken&);
  76. void handle_after_body(HTMLToken&);
  77. void handle_after_after_body(HTMLToken&);
  78. void handle_text(HTMLToken&);
  79. void handle_in_table(HTMLToken&);
  80. void handle_in_table_body(HTMLToken&);
  81. void handle_in_row(HTMLToken&);
  82. void handle_in_cell(HTMLToken&);
  83. void handle_in_table_text(HTMLToken&);
  84. void handle_in_select_in_table(HTMLToken&);
  85. void handle_in_select(HTMLToken&);
  86. void handle_in_caption(HTMLToken&);
  87. void handle_in_column_group(HTMLToken&);
  88. void handle_in_template(HTMLToken&);
  89. void handle_in_frameset(HTMLToken&);
  90. void handle_after_frameset(HTMLToken&);
  91. void handle_after_after_frameset(HTMLToken&);
  92. void the_end();
  93. void stop_parsing() { m_stop_parsing = true; }
  94. void generate_implied_end_tags(DeprecatedFlyString const& exception = {});
  95. void generate_all_implied_end_tags_thoroughly();
  96. JS::NonnullGCPtr<DOM::Element> create_element_for(HTMLToken const&, DeprecatedFlyString const& namespace_, DOM::Node& intended_parent);
  97. struct AdjustedInsertionLocation {
  98. JS::GCPtr<DOM::Node> parent;
  99. JS::GCPtr<DOM::Node> insert_before_sibling;
  100. };
  101. AdjustedInsertionLocation find_appropriate_place_for_inserting_node(JS::GCPtr<DOM::Element> override_target = nullptr);
  102. DOM::Text* find_character_insertion_node();
  103. void flush_character_insertions();
  104. JS::NonnullGCPtr<DOM::Element> insert_foreign_element(HTMLToken const&, DeprecatedFlyString const&);
  105. JS::NonnullGCPtr<DOM::Element> insert_html_element(HTMLToken const&);
  106. DOM::Element& current_node();
  107. DOM::Element& adjusted_current_node();
  108. DOM::Element& node_before_current_node();
  109. void insert_character(u32 data);
  110. void insert_comment(HTMLToken&);
  111. void reconstruct_the_active_formatting_elements();
  112. void close_a_p_element();
  113. void process_using_the_rules_for(InsertionMode, HTMLToken&);
  114. void process_using_the_rules_for_foreign_content(HTMLToken&);
  115. void parse_generic_raw_text_element(HTMLToken&);
  116. void increment_script_nesting_level();
  117. void decrement_script_nesting_level();
  118. void reset_the_insertion_mode_appropriately();
  119. void adjust_mathml_attributes(HTMLToken&);
  120. void adjust_svg_tag_names(HTMLToken&);
  121. void adjust_svg_attributes(HTMLToken&);
  122. void adjust_foreign_attributes(HTMLToken&);
  123. enum AdoptionAgencyAlgorithmOutcome {
  124. DoNothing,
  125. RunAnyOtherEndTagSteps,
  126. };
  127. AdoptionAgencyAlgorithmOutcome run_the_adoption_agency_algorithm(HTMLToken&);
  128. void clear_the_stack_back_to_a_table_context();
  129. void clear_the_stack_back_to_a_table_body_context();
  130. void clear_the_stack_back_to_a_table_row_context();
  131. void close_the_cell();
  132. InsertionMode m_insertion_mode { InsertionMode::Initial };
  133. InsertionMode m_original_insertion_mode { InsertionMode::Initial };
  134. StackOfOpenElements m_stack_of_open_elements;
  135. Vector<InsertionMode> m_stack_of_template_insertion_modes;
  136. ListOfActiveFormattingElements m_list_of_active_formatting_elements;
  137. HTMLTokenizer m_tokenizer;
  138. bool m_foster_parenting { false };
  139. bool m_frameset_ok { true };
  140. bool m_parsing_fragment { false };
  141. // https://html.spec.whatwg.org/multipage/parsing.html#scripting-flag
  142. // The scripting flag is set to "enabled" if scripting was enabled for the Document with which the parser is associated when the parser was created, and "disabled" otherwise.
  143. bool m_scripting_enabled { true };
  144. bool m_invoked_via_document_write { false };
  145. bool m_aborted { false };
  146. bool m_parser_pause_flag { false };
  147. bool m_stop_parsing { false };
  148. size_t m_script_nesting_level { 0 };
  149. JS::Realm& realm();
  150. JS::GCPtr<DOM::Document> m_document;
  151. JS::GCPtr<HTMLHeadElement> m_head_element;
  152. JS::GCPtr<HTMLFormElement> m_form_element;
  153. JS::GCPtr<DOM::Element> m_context_element;
  154. Vector<HTMLToken> m_pending_table_character_tokens;
  155. JS::GCPtr<DOM::Text> m_character_insertion_node;
  156. StringBuilder m_character_insertion_builder;
  157. };
  158. RefPtr<CSS::StyleValue> parse_dimension_value(StringView);
  159. RefPtr<CSS::StyleValue> parse_nonzero_dimension_value(StringView);
  160. }