HTMLParser.h 6.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189
  1. /*
  2. * Copyright (c) 2020, Andreas Kling <kling@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #pragma once
  7. #include <AK/NonnullRefPtrVector.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. RefPtr<DOM::Document> parse_html_document(StringView, const AK::URL&, const String& encoding);
  38. class HTMLParser : public RefCounted<HTMLParser> {
  39. friend class HTMLTokenizer;
  40. public:
  41. ~HTMLParser();
  42. static NonnullRefPtr<HTMLParser> create_for_scripting(DOM::Document&);
  43. static NonnullRefPtr<HTMLParser> create_with_uncertain_encoding(DOM::Document&, ByteBuffer const& input);
  44. static NonnullRefPtr<HTMLParser> create(DOM::Document&, StringView input, String const& encoding);
  45. void run();
  46. void run(const AK::URL&);
  47. DOM::Document& document();
  48. static NonnullRefPtrVector<DOM::Node> parse_html_fragment(DOM::Element& context_element, StringView);
  49. static String 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(const FlyString& tag_name, const FlyString& namespace_);
  57. HTMLTokenizer& tokenizer() { return m_tokenizer; }
  58. bool aborted() const { return m_aborted; }
  59. size_t script_nesting_level() const { return m_script_nesting_level; }
  60. private:
  61. HTMLParser(DOM::Document&, StringView input, const String& encoding);
  62. HTMLParser(DOM::Document&);
  63. const char* insertion_mode_name() const;
  64. DOM::QuirksMode which_quirks_mode(const HTMLToken&) const;
  65. void handle_initial(HTMLToken&);
  66. void handle_before_html(HTMLToken&);
  67. void handle_before_head(HTMLToken&);
  68. void handle_in_head(HTMLToken&);
  69. void handle_in_head_noscript(HTMLToken&);
  70. void handle_after_head(HTMLToken&);
  71. void handle_in_body(HTMLToken&);
  72. void handle_after_body(HTMLToken&);
  73. void handle_after_after_body(HTMLToken&);
  74. void handle_text(HTMLToken&);
  75. void handle_in_table(HTMLToken&);
  76. void handle_in_table_body(HTMLToken&);
  77. void handle_in_row(HTMLToken&);
  78. void handle_in_cell(HTMLToken&);
  79. void handle_in_table_text(HTMLToken&);
  80. void handle_in_select_in_table(HTMLToken&);
  81. void handle_in_select(HTMLToken&);
  82. void handle_in_caption(HTMLToken&);
  83. void handle_in_column_group(HTMLToken&);
  84. void handle_in_template(HTMLToken&);
  85. void handle_in_frameset(HTMLToken&);
  86. void handle_after_frameset(HTMLToken&);
  87. void handle_after_after_frameset(HTMLToken&);
  88. void the_end();
  89. void stop_parsing() { m_stop_parsing = true; }
  90. void generate_implied_end_tags(const FlyString& exception = {});
  91. void generate_all_implied_end_tags_thoroughly();
  92. NonnullRefPtr<DOM::Element> create_element_for(HTMLToken const&, FlyString const& namespace_, DOM::Node const& intended_parent);
  93. struct AdjustedInsertionLocation {
  94. RefPtr<DOM::Node> parent;
  95. RefPtr<DOM::Node> insert_before_sibling;
  96. };
  97. AdjustedInsertionLocation find_appropriate_place_for_inserting_node(RefPtr<DOM::Element> override_target = nullptr);
  98. DOM::Text* find_character_insertion_node();
  99. void flush_character_insertions();
  100. NonnullRefPtr<DOM::Element> insert_foreign_element(const HTMLToken&, const FlyString&);
  101. NonnullRefPtr<DOM::Element> insert_html_element(const HTMLToken&);
  102. DOM::Element& current_node();
  103. DOM::Element& adjusted_current_node();
  104. DOM::Element& node_before_current_node();
  105. void insert_character(u32 data);
  106. void insert_comment(HTMLToken&);
  107. void reconstruct_the_active_formatting_elements();
  108. void close_a_p_element();
  109. void process_using_the_rules_for(InsertionMode, HTMLToken&);
  110. void process_using_the_rules_for_foreign_content(HTMLToken&);
  111. void parse_generic_raw_text_element(HTMLToken&);
  112. void increment_script_nesting_level();
  113. void decrement_script_nesting_level();
  114. void reset_the_insertion_mode_appropriately();
  115. void adjust_mathml_attributes(HTMLToken&);
  116. void adjust_svg_tag_names(HTMLToken&);
  117. void adjust_svg_attributes(HTMLToken&);
  118. void adjust_foreign_attributes(HTMLToken&);
  119. enum AdoptionAgencyAlgorithmOutcome {
  120. DoNothing,
  121. RunAnyOtherEndTagSteps,
  122. };
  123. AdoptionAgencyAlgorithmOutcome run_the_adoption_agency_algorithm(HTMLToken&);
  124. void clear_the_stack_back_to_a_table_context();
  125. void clear_the_stack_back_to_a_table_body_context();
  126. void clear_the_stack_back_to_a_table_row_context();
  127. void close_the_cell();
  128. InsertionMode m_insertion_mode { InsertionMode::Initial };
  129. InsertionMode m_original_insertion_mode { InsertionMode::Initial };
  130. StackOfOpenElements m_stack_of_open_elements;
  131. Vector<InsertionMode> m_stack_of_template_insertion_modes;
  132. ListOfActiveFormattingElements m_list_of_active_formatting_elements;
  133. HTMLTokenizer m_tokenizer;
  134. bool m_foster_parenting { false };
  135. bool m_frameset_ok { true };
  136. bool m_parsing_fragment { false };
  137. bool m_scripting_enabled { true };
  138. bool m_invoked_via_document_write { false };
  139. bool m_aborted { false };
  140. bool m_parser_pause_flag { false };
  141. bool m_stop_parsing { false };
  142. size_t m_script_nesting_level { 0 };
  143. NonnullRefPtr<DOM::Document> m_document;
  144. RefPtr<HTMLHeadElement> m_head_element;
  145. RefPtr<HTMLFormElement> m_form_element;
  146. RefPtr<DOM::Element> m_context_element;
  147. Vector<HTMLToken> m_pending_table_character_tokens;
  148. RefPtr<DOM::Text> m_character_insertion_node;
  149. StringBuilder m_character_insertion_builder;
  150. };
  151. }