Serenity Operating System
1/*
2 * Copyright (c) 2020-2022, Andreas Kling <kling@serenityos.org>
3 *
4 * SPDX-License-Identifier: BSD-2-Clause
5 */
6
7#pragma once
8
9#include <LibJS/Heap/Cell.h>
10#include <LibWeb/DOM/Node.h>
11#include <LibWeb/HTML/Parser/HTMLTokenizer.h>
12#include <LibWeb/HTML/Parser/ListOfActiveFormattingElements.h>
13#include <LibWeb/HTML/Parser/StackOfOpenElements.h>
14
15namespace Web::HTML {
16
17#define ENUMERATE_INSERTION_MODES \
18 __ENUMERATE_INSERTION_MODE(Initial) \
19 __ENUMERATE_INSERTION_MODE(BeforeHTML) \
20 __ENUMERATE_INSERTION_MODE(BeforeHead) \
21 __ENUMERATE_INSERTION_MODE(InHead) \
22 __ENUMERATE_INSERTION_MODE(InHeadNoscript) \
23 __ENUMERATE_INSERTION_MODE(AfterHead) \
24 __ENUMERATE_INSERTION_MODE(InBody) \
25 __ENUMERATE_INSERTION_MODE(Text) \
26 __ENUMERATE_INSERTION_MODE(InTable) \
27 __ENUMERATE_INSERTION_MODE(InTableText) \
28 __ENUMERATE_INSERTION_MODE(InCaption) \
29 __ENUMERATE_INSERTION_MODE(InColumnGroup) \
30 __ENUMERATE_INSERTION_MODE(InTableBody) \
31 __ENUMERATE_INSERTION_MODE(InRow) \
32 __ENUMERATE_INSERTION_MODE(InCell) \
33 __ENUMERATE_INSERTION_MODE(InSelect) \
34 __ENUMERATE_INSERTION_MODE(InSelectInTable) \
35 __ENUMERATE_INSERTION_MODE(InTemplate) \
36 __ENUMERATE_INSERTION_MODE(AfterBody) \
37 __ENUMERATE_INSERTION_MODE(InFrameset) \
38 __ENUMERATE_INSERTION_MODE(AfterFrameset) \
39 __ENUMERATE_INSERTION_MODE(AfterAfterBody) \
40 __ENUMERATE_INSERTION_MODE(AfterAfterFrameset)
41
42class HTMLParser final : public JS::Cell {
43 JS_CELL(HTMLParser, JS::Cell);
44
45 friend class HTMLTokenizer;
46
47public:
48 ~HTMLParser();
49
50 static JS::NonnullGCPtr<HTMLParser> create_for_scripting(DOM::Document&);
51 static JS::NonnullGCPtr<HTMLParser> create_with_uncertain_encoding(DOM::Document&, ByteBuffer const& input);
52 static JS::NonnullGCPtr<HTMLParser> create(DOM::Document&, StringView input, DeprecatedString const& encoding);
53
54 void run();
55 void run(const AK::URL&);
56
57 DOM::Document& document();
58
59 static Vector<JS::Handle<DOM::Node>> parse_html_fragment(DOM::Element& context_element, StringView);
60 static DeprecatedString serialize_html_fragment(DOM::Node const& node);
61
62 enum class InsertionMode {
63#define __ENUMERATE_INSERTION_MODE(mode) mode,
64 ENUMERATE_INSERTION_MODES
65#undef __ENUMERATE_INSERTION_MODE
66 };
67
68 InsertionMode insertion_mode() const { return m_insertion_mode; }
69
70 static bool is_special_tag(DeprecatedFlyString const& tag_name, DeprecatedFlyString const& namespace_);
71
72 HTMLTokenizer& tokenizer() { return m_tokenizer; }
73
74 // https://html.spec.whatwg.org/multipage/parsing.html#abort-a-parser
75 void abort();
76
77 bool aborted() const { return m_aborted; }
78 bool stopped() const { return m_stop_parsing; }
79
80 size_t script_nesting_level() const { return m_script_nesting_level; }
81
82private:
83 HTMLParser(DOM::Document&, StringView input, DeprecatedString const& encoding);
84 HTMLParser(DOM::Document&);
85
86 virtual void visit_edges(Cell::Visitor&) override;
87
88 char const* insertion_mode_name() const;
89
90 DOM::QuirksMode which_quirks_mode(HTMLToken const&) const;
91
92 void handle_initial(HTMLToken&);
93 void handle_before_html(HTMLToken&);
94 void handle_before_head(HTMLToken&);
95 void handle_in_head(HTMLToken&);
96 void handle_in_head_noscript(HTMLToken&);
97 void handle_after_head(HTMLToken&);
98 void handle_in_body(HTMLToken&);
99 void handle_after_body(HTMLToken&);
100 void handle_after_after_body(HTMLToken&);
101 void handle_text(HTMLToken&);
102 void handle_in_table(HTMLToken&);
103 void handle_in_table_body(HTMLToken&);
104 void handle_in_row(HTMLToken&);
105 void handle_in_cell(HTMLToken&);
106 void handle_in_table_text(HTMLToken&);
107 void handle_in_select_in_table(HTMLToken&);
108 void handle_in_select(HTMLToken&);
109 void handle_in_caption(HTMLToken&);
110 void handle_in_column_group(HTMLToken&);
111 void handle_in_template(HTMLToken&);
112 void handle_in_frameset(HTMLToken&);
113 void handle_after_frameset(HTMLToken&);
114 void handle_after_after_frameset(HTMLToken&);
115
116 void the_end();
117
118 void stop_parsing() { m_stop_parsing = true; }
119
120 void generate_implied_end_tags(DeprecatedFlyString const& exception = {});
121 void generate_all_implied_end_tags_thoroughly();
122 JS::NonnullGCPtr<DOM::Element> create_element_for(HTMLToken const&, DeprecatedFlyString const& namespace_, DOM::Node& intended_parent);
123
124 struct AdjustedInsertionLocation {
125 JS::GCPtr<DOM::Node> parent;
126 JS::GCPtr<DOM::Node> insert_before_sibling;
127 };
128
129 AdjustedInsertionLocation find_appropriate_place_for_inserting_node(JS::GCPtr<DOM::Element> override_target = nullptr);
130
131 DOM::Text* find_character_insertion_node();
132 void flush_character_insertions();
133 JS::NonnullGCPtr<DOM::Element> insert_foreign_element(HTMLToken const&, DeprecatedFlyString const&);
134 JS::NonnullGCPtr<DOM::Element> insert_html_element(HTMLToken const&);
135 DOM::Element& current_node();
136 DOM::Element& adjusted_current_node();
137 DOM::Element& node_before_current_node();
138 void insert_character(u32 data);
139 void insert_comment(HTMLToken&);
140 void reconstruct_the_active_formatting_elements();
141 void close_a_p_element();
142 void process_using_the_rules_for(InsertionMode, HTMLToken&);
143 void process_using_the_rules_for_foreign_content(HTMLToken&);
144 void parse_generic_raw_text_element(HTMLToken&);
145 void increment_script_nesting_level();
146 void decrement_script_nesting_level();
147 void reset_the_insertion_mode_appropriately();
148
149 void adjust_mathml_attributes(HTMLToken&);
150 void adjust_svg_tag_names(HTMLToken&);
151 void adjust_svg_attributes(HTMLToken&);
152 void adjust_foreign_attributes(HTMLToken&);
153
154 enum AdoptionAgencyAlgorithmOutcome {
155 DoNothing,
156 RunAnyOtherEndTagSteps,
157 };
158
159 AdoptionAgencyAlgorithmOutcome run_the_adoption_agency_algorithm(HTMLToken&);
160 void clear_the_stack_back_to_a_table_context();
161 void clear_the_stack_back_to_a_table_body_context();
162 void clear_the_stack_back_to_a_table_row_context();
163 void close_the_cell();
164
165 InsertionMode m_insertion_mode { InsertionMode::Initial };
166 InsertionMode m_original_insertion_mode { InsertionMode::Initial };
167
168 StackOfOpenElements m_stack_of_open_elements;
169 Vector<InsertionMode> m_stack_of_template_insertion_modes;
170 ListOfActiveFormattingElements m_list_of_active_formatting_elements;
171
172 HTMLTokenizer m_tokenizer;
173
174 bool m_foster_parenting { false };
175 bool m_frameset_ok { true };
176 bool m_parsing_fragment { false };
177
178 // https://html.spec.whatwg.org/multipage/parsing.html#scripting-flag
179 // 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.
180 bool m_scripting_enabled { true };
181
182 bool m_invoked_via_document_write { false };
183 bool m_aborted { false };
184 bool m_parser_pause_flag { false };
185 bool m_stop_parsing { false };
186 size_t m_script_nesting_level { 0 };
187
188 JS::Realm& realm();
189
190 JS::GCPtr<DOM::Document> m_document;
191 JS::GCPtr<HTMLHeadElement> m_head_element;
192 JS::GCPtr<HTMLFormElement> m_form_element;
193 JS::GCPtr<DOM::Element> m_context_element;
194
195 Vector<HTMLToken> m_pending_table_character_tokens;
196
197 JS::GCPtr<DOM::Text> m_character_insertion_node;
198 StringBuilder m_character_insertion_builder;
199};
200
201RefPtr<CSS::StyleValue> parse_dimension_value(StringView);
202RefPtr<CSS::StyleValue> parse_nonzero_dimension_value(StringView);
203
204}