1 | use crate::reader::error::SyntaxError; |
2 | use crate::common::is_whitespace_char; |
3 | use crate::reader::events::XmlEvent; |
4 | use crate::reader::lexer::Token; |
5 | |
6 | use super::{ |
7 | ClosingTagSubstate, DoctypeSubstate, Encountered, OpeningTagSubstate, |
8 | ProcessingInstructionSubstate, PullParser, Result, State, |
9 | }; |
10 | |
11 | impl PullParser { |
12 | pub fn outside_tag(&mut self, t: Token) -> Option<Result> { |
13 | match t { |
14 | Token::Character(c) => { |
15 | if is_whitespace_char(c) { |
16 | // skip whitespace outside of the root element |
17 | if (self.config.c.trim_whitespace && self.buf.is_empty()) || |
18 | (self.depth() == 0 && self.config.c.ignore_root_level_whitespace) { |
19 | return None; |
20 | } |
21 | } else { |
22 | self.inside_whitespace = false; |
23 | if self.depth() == 0 { |
24 | return Some(self.error(SyntaxError::UnexpectedTokenOutsideRoot(t))); |
25 | } |
26 | } |
27 | |
28 | if !self.is_valid_xml_char_not_restricted(c) { |
29 | return Some(self.error(SyntaxError::InvalidCharacterEntity(c as u32))); |
30 | } |
31 | |
32 | if self.buf.is_empty() { |
33 | self.push_pos(); |
34 | } else if self.buf.len() > self.config.max_data_length { |
35 | return Some(self.error(SyntaxError::ExceededConfiguredLimit)); |
36 | } |
37 | self.buf.push(c); |
38 | None |
39 | }, |
40 | |
41 | Token::CommentEnd | Token::TagEnd | Token::EqualsSign | |
42 | Token::DoubleQuote | Token::SingleQuote | |
43 | Token::ProcessingInstructionEnd | Token::EmptyTagEnd => { |
44 | if self.depth() == 0 { |
45 | return Some(self.error(SyntaxError::UnexpectedTokenOutsideRoot(t))); |
46 | } |
47 | self.inside_whitespace = false; |
48 | |
49 | if let Some(s) = t.as_static_str() { |
50 | if self.buf.is_empty() { |
51 | self.push_pos(); |
52 | } else if self.buf.len() > self.config.max_data_length { |
53 | return Some(self.error(SyntaxError::ExceededConfiguredLimit)); |
54 | } |
55 | |
56 | self.buf.push_str(s); |
57 | } |
58 | None |
59 | }, |
60 | |
61 | Token::ReferenceStart if self.depth() > 0 => { |
62 | self.state_after_reference = State::OutsideTag; |
63 | self.into_state_continue(State::InsideReference) |
64 | }, |
65 | |
66 | Token::ReferenceEnd if self.depth() > 0 => { // Semi-colon in a text outside an entity |
67 | self.inside_whitespace = false; |
68 | if self.buf.len() > self.config.max_data_length { |
69 | return Some(self.error(SyntaxError::ExceededConfiguredLimit)); |
70 | } |
71 | Token::ReferenceEnd.push_to_string(&mut self.buf); |
72 | None |
73 | }, |
74 | |
75 | Token::CommentStart if self.config.c.coalesce_characters && self.config.c.ignore_comments => { |
76 | let next_event = self.set_encountered(Encountered::Comment); |
77 | // We need to switch the lexer into a comment mode inside comments |
78 | self.into_state(State::InsideComment, next_event) |
79 | } |
80 | |
81 | Token::CDataStart if self.depth() > 0 && self.config.c.coalesce_characters && self.config.c.cdata_to_characters => { |
82 | if self.buf.is_empty() { |
83 | self.push_pos(); |
84 | } |
85 | self.into_state_continue(State::InsideCData) |
86 | }, |
87 | |
88 | _ => { |
89 | // Encountered some markup event, flush the buffer as characters |
90 | // or a whitespace |
91 | let mut next_event = if self.buf_has_data() { |
92 | let buf = self.take_buf(); |
93 | if self.inside_whitespace && self.config.c.trim_whitespace { |
94 | None |
95 | } else if self.inside_whitespace && !self.config.c.whitespace_to_characters { |
96 | debug_assert!(buf.chars().all(|ch| ch.is_whitespace()), "ws= {buf:?}" ); |
97 | Some(Ok(XmlEvent::Whitespace(buf))) |
98 | } else if self.config.c.trim_whitespace { |
99 | Some(Ok(XmlEvent::Characters(buf.trim_matches(is_whitespace_char).into()))) |
100 | } else { |
101 | Some(Ok(XmlEvent::Characters(buf))) |
102 | } |
103 | } else { None }; |
104 | self.inside_whitespace = true; // Reset inside_whitespace flag |
105 | |
106 | // pos is popped whenever an event is emitted, so pushes must happen only if there will be an event to balance it |
107 | // and ignored comments don't pop |
108 | if t != Token::CommentStart || !self.config.c.ignore_comments { |
109 | self.push_pos(); |
110 | } |
111 | match t { |
112 | Token::OpeningTagStart if self.depth() > 0 || self.encountered < Encountered::Element || self.config.allow_multiple_root_elements => { |
113 | if let Some(e) = self.set_encountered(Encountered::Element) { |
114 | next_event = Some(e); |
115 | } |
116 | self.nst.push_empty(); |
117 | self.into_state(State::InsideOpeningTag(OpeningTagSubstate::InsideName), next_event) |
118 | }, |
119 | |
120 | Token::ClosingTagStart if self.depth() > 0 => |
121 | self.into_state(State::InsideClosingTag(ClosingTagSubstate::CTInsideName), next_event), |
122 | |
123 | Token::CommentStart => { |
124 | if let Some(e) = self.set_encountered(Encountered::Comment) { |
125 | next_event = Some(e); |
126 | } |
127 | // We need to switch the lexer into a comment mode inside comments |
128 | self.into_state(State::InsideComment, next_event) |
129 | }, |
130 | |
131 | Token::DoctypeStart if self.encountered < Encountered::Doctype => { |
132 | if let Some(e) = self.set_encountered(Encountered::Doctype) { |
133 | next_event = Some(e); |
134 | } |
135 | |
136 | // We don't have a doctype event so skip this position |
137 | // FIXME: update when we have a doctype event |
138 | self.next_pos(); |
139 | self.into_state(State::InsideDoctype(DoctypeSubstate::Outside), next_event) |
140 | }, |
141 | |
142 | Token::ProcessingInstructionStart => |
143 | self.into_state(State::InsideProcessingInstruction(ProcessingInstructionSubstate::PIInsideName), next_event), |
144 | |
145 | Token::CDataStart if self.depth() > 0 => { |
146 | self.into_state(State::InsideCData, next_event) |
147 | }, |
148 | |
149 | _ => Some(self.error(SyntaxError::UnexpectedToken(t))) |
150 | } |
151 | } |
152 | } |
153 | } |
154 | |
155 | pub fn document_start(&mut self, t: Token) -> Option<Result> { |
156 | debug_assert!(self.encountered < Encountered::Declaration); |
157 | |
158 | match t { |
159 | Token::Character(c) => { |
160 | let next_event = self.set_encountered(Encountered::AnyChars); |
161 | |
162 | if !is_whitespace_char(c) { |
163 | return Some(self.error(SyntaxError::UnexpectedTokenOutsideRoot(t))); |
164 | } |
165 | self.inside_whitespace = true; |
166 | |
167 | // skip whitespace outside of the root element |
168 | if (self.config.c.trim_whitespace && self.buf.is_empty()) || |
169 | (self.depth() == 0 && self.config.c.ignore_root_level_whitespace) { |
170 | return self.into_state(State::OutsideTag, next_event); |
171 | } |
172 | |
173 | self.push_pos(); |
174 | self.buf.push(c); |
175 | self.into_state(State::OutsideTag, next_event) |
176 | }, |
177 | |
178 | Token::CommentStart => { |
179 | let next_event = self.set_encountered(Encountered::Comment); |
180 | self.into_state(State::InsideComment, next_event) |
181 | } |
182 | |
183 | Token::OpeningTagStart => { |
184 | let next_event = self.set_encountered(Encountered::Element); |
185 | self.nst.push_empty(); |
186 | self.into_state(State::InsideOpeningTag(OpeningTagSubstate::InsideName), next_event) |
187 | }, |
188 | |
189 | Token::DoctypeStart => { |
190 | let next_event = self.set_encountered(Encountered::Doctype); |
191 | // We don't have a doctype event so skip this position |
192 | // FIXME: update when we have a doctype event |
193 | self.next_pos(); |
194 | self.into_state(State::InsideDoctype(DoctypeSubstate::Outside), next_event) |
195 | }, |
196 | |
197 | Token::ProcessingInstructionStart => { |
198 | self.push_pos(); |
199 | self.into_state_continue(State::InsideProcessingInstruction(ProcessingInstructionSubstate::PIInsideName)) |
200 | }, |
201 | |
202 | _ => Some(self.error(SyntaxError::UnexpectedToken(t))), |
203 | } |
204 | } |
205 | } |
206 | |