1//! Tools to work with source positions and mapping.
2
3#[derive(Debug)]
4/// Holds source code, allows to calculate `line:column` from byte offset.
5pub struct SourceWithLineStarts {
6 src: String,
7 marks: Vec<CharMappingMark>,
8}
9
10impl SourceWithLineStarts {
11 pub fn new(src: &str) -> Self {
12 let mut iterator = src.char_indices().peekable();
13 let mut line = 1;
14 let mut column = 0;
15 let mut marks = vec![CharMappingMark { offset: 0, line, column }];
16
17 loop {
18 match iterator.next() {
19 Some((_, '\r')) if matches!(iterator.peek(), Some((_, '\n'))) => {
20 // ignore \r followed by \n
21 column += 1;
22 }
23 Some((offset, '\r' | '\n')) => {
24 // \r or \n are linebreaks
25 line += 1;
26 column = 0;
27 marks.push(CharMappingMark { offset: offset + 1, line, column });
28 }
29 Some((offset, _)) => {
30 // any other character, just increase position
31 if column % 16 == 0 && column > 0 {
32 marks.push(CharMappingMark { offset, line, column });
33 }
34 column += 1;
35 },
36 None => break,
37 }
38 }
39
40 Self { src: src.to_owned(), marks }
41 }
42
43 fn get_position(&self, byte_offset: usize) -> (u32, u32) {
44 let byte_offset = byte_offset + 1; // include current char
45 let found = match self.marks.binary_search_by(|mark| mark.offset.cmp(&byte_offset)) {
46 Ok(x) => x,
47 Err(x) => x - 1,
48 };
49 let mark = &self.marks[found];
50 let line = mark.line;
51 let mut column = mark.column;
52 for (offset, _) in self.src[mark.offset..].char_indices() {
53 if mark.offset + offset >= byte_offset { break; }
54 column += 1;
55 }
56 (line, column)
57 }
58}
59
60#[derive(Debug)]
61struct CharMappingMark {
62 offset: usize,
63 line: u32,
64 column: u32,
65}
66
67#[derive(Default, Clone, Copy)]
68/// Positions of the start and the end of an AST node.
69pub struct SourcePos {
70 byte_offset: (usize, usize),
71}
72
73impl SourcePos {
74 /// Create positions from byte offsets:
75 /// - start - offset of the first char of the node
76 /// - end - offset of the first char after the node
77 pub fn new(start: usize, end: usize) -> Self {
78 SourcePos {
79 byte_offset: (start, end),
80 }
81 }
82
83 pub fn get_byte_offsets(&self) -> (usize, usize) {
84 self.byte_offset
85 }
86
87 /// Returns (line_start, column_start, line_end, column_end) from given positions
88 pub fn get_positions(&self, map: &SourceWithLineStarts) -> ((u32, u32), (u32, u32)) {
89 let start = map.get_position(self.byte_offset.0);
90 let end_off = if self.byte_offset.1 > 0 { self.byte_offset.1 - 1 } else { self.byte_offset.1 };
91 let end = map.get_position(end_off);
92 (start, end)
93 }
94}
95
96impl std::fmt::Debug for SourcePos {
97 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
98 self.byte_offset.fmt(f)
99 }
100}
101
102#[cfg(test)]
103mod tests {
104 use super::SourceWithLineStarts;
105 use super::SourcePos;
106
107 #[test]
108 fn no_linebreaks() {
109 let map = SourceWithLineStarts::new("qwertyuiopasdfghjklzxcvbnmQWERTYUIOPASDFGHJKLZXCVBNM");
110 for i in 0..20 {
111 assert_eq!(SourcePos::new(i, 0).get_positions(&map).0, (1, i as u32 + 1));
112 }
113 }
114
115 #[test]
116 fn unicode() {
117 let map = SourceWithLineStarts::new("!ΑαΒβΓγΔδΕεΖζΗηΘθΙιΚκΛλΜμΝνΞξΟοΠπΡρΣσςΤτΥυΦφΧχΨψΩω");
118 assert_eq!(SourcePos::new(0, 0).get_positions(&map).0, (1, 1));
119 for i in 1..20 {
120 assert_eq!(SourcePos::new(i, 0).get_positions(&map).0, (1, ((i - 1) / 2) as u32 + 2));
121 }
122 }
123
124 #[test]
125 fn many_linebreaks() {
126 let map = SourceWithLineStarts::new("\n\n\n\n\n\n123");
127 for i in 0..6 {
128 assert_eq!(SourcePos::new(i, 0).get_positions(&map).0, (i as u32 + 2, 0));
129 }
130 assert_eq!(SourcePos::new(7, 0).get_positions(&map).0, (7, 2));
131 assert_eq!(SourcePos::new(8, 0).get_positions(&map).0, (7, 3));
132 }
133
134 #[test]
135 fn after_end() {
136 let map = SourceWithLineStarts::new("123");
137 assert_eq!(SourcePos::new(100, 0).get_positions(&map).0, (1, 3));
138 let map = SourceWithLineStarts::new("123\n");
139 assert_eq!(SourcePos::new(100, 0).get_positions(&map).0, (2, 0));
140 let map = SourceWithLineStarts::new("123\n456");
141 assert_eq!(SourcePos::new(100, 0).get_positions(&map).0, (2, 3));
142 }
143}