|
| 1 | +/// Protocol-specific text position handling. |
| 2 | +/// |
| 3 | +/// This module provides types and functions for converting between different |
| 4 | +/// text position representations used by various protocols and editors. |
| 5 | +use crate::position::ByteOffset; |
| 6 | +/// Protocol-specific text position handling. |
| 7 | +/// |
| 8 | +/// This module provides types and functions for converting between different |
| 9 | +/// text position representations used by various protocols and editors. |
| 10 | +use crate::position::LineCol; |
| 11 | +/// Protocol-specific text position handling. |
| 12 | +/// |
| 13 | +/// This module provides types and functions for converting between different |
| 14 | +/// text position representations used by various protocols and editors. |
| 15 | +use crate::position::LineIndex; |
| 16 | + |
| 17 | +/// Specifies how column positions are counted in text. |
| 18 | +/// |
| 19 | +/// While motivated by LSP (Language Server Protocol) requirements, this enum |
| 20 | +/// represents a fundamental choice about text position measurement that any |
| 21 | +/// text processing system must make. Different systems count "column" positions |
| 22 | +/// differently: |
| 23 | +/// |
| 24 | +/// - Some count bytes (fast but breaks on multi-byte characters) |
| 25 | +/// - Some count UTF-16 code units (common in JavaScript/Windows ecosystems) |
| 26 | +/// - Some count Unicode codepoints (intuitive but slower) |
| 27 | +/// |
| 28 | +/// This crate provides encoding-aware position conversion to support different |
| 29 | +/// client expectations without coupling to specific protocol implementations. |
| 30 | +#[derive(Debug, Default, Clone, Copy, PartialEq, Eq)] |
| 31 | +pub enum PositionEncoding { |
| 32 | + /// Column positions count UTF-8 code units (bytes from line start) |
| 33 | + Utf8, |
| 34 | + /// Column positions count UTF-16 code units (common in VS Code and Windows editors) |
| 35 | + #[default] |
| 36 | + Utf16, |
| 37 | + /// Column positions count Unicode scalar values (codepoints) |
| 38 | + Utf32, |
| 39 | +} |
| 40 | + |
| 41 | +impl LineIndex { |
| 42 | + /// Convert a line/column position to a byte offset with encoding awareness. |
| 43 | + /// |
| 44 | + /// The `encoding` parameter specifies how the column value should be interpreted: |
| 45 | + /// - `PositionEncoding::Utf8`: column is a byte offset from line start |
| 46 | + /// - `PositionEncoding::Utf16`: column counts UTF-16 code units |
| 47 | + /// - `PositionEncoding::Utf32`: column counts Unicode codepoints |
| 48 | + /// |
| 49 | + /// This method is primarily used to convert protocol-specific positions |
| 50 | + /// (which may use different column counting methods) into byte offsets |
| 51 | + /// that can be used to index into the actual UTF-8 text. |
| 52 | + /// |
| 53 | + /// # Examples |
| 54 | + /// |
| 55 | + /// ``` |
| 56 | + /// # use djls_source::{LineIndex, LineCol, ByteOffset, PositionEncoding}; |
| 57 | + /// let text = "Hello 🌍 world"; |
| 58 | + /// let index = LineIndex::from_text(text); |
| 59 | + /// |
| 60 | + /// // UTF-16: "Hello " (6) + "🌍" (2 UTF-16 units) = position 8 |
| 61 | + /// let offset = index.line_col_to_offset( |
| 62 | + /// LineCol((0, 8)), |
| 63 | + /// text, |
| 64 | + /// PositionEncoding::Utf16 |
| 65 | + /// ); |
| 66 | + /// assert_eq!(offset, Some(ByteOffset(10))); // "Hello 🌍" is 10 bytes |
| 67 | + /// ``` |
| 68 | + #[must_use] |
| 69 | + pub fn line_col_to_offset( |
| 70 | + &self, |
| 71 | + line_col: LineCol, |
| 72 | + text: &str, |
| 73 | + encoding: PositionEncoding, |
| 74 | + ) -> Option<ByteOffset> { |
| 75 | + let line = line_col.line(); |
| 76 | + let character = line_col.column(); |
| 77 | + |
| 78 | + // Handle line bounds - if line > line_count, return document length |
| 79 | + let line_start_utf8 = match self.lines().get(line as usize) { |
| 80 | + Some(start) => *start, |
| 81 | + None => return Some(ByteOffset(u32::try_from(text.len()).unwrap_or(u32::MAX))), |
| 82 | + }; |
| 83 | + |
| 84 | + if character == 0 { |
| 85 | + return Some(ByteOffset(line_start_utf8)); |
| 86 | + } |
| 87 | + |
| 88 | + let next_line_start = self |
| 89 | + .lines() |
| 90 | + .get(line as usize + 1) |
| 91 | + .copied() |
| 92 | + .unwrap_or_else(|| u32::try_from(text.len()).unwrap_or(u32::MAX)); |
| 93 | + |
| 94 | + let line_text = text.get(line_start_utf8 as usize..next_line_start as usize)?; |
| 95 | + |
| 96 | + // Fast path optimization for ASCII text, all encodings are equivalent to byte offsets |
| 97 | + if line_text.is_ascii() { |
| 98 | + let char_offset = character.min(u32::try_from(line_text.len()).unwrap_or(u32::MAX)); |
| 99 | + return Some(ByteOffset(line_start_utf8 + char_offset)); |
| 100 | + } |
| 101 | + |
| 102 | + match encoding { |
| 103 | + PositionEncoding::Utf8 => { |
| 104 | + // UTF-8: character positions are already byte offsets |
| 105 | + let char_offset = character.min(u32::try_from(line_text.len()).unwrap_or(u32::MAX)); |
| 106 | + Some(ByteOffset(line_start_utf8 + char_offset)) |
| 107 | + } |
| 108 | + PositionEncoding::Utf16 => { |
| 109 | + // UTF-16: count UTF-16 code units |
| 110 | + let mut utf16_pos = 0; |
| 111 | + let mut utf8_pos = 0; |
| 112 | + |
| 113 | + for c in line_text.chars() { |
| 114 | + if utf16_pos >= character { |
| 115 | + break; |
| 116 | + } |
| 117 | + utf16_pos += u32::try_from(c.len_utf16()).unwrap_or(0); |
| 118 | + utf8_pos += u32::try_from(c.len_utf8()).unwrap_or(0); |
| 119 | + } |
| 120 | + |
| 121 | + // If character position exceeds line length, clamp to line end |
| 122 | + Some(ByteOffset(line_start_utf8 + utf8_pos)) |
| 123 | + } |
| 124 | + PositionEncoding::Utf32 => { |
| 125 | + // UTF-32: count Unicode code points (characters) |
| 126 | + let mut utf8_pos = 0; |
| 127 | + |
| 128 | + for (char_count, c) in line_text.chars().enumerate() { |
| 129 | + if char_count >= character as usize { |
| 130 | + break; |
| 131 | + } |
| 132 | + utf8_pos += u32::try_from(c.len_utf8()).unwrap_or(0); |
| 133 | + } |
| 134 | + |
| 135 | + // If character position exceeds line length, clamp to line end |
| 136 | + Some(ByteOffset(line_start_utf8 + utf8_pos)) |
| 137 | + } |
| 138 | + } |
| 139 | + } |
| 140 | +} |
| 141 | + |
| 142 | +#[cfg(test)] |
| 143 | +mod tests { |
| 144 | + use super::*; |
| 145 | + |
| 146 | + #[test] |
| 147 | + fn test_line_col_to_offset_utf16() { |
| 148 | + let text = "Hello 🌍 world"; |
| 149 | + let index = LineIndex::from_text(text); |
| 150 | + |
| 151 | + // "Hello " = 6 UTF-16 units, "🌍" = 2 UTF-16 units |
| 152 | + // So position (0, 8) in UTF-16 should be after the emoji |
| 153 | + let offset = index |
| 154 | + .line_col_to_offset(LineCol((0, 8)), text, PositionEncoding::Utf16) |
| 155 | + .expect("Should get offset"); |
| 156 | + assert_eq!(offset, ByteOffset(10)); // "Hello 🌍" is 10 bytes |
| 157 | + |
| 158 | + // In UTF-8, character 10 would be at the 'r' in 'world' |
| 159 | + let offset_utf8 = index |
| 160 | + .line_col_to_offset(LineCol((0, 10)), text, PositionEncoding::Utf8) |
| 161 | + .expect("Should get offset"); |
| 162 | + assert_eq!(offset_utf8, ByteOffset(10)); |
| 163 | + } |
| 164 | + |
| 165 | + #[test] |
| 166 | + fn test_line_col_to_offset_ascii_fast_path() { |
| 167 | + let text = "Hello world"; |
| 168 | + let index = LineIndex::from_text(text); |
| 169 | + |
| 170 | + // For ASCII text, all encodings should give the same result |
| 171 | + let offset_utf8 = index |
| 172 | + .line_col_to_offset(LineCol((0, 5)), text, PositionEncoding::Utf8) |
| 173 | + .expect("Should get offset"); |
| 174 | + let offset_utf16 = index |
| 175 | + .line_col_to_offset(LineCol((0, 5)), text, PositionEncoding::Utf16) |
| 176 | + .expect("Should get offset"); |
| 177 | + let offset_utf32 = index |
| 178 | + .line_col_to_offset(LineCol((0, 5)), text, PositionEncoding::Utf32) |
| 179 | + .expect("Should get offset"); |
| 180 | + |
| 181 | + assert_eq!(offset_utf8, ByteOffset(5)); |
| 182 | + assert_eq!(offset_utf16, ByteOffset(5)); |
| 183 | + assert_eq!(offset_utf32, ByteOffset(5)); |
| 184 | + } |
| 185 | +} |
0 commit comments