|
| 1 | +use std::borrow::Cow; |
| 2 | + |
| 3 | +use helix_stdx::rope::{self, RopeSliceExt}; |
| 4 | +use once_cell::sync::Lazy; |
| 5 | + |
| 6 | +use crate::indent::IndentStyle; |
| 7 | +use crate::{LineEnding, RopeSlice}; |
| 8 | + |
| 9 | +// 5 is the vim default |
| 10 | +const LINES_TO_CHECK: usize = 5; |
| 11 | + |
| 12 | +static HELIX_MODELINE_REGEX: Lazy<rope::Regex> = |
| 13 | + Lazy::new(|| rope::Regex::new(r"^(.{0,100}\s{1,100})?helix:").unwrap()); |
| 14 | +static HELIX_MODELINE_OPTION_REGEX: Lazy<rope::Regex> = |
| 15 | + Lazy::new(|| rope::Regex::new(r"[a-zA-Z0-9_-]{1,100}(?:=[a-zA-Z0-9_-]{1,100})").unwrap()); |
| 16 | +static VIM_MODELINE_REGEX: Lazy<rope::Regex> = Lazy::new(|| { |
| 17 | + rope::Regex::new( |
| 18 | + r"^(.{0,100}\s{1,100})?(vi|[vV]im[<=>]?\d{0,100}|ex):\s{0,100}(set?\s{1,100})?", |
| 19 | + ) |
| 20 | + .unwrap() |
| 21 | +}); |
| 22 | +static VIM_MODELINE_OPTION_REGEX: Lazy<rope::Regex> = |
| 23 | + Lazy::new(|| rope::Regex::new(r"[a-zA-Z0-9_-]{1,100}(?:=(?:\\:|[^:\s]){0,1000})?").unwrap()); |
| 24 | + |
| 25 | +#[derive(Default, Debug, Eq, PartialEq)] |
| 26 | +pub struct Modeline { |
| 27 | + language: Option<String>, |
| 28 | + indent_style: Option<IndentStyle>, |
| 29 | + line_ending: Option<LineEnding>, |
| 30 | +} |
| 31 | + |
| 32 | +impl Modeline { |
| 33 | + pub fn parse(text: RopeSlice) -> Self { |
| 34 | + let mut modeline = Self::default(); |
| 35 | + |
| 36 | + for line in text.lines().take(LINES_TO_CHECK).chain( |
| 37 | + text.lines_at(text.len_lines()) |
| 38 | + .reversed() |
| 39 | + .take(LINES_TO_CHECK), |
| 40 | + ) { |
| 41 | + modeline.parse_from_line(line); |
| 42 | + } |
| 43 | + |
| 44 | + modeline |
| 45 | + } |
| 46 | + |
| 47 | + pub fn language(&self) -> Option<&str> { |
| 48 | + self.language.as_deref() |
| 49 | + } |
| 50 | + |
| 51 | + pub fn indent_style(&self) -> Option<IndentStyle> { |
| 52 | + self.indent_style |
| 53 | + } |
| 54 | + |
| 55 | + pub fn line_ending(&self) -> Option<LineEnding> { |
| 56 | + self.line_ending |
| 57 | + } |
| 58 | + |
| 59 | + fn parse_from_line(&mut self, mut line: RopeSlice) { |
| 60 | + if let Some(pos) = HELIX_MODELINE_REGEX.find(line.regex_input()) { |
| 61 | + line = line.slice(line.byte_to_char(pos.end())..); |
| 62 | + while let Some(opt_pos) = HELIX_MODELINE_OPTION_REGEX.find(line.regex_input()) { |
| 63 | + let option = |
| 64 | + Cow::from(line.slice( |
| 65 | + line.byte_to_char(opt_pos.start())..line.byte_to_char(opt_pos.end()), |
| 66 | + )); |
| 67 | + let mut parts = option.as_ref().split('='); |
| 68 | + match parts.next().unwrap() { |
| 69 | + "set-language" | "lang" => { |
| 70 | + if let Some(val) = parts.next() { |
| 71 | + self.language = Some(val.to_string()); |
| 72 | + } |
| 73 | + } |
| 74 | + "indent-style" => { |
| 75 | + if let Some(val) = parts.next() { |
| 76 | + if let Some(indent_style) = IndentStyle::from_option_str(val) { |
| 77 | + self.indent_style = Some(indent_style); |
| 78 | + } |
| 79 | + } |
| 80 | + } |
| 81 | + "line-ending" => { |
| 82 | + if let Some(val) = parts.next() { |
| 83 | + if let Some(line_ending) = LineEnding::from_option_str(val) { |
| 84 | + self.line_ending = Some(line_ending); |
| 85 | + } |
| 86 | + } |
| 87 | + } |
| 88 | + _ => {} |
| 89 | + } |
| 90 | + line = line.slice(line.byte_to_char(opt_pos.end())..); |
| 91 | + let whitespace = line.chars().take_while(|c| char::is_whitespace(*c)).count(); |
| 92 | + line = line.slice(whitespace..); |
| 93 | + } |
| 94 | + } else if let Some(pos) = VIM_MODELINE_REGEX.find(line.regex_input()) { |
| 95 | + line = line.slice(line.byte_to_char(pos.end())..); |
| 96 | + while let Some(opt_pos) = VIM_MODELINE_OPTION_REGEX.find(line.regex_input()) { |
| 97 | + let option = |
| 98 | + Cow::from(line.slice( |
| 99 | + line.byte_to_char(opt_pos.start())..line.byte_to_char(opt_pos.end()), |
| 100 | + )); |
| 101 | + let mut parts = option.as_ref().splitn(2, '='); |
| 102 | + match parts.next().unwrap() { |
| 103 | + "ft" | "filetype" => { |
| 104 | + if let Some(val) = parts.next() { |
| 105 | + self.language = Some(val.to_string()); |
| 106 | + } |
| 107 | + } |
| 108 | + "sw" | "shiftwidth" => { |
| 109 | + if let Some(val) = parts.next().and_then(|val| val.parse().ok()) { |
| 110 | + if self.indent_style != Some(IndentStyle::Tabs) { |
| 111 | + self.indent_style = Some(IndentStyle::Spaces(val)); |
| 112 | + } |
| 113 | + } |
| 114 | + } |
| 115 | + "ff" | "fileformat" => { |
| 116 | + if let Some(val) = parts.next() { |
| 117 | + self.line_ending = vim_ff_to_helix_line_ending(val); |
| 118 | + } |
| 119 | + } |
| 120 | + "noet" | "noexpandtab" => { |
| 121 | + self.indent_style = Some(IndentStyle::Tabs); |
| 122 | + } |
| 123 | + "et" | "expandtab" => { |
| 124 | + if !matches!(self.indent_style, Some(IndentStyle::Spaces(_))) { |
| 125 | + self.indent_style = Some(IndentStyle::Spaces(0)); |
| 126 | + } |
| 127 | + } |
| 128 | + _ => {} |
| 129 | + } |
| 130 | + line = line.slice(line.byte_to_char(opt_pos.end())..); |
| 131 | + let whitespace = line |
| 132 | + .chars() |
| 133 | + .take_while(|c| char::is_whitespace(*c) || *c == ':') |
| 134 | + .count(); |
| 135 | + line = line.slice(whitespace..); |
| 136 | + } |
| 137 | + } |
| 138 | + } |
| 139 | +} |
| 140 | + |
| 141 | +fn vim_ff_to_helix_line_ending(val: &str) -> Option<LineEnding> { |
| 142 | + match val { |
| 143 | + "dos" => Some(LineEnding::Crlf), |
| 144 | + "unix" => Some(LineEnding::LF), |
| 145 | + #[cfg(feature = "unicode-lines")] |
| 146 | + "mac" => Some(LineEnding::CR), |
| 147 | + _ => None, |
| 148 | + } |
| 149 | +} |
| 150 | + |
| 151 | +#[cfg(test)] |
| 152 | +mod test { |
| 153 | + use super::*; |
| 154 | + |
| 155 | + #[test] |
| 156 | + fn test_modeline_parsing() { |
| 157 | + let tests = [ |
| 158 | + ( |
| 159 | + "# helix: set-language=perl", |
| 160 | + Modeline { |
| 161 | + language: Some("perl".to_string()), |
| 162 | + ..Default::default() |
| 163 | + }, |
| 164 | + ), |
| 165 | + ( |
| 166 | + "# helix: lang=perl", |
| 167 | + Modeline { |
| 168 | + language: Some("perl".to_string()), |
| 169 | + ..Default::default() |
| 170 | + }, |
| 171 | + ), |
| 172 | + ( |
| 173 | + "# helix: indent-style=3", |
| 174 | + Modeline { |
| 175 | + indent_style: Some(IndentStyle::Spaces(3)), |
| 176 | + ..Default::default() |
| 177 | + }, |
| 178 | + ), |
| 179 | + ( |
| 180 | + "# helix: indent-style=t", |
| 181 | + Modeline { |
| 182 | + indent_style: Some(IndentStyle::Tabs), |
| 183 | + ..Default::default() |
| 184 | + }, |
| 185 | + ), |
| 186 | + ( |
| 187 | + "# helix: line-ending=crlf", |
| 188 | + Modeline { |
| 189 | + line_ending: Some(LineEnding::Crlf), |
| 190 | + ..Default::default() |
| 191 | + }, |
| 192 | + ), |
| 193 | + ( |
| 194 | + "# helix: lang=perl indent-style=t line-ending=crlf", |
| 195 | + Modeline { |
| 196 | + language: Some("perl".to_string()), |
| 197 | + indent_style: Some(IndentStyle::Tabs), |
| 198 | + line_ending: Some(LineEnding::Crlf), |
| 199 | + }, |
| 200 | + ), |
| 201 | + ( |
| 202 | + "#//-- helix: lang=perl indent-style=t line-ending=crlf", |
| 203 | + Modeline { |
| 204 | + language: Some("perl".to_string()), |
| 205 | + indent_style: Some(IndentStyle::Tabs), |
| 206 | + line_ending: Some(LineEnding::Crlf), |
| 207 | + }, |
| 208 | + ), |
| 209 | + ( |
| 210 | + "vi:noai:sw=3 ts=6", |
| 211 | + Modeline { |
| 212 | + indent_style: Some(IndentStyle::Spaces(3)), |
| 213 | + ..Default::default() |
| 214 | + }, |
| 215 | + ), |
| 216 | + ( |
| 217 | + "vim: tw=77", |
| 218 | + Modeline { |
| 219 | + ..Default::default() |
| 220 | + }, |
| 221 | + ), |
| 222 | + ( |
| 223 | + "/* vim: set ai sw=5: */", |
| 224 | + Modeline { |
| 225 | + indent_style: Some(IndentStyle::Spaces(5)), |
| 226 | + ..Default::default() |
| 227 | + }, |
| 228 | + ), |
| 229 | + ( |
| 230 | + "# vim: set noexpandtab:", |
| 231 | + Modeline { |
| 232 | + indent_style: Some(IndentStyle::Tabs), |
| 233 | + ..Default::default() |
| 234 | + }, |
| 235 | + ), |
| 236 | + ( |
| 237 | + "# vim: set expandtab:", |
| 238 | + Modeline { |
| 239 | + indent_style: Some(IndentStyle::Spaces(0)), |
| 240 | + ..Default::default() |
| 241 | + }, |
| 242 | + ), |
| 243 | + ( |
| 244 | + "// vim: noai:ts=4:sw=4", |
| 245 | + Modeline { |
| 246 | + indent_style: Some(IndentStyle::Spaces(4)), |
| 247 | + ..Default::default() |
| 248 | + }, |
| 249 | + ), |
| 250 | + ( |
| 251 | + "/* vim: set noai ts=4 sw=4: */", |
| 252 | + Modeline { |
| 253 | + indent_style: Some(IndentStyle::Spaces(4)), |
| 254 | + ..Default::default() |
| 255 | + }, |
| 256 | + ), |
| 257 | + ( |
| 258 | + "/* vim: set fdm=expr ft=c fde=getline(v\\:lnum)=~'{'?'>1'\\:'1' sw=4: */", |
| 259 | + Modeline { |
| 260 | + language: Some("c".to_string()), |
| 261 | + indent_style: Some(IndentStyle::Spaces(4)), |
| 262 | + ..Default::default() |
| 263 | + }, |
| 264 | + ), |
| 265 | + ( |
| 266 | + "/* vim: set ts=8 sw=4 tw=0 noet : */", |
| 267 | + Modeline { |
| 268 | + indent_style: Some(IndentStyle::Tabs), |
| 269 | + ..Default::default() |
| 270 | + }, |
| 271 | + ), |
| 272 | + ( |
| 273 | + "vim:ff=unix ts=4 sw=4", |
| 274 | + Modeline { |
| 275 | + indent_style: Some(IndentStyle::Spaces(4)), |
| 276 | + line_ending: Some(LineEnding::LF), |
| 277 | + ..Default::default() |
| 278 | + }, |
| 279 | + ), |
| 280 | + ( |
| 281 | + "vim:tw=78:sw=2:ts=2:ft=help:norl:nowrap:", |
| 282 | + Modeline { |
| 283 | + language: Some("help".to_string()), |
| 284 | + indent_style: Some(IndentStyle::Spaces(2)), |
| 285 | + ..Default::default() |
| 286 | + }, |
| 287 | + ), |
| 288 | + ( |
| 289 | + "# vim: ft=zsh sw=2 ts=2 et", |
| 290 | + Modeline { |
| 291 | + language: Some("zsh".to_string()), |
| 292 | + indent_style: Some(IndentStyle::Spaces(2)), |
| 293 | + ..Default::default() |
| 294 | + }, |
| 295 | + ), |
| 296 | + ( |
| 297 | + "# vim:ft=sh:", |
| 298 | + Modeline { |
| 299 | + language: Some("sh".to_string()), |
| 300 | + ..Default::default() |
| 301 | + }, |
| 302 | + ), |
| 303 | + ( |
| 304 | + "\" vim:ts=8:sts=4:sw=4:expandtab:ft=vim", |
| 305 | + Modeline { |
| 306 | + language: Some("vim".to_string()), |
| 307 | + indent_style: Some(IndentStyle::Spaces(4)), |
| 308 | + ..Default::default() |
| 309 | + }, |
| 310 | + ), |
| 311 | + ( |
| 312 | + "\" vim: ts=8 noet tw=100 sw=8 sts=0 ft=vim isk+=-", |
| 313 | + Modeline { |
| 314 | + language: Some("vim".to_string()), |
| 315 | + indent_style: Some(IndentStyle::Tabs), |
| 316 | + ..Default::default() |
| 317 | + }, |
| 318 | + ), |
| 319 | + ( |
| 320 | + "; vim:ft=gitconfig:", |
| 321 | + Modeline { |
| 322 | + language: Some("gitconfig".to_string()), |
| 323 | + ..Default::default() |
| 324 | + }, |
| 325 | + ), |
| 326 | + ]; |
| 327 | + for (line, expected) in tests { |
| 328 | + let mut got = Modeline::default(); |
| 329 | + got.parse_from_line(line.into()); |
| 330 | + assert_eq!(got, expected, "{line}"); |
| 331 | + } |
| 332 | + } |
| 333 | +} |
0 commit comments