|
| 1 | +use std::io; |
| 2 | +use std::fmt; |
| 3 | +use std::fs::File; |
| 4 | +use std::io::Read; |
| 5 | + |
| 6 | +use regex; |
| 7 | +use regex::Regex; |
| 8 | + |
| 9 | +#[derive(Debug)] |
| 10 | +pub enum LicenseError { |
| 11 | + IO(io::Error), |
| 12 | + Regex(regex::Error), |
| 13 | + Parse(String), |
| 14 | +} |
| 15 | + |
| 16 | +impl fmt::Display for LicenseError { |
| 17 | + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
| 18 | + match *self { |
| 19 | + LicenseError::IO(ref err) => err.fmt(f), |
| 20 | + LicenseError::Regex(ref err) => err.fmt(f), |
| 21 | + LicenseError::Parse(ref err) => write!(f, "parsing failed, {}", err), |
| 22 | + } |
| 23 | + } |
| 24 | +} |
| 25 | + |
| 26 | +impl From<io::Error> for LicenseError { |
| 27 | + fn from(err: io::Error) -> LicenseError { |
| 28 | + LicenseError::IO(err) |
| 29 | + } |
| 30 | +} |
| 31 | + |
| 32 | +impl From<regex::Error> for LicenseError { |
| 33 | + fn from(err: regex::Error) -> LicenseError { |
| 34 | + LicenseError::Regex(err) |
| 35 | + } |
| 36 | +} |
| 37 | + |
| 38 | +// the template is parsed using a state machine |
| 39 | +enum ParsingState { |
| 40 | + Lit, |
| 41 | + LitEsc, |
| 42 | + // the u32 keeps track of brace nesting |
| 43 | + Re(u32), |
| 44 | + ReEsc(u32), |
| 45 | + Abort(String), |
| 46 | +} |
| 47 | + |
| 48 | +use self::ParsingState::*; |
| 49 | + |
| 50 | +pub struct TemplateParser { |
| 51 | + parsed: String, |
| 52 | + buffer: String, |
| 53 | + state: ParsingState, |
| 54 | + linum: u32, |
| 55 | + open_brace_line: u32, |
| 56 | +} |
| 57 | + |
| 58 | +impl TemplateParser { |
| 59 | + fn new() -> Self { |
| 60 | + Self { |
| 61 | + parsed: "^".to_owned(), |
| 62 | + buffer: String::new(), |
| 63 | + state: Lit, |
| 64 | + linum: 1, |
| 65 | + // keeps track of last line on which a regex placeholder was started |
| 66 | + open_brace_line: 0, |
| 67 | + } |
| 68 | + } |
| 69 | + |
| 70 | + /// Convert a license template into a string which can be turned into a regex. |
| 71 | + /// |
| 72 | + /// The license template could use regex syntax directly, but that would require a lot of manual |
| 73 | + /// escaping, which is inconvenient. It is therefore literal by default, with optional regex |
| 74 | + /// subparts delimited by `{` and `}`. Additionally: |
| 75 | + /// |
| 76 | + /// - to insert literal `{`, `}` or `\`, escape it with `\` |
| 77 | + /// - an empty regex placeholder (`{}`) is shorthand for `{.*?}` |
| 78 | + /// |
| 79 | + /// This function parses this input format and builds a properly escaped *string* representation |
| 80 | + /// of the equivalent regular expression. It **does not** however guarantee that the returned |
| 81 | + /// string is a syntactically valid regular expression. |
| 82 | + /// |
| 83 | + /// # Examples |
| 84 | + /// |
| 85 | + /// ``` |
| 86 | + /// # use rustfmt_nightly::config::license::TemplateParser; |
| 87 | + /// assert_eq!( |
| 88 | + /// TemplateParser::parse( |
| 89 | + /// r" |
| 90 | + /// // Copyright {\d+} The \} Rust \\ Project \{ Developers. See the {([A-Z]+)} |
| 91 | + /// // file at the top-level directory of this distribution and at |
| 92 | + /// // {}. |
| 93 | + /// // |
| 94 | + /// // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or |
| 95 | + /// // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license |
| 96 | + /// // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your |
| 97 | + /// // option. This file may not be copied, modified, or distributed |
| 98 | + /// // except according to those terms. |
| 99 | + /// " |
| 100 | + /// ).unwrap(), |
| 101 | + /// r"^ |
| 102 | + /// // Copyright \d+ The \} Rust \\ Project \{ Developers\. See the ([A-Z]+) |
| 103 | + /// // file at the top\-level directory of this distribution and at |
| 104 | + /// // .*?\. |
| 105 | + /// // |
| 106 | + /// // Licensed under the Apache License, Version 2\.0 <LICENSE\-APACHE or |
| 107 | + /// // http://www\.apache\.org/licenses/LICENSE\-2\.0> or the MIT license |
| 108 | + /// // <LICENSE\-MIT or http://opensource\.org/licenses/MIT>, at your |
| 109 | + /// // option\. This file may not be copied, modified, or distributed |
| 110 | + /// // except according to those terms\. |
| 111 | + /// " |
| 112 | + /// ); |
| 113 | + /// ``` |
| 114 | + pub fn parse(template: &str) -> Result<String, LicenseError> { |
| 115 | + let mut parser = Self::new(); |
| 116 | + for chr in template.chars() { |
| 117 | + if chr == '\n' { |
| 118 | + parser.linum += 1; |
| 119 | + } |
| 120 | + parser.state = match parser.state { |
| 121 | + Lit => parser.trans_from_lit(chr), |
| 122 | + LitEsc => parser.trans_from_litesc(chr), |
| 123 | + Re(brace_nesting) => parser.trans_from_re(chr, brace_nesting), |
| 124 | + ReEsc(brace_nesting) => parser.trans_from_reesc(chr, brace_nesting), |
| 125 | + Abort(msg) => return Err(LicenseError::Parse(msg)), |
| 126 | + }; |
| 127 | + } |
| 128 | + // check if we've ended parsing in a valid state |
| 129 | + match parser.state { |
| 130 | + Abort(msg) => return Err(LicenseError::Parse(msg)), |
| 131 | + Re(_) | ReEsc(_) => { |
| 132 | + return Err(LicenseError::Parse(format!( |
| 133 | + "escape or balance opening brace on l. {}", |
| 134 | + parser.open_brace_line |
| 135 | + ))); |
| 136 | + } |
| 137 | + LitEsc => { |
| 138 | + return Err(LicenseError::Parse(format!( |
| 139 | + "incomplete escape sequence on l. {}", |
| 140 | + parser.linum |
| 141 | + ))) |
| 142 | + } |
| 143 | + _ => (), |
| 144 | + } |
| 145 | + parser.parsed.push_str(®ex::escape(&parser.buffer)); |
| 146 | + |
| 147 | + Ok(parser.parsed) |
| 148 | + } |
| 149 | + |
| 150 | + fn trans_from_lit(&mut self, chr: char) -> ParsingState { |
| 151 | + match chr { |
| 152 | + '{' => { |
| 153 | + self.parsed.push_str(®ex::escape(&self.buffer)); |
| 154 | + self.buffer.clear(); |
| 155 | + self.open_brace_line = self.linum; |
| 156 | + Re(1) |
| 157 | + } |
| 158 | + '}' => Abort(format!( |
| 159 | + "escape or balance closing brace on l. {}", |
| 160 | + self.linum |
| 161 | + )), |
| 162 | + '\\' => LitEsc, |
| 163 | + _ => { |
| 164 | + self.buffer.push(chr); |
| 165 | + Lit |
| 166 | + } |
| 167 | + } |
| 168 | + } |
| 169 | + |
| 170 | + fn trans_from_litesc(&mut self, chr: char) -> ParsingState { |
| 171 | + self.buffer.push(chr); |
| 172 | + Lit |
| 173 | + } |
| 174 | + |
| 175 | + fn trans_from_re(&mut self, chr: char, brace_nesting: u32) -> ParsingState { |
| 176 | + match chr { |
| 177 | + '{' => { |
| 178 | + self.buffer.push(chr); |
| 179 | + Re(brace_nesting + 1) |
| 180 | + } |
| 181 | + '}' => { |
| 182 | + match brace_nesting { |
| 183 | + 1 => { |
| 184 | + // default regex for empty placeholder {} |
| 185 | + if self.buffer.is_empty() { |
| 186 | + self.parsed.push_str(".*?"); |
| 187 | + } else { |
| 188 | + self.parsed.push_str(&self.buffer); |
| 189 | + } |
| 190 | + self.buffer.clear(); |
| 191 | + Lit |
| 192 | + } |
| 193 | + _ => { |
| 194 | + self.buffer.push(chr); |
| 195 | + Re(brace_nesting - 1) |
| 196 | + } |
| 197 | + } |
| 198 | + } |
| 199 | + '\\' => { |
| 200 | + self.buffer.push(chr); |
| 201 | + ReEsc(brace_nesting) |
| 202 | + } |
| 203 | + _ => { |
| 204 | + self.buffer.push(chr); |
| 205 | + Re(brace_nesting) |
| 206 | + } |
| 207 | + } |
| 208 | + } |
| 209 | + |
| 210 | + fn trans_from_reesc(&mut self, chr: char, brace_nesting: u32) -> ParsingState { |
| 211 | + self.buffer.push(chr); |
| 212 | + Re(brace_nesting) |
| 213 | + } |
| 214 | +} |
| 215 | + |
| 216 | +pub fn load_and_compile_template(path: &str) -> Result<Regex, LicenseError> { |
| 217 | + let mut lt_file = File::open(&path)?; |
| 218 | + let mut lt_str = String::new(); |
| 219 | + lt_file.read_to_string(&mut lt_str)?; |
| 220 | + let lt_parsed = TemplateParser::parse(<_str)?; |
| 221 | + Ok(Regex::new(<_parsed)?) |
| 222 | +} |
| 223 | + |
| 224 | +#[cfg(test)] |
| 225 | +mod test { |
| 226 | + use super::TemplateParser; |
| 227 | + |
| 228 | + #[test] |
| 229 | + fn test_parse_license_template() { |
| 230 | + assert_eq!( |
| 231 | + TemplateParser::parse("literal (.*)").unwrap(), |
| 232 | + r"^literal \(\.\*\)" |
| 233 | + ); |
| 234 | + assert_eq!( |
| 235 | + TemplateParser::parse(r"escaping \}").unwrap(), |
| 236 | + r"^escaping \}" |
| 237 | + ); |
| 238 | + assert!(TemplateParser::parse("unbalanced } without escape").is_err()); |
| 239 | + assert_eq!( |
| 240 | + TemplateParser::parse(r"{\d+} place{-?}holder{s?}").unwrap(), |
| 241 | + r"^\d+ place-?holders?" |
| 242 | + ); |
| 243 | + assert_eq!(TemplateParser::parse("default {}").unwrap(), "^default .*?"); |
| 244 | + assert_eq!( |
| 245 | + TemplateParser::parse(r"unbalanced nested braces {\{{3}}").unwrap(), |
| 246 | + r"^unbalanced nested braces \{{3}" |
| 247 | + ); |
| 248 | + assert_eq!( |
| 249 | + &TemplateParser::parse("parsing error }") |
| 250 | + .unwrap_err() |
| 251 | + .to_string(), |
| 252 | + "parsing failed, escape or balance closing brace on l. 1" |
| 253 | + ); |
| 254 | + assert_eq!( |
| 255 | + &TemplateParser::parse("parsing error {\nsecond line") |
| 256 | + .unwrap_err() |
| 257 | + .to_string(), |
| 258 | + "parsing failed, escape or balance opening brace on l. 1" |
| 259 | + ); |
| 260 | + assert_eq!( |
| 261 | + &TemplateParser::parse(r"parsing error \") |
| 262 | + .unwrap_err() |
| 263 | + .to_string(), |
| 264 | + "parsing failed, incomplete escape sequence on l. 1" |
| 265 | + ); |
| 266 | + } |
| 267 | +} |
0 commit comments