Chih-Hung Hsieh | e42c505 | 2020-04-16 10:44:21 -0700 | [diff] [blame] | 1 | /// The set of user configurable options for compiling zero or more regexes. |
| 2 | #[derive(Clone, Debug)] |
| 3 | #[allow(missing_docs)] |
| 4 | pub struct RegexOptions { |
| 5 | pub pats: Vec<String>, |
| 6 | pub size_limit: usize, |
| 7 | pub dfa_size_limit: usize, |
| 8 | pub nest_limit: u32, |
| 9 | pub case_insensitive: bool, |
| 10 | pub multi_line: bool, |
| 11 | pub dot_matches_new_line: bool, |
| 12 | pub swap_greed: bool, |
| 13 | pub ignore_whitespace: bool, |
| 14 | pub unicode: bool, |
| 15 | pub octal: bool, |
| 16 | } |
| 17 | |
| 18 | impl Default for RegexOptions { |
| 19 | fn default() -> Self { |
| 20 | RegexOptions { |
| 21 | pats: vec![], |
| 22 | size_limit: 10 * (1 << 20), |
| 23 | dfa_size_limit: 2 * (1 << 20), |
| 24 | nest_limit: 250, |
| 25 | case_insensitive: false, |
| 26 | multi_line: false, |
| 27 | dot_matches_new_line: false, |
| 28 | swap_greed: false, |
| 29 | ignore_whitespace: false, |
| 30 | unicode: true, |
| 31 | octal: false, |
| 32 | } |
| 33 | } |
| 34 | } |
| 35 | |
| 36 | macro_rules! define_builder { |
| 37 | ($name:ident, $regex_mod:ident, $only_utf8:expr) => { |
| 38 | pub mod $name { |
| 39 | use super::RegexOptions; |
Joel Galenson | 3874808 | 2021-05-19 16:51:51 -0700 | [diff] [blame^] | 40 | use crate::error::Error; |
| 41 | use crate::exec::ExecBuilder; |
Chih-Hung Hsieh | e42c505 | 2020-04-16 10:44:21 -0700 | [diff] [blame] | 42 | |
Joel Galenson | 3874808 | 2021-05-19 16:51:51 -0700 | [diff] [blame^] | 43 | use crate::$regex_mod::Regex; |
Chih-Hung Hsieh | e42c505 | 2020-04-16 10:44:21 -0700 | [diff] [blame] | 44 | |
| 45 | /// A configurable builder for a regular expression. |
| 46 | /// |
| 47 | /// A builder can be used to configure how the regex is built, for example, by |
| 48 | /// setting the default flags (which can be overridden in the expression |
| 49 | /// itself) or setting various limits. |
Haibo Huang | 47619dd | 2021-01-08 17:05:43 -0800 | [diff] [blame] | 50 | #[derive(Debug)] |
Chih-Hung Hsieh | e42c505 | 2020-04-16 10:44:21 -0700 | [diff] [blame] | 51 | pub struct RegexBuilder(RegexOptions); |
| 52 | |
| 53 | impl RegexBuilder { |
| 54 | /// Create a new regular expression builder with the given pattern. |
| 55 | /// |
| 56 | /// If the pattern is invalid, then an error will be returned when |
| 57 | /// `build` is called. |
| 58 | pub fn new(pattern: &str) -> RegexBuilder { |
| 59 | let mut builder = RegexBuilder(RegexOptions::default()); |
| 60 | builder.0.pats.push(pattern.to_owned()); |
| 61 | builder |
| 62 | } |
| 63 | |
| 64 | /// Consume the builder and compile the regular expression. |
| 65 | /// |
| 66 | /// Note that calling `as_str` on the resulting `Regex` will produce the |
| 67 | /// pattern given to `new` verbatim. Notably, it will not incorporate any |
| 68 | /// of the flags set on this builder. |
| 69 | pub fn build(&self) -> Result<Regex, Error> { |
| 70 | ExecBuilder::new_options(self.0.clone()) |
| 71 | .only_utf8($only_utf8) |
| 72 | .build() |
| 73 | .map(Regex::from) |
| 74 | } |
| 75 | |
| 76 | /// Set the value for the case insensitive (`i`) flag. |
| 77 | /// |
| 78 | /// When enabled, letters in the pattern will match both upper case and |
| 79 | /// lower case variants. |
| 80 | pub fn case_insensitive( |
| 81 | &mut self, |
| 82 | yes: bool, |
| 83 | ) -> &mut RegexBuilder { |
| 84 | self.0.case_insensitive = yes; |
| 85 | self |
| 86 | } |
| 87 | |
| 88 | /// Set the value for the multi-line matching (`m`) flag. |
| 89 | /// |
| 90 | /// When enabled, `^` matches the beginning of lines and `$` matches the |
| 91 | /// end of lines. |
| 92 | /// |
| 93 | /// By default, they match beginning/end of the input. |
| 94 | pub fn multi_line(&mut self, yes: bool) -> &mut RegexBuilder { |
| 95 | self.0.multi_line = yes; |
| 96 | self |
| 97 | } |
| 98 | |
| 99 | /// Set the value for the any character (`s`) flag, where in `.` matches |
| 100 | /// anything when `s` is set and matches anything except for new line when |
| 101 | /// it is not set (the default). |
| 102 | /// |
| 103 | /// N.B. "matches anything" means "any byte" when Unicode is disabled and |
| 104 | /// means "any valid UTF-8 encoding of any Unicode scalar value" when |
| 105 | /// Unicode is enabled. |
| 106 | pub fn dot_matches_new_line( |
| 107 | &mut self, |
| 108 | yes: bool, |
| 109 | ) -> &mut RegexBuilder { |
| 110 | self.0.dot_matches_new_line = yes; |
| 111 | self |
| 112 | } |
| 113 | |
| 114 | /// Set the value for the greedy swap (`U`) flag. |
| 115 | /// |
| 116 | /// When enabled, a pattern like `a*` is lazy (tries to find shortest |
| 117 | /// match) and `a*?` is greedy (tries to find longest match). |
| 118 | /// |
| 119 | /// By default, `a*` is greedy and `a*?` is lazy. |
| 120 | pub fn swap_greed(&mut self, yes: bool) -> &mut RegexBuilder { |
| 121 | self.0.swap_greed = yes; |
| 122 | self |
| 123 | } |
| 124 | |
| 125 | /// Set the value for the ignore whitespace (`x`) flag. |
| 126 | /// |
| 127 | /// When enabled, whitespace such as new lines and spaces will be ignored |
| 128 | /// between expressions of the pattern, and `#` can be used to start a |
| 129 | /// comment until the next new line. |
| 130 | pub fn ignore_whitespace( |
| 131 | &mut self, |
| 132 | yes: bool, |
| 133 | ) -> &mut RegexBuilder { |
| 134 | self.0.ignore_whitespace = yes; |
| 135 | self |
| 136 | } |
| 137 | |
| 138 | /// Set the value for the Unicode (`u`) flag. |
| 139 | /// |
| 140 | /// Enabled by default. When disabled, character classes such as `\w` only |
| 141 | /// match ASCII word characters instead of all Unicode word characters. |
| 142 | pub fn unicode(&mut self, yes: bool) -> &mut RegexBuilder { |
| 143 | self.0.unicode = yes; |
| 144 | self |
| 145 | } |
| 146 | |
| 147 | /// Whether to support octal syntax or not. |
| 148 | /// |
| 149 | /// Octal syntax is a little-known way of uttering Unicode codepoints in |
| 150 | /// a regular expression. For example, `a`, `\x61`, `\u0061` and |
| 151 | /// `\141` are all equivalent regular expressions, where the last example |
| 152 | /// shows octal syntax. |
| 153 | /// |
| 154 | /// While supporting octal syntax isn't in and of itself a problem, it does |
| 155 | /// make good error messages harder. That is, in PCRE based regex engines, |
| 156 | /// syntax like `\0` invokes a backreference, which is explicitly |
| 157 | /// unsupported in Rust's regex engine. However, many users expect it to |
| 158 | /// be supported. Therefore, when octal support is disabled, the error |
| 159 | /// message will explicitly mention that backreferences aren't supported. |
| 160 | /// |
| 161 | /// Octal syntax is disabled by default. |
| 162 | pub fn octal(&mut self, yes: bool) -> &mut RegexBuilder { |
| 163 | self.0.octal = yes; |
| 164 | self |
| 165 | } |
| 166 | |
| 167 | /// Set the approximate size limit of the compiled regular expression. |
| 168 | /// |
| 169 | /// This roughly corresponds to the number of bytes occupied by a single |
| 170 | /// compiled program. If the program exceeds this number, then a |
| 171 | /// compilation error is returned. |
| 172 | pub fn size_limit( |
| 173 | &mut self, |
| 174 | limit: usize, |
| 175 | ) -> &mut RegexBuilder { |
| 176 | self.0.size_limit = limit; |
| 177 | self |
| 178 | } |
| 179 | |
| 180 | /// Set the approximate size of the cache used by the DFA. |
| 181 | /// |
| 182 | /// This roughly corresponds to the number of bytes that the DFA will |
| 183 | /// use while searching. |
| 184 | /// |
| 185 | /// Note that this is a *per thread* limit. There is no way to set a global |
| 186 | /// limit. In particular, if a regex is used from multiple threads |
| 187 | /// simultaneously, then each thread may use up to the number of bytes |
| 188 | /// specified here. |
| 189 | pub fn dfa_size_limit( |
| 190 | &mut self, |
| 191 | limit: usize, |
| 192 | ) -> &mut RegexBuilder { |
| 193 | self.0.dfa_size_limit = limit; |
| 194 | self |
| 195 | } |
| 196 | |
| 197 | /// Set the nesting limit for this parser. |
| 198 | /// |
| 199 | /// The nesting limit controls how deep the abstract syntax tree is allowed |
| 200 | /// to be. If the AST exceeds the given limit (e.g., with too many nested |
| 201 | /// groups), then an error is returned by the parser. |
| 202 | /// |
| 203 | /// The purpose of this limit is to act as a heuristic to prevent stack |
| 204 | /// overflow for consumers that do structural induction on an `Ast` using |
| 205 | /// explicit recursion. While this crate never does this (instead using |
| 206 | /// constant stack space and moving the call stack to the heap), other |
| 207 | /// crates may. |
| 208 | /// |
| 209 | /// This limit is not checked until the entire Ast is parsed. Therefore, |
| 210 | /// if callers want to put a limit on the amount of heap space used, then |
| 211 | /// they should impose a limit on the length, in bytes, of the concrete |
| 212 | /// pattern string. In particular, this is viable since this parser |
| 213 | /// implementation will limit itself to heap space proportional to the |
| 214 | /// length of the pattern string. |
| 215 | /// |
| 216 | /// Note that a nest limit of `0` will return a nest limit error for most |
| 217 | /// patterns but not all. For example, a nest limit of `0` permits `a` but |
| 218 | /// not `ab`, since `ab` requires a concatenation, which results in a nest |
| 219 | /// depth of `1`. In general, a nest limit is not something that manifests |
| 220 | /// in an obvious way in the concrete syntax, therefore, it should not be |
| 221 | /// used in a granular way. |
| 222 | pub fn nest_limit(&mut self, limit: u32) -> &mut RegexBuilder { |
| 223 | self.0.nest_limit = limit; |
| 224 | self |
| 225 | } |
| 226 | } |
| 227 | } |
| 228 | }; |
| 229 | } |
| 230 | |
| 231 | define_builder!(bytes, re_bytes, false); |
| 232 | define_builder!(unicode, re_unicode, true); |
| 233 | |
| 234 | macro_rules! define_set_builder { |
| 235 | ($name:ident, $regex_mod:ident, $only_utf8:expr) => { |
| 236 | pub mod $name { |
| 237 | use super::RegexOptions; |
Joel Galenson | 3874808 | 2021-05-19 16:51:51 -0700 | [diff] [blame^] | 238 | use crate::error::Error; |
| 239 | use crate::exec::ExecBuilder; |
Chih-Hung Hsieh | e42c505 | 2020-04-16 10:44:21 -0700 | [diff] [blame] | 240 | |
Joel Galenson | 3874808 | 2021-05-19 16:51:51 -0700 | [diff] [blame^] | 241 | use crate::re_set::$regex_mod::RegexSet; |
Chih-Hung Hsieh | e42c505 | 2020-04-16 10:44:21 -0700 | [diff] [blame] | 242 | |
| 243 | /// A configurable builder for a set of regular expressions. |
| 244 | /// |
| 245 | /// A builder can be used to configure how the regexes are built, for example, |
| 246 | /// by setting the default flags (which can be overridden in the expression |
| 247 | /// itself) or setting various limits. |
Haibo Huang | 47619dd | 2021-01-08 17:05:43 -0800 | [diff] [blame] | 248 | #[derive(Debug)] |
Chih-Hung Hsieh | e42c505 | 2020-04-16 10:44:21 -0700 | [diff] [blame] | 249 | pub struct RegexSetBuilder(RegexOptions); |
| 250 | |
| 251 | impl RegexSetBuilder { |
| 252 | /// Create a new regular expression builder with the given pattern. |
| 253 | /// |
| 254 | /// If the pattern is invalid, then an error will be returned when |
| 255 | /// `build` is called. |
| 256 | pub fn new<I, S>(patterns: I) -> RegexSetBuilder |
| 257 | where |
| 258 | S: AsRef<str>, |
| 259 | I: IntoIterator<Item = S>, |
| 260 | { |
| 261 | let mut builder = RegexSetBuilder(RegexOptions::default()); |
| 262 | for pat in patterns { |
| 263 | builder.0.pats.push(pat.as_ref().to_owned()); |
| 264 | } |
| 265 | builder |
| 266 | } |
| 267 | |
| 268 | /// Consume the builder and compile the regular expressions into a set. |
| 269 | pub fn build(&self) -> Result<RegexSet, Error> { |
| 270 | ExecBuilder::new_options(self.0.clone()) |
| 271 | .only_utf8($only_utf8) |
| 272 | .build() |
| 273 | .map(RegexSet::from) |
| 274 | } |
| 275 | |
| 276 | /// Set the value for the case insensitive (`i`) flag. |
| 277 | pub fn case_insensitive( |
| 278 | &mut self, |
| 279 | yes: bool, |
| 280 | ) -> &mut RegexSetBuilder { |
| 281 | self.0.case_insensitive = yes; |
| 282 | self |
| 283 | } |
| 284 | |
| 285 | /// Set the value for the multi-line matching (`m`) flag. |
| 286 | pub fn multi_line( |
| 287 | &mut self, |
| 288 | yes: bool, |
| 289 | ) -> &mut RegexSetBuilder { |
| 290 | self.0.multi_line = yes; |
| 291 | self |
| 292 | } |
| 293 | |
| 294 | /// Set the value for the any character (`s`) flag, where in `.` matches |
| 295 | /// anything when `s` is set and matches anything except for new line when |
| 296 | /// it is not set (the default). |
| 297 | /// |
| 298 | /// N.B. "matches anything" means "any byte" for `regex::bytes::RegexSet` |
| 299 | /// expressions and means "any Unicode scalar value" for `regex::RegexSet` |
| 300 | /// expressions. |
| 301 | pub fn dot_matches_new_line( |
| 302 | &mut self, |
| 303 | yes: bool, |
| 304 | ) -> &mut RegexSetBuilder { |
| 305 | self.0.dot_matches_new_line = yes; |
| 306 | self |
| 307 | } |
| 308 | |
| 309 | /// Set the value for the greedy swap (`U`) flag. |
| 310 | pub fn swap_greed( |
| 311 | &mut self, |
| 312 | yes: bool, |
| 313 | ) -> &mut RegexSetBuilder { |
| 314 | self.0.swap_greed = yes; |
| 315 | self |
| 316 | } |
| 317 | |
| 318 | /// Set the value for the ignore whitespace (`x`) flag. |
| 319 | pub fn ignore_whitespace( |
| 320 | &mut self, |
| 321 | yes: bool, |
| 322 | ) -> &mut RegexSetBuilder { |
| 323 | self.0.ignore_whitespace = yes; |
| 324 | self |
| 325 | } |
| 326 | |
| 327 | /// Set the value for the Unicode (`u`) flag. |
| 328 | pub fn unicode(&mut self, yes: bool) -> &mut RegexSetBuilder { |
| 329 | self.0.unicode = yes; |
| 330 | self |
| 331 | } |
| 332 | |
| 333 | /// Whether to support octal syntax or not. |
| 334 | /// |
| 335 | /// Octal syntax is a little-known way of uttering Unicode codepoints in |
| 336 | /// a regular expression. For example, `a`, `\x61`, `\u0061` and |
| 337 | /// `\141` are all equivalent regular expressions, where the last example |
| 338 | /// shows octal syntax. |
| 339 | /// |
| 340 | /// While supporting octal syntax isn't in and of itself a problem, it does |
| 341 | /// make good error messages harder. That is, in PCRE based regex engines, |
| 342 | /// syntax like `\0` invokes a backreference, which is explicitly |
| 343 | /// unsupported in Rust's regex engine. However, many users expect it to |
| 344 | /// be supported. Therefore, when octal support is disabled, the error |
| 345 | /// message will explicitly mention that backreferences aren't supported. |
| 346 | /// |
| 347 | /// Octal syntax is disabled by default. |
| 348 | pub fn octal(&mut self, yes: bool) -> &mut RegexSetBuilder { |
| 349 | self.0.octal = yes; |
| 350 | self |
| 351 | } |
| 352 | |
| 353 | /// Set the approximate size limit of the compiled regular expression. |
| 354 | /// |
| 355 | /// This roughly corresponds to the number of bytes occupied by a single |
| 356 | /// compiled program. If the program exceeds this number, then a |
| 357 | /// compilation error is returned. |
| 358 | pub fn size_limit( |
| 359 | &mut self, |
| 360 | limit: usize, |
| 361 | ) -> &mut RegexSetBuilder { |
| 362 | self.0.size_limit = limit; |
| 363 | self |
| 364 | } |
| 365 | |
| 366 | /// Set the approximate size of the cache used by the DFA. |
| 367 | /// |
| 368 | /// This roughly corresponds to the number of bytes that the DFA will |
| 369 | /// use while searching. |
| 370 | /// |
| 371 | /// Note that this is a *per thread* limit. There is no way to set a global |
| 372 | /// limit. In particular, if a regex is used from multiple threads |
| 373 | /// simultaneously, then each thread may use up to the number of bytes |
| 374 | /// specified here. |
| 375 | pub fn dfa_size_limit( |
| 376 | &mut self, |
| 377 | limit: usize, |
| 378 | ) -> &mut RegexSetBuilder { |
| 379 | self.0.dfa_size_limit = limit; |
| 380 | self |
| 381 | } |
| 382 | |
| 383 | /// Set the nesting limit for this parser. |
| 384 | /// |
| 385 | /// The nesting limit controls how deep the abstract syntax tree is allowed |
| 386 | /// to be. If the AST exceeds the given limit (e.g., with too many nested |
| 387 | /// groups), then an error is returned by the parser. |
| 388 | /// |
| 389 | /// The purpose of this limit is to act as a heuristic to prevent stack |
| 390 | /// overflow for consumers that do structural induction on an `Ast` using |
| 391 | /// explicit recursion. While this crate never does this (instead using |
| 392 | /// constant stack space and moving the call stack to the heap), other |
| 393 | /// crates may. |
| 394 | /// |
| 395 | /// This limit is not checked until the entire Ast is parsed. Therefore, |
| 396 | /// if callers want to put a limit on the amount of heap space used, then |
| 397 | /// they should impose a limit on the length, in bytes, of the concrete |
| 398 | /// pattern string. In particular, this is viable since this parser |
| 399 | /// implementation will limit itself to heap space proportional to the |
| 400 | /// length of the pattern string. |
| 401 | /// |
| 402 | /// Note that a nest limit of `0` will return a nest limit error for most |
| 403 | /// patterns but not all. For example, a nest limit of `0` permits `a` but |
| 404 | /// not `ab`, since `ab` requires a concatenation, which results in a nest |
| 405 | /// depth of `1`. In general, a nest limit is not something that manifests |
| 406 | /// in an obvious way in the concrete syntax, therefore, it should not be |
| 407 | /// used in a granular way. |
| 408 | pub fn nest_limit( |
| 409 | &mut self, |
| 410 | limit: u32, |
| 411 | ) -> &mut RegexSetBuilder { |
| 412 | self.0.nest_limit = limit; |
| 413 | self |
| 414 | } |
| 415 | } |
| 416 | } |
| 417 | }; |
| 418 | } |
| 419 | |
| 420 | define_set_builder!(set_bytes, bytes, false); |
| 421 | define_set_builder!(set_unicode, unicode, true); |