Implement regular expression match conditionals (#970)

Cargo.toml

@@ -27,6 +27,7 @@ lazy_static   = "1.0.0"
 lexiclean = "0.0.1"
 libc = "0.2.0"
 log = "0.4.4"
+regex = "1.5.4"
 snafu = "0.6.0"
 strum_macros = "0.21.1"
 target = "2.0.0"
@@ -50,7 +51,6 @@ features = ["derive"]
 cradle = "0.0.22"
 executable-path = "1.0.0"
 pretty_assertions = "0.7.0"
-regex             = "1.5.4"
 temptree = "0.2.0"
 which = "4.0.0"
 yaml-rust = "0.4.5"

README.adoc

@@ -884,6 +884,22 @@ $ just bar
 xyz
 ```
 
+And match against regular expressions:
+
+```make
+foo := if "hello" =~ 'hel+o' { "match" } else { "mismatch" }
+
+bar:
+  @echo {{foo}}
+```
+
+```sh
+$ just bar
+match
+```
+
+Regular expressions are provided by the https://github.com/rust-lang/regex[regex crate], whose syntax is documented on https://docs.rs/regex/1.5.4/regex/#syntax[docs.rs]. Since regular expressions commonly use backslash escape sequences, consider using single-quoted string literals, which will pass slashes to the regex parser unmolested.
+
 Conditional expressions short-circuit, which means they only evaluate one of
 their branches. This can be used to make sure that backtick expressions don't
 run when they shouldn't.

src/common.rs

@@ -25,6 +25,7 @@ pub(crate) use edit_distance::edit_distance;
 pub(crate) use lexiclean::Lexiclean;
 pub(crate) use libc::EXIT_FAILURE;
 pub(crate) use log::{info, warn};
+pub(crate) use regex::Regex;
 pub(crate) use snafu::{ResultExt, Snafu};
 pub(crate) use strum::{Display, EnumString, IntoStaticStr};
 pub(crate) use typed_arena::Arena;
@@ -46,19 +47,20 @@ pub(crate) use crate::{
 pub(crate) use crate::{
   alias::Alias, analyzer::Analyzer, assignment::Assignment,
   assignment_resolver::AssignmentResolver, ast::Ast, binding::Binding, color::Color,
-  compile_error::CompileError, compile_error_kind::CompileErrorKind, config::Config,
-  config_error::ConfigError, count::Count, delimiter::Delimiter, dependency::Dependency,
-  enclosure::Enclosure, error::Error, evaluator::Evaluator, expression::Expression,
-  fragment::Fragment, function::Function, function_context::FunctionContext,
-  interrupt_guard::InterruptGuard, interrupt_handler::InterruptHandler, item::Item,
-  justfile::Justfile, keyword::Keyword, lexer::Lexer, line::Line, list::List, loader::Loader,
-  name::Name, output_error::OutputError, parameter::Parameter, parameter_kind::ParameterKind,
-  parser::Parser, platform::Platform, position::Position, positional::Positional, recipe::Recipe,
-  recipe_context::RecipeContext, recipe_resolver::RecipeResolver, scope::Scope, search::Search,
-  search_config::SearchConfig, search_error::SearchError, set::Set, setting::Setting,
-  settings::Settings, shebang::Shebang, show_whitespace::ShowWhitespace, string_kind::StringKind,
-  string_literal::StringLiteral, subcommand::Subcommand, suggestion::Suggestion, table::Table,
-  thunk::Thunk, token::Token, token_kind::TokenKind, unresolved_dependency::UnresolvedDependency,
+  compile_error::CompileError, compile_error_kind::CompileErrorKind,
+  conditional_operator::ConditionalOperator, config::Config, config_error::ConfigError,
+  count::Count, delimiter::Delimiter, dependency::Dependency, enclosure::Enclosure, error::Error,
+  evaluator::Evaluator, expression::Expression, fragment::Fragment, function::Function,
+  function_context::FunctionContext, interrupt_guard::InterruptGuard,
+  interrupt_handler::InterruptHandler, item::Item, justfile::Justfile, keyword::Keyword,
+  lexer::Lexer, line::Line, list::List, loader::Loader, name::Name, output_error::OutputError,
+  parameter::Parameter, parameter_kind::ParameterKind, parser::Parser, platform::Platform,
+  position::Position, positional::Positional, recipe::Recipe, recipe_context::RecipeContext,
+  recipe_resolver::RecipeResolver, scope::Scope, search::Search, search_config::SearchConfig,
+  search_error::SearchError, set::Set, setting::Setting, settings::Settings, shebang::Shebang,
+  show_whitespace::ShowWhitespace, string_kind::StringKind, string_literal::StringLiteral,
+  subcommand::Subcommand, suggestion::Suggestion, table::Table, thunk::Thunk, token::Token,
+  token_kind::TokenKind, unresolved_dependency::UnresolvedDependency,
   unresolved_recipe::UnresolvedRecipe, use_color::UseColor, variables::Variables,
   verbosity::Verbosity, warning::Warning,
 };

src/conditional_operator.rs

@@ -0,0 +1,22 @@
+use crate::common::*;
+
+/// A conditional expression operator.
+#[derive(PartialEq, Debug, Copy, Clone)]
+pub(crate) enum ConditionalOperator {
+  /// `==`
+  Equality,
+  /// `!=`
+  Inequality,
+  /// `=~`
+  RegexMatch,
+}
+
+impl Display for ConditionalOperator {
+  fn fmt(&self, f: &mut Formatter) -> fmt::Result {
+    match self {
+      Self::Equality => write!(f, "=="),
+      Self::Inequality => write!(f, "!="),
+      Self::RegexMatch => write!(f, "=~"),
+    }
+  }
+}

src/error.rs

@@ -95,6 +95,9 @@ pub(crate) enum Error<'src> {
   },
   NoChoosableRecipes,
   NoRecipes,
+  RegexCompile {
+    source: regex::Error,
+  },
   Search {
     search_error: SearchError,
   },
@@ -507,6 +510,9 @@ impl<'src> ColorDisplay for Error<'src> {
       NoRecipes => {
         write!(f, "Justfile contains no recipes.")?;
       }
+      RegexCompile { source } => {
+        write!(f, "{}", source)?;
+      }
       Search { search_error } => Display::fmt(search_error, f)?,
       Shebang {
         recipe,

src/evaluator.rs

@@ -139,11 +139,17 @@ impl<'src, 'run> Evaluator<'src, 'run> {
         rhs,
         then,
         otherwise,
-        inverted,
+        operator,
       } => {
-        let lhs = self.evaluate_expression(lhs)?;
-        let rhs = self.evaluate_expression(rhs)?;
-        let condition = if *inverted { lhs != rhs } else { lhs == rhs };
+        let lhs_value = self.evaluate_expression(lhs)?;
+        let rhs_value = self.evaluate_expression(rhs)?;
+        let condition = match operator {
+          ConditionalOperator::Equality => lhs_value == rhs_value,
+          ConditionalOperator::Inequality => lhs_value != rhs_value,
+          ConditionalOperator::RegexMatch => Regex::new(&rhs_value)
+            .map_err(|source| Error::RegexCompile { source })?
+            .is_match(&lhs_value),
+        };
         if condition {
           self.evaluate_expression(then)
         } else {

src/expression.rs

@@ -26,7 +26,7 @@ pub(crate) enum Expression<'src> {
     rhs: Box<Expression<'src>>,
     then: Box<Expression<'src>>,
     otherwise: Box<Expression<'src>>,
-    inverted: bool,
+    operator: ConditionalOperator,
   },
   /// `(contents)`
   Group { contents: Box<Expression<'src>> },
@@ -52,15 +52,11 @@ impl<'src> Display for Expression<'src> {
         rhs,
         then,
         otherwise,
-        inverted,
+        operator,
       } => write!(
         f,
         "if {} {} {} {{ {} }} else {{ {} }}",
-        lhs,
-        if *inverted { "!=" } else { "==" },
-        rhs,
-        then,
-        otherwise
+        lhs, operator, rhs, then, otherwise
       ),
       Expression::StringLiteral { string_literal } => write!(f, "{}", string_literal),
       Expression::Variable { name } => write!(f, "{}", name.lexeme()),

src/lexer.rs

@@ -475,25 +475,25 @@ impl<'src> Lexer<'src> {
   /// Lex token beginning with `start` outside of a recipe body
   fn lex_normal(&mut self, start: char) -> CompileResult<'src, ()> {
     match start {
-      '&' => self.lex_digraph('&', '&', AmpersandAmpersand),
+      ' ' | '\t' => self.lex_whitespace(),
       '!' => self.lex_digraph('!', '=', BangEquals),
-      '*' => self.lex_single(Asterisk),
+      '#' => self.lex_comment(),
       '$' => self.lex_single(Dollar),
-      '@' => self.lex_single(At),
-      '[' => self.lex_delimiter(BracketL),
-      ']' => self.lex_delimiter(BracketR),
-      '=' => self.lex_choice('=', EqualsEquals, Equals),
-      ',' => self.lex_single(Comma),
-      ':' => self.lex_colon(),
+      '&' => self.lex_digraph('&', '&', AmpersandAmpersand),
       '(' => self.lex_delimiter(ParenL),
       ')' => self.lex_delimiter(ParenR),
+      '*' => self.lex_single(Asterisk),
+      '+' => self.lex_single(Plus),
+      ',' => self.lex_single(Comma),
+      ':' => self.lex_colon(),
+      '=' => self.lex_choices('=', &[('=', EqualsEquals), ('~', EqualsTilde)], Equals),
+      '@' => self.lex_single(At),
+      '[' => self.lex_delimiter(BracketL),
+      '\n' | '\r' => self.lex_eol(),
+      ']' => self.lex_delimiter(BracketR),
+      '`' | '"' | '\'' => self.lex_string(),
       '{' => self.lex_delimiter(BraceL),
       '}' => self.lex_delimiter(BraceR),
-      '+' => self.lex_single(Plus),
-      '#' => self.lex_comment(),
-      ' ' | '\t' => self.lex_whitespace(),
-      '`' | '"' | '\'' => self.lex_string(),
-      '\n' | '\r' => self.lex_eol(),
       _ if Self::is_identifier_start(start) => self.lex_identifier(),
       _ => {
         self.advance()?;
@@ -610,19 +610,22 @@ impl<'src> Lexer<'src> {
   /// Lex a double-character token of kind `then` if the second character of
   /// that token would be `second`, otherwise lex a single-character token of
   /// kind `otherwise`
-  fn lex_choice(
+  fn lex_choices(
     &mut self,
-    second: char,
-    then: TokenKind,
+    first: char,
+    choices: &[(char, TokenKind)],
     otherwise: TokenKind,
   ) -> CompileResult<'src, ()> {
-    self.advance()?;
+    self.presume(first)?;
 
-    if self.accepted(second)? {
-      self.token(then);
-    } else {
-      self.token(otherwise);
+    for (second, then) in choices {
+      if self.accepted(*second)? {
+        self.token(*then);
+        return Ok(());
       }
+    }
+
+    self.token(otherwise);
 
     Ok(())
   }
@@ -930,6 +933,7 @@ mod tests {
       Eol => "\n",
       Equals => "=",
       EqualsEquals => "==",
+      EqualsTilde => "=~",
       Indent => "  ",
       InterpolationEnd => "}}",
       InterpolationStart => "{{",
@@ -2054,7 +2058,7 @@ mod tests {
 
   error! {
     name:   tokenize_unknown,
-    input:  "~",
+    input:  "%",
     offset: 0,
     line:   0,
     column: 0,
@@ -2113,16 +2117,6 @@ mod tests {
     kind:   UnpairedCarriageReturn,
   }
 
-  error! {
-    name:   unknown_start_of_token_tilde,
-    input:  "~",
-    offset: 0,
-    line:   0,
-    column: 0,
-    width:  1,
-    kind:   UnknownStartOfToken,
-  }
-
   error! {
     name:   invalid_name_start_dash,
     input:  "-foo",

src/lib.rs

@@ -43,6 +43,7 @@ mod compile_error;
 mod compile_error_kind;
 mod compiler;
 mod completions;
+mod conditional_operator;
 mod config;
 mod config_error;
 mod count;

src/node.rs

@@ -58,15 +58,11 @@ impl<'src> Node<'src> for Expression<'src> {
         rhs,
         then,
         otherwise,
-        inverted,
+        operator,
       } => {
         let mut tree = Tree::atom(Keyword::If.lexeme());
         tree.push_mut(lhs.tree());
-        if *inverted {
-          tree.push_mut("!=");
-        } else {
-          tree.push_mut("==");
-        }
+        tree.push_mut(operator.to_string());
         tree.push_mut(rhs.tree());
         tree.push_mut(then.tree());
         tree.push_mut(otherwise.tree());

src/parser.rs

@@ -419,11 +419,14 @@ impl<'tokens, 'src> Parser<'tokens, 'src> {
   fn parse_conditional(&mut self) -> CompileResult<'src, Expression<'src>> {
     let lhs = self.parse_expression()?;
 
-    let inverted = self.accepted(BangEquals)?;
-
-    if !inverted {
+    let operator = if self.accepted(BangEquals)? {
+      ConditionalOperator::Inequality
+    } else if self.accepted(EqualsTilde)? {
+      ConditionalOperator::RegexMatch
+    } else {
       self.expect(EqualsEquals)?;
-    }
+      ConditionalOperator::Equality
+    };
 
     let rhs = self.parse_expression()?;
 
@@ -449,7 +452,7 @@ impl<'tokens, 'src> Parser<'tokens, 'src> {
       rhs: Box::new(rhs),
       then: Box::new(then),
       otherwise: Box::new(otherwise),
-      inverted,
+      operator,
     })
   }
 

src/summary.rs

@@ -18,9 +18,9 @@ use crate::compiler::Compiler;
 
 mod full {
   pub(crate) use crate::{
-    assignment::Assignment, dependency::Dependency, expression::Expression, fragment::Fragment,
-    justfile::Justfile, line::Line, parameter::Parameter, parameter_kind::ParameterKind,
-    recipe::Recipe, thunk::Thunk,
+    assignment::Assignment, conditional_operator::ConditionalOperator, dependency::Dependency,
+    expression::Expression, fragment::Fragment, justfile::Justfile, line::Line,
+    parameter::Parameter, parameter_kind::ParameterKind, recipe::Recipe, thunk::Thunk,
   };
 }
 
@@ -198,7 +198,7 @@ pub enum Expression {
     rhs: Box<Expression>,
     then: Box<Expression>,
     otherwise: Box<Expression>,
-    inverted: bool,
+    operator: ConditionalOperator,
   },
   String {
     text: String,
@@ -245,16 +245,16 @@ impl Expression {
       },
       Conditional {
         lhs,
-        rhs,
-        inverted,
-        then,
+        operator,
         otherwise,
+        rhs,
+        then,
       } => Expression::Conditional {
         lhs: Box::new(Expression::new(lhs)),
+        operator: ConditionalOperator::new(*operator),
+        otherwise: Box::new(Expression::new(otherwise)),
         rhs: Box::new(Expression::new(rhs)),
         then: Box::new(Expression::new(then)),
-        otherwise: Box::new(Expression::new(otherwise)),
-        inverted: *inverted,
       },
       StringLiteral { string_literal } => Expression::String {
         text: string_literal.cooked.clone(),
@@ -267,6 +267,23 @@ impl Expression {
   }
 }
 
+#[derive(Eq, PartialEq, Hash, Ord, PartialOrd, Debug, Clone)]
+pub enum ConditionalOperator {
+  Equality,
+  Inequality,
+  RegexMatch,
+}
+
+impl ConditionalOperator {
+  fn new(operator: full::ConditionalOperator) -> Self {
+    match operator {
+      full::ConditionalOperator::Equality => Self::Equality,
+      full::ConditionalOperator::Inequality => Self::Inequality,
+      full::ConditionalOperator::RegexMatch => Self::RegexMatch,
+    }
+  }
+}
+
 #[derive(Eq, PartialEq, Hash, Ord, PartialOrd, Debug, Clone)]
 pub struct Dependency {
   pub recipe: String,
@@ -274,8 +291,8 @@ pub struct Dependency {
 }
 
 impl Dependency {
-  fn new(dependency: &full::Dependency) -> Dependency {
-    Dependency {
+  fn new(dependency: &full::Dependency) -> Self {
+    Self {
       recipe: dependency.recipe.name().to_owned(),
       arguments: dependency.arguments.iter().map(Expression::new).collect(),
     }

src/token_kind.rs

@@ -21,6 +21,7 @@ pub(crate) enum TokenKind {
   Eol,
   Equals,
   EqualsEquals,
+  EqualsTilde,
   Identifier,
   Indent,
   InterpolationEnd,
@@ -69,6 +70,7 @@ impl Display for TokenKind {
         Plus => "'+'",
         StringToken => "string",
         Text => "command text",
+        EqualsTilde => "'=~'",
         Unspecified => "unspecified",
         Whitespace => "whitespace",
       }

tests/conditional.rs

@@ -132,7 +132,7 @@ test! {
   ",
   stdout: "",
   stderr: "
-    error: Expected '!=', '==', or '+', but found identifier
+    error: Expected '!=', '==', '=~', or '+', but found identifier
       |
     1 | a := if '' a '' { '' } else { b }
       |            ^

tests/lib.rs

@@ -25,6 +25,7 @@ mod misc;
 mod positional_arguments;
 mod quiet;
 mod readme;
+mod regexes;
 mod search;
 mod shebang;
 mod shell;

tests/regexes.rs

@@ -0,0 +1,66 @@
+use crate::common::*;
+
+#[test]
+fn match_succeeds_evaluates_to_first_branch() {
+  Test::new()
+    .justfile(
+      "
+      foo := if 'abbbc' =~ 'ab+c' {
+        'yes'
+      } else {
+        'no'
+      }
+
+      default:
+        echo {{ foo }}
+    ",
+    )
+    .stderr("echo yes\n")
+    .stdout("yes\n")
+    .run();
+}
+
+#[test]
+fn match_fails_evaluates_to_second_branch() {
+  Test::new()
+    .justfile(
+      "
+      foo := if 'abbbc' =~ 'ab{4}c' {
+        'yes'
+      } else {
+        'no'
+      }
+
+      default:
+        echo {{ foo }}
+    ",
+    )
+    .stderr("echo no\n")
+    .stdout("no\n")
+    .run();
+}
+
+#[test]
+fn bad_regex_fails_at_runtime() {
+  Test::new()
+    .justfile(
+      "
+        default:
+          echo before
+          echo {{ if '' =~ '(' { 'a' } else { 'b' } }}
+          echo after
+      ",
+    )
+    .stderr(
+      "
+        echo before
+        error: regex parse error:
+            (
+            ^
+        error: unclosed group
+      ",
+    )
+    .stdout("before\n")
+    .status(EXIT_FAILURE)
+    .run();
+}