Add compiler support for charms

Author: TheBerkin

.gitignore

@@ -1,2 +1,3 @@
 /target
 Cargo.lock
+debug.log

CHANGELOG.md

@@ -1,5 +1,13 @@
 # Changelog
 
+## 4.0.0-alpha.18 (unreleased)
+
+### New
+
+### Changes
+
+### Fixes
+
 ## 4.0.0-alpha.17
 
 ### New

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "rant"
-version = "4.0.0-alpha.17"
+version = "4.0.0-alpha.18"
 authors = ["Berkin <[email protected]>"]
 edition = "2018"
 description = "The Rant procedural templating language"
@@ -20,7 +20,7 @@ include = [
   "!tests/unincluded_*.rs"
 ]
 default-run = "rant"
-publish = false
+# publish = false
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 

README.md

@@ -84,7 +84,7 @@ Rant's CLI can run Rant code from files or the command line.
 Install it from Cargo with:
 
 ```sh
-$ cargo install rant --version 4.0.0-alpha.17 --features cli
+$ cargo install rant --version 4.0.0-alpha.18 --features cli
 ```
 
 Then run it:

src/compiler.rs

@@ -130,6 +130,8 @@ pub enum Problem {
   NestedFunctionDefMarkedConstant,
   ConstantReassignment(String),
   ConstantRedefinition(String),
+  InvalidKeyword(String),
+  WeightNotAllowed,
   FileNotFound(String),
   FileIOError(String),
 }
@@ -178,6 +180,10 @@ impl Problem {
       Problem::InvalidSink | Problem::InvalidSinkOn(_) =>         "R-0130",
       Problem::InvalidHint | Problem::InvalidHintOn(_) =>         "R-0131",
 
+      // Charms (0200 - 0249)
+      Problem::InvalidKeyword(_) =>                               "R-0200",
+      Problem::WeightNotAllowed =>                                "R-0201",
+
       // Common warnings (1000 - 1099)
       Problem::UnusedVariable(_) =>                               "R-1000",
       Problem::UnusedParameter(_) =>                              "R-1001",
@@ -230,6 +236,8 @@ impl Problem {
       Problem::NothingToCompose => "no compose value is available in this scope".to_owned(),
       Problem::ConstantReassignment(cname) => format!("reassignment of known constant '{}'", cname),
       Problem::ConstantRedefinition(cname) => format!("redefinition of known constant '{}'", cname),
+      Problem::InvalidKeyword(kw) => format!("invalid keyword: '@{}'", kw),
+      Problem::WeightNotAllowed => "@weight is not allowed in this context".to_owned(),
     }
   }
 

src/compiler/lexer.rs

@@ -3,129 +3,174 @@ use crate::InternalString;
 
 #[derive(Logos, Debug, PartialEq)]
 pub enum RantToken {
+  /// Sequence of printable non-whitespace characters
   #[error]
   #[regex(r"[\w\-_]+")]
   Fragment,
+
+  /// Sequence of printable whitespace characters
+  #[regex(r"\s+", filter_bs, priority = 2)]
+  Whitespace,
+  
+  /// Sequence of non-printable whitespace characters
+  #[regex(r"[\r\n]+\s*|\s*[\r\n]+", logos::skip, priority = 3)]
+  Blackspace,
 
+  /// `{`
   #[token("{")]
   LeftBrace,
 
+  /// `|`
   #[token("|")]
   Pipe,
 
+  /// `}`
   #[token("}")]
   RightBrace,
 
+  /// `|>`
   #[token("|>")]
   Compose,
 
+  /// `[]`
   #[token("[]")]
   ComposeValue,
 
+  /// `<>`
   #[token("<>")]
   Defer,
 
+  /// `[`
   #[token("[")]
   LeftBracket,
 
+  /// `]`
   #[token("]")]
   RightBracket,
 
+  /// `(`
   #[token("(")]
   LeftParen,
 
+  /// `)`
   #[token(")")]
   RightParen,
 
+  /// `~`
   #[token("~")]
   EmptyValue,
 
+  /// `<`
   #[token("<")]
   LeftAngle,
 
+  /// `>`
   #[token(">")]
   RightAngle,
 
+  /// `:`
   #[token(":")]
   Colon,
 
+  /// `**`
   #[token("**")]
   Temporal,
 
+  /// Labeled temporal operator, e.g. `*a*`
   #[regex(r"\*[\w\d\-_]+\*", parse_temporal_spread_label)]
   TemporalLabeled(InternalString),
 
+  /// `*`
   #[token("*")]
   Star,
 
+  /// `+`
   #[token("+")]
   Plus,
 
+  /// `&`
   #[token("&")]
   And,
 
+  /// `=`
   #[token("=")]
   Equals,
 
+  /// `!`
   #[token("!")]
   Bang,
 
+  /// `?`
   #[token("?")]
   Question,
 
+  /// `;`
   #[token(";")]
   Semi,
 
-  #[token("@")]
+  /// `@`
+  #[token("@", priority = 1)]
   At,
+
+  /// Some charm keyword, e.g. `@return`
+  #[regex(r"@[\w\d_-]+", parse_keyword, priority = 2, ignore(case))]
+  Charm(InternalString),
 
+  /// `/`
   #[token("/")]
   Slash,
 
+  /// `^`
   #[token("^")]
   Caret,
 
+  /// `$`
   #[token("$")]
   Dollar,
 
+  /// `%`
   #[token("%")]
   Percent,
 
+  /// `'`
   #[token("'")]
   Hint,
 
+  /// `_`
   #[token("_")]
   Sink,
 
+  /// `true`
   #[token("true")]
   True,
 
+  /// `false`
   #[token("false")]
   False,
 
-  #[regex(r"\s+", filter_bs, priority = 2)]
-  Whitespace,
-  
-  #[regex(r"[\r\n]+\s*|\s*[\r\n]+", logos::skip, priority = 3)]
-  Blackspace,
-  
+  /// Integer literal
   #[regex(r"\-?[0-9]+", parse_integer, priority = 2)]
   Integer(i64),
 
+  /// Float literal
   #[regex(r"\-?[0-9]+\.[0-9]+", parse_float, priority = 3)]
   Float(f64),
 
+  /// Represents inline and multi-line comments
   #[regex(r"\s*##([^#]|#[^#])*(##\s*)?", logos::skip, priority = 6)]
   #[regex(r"\s*#([^#][^\r\n]*)?\n?", logos::skip, priority = 5)]
   Comment,
 
+  /// Represents any escape sequence
   #[regex(r"\\\S", parse_escape, priority = 10)]
   #[regex(r"\\x[0-9a-fA-F][0-9a-fA-F]", parse_code_point_escape, priority = 7)]
   Escape(char),
 
+  /// Represents a verbatim string literal, e.g. `"hello world"`
   #[regex(r#""(""|[^"])*""#, parse_string_literal)]
   StringLiteral(InternalString),
 
+  /// Error token indicating an unterminated string literal, e.g. `"foo`
   #[regex(r#""(""|[^"])*"#)]
   UnterminatedStringLiteral,
 }
@@ -158,6 +203,12 @@ fn parse_string_literal(lex: &mut Lexer<RantToken>) -> Option<InternalString> {
   Some(string_content)
 }
 
+fn parse_keyword(lex: &mut Lexer<RantToken>) -> InternalString {
+  let kwd_literal = lex.slice();
+  let kwd_content = &kwd_literal[1..];
+  InternalString::from(kwd_content)
+}
+
 /// Filter function for whitespace lexer rule to exclude whitespace at start of source
 fn filter_bs(lex: &mut Lexer<RantToken>) -> Filter<()> {
   if lex.span().start > 0 {