Add initial parsing

Signed-off-by: Marcel Müller <neikos@neikos.email>

Cargo.lock

@@ -97,6 +97,12 @@ version = "2.9.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "1b8e56985ec62d17e9c1001dc89c88ecd7dc08e47eba5ec7c29c7b5eeecde967"
 
+[[package]]
+name = "camino"
+version = "1.1.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0da45bc31171d8d6960122e222a67740df867c1dd53b4d51caa297084c185cab"
+
 [[package]]
 name = "cfg-if"
 version = "1.0.1"
@@ -193,6 +199,7 @@ checksum = "2304e00983f87ffb38b55b444b5e3b60a884b5d30c0fca7d82fe33449bbe55ea"
 name = "hem"
 version = "0.1.0"
 dependencies = [
+ "camino",
  "clap",
  "insta",
  "miette",

Cargo.toml

@@ -5,6 +5,7 @@ edition = "2024"
 description = "Hemera's Expression Manipulator, editing text on the CLI"
 
 [dependencies]
+camino = "1.1.10"
 clap = { version = "4.5.40", features = ["derive"] }
 insta = "1.43.1"
 miette = { version = "7.6.0", features = ["fancy"] }

flake.nix

@@ -100,6 +100,7 @@
             rustfmt'
             rustTarget
 
+            pkgs.cargo-insta
             inputs.cargo-changelog.packages.${system}.default
           ];
         };

src/cli.rs

@@ -0,0 +1,49 @@
+use camino::Utf8PathBuf;
+use clap::Parser;
+
+#[derive(Debug, Parser)]
+pub struct Args {
+    #[clap(short, long)]
+    pub expression: String,
+
+    #[clap(short, long, default_value_t = Utf8PathBuf::from("-"))]
+    input: Utf8PathBuf,
+
+    #[clap(short, long, group = "delimiters")]
+    lines: bool,
+
+    #[clap(short, long, group = "delimiters", default_value_t = true)]
+    words: bool,
+}
+
+pub enum InputDelimiter {
+    Lines,
+    Words,
+}
+
+pub enum Input {
+    Stdin,
+    FilePath(Utf8PathBuf),
+}
+
+impl Args {
+    pub fn delimiter(&self) -> InputDelimiter {
+        if self.lines {
+            return InputDelimiter::Lines;
+        }
+
+        if self.words {
+            return InputDelimiter::Words;
+        }
+
+        unreachable!("Either lines or words has to be true")
+    }
+
+    pub fn input(&self) -> Input {
+        if self.input == "-" {
+            Input::Stdin
+        } else {
+            Input::FilePath(self.input.clone())
+        }
+    }
+}

src/expr.rs

@@ -0,0 +1,173 @@
+use std::ops::Range;
+
+use miette::LabeledSpan;
+use winnow::LocatingSlice;
+use winnow::ModalResult;
+use winnow::Parser;
+use winnow::ascii::escaped;
+use winnow::ascii::space0;
+use winnow::ascii::space1;
+use winnow::combinator::alt;
+use winnow::combinator::delimited;
+use winnow::combinator::separated;
+use winnow::error::ContextError;
+use winnow::error::StrContext;
+use winnow::token::none_of;
+
+pub fn parse(src: &str) -> miette::Result<TokenList<'_>> {
+    Ok(TokenList {
+        expressions: parse_program
+            .parse(LocatingSlice::new(src))
+            .map_err(|parse_error| {
+                let labels = vec![LabeledSpan::new(
+                    Some("Here".to_string()),
+                    parse_error.offset(),
+                    0,
+                )];
+
+                miette::diagnostic!(
+                    labels = labels,
+                    "Could not parse expression: {:?}",
+                    parse_error
+                )
+            })?,
+        src,
+    })
+}
+
+type Input<'p> = LocatingSlice<&'p str>;
+type Error = ContextError;
+
+#[derive(Debug, Clone)]
+pub struct TokenList<'p> {
+    expressions: Vec<Token<'p>>,
+    src: &'p str,
+}
+
+#[derive(Debug, Clone)]
+pub struct Token<'p> {
+    span: Range<usize>,
+    expr: Tok<'p>,
+}
+
+#[derive(Debug, Clone)]
+pub enum Tok<'p> {
+    Print,
+    PerWord,
+    PerLine,
+    Match,
+    FunctionApplication,
+    String(String),
+    Dummy(&'p ()),
+}
+
+fn parse_program<'p>(input: &mut Input<'p>) -> ModalResult<Vec<Token<'p>>, Error> {
+    delimited(space0, separated(1.., parse_expression, space1), space0)
+        .context(StrContext::Label("program"))
+        .parse_next(input)
+}
+
+fn parse_expression<'p>(input: &mut Input<'p>) -> ModalResult<Token<'p>, Error> {
+    alt((parse_builtin, parse_string))
+        .context(StrContext::Label("expression"))
+        .parse_next(input)
+}
+
+fn parse_builtin<'p>(input: &mut Input<'p>) -> ModalResult<Token<'p>, Error> {
+    alt((
+        "@w".value(Tok::PerWord)
+            .context(winnow::error::StrContext::Label("@w")),
+        "@l".value(Tok::PerLine)
+            .context(winnow::error::StrContext::Label("@l")),
+        "print"
+            .value(Tok::Print)
+            .context(winnow::error::StrContext::Label("print")),
+        "match"
+            .value(Tok::Match)
+            .context(winnow::error::StrContext::Label("match")),
+        "|".value(Tok::FunctionApplication)
+            .context(winnow::error::StrContext::Label("|")),
+    ))
+    .with_span()
+    .map(|(expr, span)| Token { expr, span })
+    .parse_next(input)
+}
+
+fn parse_string<'p>(input: &mut Input<'p>) -> ModalResult<Token<'p>, Error> {
+    let content = escaped(
+        none_of(['\'', '\\']),
+        '\\',
+        alt(("\\".value("\\"), "\'".value("\'"))),
+    )
+    .map(Tok::String);
+
+    delimited('\'', content, '\'')
+        .with_span()
+        .map(|(expr, span)| Token { expr, span })
+        .parse_next(input)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn check_simple_print() {
+        let input = "@w print";
+
+        let expr = parse(input).unwrap();
+
+        insta::assert_debug_snapshot!(expr, @r#"
+        TokenList {
+            expressions: [
+                Token {
+                    span: 0..2,
+                    expr: PerWord,
+                },
+                Token {
+                    span: 3..8,
+                    expr: Print,
+                },
+            ],
+            src: "@w print",
+        }
+        "#);
+    }
+
+    #[test]
+    fn check_complex_print() {
+        let input = "@l match 'foo' | print";
+
+        let expr = parse(input).unwrap();
+
+        insta::assert_debug_snapshot!(expr, @r#"
+        TokenList {
+            expressions: [
+                Token {
+                    span: 0..2,
+                    expr: PerLine,
+                },
+                Token {
+                    span: 3..8,
+                    expr: Match,
+                },
+                Token {
+                    span: 9..14,
+                    expr: String(
+                        "foo",
+                    ),
+                },
+                Token {
+                    span: 15..16,
+                    expr: FunctionApplication,
+                },
+                Token {
+                    span: 17..22,
+                    expr: Print,
+                },
+            ],
+            src: "@l match 'foo' | print",
+        }
+        "#);
+    }
+}

src/main.rs

@@ -1,3 +1,14 @@
-fn main() {
-    println!("Hello, world!");
+use clap::Parser;
+
+mod cli;
+mod expr;
+
+fn main() -> miette::Result<()> {
+    let args = cli::Args::parse();
+    let input_delim = args.delimiter();
+    let input = args.input();
+
+    let expression = expr::parse(&args.expression)?;
+
+    Ok(())
 }