Move line_numbers under 'aiken-lang'

And add some unit tests to ensure things are working properly.

crates/aiken-lang/src/lib.rs

@@ -10,6 +10,7 @@ pub mod expr;
 pub mod format;
 pub mod gen_uplc;
 pub mod levenshtein;
+pub mod line_numbers;
 pub mod parser;
 pub mod pretty;
 pub mod tipo;

crates/aiken-lang/src/line_numbers.rs

@@ -0,0 +1,126 @@
+use std::fmt::{self, Display};
+
+#[derive(Debug)]
+pub struct LineNumbers {
+    line_starts: Vec<usize>,
+    length: usize,
+    last: Option<usize>,
+}
+
+#[derive(Debug, PartialEq, Clone, Copy)]
+pub struct LineColumn {
+    pub line: usize,
+    pub column: usize,
+}
+
+impl Display for LineColumn {
+    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
+        f.write_str(&format!("L{};{}", self.line, self.column))
+    }
+}
+
+impl LineNumbers {
+    pub fn new(src: &str) -> Self {
+        let line_starts: Vec<usize> = std::iter::once(0)
+            .chain(src.match_indices('\n').map(|(i, _)| i + 1))
+            .collect();
+
+        let length = src.len();
+
+        Self {
+            length,
+            last: line_starts.last().cloned(),
+            line_starts: if length > 0 { line_starts } else { Vec::new() },
+        }
+    }
+
+    /// Get the line number for a byte index
+    pub fn line_number(&self, byte_index: usize) -> Option<usize> {
+        self.line_starts
+            .binary_search(&byte_index)
+            .map(|l| Some(l + 1))
+            .unwrap_or_else(|next_index| {
+                if Some(next_index) >= self.last {
+                    None
+                } else {
+                    Some(next_index)
+                }
+            })
+    }
+
+    pub fn line_and_column_number(&self, byte_index: usize) -> Option<LineColumn> {
+        let line = self.line_number(byte_index)?;
+        let column = byte_index - self.line_starts.get(line - 1).copied().unwrap_or_default() + 1;
+        Some(LineColumn { line, column })
+    }
+
+    #[allow(dead_code)]
+    pub fn byte_index(&self, line: usize, character: usize) -> usize {
+        match self.line_starts.get(line) {
+            Some(line_index) => *line_index + character,
+            None => self.length,
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use chumsky::text::Character;
+    use indoc::indoc;
+
+    fn assert_line_column(src: &str, ix: usize, lcol: Option<LineColumn>) {
+        let lines = LineNumbers::new(src);
+
+        println!("{lines:?}");
+
+        let byte = src
+            .as_bytes()
+            .get(ix)
+            .map(|b| {
+                if b.is_ascii() {
+                    format!("{}", b.to_char())
+                } else {
+                    format!("{b}")
+                }
+            })
+            .unwrap_or_else(|| "OUT-OF-BOUNDS".to_string());
+
+        assert_eq!(
+            lines.line_and_column_number(ix),
+            lcol,
+            "\n{src}\n--> at index {ix} ({byte})\n",
+        );
+    }
+
+    #[test]
+    fn out_of_range_byte_index() {
+        let src = indoc! { r#""# };
+        assert_line_column(src, 42, None);
+        assert_line_column(src, 0, None);
+    }
+
+    #[test]
+    fn basic() {
+        let src = indoc! { r#"
+            foo
+            bar
+        "# };
+
+        assert_line_column(src, 0, Some(LineColumn { line: 1, column: 1 }));
+        assert_line_column(src, 2, Some(LineColumn { line: 1, column: 3 }));
+        assert_line_column(src, 4, Some(LineColumn { line: 2, column: 1 }));
+    }
+
+    #[test]
+    fn unicode() {
+        let src = indoc! { r#"
+            💩
+            foo
+        "# };
+
+        assert_line_column(src, 0, Some(LineColumn { line: 1, column: 1 }));
+        assert_line_column(src, 2, Some(LineColumn { line: 1, column: 3 }));
+        assert_line_column(src, 5, Some(LineColumn { line: 2, column: 1 }));
+    }
+}

crates/aiken-lsp/src/edits.rs

@@ -1,5 +1,8 @@
-use crate::{line_numbers::LineNumbers, utils::span_to_lsp_range};
-use aiken_lang::ast::{Definition, ModuleKind, Span, UntypedDefinition, Use};
+use crate::utils::span_to_lsp_range;
+use aiken_lang::{
+    ast::{Definition, ModuleKind, Span, UntypedDefinition, Use},
+    line_numbers::LineNumbers,
+};
 use aiken_project::module::CheckedModule;
 use itertools::Itertools;
 use std::fs;

crates/aiken-lsp/src/lib.rs

@@ -7,7 +7,6 @@ use std::env;
 mod cast;
 mod edits;
 pub mod error;
-mod line_numbers;
 mod quickfix;
 pub mod server;
 mod utils;

crates/aiken-lsp/src/line_numbers.rs

@@ -1,48 +0,0 @@
-#[allow(dead_code)]
-#[derive(Debug)]
-pub struct LineNumbers {
-    line_starts: Vec<usize>,
-    length: usize,
-}
-
-impl LineNumbers {
-    pub fn new(src: &str) -> Self {
-        Self {
-            length: src.len(),
-            line_starts: std::iter::once(0)
-                .chain(src.match_indices('\n').map(|(i, _)| i + 1))
-                .collect(),
-        }
-    }
-
-    /// Get the line number for a byte index
-    pub fn line_number(&self, byte_index: usize) -> usize {
-        self.line_starts
-            .binary_search(&byte_index)
-            .unwrap_or_else(|next_line| next_line - 1)
-            + 1
-    }
-
-    // TODO: handle unicode characters that may be more than 1 byte in width
-    pub fn line_and_column_number(&self, byte_index: usize) -> LineColumn {
-        let line = self.line_number(byte_index);
-        let column = byte_index - self.line_starts.get(line - 1).copied().unwrap_or_default() + 1;
-        LineColumn { line, column }
-    }
-
-    // TODO: handle unicode characters that may be more than 1 byte in width
-    /// 0 indexed line and character to byte index
-    #[allow(dead_code)]
-    pub fn byte_index(&self, line: usize, character: usize) -> usize {
-        match self.line_starts.get(line) {
-            Some(line_index) => *line_index + character,
-            None => self.length,
-        }
-    }
-}
-
-#[derive(Debug, Clone, Copy)]
-pub struct LineColumn {
-    pub line: usize,
-    pub column: usize,
-}

crates/aiken-lsp/src/server.rs

@@ -8,6 +8,7 @@ use std::{
 use aiken_lang::{
     ast::{Definition, Located, ModuleKind, Span, Use},
     error::ExtraData,
+    line_numbers::LineNumbers,
     parser,
     tipo::pretty::Printer,
 };
@@ -35,7 +36,6 @@ use miette::Diagnostic;
 use crate::{
     cast::{cast_notification, cast_request},
     error::Error as ServerError,
-    line_numbers::LineNumbers,
     quickfix,
     utils::{
         path_to_uri, span_to_lsp_range, text_edit_replace, uri_to_module_name,

crates/aiken-lsp/src/server/lsp_project.rs

@@ -1,10 +1,8 @@
 use std::{collections::HashMap, path::PathBuf};
 
-use aiken_lang::ast::Tracing;
+use aiken_lang::{ast::Tracing, line_numbers::LineNumbers};
 use aiken_project::{config::Config, error::Error as ProjectError, module::CheckedModule, Project};
 
-use crate::line_numbers::LineNumbers;
-
 #[derive(Debug)]
 pub struct SourceInfo {
     /// The path to the source file from within the project root

crates/aiken-lsp/src/utils.rs

@@ -1,11 +1,11 @@
 use std::path::{Path, PathBuf};
 
-use aiken_lang::ast::Span;
+use aiken_lang::{ast::Span, line_numbers::LineNumbers};
 use itertools::Itertools;
 use lsp_types::TextEdit;
 use urlencoding::decode;
 
-use crate::{error::Error, line_numbers::LineNumbers};
+use crate::error::Error;
 
 pub const COMPILING_PROGRESS_TOKEN: &str = "compiling-aiken";
 pub const CREATE_COMPILING_PROGRESS_TOKEN: &str = "create-compiling-progress-token";
@@ -37,8 +37,12 @@ pub fn path_to_uri(path: PathBuf) -> Result<lsp_types::Url, Error> {
 }
 
 pub fn span_to_lsp_range(location: Span, line_numbers: &LineNumbers) -> lsp_types::Range {
-    let start = line_numbers.line_and_column_number(location.start);
-    let end = line_numbers.line_and_column_number(location.end);
+    let start = line_numbers
+        .line_and_column_number(location.start)
+        .expect("Spans are within bounds");
+    let end = line_numbers
+        .line_and_column_number(location.end)
+        .expect("Spans are within bounds");
 
     lsp_types::Range {
         start: lsp_types::Position {