Add test_compile crate

Cargo.lock

@@ -3803,6 +3803,25 @@ dependencies = [
  "winapi-util",
 ]
 
+[[package]]
+name = "test_compile"
+version = "0.0.1"
+dependencies = [
+ "bumpalo",
+ "pretty_assertions",
+ "roc_builtins",
+ "roc_can",
+ "roc_derive",
+ "roc_load",
+ "roc_parse",
+ "roc_problem",
+ "roc_region",
+ "roc_reporting",
+ "roc_solve",
+ "roc_target",
+ "test_solve_helpers",
+]
+
 [[package]]
 name = "test_derive"
 version = "0.0.1"

Cargo.toml

@@ -17,6 +17,7 @@ members = [
     "crates/repl_wasm",
     "crates/repl_expect",
     "crates/roc_std",
+    "crates/test_compile",
     "crates/test_utils",
     "crates/test_utils_dir",
     "crates/valgrind",

crates/compiler/parse/src/expr.rs

@@ -35,7 +35,7 @@ use roc_region::all::{Loc, Position, Region};
 
 use crate::parser::Progress::{self, *};
 
-fn expr_end<'a>() -> impl Parser<'a, (), EExpr<'a>> {
+pub fn expr_end<'a>() -> impl Parser<'a, (), EExpr<'a>> {
     |_arena, state: State<'a>, _min_indent: u32| {
         if state.has_reached_end() {
             Ok((NoProgress, (), state))

crates/test_compile/Cargo.toml

@@ -7,14 +7,22 @@ edition.workspace = true
 license.workspace = true
 version.workspace = true
 
+[dependencies]
+roc_builtins = { path = "../compiler/builtins" }
+roc_derive = { path = "../compiler/derive", features = [
+    "debug-derived-symbols",
+] }
+roc_region = { path = "../compiler/region" }
+roc_load = { path = "../compiler/load" }
+roc_parse = { path = "../compiler/parse" }
+roc_can = { path = "../compiler/can" }
+roc_problem = { path = "../compiler/problem" }
+roc_reporting = { path = "../reporting" }
+roc_target = { path = "../compiler/roc_target" }
+roc_solve = { path = "../compiler/solve" }
+test_solve_helpers = { path = "../compiler/test_solve_helpers" }
+
+bumpalo.workspace = true
+
 [dev-dependencies]
-roc_builtins = { path = "../builtins" }
-roc_derive = { path = "../derive", features = ["debug-derived-symbols"] }
-roc_load = { path = "../load" }
-roc_parse = { path = "../parse" }
-roc_problem = { path = "../problem" }
-roc_reporting = { path = "../../reporting" }
-roc_target = { path = "../roc_target" }
-roc_solve = { path = "../solve" }
-test_solve_helpers = { path = "../test_solve_helpers" }
 pretty_assertions.workspace = true

crates/test_compile/src/deindent.rs

@@ -0,0 +1,222 @@
+use bumpalo::Bump;
+use std::borrow::Cow;
+
+/// The purpose of this function is to let us run tests like this:
+///
+///     run_some_test(r#"
+///         x = 1
+///
+///         x
+///     ")
+///
+/// ...without needing to call a macro like `indoc!` to deal with the fact that
+/// multiline Rust string literals preserve all the indented spaces.
+///
+/// This function removes the indentation by removing leading newlines (e.g. after
+/// the `(r#"` opening) and then counting how many spaces precede the first line
+/// (e.g. `"        x = 1"` here) and trimming that many spaces from the beginning
+/// of each subsequent line. The end of the string is then trimmed normally, and
+/// any remaining empty lines are left empty.
+///
+/// This function is a no-op on single-line strings.
+pub fn trim_and_deindent<'a>(arena: &'a Bump, input: &'a str) -> &'a str {
+    let newline_count = input.chars().filter(|&ch| ch == '\n').count();
+
+    // If it's a single-line string, return it without allocating anything.
+    if newline_count == 0 {
+        return input.trim(); // Trim to remove spaces
+    }
+
+    // Trim leading blank lines - we expect at least one, because the opening line will be `(r#"`
+    // (Also, there may be stray blank lines at the start, which this will trim off too.)
+    let mut lines = bumpalo::collections::Vec::with_capacity_in(newline_count + 1, arena);
+
+    for line in input
+        .lines()
+        // Keep skipping until we hit a line that is neither empty nor all spaces.
+        .skip_while(|line| line.chars().all(|ch| ch == ' '))
+    {
+        lines.push(line);
+    }
+
+    // Drop trailing blank lines
+    while lines
+        .last()
+        .map_or(false, |line| line.chars().all(|ch| ch == ' '))
+    {
+        lines.pop();
+    }
+
+    // Now that we've trimmed leading and trailing blank lines,
+    // Find the smallest indent of the remaining lines. That's our indentation amount.
+    let smallest_indent = lines
+        .iter()
+        .filter(|line| !line.is_empty())
+        .map(|line| line.chars().take_while(|&ch| ch == ' ').count())
+        .min()
+        .unwrap_or(0);
+
+    // Remove this amount of indentation from each line.
+    let mut final_str_len = 0;
+
+    lines.iter_mut().for_each(|line| {
+        if line.starts_with(" ") {
+            *line = line.get(smallest_indent..).unwrap_or("");
+        }
+        final_str_len += line.len() + 1; // +1 for the newline that will be added to the end of this line.
+    });
+
+    // Convert lines into a bumpalo::String
+    let mut answer = bumpalo::collections::String::with_capacity_in(final_str_len, arena);
+
+    // Unconditionally push a newline after each line we add. We'll trim off the last one before we return.
+    for line in lines {
+        answer.push_str(line);
+        answer.push('\n');
+    }
+
+    // Trim off the extra newline we added at the end. (Saturate to 0 if we ended up with no lines.)
+    &answer.into_bump_str()[..final_str_len.saturating_sub(1)]
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use pretty_assertions::assert_eq;
+
+    #[test]
+    fn test_single_line_input() {
+        let input = "single line";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), "single line");
+    }
+
+    #[test]
+    fn test_multiline_with_indentation() {
+        let input = r#"
+        x = 1
+
+        x
+    "#;
+        let expected = "x = 1\n\nx";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_multiline_with_varying_indentation() {
+        let input = r#"
+        x = 1
+          y = 2
+        z = 3
+    "#;
+        let expected = "x = 1\n  y = 2\nz = 3";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_multiline_with_empty_lines() {
+        let input = r#"
+        x = 1
+
+        y = 2
+
+        z = 3
+    "#;
+        let expected = "x = 1\n\ny = 2\n\nz = 3";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_input_without_leading_newline() {
+        let input = "    x = 1\n    y = 2";
+        let expected = "x = 1\ny = 2";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_input_with_multiple_leading_newlines() {
+        let input = "\n\n\n    x = 1\n    y = 2";
+        let expected = "x = 1\ny = 2";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_input_with_mixed_indentation() {
+        let input = r#"
+        x = 1
+    y = 2
+            z = 3
+    "#;
+        let expected = "    x = 1\ny = 2\n        z = 3";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_input_with_only_spaces() {
+        let input = "    ";
+        let expected = "";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_input_with_only_newlines() {
+        let input = "\n\n\n";
+        let expected = "";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_input_with_tabs() {
+        let input = "\t\tx = 1\n\t\ty = 2";
+        let expected = "\t\tx = 1\n\t\ty = 2";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_input_with_mixed_spaces_and_tabs() {
+        let input = "    \tx = 1\n    \ty = 2";
+        let expected = "\tx = 1\n\ty = 2";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_input_with_trailing_spaces() {
+        let input = "    x = 1    \n    y = 2    ";
+        let expected = "x = 1    \ny = 2    ";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_input_with_empty_lines_and_spaces() {
+        let input = "    x = 1\n    \n    y = 2";
+        let expected = "x = 1\n\ny = 2";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_input_with_different_indentation_levels() {
+        let input = "    x = 1\n        y = 2\n  z = 3";
+        let expected = "  x = 1\n      y = 2\nz = 3";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_input_with_non_space_characters_at_start() {
+        let input = "x = 1\n    y = 2\n        z = 3";
+        let expected = "x = 1\n    y = 2\n        z = 3";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_empty_input() {
+        let input = "";
+        let expected = "";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+
+    #[test]
+    fn test_input_with_only_one_indented_line() {
+        let input = "    x = 1";
+        let expected = "x = 1";
+        assert_eq!(trim_and_deindent(&Bump::new(), input), expected);
+    }
+}

crates/test_compile/src/help_can.rs

@@ -0,0 +1,106 @@
+use crate::help_parse::ParseExpr;
+use bumpalo::Bump;
+use roc_can::expr::Expr;
+
+pub struct CanExpr {
+    parse_expr: ParseExpr,
+}
+
+impl Default for CanExpr {
+    fn default() -> Self {
+        Self {
+            parse_expr: ParseExpr::default(),
+        }
+    }
+}
+
+impl CanExpr {
+    pub fn can_expr<'a>(&'a self, input: &'a str) -> Result<Expr, CanExprProblem> {
+        match self.parse_expr.parse_expr(input) {
+            Ok(ast) => {
+                // todo canonicalize AST and return that result.
+                let loc_expr = roc_parse::test_helpers::parse_loc_with(arena, expr_str).unwrap_or_else(|e| {
+                    panic!(
+                        "can_expr_with() got a parse error when attempting to canonicalize:\n\n{expr_str:?} {e:?}"
+                    )
+                });
+
+                let mut var_store = VarStore::default();
+                let var = var_store.fresh();
+                let qualified_module_ids = PackageModuleIds::default();
+
+                let mut scope = Scope::new(
+                    home,
+                    "TestPath".into(),
+                    IdentIds::default(),
+                    Default::default(),
+                );
+
+                let dep_idents = IdentIds::exposed_builtins(0);
+                let mut env = Env::new(
+                    arena,
+                    expr_str,
+                    home,
+                    Path::new("Test.roc"),
+                    &dep_idents,
+                    &qualified_module_ids,
+                    None,
+                );
+
+                // Desugar operators (convert them to Apply calls, taking into account
+                // operator precedence and associativity rules), before doing other canonicalization.
+                //
+                // If we did this *during* canonicalization, then each time we
+                // visited a BinOp node we'd recursively try to apply this to each of its nested
+                // operators, and then again on *their* nested operators, ultimately applying the
+                // rules multiple times unnecessarily.
+                let loc_expr = desugar::desugar_expr(&mut env, &mut scope, &loc_expr);
+
+                scope.add_alias(
+                    Symbol::NUM_INT,
+                    Region::zero(),
+                    vec![Loc::at_zero(AliasVar::unbound(
+                        "a".into(),
+                        Variable::EMPTY_RECORD,
+                    ))],
+                    vec![],
+                    Type::EmptyRec,
+                    roc_types::types::AliasKind::Structural,
+                );
+
+                let (loc_expr, output) = canonicalize_expr(
+                    &mut env,
+                    &mut var_store,
+                    &mut scope,
+                    Region::zero(),
+                    &loc_expr.value,
+                );
+
+                let mut all_ident_ids = IdentIds::exposed_builtins(1);
+                all_ident_ids.insert(home, scope.locals.ident_ids);
+
+                let interns = Interns {
+                    module_ids: env.qualified_module_ids.clone().into_module_ids(),
+                    all_ident_ids,
+                };
+
+                CanExprOut {
+                    loc_expr,
+                    output,
+                    problems: env.problems,
+                    home: env.home,
+                    var_store,
+                    interns,
+                    var,
+                }
+            }
+            Err(syntax_error) => {
+                // todo panic due to unexpected syntax error
+            }
+        }
+    }
+
+    pub fn into_arena(self) -> Bump {
+        self.parse_expr.into_arena()
+    }
+}

crates/test_compile/src/help_parse.rs

@@ -1,20 +1,60 @@
+use bumpalo::Bump;
+use roc_parse::{
+    ast,
+    blankspace::space0_before_optional_after,
+    expr::{expr_end, loc_expr_block},
+    parser::{skip_second, EExpr, Parser, SourceError, SyntaxError},
+    state::State,
+};
+use roc_region::all::{Loc, Position};
+
+use crate::deindent::trim_and_deindent;
+
 pub struct ParseExpr {
-    arena: Box<Bump>,
+    arena: Bump,
-    ast: Box<Ast>,
+}
+
+impl Default for ParseExpr {
+    fn default() -> Self {
+        Self {
+            arena: Bump::with_capacity(4096),
+        }
+    }
 }
 
 impl ParseExpr {
-    pub fn parse(&str) -> Self {
+    pub fn parse_expr<'a>(&'a self, input: &'a str) -> Result<ast::Expr<'a>, SyntaxError<'a>> {
-        let mut arena = Bump::new();
+        self.parse_loc_expr(input)
-        let ast = parse(arena, without_indent(str));
+            .map(|loc_expr| loc_expr.value)
+            .map_err(|e| e.problem)
+    }
 
-        Self {
+    pub fn parse_loc_expr<'a>(
-            arena,
+        &'a self,
-            ast,
+        input: &'a str,
+    ) -> Result<Loc<ast::Expr<'a>>, SourceError<'a, SyntaxError<'a>>> {
+        let original_bytes = trim_and_deindent(&self.arena, input).as_bytes();
+        let state = State::new(original_bytes);
+
+        let parser = skip_second(
+            space0_before_optional_after(
+                loc_expr_block(true),
+                EExpr::IndentStart,
+                EExpr::IndentEnd,
+            ),
+            expr_end(),
+        );
+
+        match parser.parse(&self.arena, state, 0) {
+            Ok((_, loc_expr, _)) => Ok(loc_expr),
+            Err((_, fail)) => Err(SourceError {
+                problem: SyntaxError::Expr(fail, Position::default()),
+                bytes: original_bytes,
+            }),
         }
     }
 
-    pub fn ast(&self) -> &Ast {
+    pub fn into_arena(self) -> Bump {
-        self.ast
+        self.arena
     }
 }

crates/test_compile/src/lib.rs

@@ -1,2 +1,3 @@
+mod deindent;
+mod help_can;
 mod help_parse;
-mod without_indent;

crates/test_compile/src/without_indent.rs

@@ -1,100 +0,0 @@
-/// The purpose of this function is to let us run tests like this:
-///
-///     run_some_test(r#"
-///         x = 1
-///
-///         x
-///     ")
-///
-/// ...without needing to call a macro like `indoc!` to deal with the fact that
-/// multiline Rust string literals preserve all the indented spaces. This takes out
-/// the indentation as well as the leading newline in examples like the above, and it's
-/// a no-op on single-line strings.
-pub fn without_indent(input: &str) -> &str {
-    // Ignore any leading newlines, which we expect because the opening line will be `(r#"`
-    let input = input.trim_start_matches('\n');
-    let leading_spaces = input.chars().take_while(|&ch| ch == ' ').count();
-
-    input
-        .lines()
-        .map(|line| {
-            if line.starts_with(" ") {
-                line.get(leading_spaces..).unwrap_or("")
-            } else {
-                line
-            }
-        })
-        .collect::<Vec<&str>>()
-        .join("\n")
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    #[test]
-    fn test_single_line_input() {
-        let input = "single line";
-        assert_eq!(without_indent(input), "single line");
-    }
-
-    #[test]
-    fn test_multiline_with_indentation() {
-        let input = r#"
-        x = 1
-
-        x
-    "#;
-        let expected = "x = 1\n\nx";
-        assert_eq!(without_indent(input), expected);
-    }
-
-    #[test]
-    fn test_multiline_with_varying_indentation() {
-        let input = r#"
-        x = 1
-          y = 2
-        z = 3
-    "#;
-        let expected = "x = 1\n  y = 2\nz = 3";
-        assert_eq!(without_indent(input), expected);
-    }
-
-    #[test]
-    fn test_multiline_with_empty_lines() {
-        let input = r#"
-        x = 1
-
-        y = 2
-
-        z = 3
-    "#;
-        let expected = "x = 1\n\ny = 2\n\nz = 3";
-        assert_eq!(without_indent(input), expected);
-    }
-
-    #[test]
-    fn test_input_without_leading_newline() {
-        let input = "    x = 1\n    y = 2";
-        let expected = "x = 1\ny = 2";
-        assert_eq!(without_indent(input), expected);
-    }
-
-    #[test]
-    fn test_input_with_multiple_leading_newlines() {
-        let input = "\n\n\n    x = 1\n    y = 2";
-        let expected = "x = 1\ny = 2";
-        assert_eq!(without_indent(input), expected);
-    }
-
-    #[test]
-    fn test_input_with_mixed_indentation() {
-        let input = r#"
-        x = 1
-    y = 2
-            z = 3
-    "#;
-        let expected = "x = 1\ny = 2\n    z = 3";
-        assert_eq!(without_indent(input), expected);
-    }
-}