Fix function application parsing

src/parse/blankspace.rs

@@ -3,6 +3,7 @@ use bumpalo::collections::vec::Vec;
 use bumpalo::Bump;
 use parse::ast::{Expr, Space};
 use parse::parser::{and, loc, map_with_arena, unexpected, unexpected_eof, Parser, State};
+use region::Located;
 
 /// What type of comment (if any) are we currently parsing?
 #[derive(Debug, PartialEq, Eq)]
@@ -12,7 +13,10 @@ enum CommentParsing {
     No,
 }
 
-pub fn space0_before<'a, P>(parser: P, min_indent: u16) -> impl Parser<'a, Expr<'a>>
+/// Parse the given expression with 0 or more spaces before it.
+/// Returns a Located<Expr> where the location is around the Expr, ignoring the spaces.
+/// The Expr will be wrapped in a SpaceBefore if there were any interesting spaces found.
+pub fn space0_before<'a, P>(parser: P, min_indent: u16) -> impl Parser<'a, Located<Expr<'a>>>
 where
     P: Parser<'a, Expr<'a>>,
 {
@@ -20,15 +24,21 @@ where
         and(space0(min_indent), loc(parser)),
         |arena, (space_list, loc_expr)| {
             if space_list.is_empty() {
-                loc_expr.value
+                loc_expr
             } else {
-                Expr::SpaceBefore(space_list, arena.alloc(loc_expr))
+                Located {
+                    region: loc_expr.region.clone(),
+                    value: Expr::SpaceBefore(space_list, arena.alloc(loc_expr)),
+                }
             }
         },
     )
 }
 
-pub fn space1_before<'a, P>(parser: P, min_indent: u16) -> impl Parser<'a, Expr<'a>>
+/// Parse the given expression with 1 or more spaces before it.
+/// Returns a Located<Expr> where the location is around the Expr, ignoring the spaces.
+/// The Expr will be wrapped in a SpaceBefore if there were any interesting spaces found.
+pub fn space1_before<'a, P>(parser: P, min_indent: u16) -> impl Parser<'a, Located<Expr<'a>>>
 where
     P: Parser<'a, Expr<'a>>,
 {
@@ -36,15 +46,21 @@ where
         and(space1(min_indent), loc(parser)),
         |arena, (space_list, loc_expr)| {
             if space_list.is_empty() {
-                loc_expr.value
+                loc_expr
             } else {
-                Expr::SpaceBefore(space_list, arena.alloc(loc_expr))
+                Located {
+                    region: loc_expr.region.clone(),
+                    value: Expr::SpaceBefore(space_list, arena.alloc(loc_expr)),
+                }
             }
         },
     )
 }
 
-pub fn space0_after<'a, P>(parser: P, min_indent: u16) -> impl Parser<'a, Expr<'a>>
+/// Parse the given expression with 0 or more spaces after it.
+/// Returns a Located<Expr> where the location is around the Expr, ignoring the spaces.
+/// The Expr will be wrapped in a SpaceAfter if there were any interesting spaces found.
+pub fn space0_after<'a, P>(parser: P, min_indent: u16) -> impl Parser<'a, Located<Expr<'a>>>
 where
     P: Parser<'a, Expr<'a>>,
 {
@@ -52,15 +68,21 @@ where
         and(space0(min_indent), loc(parser)),
         |arena, (space_list, loc_expr)| {
             if space_list.is_empty() {
-                loc_expr.value
+                loc_expr
             } else {
-                Expr::SpaceAfter(arena.alloc(loc_expr), space_list)
+                Located {
+                    region: loc_expr.region.clone(),
+                    value: Expr::SpaceAfter(arena.alloc(loc_expr), space_list),
+                }
             }
         },
     )
 }
 
-pub fn space1_after<'a, P>(parser: P, min_indent: u16) -> impl Parser<'a, Expr<'a>>
+/// Parse the given expression with 1 or more spaces after it.
+/// Returns a Located<Expr> where the location is around the Expr, ignoring the spaces.
+/// The Expr will be wrapped in a SpaceAfter if there were any interesting spaces found.
+pub fn space1_after<'a, P>(parser: P, min_indent: u16) -> impl Parser<'a, Located<Expr<'a>>>
 where
     P: Parser<'a, Expr<'a>>,
 {
@@ -68,9 +90,12 @@ where
         and(space1(min_indent), loc(parser)),
         |arena, (space_list, loc_expr)| {
             if space_list.is_empty() {
-                loc_expr.value
+                loc_expr
             } else {
-                Expr::SpaceAfter(arena.alloc(loc_expr), space_list)
+                Located {
+                    region: loc_expr.region.clone(),
+                    value: Expr::SpaceAfter(arena.alloc(loc_expr), space_list),
+                }
             }
         },
     )
@@ -118,7 +143,9 @@ fn spaces<'a>(require_at_least_one: bool, _min_indent: u16) -> impl Parser<'a, &
                         comment_parsing = CommentParsing::Line;
                     }
                     nonblank => {
-                        return if space_list.is_empty() && require_at_least_one {
+                        return if require_at_least_one && chars_parsed <= 1 {
+                            // We've parsed 1 char and it was not a space,
+                            // but we require parsing at least one space!
                             Err(unexpected(
                                 nonblank,
                                 chars_parsed,
@@ -259,7 +286,7 @@ fn spaces<'a>(require_at_least_one: bool, _min_indent: u16) -> impl Parser<'a, &
             }
         }
 
-        if space_list.is_empty() && require_at_least_one {
+        if require_at_least_one && chars_parsed == 0 {
             Err(unexpected_eof(chars_parsed, state.attempting, state))
         } else {
             Ok((space_list.into_bump_slice(), state))

src/parse/mod.rs

@@ -8,6 +8,7 @@ pub mod parser;
 pub mod problems;
 pub mod string_literal;
 
+use bumpalo::collections::Vec;
 use bumpalo::Bump;
 use operator::Operator;
 use parse::ast::{Attempting, Expr};
@@ -15,8 +16,8 @@ use parse::blankspace::{space0, space1_before};
 use parse::ident::{ident, Ident};
 use parse::number_literal::number_literal;
 use parse::parser::{
-    and, attempt, ch, either, loc, map, map_with_arena, one_of3, one_of4, one_of6, optional,
+    and, attempt, char, either, loc, map, map_with_arena, one_of3, one_of4, one_of6, one_or_more,
-    skip_first, string, unexpected, unexpected_eof, Either, ParseResult, Parser, State,
+    optional, skip_first, string, unexpected, unexpected_eof, Either, ParseResult, Parser, State,
 };
 use parse::string_literal::string_literal;
 use region::Located;
@@ -27,9 +28,26 @@ pub fn expr<'a>(min_indent: u16) -> impl Parser<'a, Expr<'a>> {
     move |arena, state| parse_expr(min_indent, arena, state)
 }
 
+fn parse_expr_body_without_operators<'a>(
+    min_indent: u16,
+    arena: &'a Bump,
+    state: State<'a>,
+) -> ParseResult<'a, Expr<'a>> {
+    one_of6(
+        string_literal(),
+        record_literal(),
+        number_literal(),
+        when(min_indent),
+        conditional(min_indent),
+        ident_etc(min_indent),
+    )
+    .parse(arena, state)
+}
+
 fn parse_expr<'a>(min_indent: u16, arena: &'a Bump, state: State<'a>) -> ParseResult<'a, Expr<'a>> {
     let expr_parser = map_with_arena(
         and(
+            // First parse the body without operators, then try to parse possible operators after.
             loc(move |arena, state| parse_expr_body_without_operators(min_indent, arena, state)),
             optional(and(
                 and(space0(min_indent), and(loc(operator()), space0(min_indent))),
@@ -61,27 +79,10 @@ fn parse_expr<'a>(min_indent: u16, arena: &'a Bump, state: State<'a>) -> ParseRe
     attempt(Attempting::Expression, expr_parser).parse(arena, state)
 }
 
-fn parse_expr_body_without_operators<'a>(
+fn function_arg<'a>(min_indent: u16) -> impl Parser<'a, Expr<'a>> {
-    min_indent: u16,
+    // Don't parse operators, because they have a higher precedence than function application.
-    arena: &'a Bump,
+    // If we encounter one, we're done parsing function args!
-    state: State<'a>,
+    move |arena, state| parse_expr_body_without_operators(min_indent, arena, state)
-) -> ParseResult<'a, Expr<'a>> {
-    one_of6(
-        string_literal(),
-        record_literal(),
-        number_literal(),
-        when(min_indent),
-        conditional(min_indent),
-        ident_etc(min_indent),
-    )
-    .parse(arena, state)
-}
-
-pub fn loc_function_args<'a>(_min_indent: u16) -> impl Parser<'a, &'a [Located<Expr<'a>>]> {
-    move |_arena, _state| {
-        panic!("TODO stop early if we see an operator after the whitespace - precedence!");
-        // zero_or_more(after(one_or_more(whitespace(min_indent)), function_arg()))
-    }
 }
 
 pub fn when<'a>(_min_indent: u16) -> impl Parser<'a, Expr<'a>> {
@@ -98,33 +99,36 @@ pub fn conditional<'a>(min_indent: u16) -> impl Parser<'a, Expr<'a>> {
         map_with_arena(
             skip_first(
                 string(keyword::IF),
-                loc(space1_before(expr(min_indent), min_indent)),
+                space1_before(expr(min_indent), min_indent),
             ),
             |arena, loc_expr| Expr::If(arena.alloc(loc_expr)),
         ),
         map_with_arena(
             skip_first(
                 string(keyword::THEN),
-                loc(space1_before(expr(min_indent), min_indent)),
+                space1_before(expr(min_indent), min_indent),
             ),
             |arena, loc_expr| Expr::Then(arena.alloc(loc_expr)),
         ),
         map_with_arena(
             skip_first(
                 string(keyword::ELSE),
-                loc(space1_before(expr(min_indent), min_indent)),
+                space1_before(expr(min_indent), min_indent),
             ),
             |arena, loc_expr| Expr::Else(arena.alloc(loc_expr)),
         ),
         map_with_arena(
             skip_first(
                 string(keyword::CASE),
-                loc(space1_before(expr(min_indent), min_indent)),
+                space1_before(expr(min_indent), min_indent),
             ),
             |arena, loc_expr| Expr::Case(arena.alloc(loc_expr)),
         ),
     )
 }
+pub fn loc_function_args<'a>(min_indent: u16) -> impl Parser<'a, Vec<'a, Located<Expr<'a>>>> {
+    one_or_more(space1_before(function_arg(min_indent), min_indent))
+}
 
 /// When we parse an ident like `foo ` it could be any of these:
 ///
@@ -137,48 +141,48 @@ pub fn ident_etc<'a>(min_indent: u16) -> impl Parser<'a, Expr<'a>> {
     map_with_arena(
         and(
             loc(ident()),
-            either(
+            optional(either(
-                // Check if this is either a def or type annotation
+                // There may optionally be function args after this ident
-                and(space0(min_indent), either(ch('='), ch(':'))),
-                // Check if this is function application
                 loc_function_args(min_indent),
+                // If there aren't any args, there may be a '=' or ':' after it.
+                // (It's a syntax error to write e.g. `foo bar =` - so if there
+                // were any args, there is definitely no need to parse '=' or ':'!)
+                and(space0(min_indent), either(char('='), char(':'))),
+            )),
         ),
-        ),
+        |arena, (loc_ident, opt_extras)| {
-        |arena, (loc_ident, equals_or_loc_args)| {
-            match equals_or_loc_args {
-                Either::First((_space_list, Either::First(()))) => {
-                    // We have now parsed the beginning of a def (e.g. `foo =`)
-                    panic!("TODO parse def, making sure not to drop comments!");
-                }
-                Either::First((_space_list, Either::Second(()))) => {
-                    // We have now parsed the beginning of a type annotation (e.g. `foo :`)
-                    panic!("TODO parse type annotation, making sure not to drop comments!");
-                }
-                Either::Second(loc_args) => {
             // This appears to be a var, keyword, or function application.
-                    let name_expr = match loc_ident.value {
+
+            match opt_extras {
+                Some(Either::First(loc_args)) => {
+                    let loc_expr = Located {
+                        region: loc_ident.region,
+                        value: ident_to_expr(loc_ident.value),
+                    };
+
+                    Expr::Apply(arena.alloc((loc_expr, loc_args.into_bump_slice())))
+                }
+                Some(Either::Second((_space_list, Either::First(())))) => {
+                    panic!("TODO handle def, making sure not to drop comments!");
+                }
+                Some(Either::Second((_space_list, Either::Second(())))) => {
+                    panic!("TODO handle annotation, making sure not to drop comments!");
+                }
+                None => ident_to_expr(loc_ident.value),
+            }
+        },
+    )
+}
+
+fn ident_to_expr<'a>(src: Ident<'a>) -> Expr<'a> {
+    match src {
         Ident::Var(info) => Expr::Var(info.module_parts, info.value),
         Ident::Variant(info) => Expr::Variant(info.module_parts, info.value),
         Ident::Field(info) => Expr::QualifiedField(info.module_parts, info.value),
         Ident::AccessorFunction(string) => Expr::AccessorFunction(string),
         Ident::Malformed(string) => Expr::MalformedIdent(string),
-                    };
-
-                    if loc_args.is_empty() {
-                        name_expr
-                    } else {
-                        let loc_expr = Located {
-                            region: loc_ident.region,
-                            value: name_expr,
-                        };
-
-                        Expr::Apply(arena.alloc((loc_expr, loc_args)))
     }
 }
-            }
-        },
-    )
-}
 
 pub fn operator<'a>() -> impl Parser<'a, Operator> {
     one_of3(

src/parse/parser.rs

@@ -177,6 +177,19 @@ where
     move |_, state| Ok((value.clone(), state))
 }
 
+pub fn and_then<'a, P1, P2, F, Before, After>(parser: P1, transform: F) -> impl Parser<'a, After>
+where
+    P1: Parser<'a, Before>,
+    P2: Parser<'a, After>,
+    F: Fn(Before) -> P2,
+{
+    move |arena, state| {
+        parser
+            .parse(arena, state)
+            .and_then(|(output, next_state)| transform(output).parse(arena, next_state))
+    }
+}
+
 pub fn map<'a, P, F, Before, After>(parser: P, transform: F) -> impl Parser<'a, After>
 where
     P: Parser<'a, Before>,
@@ -371,10 +384,11 @@ fn line_too_long<'a>(attempting: Attempting, state: State<'a>) -> (Fail, State<'
 }
 
 /// A single char.
-pub fn ch<'a>(expected: char) -> impl Parser<'a, ()> {
+pub fn char<'a>(expected: char) -> impl Parser<'a, ()> {
     move |_arena, state: State<'a>| match state.input.chars().next() {
         Some(actual) if expected == actual => Ok(((), state.advance_without_indenting(1)?)),
-        _ => Err(unexpected_eof(1, Attempting::Keyword, state)),
+        Some(other_ch) => Err(unexpected(other_ch, 0, state, Attempting::Keyword)),
+        _ => Err(unexpected_eof(0, Attempting::Keyword, state)),
     }
 }
 
@@ -391,7 +405,7 @@ pub fn string<'a>(string: &'static str) -> impl Parser<'a, ()> {
         // TODO do this comparison in one SIMD instruction (on supported systems)
         match input.get(0..len) {
             Some(next_str) if next_str == string => Ok(((), state.advance_without_indenting(len)?)),
-            _ => Err(unexpected_eof(len, Attempting::Keyword, state)),
+            _ => Err(unexpected_eof(0, Attempting::Keyword, state)),
         }
     }
 }

tests/test_parse.rs

@@ -14,7 +14,8 @@ mod helpers;
 
 #[cfg(test)]
 mod test_parse {
-    use bumpalo::Bump;
+    use bumpalo::collections::vec::Vec;
+    use bumpalo::{self, Bump};
     use helpers::parse_with;
     use roc::operator::Operator::*;
     use roc::parse::ast::Attempting;
@@ -357,6 +358,36 @@ mod test_parse {
         assert_eq!(Ok(expected), actual);
     }
 
+    // VAR
+
+    #[test]
+    fn basic_var() {
+        let arena = Bump::new();
+        let module_parts = Vec::new_in(&arena).into_bump_slice();
+        let expected = Var(module_parts, "whee");
+        let actual = parse_with(&arena, "whee");
+
+        assert_eq!(Ok(expected), actual);
+    }
+
+    // APPLY
+
+    #[test]
+    fn basic_apply() {
+        let arena = Bump::new();
+        let module_parts = Vec::new_in(&arena).into_bump_slice();
+        let arg = Located::new(0, 5, 0, 6, Int("1"));
+        let args = bumpalo::vec![in &arena; arg];
+        let tuple = arena.alloc((
+            Located::new(0, 0, 0, 4, Var(module_parts, "whee")),
+            args.into_bump_slice(),
+        ));
+        let expected = Apply(tuple);
+        let actual = parse_with(&arena, "whee 1");
+
+        assert_eq!(Ok(expected), actual);
+    }
+
     // TODO test hex/oct/binary parsing
     //
     // TODO test for \t \r and \n in string literals *outside* unicode escape sequence!