refactor: parsing of type parameters: list, record, one_of etc

crates/nu-parser/src/parse_shape_specs.rs

@@ -1,7 +1,9 @@
 #![allow(clippy::byte_char_slices)]
 
 use crate::{lex::lex_signature, parser::parse_value, trim_quotes, TokenContents};
-use nu_protocol::{engine::StateWorkingSet, ParseError, Span, SyntaxShape, Type};
+use nu_protocol::{
+    engine::StateWorkingSet, IntoSpanned, ParseError, Span, Spanned, SyntaxShape, Type,
+};
 
 #[derive(Debug, Clone, Copy, PartialEq)]
 pub enum ShapeDescriptorUse {
@@ -58,19 +60,52 @@ pub fn parse_shape_name(
         b"filesize" => SyntaxShape::Filesize,
         b"glob" => SyntaxShape::GlobPattern,
         b"int" => SyntaxShape::Int,
-        _ if bytes.starts_with(b"list") => parse_list_shape(working_set, bytes, span, use_loc),
         b"nothing" => SyntaxShape::Nothing,
-        _ if bytes.starts_with(b"one_of") => parse_one_of_shape(working_set, bytes, span, use_loc),
         b"number" => SyntaxShape::Number,
         b"path" => SyntaxShape::Filepath,
         b"range" => SyntaxShape::Range,
-        _ if bytes.starts_with(b"record") => {
-            parse_collection_shape(working_set, bytes, span, use_loc)
+        _ if bytes.starts_with(b"one_of")
+            || bytes.starts_with(b"list")
+            || bytes.starts_with(b"record")
+            || bytes.starts_with(b"table") =>
+        {
+            let (type_name, type_params) = split_generic_params(working_set, bytes, span);
+            match type_name {
+                b"one_of" => SyntaxShape::OneOf(match type_params {
+                    Some(params) => parse_type_params(working_set, params, use_loc),
+                    None => vec![],
+                }),
+                b"list" => SyntaxShape::List(Box::new(match type_params {
+                    Some(params) => {
+                        let mut parsed_params = parse_type_params(working_set, params, use_loc);
+                        if parsed_params.len() > 1 {
+                            working_set.error(ParseError::LabeledError(
+                                "expected a single type parameter".into(),
+                                "only one parameter allowed".into(),
+                                params.span,
+                            ));
+                            SyntaxShape::Any
+                        } else {
+                            parsed_params.pop().unwrap_or(SyntaxShape::Any)
+                        }
+                    }
+                    None => SyntaxShape::Any,
+                })),
+                b"record" => SyntaxShape::Record(match type_params {
+                    Some(params) => parse_named_type_params(working_set, params, use_loc),
+                    None => vec![],
+                }),
+                b"table" => SyntaxShape::Table(match type_params {
+                    Some(params) => parse_named_type_params(working_set, params, use_loc),
+                    None => vec![],
+                }),
+                _ => {
+                    working_set.error(ParseError::UnknownType(span));
+                    SyntaxShape::Any
+                }
+            }
         }
         b"string" => SyntaxShape::String,
-        _ if bytes.starts_with(b"table") => {
-            parse_collection_shape(working_set, bytes, span, use_loc)
-        }
         _ => {
             if bytes.contains(&b'@') {
                 let mut split = bytes.splitn(2, |b| b == &b'@');
@@ -121,259 +156,31 @@ pub fn parse_shape_name(
     result
 }
 
-fn parse_collection_shape(
+fn split_generic_params<'a>(
     working_set: &mut StateWorkingSet,
-    bytes: &[u8],
+    bytes: &'a [u8],
     span: Span,
-    use_loc: ShapeDescriptorUse,
-) -> SyntaxShape {
-    assert!(bytes.starts_with(b"record") || bytes.starts_with(b"table"));
-    let is_table = bytes.starts_with(b"table");
-
-    let name = if is_table { "table" } else { "record" };
-    let prefix = (if is_table { "table<" } else { "record<" }).as_bytes();
-    let prefix_len = prefix.len();
-    let mk_shape = |ty| -> SyntaxShape {
-        if is_table {
-            SyntaxShape::Table(ty)
-        } else {
-            SyntaxShape::Record(ty)
-        }
+) -> (&'a [u8], Option<Spanned<&'a [u8]>>) {
+    let n = bytes.iter().position(|&c| c == b'<' || c == b'(');
+    let (open_delim_pos, close_delim) = match n.and_then(|n| Some((n, bytes.get(n)?))) {
+        Some((n, b'<')) => (n, b'>'),
+        Some((n, b'(')) => (n, b')'),
+        _ => return (bytes, None),
     };
 
-    if bytes == name.as_bytes() {
-        mk_shape(vec![])
-    } else if bytes.starts_with(prefix) {
-        let Some(inner_span) = prepare_inner_span(working_set, bytes, span, prefix_len, b'>')
-        else {
-            return SyntaxShape::Any;
-        };
+    let type_name = &bytes[..(open_delim_pos)];
+    let params = &bytes[(open_delim_pos + 1)..];
 
-        // record<> or table<>
-        if inner_span.end - inner_span.start == 0 {
-            return mk_shape(vec![]);
-        }
-        let source = working_set.get_span_contents(inner_span);
-        let (tokens, err) = lex_signature(
-            source,
-            inner_span.start,
-            &[b'\n', b'\r'],
-            &[b':', b','],
-            true,
-        );
+    let start = span.start + type_name.len() + 1;
 
-        if let Some(err) = err {
-            working_set.error(err);
-            // lexer errors cause issues with span overflows
-            return mk_shape(vec![]);
-        }
-
-        let mut sig = vec![];
-        let mut idx = 0;
-
-        let key_error = |span| {
-            ParseError::LabeledError(
-                format!("`{name}` type annotations key not string"),
-                "must be a string".into(),
-                span,
-            )
-        };
-
-        while idx < tokens.len() {
-            let TokenContents::Item = tokens[idx].contents else {
-                working_set.error(key_error(tokens[idx].span));
-                return mk_shape(vec![]);
-            };
-
-            if working_set
-                .get_span_contents(tokens[idx].span)
-                .starts_with(b",")
-            {
-                idx += 1;
-                continue;
-            }
-
-            let Some(key) =
-                parse_value(working_set, tokens[idx].span, &SyntaxShape::String).as_string()
-            else {
-                working_set.error(key_error(tokens[idx].span));
-                return mk_shape(vec![]);
-            };
-
-            // we want to allow such an annotation
-            // `record<name>` where the user leaves out the type
-            if idx + 1 == tokens.len() {
-                sig.push((key, SyntaxShape::Any));
-                break;
-            } else {
-                idx += 1;
-            }
-
-            let maybe_colon = working_set.get_span_contents(tokens[idx].span);
-            match maybe_colon {
-                b":" => {
-                    if idx + 1 == tokens.len() {
-                        working_set
-                            .error(ParseError::Expected("type after colon", tokens[idx].span));
-                        break;
-                    } else {
-                        idx += 1;
-                    }
-                }
-                // a key provided without a type
-                b"," => {
-                    idx += 1;
-                    sig.push((key, SyntaxShape::Any));
-                    continue;
-                }
-                // a key provided without a type
-                _ => {
-                    sig.push((key, SyntaxShape::Any));
-                    continue;
-                }
-            }
-
-            let shape_bytes = working_set.get_span_contents(tokens[idx].span).to_vec();
-            let shape = parse_shape_name(working_set, &shape_bytes, tokens[idx].span, use_loc);
-            sig.push((key, shape));
-            idx += 1;
-        }
-
-        mk_shape(sig)
-    } else {
-        working_set.error(ParseError::UnknownType(span));
-
-        SyntaxShape::Any
-    }
-}
-
-fn parse_one_of_shape(
-    working_set: &mut StateWorkingSet,
-    bytes: &[u8],
-    span: Span,
-    use_loc: ShapeDescriptorUse,
-) -> SyntaxShape {
-    assert!(bytes.starts_with(b"one_of"));
-
-    let name = "one_of";
-    let prefix = b"one_of(";
-    let prefix_len = prefix.len();
-    let mk_shape = |ty| -> SyntaxShape { SyntaxShape::OneOf(ty) };
-
-    if bytes == name.as_bytes() {
-        mk_shape(vec![])
-    } else if bytes.starts_with(prefix) {
-        let Some(inner_span) = prepare_inner_span(working_set, bytes, span, prefix_len, b')')
-        else {
-            return SyntaxShape::Any;
-        };
-
-        // one_of()
-        if inner_span.end - inner_span.start == 0 {
-            return mk_shape(vec![]);
-        }
-        let source = working_set.get_span_contents(inner_span);
-        let (tokens, err) = lex_signature(
-            source,
-            inner_span.start,
-            &[b'\n', b'\r'],
-            &[b':', b','],
-            true,
-        );
-
-        if let Some(err) = err {
-            working_set.error(err);
-            // lexer errors cause issues with span overflows
-            return mk_shape(vec![]);
-        }
-
-        let mut sig = vec![];
-        let mut idx = 0;
-
-        let key_error = |span| {
-            ParseError::LabeledError(
-                format!("`{name}` type annotations key not string"),
-                "must be a string".into(),
-                span,
-            )
-        };
-
-        while idx < tokens.len() {
-            let TokenContents::Item = tokens[idx].contents else {
-                working_set.error(key_error(tokens[idx].span));
-                return mk_shape(vec![]);
-            };
-
-            if working_set
-                .get_span_contents(tokens[idx].span)
-                .starts_with(b",")
-            {
-                idx += 1;
-                continue;
-            }
-
-            let shape_bytes = working_set.get_span_contents(tokens[idx].span).to_vec();
-            let shape = parse_shape_name(working_set, &shape_bytes, tokens[idx].span, use_loc);
-            sig.push(shape);
-            idx += 1;
-        }
-
-        mk_shape(sig)
-    } else {
-        working_set.error(ParseError::UnknownType(span));
-        SyntaxShape::Any
-    }
-}
-
-fn parse_list_shape(
-    working_set: &mut StateWorkingSet,
-    bytes: &[u8],
-    span: Span,
-    use_loc: ShapeDescriptorUse,
-) -> SyntaxShape {
-    assert!(bytes.starts_with(b"list"));
-
-    if bytes == b"list" {
-        SyntaxShape::List(Box::new(SyntaxShape::Any))
-    } else if bytes.starts_with(b"list<") {
-        let Some(inner_span) = prepare_inner_span(working_set, bytes, span, 5, b'>') else {
-            return SyntaxShape::Any;
-        };
-
-        let inner_text = String::from_utf8_lossy(working_set.get_span_contents(inner_span));
-        // remove any extra whitespace, for example `list< string >` becomes `list<string>`
-        let inner_bytes = inner_text.trim().as_bytes().to_vec();
-
-        // list<>
-        if inner_bytes.is_empty() {
-            SyntaxShape::List(Box::new(SyntaxShape::Any))
-        } else {
-            let inner_sig = parse_shape_name(working_set, &inner_bytes, inner_span, use_loc);
-
-            SyntaxShape::List(Box::new(inner_sig))
-        }
-    } else {
-        working_set.error(ParseError::UnknownType(span));
-
-        SyntaxShape::List(Box::new(SyntaxShape::Any))
-    }
-}
-
-fn prepare_inner_span(
-    working_set: &mut StateWorkingSet,
-    bytes: &[u8],
-    span: Span,
-    prefix_len: usize,
-    terminator: u8,
-) -> Option<Span> {
-    let start = span.start + prefix_len;
-
-    if bytes.ends_with(&[terminator]) {
+    if params.ends_with(&[close_delim]) {
         let end = span.end - 1;
-        Some(Span::new(start, end))
-    } else if bytes.contains(&terminator) {
-        let angle_start = bytes.split(|it| it == &terminator).collect::<Vec<_>>()[0].len() + 1;
-        let span = Span::new(span.start + angle_start, span.end);
+        (
+            type_name,
+            Some((&params[..(params.len() - 1)]).into_spanned(Span::new(start, end))),
+        )
+    } else if let Some(close_delim_pos) = params.iter().position(|it| it == &close_delim) {
+        let span = Span::new(span.start + close_delim_pos, span.end);
 
         working_set.error(ParseError::LabeledError(
             "Extra characters in the parameter name".into(),
@@ -381,9 +188,128 @@ fn prepare_inner_span(
             span,
         ));
 
-        None
+        (bytes, None)
     } else {
-        working_set.error(ParseError::Unclosed((terminator as char).into(), span));
-        None
+        working_set.error(ParseError::Unclosed((close_delim as char).into(), span));
+        (bytes, None)
     }
 }
+
+fn parse_named_type_params(
+    working_set: &mut StateWorkingSet,
+    Spanned { item: source, span }: Spanned<&[u8]>,
+    use_loc: ShapeDescriptorUse,
+) -> Vec<(String, SyntaxShape)> {
+    let (tokens, err) = lex_signature(source, span.start, &[b'\n', b'\r'], &[b':', b','], true);
+
+    if let Some(err) = err {
+        working_set.error(err);
+        return Vec::new();
+    }
+
+    let mut sig = Vec::new();
+    let mut idx = 0;
+
+    let key_error = |span| {
+        ParseError::LabeledError(
+            // format!("`{name}` type annotations key not string"),
+            "annotation key not string".into(),
+            "must be a string".into(),
+            span,
+        )
+    };
+
+    while idx < tokens.len() {
+        let TokenContents::Item = tokens[idx].contents else {
+            working_set.error(key_error(tokens[idx].span));
+            return Vec::new();
+        };
+
+        if working_set
+            .get_span_contents(tokens[idx].span)
+            .starts_with(b",")
+        {
+            idx += 1;
+            continue;
+        }
+
+        let Some(key) =
+            parse_value(working_set, tokens[idx].span, &SyntaxShape::String).as_string()
+        else {
+            working_set.error(key_error(tokens[idx].span));
+            return Vec::new();
+        };
+
+        // we want to allow such an annotation
+        // `record<name>` where the user leaves out the type
+        if idx + 1 == tokens.len() {
+            sig.push((key, SyntaxShape::Any));
+            break;
+        } else {
+            idx += 1;
+        }
+
+        let maybe_colon = working_set.get_span_contents(tokens[idx].span);
+        match maybe_colon {
+            b":" => {
+                if idx + 1 == tokens.len() {
+                    working_set.error(ParseError::Expected("type after colon", tokens[idx].span));
+                    break;
+                } else {
+                    idx += 1;
+                }
+            }
+            // a key provided without a type
+            b"," => {
+                idx += 1;
+                sig.push((key, SyntaxShape::Any));
+                continue;
+            }
+            // a key provided without a type
+            _ => {
+                sig.push((key, SyntaxShape::Any));
+                continue;
+            }
+        }
+
+        let shape_bytes = working_set.get_span_contents(tokens[idx].span).to_vec();
+        let shape = parse_shape_name(working_set, &shape_bytes, tokens[idx].span, use_loc);
+        sig.push((key, shape));
+        idx += 1;
+    }
+
+    sig
+}
+
+fn parse_type_params(
+    working_set: &mut StateWorkingSet,
+    Spanned { item: source, span }: Spanned<&[u8]>,
+    use_loc: ShapeDescriptorUse,
+) -> Vec<SyntaxShape> {
+    let (tokens, err) = lex_signature(source, span.start, &[b'\n', b'\r'], &[b':', b','], true);
+
+    if let Some(err) = err {
+        working_set.error(err);
+        return Vec::new();
+    }
+
+    let mut sig = vec![];
+    let mut idx = 0;
+
+    while idx < tokens.len() {
+        if working_set
+            .get_span_contents(tokens[idx].span)
+            .starts_with(b",")
+        {
+            idx += 1;
+            continue;
+        }
+
+        let shape_bytes = working_set.get_span_contents(tokens[idx].span).to_vec();
+        let shape = parse_shape_name(working_set, &shape_bytes, tokens[idx].span, use_loc);
+        sig.push(shape);
+        idx += 1;
+    }
+
+    sig
+}

tests/repl/test_signatures.rs

@@ -109,7 +109,7 @@ fn list_annotations_space_before() -> TestResult {
 #[test]
 fn list_annotations_unknown_separators() -> TestResult {
     let input = "def run [list: list<int, string>] {$list | length}; run [2 5 4]";
-    let expected = "unknown type";
+    let expected = "only one parameter allowed";
     fail_test(input, expected)
 }