CSY/reowolf Changeset - e406c61b1158 · Centrum Wiskunde & Informatica (CWI)

Changeset - e406c61b1158

Parent rev.

Child rev.

[Not reviewed]

0 9 1

MH - 4 years ago 2021-03-29 15:31:30
contact@maxhenger.nl

start on enum literals, extended some tests

10 files changed with 543 insertions and 45 deletions:

src/macros.rs

src/protocol/ast.rs

src/protocol/lexer.rs

src/protocol/parser/mod.rs

src/protocol/parser/type_resolver.rs

src/protocol/tests/mod.rs

src/protocol/tests/parser_imports.rs

148

src/protocol/tests/parser_inference.rs

140

src/protocol/tests/parser_monomorphs.rs

src/protocol/tests/utils.rs

107

0 comments (0 inline, 0 general)

src/macros.rs

➞

Show inline comments

 macro_rules! enabled_debug_print {
     (false, $name:literal, $format:literal) => {};
     (false, $name:literal, $format:literal, $($args:expr),*) => {};
     (true, $name:literal, $format:literal) => {
         println!("[{}] {}", $name, $format)
     };
     (true, $name:literal, $format:literal, $($args:expr),*) => {
         println!("[{}] {}", $name, format!($format, $($args),*))
     };
+}
 /*
 Change the definition of these macros to control the logging level statically
 */
 macro_rules! log {
     (@ENDPT, $logger:expr, $($arg:tt)*) => {{
         // if let Some(w) = $logger.line_writer() {
         //     let _ = writeln!(w, $($arg)*);
         // }
     }};
     ($logger:expr, $($arg:tt)*) => {{
         #[cfg(not(feature = "no_logging"))]

src/protocol/ast.rs

➞

Show inline comments

@@ @@ -995,24 +995,34 @@ pub struct LiteralStructField { @@
+}
 #[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
 pub struct LiteralStruct {
     // Phase 1: parser
     pub(crate) identifier: NamespacedIdentifier,
     pub(crate) poly_args: Vec<ParserTypeId>,
     pub(crate) fields: Vec<LiteralStructField>,
     // Phase 2: linker
     pub(crate) definition: Option<DefinitionId>
+}
 #[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
 pub struct LiteralEnum {
     // Phase 1: parser
     pub(crate) identifier: NamespacedIdentifier,
     pub(crate) poly_args: Vec<ParserTypeId>,
     // Phase 2: linker
     pub(crate) definition: Option<DefinitionId>,
     pub(crate) variant_idx: usize,
+}
 #[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
 pub enum Method {
     Get,
     Put,
     Fires,
     Create,
     Symbolic(MethodSymbolic)
+}
 #[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
 pub struct MethodSymbolic {
     pub(crate) identifier: NamespacedIdentifier,
@@ @@ -1022,24 +1032,31 @@ pub struct MethodSymbolic { @@
 #[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
 pub enum Field {
     Length,
     Symbolic(FieldSymbolic),
+}
 impl Field {
     pub fn is_length(&self) -> bool {
         match self {
             Field::Length => true,
             _ => false,
+        }
+    }
     pub fn as_symbolic(&self) -> &FieldSymbolic {
         match self {
             Field::Symbolic(v) => v,
             _ => unreachable!("attempted to get Field::Symbolic from {:?}", self)
+        }
+    }
+}
 #[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
 pub struct FieldSymbolic {
     // Phase 1: Parser
     pub(crate) identifier: Identifier,
     // Phase 3: Typing
     pub(crate) definition: Option<DefinitionId>,
     pub(crate) field_idx: usize,
+}
 #[derive(Debug, Clone, Copy, serde::Serialize, serde::Deserialize)]

src/protocol/lexer.rs

➞

Show inline comments

 use crate::protocol::ast::*;
 use crate::protocol::inputsource::*;
 const MAX_LEVEL: usize = 128;
 const MAX_NAMESPACES: u8 = 8; // only three levels are supported at the moment
 macro_rules! debug_log {
     ($format:literal) => {
         enabled_debug_print!(false, "lexer", $format);
     };
     ($format:literal, $($args:expr),*) => {
         enabled_debug_print!(false, "lexer", $format, $($args),*);
     };
+}
 macro_rules! debug_line {
     ($source:expr) => {
+        {
             let mut buffer = String::with_capacity(128);
             for idx in 0..buffer.capacity() {
                 let next = $source.lookahead(idx);
                 if next.is_none() || Some(b'\n') == next { break; }
                 buffer.push(next.unwrap() as char);
+            }
             buffer
+        }
     };
+}
 fn is_vchar(x: Option<u8>) -> bool {
     if let Some(c) = x {
         c >= 0x21 && c <= 0x7E
     } else {
         false
+    }
+}
 fn is_wsp(x: Option<u8>) -> bool {
     if let Some(c) = x {
         c == b' ' || c == b'\t'
     } else {
@@ @@ -160,24 +182,28 @@ impl Lexer<'_> { @@
         if !is_ident_start(self.source.next()) { return }
         self.source.consume();
         while is_ident_rest(self.source.next()) {
             self.source.consume()
+        }
+    }
     fn has_keyword(&self, keyword: &[u8]) -> bool {
         if !self.source.has(keyword) {
             return false;
+        }
         // Word boundary
         let next = self.source.lookahead(keyword.len());
         if next.is_none() { return true; }
         return !is_ident_rest(next);
         if let Some(next) = self.source.lookahead(keyword.len()) {
             !(next >= b'A' && next <= b'Z' || next >= b'a' && next <= b'z')
         } else {
             true
+        }
+    }
     fn consume_keyword(&mut self, keyword: &[u8]) -> Result<(), ParseError2> {
         let len = keyword.len();
         for i in 0..len {
             let expected = Some(lowercase(keyword[i]));
             let next = self.source.next();
             if next != expected {
@@ @@ -240,52 +266,53 @@ impl Lexer<'_> { @@
             self.consume_whitespace(false)?;
             had_comma = self.source.next() == Some(b',');
             if had_comma {
                 self.source.consume();
                 self.consume_whitespace(false)?;
+            }
+        }
         Ok(Some(elements))
+    }
     /// Essentially the same as `consume_comma_separated`, but will not allocate
     /// memory. Will return `true` and leave the input position at the end of
     /// the comma-separated list if well formed. Otherwise returns `false` and
     /// leaves the input position at a "random" position.
     /// memory. Will return `Ok(true)` and leave the input position at the end
     /// the comma-separated list if well formed and `Ok(false)` if the list is
     /// not present. Otherwise returns `Err(())` and leaves the input position
     /// at a "random" position.
     fn consume_comma_separated_spilled_without_pos_recovery<F: Fn(&mut Lexer) -> bool>(
         &mut self, open: u8, close: u8, func: F
     ) -> bool {
+    ) -> Result<bool, ()> {
         if Some(open) != self.source.next() {
-            return true;
+            return Ok(false);
+        }
         self.source.consume();
-        if self.consume_whitespace(false).is_err() { return false };
+        if self.consume_whitespace(false).is_err() { return Err(()) };
         let mut had_comma = true;
         loop {
             if Some(close) == self.source.next() {
                 self.source.consume();
                 return true;
+                return Ok(true);
             } else if !had_comma {
-                return false;
+                return Err(());
+            }
             if !func(self) { return false; }
             if self.consume_whitespace(false).is_err() { return false };
             if !func(self) { return Err(()); }
             if self.consume_whitespace(false).is_err() { return Err(()) };
             had_comma = self.source.next() == Some(b',');
             if had_comma {
                 self.source.consume();
-                if self.consume_whitespace(false).is_err() { return false; }
+                if self.consume_whitespace(false).is_err() { return Err(()); }
+            }
+        }
+    }
     fn consume_ident(&mut self) -> Result<Vec<u8>, ParseError2> {
         if !self.has_identifier() {
             return Err(self.error_at_pos("Expected identifier"));
+        }
         let mut result = Vec::new();
         let mut next = self.source.next();
         result.push(next.unwrap());
         self.source.consume();
         next = self.source.next();
@@ @@ -471,24 +498,25 @@ impl Lexer<'_> { @@
                         pos: port_pos.clone(),
                         variant: ParserTypeVariant::Inferred
                     }))
                 } else {
                     Err(true)
                 },
 => Ok(args[0]),
                 _ => Err(false)
+            }
         };
         // Consume the type
         debug_log!("consume_type2: {}", debug_line!(self.source));
         let pos = self.source.pos();
         let parser_type_variant = if self.has_keyword(b"msg") {
             self.consume_keyword(b"msg")?;
             ParserTypeVariant::Message
         } else if self.has_keyword(b"boolean") {
             self.consume_keyword(b"boolean")?;
             ParserTypeVariant::Bool
         } else if self.has_keyword(b"byte") {
             self.consume_keyword(b"byte")?;
             ParserTypeVariant::Byte
         } else if self.has_keyword(b"short") {
             self.consume_keyword(b"short")?;
@@ @@ -591,39 +619,44 @@ impl Lexer<'_> { @@
+            }
+        }
         self.source.seek(backup_pos);
         Ok(parser_type_id)
+    }
     /// Attempts to consume a type without returning it. If it doesn't encounter
     /// a well-formed type, then the input position is left at a "random"
     /// position.
     fn maybe_consume_type_spilled_without_pos_recovery(&mut self) -> bool {
         // Consume type identifier
         debug_log!("maybe_consume_type_spilled_...: {}", debug_line!(self.source));
         if self.has_type_keyword() {
             self.consume_any_chars();
         } else {
             let ident = self.consume_namespaced_identifier();
             if ident.is_err() { return false; }
+        }
         // Consume any polymorphic arguments that follow the type identifier
         let mut backup_pos = self.source.pos();
         if self.consume_whitespace(false).is_err() { return false; }
         if !self.maybe_consume_poly_args_spilled_without_pos_recovery() { return false; }
         match self.maybe_consume_poly_args_spilled_without_pos_recovery() {
             Ok(true) => backup_pos = self.source.pos(),
             Ok(false) => {},
             Err(()) => return false
+        }
         // Consume any array specifiers. Make sure we always leave the input
         // position at the end of the last array specifier if we do find a
         // valid type
         if self.consume_whitespace(false).is_err() { return false; }
         while let Some(b'[') = self.source.next() {
             self.source.consume();
             if self.consume_whitespace(false).is_err() { return false; }
             if self.source.next() != Some(b']') { return false; }
             self.source.consume();
             backup_pos = self.source.pos();
             if self.consume_whitespace(false).is_err() { return false; }
+        }
         self.source.seek(backup_pos);
         return true;
@@ @@ -632,26 +665,28 @@ impl Lexer<'_> { @@
     fn maybe_consume_type_spilled(&mut self) -> bool {
         let backup_pos = self.source.pos();
         if !self.maybe_consume_type_spilled_without_pos_recovery() {
             self.source.seek(backup_pos);
             return false;
+        }
         return true;
+    }
     /// Attempts to consume polymorphic arguments without returning them. If it
     /// doesn't encounter well-formed polymorphic arguments, then the input
     /// position is left at a "random" position.
     fn maybe_consume_poly_args_spilled_without_pos_recovery(&mut self) -> bool {
     /// position is left at a "random" position. Returns a boolean indicating if
     /// the poly_args list was present.
     fn maybe_consume_poly_args_spilled_without_pos_recovery(&mut self) -> Result<bool, ()> {
         debug_log!("maybe_consume_poly_args_spilled_...: {}", debug_line!(self.source));
         self.consume_comma_separated_spilled_without_pos_recovery(
             b'<', b'>', |lexer| {
                 lexer.maybe_consume_type_spilled_without_pos_recovery()
             })
+    }
     /// Consumes polymorphic arguments and its delimiters if specified. If
     /// polyargs are present then the args are consumed and the input position
     /// will be placed after the polyarg list. If polyargs are not present then
     /// the input position will remain unmodified and an empty vector will be
     /// returned.
     ///
@@ @@ -1421,34 +1456,35 @@ impl Lexer<'_> { @@
             concrete_type: ConcreteType::default(),
         }))
+    }
     fn has_struct_literal(&mut self) -> bool {
         // A struct literal is written as:
         //      namespace::StructName<maybe_one_of_these, auto>{ field: expr }
         // We will parse up until the opening brace to see if we're dealing with
         // a struct literal.
         let backup_pos = self.source.pos();
         let result = self.consume_namespaced_identifier_spilled().is_ok() &&
             self.consume_whitespace(false).is_ok() &&
             self.maybe_consume_poly_args_spilled_without_pos_recovery() &&
+            self.maybe_consume_poly_args_spilled_without_pos_recovery().is_ok() &&
             self.consume_whitespace(false).is_ok() &&
             self.source.next() == Some(b'{');
         self.source.seek(backup_pos);
         return result;
+    }
     fn consume_struct_literal_expression(&mut self, h: &mut Heap) -> Result<LiteralExpressionId, ParseError2> {
         // Consume identifier and polymorphic arguments
         debug_log!("consume_struct_literal_expression: {}", debug_line!(self.source));
         let position = self.source.pos();
         let identifier = self.consume_namespaced_identifier()?;
         self.consume_whitespace(false)?;
         let poly_args = self.consume_polymorphic_args(h, true)?;
         self.consume_whitespace(false)?;
         // Consume fields
         let fields = match self.consume_comma_separated(
             h, b'{', b'}', "Expected the end of the list of struct fields",
             |lexer, heap| {
                 let identifier = lexer.consume_identifier()?;
                 lexer.consume_whitespace(false)?;
@@ @@ -1483,38 +1519,39 @@ impl Lexer<'_> { @@
     fn has_call_expression(&mut self) -> bool {
         // We need to prevent ambiguity with various operators (because we may
         // be specifying polymorphic variables) and variables.
         if self.has_builtin_keyword() {
             return true;
+        }
         let backup_pos = self.source.pos();
         let mut result = false;
         if self.consume_namespaced_identifier_spilled().is_ok() &&
             self.consume_whitespace(false).is_ok() &&
             self.maybe_consume_poly_args_spilled_without_pos_recovery() &&
+            self.maybe_consume_poly_args_spilled_without_pos_recovery().is_ok() &&
             self.consume_whitespace(false).is_ok() &&
             self.source.next() == Some(b'(') {
             // Seems like we have a function call or an enum literal
             result = true;
+        }
         self.source.seek(backup_pos);
         return result;
+    }
     fn consume_call_expression(&mut self, h: &mut Heap) -> Result<CallExpressionId, ParseError2> {
         let position = self.source.pos();
         // Consume method identifier
         debug_log!("consume_call_expression: {}", debug_line!(self.source));
         let method;
         if self.has_keyword(b"get") {
             self.consume_keyword(b"get")?;
             method = Method::Get;
         } else if self.has_keyword(b"put") {
             self.consume_keyword(b"put")?;
             method = Method::Put;
         } else if self.has_keyword(b"fires") {
             self.consume_keyword(b"fires")?;
             method = Method::Fires;
         } else if self.has_keyword(b"create") {
             self.consume_keyword(b"create")?;
@@ @@ -1555,24 +1592,25 @@ impl Lexer<'_> { @@
             method,
             arguments,
             poly_args,
             parent: ExpressionParent::None,
             concrete_type: ConcreteType::default(),
         }))
+    }
     fn consume_variable_expression(
         &mut self,
         h: &mut Heap,
     ) -> Result<VariableExpressionId, ParseError2> {
         let position = self.source.pos();
         debug_log!("consume_variable_expression: {}", debug_line!(self.source));
         let identifier = self.consume_namespaced_identifier()?;
         Ok(h.alloc_variable_expression(|this| VariableExpression {
             this,
             position,
             identifier,
             declaration: None,
             parent: ExpressionParent::None,
             concrete_type: ConcreteType::default(),
         }))
+    }
     // ====================
@@ @@ -2385,26 +2423,50 @@ impl Lexer<'_> { @@
                     module_id: None,
                     symbols,
                 }))
             } else if Some(b'*') == next {
                 self.source.consume();
                 h.alloc_import(|this| Import::Symbols(ImportSymbols{
                     this,
                     position,
                     module_name: value,
                     module_id: None,
                     symbols: Vec::new()
                 }))
             } else if self.has_identifier() {
                 let position = self.source.pos();
                 let name = self.consume_ident()?;
                 self.consume_whitespace(false)?;
                 let alias = if self.has_string(b"as") {
                     self.consume_string(b"as")?;
                     self.consume_whitespace(true)?;
                     self.consume_ident()?
                 } else {
                     name.clone()
                 };
                 h.alloc_import(|this| Import::Symbols(ImportSymbols{
                     this,
                     position,
                     module_name: value,
                     module_id: None,
                     symbols: vec![AliasedSymbol{
                         position,
                         name,
                         alias,
                         definition_id: None
                     }]
                 }))
             } else {
-                return Err(self.error_at_pos("Expected '*' or '{'"));
+                return Err(self.error_at_pos("Expected '*', '{' or a symbol name"));
+            }
         } else {
             // No explicit alias or subimports, so implicit alias
             let alias = Vec::from(&value[last_ident_start..]);
             h.alloc_import(|this| Import::Module(ImportModule{
                 this,
                 position,
                 module_name: value,
                 alias,
                 module_id: None,
             }))
         };

src/protocol/parser/mod.rs

➞

Show inline comments

@@ @@ -166,24 +166,26 @@ impl Parser { @@
                     import.module_id = Some(target_module_id);
                     for symbol in &mut import.symbols {
                         debug_assert!(symbol.definition_id.is_none(), "symbol import already resolved");
                         let (_, target_definition_id) = self.symbol_table.resolve_symbol(module_root_id, &symbol.alias)
                             .expect("symbol import is resolved by symbol table")
                             .as_definition()
                             .expect("symbol import does not resolve to namespace symbol");
                         symbol.definition_id = Some(target_definition_id);
+                    }
+                }
+            }
             import_index += 1;
+        }
         // All imports in the AST are now annotated. We now use the symbol table
         // to construct the type table.
         let mut type_ctx = TypeCtx::new(&self.symbol_table, &mut self.heap, &self.modules);
         self.type_table.build_base_types(&mut type_ctx)?;
         Ok(())
+    }
     pub fn parse(&mut self) -> Result<(), ParseError2> {
         self.resolve_symbols_and_types()?;

src/protocol/parser/type_resolver.rs

➞

Show inline comments

@@ @@ -18,52 +18,42 @@ @@
 /// point either all types are inferred (or can be trivially implicitly
 /// determined), or we have incomplete types. In the latter casee we return an
 /// error.
 ///
 /// Inference may be applied on non-polymorphic procedures and on polymorphic
 /// procedures. When dealing with a non-polymorphic procedure we apply the type
 /// resolver and annotate the AST with the `ConcreteType`s. When dealing with
 /// polymorphic procedures we will only annotate the AST once, preserving
 /// references to polymorphic variables. Any later pass will perform just the
 /// type checking.
 ///
 /// TODO: Needs a thorough rewrite:
 ///  0. polymorph_progress is intentionally broken at the moment.
 ///  1. For polymorphic type inference we need to have an extra datastructure
 ///     for progressing the polymorphic variables and mapping them back to each
 ///     signature type that uses that polymorphic type. The two types of markers
 ///     became somewhat of a mess.
 ///  2. We're doing a lot of extra work. It seems better to apply the initial
 ///     type based on expression parents, then to apply forced constraints (arg
 ///     to a fires() call must be port-like), only then to start progressing the
 ///     types.
 ///     Furthermore, queueing of expressions can be more intelligent, currently
 ///     every child/parent of an expression is inferred again when queued. Hence
 ///     we need to queue only specific children/parents of expressions.
 ///  3. Remove the `msg` type?
 ///  4. Disallow certain types in certain operations (e.g. `Void`).
 ///  5. Implement implicit and explicit casting.
 ///  6. Investigate different ways of performing the type-on-type inference,
 ///     maybe there is a better way then flattened trees + markers?
 macro_rules! enabled_debug_print {
     (false, $name:literal, $format:literal) => {};
     (false, $name:literal, $format:literal, $($args:expr),*) => {};
     (true, $name:literal, $format:literal) => {
         println!("[{}] {}", $name, $format)
     };
     (true, $name:literal, $format:literal, $($args:expr),*) => {
         println!("[{}] {}", $name, format!($format, $($args),*))
     };
+}
 macro_rules! debug_log {
     ($format:literal) => {
         enabled_debug_print!(true, "types", $format);
     };
     ($format:literal, $($args:expr),*) => {
         enabled_debug_print!(true, "types", $format, $($args),*);
     };
+}
 use std::collections::{HashMap, HashSet, VecDeque};
 use crate::protocol::ast::*;
@@ @@ -1980,24 +1970,27 @@ impl TypeResolvingVisitor { @@
             Literal::Null => {
                 self.apply_forced_constraint(ctx, upcast_id, &MESSAGE_TEMPLATE)?
             },
             Literal::Integer(_) => {
                 self.apply_forced_constraint(ctx, upcast_id, &INTEGERLIKE_TEMPLATE)?
             },
             Literal::True | Literal::False => {
                 self.apply_forced_constraint(ctx, upcast_id, &BOOL_TEMPLATE)?
             },
             Literal::Character(_) => todo!("character literals"),
             Literal::Struct(data) => {
                 let extra = self.extra_data.get_mut(&upcast_id).unwrap();
                 for poly in &extra.poly_vars {
                     debug_log!(" * Poly: {}", poly.display_name(&ctx.heap));
+                }
                 let mut poly_progress = HashSet::new();
                 debug_assert_eq!(extra.embedded.len(), data.fields.len());
                 debug_log!(" * During (inferring types from fields and struct type):");
                 // Mutually infer field signature/expression types
                 for (field_idx, field) in data.fields.iter().enumerate() {
                     let field_expr_id = field.value;
                     let signature_type: *mut _ = &mut extra.embedded[field_idx];
                     let field_type: *mut _ = self.expr_types.get_mut(&field_expr_id).unwrap();
                     let (_, progress_arg) = Self::apply_equal2_signature_constraint(
                         ctx, upcast_id, Some(field_expr_id), extra, &mut poly_progress,
@@ @@ -2006,37 +1999,40 @@ impl TypeResolvingVisitor { @@
                     debug_log!(
                         "   - Field {} type | sig: {}, field: {}", field_idx,
                         unsafe{&*signature_type}.display_name(&ctx.heap),
                         unsafe{&*field_type}.display_name(&ctx.heap)
                     );
                     if progress_arg {
                         self.expr_queued.insert(field_expr_id);
+                    }
+                }
                 debug_log!("   - Field poly progress | {:?}", poly_progress);
                 // Same for the type of the struct itself
                 let signature_type: *mut _ = &mut extra.returned;
                 let expr_type: *mut _ = self.expr_types.get_mut(&upcast_id).unwrap();
                 let (_, progress_expr) = Self::apply_equal2_signature_constraint(
                     ctx, upcast_id, None, extra, &mut poly_progress,
                     signature_type, 0, expr_type, 0
                 )?;
                 debug_log!(
                     "   - Ret type | sig: {}, expr: {}",
                     unsafe{&*signature_type}.display_name(&ctx.heap),
                     unsafe{&*expr_type}.display_name(&ctx.heap)
                 );
                 debug_log!("   - Ret poly progress | {:?}", poly_progress);
                 if progress_expr {
                     // TODO: @cleanup, cannot call utility self.queue_parent thingo
                     if let Some(parent_id) = ctx.heap[upcast_id].parent_expr_id() {
                         self.expr_queued.insert(parent_id);
+                    }
+                }
                 // Check which expressions use the polymorphic arguments. If the
                 // polymorphic variables have been progressed then we try to
                 // progress them inside the expression as well.
                 debug_log!(" * During (reinferring from progressed polyvars):");
@@ @@ -2443,51 +2439,51 @@ impl TypeResolvingVisitor { @@
     /// Applies equal2 constraints on the signature type for each of the
     /// polymorphic variables. If the signature type is progressed then we
     /// progress the expression type as well.
     ///
     /// This function assumes that the polymorphic variables have already been
     /// progressed as far as possible by calling
     /// `apply_equal2_signature_constraint`. As such, we expect to not encounter
     /// any errors.
     ///
     /// This function returns true if the expression's type has been progressed
     fn apply_equal2_polyvar_constraint(
         heap: &Heap,
-        polymorph_data: &ExtraData, polymorph_progress: &HashSet<usize>,
+        polymorph_data: &ExtraData, _polymorph_progress: &HashSet<usize>,
         signature_type: *mut InferenceType, expr_type: *mut InferenceType
     ) -> bool {
         // Safety: all pointers should be distinct
         //         polymorph_data contains may not be modified
         debug_assert_ptrs_distinct!(signature_type, expr_type);
         let signature_type = unsafe{&mut *signature_type};
         let expr_type = unsafe{&mut *expr_type};
         // Iterate through markers in signature type to try and make progress
         // on the polymorphic variable
         let mut seek_idx = 0;
         let mut modified_sig = false;
         while let Some((poly_idx, start_idx)) = signature_type.find_body_marker(seek_idx) {
             let end_idx = InferenceType::find_subtree_end_idx(&signature_type.parts, start_idx);
             if polymorph_progress.contains(&poly_idx) {
+            // if polymorph_progress.contains(&poly_idx) {
                 // Need to match subtrees
                 let polymorph_type = &polymorph_data.poly_vars[poly_idx];
                 debug_log!("   - DEBUG: Applying {} to '{}' from '{}'", polymorph_type.display_name(heap), InferenceType::partial_display_name(heap, &signature_type.parts[start_idx..]), signature_type.display_name(heap));
                 let modified_at_marker = Self::apply_forced_constraint_types(
                     signature_type, start_idx,
                     &polymorph_type.parts, 0
                 ).expect("no failure when applying polyvar constraints");
                 modified_sig = modified_sig || modified_at_marker;
+            }
+            // }
             seek_idx = end_idx;
+        }
         // If we made any progress on the signature's type, then we also need to
         // apply it to the expression that is supposed to match the signature.
         if modified_sig {
             match InferenceType::infer_subtree_for_single_type(
                 expr_type, 0, &signature_type.parts, 0
             ) {
                 SingleInferenceResult::Modified => true,
                 SingleInferenceResult::Unmodified => false,
@@ @@ -2829,28 +2825,25 @@ impl TypeResolvingVisitor { @@
         });
+    }
     /// Inserts the extra polymorphic data struct. Assumes that the select
     /// expression's referenced (definition_id, field_idx) has been resolved.
     fn insert_initial_select_polymorph_data(
         &mut self, ctx: &Ctx, select_id: SelectExpressionId
     ) {
         use InferenceTypePart as ITP;
         // Retrieve relevant data
         let expr = &ctx.heap[select_id];
         let field = match &expr.field {
             Field::Symbolic(field) => field,
             _ => unreachable!(),
         };
         let field = expr.field.as_symbolic();
         let definition_id = field.definition.unwrap();
         let definition = ctx.heap[definition_id].as_struct();
         let field_idx = field.field_idx;
         // Generate initial polyvar types and struct type
         let num_poly_vars = definition.poly_vars.len();
         let mut poly_vars = Vec::with_capacity(num_poly_vars);
         let struct_parts_reserved = 1 + 2 * num_poly_vars;
         let mut struct_parts = Vec::with_capacity(struct_parts_reserved);
         struct_parts.push(ITP::Instance(definition_id, num_poly_vars));
@@ @@ -3111,80 +3104,98 @@ impl TypeResolvingVisitor { @@
                             String::from_utf8_lossy(&definition.poly_vars[poly_var_idx].value).to_string()
                         },
                         Definition::Component(definition) => {
                             String::from_utf8_lossy(&definition.poly_vars[poly_var_idx].value).to_string()
+                        }
                     };
                     let func_name = String::from_utf8_lossy(&symbolic.identifier.value).to_string();
                     (poly_var, func_name)
+                }
+            }
+        }
         fn get_poly_var_and_literal_name(ctx: &Ctx, poly_var_idx: usize, expr: &LiteralExpression) -> (String, String) {
             let expr = expr.value.as_struct();
             let definition = &ctx.heap[expr.definition.unwrap()];
         fn get_poly_var_and_type_name(ctx: &Ctx, poly_var_idx: usize, definition_id: DefinitionId) -> (String, String) {
             let definition = &ctx.heap[definition_id];
             match definition {
                 Definition::Enum(_) | Definition::Function(_) | Definition::Component(_) =>
                     unreachable!(),
+                    unreachable!("get_poly_var_and_type_name called on non-struct value"),
                 Definition::Struct(definition) => (
                     String::from_utf8_lossy(&definition.poly_vars[poly_var_idx].value).to_string(),
                     String::from_utf8_lossy(&definition.identifier.value).to_string()
                 ),
+            }
+        }
         // Helper function to construct initial error
         fn construct_main_error(ctx: &Ctx, poly_var_idx: usize, expr: &Expression) -> ParseError2 {
             match expr {
                 Expression::Call(expr) => {
                     let (poly_var, func_name) = get_poly_var_and_func_name(ctx, poly_var_idx, expr);
                     return ParseError2::new_error(
                         &ctx.module.source, expr.position(),
                         &format!(
                             "Conflicting type for polymorphic variable '{}' of '{}'",
                             poly_var, func_name
+                        )
+                    )
                 },
                 Expression::Literal(expr) => {
                     let (poly_var, struct_name) = get_poly_var_and_literal_name(ctx, poly_var_idx, expr);
                     let lit_struct = expr.value.as_struct();
                     let (poly_var, struct_name) = get_poly_var_and_type_name(ctx, poly_var_idx, lit_struct.definition.unwrap());
                     return ParseError2::new_error(
                         &ctx.module.source, expr.position(),
                         &format!(
                             "Conflicting type for polymorphic variable '{}' of instantiation of '{}'",
                             poly_var, struct_name
+                        )
+                    )
                 },
                 Expression::Select(expr) => {
                     let field = expr.field.as_symbolic();
                     let (poly_var, struct_name) = get_poly_var_and_type_name(ctx, poly_var_idx, field.definition.unwrap());
                     return ParseError2::new_error(
                         &ctx.module.source, expr.position(),
                         &format!(
                             "Conflicting type for polymorphic variable '{}' while accessing field '{}' of '{}'",
                             poly_var, &String::from_utf8_lossy(&field.identifier.value), struct_name
+                        )
+                    )
+                }
                 _ => unreachable!("called construct_poly_arg_error without a call/literal expression")
+            }
+        }
         // Actual checking
         let expr = &ctx.heap[expr_id];
         let (expr_args, expr_return_name) = match expr {
             Expression::Call(expr) =>
+                (
                     expr.arguments.clone(),
                     "return type"
                 ),
             Expression::Literal(expr) =>
+                (
                     expr.value.as_struct().fields
                         .iter()
                         .map(|f| f.value)
                         .collect(),
                     "literal"
                 ),
             Expression::Select(expr) =>
                 // Select expression uses the polymorphic variables of the
                 // struct it is accessing, so get the subject expression.
+                (
                     vec![expr.subject],
                     "selected field"
                 ),
             _ => unreachable!(),
         };
         // - check return type with itself
         if let Some((poly_idx, section_a, section_b)) = has_poly_mismatch(
             &poly_data.returned, &poly_data.returned
         ) {
             return construct_main_error(ctx, poly_idx, expr)
                 .with_postfixed_info(
                     &ctx.module.source, expr.position(),
                     &format!(
                         "The {} inferred the conflicting types '{}' and '{}'",

src/protocol/tests/mod.rs

➞

Show inline comments

 mod utils;
 mod lexer;
 mod parser_validation;
 mod parser_inference;
 mod parser_monomorphs;
 mod parser_imports;
 pub(crate) use utils::{Tester};
@@ \ No newline at end of file @@

src/protocol/tests/parser_imports.rs

➞

Show inline comments

@@ new file 100644 @@
 /// parser_imports.rs
 ///
 /// Simple import tests
 use super::*;
 #[test]
 fn test_module_import() {
     Tester::new("single domain name")
         .with_source("
         #module external
         struct Foo { int field }
         ")
         .with_source("
         import external;
         int caller() {
             auto a = external::Foo{ field: 0 };
             return a.field;
+        }
         ")
         .compile()
         .expect_ok();
     Tester::new("multi domain name")
         .with_source("
         #module external.domain
         struct Foo { int field }
         ")
         .with_source("
         import external.domain;
         int caller() {
             auto a = domain::Foo{ field: 0 };
             return a.field;
+        }
         ")
         .compile()
         .expect_ok();
     Tester::new("aliased domain name")
         .with_source("
         #module external
         struct Foo { int field }
         ")
         .with_source("
         import external as aliased;
         int caller() {
             auto a = aliased::Foo{ field: 0 };
             return a.field;
+        }
         ")
         .compile()
         .expect_ok();
+}
 #[test]
 fn test_single_symbol_import() {
     Tester::new("specific symbol")
         .with_source("
         #module external
         struct Foo { int field }
         ")
         .with_source("
         import external::Foo;
         int caller() {
             auto a = Foo{ field: 1 };
             auto b = Foo{ field: 2 };
             return a.field + b.field;
         }")
         .compile()
         .expect_ok();
     Tester::new("specific aliased symbol")
         .with_source("
         #module external
         struct Foo { int field }
         ")
         .with_source("
         import external::Foo as Bar;
         int caller() {
             return Bar{ field: 0 }.field;
+        }
         ")
         .compile()
         .expect_ok();
     // TODO: Re-enable once std lib is properly implemented
     // Tester::new("import all")
     //     .with_source("
     //     #module external
     //     struct Foo { int field }
     //     ")
     //     .with_source("
     //     import external::*;
     //     int caller() { return Foo{field:0}.field; }
     //     ")
     //     .compile()
     //     .expect_ok();
+}
 #[test]
 fn test_multi_symbol_import() {
     Tester::new("specific symbols")
         .with_source("
         #module external
         struct Foo { byte f }
         struct Bar { byte b }
         ")
         .with_source("
         import external::{Foo, Bar};
         byte caller() {
             return Foo{f:0}.f + Bar{b:1}.b;
+        }
         ")
         .compile()
         .expect_ok();
     Tester::new("aliased symbols")
         .with_source("
         #module external
         struct Foo { byte in_foo }
         struct Bar { byte in_bar }
         ")
         .with_source("
         import external::{Foo as Bar, Bar as Foo};
         byte caller() {
             return Foo{in_bar:0}.in_bar + Bar{in_foo:0}.in_foo;
         }")
         .compile()
         .expect_ok();
     // TODO: Re-enable once std lib is properly implemented
     // Tester::new("import all")
     //     .with_source("
     //     #module external
     //     struct Foo { byte f };
     //     struct Bar { byte b };
     //     ")
     //     .with_source("
     //     import external::*;
     //     byte caller() {
     //         auto f = Foo{f:0};
     //         auto b = Bar{b:0};
     //         return f.f + b.b;
     //     }
     //     ")
     //     .compile()
     //     .expect_ok();
+}
@@ \ No newline at end of file @@

src/protocol/tests/parser_inference.rs

➞

Show inline comments

@@ @@ -58,24 +58,81 @@ fn test_integer_inference() { @@
         })
         .for_variable("i2", |v| { v
             .assert_parser_type("auto")
             .assert_concrete_type("int");
         })
         .for_variable("l2", |v| { v
             .assert_parser_type("auto")
             .assert_concrete_type("long");
         });
     });
+}
 #[test]
 fn test_binary_expr_inference() {
     Tester::new_single_source_expect_ok(
         "compatible types",
         "int call() {
             byte b0 = 0;
             byte b1 = 1;
             short s0 = 0;
             short s1 = 1;
             int i0 = 0;
             int i1 = 1;
             long l0 = 0;
             long l1 = 1;
             auto b = b0 + b1;
             auto s = s0 + s1;
             auto i = i0 + i1;
             auto l = l0 + l1;
             return i;
         }"
     ).for_function("call", |f| { f
         .for_expression_by_source(
             "b0 + b1", "+",
             |e| { e.assert_concrete_type("byte"); }
+        )
         .for_expression_by_source(
             "s0 + s1", "+",
             |e| { e.assert_concrete_type("short"); }
+        )
         .for_expression_by_source(
             "i0 + i1", "+",
             |e| { e.assert_concrete_type("int"); }
+        )
         .for_expression_by_source(
             "l0 + l1", "+",
             |e| { e.assert_concrete_type("long"); }
         );
     });
     Tester::new_single_source_expect_err(
         "incompatible types",
         "int call() {
             byte b = 0;
             long l = 1;
             auto r = b + l;
             return 0;
         }"
     ).error(|e| { e
         .assert_ctx_has(0, "b + l")
         .assert_msg_has(0, "cannot apply")
         .assert_occurs_at(0, "+")
         .assert_msg_has(1, "has type 'byte'")
         .assert_msg_has(2, "has type 'long'");
     });
+}
 #[test]
 fn test_struct_inference() {
     Tester::new_single_source_expect_ok(
         "by function calls",
+        "
         struct Pair<T1, T2>{ T1 first, T2 second }
         Pair<T1, T2> construct<T1, T2>(T1 first, T2 second) {
             return Pair{ first: first, second: second };
+        }
         int fix_t1<T2>(Pair<byte, T2> arg) { return 0; }
         int fix_t2<T1>(Pair<T1, int> arg) { return 0; }
         int test() {
@@ @@ -146,13 +203,96 @@ fn test_struct_inference() { @@
             auto thing = construct(assigned, construct(0, 1));
             fix_poly(thing.r);
             thing.r.r = assigned;
             return 0;
+        }
         ",
     ).for_function("test", |f| { f
         .for_variable("thing", |v| { v
             .assert_parser_type("auto")
             .assert_concrete_type("Node<byte,Node<byte,byte>>");
         });
     });
+}
 #[test]
 fn test_failed_polymorph_inference() {
     Tester::new_single_source_expect_err(
         "function call inference mismatch",
+        "
         int poly<T>(T a, T b) { return 0; }
         int call() {
             byte first_arg = 5;
             long second_arg = 2;
             return poly(first_arg, second_arg);
+        }
+        "
     ).error(|e| { e
         .assert_num(3)
         .assert_ctx_has(0, "poly(first_arg, second_arg)")
         .assert_occurs_at(0, "poly")
         .assert_msg_has(0, "Conflicting type for polymorphic variable 'T'")
         .assert_occurs_at(1, "second_arg")
         .assert_msg_has(1, "inferred it to 'long'")
         .assert_occurs_at(2, "first_arg")
         .assert_msg_has(2, "inferred it to 'byte'");
     });
     Tester::new_single_source_expect_err(
         "struct literal inference mismatch",
+        "
         struct Pair<T>{ T first, T second }
         int call() {
             byte first_arg = 5;
             long second_arg = 2;
             auto pair = Pair{ first: first_arg, second: second_arg };
             return 3;
+        }
+        "
     ).error(|e| { e
         .assert_num(3)
         .assert_ctx_has(0, "Pair{ first: first_arg, second: second_arg }")
         .assert_occurs_at(0, "Pair{")
         .assert_msg_has(0, "Conflicting type for polymorphic variable 'T'")
         .assert_occurs_at(1, "second_arg")
         .assert_msg_has(1, "inferred it to 'long'")
         .assert_occurs_at(2, "first_arg")
         .assert_msg_has(2, "inferred it to 'byte'");
     });
     Tester::new_single_source_expect_err(
         "field access inference mismatch",
+        "
         struct Holder<Shazam>{ Shazam a }
         int call() {
             byte to_hold = 0;
             auto holder = Holder{ a: to_hold };
             return holder.a;
+        }
+        "
     ).error(|e| { e
         .assert_num(3)
         .assert_ctx_has(0, "holder.a")
         .assert_occurs_at(0, ".")
         .assert_msg_has(0, "Conflicting type for polymorphic variable 'Shazam'")
         .assert_msg_has(1, "inferred it to 'byte'")
         .assert_msg_has(2, "inferred it to 'int'");
     });
     // TODO: Needs better error messages anyway, but this failed before
     Tester::new_single_source_expect_err(
         "by nested field access",
+        "
         struct Node<T1, T2>{ T1 l, T2 r }
         Node<T1, T2> construct<T1, T2>(T1 l, T2 r) { return Node{ l: l, r: r }; }
         int fix_poly<T>(Node<T, T> a) { return 0; }
         int test() {
             byte assigned = 0;
             long another = 1;
             auto thing = construct(assigned, construct(another, 1));
             fix_poly(thing.r);
             thing.r.r = assigned;
             return 0;
+        }
         ",
     );
+}
@@ \ No newline at end of file @@

src/protocol/tests/parser_monomorphs.rs

➞

Show inline comments

@@ @@ -25,14 +25,17 @@ fn test_struct_monomorphs() { @@
             auto d = Number<long>{ number: 3 };
             auto e = Number<Number<short>>{ number: Number{ number: 4 }};
             return 0;
+        }
+        "
     ).for_struct("Number", |s| { s
         .assert_has_monomorph("byte")
         .assert_has_monomorph("short")
         .assert_has_monomorph("int")
         .assert_has_monomorph("long")
         .assert_has_monomorph("Number<short>")
         .assert_num_monomorphs(5);
     }).for_function("instantiator", |f| { f
         .for_variable("a", |v| {v.assert_concrete_type("Number<byte>");} )
         .for_variable("e", |v| {v.assert_concrete_type("Number<Number<short>>");} );
     });
+}
@@ \ No newline at end of file @@

src/protocol/tests/utils.rs

➞

Show inline comments

@@ @@ -389,33 +389,86 @@ impl<'a> FunctionTester<'a> { @@
         let assignment = &self.ctx.heap[assignment_id.unwrap()];
         // Construct tester and pass to tester function
         let tester = VariableTester::new(
             self.ctx, self.def.this.upcast(), local,
             assignment.as_assignment()
         );
         f(tester);
         self
+    }
     /// Finds a specific expression within a function. There are two matchers:
     /// one outer matcher (to find a rough indication of the expression) and an
     /// inner matcher to find the exact expression.
     ///
     /// The reason being that, for example, a function's body might be littered
     /// with addition symbols, so we first match on "some_var + some_other_var",
     /// and then match exactly on "+".
     pub(crate) fn for_expression_by_source<F: Fn(ExpressionTester)>(self, outer_match: &str, inner_match: &str, f: F) -> Self {
         // Seek the expression in the source code
         assert!(outer_match.contains(inner_match), "improper testing code");
         let module = seek_def_in_modules(
             &self.ctx.heap, &self.ctx.modules, self.def.this.upcast()
         ).unwrap();
         // Find the first occurrence of the expression after the definition of
         // the function, we'll check that it is included in the body later.
         let mut outer_match_idx = self.def.position.offset;
         while outer_match_idx < module.source.input.len() {
             if module.source.input[outer_match_idx..].starts_with(outer_match.as_bytes()) {
                 break;
+            }
             outer_match_idx += 1
+        }
         assert!(
             outer_match_idx < module.source.input.len(),
             "[{}] Failed to find '{}' within the source that contains {}",
             self.ctx.test_name, outer_match, self.assert_postfix()
         );
         let inner_match_idx = outer_match_idx + outer_match.find(inner_match).unwrap();
         // Use the inner match index to find the expression
         let expr_id = seek_expr_in_stmt(
             &self.ctx.heap, self.def.body,
             &|expr| expr.position().offset == inner_match_idx
         );
         assert!(
             expr_id.is_some(),
             "[{}] Failed to find '{}' within the source that contains {} \
             (note: expression was found, but not within the specified function",
             self.ctx.test_name, outer_match, self.assert_postfix()
         );
         let expr_id = expr_id.unwrap();
         // We have the expression, call the testing function
         let tester = ExpressionTester::new(
             self.ctx, self.def.this.upcast(), &self.ctx.heap[expr_id]
         );
         f(tester);
         self
+    }
     fn assert_postfix(&self) -> String {
         format!(
             "Function{{ name: {} }}",
             &String::from_utf8_lossy(&self.def.identifier.value)
+        )
+    }
+}
 pub(crate) struct VariableTester<'a> {
     ctx: TestCtx<'a>,
     definition_id: DefinitionId,
     local: &'a Local,
     assignment: &'a AssignmentExpression,
+}
 impl<'a> VariableTester<'a> {
     fn new(
         ctx: TestCtx<'a>, definition_id: DefinitionId, local: &'a Local, assignment: &'a AssignmentExpression
     ) -> Self {
         Self{ ctx, definition_id, local, assignment }
@@ @@ -447,24 +500,60 @@ impl<'a> VariableTester<'a> { @@
         );
         self
+    }
     fn assert_postfix(&self) -> String {
         format!(
             "Variable{{ name: {} }}",
             &String::from_utf8_lossy(&self.local.identifier.value)
+        )
+    }
+}
 pub(crate) struct ExpressionTester<'a> {
     ctx: TestCtx<'a>,
     definition_id: DefinitionId, // of the enclosing function/component
     expr: &'a Expression
+}
 impl<'a> ExpressionTester<'a> {
     fn new(
         ctx: TestCtx<'a>, definition_id: DefinitionId, expr: &'a Expression
     ) -> Self {
         Self{ ctx, definition_id, expr }
+    }
     pub(crate) fn assert_concrete_type(self, expected: &str) -> Self {
         let mut serialized = String::new();
         serialize_concrete_type(
             &mut serialized, self.ctx.heap, self.definition_id,
             self.expr.get_type()
         );
         assert_eq!(
             expected, &serialized,
             "[{}] Expected concrete type '{}', but got '{}' for {}",
             self.ctx.test_name, expected, &serialized, self.assert_postfix()
         );
         self
+    }
     fn assert_postfix(&self) -> String {
         format!(
             "Expression{{ debug: {:?} }}",
             self.expr
+        )
+    }
+}
 //------------------------------------------------------------------------------
 // Interface for failed compilation
 //------------------------------------------------------------------------------
 pub(crate) struct AstErrTester {
     test_name: String,
     error: ParseError2,
+}
 impl AstErrTester {
     fn new(test_name: String, error: ParseError2) -> Self {
         Self{ test_name, error }
                     serialize_parser_type(buffer, heap, *poly_arg);
+                }
                 buffer.push('>');
+            }
+        }
+    }
+}
 fn serialize_concrete_type(buffer: &mut String, heap: &Heap, def: DefinitionId, concrete: &ConcreteType) {
     // Retrieve polymorphic variables, if present (since we're dealing with a
     // concrete type we only expect procedure types)
     let poly_vars = match &heap[def] {
         Definition::Function(func) => &func.poly_vars,
         Definition::Component(comp) => &comp.poly_vars,
         _ => unreachable!("Error in testing utility: did not expect non-procedure type for concrete type serialization"),
         Definition::Function(definition) => &definition.poly_vars,
         Definition::Component(definition) => &definition.poly_vars,
         Definition::Struct(definition) => &definition.poly_vars,
         _ => unreachable!("Error in testing utility: unexpected type for concrete type serialization"),
     };
     fn serialize_recursive(
         buffer: &mut String, heap: &Heap, poly_vars: &Vec<Identifier>, concrete: &ConcreteType, mut idx: usize
     ) -> usize {
         use ConcreteTypePart as CTP;
         let part = &concrete.parts[idx];
         match part {
             CTP::Marker(poly_idx) => {
                 buffer.push_str(&String::from_utf8_lossy(&poly_vars[*poly_idx].value));
             },
+                }
                 buffer.push('>');
                 idx += 1;
+            }
+        }
         idx
+    }
     serialize_recursive(buffer, heap, poly_vars, concrete, 0);
+}
 fn seek_def_in_modules<'a>(heap: &Heap, modules: &'a [LexedModule], def_id: DefinitionId) -> Option<&'a LexedModule> {
     for module in modules {
         let root = &heap.protocol_descriptions[module.root_id];
         for definition in &root.definitions {
             if *definition == def_id {
                 return Some(module)
+            }
+        }
+    }
     None
+}
 fn seek_stmt<F: Fn(&Statement) -> bool>(heap: &Heap, start: StatementId, f: &F) -> Option<StatementId> {
     let stmt = &heap[start];
     if f(stmt) { return Some(start); }
     // This statement wasn't it, try to recurse
     let matched = match stmt {
         Statement::Block(block) => {
             for sub_id in &block.statements {
                 if let Some(id) = seek_stmt(heap, *sub_id, f) {
                     return Some(id);
+                }
+            }

0 comments (0 inline, 0 general)