CSY/reowolf Changeset - c9800c8f19d7 · Centrum Wiskunde & Informatica (CWI)

Changeset - c9800c8f19d7

Parent rev.

Child rev.

[Not reviewed]

0 17 0

mh - 3 years ago 2022-03-29 16:28:33
contact@maxhenger.nl

Rewrite tokenizer to emit markers iso ranges

17 files changed with 286 insertions and 194 deletions:

src/protocol/ast.rs

src/protocol/mod.rs

src/protocol/parser/mod.rs

src/protocol/parser/pass_definitions.rs

src/protocol/parser/pass_definitions_types.rs

src/protocol/parser/pass_imports.rs

src/protocol/parser/pass_rewriting.rs

src/protocol/parser/pass_symbols.rs

src/protocol/parser/pass_tokenizer.rs

src/protocol/parser/pass_typing.rs

src/protocol/parser/pass_validation_linking.rs

src/protocol/parser/symbol_table.rs

src/protocol/parser/token_parsing.rs

src/protocol/parser/tokens.rs

src/protocol/tests/utils.rs

src/protocol/token_writer.rs

std/std.global.pdl

0 comments (0 inline, 0 general)

src/protocol/ast.rs

➞

Show inline comments

@@ @@ -911,122 +911,119 @@ impl Definition { @@
+        }
+    }
     pub fn poly_vars(&self) -> &Vec<Identifier> {
         match self {
             Definition::Struct(def) => &def.poly_vars,
             Definition::Enum(def) => &def.poly_vars,
             Definition::Union(def) => &def.poly_vars,
             Definition::Procedure(def) => &def.poly_vars,
+        }
+    }
+}
 #[derive(Debug, Clone)]
 pub struct StructFieldDefinition {
     pub span: InputSpan,
     pub field: Identifier,
     pub parser_type: ParserType,
+}
 #[derive(Debug, Clone)]
 pub struct StructDefinition {
     pub this: StructDefinitionId,
     pub defined_in: RootId,
     // Symbol scanning
     pub span: InputSpan,
     pub identifier: Identifier,
     pub poly_vars: Vec<Identifier>,
     // Parsing
     pub fields: Vec<StructFieldDefinition>
+}
 impl StructDefinition {
     pub(crate) fn new_empty(
-        this: StructDefinitionId, defined_in: RootId, span: InputSpan,
         this: StructDefinitionId, defined_in: RootId,
         identifier: Identifier, poly_vars: Vec<Identifier>
     ) -> Self {
-        Self{ this, defined_in, span, identifier, poly_vars, fields: Vec::new() }
         Self{ this, defined_in, identifier, poly_vars, fields: Vec::new() }
+    }
+}
 #[derive(Debug, Clone, Copy)]
 pub enum EnumVariantValue {
     None,
     Integer(i64),
+}
 #[derive(Debug, Clone)]
 pub struct EnumVariantDefinition {
     pub identifier: Identifier,
     pub value: EnumVariantValue,
+}
 #[derive(Debug, Clone)]
 pub struct EnumDefinition {
     pub this: EnumDefinitionId,
     pub defined_in: RootId,
     // Symbol scanning
     pub span: InputSpan,
     pub identifier: Identifier,
     pub poly_vars: Vec<Identifier>,
     // Parsing
     pub variants: Vec<EnumVariantDefinition>,
+}
 impl EnumDefinition {
     pub(crate) fn new_empty(
-        this: EnumDefinitionId, defined_in: RootId, span: InputSpan,
         this: EnumDefinitionId, defined_in: RootId,
         identifier: Identifier, poly_vars: Vec<Identifier>
     ) -> Self {
-        Self{ this, defined_in, span, identifier, poly_vars, variants: Vec::new() }
         Self{ this, defined_in, identifier, poly_vars, variants: Vec::new() }
+    }
+}
 #[derive(Debug, Clone)]
 pub struct UnionVariantDefinition {
     pub span: InputSpan,
     pub identifier: Identifier,
     pub value: Vec<ParserType>, // if empty, then union variant does not contain any embedded types
+}
 #[derive(Debug, Clone)]
 pub struct UnionDefinition {
     pub this: UnionDefinitionId,
     pub defined_in: RootId,
     // Phase 1: symbol scanning
     pub span: InputSpan,
     pub identifier: Identifier,
     pub poly_vars: Vec<Identifier>,
     // Phase 2: parsing
     pub variants: Vec<UnionVariantDefinition>,
+}
 impl UnionDefinition {
     pub(crate) fn new_empty(
-        this: UnionDefinitionId, defined_in: RootId, span: InputSpan,
         this: UnionDefinitionId, defined_in: RootId,
         identifier: Identifier, poly_vars: Vec<Identifier>
     ) -> Self {
-        Self{ this, defined_in, span, identifier, poly_vars, variants: Vec::new() }
         Self{ this, defined_in, identifier, poly_vars, variants: Vec::new() }
+    }
+}
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub enum ProcedureKind {
     Function, // with return type
     Primitive, // without return type
     Composite,
+}
 /// Monomorphed instantiation of a procedure (or the sole instantiation of a
 /// non-polymorphic procedure).
 #[derive(Debug)]
 pub struct ProcedureDefinitionMonomorph {
     pub argument_types: Vec<TypeId>,
     pub expr_info: Vec<ExpressionInfo>
+}
 impl ProcedureDefinitionMonomorph {
     pub(crate) fn new_invalid() -> Self {
         return Self{
             argument_types: Vec::new(),
             expr_info: Vec::new(),
+        }
@@ @@ -1090,69 +1087,67 @@ pub enum ProcedureSource { @@
     // Builtin components, available to user
     CompRandomU32, // TODO: Remove, temporary thing
+}
 impl ProcedureSource {
     pub(crate) fn is_builtin(&self) -> bool {
         match self {
             ProcedureSource::FuncUserDefined | ProcedureSource::CompUserDefined => false,
             _ => true,
+        }
+    }
+}
 /// Generic storage for functions, primitive components and composite
 /// components.
 // Note that we will have function definitions for builtin functions as well. In
 // that case the span, the identifier span and the body are all invalid.
 #[derive(Debug)]
 pub struct ProcedureDefinition {
     pub this: ProcedureDefinitionId,
     pub defined_in: RootId,
     // Symbol scanning
     pub kind: ProcedureKind,
     pub span: InputSpan,
     pub identifier: Identifier,
     pub poly_vars: Vec<Identifier>,
     // Parser
     pub source: ProcedureSource,
     pub return_type: Option<ParserType>, // present on functions, not components
     pub parameters: Vec<VariableId>,
     pub scope: ScopeId,
     pub body: BlockStatementId,
     // Monomorphization of typed procedures
     pub monomorphs: Vec<ProcedureDefinitionMonomorph>,
+}
 impl ProcedureDefinition {
     pub(crate) fn new_empty(
-        this: ProcedureDefinitionId, defined_in: RootId, span: InputSpan,
         this: ProcedureDefinitionId, defined_in: RootId,
         kind: ProcedureKind, identifier: Identifier, poly_vars: Vec<Identifier>
     ) -> Self {
         Self {
             this, defined_in,
             span,
             kind, identifier, poly_vars,
             source: ProcedureSource::FuncUserDefined,
             return_type: None,
             parameters: Vec::new(),
             scope: ScopeId::new_invalid(),
             body: BlockStatementId::new_invalid(),
             monomorphs: Vec::new(),
+        }
+    }
+}
 #[derive(Debug, Clone)]
 pub enum Statement {
     Block(BlockStatement),
     EndBlock(EndBlockStatement),
     Local(LocalStatement),
     Labeled(LabeledStatement),
     If(IfStatement),
     EndIf(EndIfStatement),
     While(WhileStatement),
     EndWhile(EndWhileStatement),
     Break(BreakStatement),
     Continue(ContinueStatement),
     Synchronous(SynchronousStatement),

src/protocol/mod.rs

➞

Show inline comments

@@ @@ -39,49 +39,48 @@ pub(crate) struct ComponentState { @@
+}
 #[derive(Debug)]
 pub enum ComponentCreationError {
     ModuleDoesntExist,
     DefinitionDoesntExist,
     DefinitionNotComponent,
     InvalidNumArguments,
     InvalidArgumentType(usize),
     UnownedPort,
     InSync,
+}
 impl ProtocolDescription {
     pub fn parse(buffer: &[u8]) -> Result<Self, String> {
         let source = InputSource::new(String::new(), Vec::from(buffer));
         let mut parser = Parser::new()?;
         parser.feed(source).expect("failed to feed source");
         if let Err(err) = parser.parse() {
             println!("ERROR:\n{}", err);
             return Err(format!("{}", err))
+        }
         debug_assert_eq!(parser.modules.len(), 1, "only supporting one module here for now");
         let modules: Vec<Module> = parser.modules.into_iter()
             .map(|module| Module{
                 source: module.source,
                 root_id: module.root_id,
                 name: module.name.map(|(_, name)| name)
             })
             .collect();
         return Ok(ProtocolDescription {
             modules,
             heap: parser.heap,
             types: parser.type_table,
             pool: Mutex::new(parser.string_pool),
         });
+    }
     pub(crate) fn new_component(
         &self, module_name: &[u8], identifier: &[u8], arguments: ValueGroup
     ) -> Result<Prompt, ComponentCreationError> {
         // Find the module in which the definition can be found
         let module_root = self.lookup_module_root(module_name);
         if module_root.is_none() {
             return Err(ComponentCreationError::ModuleDoesntExist);
+        }

src/protocol/parser/mod.rs

➞

Show inline comments

@@ @@ -31,52 +31,51 @@ use crate::protocol::ast::*; @@
 use crate::protocol::input_source::*;
 use crate::protocol::ast_writer::ASTWriter;
 use crate::protocol::parser::type_table::PolymorphicVariable;
 use crate::protocol::token_writer::TokenWriter;
 const REOWOLF_PATH_ENV: &'static str = "REOWOLF_ROOT"; // first lookup reowolf path
 const REOWOLF_PATH_DIR: &'static str = "std"; // then try folder in current working directory
 #[derive(Debug, PartialEq, Eq, PartialOrd, Ord)]
 pub enum ModuleCompilationPhase {
     Tokenized,              // source is tokenized
     SymbolsScanned,         // all definitions are linked to their type class
     ImportsResolved,        // all imports are added to the symbol table
     DefinitionsParsed,      // produced the AST for the entire module
     TypesAddedToTable,      // added all definitions to the type table
     ValidatedAndLinked,     // AST is traversed and has linked the required AST nodes
     Typed,                  // Type inference and checking has been performed
     Rewritten,              // Special AST nodes are rewritten into regular AST nodes
     // When we continue with the compiler:
     // StackSize
+}
 pub struct Module {
     // Buffers
     pub source: InputSource,
     pub tokens: TokenBuffer,
-    // Identifiers
+    pub is_compiler_file: bool, // TODO: @Hack
     pub root_id: RootId,
     pub name: Option<(PragmaId, StringRef<'static>)>,
     pub version: Option<(PragmaId, i64)>,
     pub phase: ModuleCompilationPhase,
+}
 pub struct TargetArch {
     pub void_type_id: TypeId,
     pub message_type_id: TypeId,
     pub bool_type_id: TypeId,
     pub uint8_type_id: TypeId,
     pub uint16_type_id: TypeId,
     pub uint32_type_id: TypeId,
     pub uint64_type_id: TypeId,
     pub sint8_type_id: TypeId,
     pub sint16_type_id: TypeId,
     pub sint32_type_id: TypeId,
     pub sint64_type_id: TypeId,
     pub char_type_id: TypeId,
     pub string_type_id: TypeId,
     pub array_type_id: TypeId,
     pub slice_type_id: TypeId,
     pub input_type_id: TypeId,
     pub output_type_id: TypeId,
@@ @@ -135,101 +134,87 @@ pub struct Parser { @@
     pass_stack_size: PassStackSize,
     // Compiler options
     pub write_tokens_to: Option<String>,
     pub write_ast_to: Option<String>,
     pub(crate) arch: TargetArch,
+}
 impl Parser {
     pub fn new() -> Result<Self, String> {
         let mut parser = Parser{
             heap: Heap::new(),
             string_pool: StringPool::new(),
             modules: Vec::new(),
             symbol_table: SymbolTable::new(),
             type_table: TypeTable::new(),
             global_module_index: 0,
             pass_tokenizer: PassTokenizer::new(),
             pass_symbols: PassSymbols::new(),
             pass_import: PassImport::new(),
             pass_definitions: PassDefinitions::new(),
             pass_validation: PassValidationLinking::new(),
             pass_typing: PassTyping::new(),
             pass_rewriting: PassRewriting::new(),
             pass_stack_size: PassStackSize::new(),
-            write_tokens_to: Some("tokens.txt".to_string()),
+            write_tokens_to: None,
             write_ast_to: None,
             arch: TargetArch::new(),
         };
         parser.symbol_table.insert_scope(None, SymbolScope::Global);
         // Insert builtin types
         // TODO: At some point use correct values for size/alignment
         parser.arch.void_type_id    = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::Void], false, 0, 1);
         parser.arch.message_type_id = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::Message], false, 24, 8);
         parser.arch.bool_type_id    = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::Bool], false, 1, 1);
         parser.arch.uint8_type_id   = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::UInt8], false, 1, 1);
         parser.arch.uint16_type_id  = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::UInt16], false, 2, 2);
         parser.arch.uint32_type_id  = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::UInt32], false, 4, 4);
         parser.arch.uint64_type_id  = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::UInt64], false, 8, 8);
         parser.arch.sint8_type_id   = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::SInt8], false, 1, 1);
         parser.arch.sint16_type_id  = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::SInt16], false, 2, 2);
         parser.arch.sint32_type_id  = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::SInt32], false, 4, 4);
         parser.arch.sint64_type_id  = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::SInt64], false, 8, 8);
         parser.arch.char_type_id    = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::Character], false, 4, 4);
         parser.arch.string_type_id  = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::String], false, 24, 8);
         parser.arch.array_type_id   = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::Array, ConcreteTypePart::Void], true, 24, 8);
         parser.arch.slice_type_id   = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::Slice, ConcreteTypePart::Void], true, 16, 4);
         parser.arch.input_type_id   = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::Input, ConcreteTypePart::Void], true, 8, 8);
         parser.arch.output_type_id  = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::Output, ConcreteTypePart::Void], true, 8, 8);
         parser.arch.pointer_type_id = insert_builtin_type(&mut parser.type_table, vec![ConcreteTypePart::Pointer, ConcreteTypePart::Void], true, 8, 8);
         // Parse standard library
         parser.feed_standard_library()?;
         return Ok(parser)
+    }
     /// Feeds a new InputSource to the parser, which will tokenize it and store
     /// it internally for later parsing (when all modules are present). Returns
     /// the index of the new module.
     pub fn feed(&mut self, mut source: InputSource) -> Result<usize, ParseError> {
         let mut token_buffer = TokenBuffer::new();
         self.pass_tokenizer.tokenize(&mut source, &mut token_buffer)?;
         let module = Module{
             source,
             tokens: token_buffer,
             root_id: RootId::new_invalid(),
             name: None,
             version: None,
             phase: ModuleCompilationPhase::Tokenized,
         };
         let module_index = self.modules.len();
         self.modules.push(module);
         return Ok(module_index);
         return self.feed_internal(source, false);
+    }
     pub fn parse(&mut self) -> Result<(), ParseError> {
         let mut pass_ctx = PassCtx{
             heap: &mut self.heap,
             symbols: &mut self.symbol_table,
             pool: &mut self.string_pool,
             arch: &self.arch,
         };
         if let Some(filename) = &self.write_tokens_to {
             let mut writer = TokenWriter::new();
             let mut file = std::fs::File::create(std::path::Path::new(filename)).unwrap();
             writer.write(&mut file, &self.modules);
+        }
         // Advance all modules to the phase where all symbols are scanned
         for module_idx in 0..self.modules.len() {
             self.pass_symbols.parse(&mut self.modules, module_idx, &mut pass_ctx)?;
+        }
         // With all symbols scanned, perform further compilation until we can
         // add all base types to the type table.
         for module_idx in 0..self.modules.len() {
@@ @@ -332,58 +317,77 @@ impl Parser { @@
             return Err(format!("std lib root directory '{}' does not exist", base_path));
+        }
         // Try to load all standard library files. We might need a more unified
         // way to do this in the future (i.e. a "std" package, containing all
         // of the modules)
         let mut file_path = PathBuf::new();
         let mut first_file = true;
         for file in FILES {
             file_path.push(path);
             file_path.push(file);
             let source = fs::read(file_path.as_path());
             if let Err(err) = source {
                 return Err(format!(
                     "failed to read std lib file '{}' in root directory '{}', because: {}",
                     file, base_path, err
                 ));
+            }
             let source = source.unwrap();
             let input_source = InputSource::new(file.to_string(), source);
-            let module_index = self.feed(input_source);
+            let module_index = self.feed_internal(input_source, true);
             if let Err(err) = module_index {
                 // A bit of a hack, but shouldn't really happen anyway: the
                 // compiler should ship with a decent standard library (at some
                 // point)
                 return Err(format!("{}", err));
+            }
             let module_index = module_index.unwrap();
             if first_file {
                 self.global_module_index = module_index;
                 first_file = false;
+            }
+        }
         return Ok(())
+    }
     fn feed_internal(&mut self, mut source: InputSource, is_compiler_file: bool) -> Result<usize, ParseError> {
         let mut token_buffer = TokenBuffer::new();
         self.pass_tokenizer.tokenize(&mut source, &mut token_buffer)?;
         let module = Module{
             source,
             tokens: token_buffer,
             is_compiler_file,
             root_id: RootId::new_invalid(),
             name: None,
             version: None,
             phase: ModuleCompilationPhase::Tokenized,
         };
         let module_index = self.modules.len();
         self.modules.push(module);
         return Ok(module_index);
+    }
+}
 fn insert_builtin_type(type_table: &mut TypeTable, parts: Vec<ConcreteTypePart>, has_poly_var: bool, size: usize, alignment: usize) -> TypeId {
     const POLY_VARS: [PolymorphicVariable; 1] = [PolymorphicVariable{
         identifier: Identifier::new_empty(InputSpan::new()),
         is_in_use: false,
     }];
     let concrete_type = ConcreteType{ parts };
     let poly_var = if has_poly_var {
         POLY_VARS.as_slice()
     } else {
         &[]
     };
     return type_table.add_builtin_data_type(concrete_type, poly_var, size, alignment);
+}
@@ \ No newline at end of file @@

src/protocol/parser/pass_definitions.rs

➞

Show inline comments

@@ @@ -22,140 +22,137 @@ pub(crate) struct PassDefinitions { @@
     variables: ScopedBuffer<VariableId>,
     expressions: ScopedBuffer<ExpressionId>,
     statements: ScopedBuffer<StatementId>,
     parser_types: ScopedBuffer<ParserType>,
+}
 impl PassDefinitions {
     pub(crate) fn new() -> Self {
         Self{
             cur_definition: DefinitionId::new_invalid(),
             type_parser: ParserTypeParser::new(),
             buffer: String::with_capacity(128),
             struct_fields: ScopedBuffer::with_capacity(128),
             enum_variants: ScopedBuffer::with_capacity(128),
             union_variants: ScopedBuffer::with_capacity(128),
             variables: ScopedBuffer::with_capacity(128),
             expressions: ScopedBuffer::with_capacity(128),
             statements: ScopedBuffer::with_capacity(128),
             parser_types: ScopedBuffer::with_capacity(128),
+        }
+    }
     pub(crate) fn parse(&mut self, modules: &mut [Module], module_idx: usize, ctx: &mut PassCtx) -> Result<(), ParseError> {
         let module = &modules[module_idx];
         let module_range = &module.tokens.ranges[0];
         debug_assert_eq!(module.phase, ModuleCompilationPhase::ImportsResolved);
         debug_assert_eq!(module_range.range_kind, TokenRangeKind::Module);
         // Although we only need to parse the definitions, we want to go through
         // code ranges as well such that we can throw errors if we get
         // unexpected tokens at the module level of the source.
         let mut range_idx = module_range.first_child_idx;
         loop {
             let range_idx_usize = range_idx as usize;
             let cur_range = &module.tokens.ranges[range_idx_usize];
             match cur_range.range_kind {
                 TokenRangeKind::Module => unreachable!(), // should not be reachable
                 TokenRangeKind::Pragma | TokenRangeKind::Import => {
                     // Already fully parsed, fall through and go to next range
                 },
                 TokenRangeKind::Definition | TokenRangeKind::Code => {
                     // Visit range even if it is a "code" range to provide
                     // proper error messages.
                     self.visit_range(modules, module_idx, ctx, range_idx_usize)?;
                 },
         // We iterate through the entire document. If we find a marker that has
         // been handled then we skip over it. It is important that we properly
         // parse all other tokens in the document to ensure that we throw the
         // correct kind of errors.
         let num_tokens = module.tokens.tokens.len() as u32;
         let num_markers = module.tokens.markers.len();
         let mut marker_index = 0;
         let mut first_token_index = 0;
         while first_token_index < num_tokens {
             // Seek ahead to the next marker that was already handled.
             let mut last_token_index = num_tokens;
             let mut new_first_token_index = num_tokens;
             while marker_index < num_markers {
                 let marker = &module.tokens.markers[marker_index];
                 marker_index += 1;
                 if marker.handled {
                     last_token_index = marker.first_token;
                     new_first_token_index = marker.last_token;
                     break;
+                }
+            }
             if cur_range.next_sibling_idx == NO_SIBLING {
                 break;
             } else {
                 range_idx = cur_range.next_sibling_idx;
+            }
             self.visit_token_range(modules, module_idx, ctx, first_token_index, last_token_index)?;
             first_token_index = new_first_token_index;
+        }
         modules[module_idx].phase = ModuleCompilationPhase::DefinitionsParsed;
         Ok(())
+    }
     fn visit_range(
         &mut self, modules: &[Module], module_idx: usize, ctx: &mut PassCtx, range_idx: usize
     fn visit_token_range(
         &mut self, modules: &[Module], module_idx: usize, ctx: &mut PassCtx,
         token_range_begin: u32, token_range_end: u32,
     ) -> Result<(), ParseError> {
         let module = &modules[module_idx];
         let cur_range = &module.tokens.ranges[range_idx];
         debug_assert!(cur_range.range_kind == TokenRangeKind::Definition || cur_range.range_kind == TokenRangeKind::Code);
         // Detect which definition we're parsing
-        let mut iter = module.tokens.iter_range(cur_range.start, cur_range.end);
+        let mut iter = module.tokens.iter_range(token_range_begin, Some(token_range_end));
         loop {
             let next = iter.next();
             if next.is_none() {
                 return Ok(())
+            }
             // Token was not None, so peek_ident returns None if not an ident
             let ident = peek_ident(&module.source, &mut iter);
             match ident {
                 Some(KW_STRUCT) => self.visit_struct_definition(module, &mut iter, ctx)?,
                 Some(KW_ENUM) => self.visit_enum_definition(module, &mut iter, ctx)?,
                 Some(KW_UNION) => self.visit_union_definition(module, &mut iter, ctx)?,
                 Some(KW_FUNCTION) => self.visit_function_definition(module, &mut iter, ctx)?,
                 Some(KW_PRIMITIVE) | Some(KW_COMPOSITE) => self.visit_component_definition(module, &mut iter, ctx)?,
                 _ => return Err(ParseError::new_error_str_at_pos(
                     &module.source, iter.last_valid_pos(),
                     "unexpected symbol, expected a keyword marking the start of a definition"
                 )),
+            }
+        }
+    }
     fn visit_struct_definition(
         &mut self, module: &Module, iter: &mut TokenIter, ctx: &mut PassCtx
     ) -> Result<(), ParseError> {
         consume_exact_ident(&module.source, iter, KW_STRUCT)?;
         let (ident_text, _) = consume_ident(&module.source, iter)?;
         // Retrieve preallocated DefinitionId
         let module_scope = SymbolScope::Module(module.root_id);
         let definition_id = ctx.symbols.get_symbol_by_name_defined_in_scope(module_scope, ident_text)
             .unwrap().variant.as_definition().definition_id;
         self.cur_definition = definition_id;
         // Parse struct definition
         consume_polymorphic_vars_spilled(&module.source, iter, ctx)?;
         let mut fields_section = self.struct_fields.start_section();
         consume_comma_separated(
             TokenKind::OpenCurly, TokenKind::CloseCurly, &module.source, iter, ctx,
             |source, iter, ctx| {
                 let poly_vars = ctx.heap[definition_id].poly_vars();
                 let start_pos = iter.last_valid_pos();
                 let parser_type = self.type_parser.consume_parser_type(
                     iter, &ctx.heap, source, &ctx.symbols, poly_vars, definition_id,
                     module_scope, false, None
+                    module_scope, false, false, None
                 )?;
                 let field = consume_ident_interned(source, iter, ctx)?;
                 Ok(StructFieldDefinition{
                     span: InputSpan::from_positions(start_pos, field.span.end),
                     field, parser_type
                 })
             },
             &mut fields_section, "a struct field", "a list of struct fields", None
         )?;
         // Transfer to preallocated definition
         let struct_def = ctx.heap[definition_id].as_struct_mut();
         struct_def.fields = fields_section.into_vec();
         Ok(())
+    }
     fn visit_enum_definition(
         &mut self, module: &Module, iter: &mut TokenIter, ctx: &mut PassCtx
     ) -> Result<(), ParseError> {
         consume_exact_ident(&module.source, iter, KW_ENUM)?;
         let (ident_text, _) = consume_ident(&module.source, iter)?;
         // Retrieve preallocated DefinitionId
@@ @@ -200,177 +197,180 @@ impl PassDefinitions { @@
         // Retrieve preallocated DefinitionId
         let module_scope = SymbolScope::Module(module.root_id);
         let definition_id = ctx.symbols.get_symbol_by_name_defined_in_scope(module_scope, ident_text)
             .unwrap().variant.as_definition().definition_id;
         self.cur_definition = definition_id;
         // Parse union definition
         consume_polymorphic_vars_spilled(&module.source, iter, ctx)?;
         let mut variants_section = self.union_variants.start_section();
         consume_comma_separated(
             TokenKind::OpenCurly, TokenKind::CloseCurly, &module.source, iter, ctx,
             |source, iter, ctx| {
                 let identifier = consume_ident_interned(source, iter, ctx)?;
                 let mut close_pos = identifier.span.end;
                 let mut types_section = self.parser_types.start_section();
                 let has_embedded = maybe_consume_comma_separated(
                     TokenKind::OpenParen, TokenKind::CloseParen, source, iter, ctx,
                     |source, iter, ctx| {
                         let poly_vars = ctx.heap[definition_id].poly_vars();
                         self.type_parser.consume_parser_type(
                             iter, &ctx.heap, source, &ctx.symbols, poly_vars, definition_id,
                             module_scope, false, None
+                            module_scope, false, false, None
+                        )
                     },
                     &mut types_section, "an embedded type", Some(&mut close_pos)
                 )?;
                 let value = if has_embedded {
                     types_section.into_vec()
                 } else {
                     types_section.forget();
                     Vec::new()
                 };
                 Ok(UnionVariantDefinition{
                     span: InputSpan::from_positions(identifier.span.begin, close_pos),
                     identifier,
                     value
                 })
             },
             &mut variants_section, "a union variant", "a list of union variants", None
         )?;
         // Transfer to AST
         let union_def = ctx.heap[definition_id].as_union_mut();
         union_def.variants = variants_section.into_vec();
         Ok(())
+    }
     fn visit_function_definition(
         &mut self, module: &Module, iter: &mut TokenIter, ctx: &mut PassCtx
     ) -> Result<(), ParseError> {
         // Retrieve function name
         consume_exact_ident(&module.source, iter, KW_FUNCTION)?;
         let (ident_text, _) = consume_ident(&module.source, iter)?;
         let stringy = String::from_utf8_lossy(ident_text).to_string();
         // Retrieve preallocated DefinitionId
         let module_scope = SymbolScope::Module(module.root_id);
         let definition_id = ctx.symbols.get_symbol_by_name_defined_in_scope(module_scope, ident_text)
             .unwrap().variant.as_definition().definition_id;
         self.cur_definition = definition_id;
         let allow_compiler_types = module.is_compiler_file;
         consume_polymorphic_vars_spilled(&module.source, iter, ctx)?;
         // Parse function's argument list
         let mut parameter_section = self.variables.start_section();
         consume_parameter_list(
             &mut self.type_parser, &module.source, iter, ctx, &mut parameter_section, module_scope, definition_id
             &mut self.type_parser, &module.source, iter, ctx, &mut parameter_section,
             module_scope, definition_id, allow_compiler_types
         )?;
         let parameters = parameter_section.into_vec();
         // Consume return types
         consume_token(&module.source, iter, TokenKind::ArrowRight)?;
         let poly_vars = ctx.heap[definition_id].poly_vars();
         let parser_type = self.type_parser.consume_parser_type(
             iter, &ctx.heap, &module.source, &ctx.symbols, poly_vars, definition_id,
             module_scope, false, None
+            module_scope, false, allow_compiler_types, None
         )?;
         // Consume body
         let (body_id, source) = self.consume_procedure_body(module, iter, ctx, definition_id, ProcedureKind::Function)?;
         let scope_id = ctx.heap.alloc_scope(|this| Scope::new(this, ScopeAssociation::Definition(definition_id)));
         // Assign everything in the preallocated AST node
         let function = ctx.heap[definition_id].as_procedure_mut();
         function.source = source;
         function.return_type = Some(parser_type);
         function.parameters = parameters;
         function.scope = scope_id;
         function.body = body_id;
         Ok(())
+    }
     fn visit_component_definition(
         &mut self, module: &Module, iter: &mut TokenIter, ctx: &mut PassCtx
     ) -> Result<(), ParseError> {
         // Consume component variant and name
         let (_variant_text, _) = consume_any_ident(&module.source, iter)?;
         debug_assert!(_variant_text == KW_PRIMITIVE || _variant_text == KW_COMPOSITE);
         let (ident_text, _) = consume_ident(&module.source, iter)?;
         // Retrieve preallocated definition
         let module_scope = SymbolScope::Module(module.root_id);
         let definition_id = ctx.symbols.get_symbol_by_name_defined_in_scope(module_scope, ident_text)
             .unwrap().variant.as_definition().definition_id;
         self.cur_definition = definition_id;
         let allow_compiler_types = module.is_compiler_file;
         consume_polymorphic_vars_spilled(&module.source, iter, ctx)?;
         // Parse component's argument list
         let mut parameter_section = self.variables.start_section();
         consume_parameter_list(
             &mut self.type_parser, &module.source, iter, ctx, &mut parameter_section, module_scope, definition_id
             &mut self.type_parser, &module.source, iter, ctx, &mut parameter_section,
             module_scope, definition_id, allow_compiler_types
         )?;
         let parameters = parameter_section.into_vec();
         // Consume body
         let procedure_kind = ctx.heap[definition_id].as_procedure().kind;
         let (body_id, source) = self.consume_procedure_body(module, iter, ctx, definition_id, procedure_kind)?;
         let scope_id = ctx.heap.alloc_scope(|this| Scope::new(this, ScopeAssociation::Definition(definition_id)));
         // Assign everything in the AST node
         let component = ctx.heap[definition_id].as_procedure_mut();
         debug_assert!(component.return_type.is_none());
         component.source = source;
         component.parameters = parameters;
         component.scope = scope_id;
         component.body = body_id;
         Ok(())
+    }
     /// Consumes a procedure's body: either a user-defined procedure, which we
     /// parse as normal, or a builtin function, where we'll make sure we expect
     /// the particular builtin.
     ///
     /// We expect that the procedure's name is already stored in the
     /// preallocated AST node.
     fn consume_procedure_body(
         &mut self, module: &Module, iter: &mut TokenIter, ctx: &mut PassCtx, definition_id: DefinitionId, kind: ProcedureKind
     ) -> Result<(BlockStatementId, ProcedureSource), ParseError> {
         if iter.next() == Some(TokenKind::OpenCurly) && iter.peek() == Some(TokenKind::Pragma) {
             // Consume the placeholder "{ #builtin }" tokens
             iter.consume(); // opening curly brace
-            let (pragma, pragma_start, pragma_end) = consume_pragma(&module.source, iter)?;
+            let (pragma, pragma_span) = consume_pragma(&module.source, iter)?;
             if pragma != b"#builtin" {
                 return Err(ParseError::new_error_str_at_span(
-                    &module.source, InputSpan::from_positions(pragma_start, pragma_end),
+                    &module.source, pragma_span,
                     "expected a '#builtin' pragma, or a function body"
                 ));
+            }
             if iter.next() != Some(TokenKind::CloseCurly) {
                 // Just to keep the compiler writers in line ;)
                 panic!("compiler error: when using the #builtin pragma, wrap it in curly braces");
+            }
             iter.consume();
             // Retrieve module and procedure name
             assert!(module.name.is_some(), "compiler error: builtin procedure body in unnamed module");
             let (_, module_name) = module.name.as_ref().unwrap();
             let module_name = module_name.as_str();
             let definition = ctx.heap[definition_id].as_procedure();
             let procedure_name = definition.identifier.value.as_str();
             let source = match (module_name, procedure_name) {
                 ("std.global", "get") => ProcedureSource::FuncGet,
                 ("std.global", "put") => ProcedureSource::FuncPut,
                 ("std.global", "fires") => ProcedureSource::FuncFires,
                 ("std.global", "create") => ProcedureSource::FuncCreate,
                 ("std.global", "length") => ProcedureSource::FuncLength,
@@ @@ -842,49 +842,49 @@ impl PassDefinitions { @@
         Ok(ctx.heap.alloc_new_statement(|this| NewStatement{
             this,
             span: new_span,
             expression: call_id,
             next: StatementId::new_invalid(),
         }))
+    }
     fn consume_channel_statement(
         &mut self, module: &Module, iter: &mut TokenIter, ctx: &mut PassCtx
     ) -> Result<ChannelStatementId, ParseError> {
         // Consume channel specification
         let channel_span = consume_exact_ident(&module.source, iter, KW_STMT_CHANNEL)?;
         let (inner_port_type, end_pos) = if Some(TokenKind::OpenAngle) == iter.next() {
             // Retrieve the type of the channel, we're cheating a bit here by
             // consuming the first '<' and setting the initial angle depth to 1
             // such that our final '>' will be consumed as well.
             let angle_start_pos = iter.next_start_position();
             iter.consume();
             let definition_id = self.cur_definition;
             let poly_vars = ctx.heap[definition_id].poly_vars();
             let parser_type = self.type_parser.consume_parser_type(
                 iter, &ctx.heap, &module.source, &ctx.symbols, poly_vars,
                 definition_id, SymbolScope::Module(module.root_id),
                 true, Some(angle_start_pos)
+                true, false, Some(angle_start_pos)
             )?;
             (parser_type.elements, parser_type.full_span.end)
         } else {
             // Assume inferred
+            (
                 vec![ParserTypeElement{
                     element_span: channel_span,
                     variant: ParserTypeVariant::Inferred
                 }],
                 channel_span.end
+            )
         };
         let from_identifier = consume_ident_interned(&module.source, iter, ctx)?;
         consume_token(&module.source, iter, TokenKind::ArrowRight)?;
         let to_identifier = consume_ident_interned(&module.source, iter, ctx)?;
         consume_token(&module.source, iter, TokenKind::SemiColon)?;
         // Construct ports
         let port_type_span = InputSpan::from_positions(channel_span.begin, end_pos);
         let port_type_len = inner_port_type.len() + 1;
         let mut from_port_type = ParserType{ elements: Vec::with_capacity(port_type_len), full_span: port_type_span };
         from_port_type.elements.push(ParserTypeElement{
@@ @@ -938,49 +938,50 @@ impl PassDefinitions { @@
             relative_pos_in_parent: 0,
             in_sync: SynchronousStatementId::new_invalid(),
         });
         return Ok(stmt_id);
+    }
     /// Attempts to consume a memory statement (a statement along the lines of
     /// `type var_name = initial_expr`). Will return `Ok(None)` if it didn't
     /// seem like there was a memory statement, `Ok(Some(...))` if there was
     /// one, and `Err(...)` if its reasonable to assume that there was a memory
     /// statement, but we failed to parse it.
     fn maybe_consume_memory_statement_without_semicolon(
         &mut self, module: &Module, iter: &mut TokenIter, ctx: &mut PassCtx
     ) -> Result<Option<MemoryStatementId>, ParseError> {
         // This is a bit ugly. It would be nicer if we could somehow
         // consume the expression with a type hint if we do get a valid
         // type, but we don't get an identifier following it
         let iter_state = iter.save();
         let definition_id = self.cur_definition;
         let poly_vars = ctx.heap[definition_id].poly_vars();
         let parser_type = self.type_parser.consume_parser_type(
             iter, &ctx.heap, &module.source, &ctx.symbols, poly_vars,
             definition_id, SymbolScope::Definition(definition_id), true, None
             definition_id, SymbolScope::Definition(definition_id),
             true, false, None
         );
         if let Ok(parser_type) = parser_type {
             if Some(TokenKind::Ident) == iter.next() {
                 // Assume this is a proper memory statement
                 let identifier = consume_ident_interned(&module.source, iter, ctx)?;
                 let memory_span = InputSpan::from_positions(parser_type.full_span.begin, identifier.span.end);
                 let assign_span = consume_token(&module.source, iter, TokenKind::Equal)?;
                 let initial_expr_id = self.consume_expression(module, iter, ctx)?;
                 let initial_expr_end_pos = iter.last_valid_pos();
                 // Create the AST variable
                 let local_id = ctx.heap.alloc_variable(|this| Variable{
                     this,
                     kind: VariableKind::Local,
                     identifier: identifier.clone(),
                     parser_type,
                     relative_pos_in_parent: 0,
                     unique_id_in_scope: -1,
                 });
                 // Create the initial assignment expression
                 // Note: we set the initial variable declaration here
@@ @@ -1545,49 +1546,49 @@ impl PassDefinitions { @@
                 this, span,
                 value: Literal::Character(character),
                 parent: ExpressionParent::None,
                 type_index: -1,
             }).upcast()
         } else if next == Some(TokenKind::Ident) {
             // May be a variable, a type instantiation or a function call. If we
             // have a single identifier that we cannot find in the type table
             // then we're going to assume that we're dealing with a variable.
             let ident_span = iter.next_span();
             let ident_text = module.source.section_at_span(ident_span);
             let symbol = ctx.symbols.get_symbol_by_name(SymbolScope::Module(module.root_id), ident_text);
             if symbol.is_some() {
                 // The first bit looked like a symbol, so we're going to follow
                 // that all the way through, assume we arrive at some kind of
                 // function call or type instantiation
                 use ParserTypeVariant as PTV;
                 let symbol_scope = SymbolScope::Definition(self.cur_definition);
                 let poly_vars = ctx.heap[self.cur_definition].poly_vars();
                 let parser_type = self.type_parser.consume_parser_type(
                     iter, &ctx.heap, &module.source, &ctx.symbols, poly_vars, self.cur_definition,
                     symbol_scope, true, None
+                    symbol_scope, true, false, None
                 )?;
                 debug_assert!(!parser_type.elements.is_empty());
                 match parser_type.elements[0].variant {
                     PTV::Definition(target_definition_id, _) => {
                         let definition = &ctx.heap[target_definition_id];
                         match definition {
                             Definition::Struct(_) => {
                                 // Struct literal
                                 let mut last_token = iter.last_valid_pos();
                                 let mut struct_fields = Vec::new();
                                 consume_comma_separated(
                                     TokenKind::OpenCurly, TokenKind::CloseCurly, &module.source, iter, ctx,
                                     |source, iter, ctx| {
                                         let identifier = consume_ident_interned(source, iter, ctx)?;
                                         consume_token(source, iter, TokenKind::Colon)?;
                                         let value = self.consume_expression(module, iter, ctx)?;
                                         Ok(LiteralStructField{ identifier, value, field_idx: 0 })
                                     },
                                     &mut struct_fields, "a struct field", "a list of struct fields", Some(&mut last_token)
                                 )?;
                                 ctx.heap.alloc_literal_expression(|this| LiteralExpression{
                                     this,
                                     span: InputSpan::from_positions(ident_span.begin, last_token),
@@ @@ -1710,49 +1711,49 @@ impl PassDefinitions { @@
                     let bound_to = self.consume_prefix_expression(module, iter, ctx)?;
                     consume_token(&module.source, iter, TokenKind::Equal)?;
                     let bound_from = self.consume_prefix_expression(module, iter, ctx)?;
                     let full_span = InputSpan::from_positions(
                         operator_span.begin, ctx.heap[bound_from].full_span().end,
                     );
                     ctx.heap.alloc_binding_expression(|this| BindingExpression{
                         this, operator_span, full_span, bound_to, bound_from,
                         parent: ExpressionParent::None,
                         type_index: -1,
                     }).upcast()
                 } else if ident_text == KW_CAST {
                     // Casting expression
                     iter.consume();
                     let to_type = if Some(TokenKind::OpenAngle) == iter.next() {
                         let angle_start_pos = iter.next_start_position();
                         iter.consume();
                         let definition_id = self.cur_definition;
                         let poly_vars = ctx.heap[definition_id].poly_vars();
                         self.type_parser.consume_parser_type(
                             iter, &ctx.heap, &module.source, &ctx.symbols,
                             poly_vars, definition_id, SymbolScope::Module(module.root_id),
                             true, Some(angle_start_pos)
+                            true, false, Some(angle_start_pos)
                         )?
                     } else {
                         // Automatic casting with inferred target type
                         ParserType{
                             elements: vec![ParserTypeElement{
                                 element_span: ident_span,
                                 variant: ParserTypeVariant::Inferred,
                             }],
                             full_span: ident_span
+                        }
                     };
                     consume_token(&module.source, iter, TokenKind::OpenParen)?;
                     let subject = self.consume_expression(module, iter, ctx)?;
                     let mut full_span = iter.next_span();
                     full_span.begin = to_type.full_span.begin;
                     consume_token(&module.source, iter, TokenKind::CloseParen)?;
                     ctx.heap.alloc_cast_expression(|this| CastExpression{
                         this,
                         cast_span: to_type.full_span,
                         full_span, to_type, subject,
                         parent: ExpressionParent::None,
                         type_index: -1,
@@ @@ -1846,48 +1847,48 @@ impl PassDefinitions { @@
             TokenKind::OpenParen, TokenKind::CloseParen, &module.source, iter, ctx,
             |_source, iter, ctx| self.consume_expression(module, iter, ctx),
             &mut section, "an expression", "a list of expressions", end_pos
         )?;
         Ok(section.into_vec())
+    }
+}
 /// Consumes polymorphic variables and throws them on the floor.
 fn consume_polymorphic_vars_spilled(source: &InputSource, iter: &mut TokenIter, _ctx: &mut PassCtx) -> Result<(), ParseError> {
     maybe_consume_comma_separated_spilled(
         TokenKind::OpenAngle, TokenKind::CloseAngle, source, iter, _ctx,
         |source, iter, _ctx| {
             consume_ident(source, iter)?;
             Ok(())
         }, "a polymorphic variable"
     )?;
     Ok(())
+}
 /// Consumes the parameter list to functions/components
 fn consume_parameter_list(
     parser: &mut ParserTypeParser, source: &InputSource, iter: &mut TokenIter,
     ctx: &mut PassCtx, target: &mut ScopedSection<VariableId>,
     scope: SymbolScope, definition_id: DefinitionId
+    scope: SymbolScope, definition_id: DefinitionId, allow_compiler_types: bool
 ) -> Result<(), ParseError> {
     consume_comma_separated(
         TokenKind::OpenParen, TokenKind::CloseParen, source, iter, ctx,
         |source, iter, ctx| {
             let poly_vars = ctx.heap[definition_id].poly_vars(); // Rust being rust, multiple lookups
             let parser_type = parser.consume_parser_type(
                 iter, &ctx.heap, source, &ctx.symbols, poly_vars, definition_id,
                 scope, false, None
+                scope, false, allow_compiler_types, None
             )?;
             let identifier = consume_ident_interned(source, iter, ctx)?;
             let parameter_id = ctx.heap.alloc_variable(|this| Variable{
                 this,
                 kind: VariableKind::Parameter,
                 parser_type,
                 identifier,
                 relative_pos_in_parent: 0,
                 unique_id_in_scope: -1,
             });
             Ok(parameter_id)
         },
         target, "a parameter", "a parameter list", None
+    )
+}
@@ \ No newline at end of file @@

src/protocol/parser/pass_definitions_types.rs

➞

Show inline comments

@@ @@ -38,63 +38,65 @@ enum ParseState { @@
 #[derive(Debug)]
 pub(crate) struct ParserTypeParser {
     entries: Vec<Entry>,
     depths: Vec<DepthElement>,
     parse_state: ParseState,
     first_pos: InputPosition,
     last_pos: InputPosition,
+}
 impl ParserTypeParser {
     pub(crate) fn new() -> Self {
         return Self{
             entries: Vec::with_capacity(16),
             depths: Vec::with_capacity(16),
             parse_state: ParseState::TypeMaybePolyArgs,
             first_pos: InputPosition{ line: 0, offset: 0 },
             last_pos: InputPosition{ line: 0, offset: 0 }
+        }
+    }
     pub(crate) fn consume_parser_type(
         &mut self, iter: &mut TokenIter, heap: &Heap, source: &InputSource,
         symbols: &SymbolTable, poly_vars: &[Identifier],
         wrapping_definition: DefinitionId, cur_scope: SymbolScope,
         allow_inference: bool, inside_angular_bracket: Option<InputPosition>,
         allow_inference: bool, allow_compiler_types: bool,
         inside_angular_bracket: Option<InputPosition>,
     ) -> Result<ParserType, ParseError> {
         // Prepare
         self.entries.clear();
         self.depths.clear();
         // Setup processing
         if let Some(bracket_pos) = inside_angular_bracket {
             self.push_depth(DepthKind::PolyArgs, u32::MAX, bracket_pos);
+        }
         let initial_state = match iter.next() {
             Some(TokenKind::Ident) => {
+            Some(TokenKind::Ident) | Some(TokenKind::Pragma) => {
                 let element = Self::consume_parser_type_element(
                     iter, source, heap, symbols, wrapping_definition, poly_vars, cur_scope, allow_inference
                     iter, source, heap, symbols, wrapping_definition, poly_vars, cur_scope,
                     allow_inference, allow_compiler_types
                 )?;
                 self.first_pos = element.element_span.begin;
                 self.last_pos = element.element_span.end;
                 self.entries.push(Entry{
                     element,
                     depth: self.cur_depth(),
                 });
                 // Due to the nature of the subsequent type parsing algorithm,
                 // we check the opening polymorphic argument list paren here.
                 if let Some(TokenKind::OpenAngle) = iter.next() {
                     self.consume_open_angle(iter);
                     ParseState::PolyArgStart
                 } else {
                     ParseState::TypeMaybePolyArgs
+                }
             },
             Some(TokenKind::OpenParen) => {
                 let tuple_start_pos = iter.next_start_position();
                 self.first_pos = tuple_start_pos; // last pos will be set later, this is a tuple
                 let tuple_entry_index = self.entries.len() as u32;
                 let tuple_depth = self.cur_depth();
@@ @@ -133,77 +135,80 @@ impl ParserTypeParser { @@
                         _ => return Err(ParseError::new_error_str_at_pos(
                             source, iter.last_valid_pos(),
                             "unexpected token: expected ',', '<', '>', '<<', ')' or '['"
                         )),
+                    }
                 },
                 ParseState::TypeNeverPolyArgs => {
                     // Allowed tokens: , > >> ) [
                     match next {
                         Some(TokenKind::Comma) => self.consume_comma(iter),
                         Some(TokenKind::CloseAngle) => self.consume_close_angle(source, iter)?,
                         Some(TokenKind::ShiftRight) => self.consume_double_close_angle(source, iter)?,
                         Some(TokenKind::CloseParen) => self.consume_close_paren(source, iter)?,
                         Some(TokenKind::OpenSquare) => self.consume_square_parens(source, iter)?,
                         _ => return Err(ParseError::new_error_str_at_pos(
                             source, iter.last_valid_pos(),
                             "unexpected token: expected ',', '>', '>>', ')' or '['"
                         )),
+                    }
                 },
                 ParseState::PolyArgStart => {
                     // Allowed tokens: ident (
                     match next {
                         Some(TokenKind::Ident) => self.consume_type_idents(
                             source, heap, symbols, wrapping_definition, poly_vars, cur_scope, allow_inference, iter
                             source, heap, symbols, wrapping_definition, poly_vars, cur_scope,
                             allow_inference, allow_compiler_types, iter
                         )?,
                         Some(TokenKind::OpenParen) => self.consume_open_paren(iter),
                         _ => return Err(ParseError::new_error_str_at_pos(
                             source, iter.last_valid_pos(),
                             "unexpected token: expected typename or '('"
                         )),
+                    }
                 },
                 ParseState::TupleStart => {
                     // Allowed tokens: ident ( )
                     // We'll strip the nested tuple later in this function
                     match next {
                         Some(TokenKind::Ident) => self.consume_type_idents(
                             source, heap, symbols, wrapping_definition, poly_vars, cur_scope, allow_inference, iter
                             source, heap, symbols, wrapping_definition, poly_vars, cur_scope,
                             allow_inference, allow_compiler_types, iter
                         )?,
                         Some(TokenKind::OpenParen) => self.consume_open_paren(iter),
                         Some(TokenKind::CloseParen) => self.consume_close_paren(source, iter)?,
                         _ => return Err(ParseError::new_error_str_at_pos(
                             source, iter.last_valid_pos(),
                             "unexpected token: expected typename or ')'"
                         )),
+                    }
                 },
                 ParseState::ParsedComma => {
                     // Allowed tokens: ident ( > >> )
                     match next {
                         Some(TokenKind::Ident) => self.consume_type_idents(
                             source, heap, symbols, wrapping_definition, poly_vars, cur_scope, allow_inference, iter
                             source, heap, symbols, wrapping_definition, poly_vars, cur_scope,
                             allow_inference, allow_compiler_types, iter
                         )?,
                         Some(TokenKind::OpenParen) => self.consume_open_paren(iter),
                         Some(TokenKind::CloseAngle) => self.consume_close_angle(source, iter)?,
                         Some(TokenKind::ShiftRight) => self.consume_double_close_angle(source, iter)?,
                         Some(TokenKind::CloseParen) => self.consume_close_paren(source, iter)?,
                         _ => return Err(ParseError::new_error_str_at_pos(
                             source, iter.last_valid_pos(),
                             "unexpected token: expected typename, '(', '>', '>>' or ')'"
                         ))
+                    }
+                }
+            }
+        }
         // If here then we have found the correct number of closing braces.
         // However we might still have any number of array postfixed
         if inside_angular_bracket.is_none() {
             while Some(TokenKind::OpenSquare) == iter.next() {
                 self.consume_square_parens(source, iter)?;
+            }
+        }
         // Type should be completed. But we still need to check the polymorphic
         // arguments and strip tuples with just one embedded type.
@@ @@ -267,52 +272,54 @@ impl ParserTypeParser { @@
         // Convert the results from parsing into the `ParserType`
         let mut elements = Vec::with_capacity(self.entries.len());
         debug_assert!(!self.entries.is_empty());
         for entry in self.entries.drain(..) {
             if ParserTypeVariant::Tuple(1) == entry.element.variant {
                 // We strip these ones
             } else {
                 elements.push(entry.element);
+            }
+        }
         return Ok(ParserType{
             elements,
             full_span: InputSpan::from_positions(self.first_pos, self.last_pos),
         });
+    }
     // --- Parsing Utilities
     #[inline]
     fn consume_type_idents(
         &mut self, source: &InputSource, heap: &Heap, symbols: &SymbolTable,
         wrapping_definition: DefinitionId, poly_vars: &[Identifier],
         cur_scope: SymbolScope, allow_inference: bool, iter: &mut TokenIter
         cur_scope: SymbolScope, allow_inference: bool, allow_compiler_types: bool,
         iter: &mut TokenIter
     ) -> Result<(), ParseError> {
         let element = Self::consume_parser_type_element(
             iter, source, heap, symbols, wrapping_definition, poly_vars, cur_scope, allow_inference
             iter, source, heap, symbols, wrapping_definition, poly_vars, cur_scope,
             allow_inference, allow_compiler_types
         )?;
         let depth = self.cur_depth();
         self.last_pos = element.element_span.end;
         self.entries.push(Entry{ element, depth });
         self.parse_state = ParseState::TypeMaybePolyArgs;
         return Ok(());
+    }
     #[inline]
     fn consume_open_angle(&mut self, iter: &mut TokenIter) {
         // Note: open angular bracket is only consumed when we just parsed an
         //  ident-based type. So the last element of the `entries` array is the
         //  one that this angular bracket starts the polymorphic arguments for.
         let angle_start_pos = iter.next_start_position();
         let entry_index = (self.entries.len() - 1) as u32;
         self.push_depth(DepthKind::PolyArgs, entry_index, angle_start_pos);
         self.parse_state = ParseState::PolyArgStart;
         iter.consume();
+    }
     #[inline]
     fn consume_close_angle(&mut self, source: &InputSource, iter: &mut TokenIter) -> Result<(), ParseError> {
@@ @@ -407,53 +414,77 @@ impl ParserTypeParser { @@
         });
         // Need to increment the depth of the child types
         self.entries[insert_at + 1].depth += 1; // element we applied the array type to
         if num_embedded != 0 {
             for index in insert_at + 2..self.entries.len() {
                 let element = &mut self.entries[index];
                 if element.depth >= insert_depth + 1 {
                     element.depth += 1;
                 } else {
                     break;
+                }
+            }
+        }
         return Ok(())
+    }
     /// Consumes a namespaced identifier that should resolve to some kind of
     /// type. There may be commas or polymorphic arguments remaining after this
     /// function has finished.
     fn consume_parser_type_element(
         iter: &mut TokenIter, source: &InputSource, heap: &Heap, symbols: &SymbolTable,
         wrapping_definition: DefinitionId, poly_vars: &[Identifier],
         mut scope: SymbolScope, allow_inference: bool,
+        mut scope: SymbolScope, allow_inference: bool, allow_compiler_types: bool,
     ) -> Result<ParserTypeElement, ParseError> {
         use ParserTypeVariant as PTV;
         let (mut type_text, mut type_span) = consume_any_ident(source, iter)?;
         // Early check for special builtin types available to the compiler
         if iter.next() == Some(TokenKind::Pragma) {
             let (type_text, pragma_span) = consume_pragma(source, iter)?;
             let variant = match type_text {
                 PRAGMA_TYPE_VOID => Some(PTV::Void),
                 PRAGMA_TYPE_PORTLIKE => Some(PTV::InputOrOutput),
                 PRAGMA_TYPE_INTEGERLIKE => Some(PTV::IntegerLike),
                 PRAGMA_TYPE_ARRAYLIKE => Some(PTV::ArrayLike),
                 _ => None,
             };
             if !allow_compiler_types || variant.is_none() {
                 return Err(ParseError::new_error_str_at_span(
                     source, pragma_span, "unexpected pragma in type"
                 ));
+            }
             return Ok(ParserTypeElement{
                 variant: variant.unwrap(),
                 element_span: pragma_span,
             });
+        }
         // No special type, parse as normal
         let (mut type_text, mut type_span) = consume_any_ident(source, iter)?;
         let variant = match type_text {
             KW_TYPE_MESSAGE => PTV::Message,
             KW_TYPE_BOOL => PTV::Bool,
             KW_TYPE_UINT8 => PTV::UInt8,
             KW_TYPE_UINT16 => PTV::UInt16,
             KW_TYPE_UINT32 => PTV::UInt32,
             KW_TYPE_UINT64 => PTV::UInt64,
             KW_TYPE_SINT8 => PTV::SInt8,
             KW_TYPE_SINT16 => PTV::SInt16,
             KW_TYPE_SINT32 => PTV::SInt32,
             KW_TYPE_SINT64 => PTV::SInt64,
             KW_TYPE_IN_PORT => PTV::Input,
             KW_TYPE_OUT_PORT => PTV::Output,
             KW_TYPE_CHAR => PTV::Character,
             KW_TYPE_STRING => PTV::String,
             KW_TYPE_INFERRED => {
                 if !allow_inference {
                     return Err(ParseError::new_error_str_at_span(
                         source, type_span, "type inference is not allowed here"
                     ));
+                }
                 PTV::Inferred
             },

src/protocol/parser/pass_imports.rs

➞

Show inline comments

@@ @@ -4,86 +4,80 @@ use super::{Module, ModuleCompilationPhase, PassCtx}; @@
 use super::tokens::*;
 use super::token_parsing::*;
 use crate::protocol::input_source::{InputSource as InputSource, InputSpan, ParseError};
 use crate::collections::*;
 /// Parses all the imports in the module tokens. Is applied after the
 /// definitions and name of modules are resolved. Hence we should be able to
 /// resolve all symbols to their appropriate module/definition.
 pub(crate) struct PassImport {
     imports: Vec<ImportId>,
     found_symbols: Vec<(AliasedSymbol, SymbolDefinition)>,
     scoped_symbols: Vec<Symbol>,
+}
 impl PassImport {
     pub(crate) fn new() -> Self {
         Self{
             imports: Vec::with_capacity(32),
             found_symbols: Vec::with_capacity(32),
             scoped_symbols: Vec::with_capacity(32),
+        }
+    }
     pub(crate) fn parse(&mut self, modules: &mut [Module], module_idx: usize, ctx: &mut PassCtx) -> Result<(), ParseError> {
         let module = &modules[module_idx];
         let module_range = &module.tokens.ranges[0];
         debug_assert!(modules.iter().all(|m| m.phase >= ModuleCompilationPhase::SymbolsScanned));
         debug_assert_eq!(module.phase, ModuleCompilationPhase::SymbolsScanned);
         debug_assert_eq!(module_range.range_kind, TokenRangeKind::Module);
         let mut range_idx = module_range.first_child_idx;
         loop {
             let range_idx_usize = range_idx as usize;
             let cur_range = &module.tokens.ranges[range_idx_usize];
         let module_root_id = module.root_id;
         let num_markers = module.tokens.markers.len();
             if cur_range.range_kind == TokenRangeKind::Import {
                 self.visit_import_range(modules, module_idx, ctx, range_idx_usize)?;
+            }
             if cur_range.next_sibling_idx == NO_SIBLING {
                 break;
             } else {
                 range_idx = cur_range.next_sibling_idx;
         for marker_index in 0..num_markers {
             let marker = &modules[module_idx].tokens.markers[marker_index];
             match marker.kind {
                 TokenMarkerKind::Import => {
                     self.visit_import_marker(modules, module_idx, ctx, marker_index)?;
                 },
                 TokenMarkerKind::Definition | TokenMarkerKind::Pragma => {},
+            }
+        }
-        let root = &mut ctx.heap[module.root_id];
+        let root = &mut ctx.heap[module_root_id];
         root.imports.extend(self.imports.drain(..));
         let module = &mut modules[module_idx];
         module.phase = ModuleCompilationPhase::ImportsResolved;
         Ok(())
+    }
     pub(crate) fn visit_import_range(
         &mut self, modules: &[Module], module_idx: usize, ctx: &mut PassCtx, range_idx: usize
     pub(crate) fn visit_import_marker(
         &mut self, modules: &mut [Module], module_idx: usize, ctx: &mut PassCtx, marker_index: usize
     ) -> Result<(), ParseError> {
         let module = &modules[module_idx];
         let import_range = &module.tokens.ranges[range_idx];
         debug_assert_eq!(import_range.range_kind, TokenRangeKind::Import);
         let marker = &module.tokens.markers[marker_index];
-        let mut iter = module.tokens.iter_range(import_range.start, import_range.end);
+        let mut iter = module.tokens.iter_range(marker.first_token, None);
         // Consume "import"
         let (_import_ident, import_span) =
             consume_any_ident(&module.source, &mut iter)?;
         debug_assert_eq!(_import_ident, KW_IMPORT);
         // Consume module name
         let (module_name, module_name_span) = consume_domain_ident(&module.source, &mut iter)?;
         let target_root_id = ctx.symbols.get_module_by_name(module_name);
         if target_root_id.is_none() {
             return Err(ParseError::new_error_at_span(
                 &module.source, module_name_span,
                 format!("could not resolve module '{}'", String::from_utf8_lossy(module_name))
             ));
+        }
         let module_name = ctx.pool.intern(module_name);
         let module_identifier = Identifier{ span: module_name_span, value: module_name };
         let target_root_id = target_root_id.unwrap();
         // Check for subsequent characters (alias, multiple imported symbols)
         let next = iter.next();
         let import_id;
         if has_ident(&module.source, &mut iter, b"as") {
@@ @@ -294,27 +288,33 @@ impl PassImport { @@
             import_id = ctx.heap.alloc_import(|this| Import::Module(ImportModule{
                 this,
                 span: InputSpan::from_positions(import_span.begin, module_identifier.span.end),
                 module: module_identifier,
                 alias: alias_identifier,
                 module_id: target_root_id,
             }));
             if let Err((new_symbol, old_symbol)) = ctx.symbols.insert_symbol(SymbolScope::Module(module.root_id), Symbol{
                 name: alias,
                 variant: SymbolVariant::Module(SymbolModule{
                     root_id: target_root_id,
                     introduced_at: import_id
                 })
             }) {
                 return Err(construct_symbol_conflict_error(modules, module_idx, ctx, &new_symbol, &old_symbol));
+            }
+        }
         // By now the `import_id` is set, just need to make sure that the import
         // properly ends with a semicolon
         consume_token(&module.source, &mut iter, TokenKind::SemiColon)?;
         self.imports.push(import_id);
         // Update the marker
         let marker_last_token = iter.token_index();
         let marker = &mut modules[module_idx].tokens.markers[marker_index];
         marker.last_token = marker_last_token;
         marker.handled = true;
         Ok(())
+    }
+}

src/protocol/parser/pass_rewriting.rs

➞

Show inline comments

@@ @@ -28,48 +28,52 @@ impl PassRewriting { @@
+}
 impl Visitor for PassRewriting {
     fn visit_module(&mut self, ctx: &mut Ctx) -> VisitorResult {
         let module = ctx.module();
         debug_assert_eq!(module.phase, ModuleCompilationPhase::Typed);
         let root_id = module.root_id;
         let root = &ctx.heap[root_id];
         let definition_section = self.definition_buffer.start_section_initialized(&root.definitions);
         for definition_index in 0..definition_section.len() {
             let definition_id = definition_section[definition_index];
             self.visit_definition(ctx, definition_id)?;
+        }
         definition_section.forget();
         ctx.module_mut().phase = ModuleCompilationPhase::Rewritten;
         return Ok(())
+    }
     // --- Visiting procedures
     fn visit_procedure_definition(&mut self, ctx: &mut Ctx, id: ProcedureDefinitionId) -> VisitorResult {
         let definition = &ctx.heap[id];
         if definition.source.is_builtin() {
             return Ok(());
+        }
         let body_id = definition.body;
         self.current_scope = definition.scope;
         self.current_procedure_id = id;
         return self.visit_block_stmt(ctx, body_id);
+    }
     // --- Visiting statements (that are not the select statement)
     fn visit_block_stmt(&mut self, ctx: &mut Ctx, id: BlockStatementId) -> VisitorResult {
         let block_stmt = &ctx.heap[id];
         let stmt_section = self.statement_buffer.start_section_initialized(&block_stmt.statements);
         self.current_scope = block_stmt.scope;
         for stmt_idx in 0..stmt_section.len() {
             self.visit_stmt(ctx, stmt_section[stmt_idx])?;
+        }
         stmt_section.forget();
         return Ok(())
+    }
     fn visit_labeled_stmt(&mut self, ctx: &mut Ctx, id: LabeledStatementId) -> VisitorResult {
         let labeled_stmt = &ctx.heap[id];
         let body_id = labeled_stmt.body;

src/protocol/parser/pass_symbols.rs

➞

Show inline comments

@@ @@ -24,252 +24,248 @@ impl PassSymbols { @@
         Self{
             symbols: Vec::with_capacity(128),
             pragmas: Vec::with_capacity(8),
             imports: Vec::with_capacity(32),
             definitions: Vec::with_capacity(128),
             buffer: String::with_capacity(128),
             has_pragma_version: false,
             has_pragma_module: false,
+        }
+    }
     fn reset(&mut self) {
         self.symbols.clear();
         self.pragmas.clear();
         self.imports.clear();
         self.definitions.clear();
         self.has_pragma_version = false;
         self.has_pragma_module = false;
+    }
     pub(crate) fn parse(&mut self, modules: &mut [Module], module_idx: usize, ctx: &mut PassCtx) -> Result<(), ParseError> {
         self.reset();
         let module = &mut modules[module_idx];
-        let module_range = &module.tokens.ranges[0];
+        let module_is_compiler_file = module.is_compiler_file;
         debug_assert_eq!(module.phase, ModuleCompilationPhase::Tokenized);
         debug_assert_eq!(module_range.range_kind, TokenRangeKind::Module);
         debug_assert!(module.root_id.is_invalid()); // not set yet,
         debug_assert!(module.root_id.is_invalid()); // not set yet
         // Preallocate root in the heap
         let root_id = ctx.heap.alloc_protocol_description(|this| {
             Root{
                 this,
                 pragmas: Vec::new(),
                 imports: Vec::new(),
                 definitions: Vec::new(),
+            }
         });
         module.root_id = root_id;
         // Retrieve first range index, then make immutable borrow
         let mut range_idx = module_range.first_child_idx;
         // Visit token ranges to detect definitions and pragmas
         loop {
         // Use pragma token markers to detects symbol definitions and pragmas
         let num_markers = module.tokens.markers.len();
         for marker_index in 0..num_markers {
             let module = &modules[module_idx];
             let range_idx_usize = range_idx as usize;
             let cur_range = &module.tokens.ranges[range_idx_usize];
             let next_sibling_idx = cur_range.next_sibling_idx;
             let range_kind = cur_range.range_kind;
             let marker = &module.tokens.markers[marker_index];
             // Parse if it is a definition or a pragma
             if range_kind == TokenRangeKind::Definition {
                 self.visit_definition_range(modules, module_idx, ctx, range_idx_usize)?;
             } else if range_kind == TokenRangeKind::Pragma {
                 self.visit_pragma_range(modules, module_idx, ctx, range_idx_usize)?;
+            }
             if next_sibling_idx == NO_SIBLING {
                 break;
             } else {
                 range_idx = next_sibling_idx;
             match marker.kind {
                 TokenMarkerKind::Pragma => {
                     self.visit_pragma_marker(modules, module_idx, ctx, marker_index)?;
                 },
                 TokenMarkerKind::Definition => {
                     self.visit_definition_marker(modules, module_idx, ctx, marker_index)?;
+                }
                 TokenMarkerKind::Import => {}, // we don't care yet
+            }
+        }
         // Add the module's symbol scope and the symbols we just parsed
         let module_scope = SymbolScope::Module(root_id);
         ctx.symbols.insert_scope(Some(SymbolScope::Global), module_scope);
         for symbol in self.symbols.drain(..) {
             ctx.symbols.insert_scope(Some(module_scope), SymbolScope::Definition(symbol.variant.as_definition().definition_id));
             if let Err((new_symbol, old_symbol)) = ctx.symbols.insert_symbol(module_scope, symbol) {
                 return Err(construct_symbol_conflict_error(modules, module_idx, ctx, &new_symbol, &old_symbol))
+            }
+        }
         if module_is_compiler_file {
             debug_assert!(self.symbols.is_empty());
             ctx.symbols.get_all_symbols_defined_in_scope(module_scope, &mut self.symbols);
             for symbol in self.symbols.drain(..) {
                 ctx.symbols.insert_symbol_in_global_scope(symbol);
+            }
+        }
         // Modify the preallocated root
         let root = &mut ctx.heap[root_id];
         root.pragmas.extend(self.pragmas.drain(..));
         root.definitions.extend(self.definitions.drain(..));
         // Modify module
         let module = &mut modules[module_idx];
         module.phase = ModuleCompilationPhase::SymbolsScanned;
         Ok(())
+    }
-    fn visit_pragma_range(&mut self, modules: &mut [Module], module_idx: usize, ctx: &mut PassCtx, range_idx: usize) -> Result<(), ParseError> {
+    fn visit_pragma_marker(&mut self, modules: &mut [Module], module_idx: usize, ctx: &mut PassCtx, marker_index: usize) -> Result<(), ParseError> {
         let module = &mut modules[module_idx];
         let range = &module.tokens.ranges[range_idx];
         let mut iter = module.tokens.iter_range(range.start, module.tokens.tokens.len() as u32);
         let marker = &module.tokens.markers[marker_index];
         let mut iter = module.tokens.iter_range(marker.first_token, None);
         // Consume pragma name
-        let (pragma_section, pragma_start, _) = consume_pragma(&module.source, &mut iter)?;
+        let (pragma_section, mut pragma_span) = consume_pragma(&module.source, &mut iter)?;
         // Consume pragma values
         if pragma_section == b"#module" {
             // Check if name is defined twice within the same file
             if self.has_pragma_module {
-                return Err(ParseError::new_error_str_at_pos(&module.source, pragma_start, "module name is defined twice"));
+                return Err(ParseError::new_error_str_at_span(&module.source, pragma_span, "module name is defined twice"));
+            }
             // Consume the domain-name
+            // Consume the domain-name, then record end of pragma
             let (module_name, module_span) = consume_domain_ident(&module.source, &mut iter)?;
             // TODO: Fix with newer token range parsing
             module.tokens.ranges[range_idx as usize].end = iter.token_index();
             // if iter.next().is_some() {
             //     return Err(ParseError::new_error_str_at_pos(&module.source, iter.last_valid_pos(), "expected end of #module pragma after module name"));
             // }
             let marker_last_token = iter.token_index();
             // Add to heap and symbol table
-            let pragma_span = InputSpan::from_positions(pragma_start, module_span.end);
+            pragma_span.end = module_span.end;
             let module_name = ctx.pool.intern(module_name);
             let pragma_id = ctx.heap.alloc_pragma(|this| Pragma::Module(PragmaModule{
                 this,
                 span: pragma_span,
                 value: Identifier{ span: module_span, value: module_name.clone() },
             }));
             self.pragmas.push(pragma_id);
             if let Err(other_module_root_id) = ctx.symbols.insert_module(module_name.clone(), module.root_id) {
                 // Naming conflict
                 let this_module = &modules[module_idx];
                 let other_module = seek_module(modules, other_module_root_id).unwrap();
                 let other_module_pragma_id = other_module.name.as_ref().map(|v| (*v).0).unwrap();
                 let other_pragma = ctx.heap[other_module_pragma_id].as_module();
                 return Err(ParseError::new_error_str_at_span(
                     &this_module.source, pragma_span, "conflict in module name"
                 ).with_info_str_at_span(
                     &other_module.source, other_pragma.span, "other module is defined here"
                 ));
+            }
             let marker = &mut module.tokens.markers[marker_index];
             marker.last_token = marker_last_token;
             marker.handled = true;
             module.name = Some((pragma_id, module_name));
             self.has_pragma_module = true;
         } else if pragma_section == b"#version" {
             // Check if version is defined twice within the same file
             if self.has_pragma_version {
-                return Err(ParseError::new_error_str_at_pos(&module.source, pragma_start, "module version is defined twice"));
+                return Err(ParseError::new_error_str_at_span(&module.source, pragma_span, "module version is defined twice"));
+            }
             // Consume the version pragma
             let (version, version_span) = consume_integer_literal(&module.source, &mut iter, &mut self.buffer)?;
             let marker_last_token = iter.token_index();
             pragma_span.end = version_span.end;
             let pragma_id = ctx.heap.alloc_pragma(|this| Pragma::Version(PragmaVersion{
                 this,
-                span: InputSpan::from_positions(pragma_start, version_span.end),
+                span: pragma_span,
                 version,
             }));
             self.pragmas.push(pragma_id);
             let marker = &mut module.tokens.markers[marker_index];
             marker.last_token = marker_last_token;
             marker.handled = true;
             module.version = Some((pragma_id, version as i64));
             self.has_pragma_version = true;
         } else {
             // Custom pragma, maybe we support this in the future, but for now
             // we don't.
             return Err(ParseError::new_error_str_at_pos(&module.source, pragma_start, "illegal pragma name"));
+        }
         } // else: custom pragma used for something else, will be handled later (or rejected with an error)
         Ok(())
+    }
-    fn visit_definition_range(&mut self, modules: &[Module], module_idx: usize, ctx: &mut PassCtx, range_idx: usize) -> Result<(), ParseError> {
+    fn visit_definition_marker(&mut self, modules: &[Module], module_idx: usize, ctx: &mut PassCtx, marker_index: usize) -> Result<(), ParseError> {
         let module = &modules[module_idx];
         let range = &module.tokens.ranges[range_idx];
         let definition_span = InputSpan::from_positions(
             module.tokens.start_pos(range),
             module.tokens.end_pos(range)
         );
         let mut iter = module.tokens.iter_range(range.start, range.end);
         let marker = &module.tokens.markers[marker_index];
         let mut iter = module.tokens.iter_range(marker.first_token, None);
         // First ident must be type of symbol
         let (kw_text, _) = consume_any_ident(&module.source, &mut iter).unwrap();
         // Retrieve identifier of definition
         let identifier = consume_ident_interned(&module.source, &mut iter, ctx)?;
         println!("DEBUG: Parsing {} --- {}", String::from_utf8_lossy(kw_text).to_string(), identifier.value.as_str());
         let mut poly_vars = Vec::new();
         maybe_consume_comma_separated(
             TokenKind::OpenAngle, TokenKind::CloseAngle, &module.source, &mut iter, ctx,
             |source, iter, ctx| consume_ident_interned(source, iter, ctx),
             &mut poly_vars, "a polymorphic variable", None
         )?;
         let ident_text = identifier.value.clone(); // because we need it later
         let ident_span = identifier.span.clone();
         // Reserve space in AST for definition and add it to the symbol table
         let definition_class;
         let ast_definition_id;
         match kw_text {
             KW_STRUCT => {
                 let struct_def_id = ctx.heap.alloc_struct_definition(|this| {
-                    StructDefinition::new_empty(this, module.root_id, definition_span, identifier, poly_vars)
                     StructDefinition::new_empty(this, module.root_id, identifier, poly_vars)
                 });
                 definition_class = DefinitionClass::Struct;
                 ast_definition_id = struct_def_id.upcast();
             },
             KW_ENUM => {
                 let enum_def_id = ctx.heap.alloc_enum_definition(|this| {
-                    EnumDefinition::new_empty(this, module.root_id, definition_span, identifier, poly_vars)
                     EnumDefinition::new_empty(this, module.root_id, identifier, poly_vars)
                 });
                 definition_class = DefinitionClass::Enum;
                 ast_definition_id = enum_def_id.upcast();
             },
             KW_UNION => {
                 let union_def_id = ctx.heap.alloc_union_definition(|this| {
-                    UnionDefinition::new_empty(this, module.root_id, definition_span, identifier, poly_vars)
                     UnionDefinition::new_empty(this, module.root_id, identifier, poly_vars)
                 });
                 definition_class = DefinitionClass::Union;
                 ast_definition_id = union_def_id.upcast()
             },
             KW_FUNCTION => {
                 let proc_def_id = ctx.heap.alloc_procedure_definition(|this| {
-                    ProcedureDefinition::new_empty(this, module.root_id, definition_span, ProcedureKind::Function, identifier, poly_vars)
                     ProcedureDefinition::new_empty(this, module.root_id, ProcedureKind::Function, identifier, poly_vars)
                 });
                 definition_class = DefinitionClass::Function;
                 ast_definition_id = proc_def_id.upcast();
             },
             KW_PRIMITIVE | KW_COMPOSITE => {
                 let procedure_kind = if kw_text == KW_PRIMITIVE {
                     ProcedureKind::Primitive
                 } else {
                     ProcedureKind::Composite
                 };
                 let proc_def_id = ctx.heap.alloc_procedure_definition(|this| {
-                    ProcedureDefinition::new_empty(this, module.root_id, definition_span, procedure_kind, identifier, poly_vars)
                     ProcedureDefinition::new_empty(this, module.root_id, procedure_kind, identifier, poly_vars)
                 });
                 definition_class = DefinitionClass::Component;
                 ast_definition_id = proc_def_id.upcast();
             },
             _ => unreachable!("encountered keyword '{}' in definition range", String::from_utf8_lossy(kw_text)),
+        }
         let symbol = Symbol{
             name: ident_text,
             variant: SymbolVariant::Definition(SymbolDefinition{
                 defined_in_module: module.root_id,
                 defined_in_scope: SymbolScope::Module(module.root_id),
                 definition_span,
                 identifier_span: ident_span,
                 imported_at: None,
                 class: definition_class,
                 definition_id: ast_definition_id,
             }),
         };
         self.symbols.push(symbol);
         self.definitions.push(ast_definition_id);
         Ok(())
+    }
+}
@@ \ No newline at end of file @@

src/protocol/parser/pass_tokenizer.rs

➞

Show inline comments

@@ @@ -45,58 +45,61 @@ impl PassTokenizer { @@
         self.stack_idx = 0;
         target.ranges.push(TokenRange{
             parent_idx: NO_RELATION,
             range_kind: TokenRangeKind::Module,
             curly_depth: 0,
             start: 0,
             end: 0,
             num_child_ranges: 0,
             first_child_idx: NO_RELATION,
             last_child_idx: NO_RELATION,
             next_sibling_idx: NO_RELATION,
         });
         // Main tokenization loop
         while let Some(c) = source.next() {
             let token_index = target.tokens.len() as u32;
             if is_char_literal_start(c) {
                 self.consume_char_literal(source, target)?;
             } else if is_string_literal_start(c) {
                 self.consume_string_literal(source, target)?;
             } else if is_identifier_start(c) {
                 let ident = self.consume_identifier(source, target)?;
                 if demarks_definition(ident) {
                 if demarks_symbol(ident) {
                     self.emit_marker(target, TokenMarkerKind::Definition, token_index);
                     self.push_range(target, TokenRangeKind::Definition, token_index);
                 } else if demarks_import(ident) {
                     self.emit_marker(target, TokenMarkerKind::Import, token_index);
                     self.push_range(target, TokenRangeKind::Import, token_index);
+                }
             } else if is_integer_literal_start(c) {
                 self.consume_number(source, target)?;
             } else if is_pragma_start_or_pound(c) {
                 let was_pragma = self.consume_pragma_or_pound(c, source, target)?;
                 if was_pragma {
                     self.emit_marker(target, TokenMarkerKind::Pragma, token_index);
                     self.push_range(target, TokenRangeKind::Pragma, token_index);
+                }
             } else if self.is_line_comment_start(c, source) {
                 self.consume_line_comment(source, target)?;
             } else if self.is_block_comment_start(c, source) {
                 self.consume_block_comment(source, target)?;
             } else if is_whitespace(c) {
                 self.consume_whitespace(source);
                 let range = &target.ranges[self.stack_idx];
                 if range.range_kind == TokenRangeKind::Pragma {
                     self.pop_range(target, target.tokens.len() as u32);
+                }
             } else {
                 let was_punctuation = self.maybe_parse_punctuation(c, source, target)?;
                 if let Some((token, token_pos)) = was_punctuation {
                     if token == TokenKind::OpenCurly {
                         self.curly_stack.push(token_pos);
                     } else if token == TokenKind::CloseCurly {
                         // Check if this marks the end of a range we're
                         // currently processing
                         if self.curly_stack.is_empty() {
                             return Err(ParseError::new_error_str_at_pos(
                                 source, token_pos, "unmatched closing curly brace '}'"
                             ));
@@ @@ -634,48 +637,64 @@ impl PassTokenizer { @@
         parent_range.last_child_idx = new_range_idx;
         parent_range.end = code_end_idx;
         parent_range.num_child_ranges += 1;
         let curly_depth = self.curly_stack.len() as u32;
         target.ranges.push(TokenRange{
             parent_idx,
             range_kind: TokenRangeKind::Code,
             curly_depth,
             start: code_start_idx,
             end: code_end_idx,
             num_child_ranges: 0,
             first_child_idx: NO_RELATION,
             last_child_idx: NO_RELATION,
             next_sibling_idx,
         });
         // Fix up the sibling indices
         if sibling_idx != NO_RELATION {
             let sibling_range = &mut target.ranges[sibling_idx as usize];
             sibling_range.next_sibling_idx = new_range_idx;
+        }
+    }
     fn emit_marker(&mut self, target: &mut TokenBuffer, kind: TokenMarkerKind, first_token: u32) {
         debug_assert!(
             target.markers
                 .last().map(|v| v.first_token < first_token)
                 .unwrap_or(true)
         );
         target.markers.push(TokenMarker{
             kind,
             curly_depth: self.curly_stack.len() as u32,
             first_token,
             last_token: u32::MAX,
             handled: false,
         });
+    }
     fn push_range(&mut self, target: &mut TokenBuffer, range_kind: TokenRangeKind, first_token_idx: u32) {
         let new_range_idx = target.ranges.len() as i32;
         let parent_idx = self.stack_idx as i32;
         let parent_range = &mut target.ranges[self.stack_idx];
         if parent_range.first_child_idx == NO_RELATION {
             parent_range.first_child_idx = new_range_idx;
+        }
         let last_registered_idx = parent_range.end;
         if last_registered_idx != first_token_idx {
             self.add_code_range(target, parent_idx, last_registered_idx, first_token_idx, new_range_idx + 1);
+        }
         // Push the new range
         self.stack_idx = target.ranges.len();
         let curly_depth = self.curly_stack.len() as u32;
         target.ranges.push(TokenRange{
             parent_idx,
             range_kind,
             curly_depth,
             start: first_token_idx,
             end: first_token_idx, // modified when popped
             num_child_ranges: 0,
@@ @@ -706,49 +725,49 @@ impl PassTokenizer { @@
         parent.end = end_token_idx;
         parent.num_child_ranges += 1;
         // Fix up the sibling (if it exists)
         if prev_sibling_idx != NO_RELATION {
             let sibling = &mut target.ranges[prev_sibling_idx as usize];
             sibling.next_sibling_idx = popped_idx;
+        }
+    }
     fn check_ascii(&self, source: &InputSource) -> Result<(), ParseError> {
         match source.next() {
             Some(c) if !c.is_ascii() => {
                 Err(ParseError::new_error_str_at_pos(source, source.pos(), "encountered a non-ASCII character"))
             },
             _else => {
                 Ok(())
             },
+        }
+    }
+}
 // Helpers for characters
-fn demarks_definition(ident: &[u8]) -> bool {
+fn demarks_symbol(ident: &[u8]) -> bool {
     return
         ident == KW_STRUCT ||
             ident == KW_ENUM ||
             ident == KW_UNION ||
             ident == KW_FUNCTION ||
             ident == KW_PRIMITIVE ||
             ident == KW_COMPOSITE
+}
 fn demarks_import(ident: &[u8]) -> bool {
     return ident == KW_IMPORT;
+}
 fn is_whitespace(c: u8) -> bool {
     c.is_ascii_whitespace()
+}
 fn is_char_literal_start(c: u8) -> bool {
     return c == b'\'';
+}
 fn is_string_literal_start(c: u8) -> bool {
     return c == b'"';
+}

src/protocol/parser/pass_typing.rs

➞

Show inline comments

@@ @@ -1202,71 +1202,76 @@ impl PassTyping { @@
 // -----------------------------------------------------------------------------
 // PassTyping - Visitor-like implementation
 // -----------------------------------------------------------------------------
 type VisitorResult = Result<(), ParseError>;
 type VisitExprResult = Result<InferNodeIndex, ParseError>;
 impl PassTyping {
     // Definitions
     fn visit_definition(&mut self, ctx: &mut Ctx, id: DefinitionId) -> VisitorResult {
         return visitor_recursive_definition_impl!(self, &ctx.heap[id], ctx);
+    }
     fn visit_enum_definition(&mut self, _: &mut Ctx, _: EnumDefinitionId) -> VisitorResult { return Ok(()) }
     fn visit_struct_definition(&mut self, _: &mut Ctx, _: StructDefinitionId) -> VisitorResult { return Ok(()) }
     fn visit_union_definition(&mut self, _: &mut Ctx, _: UnionDefinitionId) -> VisitorResult { return Ok(()) }
     fn visit_procedure_definition(&mut self, ctx: &mut Ctx, id: ProcedureDefinitionId) -> VisitorResult {
         let procedure_def = &ctx.heap[id];
         self.procedure_id = id;
         self.procedure_kind = procedure_def.kind;
         let body_id = procedure_def.body;
         let procedure_is_builtin = procedure_def.source.is_builtin();
         debug_log!("{}", "-".repeat(50));
         debug_log!("Visiting procedure: '{}' (id: {}, kind: {:?})", procedure_def.identifier.value.as_str(), id.0.index, procedure_def.kind);
         debug_log!("{}", "-".repeat(50));
         // Visit parameters
         let section = self.var_buffer.start_section_initialized(procedure_def.parameters.as_slice());
         for param_id in section.iter_copied() {
             let param = &ctx.heap[param_id];
             let var_type = self.determine_inference_type_from_parser_type_elements(&param.parser_type.elements, true);
             debug_assert!(var_type.is_done, "expected function arguments to be concrete types");
             self.var_data.push(VarData{
                 var_id: param_id,
                 var_type,
                 used_at: Vec::new(),
                 linked_var: None
             })
+        }
         section.forget();
         // Visit all of the expressions within the body
         self.parent_index = None;
         return self.visit_block_stmt(ctx, body_id);
         if !procedure_is_builtin {
             return self.visit_block_stmt(ctx, body_id);
         } else {
             return Ok(());
+        }
+    }
     // Statements
     fn visit_stmt(&mut self, ctx: &mut Ctx, id: StatementId) -> VisitorResult {
         return visitor_recursive_statement_impl!(self, &ctx.heap[id], ctx, Ok(()));
+    }
     fn visit_block_stmt(&mut self, ctx: &mut Ctx, id: BlockStatementId) -> VisitorResult {
         // Transfer statements for traversal
         let block = &ctx.heap[id];
         let section = self.stmt_buffer.start_section_initialized(block.statements.as_slice());
         for stmt_id in section.iter_copied() {
             self.visit_stmt(ctx, stmt_id)?;
+        }
         section.forget();
         Ok(())
+    }
     fn visit_local_stmt(&mut self, ctx: &mut Ctx, id: LocalStatementId) -> VisitorResult {
         return visitor_recursive_local_impl!(self, &ctx.heap[id], ctx);
+    }

src/protocol/parser/pass_validation_linking.rs

➞

Show inline comments

@@ @@ -179,57 +179,61 @@ impl Visitor for PassValidationLinking { @@
+        }
         section.forget();
         ctx.module_mut().phase = ModuleCompilationPhase::ValidatedAndLinked;
         Ok(())
+    }
     //--------------------------------------------------------------------------
     // Definition visitors
     //--------------------------------------------------------------------------
     fn visit_procedure_definition(&mut self, ctx: &mut Ctx, id: ProcedureDefinitionId) -> VisitorResult {
         self.reset_state();
         let definition = &ctx.heap[id];
         self.proc_id = id;
         self.proc_kind = definition.kind;
         self.expr_parent = ExpressionParent::None;
         // Visit parameters
         let scope_id = definition.scope;
         let old_scope = self.push_scope(ctx, true, scope_id);
         let definition = &ctx.heap[id];
         let body_id = definition.body;
         let definition_is_builtin = definition.source.is_builtin();
         let section = self.variable_buffer.start_section_initialized(&definition.parameters);
         for variable_idx in 0..section.len() {
             let variable_id = section[variable_idx];
             self.checked_at_single_scope_add_local(ctx, self.cur_scope, -1, variable_id)?;
+        }
         section.forget();
         // Visit statements in function body
         self.visit_block_stmt(ctx, body_id)?;
         // Visit statements in function body, if present at all
         if !definition_is_builtin {
             self.visit_block_stmt(ctx, body_id)?;
+        }
         self.pop_scope(old_scope);
         self.resolve_pending_control_flow_targets(ctx)?;
         Ok(())
+    }
     //--------------------------------------------------------------------------
     // Statement visitors
     //--------------------------------------------------------------------------
     fn visit_block_stmt(&mut self, ctx: &mut Ctx, id: BlockStatementId) -> VisitorResult {
         // Get end of block
         let block_stmt = &ctx.heap[id];
         let end_block_id = block_stmt.end_block;
         let scope_id = block_stmt.scope;
         // Traverse statements in block
         let statement_section = self.statement_buffer.start_section_initialized(&block_stmt.statements);
         let old_scope = self.push_scope(ctx, false, scope_id);
         assign_and_replace_next_stmt!(self, ctx, id.upcast());
         for stmt_idx in 0..statement_section.len() {
             self.relative_pos_in_parent = stmt_idx as i32;

src/protocol/parser/symbol_table.rs

➞

Show inline comments

@@ @@ -64,49 +64,48 @@ impl ScopedSymbols { @@
     fn get_symbol<'a>(&'a self, name: &StringRef) -> Option<&'a Symbol> {
         for symbol in self.symbols.iter() {
             if symbol.name == *name {
                 return Some(symbol);
+            }
+        }
         None
+    }
+}
 #[derive(Debug, Clone)]
 pub struct SymbolModule {
     pub root_id: RootId,
     pub introduced_at: ImportId,
+}
 #[derive(Debug, Clone)]
 pub struct SymbolDefinition {
     // Definition location (not necessarily the place where the symbol
     // is introduced, as it may be imported). Builtin symbols will have invalid
     // spans and module IDs
     pub defined_in_module: RootId,
     pub defined_in_scope: SymbolScope,
     pub definition_span: InputSpan, // full span of definition
     pub identifier_span: InputSpan, // span of just the identifier
     // Location where the symbol is introduced in its scope
     pub imported_at: Option<ImportId>,
     // Definition in the heap, with a utility enum to determine its
     // class if the ID is not needed.
     pub class: DefinitionClass,
     pub definition_id: DefinitionId,
+}
 impl SymbolDefinition {
     /// Clones the entire data structure, but replaces the `imported_at` field
     /// with the supplied `ImportId`.
     pub(crate) fn into_imported(mut self, imported_at: ImportId) -> Self {
         self.imported_at = Some(imported_at);
         self
+    }
+}
 #[derive(Debug, Clone)]
 pub enum SymbolVariant {
     Module(SymbolModule),
     Definition(SymbolDefinition),
+}
@@ @@ -210,48 +209,56 @@ impl SymbolTable { @@
     // Note: we do not return a reference because Rust doesn't like it.
     pub(crate) fn insert_symbol(&mut self, in_scope: SymbolScope, symbol: Symbol) -> Result<(), (Symbol, Symbol)> {
         debug_assert!(self.scope_lookup.contains_key(&in_scope), "inserting symbol {}, but scope {:?} does not exist", symbol.name.as_str(), in_scope);
         let mut seek_scope = in_scope;
         loop {
             let scoped_symbols = self.scope_lookup.get(&seek_scope).unwrap();
             for existing_symbol in scoped_symbols.symbols.iter() {
                 if symbol.name == existing_symbol.name {
                     return Err((symbol, existing_symbol.clone()))
+                }
+            }
             match scoped_symbols.parent_scope {
                 Some(parent_scope) => { seek_scope = parent_scope; },
                 None => { break; }
+            }
+        }
         // If here, then there is no collision
         let scoped_symbols = self.scope_lookup.get_mut(&in_scope).unwrap();
         scoped_symbols.symbols.push(symbol);
         Ok(())
+    }
     /// Insert a symbol in the global scope. Naturally there will be a
     /// collision (as the symbol originates from a module), so we do *not* check
     /// for this.
     pub(crate) fn insert_symbol_in_global_scope(&mut self, symbol: Symbol) {
         let scoped_symbols = self.scope_lookup.get_mut(&SymbolScope::Global).unwrap();
         scoped_symbols.symbols.push(symbol);
+    }
     /// Retrieves a symbol by name by searching in a particular scope and that scope's parents. The
     /// returned symbol may both be imported as defined within any of the searched scopes.
     pub(crate) fn get_symbol_by_name(
         &self, mut in_scope: SymbolScope, name: &[u8]
     ) -> Option<&Symbol> {
         let string_ref = StringRef::new(name);
         loop {
             let scope = self.scope_lookup.get(&in_scope);
             if scope.is_none() {
                 return None;
+            }
             let scope = scope.unwrap();
             if let Some(symbol) = scope.get_symbol(&string_ref) {
                 return Some(symbol);
             } else {
                 // Could not find symbol in current scope, seek in the parent scope if it exists
                 match &scope.parent_scope {
                     Some(parent_scope) => { in_scope = *parent_scope; },
                     None => return None,
+                }
+            }
+        }
+    }

src/protocol/parser/token_parsing.rs

➞

Show inline comments

@@ @@ -65,48 +65,57 @@ pub(crate) const KW_TYPE_UINT64_STR: &'static str = "u64"; @@
 pub(crate) const KW_TYPE_SINT8_STR:    &'static str = "s8";
 pub(crate) const KW_TYPE_SINT16_STR:   &'static str = "s16";
 pub(crate) const KW_TYPE_SINT32_STR:   &'static str = "s32";
 pub(crate) const KW_TYPE_SINT64_STR:   &'static str = "s64";
 pub(crate) const KW_TYPE_CHAR_STR:     &'static str = "char";
 pub(crate) const KW_TYPE_STRING_STR:   &'static str = "string";
 pub(crate) const KW_TYPE_INFERRED_STR: &'static str = "auto";
 pub(crate) const KW_TYPE_IN_PORT:  &'static [u8] = KW_TYPE_IN_PORT_STR.as_bytes();
 pub(crate) const KW_TYPE_OUT_PORT: &'static [u8] = KW_TYPE_OUT_PORT_STR.as_bytes();
 pub(crate) const KW_TYPE_MESSAGE:  &'static [u8] = KW_TYPE_MESSAGE_STR.as_bytes();
 pub(crate) const KW_TYPE_BOOL:     &'static [u8] = KW_TYPE_BOOL_STR.as_bytes();
 pub(crate) const KW_TYPE_UINT8:    &'static [u8] = KW_TYPE_UINT8_STR.as_bytes();
 pub(crate) const KW_TYPE_UINT16:   &'static [u8] = KW_TYPE_UINT16_STR.as_bytes();
 pub(crate) const KW_TYPE_UINT32:   &'static [u8] = KW_TYPE_UINT32_STR.as_bytes();
 pub(crate) const KW_TYPE_UINT64:   &'static [u8] = KW_TYPE_UINT64_STR.as_bytes();
 pub(crate) const KW_TYPE_SINT8:    &'static [u8] = KW_TYPE_SINT8_STR.as_bytes();
 pub(crate) const KW_TYPE_SINT16:   &'static [u8] = KW_TYPE_SINT16_STR.as_bytes();
 pub(crate) const KW_TYPE_SINT32:   &'static [u8] = KW_TYPE_SINT32_STR.as_bytes();
 pub(crate) const KW_TYPE_SINT64:   &'static [u8] = KW_TYPE_SINT64_STR.as_bytes();
 pub(crate) const KW_TYPE_CHAR:     &'static [u8] = KW_TYPE_CHAR_STR.as_bytes();
 pub(crate) const KW_TYPE_STRING:   &'static [u8] = KW_TYPE_STRING_STR.as_bytes();
 pub(crate) const KW_TYPE_INFERRED: &'static [u8] = KW_TYPE_INFERRED_STR.as_bytes();
 // Builtin pragma types
 // Not usable by the programmer, but usable in the standard library. These hint
 // at the fact that we need a different system (e.g. function overloading)
 pub(crate) const PRAGMA_TYPE_VOID: &'static [u8] = b"#type_void";
 pub(crate) const PRAGMA_TYPE_PORTLIKE: &'static [u8] = b"#type_portlike";
 pub(crate) const PRAGMA_TYPE_INTEGERLIKE: &'static [u8] = b"#type_integerlike";
 pub(crate) const PRAGMA_TYPE_ARRAYLIKE: &'static [u8] = b"#type_arraylike";
 /// A special trait for when consuming comma-separated things such that we can
 /// push them onto a `Vec` and onto a `ScopedSection`. As we monomorph for
 /// very specific comma-separated cases I don't expect polymorph bloat.
 /// Also, I really don't like this solution.
 pub(crate) trait Extendable {
     type Value;
     fn push(&mut self, v: Self::Value);
+}
 impl<T> Extendable for Vec<T> {
     type Value = T;
     #[inline]
     fn push(&mut self, v: Self::Value) {
         (self as &mut Vec<T>).push(v);
+    }
+}
 impl<T: Sized> Extendable for ScopedSection<T> {
     type Value = T;
     #[inline]
     fn push(&mut self, v: Self::Value) {
@@ @@ -428,55 +437,55 @@ pub(crate) fn consume_string_literal( @@
     Ok(span)
+}
 fn parse_escaped_character(source: &InputSource, literal_span: InputSpan, v: u8) -> Result<char, ParseError> {
     let result = match v {
         b'r' => '\r',
         b'n' => '\n',
         b't' => '\t',
         b'0' => '\0',
         b'\\' => '\\',
         b'\'' => '\'',
         b'"' => '"',
         v => {
             let msg = if v.is_ascii_graphic() {
                 format!("unsupported escape character '{}'", v as char)
             } else {
                 format!("unsupported escape character with (unsigned) byte value {}", v)
             };
             return Err(ParseError::new_error_at_span(source, literal_span, msg))
         },
     };
     Ok(result)
+}
-pub(crate) fn consume_pragma<'a>(source: &'a InputSource, iter: &mut TokenIter) -> Result<(&'a [u8], InputPosition, InputPosition), ParseError> {
+pub(crate) fn consume_pragma<'a>(source: &'a InputSource, iter: &mut TokenIter) -> Result<(&'a [u8], InputSpan), ParseError> {
     if Some(TokenKind::Pragma) != iter.next() {
         return Err(ParseError::new_error_str_at_pos(source, iter.last_valid_pos(), "expected a pragma"));
+    }
-    let (pragma_start, pragma_end) = iter.next_positions();
+    let pragma_span = iter.next_span();
     iter.consume();
-    Ok((source.section_at_pos(pragma_start, pragma_end), pragma_start, pragma_end))
+    Ok((source.section_at_span(pragma_span), pragma_span))
+}
 pub(crate) fn has_ident(source: &InputSource, iter: &mut TokenIter, expected: &[u8]) -> bool {
     peek_ident(source, iter).map_or(false, |section| section == expected)
+}
 pub(crate) fn peek_ident<'a>(source: &'a InputSource, iter: &mut TokenIter) -> Option<&'a [u8]> {
     if Some(TokenKind::Ident) == iter.next() {
         let (start, end) = iter.next_positions();
         return Some(source.section_at_pos(start, end))
+    }
     None
+}
 /// Consumes any identifier and returns it together with its span. Does not
 /// check if the identifier is a reserved keyword.
 pub(crate) fn consume_any_ident<'a>(
     source: &'a InputSource, iter: &mut TokenIter
 ) -> Result<(&'a [u8], InputSpan), ParseError> {
     if Some(TokenKind::Ident) != iter.next() {
         return Err(ParseError::new_error_str_at_pos(source, iter.last_valid_pos(), "expected an identifier"));
+    }
     let (ident_start, ident_end) = iter.next_positions();

src/protocol/parser/tokens.rs

➞

Show inline comments

@@ @@ -149,107 +149,124 @@ impl TokenKind { @@
             TK::LessEquals => "<=",
             TK::ShiftRight => ">>",
             TK::GreaterEquals => ">=",
             TK::ShiftLeftEquals => "<<=",
             TK::ShiftRightEquals => ">>=",
             // Lets keep these in explicitly for now, in case we want to add more symbols
             TK::Ident | TK::Pragma | TK::Integer | TK::String | TK::Character |
             TK::LineComment | TK::BlockComment | TK::SpanEnd => unreachable!(),
+        }
+    }
+}
 /// Represents a single token at a particular position.
 pub struct Token {
     pub kind: TokenKind,
     pub pos: InputPosition,
+}
 impl Token {
     pub(crate) fn new(kind: TokenKind, pos: InputPosition) -> Self {
         Self{ kind, pos }
+    }
+}
 #[derive(Debug, Clone, Copy)]
 pub enum TokenMarkerKind {
     Pragma,
     Import,
     Definition,
+}
 /// A marker for a specific token. These are stored separately from the array of
 /// tokens. These are used for initial symbol, module name, and import
 /// discovery.
 #[derive(Debug)]
 pub struct TokenMarker {
     pub kind: TokenMarkerKind,
     pub curly_depth: u32,
     // Indices into token buffer. The first token is inclusive and set upon
     // tokenization, the last token is set at a later stage in parsing (e.g.
     // at symbol discovery we may parse some of the `Pragma` tokens and set the
     // last parsed token)
     pub first_token: u32,
     pub last_token: u32,
     pub handled: bool,
+}
 /// The kind of token ranges that are specially parsed by the tokenizer.
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub enum TokenRangeKind {
     Module,
     Pragma,
     Import,
     Definition,
     Code,
+}
 pub const NO_RELATION: i32 = -1;
 pub const NO_SIBLING: i32 = NO_RELATION;
 /// A range of tokens with a specific meaning. Such a range is part of a tree
 /// where each parent tree envelops all of its children.
 #[derive(Debug)]
 pub struct TokenRange {
     // Index of parent in `TokenBuffer.ranges`, does not have a parent if the
     // range kind is Module, in that case the parent index is -1.
     pub parent_idx: i32,
     pub range_kind: TokenRangeKind,
     pub curly_depth: u32,
     // Offsets into `TokenBuffer.ranges`: the tokens belonging to this range.
     pub start: u32,             // first token (inclusive index)
     pub end: u32,               // last token (exclusive index)
     // Child ranges
     pub num_child_ranges: u32,  // Number of subranges
     pub first_child_idx: i32,   // First subrange (or -1 if no subranges)
     pub last_child_idx: i32,    // Last subrange (or -1 if no subranges)
     pub next_sibling_idx: i32,  // Next subrange (or -1 if no next subrange)
+}
 pub struct TokenBuffer {
     pub tokens: Vec<Token>,
     pub markers: Vec<TokenMarker>,
     pub ranges: Vec<TokenRange>,
+}
 impl TokenBuffer {
     pub(crate) fn new() -> Self {
         Self{ tokens: Vec::new(), ranges: Vec::new() }
+    }
     pub(crate) fn iter_range<'a>(&'a self, inclusive_start: u32, exclusive_end: u32) -> TokenIter<'a> {
         debug_assert!(exclusive_end as usize <= self.tokens.len());
         TokenIter::new(self, inclusive_start as usize, exclusive_end as usize)
+    }
     pub(crate) fn start_pos(&self, range: &TokenRange) -> InputPosition {
         self.tokens[range.start as usize].pos
         return Self{
             tokens: Vec::new(),
             markers: Vec::new(),
             ranges: Vec::new()
         };
+    }
     pub(crate) fn end_pos(&self, range: &TokenRange) -> InputPosition {
         let last_token = &self.tokens[range.end as usize - 1];
         if last_token.kind == TokenKind::SpanEnd {
             return last_token.pos
         } else {
             debug_assert!(!last_token.kind.has_span_end());
             return last_token.pos.with_offset(last_token.kind.num_characters());
+        }
     pub(crate) fn iter_range(
         &self, inclusive_start: u32, exclusive_end: Option<u32>
     ) -> TokenIter {
         let exclusive_end = exclusive_end.unwrap_or(self.tokens.len() as u32) as usize;
         debug_assert!(exclusive_end <= self.tokens.len());
         TokenIter::new(self, inclusive_start as usize, exclusive_end)
+    }
+}
 /// Iterator over tokens within a specific `TokenRange`.
 pub(crate) struct TokenIter<'a> {
     tokens: &'a Vec<Token>,
     cur: usize,
     end: usize,
+}
 impl<'a> TokenIter<'a> {
     fn new(buffer: &'a TokenBuffer, start: usize, end: usize) -> Self {
         Self{ tokens: &buffer.tokens, cur: start, end }
+    }
     /// Returns the next token (may include comments), or `None` if at the end
     /// of the range.
     pub(crate) fn next_including_comments(&self) -> Option<TokenKind> {
         if self.cur >= self.end {
             return None;
+        }
         let token = &self.tokens[self.cur];
         Some(token.kind)

src/protocol/tests/utils.rs

➞

Show inline comments

@@ @@ -579,49 +579,50 @@ impl<'a> FunctionTester<'a> { @@
         );
         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.span.begin.offset as usize;
         let body = &self.ctx.heap[self.def.body];
         let mut outer_match_idx = body.span.begin.offset as usize;
         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.upcast(),
             &|expr| expr.operation_span().begin.offset as usize == 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()

src/protocol/token_writer.rs

➞

Show inline comments

 #![allow(dead_code)]
 use std::fmt::Write;
 use std::io::Write as IOWrite;
 use crate::protocol::input_source::{InputSource, InputSpan};
 use crate::protocol::parser::Module;
-use crate::protocol::tokens::{Token, TokenBuffer, TokenKind, TokenRange, TokenIter, TokenRangeKind};
 use crate::protocol::tokens::{Token, TokenKind, TokenRange};
 pub(crate) struct TokenWriter {
     buffer: String,
+}
 impl TokenWriter {
     pub(crate) fn new() -> Self {
         return Self{
             buffer: String::with_capacity(4096),
+        }
+    }
     pub(crate) fn write<W: IOWrite>(&mut self, w: &mut W, modules: &[Module]) {
         self.buffer.clear();
         for module in modules {
             self.write_module_tokens(module);
+        }
         w.write_all(self.buffer.as_bytes()).expect("write tokens");
+    }
     fn write_module_tokens(&mut self, module: &Module) {
         self.write_dashed_indent(0);

std/std.global.pdl

➞

Show inline comments

 #module std.global
 // Note: parsing of token ranges and pragma needs to change. For now we insert
 // spaces to work with the current system. Needs to be a system where the
 // pragmas, "func" keywords (and similar keywords) indicate initial points to
 // start parsing.
 func get<T>(in<T> input) -> T { #builtin }
 func put<T>(out<T> output, T value) -> #type_void { #builtin }
 func fires<T>(#type_portlike <T>) -> bool { #builtin }
 func create<T>(#type_integerlike length) -> T[] { #builtin }
 func length<T>(#type_arraylike <T> array) -> u32 { #builtin }
 func fires<T>(#type_portlike<T> port) -> bool { #builtin }
 func create<T>(#type_integerlike len) -> T[] { #builtin }
 func length<T>(#type_arraylike<T> array) -> u32 { #builtin }
 func assert(bool condition) -> #type_void { #builtin }
 func print(string message) -> #type_void { #builtin }
@@ \ No newline at end of file @@

0 comments (0 inline, 0 general)