refactor: Refactor of attribute lookahead processing (#84)

parser_library/src/analyzer.cpp

@@ -47,7 +47,7 @@ analyzer::analyzer(const std::string& text,
           *parser_,
           tracer)
 {
-    parser_->initialize(&hlasm_ctx_ref_, &lsp_proc_);
+    parser_->initialize(&hlasm_ctx_ref_, &lsp_proc_, &lib_provider, &mngr_);
     parser_->setErrorHandler(std::make_shared<error_strategy>());
     parser_->removeErrorListeners();
     parser_->addErrorListener(&listener_);

parser_library/src/context/hlasm_context.cpp

@@ -19,6 +19,7 @@
 
 #include "ebcdic_encoding.h"
 #include "expressions/conditional_assembly/terms/ca_constant.h"
+#include "expressions/conditional_assembly/terms/ca_symbol_attribute.h"
 #include "instruction.h"
 
 namespace hlasm_plugin::parser_library::context {
@@ -262,6 +263,45 @@ void hlasm_context::set_source_indices(size_t begin_index, size_t end_index, siz
     source_stack_.back().end_line = end_line;
 }
 
+std::pair<source_position, source_snapshot> hlasm_context::get_begin_snapshot(bool ignore_macros) const
+{
+    context::source_position statement_position;
+
+    bool is_in_macros = ignore_macros ? false : scope_stack_.size() > 1;
+
+    if (!is_in_macros && current_copy_stack().empty())
+    {
+        statement_position.file_offset = current_source().begin_index;
+        statement_position.file_line = current_source().current_instruction.pos.line;
+    }
+    else
+    {
+        statement_position.file_offset = current_source().end_index;
+        statement_position.file_line = current_source().end_line + 1;
+    }
+
+    context::source_snapshot snapshot = current_source().create_snapshot();
+
+    if (snapshot.copy_frames.size() && is_in_macros)
+        ++snapshot.copy_frames.back().statement_offset;
+
+    return std::make_pair(std::move(statement_position), std::move(snapshot));
+}
+
+std::pair<source_position, source_snapshot> hlasm_context::get_end_snapshot() const
+{
+    context::source_position statement_position;
+    statement_position.file_offset = current_source().end_index;
+    statement_position.file_line = current_source().end_line + 1;
+
+    context::source_snapshot snapshot = current_source().create_snapshot();
+
+    if (snapshot.copy_frames.size())
+        ++snapshot.copy_frames.back().statement_offset;
+
+    return std::make_pair(std::move(statement_position), std::move(snapshot));
+}
+
 void hlasm_context::push_statement_processing(const processing::processing_kind kind)
 {
     assert(!proc_stack_.empty());
@@ -322,6 +362,11 @@ const std::deque<code_scope>& hlasm_context::scope_stack() const { return scope_
 
 const source_context& hlasm_context::current_source() const { return source_stack_.back(); }
 
+const std::vector<copy_member_invocation>& hlasm_context::current_copy_stack() const
+{
+    return source_stack_.back().copy_stack;
+}
+
 std::vector<copy_member_invocation>& hlasm_context::current_copy_stack() { return source_stack_.back().copy_stack; }
 
 std::vector<id_index> hlasm_context::whole_copy_stack() const
@@ -366,11 +411,8 @@ var_sym_ptr hlasm_context::get_var_sym(id_index name)
 
 void hlasm_context::add_sequence_symbol(sequence_symbol_ptr seq_sym)
 {
-    if (curr_scope()->is_in_macro())
-        throw std::runtime_error("adding sequence symbols to macro definition not allowed");
-
-    if (curr_scope()->sequence_symbols.find(seq_sym->name) == curr_scope()->sequence_symbols.end())
-        curr_scope()->sequence_symbols.emplace(seq_sym->name, std::move(seq_sym));
+    auto& opencode = scope_stack_.front();
+    opencode.sequence_symbols.try_emplace(seq_sym->name, std::move(seq_sym));
 }
 
 const sequence_symbol* hlasm_context::get_sequence_symbol(id_index name) const
@@ -394,6 +436,14 @@ const sequence_symbol* hlasm_context::get_sequence_symbol(id_index name) const
         return nullptr;
 }
 
+const sequence_symbol* hlasm_context::get_opencode_sequence_symbol(id_index name) const
+{
+    if (const auto& sym = scope_stack_.front().sequence_symbols.find(name);
+        sym != scope_stack_.front().sequence_symbols.end())
+        return sym->second.get();
+    return nullptr;
+}
+
 void hlasm_context::set_branch_counter(A_t value)
 {
     curr_scope()->branch_counter = value;
@@ -541,6 +591,8 @@ C_t hlasm_context::get_type_attr(var_sym_ptr var_symbol, const std::vector<size_
     if (value.empty())
         return "O";
 
+    expressions::ca_symbol_attribute::try_extract_leading_symbol(value);
+
     auto res = expressions::ca_constant::try_self_defining_term(value);
     if (res)
         return "N";

parser_library/src/context/hlasm_context.h

@@ -95,6 +95,9 @@ class hlasm_context
     // sets current source file indices
     void set_source_indices(size_t begin_index, size_t end_index, size_t end_line);
 
+    std::pair<source_position, source_snapshot> get_begin_snapshot(bool ignore_macros) const;
+    std::pair<source_position, source_snapshot> get_end_snapshot() const;
+
     // pushes new kind of statement processing
     void push_statement_processing(const processing::processing_kind kind);
     // pushes new kind of statement processing as well as new source
@@ -107,6 +110,7 @@ class hlasm_context
     // gets macro nest
     const std::deque<code_scope>& scope_stack() const;
     // gets copy nest of current statement processing
+    const std::vector<copy_member_invocation>& current_copy_stack() const;
     std::vector<copy_member_invocation>& current_copy_stack();
     // gets names of whole copy nest
     std::vector<id_index> whole_copy_stack() const;
@@ -139,6 +143,9 @@ class hlasm_context
     // return sequence symbol in current scope
     // returns nullptr if there is none in the current scope
     const sequence_symbol* get_sequence_symbol(id_index name) const;
+    // return opencode sequence symbol
+    // returns nullptr if there is none
+    const sequence_symbol* get_opencode_sequence_symbol(id_index name) const;
 
     void set_branch_counter(A_t value);
     A_t get_branch_counter() const;
@@ -205,7 +212,7 @@ class hlasm_context
             return glob->second;
         }
 
-        set_sym_ptr val(std::make_shared<set_symbol<T>>(id, is_scalar, true));
+        auto val = std::make_shared<set_symbol<T>>(id, is_scalar, true);
 
         globals_.insert({ id, val });
         curr_scope()->variables.insert({ id, val });

parser_library/src/context/ordinary_assembly/ordinary_assembly_context.cpp

@@ -19,8 +19,7 @@
 
 #include "alignment.h"
 
-using namespace hlasm_plugin::parser_library;
-using namespace hlasm_plugin::parser_library::context;
+namespace hlasm_plugin::parser_library::context {
 
 void ordinary_assembly_context::create_private_section()
 {
@@ -55,6 +54,15 @@ bool ordinary_assembly_context::create_symbol(
     return ok;
 }
 
+void ordinary_assembly_context::add_symbol_reference(symbol sym) { symbol_refs_.try_emplace(sym.name, std::move(sym)); }
+
+const symbol* ordinary_assembly_context::get_symbol_reference(context::id_index name) const
+{
+    auto tmp = symbol_refs_.find(name);
+
+    return tmp == symbol_refs_.end() ? nullptr : &tmp->second;
+}
+
 const symbol* ordinary_assembly_context::get_symbol(id_index name) const
 {
     auto tmp = symbols_.find(name);
@@ -255,4 +263,6 @@ std::pair<address, space_ptr> ordinary_assembly_context::reserve_storage_area_sp
         return std::make_pair(ret_addr, sp);
     }
     return std::make_pair(curr_section_->current_location_counter().reserve_storage_area(length, align).first, nullptr);
-}
+}
+
+} // namespace hlasm_plugin::parser_library::context

parser_library/src/context/ordinary_assembly/ordinary_assembly_context.h

@@ -25,9 +25,7 @@
 #include "symbol_dependency_tables.h"
 
 
-namespace hlasm_plugin {
-namespace parser_library {
-namespace context {
+namespace hlasm_plugin::parser_library::context {
 
 // class holding complete information about the 'ordinary assembly' (assembler and machine instructions)
 // it contains 'sections' ordinary 'symbols' and all dependencies between them
@@ -37,6 +35,8 @@ class ordinary_assembly_context : public dependency_solver
     std::vector<std::unique_ptr<section>> sections_;
     // list of visited symbols
     std::unordered_map<id_index, symbol> symbols_;
+    // list of lookaheaded symbols
+    std::unordered_map<id_index, symbol> symbol_refs_;
 
     section* curr_section_;
 
@@ -57,6 +57,9 @@ class ordinary_assembly_context : public dependency_solver
     [[nodiscard]] bool create_symbol(
         id_index name, symbol_value value, symbol_attributes attributes, location symbol_location);
 
+    void add_symbol_reference(symbol sym);
+    const symbol* get_symbol_reference(context::id_index name) const;
+
     // gets symbol by name
     virtual const symbol* get_symbol(id_index name) const override;
     symbol* get_symbol(id_index name);
@@ -111,7 +114,6 @@ class ordinary_assembly_context : public dependency_solver
     std::pair<address, space_ptr> reserve_storage_area_space(size_t length, alignment align);
 };
 
-} // namespace context
-} // namespace parser_library
-} // namespace hlasm_plugin
+} // namespace hlasm_plugin::parser_library::context
+
 #endif

parser_library/src/context/ordinary_assembly/symbol_attributes.h

@@ -44,7 +44,8 @@ enum class symbol_origin
     SECT,
     MACH,
     EQU,
-    DAT
+    DAT,
+    UNKNOWN
 };
 
 // structure wrapping attributes of the symbol

parser_library/src/context/source_snapshot.h

@@ -20,9 +20,7 @@
 #include "id_storage.h"
 #include "range.h"
 
-namespace hlasm_plugin {
-namespace parser_library {
-namespace context {
+namespace hlasm_plugin::parser_library::context {
 
 // helper structure representing position in source file
 struct source_position
@@ -69,6 +67,24 @@ struct source_snapshot
     size_t end_line;
     std::vector<copy_frame> copy_frames;
 
+    source_snapshot()
+        : begin_index(0)
+        , end_index(0)
+        , end_line(0)
+    {}
+
+    source_snapshot(location instruction,
+        size_t begin_index,
+        size_t end_index,
+        size_t end_line,
+        std::vector<copy_frame> copy_frames)
+        : instruction(std::move(instruction))
+        , begin_index(begin_index)
+        , end_index(end_index)
+        , end_line(end_line)
+        , copy_frames(std::move(copy_frames))
+    {}
+
     bool operator==(const source_snapshot& oth) const
     {
         if (!(end_line == oth.end_line && begin_index == oth.begin_index && end_index == oth.end_index
@@ -83,7 +99,6 @@ struct source_snapshot
     }
 };
 
-} // namespace context
-} // namespace parser_library
-} // namespace hlasm_plugin
+} // namespace hlasm_plugin::parser_library::context
+
 #endif

parser_library/src/expressions/conditional_assembly/ca_expression.h

@@ -40,7 +40,7 @@ class ca_expression : public diagnosable_op_impl
     ca_expression(context::SET_t_enum expr_kind, range expr_range);
 
     // retrieves set of attributed symbols that are not yet defined
-    virtual undef_sym_set get_undefined_attributed_symbols(const context::dependency_solver& solver) const = 0;
+    virtual undef_sym_set get_undefined_attributed_symbols(const evaluation_context& eval_ctx) const = 0;
 
     // builds parts of the expression tree that could not be built during parsing
     virtual void resolve_expression_tree(context::SET_t_enum kind) = 0;

parser_library/src/expressions/conditional_assembly/ca_operator_binary.cpp

@@ -30,10 +30,10 @@ ca_binary_operator::ca_binary_operator(
     , right_expr(std::move(right_expr))
 {}
 
-undef_sym_set ca_binary_operator::get_undefined_attributed_symbols(const context::dependency_solver& solver) const
+undef_sym_set ca_binary_operator::get_undefined_attributed_symbols(const evaluation_context& eval_ctx) const
 {
-    auto tmp = left_expr->get_undefined_attributed_symbols(solver);
-    tmp.merge(right_expr->get_undefined_attributed_symbols(solver));
+    auto tmp = left_expr->get_undefined_attributed_symbols(eval_ctx);
+    tmp.merge(right_expr->get_undefined_attributed_symbols(eval_ctx));
     return tmp;
 }
 

parser_library/src/expressions/conditional_assembly/ca_operator_binary.h

@@ -29,7 +29,7 @@ class ca_binary_operator : public ca_expression
 
     ca_binary_operator(ca_expr_ptr left_expr, ca_expr_ptr right_expr, context::SET_t_enum expr_kind, range expr_range);
 
-    virtual undef_sym_set get_undefined_attributed_symbols(const context::dependency_solver& solver) const override;
+    virtual undef_sym_set get_undefined_attributed_symbols(const evaluation_context& eval_ctx) const override;
 
     virtual void resolve_expression_tree(context::SET_t_enum kind) override;
 

parser_library/src/expressions/conditional_assembly/ca_operator_unary.cpp

@@ -27,9 +27,9 @@ ca_unary_operator::ca_unary_operator(ca_expr_ptr expr, context::SET_t_enum expr_
     , expr(std::move(expr))
 {}
 
-undef_sym_set ca_unary_operator::get_undefined_attributed_symbols(const context::dependency_solver& solver) const
+undef_sym_set ca_unary_operator::get_undefined_attributed_symbols(const evaluation_context& eval_ctx) const
 {
-    return expr->get_undefined_attributed_symbols(solver);
+    return expr->get_undefined_attributed_symbols(eval_ctx);
 }
 
 void ca_unary_operator::resolve_expression_tree(context::SET_t_enum kind)

parser_library/src/expressions/conditional_assembly/ca_operator_unary.h

@@ -27,7 +27,7 @@ class ca_unary_operator : public ca_expression
 
     ca_unary_operator(ca_expr_ptr expr, context::SET_t_enum expr_kind, range expr_range);
 
-    virtual undef_sym_set get_undefined_attributed_symbols(const context::dependency_solver& solver) const override;
+    virtual undef_sym_set get_undefined_attributed_symbols(const evaluation_context& eval_ctx) const override;
 
     virtual void resolve_expression_tree(context::SET_t_enum kind) override;
 

parser_library/src/expressions/conditional_assembly/terms/ca_constant.cpp

@@ -23,10 +23,7 @@ ca_constant::ca_constant(context::A_t value, range expr_range)
     , value(value)
 {}
 
-undef_sym_set ca_constant::get_undefined_attributed_symbols(const context::dependency_solver&) const
-{
-    return undef_sym_set();
-}
+undef_sym_set ca_constant::get_undefined_attributed_symbols(const evaluation_context&) const { return undef_sym_set(); }
 
 void ca_constant::resolve_expression_tree(context::SET_t_enum kind)
 {

parser_library/src/expressions/conditional_assembly/terms/ca_constant.h

@@ -32,7 +32,7 @@ class ca_constant : public ca_expression
 
     ca_constant(context::A_t value, range expr_range);
 
-    virtual undef_sym_set get_undefined_attributed_symbols(const context::dependency_solver& solver) const override;
+    virtual undef_sym_set get_undefined_attributed_symbols(const evaluation_context& eval_ctx) const override;
 
     virtual void resolve_expression_tree(context::SET_t_enum kind) override;
 

parser_library/src/expressions/conditional_assembly/terms/ca_expr_list.cpp

@@ -28,11 +28,11 @@ ca_expr_list::ca_expr_list(std::vector<ca_expr_ptr> expr_list, range expr_range)
     , expr_list(std::move(expr_list))
 {}
 
-undef_sym_set ca_expr_list::get_undefined_attributed_symbols(const context::dependency_solver& solver) const
+undef_sym_set ca_expr_list::get_undefined_attributed_symbols(const evaluation_context& eval_ctx) const
 {
     undef_sym_set tmp;
     for (auto&& expr : expr_list)
-        tmp.merge(expr->get_undefined_attributed_symbols(solver));
+        tmp.merge(expr->get_undefined_attributed_symbols(eval_ctx));
     return tmp;
 }
 

parser_library/src/expressions/conditional_assembly/terms/ca_expr_list.h

@@ -30,7 +30,7 @@ class ca_expr_list : public ca_expression
 
     ca_expr_list(std::vector<ca_expr_ptr> expr_list, range expr_range);
 
-    virtual undef_sym_set get_undefined_attributed_symbols(const context::dependency_solver& solver) const override;
+    virtual undef_sym_set get_undefined_attributed_symbols(const evaluation_context& eval_ctx) const override;
 
     virtual void resolve_expression_tree(context::SET_t_enum kind) override;
 

parser_library/src/expressions/conditional_assembly/terms/ca_function.cpp

@@ -45,13 +45,13 @@ ca_function::ca_function(context::id_index function_name,
     , duplication_factor(std::move(duplication_factor))
 {}
 
-undef_sym_set ca_function::get_undefined_attributed_symbols(const context::dependency_solver& solver) const
+undef_sym_set ca_function::get_undefined_attributed_symbols(const evaluation_context& eval_ctx) const
 {
     undef_sym_set ret;
     for (auto&& expr : parameters)
-        ret.merge(expr->get_undefined_attributed_symbols(solver));
+        ret.merge(expr->get_undefined_attributed_symbols(eval_ctx));
     if (duplication_factor)
-        ret.merge(duplication_factor->get_undefined_attributed_symbols(solver));
+        ret.merge(duplication_factor->get_undefined_attributed_symbols(eval_ctx));
     return ret;
 }