Added type checked support for GOTO and LABEL in prevision of adding conditionals jumps and loops

CMakeLists.txt

@@ -13,7 +13,7 @@ add_executable(sugar
         include/molasses/lexer.h
         src/molasses/parser_primitives.cpp
         include/molasses/parser_primitives.h
-        include/molasses/generator_primitives.h include/molasses/errors.h include/molasses/parser_types.h src/molasses/generator_primitives.cpp)
+        include/molasses/generator_primitives.h include/molasses/errors.h include/molasses/parser_types.h src/molasses/generator_primitives.cpp include/generalized_parsing.h)
 
 function(add_expect_test [testname filename])
     add_test(
@@ -31,6 +31,7 @@ add_expect_test(id003 ./tests/003.exp)
 add_expect_test(id004 ./tests/004.exp)
 add_expect_test(id005 ./tests/005.exp)
 add_expect_test(id006 ./tests/006.exp)
+add_expect_test(id007 ./tests/007.exp)
 
 
 

include/generalized_parsing.h

@@ -0,0 +1,29 @@
+#pragma once
+#include "molasses/lexer.h"
+#include "molasses/parser_primitives.h"
+#include <memory>
+#include <span>
+#include <vector>
+
+struct parser {
+    /**
+     *
+     * @param ctx
+     * @param lexer_data
+     * @param current
+     * @return a span of the tokens left before reaching the target, assuming the parser is as greedy as possible
+     */
+    virtual std::span<molasses::symbol> identify(molasses::parser_context& ctx, const molasses::lexed_output& lexer_data, std::span<molasses::symbol> current) = 0;
+};
+
+struct ast_node {
+    virtual std::vector<molasses::symbol> apply(molasses::parser_context& ctx, const molasses::lexed_output& lexer_data, std::span<molasses::symbol> current) = 0;
+};
+
+struct parser_branch : public parser {
+    std::vector<std::shared_ptr<parser>> choice;
+};
+
+struct parser_sequence : public parser {
+    std::vector<std::shared_ptr<parser>> sequence;
+};

include/molasses/errors.h

@@ -69,6 +69,15 @@ namespace molasses {
               )) {}
         // TODO: Better error message
     };
+    struct orphan_goto_error : public parser_error {
+        orphan_goto_error(const symbol& sym, const std::string& label_name)
+            : parser_error(details::concatenate_builder(
+                  "A __GOTO__ has been found to be without a label\n", "\tAt ", sym.file_name, ":", sym.line, ":", sym.column,
+                  "\n", "\tExpected a label ", label_name,
+                  "\n"
+              )) {}
+        // TODO: Better error message
+    };
     struct unknown_token_error : public parser_error {
         explicit unknown_token_error(const symbol& sym)
             : parser_error(details::concatenate_builder(

include/molasses/generator_primitives.h

@@ -9,7 +9,10 @@ namespace molasses {
 	std::vector<std::string> generate_return();
 
     template<architecture_t Arch = architecture>
-	std::vector<std::string> generate_label(const std::string& target);
+    std::vector<std::string> generate_label(const std::string& target);
+
+    template<architecture_t Arch = architecture>
+    std::vector<std::string> generate_goto(const std::string& target);
 
     template<architecture_t Arch = architecture>
     std::vector<std::string> generate_push_int32(int32_t target);

include/molasses/parser_primitives.h

@@ -57,6 +57,13 @@ namespace molasses {
     template<architecture_t Arch = architecture>
     parser_context register_i32_operations(parser_context);
 	
-	bool type_check(const parser_context&, const lexed_output&, const std::vector<symbol>&, std::vector<std::string> execution_input, const std::vector<std::string>& execution_output);
+	bool type_check(
+        const parser_context& parser_state,
+        const lexed_output& lexer_state,
+        const std::vector<symbol>& consumed_stream,
+        std::vector<std::string> execution_input,
+        const std::vector<std::string>& execution_output,
+        const std::vector<std::pair<size_t, size_t>>& sub_bodies
+    );
 }
 

include/molasses/parser_types.h

@@ -98,12 +98,14 @@ namespace molasses {
         std::vector<std::string> _args;
         std::vector<std::string> _rets;
         std::vector<symbol> _body;
+        std::vector<std::pair<size_t, size_t>> _simple_sub_bodies;
 
-        procedure_operation(std::string name, std::vector<std::string> args, std::vector<std::string> rets, std::vector<symbol> body)
+        procedure_operation(std::string name, std::vector<std::string> args, std::vector<std::string> rets, std::vector<symbol> body, std::vector<std::pair<size_t, size_t>> sub_bodies)
             : _name(std::forward<std::string>(name))
             , _args(std::forward<std::vector<std::string>>(args))
             , _rets(std::forward<std::vector<std::string>>(rets))
             , _body(std::forward<std::vector<symbol>>(body))
+            , _simple_sub_bodies(std::forward<std::vector<std::pair<size_t, size_t>>>(sub_bodies))
         {}
 
         [[nodiscard]] std::string name() const final {

src/molasses/generator_primitives_x86_64_linux.cpp

@@ -536,11 +536,18 @@ namespace molasses {
     }
 
     template<>
-	std::vector<std::string> generate_label<architecture_t::x86_64_linux>(const std::string& target) {
-		return {
+    std::vector<std::string> generate_label<architecture_t::x86_64_linux>(const std::string& target) {
+        return {
             marshal(target)+":\n"
-		};
-	}
+        };
+    }
+
+    template<>
+    std::vector<std::string> generate_goto<architecture_t::x86_64_linux>(const std::string& target) {
+        return {
+            "    jmp "+marshal(target)+"\n"
+        };
+    }
 
     template<>
 	std::vector<std::string> generate_return<architecture_t::x86_64_linux>() {

src/molasses/parser_primitives.cpp

@@ -1,8 +1,9 @@
-#include <algorithm>
-#include <iostream>
 #include "molasses/parser_primitives.h"
-#include "molasses/generator_primitives.h"
 #include "molasses/errors.h"
+#include "molasses/generator_primitives.h"
+#include <algorithm>
+#include <iostream>
+#include <span>
 
 namespace molasses {
 	std::vector<std::string> operator>>(std::vector<std::string> current_stack, const operation& next_op) {
@@ -90,24 +91,40 @@ namespace molasses {
 		const lexed_output& lexer_state,
 		const std::vector<symbol>& consumed_stream,
 		std::vector<std::string> execution_input,
-		const std::vector<std::string>& execution_output
+		const std::vector<std::string>& execution_output,
+        const std::vector<std::pair<size_t, size_t>>& sub_bodies
 	) {
 		auto& type_stack = execution_input;
-		
-		for(const auto& symbol : consumed_stream) {
+
+        std::map<size_t, std::vector<std::string>> effective_snapshots;
+        size_t idx = 0;
+
+		for(auto it = consumed_stream.begin(); it != consumed_stream.end(); ++it, ++idx) {
+            const auto& symbol = *it;
 			const auto& symbol_text = lexer_state.dictionary.at(symbol);
-            if(symbol.is_string) {
+            // Skip GOTOs and LABELs
+            if(auto ahead = it; ++ahead != consumed_stream.end() and (lexer_state.dictionary.at(*ahead) == "__GOTO__" or lexer_state.dictionary.at(*ahead) == "__LABEL__")) {
+                effective_snapshots[idx] = type_stack;
+                it = ahead;
+                ++idx;
+            } else if(symbol.is_string) {
                 type_stack.emplace_back("u8 ptr");
             } else if(auto is_int32 = try_parse_int32(symbol_text); is_int32) {
                 type_stack.emplace_back("i32");
             } else if(auto is_int64 = try_parse_int64(symbol_text); is_int64) {
                 type_stack.emplace_back("i64");
             } else if(auto is_op = parser_state.lookup_operation(symbol_text); is_op) {
-				type_stack = type_stack >> *is_op;
-			}
+                type_stack = type_stack >> *is_op;
+            }
 		}
+        if(type_stack != execution_output) return false;
+        for(auto [start, end] : sub_bodies) {
+            if(effective_snapshots[start] != effective_snapshots[end]) {
+                return false;
+            }
+        }
 		
-		return type_stack == execution_output;
+		return b">true;
 	}
 
 	generate_context parse(parser_context ctx, const lexed_output& lexer_data) {
@@ -115,14 +132,19 @@ namespace molasses {
 			DO_KW = 1,
 			SEPARATOR_KW,
 			PROC_KW,
-			END_KW
+			END_KW,
+            LABEL_KW,
+            GOTO_KW
 		};
 		
 		lexed_output fake;
-		fake.dictionary[PROC_KW] = "__PROC__";
+        fake.dictionary[PROC_KW] = "__PROC__";
 		fake.dictionary[SEPARATOR_KW] = "__--__";
 		fake.dictionary[DO_KW] = "__DO__";
 		fake.dictionary[END_KW] = "__END__";
+        fake.dictionary[LABEL_KW] = "__LABEL__";
+        fake.dictionary[GOTO_KW] = "__GOTO__";
+
 		
 		auto tokens = concatenate(fake, lexer_data);
 		
@@ -197,16 +219,40 @@ namespace molasses {
 			
 			// Process body
 			std::vector<symbol> body;
+            std::vector<std::pair<size_t, size_t>> sub_bodies;
+            std::map<std::string, size_t> found_labels;
+            std::map<std::string, size_t> found_gotos;
 			while(*it != END_KW) {
-				body.emplace_back(*it);
-                last_valid = it;
-				++it;
+                if(auto ahead = it; ++ahead != tokens.symbols.end() and (*ahead == GOTO_KW or *ahead == LABEL_KW)) {
+                    if(*ahead == GOTO_KW) {
+                        found_gotos[tokens.dictionary[*it]] = body.size();
+                    } else if(*ahead == LABEL_KW) {
+                        found_labels[tokens.dictionary[*it]] = body.size();
+                        // TODO: Handle duplicate labels
+                    }
+                    body.emplace_back(*it);
+                    body.emplace_back(*ahead);
+                    last_valid = ahead;
+                    it = ++ahead;
+                } else {
+                    body.emplace_back(*it);
+                    last_valid = it;
+                    ++it;
+                }
                 check_for_unexpected_stream_end("__DO__ block needs a matching __END__", details::concatenate_builder("Parsing procedure ",tokens.dictionary.at(name_symbol),", file ",last_valid->file_name, ":", last_valid->line, ":", last_valid->column));
 			}
             last_valid = it;
 			++it;
+
+            for(auto& [dest, index] : found_gotos) {
+                if(not found_labels.contains(dest)) {
+                    throw orphan_goto_error(body[index], dest);
+                }
+
+                sub_bodies.emplace_back(std::min(index, found_labels[dest]), std::max(index, found_labels[dest]));
+            }
 			
-			return std::make_pair(it, std::make_shared<procedure_operation>(name, argument_types, return_types, body));
+			return std::make_pair(it, std::make_shared<procedure_operation>(name, argument_types, return_types, body, sub_bodies));
 #undef check_for_unexpected_stream_end
         };
 
@@ -220,7 +266,7 @@ namespace molasses {
         } while (progress != tokens.symbols.end());
 
 		for(auto& proc : parsed_procedures) {
-			if(not type_check(ctx, tokens, proc->_body, proc->_args, proc->_rets)) {
+			if(not type_check(ctx, tokens, proc->_body, proc->_args, proc->_rets, proc->_simple_sub_bodies)) {
 				throw procedure_stack_error();
 			}
 		}
@@ -239,9 +285,21 @@ namespace molasses {
 		    ops.push_back(instruction);
 	    }
 		
-        for(auto elem : _body) {
+        for(auto it = _body.begin(); it != _body.end(); ++it) {
+            auto elem = *it;
             auto token = lexer_data.dictionary.at(elem);
-            if(elem.is_string) {
+            if(auto ahead = it; ++ahead != _body.end() and (lexer_data.dictionary.at(*ahead) == "__GOTO__" or lexer_data.dictionary.at(*ahead) == "__LABEL__")) {
+                if(lexer_data.dictionary.at(*ahead) == "__GOTO__") {
+                    for(auto&& instruction : generate_goto(name() + " in " + token)) {
+                        ops.push_back(instruction);
+                    }
+                } else if(lexer_data.dictionary.at(*ahead) == "__LABEL__") {
+                    for(auto&& instruction : generate_label(name() + " in " + token)) {
+                        ops.push_back(instruction);
+                    }
+                }
+                it = ahead;
+            } else if(elem.is_string) {
                 for(auto&& instruction : generate_push_string_ptr(elem)) {
                     ops.push_back(instruction);
                 }

tests/007.exp

@@ -0,0 +1,82 @@
+#!/usr/bin/expect
+
+set SUGAR_EXECUTABLE $::env(SUGAR_EXECUTABLE)
+set BUILD_NAME 007.
+
+proc abort {reason} {
+    puts "test failed $reason"
+    exit 1
+}
+
+spawn -noecho $SUGAR_EXECUTABLE tests/007/exit-with-3.mol tests/007/library.mol lex-all merge-all
+expect {
+    error { abort "failed to parse" }
+    eof { abort "cannot find the symbol main in lexed output" }
+    main
+}
+expect eof
+lassign [wait] pid spawnid os_error_flag value
+if {$value != 0} {
+    abort "compiler crashed"
+}
+
+# -------------------------------------------------------------------
+
+spawn -noecho $SUGAR_EXECUTABLE tests/007/exit-with-3.mol tests/007/library.mol lex-all merge-all parse
+expect {
+    error { abort "failed to parse" }
+    eof { abort "cannot find the  main procedure" }
+    main
+}
+expect eof
+lassign [wait] pid spawnid os_error_flag value
+if {$value != 0} {
+    abort "compiler crashed"
+}
+
+# -------------------------------------------------------------------
+
+spawn -noecho $SUGAR_EXECUTABLE tests/007/exit-with-3.mol tests/007/library.mol lex-all merge-all parse
+expect {
+    error { abort "failed to parse" }
+    eof { abort "cannot find the exit-syscall-number procedure" }
+    exit-syscall-number
+}
+expect eof
+lassign [wait] pid spawnid os_error_flag value
+if {$value != 0} {
+    abort "compiler crashed"
+}
+
+# -------------------------------------------------------------------
+
+spawn -noecho $SUGAR_EXECUTABLE tests/007/exit-with-3.mol lex tests/007/library.mol lex merge parse /tmp/sugar.generated.$BUILD_NAME generate assemble
+expect {
+    error { abort "failed to compile" }
+    eof { abort "didn't run clang" }
+    clang
+}
+expect {
+    error { abort "failed to link" }
+    eof { abort "didn't run ld" }
+    ld
+}
+expect eof
+lassign [wait] pid spawnid os_error_flag value
+if {$value != 0} {
+    abort "compiler crashed"
+}
+
+# -------------------------------------------------------------------
+
+spawn -noecho /tmp/sugar.generated.$BUILD_NAME
+expect {
+    error { abort "failed to compile" }
+    eof { abort "didn't output" }
+    Hello
+}
+expect eof
+lassign [wait] pid spawnid os_error_flag value
+if {$value != 3} {
+    abort "executable didn't return exit code 3 but $value instead"
+}

tests/007/exit-with-3.mol

@@ -0,0 +1,7 @@
+__PROC__ main
+__--__
+__DO__
+
+11_i64 "\tHello world\n" write-out
+1 2_i32 + i32-to-i64 exit
+__END__

tests/007/library.mol

@@ -0,0 +1,39 @@
+__PROC__ write-syscall-number
+__--__
+i64
+__DO__
+1_i64
+__END__
+
+__PROC__ stdout-fd
+__--__
+i64
+__DO__
+1_i64
+__END__
+
+__PROC__ exit-syscall-number
+__--__
+i64
+__DO__
+60_i64
+__END__
+
+__PROC__ exit
+i64
+__--__
+__DO__
+"POTAT" __LABEL__
+exit-syscall-number syscall1 drop_i64
+60_i64
+"POTAT" __GOTO__
+drop_i64
+__END__
+
+__PROC__ write-out
+i64
+u8 ptr
+__--__
+__DO__
+u8-ptr_to_i64 stdout-fd write-syscall-number syscall3 drop_i64
+__END__