Added __JUMP_IF__ operation, its generator, and a a test

- Duplicate labels in a PROC now fail
- Bad token line indication when the token concludes a line

CMakeLists.txt

@@ -37,6 +37,7 @@ 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)
+add_expect_test(id008 ./tests/008.exp)
 
 
 

include/molasses/errors.h

@@ -61,6 +61,13 @@ namespace molasses {
                   "\tExpected ", expected, "\n", "\tFound ", found, "\n"
               )) {}
     };
+    struct duplicate_label_error : public parser_error {
+        duplicate_label_error(const symbol& sym, const std::string& found)
+            : parser_error(details::concatenate_builder(
+                  "Unexpected label duplication\n", "\tAt ", sym.file_name, ":", sym.line, ":", sym.column, "\n",
+                  "\tFound ", found, " duplicated label\n"
+              )) {}
+    };
     struct expecting_token_error : public parser_error {
         expecting_token_error(const std::string& expected, const std::string& context)
             : parser_error(details::concatenate_builder(

include/molasses/generator_primitives.h

@@ -14,6 +14,9 @@ namespace molasses {
     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_jump_if(const std::string& target);
+
     template<architecture_t Arch = architecture>
     std::vector<std::string> generate_push_int32(int32_t target);
 

src/molasses/generator_primitives_x86_64_linux.cpp

@@ -549,6 +549,16 @@ namespace molasses {
         };
     }
 
+    template<>
+    std::vector<std::string> generate_jump_if<architecture_t::x86_64_linux>(const std::string& target) {
+        return {
+            "    popq %rax\n",
+            "    pushq %rax\n",
+            "    cmpq $0, %rax\n"
+            "    jne "+marshal(target)+"\n"
+        };
+    }
+
     template<>
 	std::vector<std::string> generate_return<architecture_t::x86_64_linux>() {
 		return {

src/molasses/lexer.cpp

@@ -44,10 +44,6 @@ namespace molasses {
 		};
 		
 		for(auto& character : source) {
-            if(character == '\n') {
-                line++;
-                column = 0;
-            }
             column++;
             if(state == state_machine_t::string_escape) {
                 switch(character) {
@@ -79,6 +75,10 @@ namespace molasses {
                 } else {
                     builder << character;
                 }
+                if(character == '\n') {
+                    line++;
+                    column = 0;
+                }
             } else {
                 if(state == state_machine_t::string_end) {
                     std::stringstream quoted;

src/molasses/parser_primitives.cpp

@@ -107,6 +107,13 @@ namespace molasses {
                 effective_snapshots[idx] = type_stack;
                 it = ahead;
                 ++idx;
+            } else if(auto ahead = it; ++ahead != consumed_stream.end() and (lexer_state.dictionary.at(*ahead) == "__JUMP_IF__")) {
+                if(type_stack.empty()) return false;
+                if(type_stack.back() != "i64") return false;
+                //type_stack.pop_back(); // JUMP_IF does not consume its test variable
+                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) {
@@ -134,7 +141,8 @@ namespace molasses {
 			PROC_KW,
 			END_KW,
             LABEL_KW,
-            GOTO_KW
+            GOTO_KW,
+            JUMP_IF_KW
 		};
 		
 		lexed_output fake;
@@ -144,6 +152,7 @@ namespace molasses {
 		fake.dictionary[END_KW] = "__END__";
         fake.dictionary[LABEL_KW] = "__LABEL__";
         fake.dictionary[GOTO_KW] = "__GOTO__";
+        fake.dictionary[JUMP_IF_KW] = "__JUMP_IF__";
 
 		
 		auto tokens = concatenate(fake, lexer_data);
@@ -176,12 +185,11 @@ namespace molasses {
 				throw expecting_token_error(expected, context); \
 			}}while(false)
 
-            decltype(it) last_valid;
+            decltype(it) last_valid = it;
 
             if(*it != PROC_KW) {
 				throw unexpected_token_error(*it, tokens.dictionary[*it],tokens.dictionary[PROC_KW]);
 			}
-            last_valid = it;
 			++it;
             check_for_unexpected_stream_end("Procedure-Name", details::concatenate_builder("In top level, file ",last_valid->file_name, ":", last_valid->line, ":", last_valid->column));
 			
@@ -222,13 +230,19 @@ namespace molasses {
             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;
+            std::map<std::string, size_t> found_jump_ifs;
 			while(*it != END_KW) {
-                if(auto ahead = it; ++ahead != tokens.symbols.end() and (*ahead == GOTO_KW or *ahead == LABEL_KW)) {
+                if(auto ahead = it; ++ahead != tokens.symbols.end() and (*ahead == GOTO_KW or *ahead == JUMP_IF_KW or *ahead == LABEL_KW)) {
                     if(*ahead == GOTO_KW) {
                         found_gotos[tokens.dictionary[*it]] = body.size();
+                    } else if(*ahead == JUMP_IF_KW) {
+                        found_jump_ifs[tokens.dictionary[*it]] = body.size();
                     } else if(*ahead == LABEL_KW) {
-                        found_labels[tokens.dictionary[*it]] = body.size();
-                        // TODO: Handle duplicate labels
+                        auto label_value = tokens.dictionary[*it];
+                        if(found_labels.contains(label_value)) {
+                            throw duplicate_label_error(*it, label_value);
+                        }
+                        found_labels[label_value] = body.size();
                     }
                     body.emplace_back(*it);
                     body.emplace_back(*ahead);
@@ -249,6 +263,14 @@ namespace molasses {
                     throw orphan_goto_error(body[index], dest);
                 }
 
+                sub_bodies.emplace_back(std::min(index, found_labels[dest]), std::max(index, found_labels[dest]));
+            }
+
+            for(auto& [dest, index] : found_jump_ifs) {
+                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]));
             }
 			
@@ -288,7 +310,14 @@ namespace molasses {
         for(auto it = _body.begin(); it != _body.end(); ++it) {
             auto elem = *it;
             auto token = lexer_data.dictionary.at(elem);
-            if(auto ahead = it; ++ahead != _body.end() and (lexer_data.dictionary.at(*ahead) == "__GOTO__" or lexer_data.dictionary.at(*ahead) == "__LABEL__")) {
+            if(
+                auto ahead = it;
+                ++ahead != _body.end() and (
+                    lexer_data.dictionary.at(*ahead) == "__GOTO__"
+                    or lexer_data.dictionary.at(*ahead) == "__LABEL__"
+                    or lexer_data.dictionary.at(*ahead) == "__JUMP_IF__"
+                )
+            ) {
                 if(lexer_data.dictionary.at(*ahead) == "__GOTO__") {
                     for(auto&& instruction : generate_goto(name() + " in " + token)) {
                         ops.push_back(instruction);
@@ -297,6 +326,10 @@ namespace molasses {
                     for(auto&& instruction : generate_label(name() + " in " + token)) {
                         ops.push_back(instruction);
                     }
+                } else if(lexer_data.dictionary.at(*ahead) == "__JUMP_IF__") {
+                    for(auto&& instruction : generate_jump_if(name() + " in " + token)) {
+                        ops.push_back(instruction);
+                    }
                 }
                 it = ahead;
             } else if(elem.is_string) {

tests/001.exp

@@ -1,7 +1,7 @@
 #!/usr/bin/expect
 
 set SUGAR_EXECUTABLE $::env(SUGAR_EXECUTABLE)
-set BUILD_NAME 001.
+set BUILD_NAME 001
 
 proc abort {reason} {
     puts "test failed $reason"

tests/002.exp

@@ -1,7 +1,7 @@
 #!/usr/bin/expect
 
 set SUGAR_EXECUTABLE $::env(SUGAR_EXECUTABLE)
-set BUILD_NAME 002.
+set BUILD_NAME 002
 
 proc abort {reason} {
     puts "test failed $reason"

tests/003.exp

@@ -1,7 +1,7 @@
 #!/usr/bin/expect
 
 set SUGAR_EXECUTABLE $::env(SUGAR_EXECUTABLE)
-set BUILD_NAME 003.
+set BUILD_NAME 003
 
 proc abort {reason} {
     puts "test failed $reason"

tests/005.exp

@@ -1,7 +1,7 @@
 #!/usr/bin/expect
 
 set SUGAR_EXECUTABLE $::env(SUGAR_EXECUTABLE)
-set BUILD_NAME 005.
+set BUILD_NAME 005
 
 proc abort {reason} {
     puts "test failed $reason"

tests/006.exp

@@ -1,7 +1,7 @@
 #!/usr/bin/expect
 
 set SUGAR_EXECUTABLE $::env(SUGAR_EXECUTABLE)
-set BUILD_NAME 006.
+set BUILD_NAME 006
 
 proc abort {reason} {
     puts "test failed $reason"

tests/007.exp

@@ -1,7 +1,7 @@
 #!/usr/bin/expect
 
 set SUGAR_EXECUTABLE $::env(SUGAR_EXECUTABLE)
-set BUILD_NAME 007.
+set BUILD_NAME 007
 
 proc abort {reason} {
     puts "test failed $reason"

tests/007/exit-with-3.mol

@@ -2,6 +2,6 @@ __PROC__ main
 __--__
 __DO__
 
-11_i64 "\tHello world\n" write-out
+13_i64 "\tHello world\n" write-out
 1 2_i32 + i32-to-i64 exit
 __END__

tests/008.exp

@@ -0,0 +1,82 @@
+#!/usr/bin/expect
+
+set SUGAR_EXECUTABLE $::env(SUGAR_EXECUTABLE)
+set BUILD_NAME 008
+
+proc abort {reason} {
+    puts "test failed $reason"
+    exit 1
+}
+
+spawn -noecho $SUGAR_EXECUTABLE tests/008/exit-with-3.mol tests/008/library.mol lex-all merge-all
+expect {
+    error { abort "failed to parse" }
+    eof { abort "cannot find the symbol main in lexed output" }
+    main
+}
+expect eof
+lassign [wait] pid spawnid os_error_flag value
+if {$value != 0} {
+    abort "compiler crashed"
+}
+
+# -------------------------------------------------------------------
+
+spawn -noecho $SUGAR_EXECUTABLE tests/008/exit-with-3.mol tests/008/library.mol lex-all merge-all parse
+expect {
+    error { abort "failed to parse" }
+    eof { abort "cannot find the  main procedure" }
+    main
+}
+expect eof
+lassign [wait] pid spawnid os_error_flag value
+if {$value != 0} {
+    abort "compiler crashed"
+}
+
+# -------------------------------------------------------------------
+
+spawn -noecho $SUGAR_EXECUTABLE tests/008/exit-with-3.mol tests/008/library.mol lex-all merge-all parse
+expect {
+    error { abort "failed to parse" }
+    eof { abort "cannot find the exit-syscall-number procedure" }
+    exit-syscall-number
+}
+expect eof
+lassign [wait] pid spawnid os_error_flag value
+if {$value != 0} {
+    abort "compiler crashed"
+}
+
+# -------------------------------------------------------------------
+
+spawn -noecho $SUGAR_EXECUTABLE tests/008/exit-with-3.mol lex tests/008/library.mol lex merge parse /tmp/sugar.generated.$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/008/exit-with-3.mol

@@ -0,0 +1,13 @@
+__PROC__ main
+__--__
+__DO__
+
+10_i64
+"SAUSAGE" __LABEL__
+13_i64 "\tHello world\n" write-out
+
+-1_i64 +_i64
+"SAUSAGE" __JUMP_IF__ drop_i64
+
+1 2_i32 + i32-to-i64 exit
+__END__

tests/008/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__