kraken/stdlib/c_generator.krak

import io:*
import mem:*
import map:*
import hash_map:*
import stack:*
import str:*
import util:*
import tree:*
import symbol:*
import ast_nodes:*
// for error with syntax tree
import pass_common:*
import poset:*

obj c_generator (Object) {
    var id_counter: int
    var ast_name_map: hash_map<*ast_node, str>
    var used_names: hash_set<str>
    var function_type_map: map<type, str>
    var replacement_map: map<str, str>
    var longest_replacement: int
    var function_typedef_string: str
    var linker_string: str
    fun construct(): *c_generator {
        id_counter = 0
        ast_name_map.construct()
        used_names.construct()
        // to avoid using c keywords
        used_names.add(str("extern"))
        used_names.add(str("register"))
        function_type_map.construct()
        function_typedef_string.construct()
        linker_string.construct()

        replacement_map.construct()
     	replacement_map[str("+")] = str("plus")
        replacement_map[str("-")] = str("minus")
        replacement_map[str("*")] = str("star")
        replacement_map[str("/")] = str("div")
        replacement_map[str("%")] = str("mod")
        replacement_map[str("^")] = str("carat")
        replacement_map[str("&")] = str("amprsd")
        replacement_map[str("|")] = str("pipe")
        replacement_map[str("~")] = str("tilde")
        replacement_map[str("!")] = str("exlmtnpt")
        replacement_map[str(",")] = str("comma")
        replacement_map[str("=")] = str("eq")
        replacement_map[str("++")] = str("dbplus")
        replacement_map[str("--")] = str("dbminus")
        replacement_map[str("<<")] = str("dbleft")
        replacement_map[str(">>")] = str("dbright")
        replacement_map[str("::")] = str("scopeop")
        replacement_map[str(":")] = str("colon")
        replacement_map[str("==")] = str("dbq")
        replacement_map[str("!=")] = str("notequals")
        replacement_map[str("&&")] = str("doubleamprsnd")
        replacement_map[str("||")] = str("doublepipe")
        replacement_map[str("+=")] = str("plusequals")
        replacement_map[str("-=")] = str("minusequals")
        replacement_map[str("/=")] = str("divequals")
        replacement_map[str("%=")] = str("modequals")
        replacement_map[str("^=")] = str("caratequals")
        replacement_map[str("&=")] = str("amprsdequals")
        replacement_map[str("|=")] = str("pipeequals")
        replacement_map[str("*=")] = str("starequals")
        replacement_map[str("<<=")] = str("doublerightequals")
        replacement_map[str("<")] = str("lt")
        replacement_map[str(">")] = str("gt")
        replacement_map[str(">>=")] = str("doubleleftequals")
        replacement_map[str("(")] = str("openparen")
        replacement_map[str(")")] = str("closeparen")
        replacement_map[str("[")] = str("obk")
        replacement_map[str("]")] = str("cbk")
        replacement_map[str(" ")] = str("_")
        replacement_map[str(".")] = str("dot")
        replacement_map[str("->")] = str("arrow")

        longest_replacement = 0
        replacement_map.for_each(fun(key: str, value: str) {
            if (key.length() > longest_replacement)
                longest_replacement = key.length()
        })

        return this
    }
    fun copy_construct(old: *c_generator) {
        id_counter = old->id_counter
        ast_name_map.copy_construct(&old->ast_name_map)
        used_names.copy_construct(&old->used_names)
        function_type_map.copy_construct(&old->function_type_map)
        function_typedef_string.copy_construct(&old->function_typedef_string)
        replacement_map.copy_construct(&old->replacement_map)
        longest_replacement = old->longest_replacement
        linker_string.copy_construct(&old->linker_string)
    }
    fun operator=(other: ref c_generator) {
        destruct()
        copy_construct(&other)
    }
    fun destruct() {
        ast_name_map.destruct()
        used_names.destruct()
        function_type_map.destruct()
        function_typedef_string.destruct()
        replacement_map.destruct()
        linker_string.destruct()
    }
    fun get_id(): str return to_string(id_counter++);
    fun generate_function_prototype_and_header(child: *ast_node):pair<str,str> {
        var backing = child->function
        var parameter_types = str()
        var parameters = str()
        var decorated_name = str()

        if (backing.is_extern)
            decorated_name = backing.name
        else
            decorated_name = generate_function(child)

        backing.parameters.for_each(fun(parameter: *ast_node) {
            if (parameter_types != "") { parameter_types += ", "; parameters += ", ";}
            parameter_types += type_to_c(parameter->identifier.type)
            parameters += type_to_c(parameter->identifier.type) + " " + get_name(parameter)
        })
        if (backing.is_variadic) {
            parameter_types += ", ..."
            parameters += ", ..."
        }
        return make_pair(type_to_c(backing.type->return_type) + " " + decorated_name + "(" + parameter_types + ");\n",
                         type_to_c(backing.type->return_type) + " " + decorated_name + "(" + parameters + ")")
    }
    fun generate_c(name_ast_map: map<str, pair<*tree<symbol>,*ast_node>>, ast_to_syntax_in: map<*ast_node, *tree<symbol>> ): pair<str,str> {
        var prequal: str = "#include <stdbool.h>\n"
        var plain_typedefs: str = "\n/**Plain Typedefs**/\n"
        var top_level_c_passthrough: str = ""
        var variable_extern_declarations: str = ""
        var structs: str = "\n/**Type Structs**/\n"
        function_typedef_string = "\n/**Typedefs**/\n"
        var function_prototypes: str = "\n/**Function Prototypes**/\n"
        var function_definitions: str = "\n/**Function Definitions**/\n"
        var variable_declarations: str = "\n/**Variable Declarations**/\n"

        // moved out from below so that it can be used for methods as well as regular functions (and eventually lambdas...)
        var generate_function_definition = fun(child: *ast_node) {
            var backing = child->function
            var prototype_and_header = generate_function_prototype_and_header(child)
            function_prototypes += prototype_and_header.first
            if (!backing.is_extern)
                function_definitions += prototype_and_header.second
            if (backing.body_statement) {
                function_definitions += " {\n" + generate(backing.body_statement)
                function_definitions += ";\n}\n"
            }
        }

        var type_poset = poset<*ast_node>()
        // iterate through asts
        name_ast_map.for_each(fun(name: str, tree_pair: pair<*tree<symbol>,*ast_node>) {
            // iterate through children for each ast
            // do lambdas seperatly, so we can reconstitute the enclosing object if it has one
            tree_pair.second->translation_unit.lambdas.for_each(fun(child: *ast_node) {
                generate_function_definition(child)
            })
            tree_pair.second->translation_unit.children.for_each(fun(child: *ast_node) {
                match (*child) {
                    ast_node::if_comp(backing) error("if_comp not currently supported")
                    ast_node::simple_passthrough(backing) error("simple_passthrough removed")
                    ast_node::declaration_statement(backing) variable_declarations += generate_declaration_statement(child) + ";\n" // false - don't do defer
                    // shouldn't need to do anything with return, as the intrinsic should be something like link
                    ast_node::compiler_intrinsic(backing) generate_compiler_intrinsic(child)
                    ast_node::function(backing) {
                        // check for and add to parameters if a closure
                        generate_function_definition(child)
                    }
                    ast_node::template(backing) {
                        backing.instantiated.for_each(fun(node: *ast_node) {
                            match (*node) {
                                ast_node::function(backing) generate_function_definition(node)
                                ast_node::type_def(backing) {
                                    type_poset.add_vertex(node)
                                    backing.variables.for_each(fun(i: *ast_node) {
                                        var var_type = get_ast_type(i->declaration_statement.identifier)
                                        if (!var_type->indirection && var_type->type_def)
                                            type_poset.add_relationship(node, var_type->type_def)
                                    })
                                }
                            }
                        })
                    }
                    ast_node::type_def(backing) {
                        type_poset.add_vertex(child)
                        backing.variables.for_each(fun(i: *ast_node) {
                            var var_type = get_ast_type(i->declaration_statement.identifier)
                            if (!var_type->indirection && var_type->type_def)
                                type_poset.add_relationship(child, var_type->type_def)
                        })
                    }
                    ast_node::adt_def(backing) error("ADT remaining!")
                }
            })
        })
        type_poset.get_sorted().for_each(fun(vert: *ast_node) {
            var base_name = get_name(vert)
            plain_typedefs += str("typedef ")
            if (vert->type_def.is_union) {
                plain_typedefs += "union "
                structs += "union "
            } else {
                plain_typedefs += "struct "
                structs += "struct "
            }
            plain_typedefs += base_name + "_dummy " + base_name + ";\n"
            structs += base_name + "_dummy {\n"
            vert->type_def.variables.for_each(fun(variable_declaration: *ast_node) structs += generate_declaration_statement(variable_declaration) + ";\n";)
            // generate the methods (note some of these may be templates)
            vert->type_def.methods.for_each(fun(method: *ast_node) {
                if (is_template(method))
                    method->template.instantiated.for_each(fun(m: *ast_node) generate_function_definition(m);)
                else
                    generate_function_definition(method);
            })
            structs += "};\n"
        })

        return make_pair(prequal+plain_typedefs+function_typedef_string+top_level_c_passthrough+variable_extern_declarations+structs+function_prototypes+variable_declarations+function_definitions + "\n", linker_string)
    }
    fun generate_declaration_statement(node: *ast_node): str {
        var identifier = node->declaration_statement.identifier
        var ident_type = identifier->identifier.type
        var to_ret = type_to_c(identifier->identifier.type) + " " + get_name(identifier)
        if (identifier->identifier.is_extern)
            to_ret = "extern " + to_ret
        if (node->declaration_statement.expression) {
            // in case of recursive closures, make sure variable is declared before assignment
            /*to_ret += ";\n"*/
            /*to_ret += get_name(identifier) + " = " + generate(node->declaration_statement.expression)*/
            to_ret += " = " + generate(node->declaration_statement.expression)
        }
        if (node->declaration_statement.init_method_call) {
            error("init_method_call remaining")
        }
        return to_ret
    }
    fun generate_assignment_statement(node: *ast_node): str {
        return generate(node->assignment_statement.to) + " = " + generate(node->assignment_statement.from)
    }
    fun generate_if_statement(node: *ast_node): str {
        var if_str = "if (" + generate(node->if_statement.condition) + ") {\n" + generate(node->if_statement.then_part) + "}"
        if (node->if_statement.else_part)
            if_str += " else {\n" + generate(node->if_statement.else_part) + "}"
        return if_str + "\n"
    }
    fun generate_while_loop(node: *ast_node): str {
        return "while (" + generate(node->while_loop.condition) + ")\n" + generate(node->while_loop.statement)
    }
    fun generate_for_loop(node: *ast_node): str {
        var init = str(";")
        if (node->for_loop.init)
            init = generate(node->for_loop.init)
        var cond = str(";")
        if (node->for_loop.condition)
            cond = generate(node->for_loop.condition)
        // gotta take off last semicolon
        var update = str()
        if (node->for_loop.update) {
            update = generate(node->for_loop.update)
            if (update.length() < 2)
                error("update less than 2! Likely legal, but need easy compiler mod here")
            update = update.slice(0,-2)
        }
        return "for (" + init + cond + "; " + update + ")\n" + generate(node->for_loop.body)
    }
    fun generate_identifier(node: *ast_node): str {
        if (get_ast_type(node)->is_ref)
            error("still existin ref in identifier")
        return get_name(node)
    }
    fun generate_return_statement(node: *ast_node): str {
        if (node->return_statement.return_value)
            return "return " + generate(node->return_statement.return_value)
        return str("return")
    }
    fun generate_branching_statement(node: *ast_node): str {
        match(node->branching_statement.b_type) {
            branching_type::break_stmt() return str("break")
            branching_type::continue_stmt() return str("continue")
        }
    }
    fun generate_cast(node: *ast_node): str {
        return "((" + type_to_c(node->cast.to_type) + ")(" + generate(node->cast.value) + "))"
    }
    fun generate_value(node: *ast_node): str {
        var value = node->value.string_value
        if (node->value.value_type->base == base_type::character() && node->value.value_type->indirection == 0)
            return "'" + value + "'"
        if (node->value.value_type->base != base_type::character() || node->value.value_type->indirection != 1)
            return value

        var to_ret = str("\"") //"
        value.for_each(fun(c: char) {
            if (c == '\n')
                to_ret += "\\n"
            else if (c == '\\')
                to_ret += "\\\\"
            else if (c == '"')
                to_ret += "\\\""
            else
                to_ret += c
        })
        return to_ret + "\""
    }
    fun generate_code_block(node: *ast_node): str {
        var to_ret = str("{\n")
        node->code_block.children.for_each(fun(child: *ast_node) to_ret += generate(child) + ";\n";)
        return to_ret + "}"
    }
    // this generates the function as a value, not the actual function
    fun generate_function(node: *ast_node): str {
        return get_name(node)
    }
    fun generate_function_call(node: *ast_node): str {
        var func_name = generate(node->function_call.func)
        var call_string = str()
        var func_return_type = get_ast_type(node)

        var parameters = node->function_call.parameters
        if ( parameters.size == 2 && (func_name == "+" || func_name == "-" || func_name == "*" || func_name == "/"
            || func_name == "<" || func_name == ">" || func_name == "<=" || func_name == ">="
            || func_name == "==" || func_name == "!=" || func_name == "%" || func_name == "^"
            || func_name == "|" || func_name == "&" || func_name == ">>" || func_name == "<<"
        ))
            return "(" + generate(parameters[0]) + func_name + generate(parameters[1]) + ")"
        if ( parameters.size == 2 && (func_name == "||" || func_name == "&&"))
            error("Remaining || or &&")
        // don't propegate enclosing function down right of access
        // XXX what about enclosing object? should it be the thing on the left?
        if (func_name == "." || func_name == "->")
            return "(" + generate(parameters[0]) + func_name + generate(parameters[1]) + ")"
        if (func_name == "[]")
            return "(" + generate(parameters[0]) + "[" + generate(parameters[1]) + "])"
        // the post ones need to be post-ed specifically, and take the p off
        if (func_name == "++p" || func_name == "--p")
            return "(" + generate(parameters[0]) + ")" + func_name.slice(0,-2)

        // So we don't end up copy_constructing etc, we just handle the unary operators right here
        if (func_name == "*" || func_name == "&")
            return "(" + func_name + generate(parameters[0]) + ")"

        var func_type = get_ast_type(node->function_call.func)
        // regular parameter generation
        for (var i = 0; i < parameters.size; i++;) {
            var param = parameters[i]
            var in_function_param_type = null<type>()
            // grab type from param itself if we're out of param types (because variadic function)
            if (i < func_type->parameter_types.size)
                in_function_param_type = func_type->parameter_types[i]
            else
                in_function_param_type = get_ast_type(param)->clone_without_ref()
            if (call_string != "")
                call_string += ", "

            call_string += generate(param)
        }
        call_string = func_name + "(" + call_string + ")"
        return call_string
    }

    fun generate_compiler_intrinsic(node: *ast_node): str {
        if (node->compiler_intrinsic.intrinsic == "sizeof") {
            if (node->compiler_intrinsic.parameters.size || node->compiler_intrinsic.type_parameters.size != 1)
                error("wrong parameters to sizeof compiler intrinsic")
            return "sizeof(" + type_to_c(node->compiler_intrinsic.type_parameters[0]) + ")"
        } else if (node->compiler_intrinsic.intrinsic == "link") {
            node->compiler_intrinsic.parameters.for_each(fun(value: *ast_node) {
                linker_string += str("-l") + value->value.string_value + " "
            })
            return str()
        }
        error(node->compiler_intrinsic.intrinsic + ": unknown intrinsic")
        return str("ERROR")
    }

    fun generate(node: *ast_node): str {
        if (!node) return str("/*NULL*/")
        match (*node) {
            ast_node::declaration_statement(backing)    return generate_declaration_statement(node)
            ast_node::assignment_statement(backing)     return generate_assignment_statement(node)
            ast_node::if_statement(backing)             return generate_if_statement(node)
            ast_node::while_loop(backing)               return generate_while_loop(node)
            ast_node::for_loop(backing)                 return generate_for_loop(node)
            ast_node::function(backing)                 return generate_function(node)
            ast_node::function_call(backing)            return generate_function_call(node)
            ast_node::compiler_intrinsic(backing)       return generate_compiler_intrinsic(node)
            ast_node::code_block(backing)               return generate_code_block(node)
            ast_node::return_statement(backing)         return generate_return_statement(node)
            ast_node::branching_statement(backing)      return generate_branching_statement(node)
            ast_node::defer_statement(backing)          error("unremoved defer")
            ast_node::match_statement(backing)          error("unremoved match")
            ast_node::cast(backing)                     return generate_cast(node)
            ast_node::value(backing)                    return generate_value(node)
            ast_node::identifier(backing)               return generate_identifier(node)
        }
        error(str("COULD NOT GENERATE ") + get_ast_name(node))
        return str("/* COULD NOT GENERATE */")
    }
    fun type_to_c(type: *type): str {
        var indirection = str()
        if (type->is_ref) error("still ref in type_to_c") //indirection += "/*ref*/ *"
        for (var i = 0; i < type->indirection; i++;) indirection += "*"
        match (type->base) {
            base_type::none()               return str("none") + indirection
            base_type::template()           return str("template") + indirection
            base_type::template_type()      return str("template_type") + indirection
            base_type::void_return()        return str("void") + indirection
            base_type::boolean()            return str("bool") + indirection
            base_type::character()          return str("char") + indirection
            base_type::ucharacter()         return str("unsigned char") + indirection
            base_type::short_int()          return str("short") + indirection
            base_type::ushort_int()         return str("unsigned short") + indirection
            base_type::integer()            return str("int") + indirection
            base_type::uinteger()           return str("unsigned int") + indirection
            base_type::long_int()           return str("long") + indirection
            base_type::ulong_int()          return str("unsigned long") + indirection
            base_type::floating()           return str("float") + indirection
            base_type::double_precision()   return str("double") + indirection
            base_type::object()             return get_name(type->type_def) + indirection
            base_type::function() {
                type = type->clone_with_indirection(0,false)
                if (!function_type_map.contains_key(*type)) {
                    var temp_name = str("function") + get_id()
                    var temp = str()
                    type->parameter_types.for_each(fun(parameter_type: *type) {
                        temp += str(", ") + type_to_c(parameter_type) + " "
                        temp_name += "_" + cify_name(type_to_c(parameter_type))
                    })
                    if (type->is_raw)
                        function_typedef_string += str("typedef ") + type_to_c(type->return_type) + " (*" + temp_name + ")(" + temp.slice(1,-1) + ");\n"
                    else
                        error(type->to_string() + " is not raw!")
                    // again, the indirection
                    function_type_map[*type] = temp_name
                }
                return function_type_map[*type] + indirection
            }
        }
        return str("impossible type") + indirection
    }
    fun type_decoration(type: *type): str {
        return cify_name(type->to_string())
    }
    fun get_name(node: *ast_node): str {
        var maybe_it = ast_name_map.get_ptr_or_null(node);
        if (maybe_it)
            return *maybe_it
        var result = str("impossible name")
        var make_unique = true
        match (*node) {
            ast_node::type_def(backing) {
                var upper = backing.scope[str("~enclosing_scope")][0]
                result = cify_name(backing.name)
                if (is_template(upper))
                    upper->template.instantiated_map.reverse_get(node).for_each(fun(t: ref type) result += str("_") + type_decoration(&t);)
            }
            ast_node::function(backing) {
                // be careful, operators like . come through this
                if (backing.name == "main" || backing.is_extern || !backing.body_statement) {
                    result = backing.name
                    make_unique = false
                } else {
                    result = "fun_"
                    var upper = backing.scope.get_with_default(str("~enclosing_scope"), vec(null<ast_node>()))[0]
                    if (upper && is_type_def(upper))
                        result += get_name(upper) + "_"
                    result += cify_name(node->function.name)
                    node->function.parameters.for_each(fun(param: *ast_node) result += str("_") + type_decoration(param->identifier.type);)
                }
            }
            ast_node::identifier(backing) {
                if (backing.name == "this" || backing.is_extern)
                    make_unique = false
                result = backing.name
            }
        }
        if (result == "impossible name")
            error("HUGE PROBLEMS")
        if (make_unique && used_names.contains(result))
            result += get_id()
        ast_name_map.set(node, result)
        used_names.add(result)
        return result
    }
    fun cify_name(name: str): str {
        var to_ret = str()
        for (var i = 0; i < name.length(); i++;) {
            var replaced = false
            for (var j = longest_replacement; j > 0; j--;) {
                if (i + j <= name.length() && replacement_map.contains_key(name.slice(i,i+j))) {
                   to_ret += replacement_map[name.slice(i,i+j)]
                   replaced = true
                   i += j-1;
                   break
                }
            }
            if (!replaced)
               to_ret += name[i]
        }
        return to_ret
    }
}