kraken/k.krak

import io:*
import grammer:*
import lexer:*
import parser:*
import str:*
import serialize:*
import os:*
import set:*
import vec:*
import vec_literals:*
import poset:*
import util:*
import ast:*
import tree:*
import symbol:*

fun main(argc: int, argv: **char): int {
    // delay construction until we either load it or copy construct it
    var gram: grammer
    var base_dir = str("/").join(str(argv[0]).split('/').slice(0,-2))
    var file_name = base_dir + "/krakenGrammer.kgm"
    var compiled_name = file_name + str(".comp_new")
    var compiled_version = 1
    var file_contents = read_file(file_name)
    var loaded_and_valid = false

    if (argc <= 1) {
        println("No input file!\n Call with one argument (the input file), or two arguments (input file and output name)")
        exit(1)
    } else if (str(argv[1]) == "-v" || str(argv[1]) == "--version") {
        println("0.0 pre")
        exit(0)
    }
    var opt_str = str("-O2")
    var compile_c = true
    var positional_args = vec<str>()
    for (var i = 1; i < argc; i++;) {
        var arg_str = str(argv[i])
        if (arg_str.length() > 2 && arg_str.slice(0,2) == "-O") {
            opt_str = arg_str
        } else if (arg_str == "--no-c-compile") {
            compile_c = false
        } else {
            positional_args.add(arg_str)
        }
    }

    if (file_exists(compiled_name)) {
        var pos = 0
        var binary = read_file_binary(compiled_name)
        var saved_version = 0
        unpack(saved_version, pos) = unserialize<int>(binary, pos)
        if (saved_version == compiled_version) {
            var cached_contents = str()
            unpack(cached_contents, pos) = unserialize<str>(binary, pos)
            if (cached_contents == file_contents) {
                loaded_and_valid = true
                pos = gram.unserialize(binary, pos)
            } else println("contents different")
        } else println("version number different")
    } else {
        println("cached file does not exist")
    }
    if (!loaded_and_valid) {
        println("Not loaded_and_valid, re-generating and writing out")
        gram.copy_construct(&load_grammer(file_contents))
        println("grammer loaded, calculate_first_set")
        gram.calculate_first_set()
        println("grammer loaded, calculate_state_automaton")
        gram.calculate_state_automaton()
        println("calculated, writing out")
        write_file_binary(compiled_name, serialize(compiled_version) + serialize(file_contents) + serialize(gram))
        println("done writing")
    }

    var lex = lexer(gram.terminals)
    var parse.construct(&gram, &lex): parser

    var kraken_file_name = positional_args[0]
    var executable_name = str(".").join(kraken_file_name.split('.').slice(0,-2))
    if (positional_args.size > 1)
        executable_name = positional_args[1]

    var pass_poset = poset<pair<str, int>>()
    var name_ast_map = map<str, *tree<ast>>()
    var import_paths =  vec(str(), base_dir + "/stdlib/")
    var passes = vec<fun(str): *tree<ast>>()
    passes = vec(
        fun(file_name: str): *tree<ast> {
            var file = str()
            for (var i = 0; i < import_paths.size; i++;) {
                if (file_exists(import_paths[i] + file_name)) {
                    if (file != "")
                        error("File: " + file_name + ", found in multiple import paths - at least two of [" + str(",").join(import_paths) + "]")
                    file = read_file(import_paths[i] + file_name)
                }
            }
            if (file == "")
                error("File: " + file_name + ", not found in any import path - none of [" + str(",").join(import_paths) + "]")
            printerr(file_name + ", ")
            var parse_tree = parse.parse_input(file, file_name)
            trim(parse_tree)
            return syntax_to_ast(file_name, parse_tree)
        },
        fun(file_name: str): *tree<ast> {
            println("Checking for imports in " + file_name)
            name_ast_map[file_name]->children.for_each(fun(n: *tree<ast>) {
                match (n->data) {
                    ast::_import(b) {
                        var imported_file_name = n->children[0]->data._identifier.first + ".krak"
                        for (var i = 0; i < passes.size; i++;) {
                            if (i == 0)
                                pass_poset.add_relationship(make_pair(file_name, 2), make_pair(imported_file_name, 0))
                            else
                                pass_poset.add_relationship(make_pair(imported_file_name, i), make_pair(imported_file_name, i-1))
                        }
                    }
                }
            })
            return name_ast_map[file_name]
        },
        fun(file_name: str): *tree<ast> {
            println("Doing thing 3 to " + file_name)
            return name_ast_map[file_name]
        }
    )
    for (var i = 0; i < passes.size; i++;) {
        if (i == 0)
            pass_poset.add_vertex(make_pair(kraken_file_name, i))
        else
            pass_poset.add_relationship(make_pair(kraken_file_name, i), make_pair(kraken_file_name, i-1))
    }
    while (pass_poset.size() != 0) {
        var file_pass = pass_poset.pop()
        printlnerr("doing pass " + to_string(file_pass.second) + " on " + file_pass.first)
        name_ast_map[file_pass.first] = passes[file_pass.second](file_pass.first)
    }

    println()
    println()
    println("Finished with trees:")
    name_ast_map.for_each(fun(key: str, value: *tree<ast>) {
        printlnerr(key + ":")
        print_tree(value, 1)
        printlnerr("done")
    })

    var kraken_c_output_name = kraken_file_name + ".c"
    var c_code = str("//don't you wish this was real kraken\n")
    var c_flags = str("")
    write_file(kraken_c_output_name, c_code)

    if (compile_c) {
        var compile_string = "cc -g " + opt_str + " -Wno-int-to-pointer-cast -Wno-pointer-to-int-cast -Wno-incompatible-pointer-types -std=c99 " + c_flags + " " + kraken_c_output_name + " -o " + executable_name
        /*printlnerr(compile_string)*/
        /*system(compile_string)*/
    }

    return 0
}

fun syntax_to_ast(file_name: str, syntax: *tree<symbol>): *tree<ast> {
    var syntax_to_ast_helper: fun(*tree<symbol>): *tree<ast> = fun(syntax: *tree<symbol>): *tree<ast> {
        printlnerr("syntax_to_ast " + syntax->data.name)
        if (syntax->data.name == "import") {
            return _import(from_vector(syntax->children.slice(2,-1).filter(fun(s:*tree<symbol>):bool {
                return s->data.name == "identifier" || s->data.data == "*"
            }).map(fun(s: *tree<symbol>): str {
                return concat(s)
            })), vec(syntax_to_ast_helper(syntax->children[1])))
        } else if (syntax->data.name == "function")
            return _function(concat(get_node("func_identifier", syntax)), null<type>(),
                (get_nodes("typed_parameter", syntax) +
                 get_nodes("statement", syntax)).map(syntax_to_ast_helper))
        else if (syntax->data.name == "typed_parameter")
            return _identifier(concat(get_node("identifier", syntax)), null<type>())
        else if (syntax->data.name == "type_def")
            return _type_def(concat(get_node("identifier", syntax)),
                             get_nodes("declaration_statement", syntax).map(syntax_to_ast_helper))
        else if (syntax->data.name == "adt_def")
            return _type_def(concat(get_node("identifier", syntax)),
                             get_nodes("adt_option", syntax).map(fun(s: *tree<symbol>): *tree<ast> {
                                return _identifier(concat(get_node("identifier", s)), null<type>())
                             }))
        else if (syntax->data.name == "statement")
            return syntax_to_ast_helper(syntax->children[0])
        else if (syntax->data.name == "code_block")
            return _block(syntax->children.map(syntax_to_ast_helper))
        else if (syntax->data.name == "return_statement")
            return _return(syntax->children.map(syntax_to_ast_helper))
        else if (syntax->data.name == "defer_statement")
            return _defer(syntax->children.map(syntax_to_ast_helper))
        else if (syntax->data.name == "match_statement") {
            return _match(vec(syntax_to_ast_helper(get_node("boolean_expression", syntax))) +
                              get_nodes("case_statement", syntax).map(fun(s: *tree<symbol>): *tree<ast> {
                                    return _case(s->children.map(syntax_to_ast_helper))
                              }))
        } else if (syntax->data.name == "declaration_statement") {
            var children = vec(_identifier(concat(get_node("identifier", syntax)), null<type>()))
            children += get_nodes("boolean_expression", syntax).map(syntax_to_ast_helper)
            return _declaration(children)
        } else if (syntax->data.name == "assignment_statement")
            return _assignment(vec(_binding(concat(syntax->children[1]), null<tree<ast>>()),
                                   syntax_to_ast_helper(syntax->children[0]),
                                   syntax_to_ast_helper(syntax->children[2])))
        else if (syntax->data.name == "function_call")
            return _call(syntax->children.map(fun(s: *tree<symbol>): *tree<ast> {
                                return syntax_to_ast_helper(s->children[0])
                         }))
        else if (syntax->data.name == "boolean_expression" ||
                 syntax->data.name == "and_boolean_expression" ||
                 syntax->data.name == "bitwise_or" ||
                 syntax->data.name == "bitwise_xor" ||
                 syntax->data.name == "bitwise_and" ||
                 syntax->data.name == "bool_exp" ||
                 syntax->data.name == "expression" ||
                 syntax->data.name == "shiftand" ||
                 syntax->data.name == "term" ||
                 syntax->data.name == "factor" ||
                 syntax->data.name == "unarad" ||
                 syntax->data.name == "access_operation") {
            if (syntax->children.size == 1) {
                return syntax_to_ast_helper(syntax->children[0])
            } else if (syntax->children.size == 2) {
                if (syntax->children[0]->data.terminal) {
                    return _call(vec(_binding(concat(syntax->children[0]), null<tree<ast>>()),
                                     syntax_to_ast_helper(syntax->children[1])))
                } else {
                    return _call(vec(_binding(concat(syntax->children[1]), null<tree<ast>>()),
                                     syntax_to_ast_helper(syntax->children[0])))
                }
            } else {
                return _call(vec(_binding(concat(syntax->children[1]), null<tree<ast>>()),
                                 syntax_to_ast_helper(syntax->children[0]),
                                 syntax_to_ast_helper(syntax->children[2])))
            }
        } else if (syntax->data.name == "number")
            return _value(concat(syntax), null<type>())
        else if (syntax->data.name == "scoped_identifier" || syntax->data.name == "identifier")
            return _binding(concat(syntax), null<tree<ast>>())
        else
            return null<tree<ast>>()
    }
    var result = _translation_unit(file_name, syntax->children.map(syntax_to_ast_helper))
    printlnerr("made")
    print_tree(result, 1)
    printlnerr("from")
    print_tree(syntax, 1)
    return result
}
fun print_tree<T>(t: *tree<T>, level: int) {
    printlnerr("\t" * level + to_string(t->data))
    for (var i = 0; i < t->children.size; i++;)
        if (t->children[i])
            print_tree(t->children[i], level+1)
        else
            printlnerr("\t" * (level + 1) + "null!")
}
fun get_node(lookup: *char, parent: *tree<symbol>): *tree<symbol> {
    return get_node(str(lookup), parent)
}
fun get_node(lookup: str, parent: *tree<symbol>): *tree<symbol> {
    var results = get_nodes(lookup, parent)
    if (results.size > 1)
        error(parent, "get node too many results!")
    if (results.size)
        return results[0]
    return null<tree<symbol>>()
}
fun get_nodes(lookup: *char, parent: *tree<symbol>): vec<*tree<symbol>> {
    return get_nodes(str(lookup), parent)
}
fun get_nodes(lookup: str, parent: *tree<symbol>): vec<*tree<symbol>> {
    return parent->children.filter(fun(node: *tree<symbol>):bool return node->data.name == lookup;)
}
fun concat(node: *tree<symbol>): str {
    var str.construct(): str
    if (node->data.data != "no_value")
        str += node->data.data
    node->children.for_each(fun(child: *tree<symbol>) str += concat(child);)
    return str
}
fun get_first_terminal(source: *tree<symbol>): *tree<symbol> {
    if (!source)
        return null<tree<symbol>>()
    if (source->data.terminal)
        return source
    if (source->children.size == 0)
        return null<tree<symbol>>()
    return get_first_terminal(source->children.first())
}
fun error(source: *tree<symbol>, message: *char) error(source, str(message));
fun error(source: *tree<symbol>, message: str) {
    var first = get_first_terminal(source)
    if (first)
        error("***error |" + concat(source) + "| *** " + first->data.source + ": " + first->data.position + " " + message)
    error(message)
}
fun trim(parse_tree: *tree<symbol>) {
    remove_node(symbol("$NULL$", false), parse_tree)
    remove_node(symbol("WS", false), parse_tree)
    // the terminals have " around them, which we have to escape
    remove_node(symbol("\"\\(\"", true), parse_tree)
    remove_node(symbol("\"\\)\"", true), parse_tree)
    remove_node(symbol("\"template\"", true), parse_tree)
    remove_node(symbol("\"return\"", true), parse_tree)
    remove_node(symbol("\"defer\"", true), parse_tree)
    remove_node(symbol("\";\"", true), parse_tree)
    remove_node(symbol("line_end", false), parse_tree)
    remove_node(symbol("\"{\"", true), parse_tree)
    remove_node(symbol("\"}\"", true), parse_tree)
    remove_node(symbol("\"(\"", true), parse_tree)
    remove_node(symbol("\")\"", true), parse_tree)
    remove_node(symbol("\"if\"", true), parse_tree)
    remove_node(symbol("\"while\"", true), parse_tree)
    remove_node(symbol("\"__if_comp__\"", true), parse_tree)
    remove_node(symbol("\"comp_simple_passthrough\"", true), parse_tree)
    /*remove_node(symbol("obj_nonterm", false), parse_tree)*/
    remove_node(symbol("adt_nonterm", false), parse_tree)

    collapse_node(symbol("case_statement_list", false), parse_tree)
    collapse_node(symbol("opt_param_assign_list", false), parse_tree)
    collapse_node(symbol("param_assign_list", false), parse_tree)
    collapse_node(symbol("opt_typed_parameter_list", false), parse_tree)
    collapse_node(symbol("opt_parameter_list", false), parse_tree)
    collapse_node(symbol("intrinsic_parameter_list", false), parse_tree)
    collapse_node(symbol("identifier_list", false), parse_tree)
    collapse_node(symbol("adt_option_list", false), parse_tree)
    collapse_node(symbol("statement_list", false), parse_tree)
    collapse_node(symbol("parameter_list", false), parse_tree)
    collapse_node(symbol("typed_parameter_list", false), parse_tree)
    collapse_node(symbol("unorderd_list_part", false), parse_tree)
    collapse_node(symbol("if_comp_pred", false), parse_tree)
    collapse_node(symbol("declaration_block", false), parse_tree)
    collapse_node(symbol("type_list", false), parse_tree)
    collapse_node(symbol("opt_type_list", false), parse_tree)
    collapse_node(symbol("template_param_list", false), parse_tree)
    collapse_node(symbol("trait_list", false), parse_tree)
    collapse_node(symbol("dec_type", false), parse_tree)
}
fun remove_node(remove: symbol, parse_tree: *tree<symbol>) {
    var to_process = stack<*tree<symbol>>()
    to_process.push(parse_tree)
    while(!to_process.empty()) {
        var node = to_process.pop()
        for (var i = 0; i < node->children.size; i++;) {
            if (!node->children[i] || node->children[i]->data.equal_wo_data(remove)) {
                node->children.remove(i)
                i--;
            } else {
                to_process.push(node->children[i])
            }
        }
    }
}
fun collapse_node(remove: symbol, parse_tree: *tree<symbol>) {
    var to_process = stack<*tree<symbol>>()
    to_process.push(parse_tree)
    while(!to_process.empty()) {
        var node = to_process.pop()
        for (var i = 0; i < node->children.size; i++;) {
            if (node->children[i]->data.equal_wo_data(remove)) {
                var add_children = node->children[i]->children;
                // stick child's children between the current children divided
                // on i, without including i
                node->children = node->children.slice(0,i) +
                                    add_children + node->children.slice(i+1,-1)
                i--;
            } else {
                to_process.push(node->children[i])
            }
        }
    }
}