# SPDX-License-Identifier: MIT
# Copyright 2021 Jookia <contact@jookia.org>

import log

class Token:
    def __init__(self, type, value):
        self.type = type
        self.value = value

    def __repr__(self):
        return "Token(type %s, value '%s')" % (self.type, self.value)

def tokenize(code):
    tokens = []
    token = ""
    text = ""
    mode = "normal" # normal/note/text
    for symbol in code:
        if symbol == '\n':
            symbol_print = "new line"
        else:
            symbol_print = "symbol '%s'" % (symbol)
        log.log(log.LEXER, log.TRACE, "Read %s" % (symbol_print))
        if symbol == " " or symbol == "\t" or symbol == "\n":
            log.log(log.LEXER, log.TRACE, "Read token '%s'" % (token))
            if token == "":
                pass
            elif token == "BeginNote":
                log.log(log.LEXER, log.TRACE, "Switching to note mode")
                mode = "note"
            elif token == "EndNote":
                log.log(log.LEXER, log.TRACE, "Ending note mode")
                mode = "normal"
            elif token == "BeginText":
                log.log(log.LEXER, log.TRACE, "Switching to text mode")
                mode = "text"
            elif token == "EndText":
                log.log(log.LEXER, log.TRACE, "Ending text mode")
                content = text[1:-8]
                log.log(log.LEXER, log.DEBUG, "Appending text '%s'" % (content))
                tokens.append(Token("text", content))
                mode = "normal"
                text = ""
            elif token != "":
                if mode == "normal":
                    keywords = ["NewLang", "Done", "Set", "To", "EndSet",
                                "If", "Then", "Else", "EndIf"]
                    if token in keywords:
                        type = "keyword"
                        token = token.lower()
                    else:
                        type = "symbol"
                    tok = Token(type, token)
                    log.log(log.LEXER, log.DEBUG, "Appending token %s" % (tok))
                    tokens.append(tok)
                else:
                    log.log(log.LEXER, log.TRACE, "Skipping token '%s'" % (token))
            token = ""
        else:
            token += symbol
        if mode == "text":
            text += symbol
    log.log(log.LEXER, log.TRACE, "Done lexing, added EOF")
    tokens.append(Token("EOF", None))
    return tokens

class Reference:
    def __init__(self, value):
        self.value = value

    def __repr__(self):
        return "Reference('%s')" % (self.value)

class Text:
    def __init__(self, value):
        self.value = value

    def __repr__(self):
        return "Text('%s')" % (self.value)

class Statement:
    def __init__(self, subject, verb, arguments):
        self.subject = subject
        self.verb = verb
        self.arguments = arguments

    def __repr__(self):
        return "Statement(subject %s, verb '%s', arguments %s)" % (self.subject, self.verb, self.arguments)

class Set:
    def __init__(self, subject, statement):
        self.subject = subject
        self.statement = statement

    def __repr__(self):
        return "Set(subject %s, statement %s)" % (self.subject, self.statement)

class Conditional:
    def __init__(self, test, success, failure):
        self.test = test
        self.success = success
        self.failure = failure

    def __repr__(self):
        return "Conditional(test %s, success %s, failure %s)" % (self.test, self.success, self.failure)

class Parser:
    def __init__(self, tokens):
        self.tokens = tokens
        self.pos = 0

    def next(self):
        token = self.tokens[self.pos]
        if self.pos < (len(self.tokens) - 1):
            self.pos += 1
        log.log(log.PARSER, log.TRACE, "Read %s" % (token))
        return token

    def peek(self):
        token = self.tokens[self.pos]
        log.log(log.PARSER, log.TRACE, "Peeked %s" % (token))
        return token

    def parse_version(self):
        token = self.next()
        if token.type != "keyword" or token.value != "newlang":
            log.log(log.PARSER, log.NORMAL, "Expected NewLang keyword")
            return None
        token = self.next()
        log.log(log.PARSER, log.DEBUG, "Parsed language version %s" % (token.value))
        return token.value

    def parse_value(self, type, value):
        if type == "symbol":
            return Reference(value)
        elif type == "text":
            return Text(value)
        else:
            log.log(log.PARSER, log.NORMAL, "Unexpected type %s" % (type))
            return None

    def parse_arguments(self, terminator):
        args = []
        while True:
            token = self.next()
            if token.type == "keyword":
                if token.value == terminator:
                    return args
                else:
                    log.log(log.PARSER, log.NORMAL, "Unexpected keyword %s" % (token.value))
                    return None
            else:
                arg = self.parse_value(token.type, token.value)
                if not arg:
                    log.log(log.PARSER, log.NORMAL, "While parsing argument")
                    return None
                log.log(log.PARSER, log.TRACE, "Parsed %s" % (arg))
                args.append(arg)

    def parse_statement(self, terminator):
        token = self.next()
        subject = self.parse_value(token.type, token.value)
        if not subject:
            log.log(log.PARSER, log.NORMAL, "While parsing subject")
            return None
        token = self.next()
        if token.type == "keyword":
            if token.value == terminator:
                verb = None
            else:
                log.log(log.PARSER, log.NORMAL, "Unexpected keyword %s" % (token.value))
                return None
        elif token.type == "symbol":
            verb = token.value
        else:
            verb = token.value
        if verb:
            arguments = self.parse_arguments(terminator)
            if arguments is None:
                log.log(log.PARSER, log.NORMAL, "While parsing arguments")
                return None
        else:
            arguments = []
        statement = Statement(subject, verb, arguments)
        log.log(log.PARSER, log.DEBUG, "Parsed %s" % (statement))
        return statement

    def parse_set(self):
        token = self.next()
        if token.type != "symbol":
            log.log(log.PARSER, log.NORMAL, "Expect symbol, got %s" % (token.type))
            return None
        subject = token.value
        token = self.next()
        if token.type != "keyword" or token.value != "to":
            log.log(log.PARSER, log.NORMAL, "Expected To, got %s %s" % (token.type, token.value))
            return None
        log.log(log.PARSER, log.TRACE, "Parsing set value...")
        ast = self.parse_statement("endset")
        if not ast:
            log.log(log.PARSER, log.NORMAL, "While parsing statement")
            return None
        set = Set(subject, ast)
        log.log(log.PARSER, log.DEBUG, "Parsed %s" % (set))
        return set

    def parse_if(self):
        log.log(log.PARSER, log.TRACE, "Parsing if test condition...")
        test = self.parse_statement("then")
        if not test:
            log.log(log.PARSER, log.NORMAL, "While parsing test condition")
            return None
        log.log(log.PARSER, log.TRACE, "Parsing if success statement...")
        success = self.parse_statement("else")
        if not success:
            log.log(log.PARSER, log.NORMAL, "While parsing success statement")
            return None
        log.log(log.PARSER, log.TRACE, "Parsing if failure statement...")
        failure = self.parse_statement("endif")
        if not failure:
            log.log(log.PARSER, log.NORMAL, "While parsing failure statement")
            return None
        conditional = Conditional(test, success, failure)
        log.log(log.PARSER, log.DEBUG, "Parsed %s" % (conditional))
        return conditional

    def parse_directive(self):
        token = self.peek()
        if token.type != "keyword" and token.type != "symbol":
            log.log(log.PARSER, log.NORMAL, "Expected keyword or symbol here, got %s" % (token.type))
            return None
        if token.type == "keyword":
            self.next()
            if token.value == "set":
                ast = self.parse_set()
                if not ast:
                    log.log(log.PARSER, log.NORMAL, "While parsing set directive")
                    return None
                return ast
            elif token.value == "if":
                ast = self.parse_if()
                if not ast:
                    log.log(log.PARSER, log.NORMAL, "While parsing set directive")
                    return None
                return ast
            else:
                log.log(log.PARSER, log.NORMAL, "Unexpected keyword %s" % (token.value))
                return None
        else:
            ast = self.parse_statement("done")
            if not ast:
                log.log(log.PARSER, log.NORMAL, "While parsing statement")
                return None
            return ast

    def parse_file(self):
        log.log(log.PARSER, log.TRACE, "Parsing file...")
        ast = []
        version = self.parse_version()
        if not version:
            log.log(log.PARSER, log.NORMAL, "While parsing version identifier at start of file")
            return None
        if version != "0":
            log.log(log.PARSER, log.NORMAL, "Invalid version identifier %s" % (version))
            return None
        while self.peek().type != "EOF":
            directive = self.parse_directive()
            if directive == None:
                log.log(log.PARSER, log.NORMAL, "While parsing directive in file")
                return None
            else:
                ast.append(directive)
        log.log(log.PARSER, log.DEBUG, "Parsed file")
        return ast

def parse_file(code):
    tokens = tokenize(code)
    parser = Parser(tokens)
    return parser.parse_file()