Sheerka-Old/src/parsers/DefRuleParser.py

from dataclasses import dataclass
from typing import List

import core.utils
from core.builtin_concepts import ReturnValueConcept
from core.builtin_concepts_ids import BuiltinConcepts
from core.global_symbols import NotInit
from core.sheerka.services.SheerkaExecute import ParserInput
from core.sheerka.services.SheerkaRuleManager import SheerkaRuleManager, CompiledCondition
from core.sheerka.services.sheerka_service import FailedToCompileError
from core.tokenizer import Keywords, TokenKind
from parsers.BaseCustomGrammarParser import BaseCustomGrammarParser, NameNode, KeywordNotFound, SyntaxErrorNode
from parsers.BaseParser import Node, UnexpectedEofParsingError
from parsers.FormatRuleActionParser import FormatAstNode
from sheerkarete.conditions import AndConditions


@dataclass()
class DefRuleNode(Node):
    tokens: dict
    name: NameNode = NotInit
    python: List[CompiledCondition] = NotInit
    rete: List[AndConditions] = NotInit


@dataclass()
class DefExecRuleNode(DefRuleNode):
    then: ReturnValueConcept = NotInit


@dataclass
class DefFormatRuleNode(DefRuleNode):
    print: FormatAstNode = NotInit


class DefRuleParser(BaseCustomGrammarParser):
    DEF_KEYWORDS = [Keywords.RULE, Keywords.AS]
    DEF_KEYWORDS_VALUES = [k.value for k in DEF_KEYWORDS]

    RULE_KEYWORDS = [Keywords.WHEN, Keywords.THEN, Keywords.PRINT]
    RULE_KEYWORDS_VALUES = [k.value for k in RULE_KEYWORDS]

    def __init__(self, **kwargs):
        BaseCustomGrammarParser.__init__(self, "DefRule", 60)

    def parse(self, context, parser_input: ParserInput):
        if not isinstance(parser_input, ParserInput):
            return None

        # rule parser can only manage string text
        if parser_input.from_tokens:
            ret = context.sheerka.ret(
                self.name,
                False,
                context.sheerka.new(BuiltinConcepts.NOT_FOR_ME, body=parser_input))
            self.log_result(context, parser_input, ret)
            return ret

        context.log(f"Parsing '{parser_input}' with DefRuleParser", self.name)
        sheerka = context.sheerka

        if parser_input.is_empty():
            return sheerka.ret(self.name,
                               False,
                               sheerka.new(BuiltinConcepts.IS_EMPTY))

        if not self.reset_parser(context, parser_input):
            return self.sheerka.ret(self.name,
                                    False,
                                    context.sheerka.new(BuiltinConcepts.ERROR, body=self.error_sink))

        self.parser_input.next_token()
        node = self.parse_def_rule()

        body = self.get_return_value_body(sheerka, parser_input.as_text(), node, node, self.error_sink)
        ret = sheerka.ret(self.name, not self.has_error, body)

        self.log_result(context, parser_input.as_text(), ret)
        return ret

    def parse_def_rule(self):
        token = self.parser_input.token
        if token.value == Keywords.DEF.value:
            return self.parse_rule_name()
        elif token.value in (Keywords.WHEN.value, Keywords.PRINT.value):
            return self.parse_rule()
        else:
            self.add_error(KeywordNotFound([], [Keywords.WHEN.value]))
            return None

    def parse_rule_name(self):
        """
        Parses def rule xxx as yyyy
        """
        self.parser_input.next_token()  # eat def
        token = self.parser_input.token
        if token.value != Keywords.RULE.value:
            self.add_error(KeywordNotFound([token], [Keywords.RULE.value]))
            return None

        buffer = []
        while self.parser_input.next_token(skip_whitespace=False):
            token = self.parser_input.token
            if token.value == Keywords.AS.value:
                break
            else:
                buffer.append(token)
        else:  # 'as' keyword not found
            self.add_error(KeywordNotFound([], [Keywords.AS.value]))
            return None

        if not self.parser_input.next_token():  # eat as
            self.add_error(UnexpectedEofParsingError("While parsing 'when'."))
            return None

        rule = self.parse_rule()

        name_node = self.get_concept_name(buffer)
        if name_node is None:
            return rule

        rule.name = name_node
        return rule

    def parse_rule(self):
        """"
        Parses 'when xxx then yyy'
        or 'when xxx print yyy'
        """
        parts = self.get_parts(self.RULE_KEYWORDS_VALUES, strip_tokens=True)
        if Keywords.THEN in parts and Keywords.PRINT in parts:
            self.add_error(SyntaxErrorNode([], "Cannot have both 'print' and 'then' keywords"))
            return None

        if Keywords.THEN not in parts and Keywords.PRINT not in parts:
            self.add_error(KeywordNotFound([], [Keywords.THEN.value, Keywords.PRINT.value]))
            return None

        return self.parse_format_rule(parts) if Keywords.PRINT in parts else self.parse_exec_rule(parts)

    def parse_exec_rule(self, parts):
        node = DefExecRuleNode(parts)
        try:
            compiled_result = self.get_when(parts[Keywords.WHEN])
            if compiled_result is None:
                return node
            node.python = compiled_result.python_conditions
            node.rete = compiled_result.rete_conditions

            parsed = self.get_then(parts[Keywords.THEN])
            if parsed is None:
                return node
            node.then = parsed
        except KeyError as e:
            self.add_error(KeywordNotFound([], [e.args[0].value]))
            return None

        return node

    def parse_format_rule(self, parts):
        node = DefFormatRuleNode(parts)
        try:
            compiled_result = self.get_when(parts[Keywords.WHEN])
            if compiled_result is None:
                return node
            node.python = compiled_result.python_conditions
            node.rete = compiled_result.rete_conditions

            parsed = self.get_print(parts[Keywords.PRINT])
            if parsed is None:
                return node
            node.print = parsed
        except KeyError as e:
            self.add_error(KeywordNotFound([], [e.args[0].value]))
            return None

        return node

    def get_when(self, tokens):
        """
        Validate the when part of the rule.
        :param tokens:
        :return:
        """
        source = core.utils.get_text_from_tokens(core.utils.strip_tokens(tokens[1:]))
        try:
            rule_manager_service = self.sheerka.services[SheerkaRuleManager.NAME]
            compiled_result = rule_manager_service.compile_when(self.context, self.name, source)
        except FailedToCompileError as ex:
            for c in ex.cause:
                self.add_error(c)
            return None

        return compiled_result

    def get_then(self, tokens):
        source = core.utils.get_text_from_tokens(core.utils.strip_tokens(tokens[1:]))
        res = self.sheerka.services[SheerkaRuleManager.NAME].compile_exec(self.context, source)

        if not res.status:
            self.add_error(res.value)
            return None

        return res

    def get_print(self, tokens):
        """
        Validate the print part
        :param tokens:
        :return:
        """
        source = core.utils.get_text_from_tokens(core.utils.strip_tokens(tokens[1:]))
        res = self.sheerka.services[SheerkaRuleManager.NAME].compile_print(self.context, source)
        if not res.status:
            self.add_error(res.value)
            return None

        return res.body

    def get_concept_name(self, tokens):
        name_tokens = core.utils.strip_tokens(tokens)
        if len(name_tokens) == 0:
            self.add_error(SyntaxErrorNode([], "Name is mandatory"))
            return None

        for token in name_tokens:
            if token.type == TokenKind.NEWLINE:
                self.add_error(SyntaxErrorNode([token], "Newline are not allowed in name."))
                return None

        name_node = NameNode(name_tokens)  # skip the first token
        return name_node