gazprea-fuzzer-python/ast_parser/general_unparser.py

import xml
import xml.dom.minidom
import xml.etree.ElementTree as ET

from constants import *


class GeneralUnparser:
    def __init__(self, input: str or ET.Element,
                 from_xml: bool = False,
                 indentation_character: str = " ",
                 endline: str = "\n",
                 instream_begin_delimiter: str = "",
                 instream_end_delimiter: str = "",
                 outstream_begin_delimiter: str = "",
                 outstream_end_delimiter: str = "",
                 return_string: str = "return",
                 function_return_type_indicator_predicate: str = "returns",
                 arg_start_delimiter: str = "(",
                 arg_end_delimiter: str = ")",
                 arg_separator: str = ",",
                 loop_start_delimiter: str = "loop while (",
                 loop_end_delimiter: str = ")",
                 assignment_character: str = '=',
                 conditional_start_delimiter: str = "if (",
                 conditional_end_delimiter: str = ")",
                 conditional_case_delimiter: str = "else if (",
                 conditional_else_delimiter: str = "else",
                 block_start_delimiter: str = "{",
                 block_end_delimiter: str = "}",
                 strip_conditionals: bool = False,
                 ):
        self.endline = endline
        self.indentation_character = indentation_character
        self.indentation = 0

        self.stream_delimiters = {
            "instream_begin": instream_begin_delimiter,
            "instream_end": instream_end_delimiter,
            "outstream_begin": outstream_begin_delimiter,
            "outstream_end": outstream_end_delimiter
        }

        self.arg_delimiters = (arg_start_delimiter, arg_end_delimiter, arg_separator)
        self.loop_delimiters = (loop_start_delimiter, loop_end_delimiter)
        self.conditional_delimiters = (conditional_start_delimiter,
                                       conditional_end_delimiter,
                                       conditional_case_delimiter,
                                       conditional_else_delimiter)

        self.return_string = return_string
        self.return_indicator = function_return_type_indicator_predicate
        self.assignment_character = assignment_character
        self.block_delimiters = (block_start_delimiter, block_end_delimiter)

        self.strip_conditionals = strip_conditionals

        if from_xml:
            self.xml = input
            self.source = None
        else:
            self.source = input
            self.xml = None

    def unparse(self):
        """
        @brief unparses the xml into valid gazprea code

        :return: a string of valid gazprea code
        """
        self.source = ""

        self.setup()

        for node in self.xml:
            self.unparse_node(node)

    def unparse_node(self, node):
        if node.tag not in [GAZ_VAR_TAG, GAZ_RHS_TAG, GAZ_LHS_TAG, GAZ_LIT_TAG, GAZ_OPERATOR_TAG, GAZ_BRACKET_TAG]:
            self.source += self.indentation_character * self.indentation

        if node.tag == GAZ_BLOCK_TAG:
            self.unparse_block(node)
        elif node.tag == GAZ_DECLARATION_TAG:
            self.unparse_declaration(node)
        elif node.tag == GAZ_RETURN_TAG:
            self.unparse_return(node)
        elif node.tag == GAZ_OPERATOR_TAG:
            self.unparse_operator(node)
        elif node.tag == GAZ_UNARY_OPERATOR_TAG:
            self.unparse_unary(node)
        elif node.tag == GAZ_STREAM_TAG:
            self.unparse_stream(node)
        elif node.tag == GAZ_LIT_TAG:
            self.unparse_literal(node)
        elif node.tag == GAZ_PROCEDURE_TAG or node.tag == GAZ_FUNCTION_TAG:
            self.unparse_routine(node)
        elif node.tag == GAZ_VAR_TAG:
            self.unparse_variable(node)
        elif node.tag == GAZ_RHS_TAG or node.tag == GAZ_LHS_TAG:
            self.unparse_xhs(node)
        elif node.tag == GAZ_LIT_TAG:
            self.unparse_literal(node)
        elif node.tag == GAZ_ASSIGNMENT_TAG:
            self.unparse_assignment(node)
        elif node.tag == GAZ_IF_TAG:
            self.unparse_conditional(node)
        elif node.tag == GAZ_LOOP_TAG:
            self.unparse_loop(node)
        elif node.tag == GAZ_BRACKET_TAG:
            self.unparse_brackets(node)
        elif node.tag == GAZ_BREAK_TAG:
            self.unparse_break(node)
        elif node.tag is None:
            raise NoneTagException("None tag encountered", self.xml)
        else:
            with open("unknown_tag.xml", "w") as f:
                f.write(self.pretty_xml())
            raise Exception("Unknown tag: " + node.tag)

    def unparse_block(self, node):
        if node.get(GAZ_TY_KEY) is None:
            self.source += f"{self.indentation * self.indentation_character}"

        self.source += f"{self.block_delimiters[0]}\n"
        self.indentation += 4

        for child in node:
            self.unparse_node(child)
        self.indentation -= 4

        if node.get(GAZ_TY_KEY) is None:
            self.source += f"{self.indentation * self.indentation_character}{self.block_delimiters[1]}\n\n"
        elif node.get(GAZ_TY_KEY) in [GAZ_TRUE_BLOCK_TAG, GAZ_FALSE_BLOCK_TAG]:
            self.source += f"{self.indentation * self.indentation_character}{self.block_delimiters[1]}"

    def unparse_declaration(self, node, declaration_op: str = '='):
        variable = node.find(GAZ_VAR_TAG)
        rhs = node.find(GAZ_RHS_TAG)
        self.unparse_variable(variable, True)
        self.source += f" {declaration_op} "
        self.unparse_node(rhs)
        self.source += self.endline

    def unparse_variable(self, node, is_declaration: bool = False):
        if node is None:
            raise ValueError("Node is None")
        mut = node.get(GAZ_QUALIFIER_KEY)
        type = node.get(GAZ_TY_KEY)
        name = node.get(GAZ_NAME_KEY)

        self.source += self.format_variable(mut, self.translate_type(type), name, is_declaration)

    def unparse_stream(self, node):
        if node.get(GAZ_TY_KEY) == GAZ_OUT_STREAM:
            self.unparse_outstream(node)
        elif node.get(GAZ_TY_KEY) == GAZ_IN_STREAM:
            self.unparse_instream(node)

    def unparse_instream(self, node):
        self.source += self.stream_delimiters["instream_begin"]
        for child in node:
            self.unparse_node(child)

        self.source += self.stream_delimiters["instream_end"]
        self.source += self.endline

    def unparse_outstream(self, node):
        self.source += self.stream_delimiters["outstream_begin"]
        for child in node:
            self.unparse_node(child)

        self.source += self.stream_delimiters["outstream_end"]
        self.source += self.endline

    def unparse_literal(self, node):
        self.source += "{}".format(self.translate_value(node.get(GAZ_VAL_KEY)))

    def unparse_xhs(self, node):
        for child in node:
            self.unparse_node(child)

    def unparse_operator(self, node):
        ty = node.get(GAZ_TY_KEY)
        self.unparse_xhs(node.find(GAZ_LHS_TAG))
        self.source += " {} ".format(self.translate_op(node.get("op"), ty))
        self.unparse_xhs(node.find(GAZ_RHS_TAG))

    def unparse_return(self, node):
        self.source += self.return_string + " "
        for child in node:
            self.unparse_node(child)
        self.source += self.endline

    def unparse_routine(self, node):
        return_type = ""
        if node.get(GAZ_RETURN_KEY) not in [None, GAZ_VOID_TYPE, ""]:
            return_type = self.return_indicator + " " + self.translate_type(node.get(GAZ_RETURN_KEY))

        args = []
        for child in node:
            if child.tag == GAZ_ARG_TAG:
                args.append(child)

        args = self.unparse_argument(args)

        name = node.get(GAZ_NAME_KEY)

        self.source += self.function_declaration(node.tag, args, name, return_type)

        for child in node:
            if child.tag == GAZ_ARG_TAG:
                continue
            self.unparse_node(child)

        # cls.source += "}\n\n" #blocks are already there

    def function_declaration(self, xml_tag, args, name, return_type):
        return "{} {}{} {} ".format(
            xml_tag,
            name,
            args,
            return_type,
        )

    def unparse_argument(self, nodes):
        if len(nodes) == 0:
            return self.arg_delimiters[0] + self.arg_delimiters[1]
        args = self.arg_delimiters[0]
        for i in range(len(nodes)):
            args += self.unparse_single_arg(nodes[i])
            if i < len(nodes) - 1:
                args += self.arg_delimiters[2] + " "

        args += self.arg_delimiters[1]
        return args

    def unparse_assignment(self, element_in: ET.Element):
        self.unparse_variable(element_in.find(GAZ_VAR_TAG), False)
        self.source += " {} ".format(self.assignment_character)
        self.unparse_node(element_in.find(GAZ_RHS_TAG))
        self.source += self.endline

    def unparse_conditional(self, element_in: ET.Element):
        self.source += self.conditional_delimiters[0]
        i = 0
        for node in element_in:
            if node.tag == GAZ_IF_TAG:
                if self.strip_conditionals:
                    self.source += (self.conditional_delimiters[2] + " ").strip()
                else:
                    self.source += (self.conditional_delimiters[2])  # TODO this does wrong
                self.unparse_conditional(node)
            elif node.tag != GAZ_BLOCK_TAG:
                self.unparse_node(node)
                if self.strip_conditionals:
                    self.source += (self.conditional_delimiters[1] + " ").strip()
                else:
                    self.source += (self.conditional_delimiters[1] + " ")
            elif node.tag == GAZ_BLOCK_TAG:
                self.source += self.indentation * self.indentation_character
                if node.get(GAZ_TY_KEY) == GAZ_TRUE_BLOCK_TAG:
                    self.unparse_node(node)
                elif node.get(GAZ_TY_KEY) == GAZ_FALSE_BLOCK_TAG:
                    if self.strip_conditionals:
                        self.source += (self.conditional_delimiters[3] + " ").strip()
                    else:
                        self.source += (" " + self.conditional_delimiters[3] + " ")
                    self.unparse_node(node)
                else:
                    self.unparse_node(node)
            i += 1

    def unparse_loop(self, element_in: ET.Element):
        self.source += self.loop_delimiters[0]
        i = 0
        for node in element_in:
            if node.tag != GAZ_BLOCK_TAG:
                self.unparse_node(node)
                if self.strip_conditionals:
                    self.source += (self.loop_delimiters[1] + " ").strip()
                else:
                    self.source += (self.loop_delimiters[1] + " ")
            elif node.tag == GAZ_BLOCK_TAG:
                self.unparse_node(node)
            i += 1

    def unparse_unary(self, element_in: ET.Element):
        self.source += "{}".format(self.translate_op(element_in.get("op")))
        self.unparse_xhs(element_in.find(GAZ_RHS_TAG))

    def unparse_brackets(self, element_in: ET.Element):
        self.source += "("
        self.unparse_xhs(element_in.find(GAZ_RHS_TAG))
        self.source += ")"

    def unparse_break(self, element_in: ET.Element):
        self.source += "break" + self.endline

    def unparse_single_arg(self, param):
        return self.format_single_arg(self.translate_type(param.get(GAZ_TY_KEY)), param.get(GAZ_NAME_KEY))

    def unparse_routine_call(self, element_in: ET.Element):
        pass

    def unparse_statement(self, element_in: ET.Element):
        pass

    def unparse_top_block(self, element_in: ET.Element):
        pass

    def unparse_expression(self, element_in: ET.Element):
        pass

    def format_variable(self, mut, name, value, declaration: bool = False):
        raise NotImplementedError

    def translate_value(self, val):
        raise NotImplementedError

    def translate_op(self, param, ty=None):
        raise NotImplementedError

    def translate_type(self, ty):
        raise NotImplementedError

    def format_single_arg(self, param, param1):
        raise NotImplementedError

    def setup(self):
        raise NotImplementedError

    def pretty_xml(self):
        dom = xml.dom.minidom.parseString(ET.tostring(self.xml).decode('utf-8'))
        pretty: str = dom.toprettyxml()
        return pretty