Python core.StateMachine類代碼示例

class X:
    def __init__(self, Name):
        sh = StringIO("[:\\P{Script=%s}:]" % Name)
        self.name = Name
        self.charset = regex.snap_set_expression(sh, {})
        self.sm = StateMachine()
        self.sm.add_transition(self.sm.init_state_index, self.charset, AcceptanceF=True)
        self.id = self.sm.get_id()

    def check(self, SM):
        """This function throws an exception as soon as one single value
           is not matched according to the expectation.
        """
        print "Name = " + self.name, 
        for interval in self.charset.get_intervals(PromiseToTreatWellF=True):
            for i in range(interval.begin, interval.end):
                utf8_seq = unicode_to_utf8(i)

                # Apply sequence to state machine
                s_idx = result.init_state_index
                for byte in utf8_seq:
                    s_idx = result.states[s_idx].target_map.get_resulting_target_state_index(byte)

                # All acceptance flags must belong to the original state machine
                for cmd in result.states[s_idx].single_entry:
                    if cmd.__class__ != SeAccept: continue
                    # HERE: As soon as something is wrong --> fire an exception
                    assert cmd.acceptance_id() == self.id
        print " (OK=%i)" % self.id

def test_on_UCS_range(Trafo, Source, Drain, CharacterBackwardTrafo):

    sm     = StateMachine()
    acc_db = {}
    for x in range(Source.begin, Source.end):
        ti = sm.add_transition(sm.init_state_index, x, AcceptanceF=True)
        acc_id    = len(acc_db)
        sm.states[ti].mark_acceptance_id(acc_id)
        acc_db[x] = acc_id

    if Setup.bad_lexatom_detection_f:
        acc_db[None] = E_IncidenceIDs.BAD_LEXATOM
    else:
        acc_db[None] = None

    state_n_before, result = transform(Trafo, sm)
    # assert state_n_before == len(result.states)

    init_state = result.get_init_state()
    count      = 0
    for y in range(Drain.begin, Drain.end):
        # Translate character into 
        x  = CharacterBackwardTrafo(y)
        # Transit on the translated charater
        ti = init_state.target_map.get_resulting_target_state_index(y)
        # Compare resulting state with the expected state's acceptance
        assert_only_acceptance_id(sm.states, ti, acc_db, x, y)

        count += 1

    print "<terminated: %i transitions ok>" % count

def do(sh):
    """Converts a uni-code string into a state machine that parses 
       its letters sequentially. Each state in the sequence correponds
       to the sucessful triggering of a letter. Only the last state, though,
       is an acceptance state. Any bailing out before is 'not accepted'. 
       Example:

       "hey" is translated into the state machine:

           (0)-- 'h' -->(1)-- 'e' -->(2)-- 'y' --> ACCEPTANCE
            |            |            |
           FAIL         FAIL         FAIL
    
      Note: The state indices are globally unique. But, they are not necessarily
            0, 1, 2, ... 
    """
    assert     sh.__class__.__name__ == "StringIO" \
            or sh.__class__.__name__ == "file"

    # resulting state machine
    result    = StateMachine()
    state_idx = result.init_state_index

    # Only \" is a special character '"', any other backslashed character
    # remains as the sequence 'backslash' + character
    for char_code in get_character_code_sequence(sh):
        state_idx = result.add_transition(state_idx, char_code)

    # when the last state has trigger it is supposed to end up in 'acceptance'
    result.states[state_idx].set_acceptance()
    return result

def test(TestString):
    print "expression    = \"" + TestString + "\""
    sm = StateMachine()
    try:
        trigger_set = character_set.do(StringIO(TestString + "]"))
        sm.add_transition(sm.init_state_index, trigger_set, AcceptanceF=True)
        print "state machine\n", sm 
    except RegularExpressionException, x:
        print repr(x)

def snap_non_control_character(stream, PatternDict):
    __debug_entry("non-control characters", stream)

    # (*) read first character
    char_code = utf8.__read_one_utf8_code_from_stream(stream)
    if char_code is None:
        error_msg("Character could not be interpreted as UTF8 code or End of File reached prematurely.", 
                  stream)
    result = StateMachine()
    result.add_transition(result.init_state_index, char_code, AcceptanceF=True)
    return __debug_exit(result, stream)

def get_any():
    """RETURNS:

       A state machine that 'eats' any character, but only one. 

           (0)--- \Any --->(( 0 ))
    """
    result = StateMachine()
    result.add_transition(result.init_state_index, NumberSet(Interval(-sys.maxint, sys.maxint)), AcceptanceF=True)

    return result

def create_ALL_BUT_NEWLINE_state_machine():
    global Setup
    result = StateMachine()
    # NOTE: Buffer control characters are supposed to be filtered out by the code
    #       generator.
    trigger_set = NumberSet(Interval(ord("\n")).inverse()) 

    if Setup.get_character_value_limit() != sys.maxint:
        trigger_set.intersect_with(Interval(0, Setup.get_character_value_limit()))

    result.add_transition(result.init_state_index, trigger_set, AcceptanceF=True) 
    return result

def create_ALL_BUT_NEWLINE_state_machine(stream):
    global Setup
    result = StateMachine()
    # NOTE: Buffer control characters are supposed to be filtered out by the code
    #       generator.
    trigger_set = NumberSet(Interval(ord("\n"))).get_complement(Setup.buffer_codec.source_set)
    if trigger_set.is_empty():
        error_msg("The set of admissible characters contains only newline.\n"
                  "The '.' for 'all but newline' is an empty set.",
                  SourceRef.from_FileHandle(stream))

    result.add_transition(result.init_state_index, trigger_set, AcceptanceF=True) 
    return result

def test(ByteSequenceDB):

    L = len(ByteSequenceDB[0])

    for seq in ByteSequenceDB:
        assert len(seq) == L
        for x in seq:
            assert isinstance(x, Interval)

    first_different_byte_index = -1
    for i in range(L):
        x0 = ByteSequenceDB[0][i]
        for seq in ByteSequenceDB[1:]:
            if not seq[i].is_equal(x0): 
                first_different_byte_index = i
                break
        if first_different_byte_index != -1: 
            break
    if first_different_byte_index == -1:
        first_different_byte_index = 0

    print "# Best To be Displayed by:"
    print "#"
    print "#  > " + sys.argv[0] + " " + sys.argv[1] + " | dot -Tsvg -o tmp.svg"
    print "#"
    print "# -------------------------"
    print "# Byte Sequences:     "
    i = -1
    for seq in ByteSequenceDB:
        i += 1
        print "# (%i) " % i,
        for x in seq:
            print "    " + x.get_string(Option="hex"), 
        print
    print "#    L    = %i" % L
    print "#    DIdx = %i" % first_different_byte_index



    sm = StateMachine()
    end_index = state_machine.index.get()
    sm.states[end_index] = State()

    trafo.plug_state_sequence_for_trigger_set_sequence(sm, sm.init_state_index, end_index, 
                                                       ByteSequenceDB, L, first_different_byte_index)

    if len(sm.get_orphaned_state_index_list()) != 0:
        print "Error: Orphaned States Detected!"

    print sm.get_graphviz_string(Option="hex")

def get_setup(L0, L1, FSM0, FSM1, FSM2):
    # SPECIALITIES: -- sm0 and sm1 have an intersection between their second 
    #                  transition.
    #               -- sm1 transits further upon acceptance.
    #               -- sm2 has only one transition.
    ci_list = [
        CountInfo(dial_db.new_incidence_id(), NumberSet.from_range(L0, L1), 
                  CountAction(E_CharacterCountType.COLUMN, 0)),
    ]

    # Generate State Machine that does not have any intersection with 
    # the loop transitions.
    sm0 = StateMachine()
    si = sm0.add_transition(sm0.init_state_index, FSM0)
    si = sm0.add_transition(si, NS_A, AcceptanceF=True)
    sm0.states[si].mark_acceptance_id(dial_db.new_incidence_id())

    sm1 = StateMachine()
    si0 = sm1.add_transition(sm1.init_state_index, FSM1)
    si  = sm1.add_transition(si0, NS_A, AcceptanceF=True)
    iid1 = dial_db.new_incidence_id()
    sm1.states[si].mark_acceptance_id(iid1)
    si  = sm1.add_transition(si, NS_B, si0)
    sm1.states[si].mark_acceptance_id(iid1)

    sm2 = StateMachine()
    si = sm2.add_transition(sm2.init_state_index, FSM2, AcceptanceF=True)
    sm2.states[si].mark_acceptance_id(dial_db.new_incidence_id())

    return ci_list, [sm0, sm1, sm2]

def do(stream, PatternDict):
    trigger_set = snap_set_expression(stream, PatternDict)

    if trigger_set is None: 
        raise RegularExpressionException("Regular Expression: character_set_expression called for something\n" + \
                                         "that does not start with '[:', '[' or '\\P'")
    if trigger_set.is_empty():
        raise RegularExpressionException("Regular Expression: Character set expression results in empty set.")

    # Create state machine that triggers with the trigger set to SUCCESS
    # NOTE: The default for the ELSE transition is FAIL.
    sm = StateMachine()
    sm.add_transition(sm.init_state_index, trigger_set, AcceptanceF=True)

    return __debug_exit(sm, stream)

def create_range_skipper_code(Language, TestStr, CloserSequence, QuexBufferSize=1024, 
                              CommentTestStrF=False, ShowPositionF=False):
    assert QuexBufferSize >= len(CloserSequence) + 2

    end_str = __prepare(Language)

    door_id_on_skip_range_open = dial_db.new_door_id()

    data = { 
        "closer_sequence":    CloserSequence, 
        "closer_pattern":     Pattern(StateMachine.from_sequence(CloserSequence), 
                                      PatternString="<skip range closer>"),
        "mode_name":          "MrUnitTest",
        "on_skip_range_open": CodeFragment([end_str]),
        "door_id_after":      DoorID.continue_without_on_after_match(),
    }

    skipper_code = range_skipper.do(data, Analyzer)
    __require_variables()

    return create_customized_analyzer_function(Language, TestStr, skipper_code,
                                               QuexBufferSize, CommentTestStrF, ShowPositionF, end_str,
                                               MarkerCharList  = [], 
                                               LocalVariableDB = deepcopy(variable_db.get()),
                                               DoorIdOnSkipRangeOpen=door_id_on_skip_range_open) 

 def __init__(self, Name):
     sh = StringIO("[:\\P{Script=%s}:]" % Name)
     self.name = Name
     self.charset = regex.snap_set_expression(sh, {})
     self.sm = StateMachine()
     self.sm.add_transition(self.sm.init_state_index, self.charset, AcceptanceF=True)
     self.id = self.sm.get_id()

    def __sm_newline_default(self):
        """Default newline: '(\n)|(\r\n)'
        """
        global cc_type_name_db

        newline_set = NumberSet(ord('\n'))
        retour_set  = NumberSet(ord('\r'))

        before = self.specifier_count_op_map.find_occupier(newline_set, set())
        if before is not None:
            error.warning("Trying to implement default newline: '\\n' or '\\r\\n'.\n" 
                          "The '\\n' option is not possible, since it has been occupied by '%s'.\n" \
                          "No newline can be defined by default."
                          % cc_type_name_db[before.cc_type], before.sr, 
                          SuppressCode=NotificationDB.warning_default_newline_0A_impossible)
            # In this case, no newline can be defined!
            return

        sm = StateMachine.from_character_set(newline_set)

        if Setup.dos_carriage_return_newline_f:
            before = self.specifier_count_op_map.find_occupier(retour_set, set())
            if before is not None:
                error.warning("Trying to implement default newline: '\\n' or '\\r\\n'.\n" 
                          "The '\\r\\n' option is not possible, since '\\r' has been occupied by '%s'." \
                          % cc_type_name_db[before.cc_type],
                          before.sr, 
                          SuppressCode=NotificationDB.warning_default_newline_0D_impossible)
            else:
                sm.add_transition_sequence(sm.init_state_index, [retour_set, newline_set])

        return sm

def create_state_machine(SM, Result, Class_StateMachine, Class_State):
    # If all states are of size one, this means, that there were no states that
    # could have been combined. In this case a simple copy of the original
    # state machine will do.
    if len(filter(lambda state_set: len(state_set) != 1, Result.state_set_list)) == 0:
        return SM.clone()
    
    # Define a mapping from the state set to a new target state index
    #
    # map:  state_set_index  --->  index of the state that represents it
    #
    map_new_state_index = dict([(i, state_machine_index.get()) for i in xrange(len(Result.state_set_list))])
                
    # The state set that contains the initial state becomes the initial state of 
    # the new state machine.   
    state_set_containing_initial_state_i = Result.map[SM.init_state_index]
    new_init_state_index                 = map_new_state_index[state_set_containing_initial_state_i]

    result = StateMachine(new_init_state_index)

    # Ensure that each target state index has a state inside the state machine
    # Build up the state machine out of the state sets
    for state_set_idx, state_set in enumerate(Result.state_set_list):

        new_state_index = map_new_state_index[state_set_idx]

        # Merge all core information of the states inside the state set.
        # If one state set contains an acceptance state, then the result is 'acceptance'.
        # (Note: The initial split separates acceptance states from those that are not
        #  acceptance states. There can be no state set containing acceptance and 
        #  non-acceptance states) 
        # (Note, that the prototype's info has not been included yet, consider whole set)
        result.states[new_state_index] = Class_State.new_merged_core_state(SM.states[i] for i in state_set)

    for state_set_idx, state_set in enumerate(Result.state_set_list):
        # The prototype: States in one set behave all equivalent with respect to target state sets
        # thus only one state from the start set has to be considered.      
        prototype    = SM.states[state_set[0]]
        representative = result.states[map_new_state_index[state_set_idx]]

        # The representative must have all transitions that the prototype has
        for target_state_index, trigger_set in prototype.target_map.get_map().iteritems():
            target_state_set_index = Result.map[target_state_index]
            target_index           = map_new_state_index[target_state_set_index]
            representative.add_transition(trigger_set, target_index)

    return result