def test_readme():
"""The basic examples in the README"""
# Basic DFA that accepts ab+
E = cpppo.state( "E" )
A = cpppo.state_input( "A" )
B = cpppo.state_input( "B", terminal=True )
E['a'] = A
A['b'] = B
B['b'] = B
data = cpppo.dotdict()
source = cpppo.peekable( str( 'abbbb,ab' ))
with cpppo.dfa( initial=E ) as abplus:
for i,(m,s) in enumerate( abplus.run( source=source, path="ab+", data=data )):
log.info( "%s #%3d -> %10.10s; next byte %3d: %-10.10r: %r", m.name_centered(),
i, s, source.sent, source.peek(), data )
assert i == 5
assert source.peek() == str(',')
# Composite state machine accepting ab+, ignoring ,[ ]* separators
CSV = cpppo.dfa( "CSV", initial=E, terminal=True )
SEP = cpppo.state_drop( "SEP" )
CSV[','] = SEP
SEP[' '] = SEP
SEP[None] = CSV
source = cpppo.peekable( str( 'abbbb, ab' ))
with cpppo.dfa( initial=CSV ) as r2:
for i,(m,s) in enumerate( r2.run( source=source, path="readme_CSV", data=data )):
log.info( "%s #%3d -> %10.10s; next byte %3d: %-10.10r: %r", m.name_centered(),
i, s, source.sent, source.peek(), data )
assert i == 14
assert source.peek() is None
def main():
"""The basic examples in the README"""
# Basic DFA that accepts ab+
E = cpppo.state( 'E' )
A = cpppo.state_input( 'A' )
B = cpppo.state_input( 'B', terminal=True )
E['a'] = A
A['b'] = B
B['b'] = B
BASIC = cpppo.dfa( 'ab+', initial=E, context='basic' )
# Composite state machine accepting ab+, ignoring ,[ ]* separators
ABP = cpppo.dfa( 'ab+', initial=E, terminal=True )
SEP = cpppo.state_drop( 'SEP' )
ABP[','] = SEP
SEP[' '] = SEP
SEP[None] = ABP
CSV = cpppo.dfa( 'CSV', initial=ABP, context='csv' )
# A regular expression; he default dfa name is the regular expression itself.
REGEX = cpppo.regex( initial='(ab+)((,[ ]*)(ab+))*', context='regex' )
data = cpppo.dotdict()
for machine in [ BASIC, CSV, REGEX ]:
path = machine.context() + '.input' # default for state_input data
source = cpppo.peekable( str( 'abbbb, ab' ))
with machine:
for i,(m,s) in enumerate( machine.run( source=source, data=data )):
print( "%s #%3d; next byte %3d: %-10.10r: %r" % (
m.name_centered(), i, source.sent, source.peek(), data.get(path) ))
print( "Accepted: %r; remaining: %r\n" % ( data.get(path), ''.join( source )))
print( "Final: %r" % ( data ))
def tnet_machine( name="TNET", context="tnet" ):
"""Accept a sentence of input bytes matching a tnetstring, and then
loop. Sub-machine terminates at earliest match (non-greedy), causing
echo.transition to trigger .process (which resets our sub-machine to initial
state), and then we move to the next state (loops), allowing us to
immediately run."""
class tnet_parser( cpppo.state_input ):
TYPES = (b'#'[0], b'}'[0], b']'[0], b','[0],
b'$'[0], b'!'[0], b'~'[0], b'^'[0])
def process( self, source, machine=None, path=None, data=None ):
"""Convert the collected data according to the type"""
tntype = next( source )
ours = self.context( path )
raw = ours + '...data.input'
src = b'' if raw not in data else (
data[raw].tostring() if sys.version_info[0] < 3
else data[raw].tobytes() )
if tntype == b','[0]:
log.info("%5d bytes data: %s", len( src ), cpppo.reprlib.repr( src ))
data[ours] = src
elif tntype == b'$'[0]:
log.info("%5d string data: %s", len( src ), cpppo.reprlib.repr( src ))
data[ours] = src.decode( 'utf-8' )
elif tntype == b'#'[0]:
data[ours] = int( src )
log.info("%5d int data: %s == %s", len( src ), cpppo.reprlib.repr( src ),
cpppo.reprlib.repr( data[ours] ))
elif tntype == b'~'[0]:
assert 0 == len( src )
data[ours] = None
else:
assert False, "Invalid tnetstring type: %s" % tntype
bytes_conf = {
"alphabet": cpppo.type_bytes_iter,
"typecode": cpppo.type_bytes_array_symbol,
}
SIZE = cpppo.dfa( name="SIZE",
initial=cpppo.integer_bytes(
name="INT", context="size", decode='ascii', terminal=True ))
COLON = cpppo.state_drop( name="COLON", **bytes_conf )
DATA = data_parser( name="DATA", context="data", repeat="..size" )
TYPE = tnet_parser( name="TYPE", context="type", terminal=True,
**bytes_conf )
SIZE[b':'[0]] = COLON
COLON[None] = DATA
for t in tnet_parser.TYPES:
DATA[t] = TYPE
# Recognize a TNET string and then terminate, resetting to automatically
# recognize another
return cpppo.dfa( name=name, context=context, initial=SIZE, terminal=True )
def tnet_machine( name="TNET", context="tnet" ):
"""Accept a sentence of input bytes matching a tnetstring, and then
loop. Sub-machine terminates at earliest match (non-greedy), causing
echo.transition to trigger .process (which resets our sub-machine to initial
state), and then we move to the next state (loops), allowing us to
immediately run."""
class tnet_parser( cpppo.state_input ):
TYPES = (b'#'[0], b'}'[0], b']'[0], b','[0],
b'$'[0], b'!'[0], b'~'[0], b'^'[0])
def process( self, source, machine=None, path=None, data=None ):
"""Convert the collected data according to the type"""
tntype = next( source )
ours = self.context( path )
raw = ours + '...data.input'
src = ( data[raw].tostring() if sys.version_info[0] < 3
else data[raw].tobytes() )
if tntype == b','[0]:
log.info("%5d bytes data: %s", len( src ), cpppo.reprlib.repr( src ))
data[ours] = src
elif tntype == b'$'[0]:
log.info("%5d string data: %s", len( src ), cpppo.reprlib.repr( src ))
data[ours] = src.decode( 'utf-8' )
elif tntype == b'#'[0]:
data[ours] = int( src )
log.info("%5d int data: %s == %s", len( src ), cpppo.reprlib.repr( src ),
cpppo.reprlib.repr( data[ours] ))
elif tntype == b'~'[0]:
assert 0 == len( src )
data[ours] = None
else:
assert False, "Invalid tnetstring type: %s" % tntype
bytes_conf = {
"alphabet": cpppo.type_bytes_iter,
"typecode": cpppo.type_bytes_array_symbol,
}
SIZE = cpppo.dfa( name="SIZE",
initial=cpppo.integer_bytes(
name="INT", context="size", decode='ascii', terminal=True ))
COLON = cpppo.state_drop( name="COLON", **bytes_conf )
DATA = data_parser( name="DATA", context="data", repeat="..size" )
TYPE = tnet_parser( name="TYPE", context="type", terminal=True,
**bytes_conf )
SIZE[b':'[0]] = COLON
COLON[None] = DATA
for t in tnet_parser.TYPES:
DATA[t] = TYPE
# Recognize a TNET string and then terminate, resetting to automatically
# recognize another
return cpppo.dfa( name=name, context=context, initial=SIZE, terminal=True )
def test_decide():
"""Allow state transition decisions based on collected context other than just
the next source symbol.
"""
e = cpppo.state("enter")
e["a"] = a = cpppo.state_input("a", context="a")
a[" "] = s1 = cpppo.state_drop("s1")
s1[" "] = s1
s1[None] = i1 = cpppo.integer("i1", context="i1")
i1[" "] = s2 = cpppo.state_drop("s2")
s2[" "] = s2
s2[None] = i2 = cpppo.integer("i2", context="i2")
less = cpppo.state("less", terminal=True)
greater = cpppo.state("greater", terminal=True)
equal = cpppo.state("equal", terminal=True)
i2[None] = cpppo.decide("isless", less, predicate=lambda machine, source, path, data: data.i1 < data.i2)
i2[None] = cpppo.decide("isgreater", greater, predicate=lambda machine, source, path, data: data.i1 > data.i2)
i2[None] = equal
source = cpppo.peekable(str("a 1 2"))
data = cpppo.dotdict()
with cpppo.dfa("comparo", initial=e) as comparo:
for i, (m, s) in enumerate(comparo.run(source=source, data=data)):
log.info(
"%s #%3d -> %10.10s; next byte %3d: %-10.10r: %r",
m.name_centered(),
i,
s,
source.sent,
source.peek(),
data,
)
assert i == 11
assert s is less
source = cpppo.peekable(str("a 33 33"))
data = cpppo.dotdict()
with cpppo.dfa("comparo", initial=e) as comparo:
for i, (m, s) in enumerate(comparo.run(source=source, data=data)):
log.info(
"%s #%3d -> %10.10s; next byte %3d: %-10.10r: %r",
m.name_centered(),
i,
s,
source.sent,
source.peek(),
data,
)
assert i == 13
assert s is equal
def __init__( self, name=None, **kwds ):
name = name or kwds.setdefault( 'context', self.__class__.__name__ )
# Parse the status, and status_ext.size
stat = USINT( 'status', context=None )
stat[True] = size = USINT( '_ext.size', extension='_ext.size' )
# Prepare a state-machine to parse each UINT into .UINT, and move it onto the .data list
exts = UINT( 'ext_status', extension='.ext_status' )
exts[None] = move_if( 'data', source='.ext_status',
destination='.data', initializer=lambda **kwds: [] )
exts[None] = cpppo.state( 'done', terminal=True )
# Parse each status_ext.data in a sub-dfa, repeating status_ext.size times
each = cpppo.dfa( 'each', extension='_ext',
initial=exts, repeat='_ext.size',
terminal=True )
# Only enter the state_ext.data dfa if status_ext.size is non-zero
size[None] = cpppo.decide( '_ext.size',
predicate=lambda path=None, data=None, **kwds: data[path+'_ext.size'],
state=each )
# Otherwise, we're done!
size[None] = octets_noop( 'done',
terminal=True )
super( status, self ).__init__( name=name, initial=stat, **kwds )
def data_parser( **kwds ):
"""Parses raw bytes into .data, by default using ..size to denote the amount. """
kwds.setdefault( "name", "DATA" )
kwds.setdefault( "context", "data" )
kwds.setdefault( "repeat", "..size" )
return cpppo.dfa(
initial=cpppo.state_input( name="BYTE", terminal=True, **bytes_conf ), **kwds )
def data_parser( **kwds ):
"""Parses raw bytes into .data, by default using ..size to denote the amount. """
kwds.setdefault( "name", "DATA" )
kwds.setdefault( "context", "data" )
kwds.setdefault( "repeat", "..size" )
return cpppo.dfa(
initial=cpppo.state_input( name="BYTE", terminal=True, **bytes_conf ), **kwds )
def __init__(self, name=None, **kwds):
name = name or kwds.setdefault('context', self.__class__.__name__)
# Parse the status, and status_ext.size
stat = USINT('status', context=None)
stat[True] = size = USINT('_ext.size', extension='_ext.size')
# Prepare a state-machine to parse each UINT into .UINT, and move it onto the .data list
exts = UINT('ext_status', extension='.ext_status')
exts[None] = move_if('data',
source='.ext_status',
destination='.data',
initializer=lambda **kwds: [])
exts[None] = cpppo.state('done', terminal=True)
# Parse each status_ext.data in a sub-dfa, repeating status_ext.size times
each = cpppo.dfa('each',
extension='_ext',
initial=exts,
repeat='_ext.size',
terminal=True)
# Only enter the state_ext.data dfa if status_ext.size is non-zero
size[None] = cpppo.decide('_ext.size',
predicate=lambda path=None, data=None, **
kwds: data[path + '_ext.size'],
state=each)
# Otherwise, we're done!
size[None] = octets_noop('done', terminal=True)
super(status, self).__init__(name=name, initial=stat, **kwds)
def __init__(self, name=None, **kwds):
"""Parse CPF list items 'til .count reached, which should be simultaneous with symbol exhaustion, if
caller specified a symbol limit.
"""
name = name or kwds.setdefault('context', self.__class__.__name__)
# A number, and then each CPF item consistes of a type, length and then parsable data.
ityp = UINT(context='type_id')
ityp[True] = ilen = UINT(context='length')
ilen[None] = cpppo.decide('empty',
predicate=lambda path=None, data=None, **
kwds: not data[path].length,
state=octets_noop('done', terminal=True))
# Prepare a parser for each recognized CPF item type. It must establish one level of
# context, because we need to pass it a limit='..length' denoting the length we just parsed.
for typ, cls in (self.item_parsers or {}).items():
ilen[None] = cpppo.decide(cls.__name__,
state=cls(terminal=True,
limit='..length'),
predicate=lambda path=None, data=None, **
kwds: data[path].type_id == typ)
# If we don't recognize the CPF item type, just parse remainder into .input (so we could re-generate)
ilen[None] = urec = octets('unrecognized', context=None, terminal=True)
urec[True] = urec
# Each item is collected into '.item__', 'til no more input available, and then moved into
# place into '.item' (init to [])
item = cpppo.dfa('each', context='item__', initial=ityp)
item[None] = move_if('move',
source='.item__',
destination='.item',
initializer=lambda **kwds: [])
item[None] = cpppo.state('done', terminal=True)
# Parse count, and then exactly .count CPF items.
loop = UINT(context='count')
loop[None] = cpppo.dfa('all',
initial=item,
repeat='.count',
terminal=True)
super(CPF, self).__init__(name=name, initial=loop, **kwds)
def test_decide():
"""Allow state transition decisions based on collected context other than just
the next source symbol.
"""
e = cpppo.state("enter")
e['a'] = a = cpppo.state_input("a", context='a')
a[' '] = s1 = cpppo.state_drop("s1")
s1[' '] = s1
s1[None] = i1 = cpppo.integer("i1", context='i1')
i1[' '] = s2 = cpppo.state_drop("s2")
s2[' '] = s2
s2[None] = i2 = cpppo.integer("i2", context='i2')
less = cpppo.state("less", terminal=True)
greater = cpppo.state("greater", terminal=True)
equal = cpppo.state("equal", terminal=True)
i2[None] = cpppo.decide(
"isless",
less,
predicate=lambda machine, source, path, data: data.i1 < data.i2)
i2[None] = cpppo.decide(
"isgreater",
greater,
predicate=lambda machine, source, path, data: data.i1 > data.i2)
i2[None] = equal
source = cpppo.peekable(str('a 1 2'))
data = cpppo.dotdict()
with cpppo.dfa("comparo", initial=e) as comparo:
for i, (m, s) in enumerate(comparo.run(source=source, data=data)):
log.info("%s #%3d -> %10.10s; next byte %3d: %-10.10r: %r",
m.name_centered(), i, s, source.sent, source.peek(), data)
assert i == 12
assert s is less
source = cpppo.peekable(str('a 33 33'))
data = cpppo.dotdict()
with cpppo.dfa("comparo", initial=e) as comparo:
for i, (m, s) in enumerate(comparo.run(source=source, data=data)):
log.info("%s #%3d -> %10.10s; next byte %3d: %-10.10r: %r",
m.name_centered(), i, s, source.sent, source.peek(), data)
assert i == 14
assert s is equal
def register_service_parser( cls, number, name, short, machine ):
"""Registers a parser with the Object. May be invoked during import; no logging."""
assert number not in cls.service and name not in cls.service, \
"Duplicate service #%d: %r registered for Object %s" % ( number, name, cls.__name__ )
cls.service[number] = name
cls.service[name] = number
cls.transit[number] = chr( number ) if sys.version_info.major < 3 else number
cls.parser.initial[cls.transit[number]] \
= cpppo.dfa( name=short, initial=machine, terminal=True )
def __init__( self, name=None, **kwds ):
"""Parse CPF list items 'til .count reached, which should be simultaneous with symbol exhaustion, if
caller specified a symbol limit.
"""
name = name or kwds.setdefault( 'context', self.__class__.__name__ )
# A number, and then each CPF item consistes of a type, length and then parsable data.
ityp = UINT( context='type_id' )
ityp[True] = ilen = UINT( context='length' )
ilen[None] = cpppo.decide( 'empty',
predicate=lambda path=None, data=None, **kwds: not data[path].length,
state=octets_noop( 'done', terminal=True ))
# Prepare a parser for each recognized CPF item type. It must establish one level of
# context, because we need to pass it a limit='..length' denoting the length we just parsed.
for typ,cls in ( self.item_parsers or {} ).items():
ilen[None] = cpppo.decide( cls.__name__, state=cls( terminal=True, limit='..length' ),
predicate=lambda path=None, data=None, **kwds: data[path].type_id == typ )
# If we don't recognize the CPF item type, just parse remainder into .input (so we could re-generate)
ilen[None] = urec = octets( 'unrecognized', context=None,
terminal=True )
urec[True] = urec
# Each item is collected into '.item__', 'til no more input available, and then moved into
# place into '.item' (init to [])
item = cpppo.dfa( 'each', context='item__',
initial=ityp )
item[None] = move_if( 'move', source='.item__',
destination='.item', initializer=lambda **kwds: [] )
item[None] = cpppo.state( 'done', terminal=True )
# Parse count, and then exactly .count CPF items.
loop = UINT( context='count' )
loop[None] = cpppo.dfa( 'all',
initial=item, repeat='.count',
terminal=True )
super( CPF, self ).__init__( name=name, initial=loop, **kwds )
def register_service_parser(cls, number, name, short, machine):
"""Registers a parser with the Object. May be invoked during import; no logging."""
assert number not in cls.service and name not in cls.service, \
"Duplicate service #%d: %r registered for Object %s" % ( number, name, cls.__name__ )
cls.service[number] = name
cls.service[name] = number
cls.transit[number] = chr(
number) if sys.version_info.major < 3 else number
cls.parser.initial[cls.transit[number]] \
= cpppo.dfa( name=short, initial=machine, terminal=True )
def test_struct():
dtp = cpppo.type_bytes_array_symbol
abt = cpppo.type_bytes_iter
ctx = 'val'
a = cpppo.state_input( "First", alphabet=abt, typecode=dtp, context=ctx )
a[True] = b = cpppo.state_input( "Second", alphabet=abt, typecode=dtp, context=ctx )
b[True] = c = cpppo.state_input( "Third", alphabet=abt, typecode=dtp, context=ctx )
c[True] = d = cpppo.state_input( "Fourth", alphabet=abt, typecode=dtp, context=ctx )
d[None] = cpppo.state_struct( "int32", context=ctx,
format=str("<i"),
terminal=True )
machine = cpppo.dfa( initial=a )
with machine:
material = b'\x01\x02\x03\x80\x99'
segment = 3
source = cpppo.chainable()
log.info( "States; %r input, by %d", material, segment )
inp = None
data = cpppo.dotdict()
path = "struct"
sequence = machine.run( source=source, path=path, data=data )
for num in range( 10 ):
try:
mch,sta = next( sequence )
inp = source.peek()
except StopIteration:
inp = source.peek()
log.info( "%s <- %-10.10r test done", cpppo.centeraxis( mch, 25, clip=True ), inp )
break
log.info( "%s <- %-10.10r test rcvd", cpppo.centeraxis( mch, 25, clip=True ), inp )
if sta is None:
log.info( "%s <- %-10.10r test no next state", cpppo.centeraxis( mch, 25, clip=True ), inp )
if inp is None:
if not material:
log.info( "%s <- %-10.10r test source finished", cpppo.centeraxis( mch, 25, clip=True ), inp )
# Will load consecutive empty iterables; chainable must handle
source.chain( material[:segment] )
material = material[segment:]
inp = source.peek()
log.info( "%s <- %-10.10r test chain", cpppo.centeraxis( mch, 25, clip=True ), inp )
if num == 0: assert inp == b'\x01'[0]; assert sta.name == "First"
if num == 1: assert inp == b'\x02'[0]; assert sta.name == "Second"
if num == 2: assert inp == b'\x03'[0]; assert sta.name == "Third"
if num == 3: assert inp == b'\x80'[0]; assert sta is None
if num == 4: assert inp == b'\x80'[0]; assert sta.name == "Fourth"
if num == 5: assert inp == b'\x99'[0]; assert sta.name == "int32"
if num == 6: assert inp == b'\x99'[0]; assert sta.name == "int32"
assert inp == b'\x99'[0]
assert num == 6
assert sta.name == "int32"
assert data.struct.val == -2147286527
def main():
"""The basic examples in the README"""
# Basic DFA that accepts ab+
E = cpppo.state('E')
A = cpppo.state_input('A')
B = cpppo.state_input('B', terminal=True)
E['a'] = A
A['b'] = B
B['b'] = B
BASIC = cpppo.dfa('ab+', initial=E, context='basic')
# Composite state machine accepting ab+, ignoring ,[ ]* separators
ABP = cpppo.dfa('ab+', initial=E, terminal=True)
SEP = cpppo.state_drop('SEP')
ABP[','] = SEP
SEP[' '] = SEP
SEP[None] = ABP
CSV = cpppo.dfa('CSV', initial=ABP, context='csv')
# A regular expression; he default dfa name is the regular expression itself.
REGEX = cpppo.regex(initial='(ab+)((,[ ]*)(ab+))*', context='regex')
data = cpppo.dotdict()
for machine in [BASIC, CSV, REGEX]:
path = machine.context() + '.input' # default for state_input data
source = cpppo.peekable(str('abbbb, ab'))
with machine:
for i, (m, s) in enumerate(machine.run(source=source, data=data)):
print("%s #%3d; next byte %3d: %-10.10r: %r" %
(m.name_centered(), i, source.sent, source.peek(),
data.get(path)))
print("Accepted: %r; remaining: %r\n" %
(data.get(path), ''.join(source)))
print("Final: %r" % (data))
def test_dfa():
# Simple DFA with states consuming no input. A NULL (None) state transition
# doesn't require input for state change. The Default (True) transition
# requires input to make the transition, but none of these states consume
# it, so it'll be left over at the end.
a = cpppo.state("Initial")
a[None] = b = cpppo.state("Middle")
b[True] = cpppo.state("Terminal", terminal=True)
source = cpppo.chainable()
i = a.run(source=source)
m, s = next(i)
assert m is None
assert s is not None and s.name == "Middle"
try:
next(i)
assert False, "Expected no more non-transition events"
except StopIteration:
pass
machine = cpppo.dfa(initial=a)
with machine:
log.info("DFA:")
for initial in machine.initial.nodes():
for inp, target in initial.edges():
log.info(
"%s <- %-10.10r -> %s" %
(cpppo.centeraxis(initial, 25, clip=True), inp, target))
# Running with no input will yield the initial state, with None input; since it is a NULL
# state (no input processed), it will simply attempt to transition. This will require the
# next input from source, which is empty, so it will return input,state=(None, None)
# indicating a non-terminal state and no input left. This gives the caller an opportunity
# to reload input and try again. If a loop is detected (same state and input conditions
# seen repeatedly), the DFA will terminate; if not in a terminal state, an exception will be
# raised.
log.info("States; No input")
source = cpppo.chainable()
sequence = machine.run(source=source)
for num in range(10):
try:
mch, sta = next(sequence)
except StopIteration:
sequence = None
break
except cpppo.NonTerminal as e:
assert "non-terminal state" in str(e)
break
inp = source.peek()
log.info("%s <- %r" % (cpppo.centeraxis(mch, 25, clip=True), inp))
if num == 0:
assert inp is None
assert sta.name == "Initial"
if num == 1:
assert inp is None
assert sta.name == "Middle"
if num == 2:
assert inp is None
assert sta is None # And no more no-input transitions
assert num < 3 # If we get here, we didn't detect loop
assert num == 3
# since the iterator did not stop cleanly (after processing a state's input,
# and then trying to determine the next state), it'll continue indefinitely
assert sta is None
assert sequence is not None
# Try with some input loaded into source stream, using an identical generator sequence.
# Only the first element is gotten, and is reused for every NULL state transition, and is
# left over at the end.
log.info("States; 'abc' input")
assert source.peek() is None
source.chain(b'abc')
assert source.peek() == b'a'[0] # python2: str, python3: int
sequence = machine.run(source=source)
for num in range(10):
try:
mch, sta = next(sequence)
except StopIteration:
break
inp = source.peek()
log.info("%s <- %r", cpppo.centeraxis(mch, 25, clip=True), inp)
if num == 0:
assert inp == b'a'[0]
assert sta.name == "Initial"
if num == 1:
assert inp == b'a'[0]
assert sta.name == "Middle"
if num == 2:
assert inp == b'a'[0]
assert sta.name == "Terminal"
assert num < 3
assert num == 3
assert inp == b'a'[0]
assert sta.name == "Terminal"
def test_state():
"""A state is expected to process its input (perhaps nothing, if its a no-input state), and then use
the next input symbol to transition to another state. Each state has a context into a data
artifact, into which it will collect its results.
We must ensure that all state transitions are configured in the target alphabet; if an encoder
is supplied, then all input symbols and all transition symbols will be encoded using it. In
this test, all string literals are Unicode (in both Python 2 and 3), so we use a unicode encoder
to convert them to symbols."""
unicodekwds = {
'alphabet': unicode if sys.version_info[0] < 3 else str,
'encoder': cpppo.type_unicode_encoder,
}
s1 = cpppo.state('one', **unicodekwds)
s2 = cpppo.state_drop('two', **unicodekwds)
s1['a'] = s2
assert s1['a'] is s2
source = cpppo.peeking('abc')
# We can run state instances with/without acquisition
g = s1.run(source=source)
assert next(g) == (None, s2)
assert source.peek() == 'a'
with pytest.raises(StopIteration):
next(g)
with s1:
g = s1.run(source=source)
assert source.peek() == 'a'
assert next(g) == (None, s2)
assert source.peek() == 'a'
try:
next(g)
assert False, "Should have terminated"
except StopIteration:
pass
assert source.peek() == 'a'
# A state machine accepting a sequence of unicode a's
a_s = cpppo.state("a_s", **unicodekwds)
an_a = cpppo.state_input("a",
terminal=True,
typecode=cpppo.type_unicode_array_symbol,
**unicodekwds)
a_s['a'] = an_a
an_a['a'] = an_a
source = cpppo.peeking('aaaa')
data = cpppo.dotdict()
with cpppo.dfa(initial=a_s) as aplus:
for i, (m, s) in enumerate(aplus.run(source=source)):
log.info("%s #%3d -> %10.10s; next byte %3d: %-10.10r: %r",
m.name_centered(), i, s, source.sent, source.peek(), data)
assert i == 5
assert source.peek() is None
assert len(data) == 0
# Accepting a's separated by comma and space/pi (for kicks). When the lower level a's machine
# doesn't recognize the symbol, then the higher level machine will recognize and discard
sep = cpppo.state_drop("sep", **unicodekwds)
csv = cpppo.dfa("csv", initial=a_s, terminal=True, **unicodekwds)
csv[','] = sep
sep[' '] = sep
sep['π'] = sep
sep[None] = csv
source = cpppo.peeking('aaaa, a,π a')
data = cpppo.dotdict()
with cpppo.dfa(initial=csv) as csvaplus:
for i, (m, s) in enumerate(
csvaplus.run(source=source, path="csv", data=data)):
log.info("%s #%3d -> %10.10s; next byte %3d: %-10.10r: %r",
m.name_centered(), i, s, source.sent, source.peek(), data)
assert i == 18
assert source.peek() is None
assert data.csv.input.tounicode() == 'aaaaaa'
def __init__(self, name=None, **kwds):
name = name or kwds.setdefault('context', self.__class__.__name__)
# Get the size, and chain remaining machine onto rest. When used as a Route Path, the size
# is padded, so insert a state to drop the pad, and chain rest to that instead.
size = rest = USINT(context='size')
if self.padsize:
size[True] = rest = octets_drop('pad', repeat=1)
# After capturing each segment__ (pseg), move it onto the path segment list, and loop
pseg = octets_noop('type', terminal=True)
# ...segment parsers...
pmov = move_if('move',
initializer=lambda **kwds: [],
source='..segment__',
destination='..segment',
state=pseg)
# Wire each different segment type parser between pseg and pmov
pseg[b'\x28'[0]] = e_8t = octets_drop('type', repeat=1)
e_8t[True] = e_8v = USINT('elem_8bit', context='element')
e_8v[None] = pmov
pseg[b'\x29'[0]] = e16t = octets_drop('type', repeat=2)
e16t[True] = e16v = UINT('elem16bit', context='element')
e16v[None] = pmov
pseg[b'\x2a'[0]] = e32t = octets_drop('type', repeat=2)
e32t[True] = e32v = UDINT('elem32bit', context='element')
e32v[None] = pmov
pseg[b'\x20'[0]] = c_8t = octets_drop('type', repeat=1)
c_8t[True] = c_8v = USINT('clas_8bit', context='class')
c_8v[None] = pmov
pseg[b'\x21'[0]] = c16t = octets_drop('type', repeat=2)
c16t[True] = c16v = UINT('clas16bit', context='class')
c16v[None] = pmov
pseg[b'\x24'[0]] = i_8t = octets_drop('type', repeat=1)
i_8t[True] = i_8v = USINT('inst_8bit', context='instance')
i_8v[None] = pmov
pseg[b'\x25'[0]] = i16t = octets_drop('type', repeat=2)
i16t[True] = i16v = UINT('inst16bit', context='instance')
i16v[None] = pmov
pseg[b'\x30'[0]] = a_8t = octets_drop('type', repeat=1)
a_8t[True] = a_8v = USINT('attr_8bit', context='attribute')
a_8v[None] = pmov
pseg[b'\x31'[0]] = a16t = octets_drop('type', repeat=2)
a16t[True] = a16v = UINT('attr16bit', context='attribute')
a16v[None] = pmov
pseg[b'\x91'[0]] = symt = octets_drop('type', repeat=1)
symt[True] = syml = USINT('sym_len', context='symbolic.length')
syml[None] = symv = cpppo.string_bytes('symbolic',
context='symbolic',
limit='.length',
initial='.*',
decode='iso-8859-1')
# An odd-length ANSI Extended Symbolic name means an odd total. Pad
symo = octets_drop('pad', repeat=1)
symo[None] = pmov
symv[None] = cpppo.decide('odd',
predicate=lambda path=None, data=None, **
kwds: len(data[path].symbolic) % 2,
state=symo)
symv[None] = pmov
# Route Path port/link-address. See Vol 1-3.13, Table C-1.3 Port Segment Encoding.
# segment: 0b000spppp
# |\\\\+-> port number 0x01-0x0E; 0x0F=>extended
# |
# +------> link size+address; 0=>numeric, 1=>size+string
#
def port_fix(path=None, data=None, **kwds):
"""Discard port values about 0x0F; return True (transition) if remaining port value is 0x0F
(Optional Extended port)"""
data[path].port &= 0x0F
if data[path].port == 0x0F:
# Port is extended; discard and prepare to collect new port number
data[path].port = cpppo.dotdict()
return True
# Port is OK; don't transition
return False
# [01-0E][LL] port 01-0E, link-address #LL
pseg[b'\x01'[0]] = pnum = USINT('port_num', context='port')
pseg[b'\x02'[0]] = pnum
pseg[b'\x03'[0]] = pnum
pseg[b'\x04'[0]] = pnum
pseg[b'\x05'[0]] = pnum
pseg[b'\x06'[0]] = pnum
pseg[b'\x07'[0]] = pnum
pseg[b'\x08'[0]] = pnum
pseg[b'\x09'[0]] = pnum
pseg[b'\x0a'[0]] = pnum
pseg[b'\x0b'[0]] = pnum
pseg[b'\x0c'[0]] = pnum
pseg[b'\x0d'[0]] = pnum
pseg[b'\x0e'[0]] = pnum
# [0F][PPPP][LL] port 0xPPPP, link-address 0xLL
pseg[b'\x0f'[0]] = pnum
# A big port#; re-scan a UINT into .port (won't work 'til port_fix is called)
pnbg = UINT('port_nbg', context='port')
pnbg[True] = pnlk = USINT('link_num', context='link')
# Fix the port#; if 0x0F, setup for extended port and transition to pnbg. Otherwise,
# (not extended port), just go the the port numeric link.
pnum[None] = cpppo.decide('port_nfix', predicate=port_fix, state=pnbg)
pnum[None] = pnlk
pnlk[None] = pmov # and done; move segment, get next
# [11-1E][SS]'123.123.123.123'[00] port 0x01-0E, link address '123.123.123.123' (pad if size 0xSS odd)
pseg[b'\x11'] = padr = USINT('port_adr', context='port')
pseg[b'\x12'[0]] = padr
pseg[b'\x13'[0]] = padr
pseg[b'\x14'[0]] = padr
pseg[b'\x15'[0]] = padr
pseg[b'\x16'[0]] = padr
pseg[b'\x17'[0]] = padr
pseg[b'\x18'[0]] = padr
pseg[b'\x19'[0]] = padr
pseg[b'\x1a'[0]] = padr
pseg[b'\x1b'[0]] = padr
pseg[b'\x1c'[0]] = padr
pseg[b'\x1d'[0]] = padr
pseg[b'\x1e'[0]] = padr
# [1F][SS][PPPP]'123.123.123.123'[00] port 0xPPPP, link address '123.123.123.123' (pad if size SS odd)
pseg[b'\x1f'[0]] = padr
# Harvest the addresses into .link
adrv = cpppo.string_bytes('link_add',
context='link',
limit='.length',
initial='.*',
decode='iso-8859-1')
# An odd-length link address means an odd total. Pad
adro = octets_drop('link_pad', repeat=1)
adro[None] = pmov
adrv[None] = cpppo.decide('link_odd',
predicate=lambda path=None, data=None, **
kwds: len(data[path + '.link']) % 2,
state=adro)
adrv[None] = pmov
# A big port#; re-scan a UINT into .port (won't work 'til port_fix is called)
pabg = UINT('port_abg', context='port')
pabg[None] = adrv
#
padr[True] = adrl = USINT('link_len', context='link.length')
adrl[None] = cpppo.decide('port_afix', predicate=port_fix, state=pabg)
adrl[None] = adrv
# Parse all segments in a sub-dfa limited by the parsed path.size (in words; double)
rest[None] = cpppo.dfa('each',
context='segment__',
initial=pseg,
terminal=True,
limit=lambda path=None, data=None, **kwds: data[
path + '..size'] * 2)
super(EPATH, self).__init__(name=name, initial=size, **kwds)
def test_dfa():
# Simple DFA with states consuming no input. A NULL (None) state transition
# doesn't require input for state change. The Default (True) transition
# requires input to make the transition, but none of these states consume
# it, so it'll be left over at the end.
a = cpppo.state( "Initial" )
a[None] = b = cpppo.state( "Middle" )
b[True] = cpppo.state( "Terminal", terminal=True )
source = cpppo.chainable()
i = a.run( source=source )
m,s = next( i )
assert m is None
assert s is not None and s.name == "Middle"
try:
next( i )
assert False, "Expected no more non-transition events"
except StopIteration:
pass
machine = cpppo.dfa( initial=a )
with machine:
log.info( "DFA:" )
for initial in machine.initial.nodes():
for inp,target in initial.edges():
log.info( "%s <- %-10.10r -> %s" % ( cpppo.centeraxis( initial, 25, clip=True ),
inp, target ))
# Running with no input will yield the initial state, with None input; since it is a NULL
# state (no input processed), it will simply attempt to transition. This will require the
# next input from source, which is empty, so it will return input,state=(None, None)
# indicating a non-terminal state and no input left. This gives the caller an opportunity
# to reload input and try again. If a loop is detected (same state and input conditions
# seen repeatedly), the DFA will terminate; if not in a terminal state, an exception will be
# raised.
log.info( "States; No input" )
source = cpppo.chainable()
sequence = machine.run( source=source )
for num in range( 10 ):
try:
mch,sta = next( sequence )
except StopIteration:
sequence = None
break
except cpppo.NonTerminal as e:
assert "non-terminal state" in str( e )
break
inp = source.peek()
log.info( "%s <- %r" % ( cpppo.centeraxis( mch, 25, clip=True ), inp ))
if num == 0: assert inp is None; assert sta.name == "Initial"
if num == 1: assert inp is None; assert sta.name == "Middle"
if num == 2: assert inp is None; assert sta is None # And no more no-input transitions
assert num < 3 # If we get here, we didn't detect loop
assert num == 3
# since the iterator did not stop cleanly (after processing a state's input,
# and then trying to determine the next state), it'll continue indefinitely
assert sta is None
assert sequence is not None
# Try with some input loaded into source stream, using an identical generator sequence.
# Only the first element is gotten, and is reused for every NULL state transition, and is
# left over at the end.
log.info( "States; 'abc' input" )
assert source.peek() is None
source.chain( b'abc' )
assert source.peek() == b'a'[0] # python2: str, python3: int
sequence = machine.run( source=source )
for num in range( 10 ):
try:
mch,sta = next( sequence )
except StopIteration:
break
inp = source.peek()
log.info( "%s <- %r", cpppo.centeraxis( mch, 25, clip=True ), inp )
if num == 0: assert inp == b'a'[0]; assert sta.name == "Initial"
if num == 1: assert inp == b'a'[0]; assert sta.name == "Middle"
if num == 2: assert inp == b'a'[0]; assert sta.name == "Terminal"
assert num < 3
assert num == 3
assert inp == b'a'[0]
assert sta.name == "Terminal"
def test_state():
"""A state is expected to process its input (perhaps nothing, if its a no-input state), and then use
the next input symbol to transition to another state. Each state has a context into a data
artifact, into which it will collect its results.
We must ensure that all state transitions are configured in the target alphabet; if an encoder
is supplied, then all input symbols and all transition symbols will be encoded using it. In
this test, all string literals are Unicode (in both Python 2 and 3), so we use a unicode encoder
to convert them to symbols."""
unicodekwds = {
'alphabet': unicode if sys.version_info[0] < 3 else str,
'encoder': cpppo.type_unicode_encoder,
}
s1 = cpppo.state(
'one', **unicodekwds )
s2 = cpppo.state_drop(
'two', **unicodekwds )
s1['a'] = s2
assert s1['a'] is s2
source = cpppo.peeking( 'abc' )
# We can run state instances with/without acquisition
g = s1.run( source=source )
assert next( g ) == (None, s2)
assert source.peek() == 'a'
with pytest.raises(StopIteration):
next( g )
with s1:
g = s1.run( source=source )
assert source.peek() == 'a'
assert next( g ) == (None, s2)
assert source.peek() == 'a'
try:
next( g )
assert False, "Should have terminated"
except StopIteration:
pass
assert source.peek() == 'a'
# A state machine accepting a sequence of unicode a's
a_s = cpppo.state( "a_s", **unicodekwds )
an_a = cpppo.state_input( "a", terminal=True,
typecode=cpppo.type_unicode_array_symbol,
**unicodekwds )
a_s['a'] = an_a
an_a['a'] = an_a
source = cpppo.peeking( 'aaaa' )
data = cpppo.dotdict()
with cpppo.dfa( initial=a_s ) as aplus:
for i,(m,s) in enumerate( aplus.run( source=source )):
log.info( "%s #%3d -> %10.10s; next byte %3d: %-10.10r: %r", m.name_centered(),
i, s, source.sent, source.peek(), data )
assert i == 5
assert source.peek() is None
assert len( data ) == 0
# Accepting a's separated by comma and space/pi (for kicks). When the lower level a's machine
# doesn't recognize the symbol, then the higher level machine will recognize and discard
sep = cpppo.state_drop( "sep", **unicodekwds )
csv = cpppo.dfa( "csv", initial=a_s , terminal=True, **unicodekwds )
csv[','] = sep
sep[' '] = sep
sep['π'] = sep
sep[None] = csv
source = cpppo.peeking( 'aaaa, a,π a' )
data = cpppo.dotdict()
with cpppo.dfa( initial=csv ) as csvaplus:
for i,(m,s) in enumerate( csvaplus.run( source=source, path="csv", data=data )):
log.info( "%s #%3d -> %10.10s; next byte %3d: %-10.10r: %r", m.name_centered(),
i, s, source.sent, source.peek(), data )
assert i == 18
assert source.peek() is None
assert data.csv.input.tounicode() == 'aaaaaa'
class Object(object):
"""An EtherNet/IP device.Object is capable of parsing and processing a number of requests. It has
a class_id and an instance_id; an instance_id of 0 indicates the "class" instance of the
device.Object, which has different (class level) Attributes (and may respond to different commands)
than the other instance_id's.
Each Object has a single class-level parser, which is used to register all of its available
service request parsers. The next available symbol designates the type of service request,
eg. 0x01 ==> Get Attributes All. These parsers enumerate the requests that are *possible* on
the Object. Later, when the Object is requested to actually process the request, a decision can
be made about whether the request is *allowed*.
The parser knows how to parse any requests it must handle, and any replies it can generate, and
puts the results into the provided data artifact.
Assuming Obj is an instance of Object, and the source iterator produces the incoming symbols:
0x52, 0x04, 0x91, 0x05, 0x53, 0x43, 0x41, 0x44, #/* R...SCAD */
0x41, 0x00, 0x14, 0x00, 0x02, 0x00, 0x00, 0x00, #/* A....... */
then we could run the parser:
data = cpppo.dotdict()
with Obj.parse as machine:
for m,w in machine.run( source=source, data=data ):
pass
and it would parse a recognized command (or reply, but that would be unexpected), and produce
the following entries (in data, under the current context):
'service': 0x52,
'path.segment': [{'symbolic': 'SCADA', 'length': 5}],
'read_frag.elements': 20,
'read_frag.offset': 2,
Then, we could process the request:
proceed = Obj.request( data )
and this would process a request, converting it into a reply (any data elements unchanged by the
reply remain):
'service': 0xd2, # changed: |= 0x80
'status': 0x00, # default if not specified
'path.segment': [{'symbolic': 'SCADA', 'length': 5}], # unchanged
'read_frag.elements': 20, # unchanged
'read_frag.offset': 2, # unchanged
'read_frag.type': 0x00c3, # produced for reply
'read_frag.data': [ # produced for response
0x104c, 0x0008,
0x0003, 0x0002, 0x0002, 0x0002,
0x000e, 0x0000, 0x0000, 0x42e6,
0x0007, 0x40c8, 0x40c8, 0x0000,
0x00e4, 0x0000, 0x0064, 0x02b2,
0x80c8
]
'input': bytearray( [ # encoded response payload
0xd2, 0x00, #/* ....,... */
0x00, 0x00, 0xc3, 0x00, 0x4c, 0x10, 0x08, 0x00, #/* ....L... */
0x03, 0x00, 0x02, 0x00, 0x02, 0x00, 0x02, 0x00, #/* ........ */
0x0e, 0x00, 0x00, 0x00, 0x00, 0x00, 0xe6, 0x42, #/* .......B */
0x07, 0x00, 0xc8, 0x40, 0xc8, 0x40, 0x00, 0x00, #/* ...@.@.. */
0xe4, 0x00, 0x00, 0x00, 0x64, 0x00, 0xb2, 0x02, #/* ....d... */
0xc8, 0x80 #/* .@ */
]
The response payload is also produced as a bytes array in data.input, encoded and ready for
transmission, or encapsulation by the next higher level of request processor (eg. a
Message_Router, encapsulating the response into an EtherNet/IP response).
"""
max_instance = 0
lock = threading.Lock()
service = {} # Service number/name mappings
transit = {} # Symbol to transition to service parser on
# The parser doesn't add a layer of context; run it with a path= keyword to add a layer
parser = cpppo.dfa(service, initial=cpppo.state('select'), terminal=True)
@classmethod
def register_service_parser(cls, number, name, short, machine):
"""Registers a parser with the Object. May be invoked during import; no logging."""
assert number not in cls.service and name not in cls.service, \
"Duplicate service #%d: %r registered for Object %s" % ( number, name, cls.__name__ )
cls.service[number] = name
cls.service[name] = number
cls.transit[number] = chr(
number) if sys.version_info.major < 3 else number
cls.parser.initial[cls.transit[number]] \
= cpppo.dfa( name=short, initial=machine, terminal=True )
GA_ALL_NAM = "Get Attributes All"
GA_ALL_CTX = "get_attributes_all"
GA_ALL_REQ = 0x01
GA_ALL_RPY = GA_ALL_REQ | 0x80
GA_SNG_NAM = "Get Attribute Single"
GA_SNG_REQ = 0x0e
GA_SNG_RPY = GA_SNG_REQ | 0x80
SA_SNG_NAM = "Set Attribute Single"
SA_SNG_REQ = 0x10
SA_SNG_RPY = SA_SNG_REQ | 0x80
def __init__(self, name=None, instance_id=None):
"""Create the instance (default to the next available instance_id). An instance_id of 0 holds
the "class" attributes/commands.
"""
self.name = name or self.__class__.__name__
# Allocate and/or keep track of maximum instance ID assigned thus far.
if instance_id is None:
instance_id = self.__class__.max_instance + 1
if instance_id > self.__class__.max_instance:
self.__class__.max_instance = instance_id
self.instance_id = instance_id
(log.normal if self.instance_id else log.info)(
"%24s, Class ID 0x%04x, Instance ID %3d created", self,
self.class_id, self.instance_id)
instance = lookup(self.class_id, instance_id)
assert instance is None, \
"CIP Object class %x, instance %x already exists" % ( self.class_id, self.instance_id )
#
# directory.1.2.None == self
# self.attribute == directory.1.2 (a dotdict), for direct access of our attributes
#
self.attribute = directory.setdefault(
str(self.class_id) + '.' + str(instance_id), cpppo.dotdict())
self.attribute['0'] = self
# Check that the class-level instance (0) has been created; if not, we'll create one using
# the default parameters. If this isn't appropriate, then the user should create it using
# the appropriate parameters.
if lookup(self.class_id, 0) is None:
self.__class__(name='meta-' + self.name, instance_id=0)
if self.instance_id == 0:
# Set up the default Class-level values.
self.attribute['1'] = Attribute('Revision', INT, default=0)
self.attribute['2'] = MaxInstance('Max Instance',
INT,
class_id=self.class_id)
self.attribute['3'] = NumInstances('Num Instances',
INT,
class_id=self.class_id)
# A UINT array; 1st UINT is size (default 0)
self.attribute['4'] = Attribute('Optional Attributes',
INT,
default=0)
def __str__(self):
return self.name
def __repr__(self):
return "(0x%02x,%3d) %s" % (self.class_id, self.instance_id, self)
def request(self, data):
"""Handle a request, converting it into a response. Must be a dotdict data artifact such as is
produced by the Object's parser. For example, a request data containing either of the
following:
{
'service': 0x01,
'get_attributes_all': True,
}
should run the Get Attribute All service, and return True if the channel should continue.
In addition, we produce the bytes used by any higher level encapsulation.
TODO: Validate the request.
"""
result = b''
if log.isEnabledFor(logging.DETAIL):
log.detail("%s Request: %s", self, enip_format(data))
try:
# Validate the request. As we process, ensure that .status is set to reflect the
# failure mode, should an exception be raised. Return True iff the communications
# channel should continue.
data.status = 0x08 # Service not supported, if not recognized
data.pop('status_ext', None)
if (data.get('service') == self.GA_ALL_REQ
or 'get_attributes_all' in data and data.setdefault(
'service', self.GA_ALL_REQ) == self.GA_ALL_REQ):
pass
else:
raise AssertionError("Unrecognized Service Request")
# A recognized request; process the request data artifact, converting it into a reply.
data.service |= 0x80
if data.service == self.GA_ALL_RPY:
# Get Attributes All. Collect up the bytes representing the attributes. Replace
# the place-holder .get_attribute_all=True with a real dotdict.
data.status = 0x08 # Service not supported, if we fail to access an Attribute
result = b''
a_id = 1
while str(a_id) in self.attribute:
result += self.attribute[str(a_id)].produce()
a_id += 1
data.get_attributes_all = cpppo.dotdict()
data.get_attributes_all.data = bytearray(result)
data.status = 0x00
data.pop('status_ext', None)
# TODO: Other request processing here...
else:
raise AssertionError("Unrecognized Service Reply")
except Exception as exc:
log.warning(
"%r Service 0x%02x %s failed with Exception: %s\nRequest: %s\n%s\nStack %s",
self, data.service if 'service' in data else 0,
(self.service[data.service] if 'service' in data
and data.service in self.service else "(Unknown)"), exc,
enip_format(data),
''.join(traceback.format_exception(*sys.exc_info())),
''.join(traceback.format_stack()))
assert data.status != 0x00, \
"Implementation error: must specify .status error code before raising Exception"
pass
# Always produce a response payload; if a failure occurred, will contain an error status.
# If this fails, we'll raise an exception for higher level encapsulation to handle.
data.input = bytearray(self.produce(data))
log.detail("%s Response: %s: %s", self, self.service[data.service],
enip_format(data))
return True # We shouldn't be able to terminate a connection at this level
@classmethod
def produce(cls, data):
result = b''
if (data.get('service') == cls.GA_ALL_REQ
or 'get_attributes_all' in data and data.setdefault(
'service', cls.GA_ALL_REQ) == cls.GA_ALL_REQ):
# Get Attributes All
result += USINT.produce(data.service)
result += EPATH.produce(data.path)
elif data.get('service') == cls.GA_ALL_RPY:
# Get Attributes All Reply
result += USINT.produce(data.service)
result += b'\x00' # reserved
result += status.produce(data)
result += octets_encode(data.get_attributes_all.data)
else:
assert False, "%s doesn't recognize request/reply format: %r" % (
cls.__name__, data)
return result
def __init__( self, name=None, **kwds ):
name = name or kwds.setdefault( 'context', self.__class__.__name__ )
# Get the size, and chain remaining machine onto rest. When used as a Route Path, the size
# is padded, so insert a state to drop the pad, and chain rest to that instead.
size = rest = USINT( context='size' )
if self.padsize:
size[True] = rest = octets_drop( 'pad', repeat=1 )
# After capturing each segment__ (pseg), move it onto the path segment list, and loop
pseg = octets_noop( 'type', terminal=True )
# ...segment parsers...
pmov = move_if( 'move', initializer=lambda **kwds: [],
source='..segment__', destination='..segment',
state=pseg )
# Wire each different segment type parser between pseg and pmov
pseg[b'\x28'[0]]= e_8t = octets_drop( 'type', repeat=1 )
e_8t[True] = e_8v = USINT( 'elem_8bit', context='element')
e_8v[None] = pmov
pseg[b'\x29'[0]]= e16t = octets_drop( 'type', repeat=2 )
e16t[True] = e16v = UINT( 'elem16bit', context='element')
e16v[None] = pmov
pseg[b'\x2a'[0]]= e32t = octets_drop( 'type', repeat=2 )
e32t[True] = e32v = UDINT( 'elem32bit', context='element')
e32v[None] = pmov
pseg[b'\x20'[0]]= c_8t = octets_drop( 'type', repeat=1 )
c_8t[True] = c_8v = USINT( 'clas_8bit', context='class')
c_8v[None] = pmov
pseg[b'\x21'[0]]= c16t = octets_drop( 'type', repeat=2 )
c16t[True] = c16v = UINT( 'clas16bit', context='class')
c16v[None] = pmov
pseg[b'\x24'[0]]= i_8t = octets_drop( 'type', repeat=1 )
i_8t[True] = i_8v = USINT( 'inst_8bit', context='instance')
i_8v[None] = pmov
pseg[b'\x25'[0]]= i16t = octets_drop( 'type', repeat=2 )
i16t[True] = i16v = UINT( 'inst16bit', context='instance')
i16v[None] = pmov
pseg[b'\x30'[0]]= a_8t = octets_drop( 'type', repeat=1 )
a_8t[True] = a_8v = USINT( 'attr_8bit', context='attribute')
a_8v[None] = pmov
pseg[b'\x31'[0]]= a16t = octets_drop( 'type', repeat=2 )
a16t[True] = a16v = UINT( 'attr16bit', context='attribute')
a16v[None] = pmov
pseg[b'\x91'[0]]= symt = octets_drop( 'type', repeat=1 )
symt[True] = syml = USINT( 'sym_len', context='symbolic.length' )
syml[None] = symv = cpppo.string_bytes(
'symbolic', context='symbolic', limit='.length',
initial='.*', decode='iso-8859-1' )
# An odd-length ANSI Extended Symbolic name means an odd total. Pad
symo = octets_drop( 'pad', repeat=1 )
symo[None] = pmov
symv[None] = cpppo.decide( 'odd',
predicate=lambda path=None, data=None, **kwds: len( data[path].symbolic ) % 2,
state=symo )
symv[None] = pmov
# Route Path port/link-address. See Vol 1-3.13, Table C-1.3 Port Segment Encoding.
# segment: 0b000spppp
# |\\\\+-> port number 0x01-0x0E; 0x0F=>extended
# |
# +------> link size+address; 0=>numeric, 1=>size+string
#
def port_fix( path=None, data=None, **kwds ):
"""Discard port values about 0x0F; return True (transition) if remaining port value is 0x0F
(Optional Extended port)"""
data[path].port &= 0x0F
if data[path].port == 0x0F:
# Port is extended; discard and prepare to collect new port number
data[path].port = cpppo.dotdict()
return True
# Port is OK; don't transition
return False
# [01-0E][LL] port 01-0E, link-address #LL
pseg[b'\x01'[0]]= pnum = USINT( 'port_num', context='port' )
pseg[b'\x02'[0]] = pnum
pseg[b'\x03'[0]] = pnum
pseg[b'\x04'[0]] = pnum
pseg[b'\x05'[0]] = pnum
pseg[b'\x06'[0]] = pnum
pseg[b'\x07'[0]] = pnum
pseg[b'\x08'[0]] = pnum
pseg[b'\x09'[0]] = pnum
pseg[b'\x0a'[0]] = pnum
pseg[b'\x0b'[0]] = pnum
pseg[b'\x0c'[0]] = pnum
pseg[b'\x0d'[0]] = pnum
pseg[b'\x0e'[0]] = pnum
# [0F][PPPP][LL] port 0xPPPP, link-address 0xLL
pseg[b'\x0f'[0]] = pnum
# A big port#; re-scan a UINT into .port (won't work 'til port_fix is called)
pnbg = UINT( 'port_nbg', context='port' )
pnbg[True] = pnlk = USINT( 'link_num', context='link' )
# Fix the port#; if 0x0F, setup for extended port and transition to pnbg. Otherwise,
# (not extended port), just go the the port numeric link.
pnum[None] = cpppo.decide( 'port_nfix', predicate=port_fix,
state=pnbg )
pnum[None] = pnlk
pnlk[None] = pmov # and done; move segment, get next
# [11-1E][SS]'123.123.123.123'[00] port 0x01-0E, link address '123.123.123.123' (pad if size 0xSS odd)
pseg[b'\x11'] = padr = USINT( 'port_adr', context='port' )
pseg[b'\x12'[0]] = padr
pseg[b'\x13'[0]] = padr
pseg[b'\x14'[0]] = padr
pseg[b'\x15'[0]] = padr
pseg[b'\x16'[0]] = padr
pseg[b'\x17'[0]] = padr
pseg[b'\x18'[0]] = padr
pseg[b'\x19'[0]] = padr
pseg[b'\x1a'[0]] = padr
pseg[b'\x1b'[0]] = padr
pseg[b'\x1c'[0]] = padr
pseg[b'\x1d'[0]] = padr
pseg[b'\x1e'[0]] = padr
# [1F][SS][PPPP]'123.123.123.123'[00] port 0xPPPP, link address '123.123.123.123' (pad if size SS odd)
pseg[b'\x1f'[0]] = padr
# Harvest the addresses into .link
adrv = cpppo.string_bytes(
'link_add', context='link', limit='.length',
initial='.*', decode='iso-8859-1' )
# An odd-length link address means an odd total. Pad
adro = octets_drop( 'link_pad', repeat=1 )
adro[None] = pmov
adrv[None] = cpppo.decide( 'link_odd',
predicate=lambda path=None, data=None, **kwds: len( data[path+'.link'] ) % 2,
state=adro )
adrv[None] = pmov
# A big port#; re-scan a UINT into .port (won't work 'til port_fix is called)
pabg = UINT( 'port_abg', context='port' )
pabg[None] = adrv
#
padr[True] = adrl = USINT( 'link_len', context='link.length' )
adrl[None] = cpppo.decide( 'port_afix', predicate=port_fix,
state=pabg )
adrl[None] = adrv
# Parse all segments in a sub-dfa limited by the parsed path.size (in words; double)
rest[None] = cpppo.dfa( 'each', context='segment__',
initial=pseg, terminal=True,
limit=lambda path=None, data=None, **kwds: data[path+'..size'] * 2 )
super( EPATH, self ).__init__( name=name, initial=size, **kwds )