def input_coercion(name, x_fst, *contexts):
"""Compiles rules like X <-- LC1 _ RC1, ..., LCk _ RCk
name -- name to be given to the rule FST
x_fst -- the center (X) of the rule
\*contexts -- list of contexts, i.e. pairs of left and right context
Returns a triple:
rule_fst -- the compiled rule
selector_fst -- FST which selects examples which are relevant for
this rule
scrambler_fst -- an encoded FST which produces negative examples
"""
global pistar_fst
postcondition_fst = x_to_condition(x_fst)
x_all_fst = pistar_fst.copy()
temp_fst = x_fst.copy()
temp_fst.output_project()
x_all_fst.compose(temp_fst) # PI* .o. X.l
precondition_fst = x_to_condition(x_all_fst)
context_condition_fst = contexts_to_condition(*contexts)
precondition_fst.intersect(context_condition_fst)
rule_fst = generalized_restriction(precondition_fst, postcondition_fst)
rule_fst.set_name(name)
if cfg.verbosity >= 20:
twbt.ppfst(rule_fst, True)
selector_fst = selector_from_x(x_all_fst)
scrambler_fst = correct_to_incorrect(x_fst, "input")
return rule_fst, selector_fst, scrambler_fst
def read_fst(filename="examples.fst"):
"""Reads in a previously stored example FST file
"""
import hfst
exfile = hfst.HfstInputStream(filename)
cfg.examples_fst = exfile.read()
pair_symbols = cfg.examples_fst.get_property("x-pair_symbols")
# print("pair_symbols", pair_symbols) ##
pair_symbol_lst = re.split(r" +", pair_symbols)
for pair in pair_symbol_lst:
cfg.pair_symbol_set.add(pair)
(insym, outsym) = cfg.pairsym2sympair(pair)
cfg.symbol_pair_set.add((insym, outsym))
cfg.input_symbol_set.add(insym)
cfg.output_symbol_set.add(outsym)
cfg.all_pairs_fst = hfst.empty_fst()
for insym, outsym in cfg.symbol_pair_set:
in_quoted = re.sub(r"([{}])", r"%\1", insym)
#print(in_quoted, outsym)### tilts if insym contains bad chars
pair_fst = hfst.regex(in_quoted + ':' + outsym)
cfg.all_pairs_fst.disjunct(pair_fst)
cfg.all_pairs_fst.remove_epsilons()
cfg.all_pairs_fst.minimize()
if cfg.verbosity >= 30:
twbt.ppfst(cfg.all_pairs_fst, title="cfg.all_pairs_fst")
return
def read_examples(filename="test.pstr", build_fsts=True):
"""Reads the examples from the file whose name is 'filename'.
The file must contain one example per line and each line consists of
a space separated sequence of pair-symbols. The examples are processed into
"""
if build_fsts:
import hfst
examples_bfst = hfst.HfstBasicTransducer()
exfile = open(filename, "r")
for line_nl in exfile:
line = line_nl.strip()
if not line or line.startswith("!"):
continue
pairsym_lst = re.split("\s+", line)
symbol_pair_lst = [
cfg.pairsym2sympair(pairsym) for pairsym in pairsym_lst
]
# print("symbol_pair_lst:", symbol_pair_lst) ##
pair_symbol_str = " ".join([
cfg.sympair2pairsym(insym, outsym)
for insym, outsym in symbol_pair_lst
])
# print("pair_symbol_lst:", pair_symbol_lst) ##
cfg.example_lst.append(pair_symbol_str)
cfg.example_set.add(pair_symbol_str) # spaces normalized
#LINE_FST = hfst.tokenized_fst(symbol_pair_lst)
# twbt.printfst(LINE_FST, True) ##
if build_fsts:
examples_bfst.disjunct(symbol_pair_lst, 0)
for insym, outsym in symbol_pair_lst:
cfg.symbol_pair_set.add((insym, outsym))
exfile.close()
if cfg.verbosity >= 30:
print("List of examples:", cfg.example_lst)
print("List of alphabet symbol pairs:", sorted(cfg.symbol_pair_set))
if build_fsts:
cfg.examples_fst = hfst.HfstTransducer(examples_bfst)
cfg.examples_fst.set_name(filename)
cfg.examples_fst.minimize()
if cfg.verbosity >= 30:
twbt.ppfst(cfg.examples_fst, False,
title="Example file as FST") ##
for insym, outsym in cfg.symbol_pair_set:
cfg.input_symbol_set.add(insym)
cfg.output_symbol_set.add(outsym)
for insym, outsym in cfg.symbol_pair_set:
pair_symbol = cfg.sympair2pairsym(insym, outsym)
cfg.pair_symbol_set.add(pair_symbol)
if build_fsts:
pair_symbol_lst = [
insym + ':' + outsym for insym, outsym in cfg.symbol_pair_set
]
pair_symbol_str = " ".join(sorted(pair_symbol_lst))
# print("symbol pairs:", pair_symbol_str) ##
cfg.examples_fst.set_property("x-pair_symbols", pair_symbol_str)
return
def e(str):
"""Convert a two-level component expression into a FST.
str -- a string containing a (two-level) regular expression
Returns an FST which performs the mapping represented by str
corresponding to the expression.
"""
global XRC
# print("Regex string:", str) ##
if str == "":
return (XRC.compile("[]"))
F = XRC.compile(str)
F.minimize()
F.set_name(str)
if cfg.verbosity >= 5:
twbt.ppfst(F) ##
return (F)
def read_fst(filename="examples.fst"):
"""Reads in a previously stored example FST file
"""
exfile = hfst.HfstInputStream(filename)
cfg.examples_fst = exfile.read()
pair_symbols = cfg.examples_fst.get_property("x-pair_symbols")
# print("pair_symbols", pair_symbols) ##
pair_symbol_lst = re.split(r" +", pair_symbols)
for pair in pair_symbol_lst:
cfg.pair_symbol_set.add(pair)
(insym, outsym) = cfg.pairsym2sympair(pair)
cfg.symbol_pair_set.add((insym, outsym))
cfg.input_symbol_set.add(insym)
cfg.output_symbol_set.add(outsym)
cfg.all_pairs_fst = pairs_to_fst(cfg.symbol_pair_set)
if cfg.verbosity >= 30:
twbt.ppfst(cfg.all_pairs_fst, title="cfg.all_pairs_fst")
return
def init():
"""Initializes the module by computing several common FSTs
Assumes that twexamp.read_fst() has read in cfg.examples_fst and
initialized sone symbol sets.
"""
global pistar_fst, pistar_fsa, diamond_sym, diamond_fst
global trim_pre_fst, trim_post_fst
assert cfg.examples_fst, "cfg.examples_fst not loaded (by twexamp module)"
cfg.definitions["PAIRS"] = cfg.all_pairs_fst.copy()
cfg.definitions["PI"] = cfg.all_pairs_fst.copy()
diamond_sym = 'DIAMOND'
diamond_fst = hfst.regex(diamond_sym)
pi_fst = cfg.all_pairs_fst.copy()
pistar_fst = cfg.all_pairs_fst.copy()
pistar_fst.repeat_star()
pistar_fst.remove_epsilons()
pistar_fst.minimize()
pistar_fsa = hfst.fst_to_fsa(pistar_fst, separator='^')
pi_in_fst = pi_fst.copy()
pi_in_fst.input_project()
pi_out_fst = pi_fst.copy()
pi_out_fst.output_project()
pi_in_star_fst = pistar_fst.copy()
pi_in_star_fst.input_project()
pi_out_star_fst = pistar_fst.copy()
pi_out_star_fst.output_project()
if cfg.verbosity >= 20:
twbt.ppfst(pistar_fst, title="pistar_fst")
fst1 = fs.star(fs.crossprod(fs.expr("ZERO"), pi_in_fst))
fst2 = fs.star(fs.concat(fst1, fs.expr("ZERO:BEGIN")))
fst3 = fs.concat(fst2, pi_in_star_fst)
fst4 = fs.star(
fs.concat(fs.expr("ZERO:END"),
fs.star(fs.crossprod(fs.expr("ZERO"), pi_in_fst))))
trim_pre_fst = fs.concat(fst3, fst4)
trim_pre_fst.set_name("trim_pre_fst")
#trim_pre_fst = XRC.compile(
# "[[ZERO .x. [PI].u]* ZERO:BEGIN]* " \
# "[[PI].u]* " \
# "[ZERO:END [ZERO .x. [PI].u]*]*"
#)
fst1 = fs.star(fs.crossprod(pi_out_fst, fs.expr("ZERO")))
fst2 = fs.star(fs.concat(fst1, fs.expr("BEGIN:ZERO")))
fst3 = fs.concat(fst2, pi_out_star_fst)
fst4 = fs.star(
fs.concat(fs.expr("END:ZERO"),
fs.star(fs.crossprod(pi_out_fst, fs.expr("ZERO")))))
trim_post_fst = fs.concat(fst3, fst4)
trim_post_fst.set_name("trim_post_fst")
#trim_post_fst = XRC.compile(
# "[[[PI].l .x. ZERO]* BEGIN:ZERO]* " \
# "[[PI].l]* " \
# "[END:ZERO [[PI].l .x. ZERO]*]*"
#)
if cfg.verbosity >= 20:
twbt.ppfst(trim_pre_fst)
twbt.ppfst(trim_post_fst)
return
selector_fst -- FST which selects examples which are relevant for this rule
scrambler_fst -- empty_fst (negative examples not relevant for these rules)
"""
context_condition_fst = contexts_to_condition(*contexts)
x_condition_fst = x_to_condition(x_fst)
context_condition_fst.intersect(x_condition_fst)
null_fst = hfst.empty_fst()
rule_fst = generalized_restriction(context_condition_fst, null_fst)
rule_fst.set_name(name)
# twbt.ppfst(rule_fst, True) ##
selector_fst = selector_from_x(x_fst)
scrambler_fst = hfst.empty_fst()
return rule_fst, selector_fst, scrambler_fst
if __name__ == "__main__":
twex.read_examples()
init(1)
#define("V", "PI .o.[a|e|i|o|ä|ö]")
#define("C", "[PI .o. [h|l|n|s|t|v]] | %{ij%}:j")
R1 = doublearrow("{ao}:o <=> _ {ij}:", e("%{ao%}:o"),
(e("[]"), e("[%{ij%} .o. PI]")))
twbt.ppfst(R1, True)
rule2_fst = doublearrow("{ij}:j <=> V :Ø* _ :Ø* V", "%{ij%}:j",
("V [PI .o. Ø]*", "[PI .o. Ø]* V"))
twbt.ppfst(rule2_fst, True)
R3 = doublearrow("{tl}:l <=> _ CLOSED", "%{tl%}:l",
("[]", "V %{ij%}:i* C [C | [PI .o. Ø]* END]"))
twbt.ppfst(R3, True)
description="A compiler and tester for two-level rules")
arpar.add_argument("start",
help="start parseing from",
default="expr_start")
args = arpar.parse_args()
twexamp.read_fst(filename="nounex.fst")
parser = init()
for line_nl in sys.stdin:
line = line_nl.strip()
#print(line)
result = parser.parse(line,
start=args.start,
semantics=TwolFstSemantics())
if args.start == "def_start":
op, left, right, source = result
print(left, "=")
twbt.ppfst(right)
elif args.start == "rul_start":
op, left, right, source = result
twbt.ppfst(left)
print(op)
for lc, rc in right:
twbt.ppfst(lc, title="left context")
twbt.ppfst(rc, title="right context")
elif args.start == "expr_start":
fst = result
#print(fst)
twbt.ppfst(fst, True)
elif op == "?":
print("Incorrect: " + line)
help="name of the examples fst or example pair symbol string file",
default="examples.fst")
arpar.add_argument("rules", help="name of the rule file", default="test.rules")
args = arpar.parse_args()
cfg.verbosity = args.verbosity
if args.recursion:
sys.setrecursionlimit(args.recursion)
if args.examples.endswith(".fst"):
twexamp.read_fst(args.examples)
else:
twexamp.read_examples(args.examples)
if cfg.verbosity >= 30:
twbt.ppfst(cfg.examples_fst, title="examples_fst")
parser = twparser.init()
examples_fsa = hfst.fst_to_fsa(cfg.examples_fst, separator="^")
examples_up_fsa = cfg.examples_fst.copy()
examples_up_fsa.input_project()
if cfg.verbosity >= 30:
twbt.ppfst(examples_up_fsa, title="examples_up_fsa")
twrule.init()
skip = False
all_rules_fst_lst = []
rule_file = open(args.rules, 'r')
