def tonal_space_local_alignment(sem1, sem2):
"""
Like L{tonal_space_alignment}, but uses local alignment of seq2 within seq1.
Also returns the distance metric (like L{tonal_space_local_distance}.
"""
from jazzparser.utils.distance import local_levenshtein_distance
# Get a list of (coord,fun) pairs for the logical forms
seq1 = _lf_to_coord_funs(sem1)
seq2 = _lf_to_coord_funs(sem2)
# Produce a version of the paths made up of steps and functions,
# rather than points and functions
steps1 = _steps_list(seq1)
steps2 = _steps_list(seq2)
dists,pointers = local_levenshtein_distance(
steps1,
steps2,
delins_cost=2,
subst_cost_fun=_subst_cost)
# We now have the matrix of costs and the pointers that generated
# those costs.
# Trace back to find out what produces the optimal alignment
# Start at the top right corner
i,j = (len(dists)-1), (len(dists[0])-1)
oplist = []
while not i == j == 0:
if pointers[i][j] == "I":
oplist.append("I")
j -= 1
elif pointers[i][j] == "D":
oplist.append("D")
i -= 1
elif pointers[i][j] == ".":
oplist.append(".")
i -= 1
else:
# Substitution: find out what kind
step1 = steps1[i-1]
step2 = steps2[j-1]
subst_type = _subst_type(step1, step2)
if subst_type == "both":
oplist.append("S")
elif subst_type == "fun":
oplist.append("Sf")
elif subst_type == "root":
oplist.append("Sr")
else:
oplist.append("A")
j -= 1
i -= 1
oplist = list(reversed(oplist))
distance = float(dists[-1][-1]) / 2.0
return oplist, steps1, steps2, distance
def tonal_space_local_alignment(sem1, sem2):
"""
Like L{tonal_space_alignment}, but uses local alignment of seq2 within seq1.
Also returns the distance metric (like L{tonal_space_local_distance}.
"""
from jazzparser.utils.distance import local_levenshtein_distance
# Get a list of (coord,fun) pairs for the logical forms
seq1 = _lf_to_coord_funs(sem1)
seq2 = _lf_to_coord_funs(sem2)
# Produce a version of the paths made up of steps and functions,
# rather than points and functions
steps1 = _steps_list(seq1)
steps2 = _steps_list(seq2)
dists, pointers = local_levenshtein_distance(steps1,
steps2,
delins_cost=2,
subst_cost_fun=_subst_cost)
# We now have the matrix of costs and the pointers that generated
# those costs.
# Trace back to find out what produces the optimal alignment
# Start at the top right corner
i, j = (len(dists) - 1), (len(dists[0]) - 1)
oplist = []
while not i == j == 0:
if pointers[i][j] == "I":
oplist.append("I")
j -= 1
elif pointers[i][j] == "D":
oplist.append("D")
i -= 1
elif pointers[i][j] == ".":
oplist.append(".")
i -= 1
else:
# Substitution: find out what kind
step1 = steps1[i - 1]
step2 = steps2[j - 1]
subst_type = _subst_type(step1, step2)
if subst_type == "both":
oplist.append("S")
elif subst_type == "fun":
oplist.append("Sf")
elif subst_type == "root":
oplist.append("Sr")
else:
oplist.append("A")
j -= 1
i -= 1
oplist = list(reversed(oplist))
distance = float(dists[-1][-1]) / 2.0
return oplist, steps1, steps2, distance
def tonal_space_local_distance(sem1, sem2):
"""
Like L{tonal_space_distance}, but uses local alignment of seq2 within
seq1.
"""
from jazzparser.utils.distance import local_levenshtein_distance
# Get a list of (coord,fun) pairs for the logical forms
seq1 = _lf_to_coord_funs(sem1)
seq2 = _lf_to_coord_funs(sem2)
# Produce a version of the paths made up of steps and functions,
# rather than points and functions
steps1 = _steps_list(seq1)
steps2 = _steps_list(seq2)
dists, pointers = local_levenshtein_distance(steps1,
steps2,
delins_cost=2,
subst_cost_fun=_subst_cost)
return float(dists[-1][-1]) / 2.0
def tonal_space_local_distance(sem1, sem2):
"""
Like L{tonal_space_distance}, but uses local alignment of seq2 within
seq1.
"""
from jazzparser.utils.distance import local_levenshtein_distance
# Get a list of (coord,fun) pairs for the logical forms
seq1 = _lf_to_coord_funs(sem1)
seq2 = _lf_to_coord_funs(sem2)
# Produce a version of the paths made up of steps and functions,
# rather than points and functions
steps1 = _steps_list(seq1)
steps2 = _steps_list(seq2)
dists,pointers = local_levenshtein_distance(
steps1,
steps2,
delins_cost=2,
subst_cost_fun=_subst_cost)
return float(dists[-1][-1]) / 2.0
