def buildSyllable(sTree, segments, noLabel):
c = tb.tree_category(sTree)
if c == "C0" or c == "C1" or c == "C2" or c == "C3" or c == "C4" or c == "C5" or c == "Consonant" or c == "Stress":
if noLabel:
segments.append(tb.tree_children(sTree)[0])
else:
segments.append(tb.tree_children(sTree)[0] + "_C")
elif tb.tree_category(sTree) == "Vowel":
if noLabel:
segments.append(tb.tree_children(sTree)[0])
else:
segments.append(tb.tree_children(sTree)[0] + "_V")
else:
for child in tb.tree_children(sTree):
buildSyllable(child, segments, noLabel)
def buildWord(wTree, sCat, syllables, segSep, sylSep, noLabel):
try:
if sCat.search(
tb.tree_category(wTree)
): #=="Syllable" or tb.tree_category(wTree)=="SyllableIF" or tb.tree_category(wTree)=="SyllableI" or tb.tree_category(wTree)=="SyllableF":
segments = []
buildSyllable(wTree, segments, noLabel)
syllables.append(segSep.join(segments))
else:
for child in tb.tree_children(wTree):
buildWord(child, sCat, syllables, segSep, sylSep, noLabel)
except:
print "%s\n" % wTree
sys.exit()
for child in tb.tree_children(wTree):
buildWord(child, sCat, syllables, segSep, sylSep, noLabel)
def rightbranch(tree):
"""
Transform a subtree lying on a right branch.
"""
def leftbranch(subtree, transformed_right):
"""
Transform a subtree lying on a left branch.
transformed_right is transformed right material between this node and Anc.
"""
if tb.is_preterminal(subtree):
return [subtree, transformed_right]
else:
left, right = tb.tree_children(subtree)
return leftbranch(
left,
tb.make_nonterminal(
make_pair(Anc, tb.tree_label(left)),
rightbranch(right) + [transformed_right]))
if tb.is_preterminal(tree):
return [flag(tree)]
else:
Anc = tb.tree_label(tree)
left, right = tb.tree_children(tree)
return leftbranch(
left,
tb.make_nonterminal(make_pair(Anc, tb.tree_label(left)),
rightbranch(right)))
def visit(node, sofar):
assert tb.is_phrasal(node)
label = tb.tree_label(node)
assert is_pair(label)
A, X = pair_categories(label)
children = tb.tree_children(node)
assert len(children) > 0
assert tb.is_preterminal(children[0])
if len(children) == 1:
assert is_flagged(children[0])
xf = cat_c(A, X, unflag(children[0]))
elif len(children) == 2:
if is_flagged(children[0]):
A1, B = pair_categories(tb.tree_label(children[1]))
assert A1 == A
xf = cat_b(A, X, B, unflag(children[0]))
else:
C, a = pair_categories(tb.tree_label(children[1]))
assert a == tb.tree_label(children[0]), \
"error in label of node.children[1] a = {}, node = {}".format(a, node)
xf = cat_e(A, X, C, children[0])
elif len(children) == 3:
assert not is_flagged(children[0])
C, a = pair_categories(tb.tree_label(children[1]))
A1, B = pair_categories(tb.tree_label(children[2]))
assert A == A1
xf = cat_d(A, X, B, C, children[0])
else:
sys.exit("error: ill-formed subtree {}\n in tree {}".format(
node, xtree))
sofar.append(xf)
for child in children[1:]:
sofar = visit(child, sofar)
return sofar
def leftbranch(subtree, continuation):
if tb.is_preterminal(subtree):
return [subtree] + continuation
else:
left, right = tb.tree_children(subtree)
return leftbranch(left, [
tb.make_nonterminal(make_pair(Anc, tb.tree_label(left)),
rightbranch(right) + continuation)
])
def leftbranch(subtree, continuation, X1):
if tb.is_preterminal(subtree):
return [relabel(subtree, X1)] + continuation
else:
left, right = tb.tree_children(subtree)
X2 = tb.tree_label(left) + '>'
return leftbranch(left, [
tb.make_nonterminal(make_pair(Anc, tb.tree_label(left)),
rightbranch(right, X2) + continuation)
], X1)
def visit(node, wordssofar, segssofar):
"""Does a preorder visit of the nodes in the tree"""
if tb.is_terminal(node):
if not ignore_terminal_rex.match(node):
segssofar.append(simplify_terminal(node))
return wordssofar,segssofar
for child in tb.tree_children(node):
wordssofar,segssofar = visit(child, wordssofar, segssofar)
if word_rex.match(tb.tree_label(node)):
if segssofar != []:
wordssofar.append(''.join(segssofar))
segssofar = []
return wordssofar,segssofar
def visit(node, wordssofar, segssofar):
"""Does a preorder visit of the nodes in the tree"""
if tb.is_terminal(node):
if not ignore_terminal_rex.match(node):
segssofar.append(simplify_terminal(node))
return wordssofar, segssofar
for child in tb.tree_children(node):
wordssofar, segssofar = visit(child, wordssofar, segssofar)
if word_rex.match(tb.tree_label(node)):
if segssofar != []:
wordssofar.append(''.join(segssofar))
segssofar = []
return wordssofar, segssofar
def visitTree(tree, words, wCat, sCat, segSep, sylSep, noLabel):
"""
Performs a pre-order traversal of tree, and collects syllable and word boundaries
"""
def buildSyllable(sTree, segments, noLabel):
c = tb.tree_category(sTree)
if c == "C0" or c == "C1" or c == "C2" or c == "C3" or c == "C4" or c == "C5" or c == "Consonant" or c == "Stress":
if noLabel:
segments.append(tb.tree_children(sTree)[0])
else:
segments.append(tb.tree_children(sTree)[0] + "_C")
elif tb.tree_category(sTree) == "Vowel":
if noLabel:
segments.append(tb.tree_children(sTree)[0])
else:
segments.append(tb.tree_children(sTree)[0] + "_V")
else:
for child in tb.tree_children(sTree):
buildSyllable(child, segments, noLabel)
def buildWord(wTree, sCat, syllables, segSep, sylSep, noLabel):
try:
if sCat.search(
tb.tree_category(wTree)
): #=="Syllable" or tb.tree_category(wTree)=="SyllableIF" or tb.tree_category(wTree)=="SyllableI" or tb.tree_category(wTree)=="SyllableF":
segments = []
buildSyllable(wTree, segments, noLabel)
syllables.append(segSep.join(segments))
else:
for child in tb.tree_children(wTree):
buildWord(child, sCat, syllables, segSep, sylSep, noLabel)
except:
print "%s\n" % wTree
sys.exit()
for child in tb.tree_children(wTree):
buildWord(child, sCat, syllables, segSep, sylSep, noLabel)
# print("VisitT %s"%tb.tree_category(tree))
if wCat.search(tb.tree_category(tree)):
syllables = []
buildWord(tree, sCat, syllables, segSep, sylSep, noLabel)
words.append(sylSep.join(syllables))
else:
for child in tb.tree_children(tree):
visitTree(child, words, wCat, sCat, segSep, sylSep, noLabel)
def lcx2tree_labels0(xtree):
"""
Maps an lcx2 tree to the corresponding labels, as in my 1996 paper.
"""
def visit(node, sofar):
assert tb.is_phrasal(node)
label = tb.tree_label(node)
assert is_pair(label)
A, X = pair_categories(label)
children = tb.tree_children(node)
assert len(children) > 0
assert tb.is_preterminal(children[0])
if len(children) == 1:
assert is_flagged(children[0])
xf = cat_c(A, X, unflag(children[0]))
elif len(children) == 2:
if is_flagged(children[0]):
A1, B = pair_categories(tb.tree_label(children[1]))
assert A1 == A
xf = cat_b(A, X, B, unflag(children[0]))
else:
C, a = pair_categories(tb.tree_label(children[1]))
assert a == tb.tree_label(children[0]), \
"error in label of node.children[1] a = {}, node = {}".format(a, node)
xf = cat_e(A, X, C, children[0])
elif len(children) == 3:
assert not is_flagged(children[0])
C, a = pair_categories(tb.tree_label(children[1]))
A1, B = pair_categories(tb.tree_label(children[2]))
assert A == A1
xf = cat_d(A, X, B, C, children[0])
else:
sys.exit("error: ill-formed subtree {}\n in tree {}".format(
node, xtree))
sofar.append(xf)
for child in children[1:]:
sofar = visit(child, sofar)
return sofar
root = tb.tree_label(xtree)
rchildren = tb.tree_children(xtree)
assert len(rchildren) == 2, "nonbinary xtree = {}".format(xtree)
sofar = [cat_a(root, rchildren[0])]
return visit(rchildren[1], sofar)
def visit(node, wordssofar, segssofar):
"""Does a preorder visit of the nodes in the tree"""
if tb.is_terminal(node):
if not ignore_terminal_rex.match(node):
segssofar.append(node)
return wordssofar,segssofar
for child in tb.tree_children(node):
wordssofar,segssofar = visit(child, wordssofar, segssofar)
mo = score_cat_rex.match(tb.tree_label(node))
if mo:
if segssofar != []:
word = ''.join(segssofar)
segssofar = []
try:
topic = mo.group('topic')
if topic != None:
wordssofar.append((word,topic))
else:
wordssofar.append(word)
except IndexError:
wordssofar.append(word)
return wordssofar,segssofar
def relabel(tree, label):
return tb.make_nonterminal(tree[0] + ' ' + label,
tb.tree_children(tree))