def find_tree(treebankdir, treebank_re, interesting_tree_p=lambda tree: True):
"""find_tree displays trees for which interesting_tree_p(tree) is True.
treebankdir and treebank_re specify the treebank files to be searched.
interesting_tree_p is a function which is called on each tree in turn.
If interesting_tree_p does not return False, the tree is displayed
using drawtree."""
for filename in lx.findfiles(treebankdir, treebank_re):
trees = read_file(filename)
treeno = 0
for root in trees:
treeno += 1
tree = root[1]
if interesting_tree_p(tree):
drawtree.tb_tk(tree)
inp = raw_input(filename+":"+str(treeno)+" find next tree? [Y/n/p] ").strip()
if len(inp) != 0:
if inp[0] == "P" or inp[0] == "p":
psfilename = raw_input("Saving postscript image of tree; enter filename for image file? ").strip()
if psfilename != "":
drawtree.tktb_postscript(psfilename)
elif inp[0] != "Y" and inp[0] != "y":
return
def trees(treebankdir, treebank_re):
"""trees is a generator that iterates through
the trees in a treebank. treebankdir is the name
of the directory in which the treebank files are
located, and treebank_re is a regular expression
that all treebank files must match."""
for filename in lx.findfiles(treebankdir, treebank_re):
trees = read_file(filename)
for root in trees:
tree = root[1]
yield tree
"""regularverbs.py extracts regular verbs and their inflections
from CHILDES files."""
import lx, re, xmlchat
childes_dir = "/usr/local/data/CHILDES/data-xml/English-USA/Brent"
for xmlfile in lx.findfiles(childes_dir, re.compile(r".*\.xml")):
roles, utterances = xmlchat.readfile(xmlfile)
for utt in utterances:
id = utt['who']
role = roles[id]['role']
if role in ["Mother","Father"]:
for morph in utt.get('mor', []):
if morph['c'] == 'v':
print morph