def printDiffs(wordcount, clusters):
diffCounts = {}
for clusterSet in clusters.values():
for cluster in clusterSet:
if len(cluster) > 1:
ranked = sorted(list(cluster), key=wordcount.get, reverse=True)
diffs = StringDistance.pythonDiff(ranked[0], ranked[1], path=True)
for diff in diffs:
if not diff in diffCounts:
diffCounts[diff] = 0
diffCounts[diff] += 1
for diff, ct in diffCounts.iteritems():
print(diff, ct, sep='\t')
def printEdits(wordcount, clusters):
editCounts = {}
for clusterSet in clusters.values():
for cluster in clusterSet:
if len(cluster) > 1:
ranked = sorted(list(cluster), key=wordcount.get, reverse=True)
edits = StringDistance.levenshtein(ranked[0], ranked[1], path=True)
for edit in edits[0]:
if not edit in editCounts:
editCounts[edit] = 0
editCounts[edit] += 1
for edit, ct in editCounts.iteritems():
print(edit, ct, sep='\t')
