transpose.py

def transpose_corpus(corpus, output, limit):
    """ Transpose a corpus in Matrix Market format.
    """
    corpus = corpora.MmCorpus(corpus)
    corpus2 = [list() for i in range(corpus.num_terms)]
    for (i, doc) in enumerate(corpus):
        for (j, count) in doc:
            corpus2[j].append((i, count))
        if limit > 0 and i >= limit:
            break
    corpora.MmCorpus.serialize(output, corpus2)