import gensim
import sqlite3bowcorpus

class LsiWrapper(object):

    def __init__(self, corpus, lsi_model_file, num_topics):
        self.corpus = corpus
        self.lsi_model_file = lsi_model_file
        self.lsimodel = gensim.models.lsimodel.LsiModel(id2word = self.corpus.tid_to_term, numTopics = num_topics)
        self.lsimodel.load(lsi_model_file)

    def topic_space(self, text):
        '''Compute the topic space representation for bag of words representation,
        bow.
        '''
        bow = self.corpus.text2bow(text)
        tranformation = self.lsimodel[[bow]]
        return list(tranformation[0][0])

def compute_lsi(dbname, corpus_model_file, num_topics, savefile):
    corpus = sqlite3bowcorpus.Sqlite3BowCorpus(dbname, corpus_model_file)
    corpus.request_documents_cached()
    lsimodel = gensim.models.lsimodel.LsiModel(corpus, num_topics, corpus.tid_to_term)
    lsimodel.save(savefile)

def compute_lsi_main():
    dbname = sys.argv[1]
    corpus_model_file = sys.argv[2]
    num_topics = int(sys.argv[3])
    savefile = sys.argv[4]
    compute_lsi(dbname, corpus_model_file, num_topics, savefile)

def print_lsi_main():
    dbname = sys.argv[1]
    corpus_model_file = sys.argv[2]
    num_topics = int(sys.argv[3])
    lsi_model_file = sys.argv[4]
    corpus = sqlite3bowcorpus.Sqlite3BowCorpus(dbname, corpus_model_file)
    lsi_model = LsiWrapper(corpus, lsi_model_file, num_topics)
    lsi_model.topic_space(sys.argv[5])

if __name__ == '__main__':
    import sys
    print_lsi_main()