def write_vector(vector, path):
#no incluye colnames
aux = []
for v in vector[1:]:
aux.append(str(v[0]))
aux = '\n'.join(aux)
save_data(path, aux)
data2 = transform_raw_charset(all_cs, data2)
all_cs = None
delta = {} # {cs_from:{cs_to:[subs], ...}, ...}
for i in range(1, max(max(data1), max(data2)) + 1):
if i in data1:
cs_from = data1[i]
del data1[i]
else:
cs_from = 0 #frozenset([]) # creation
if i in data2:
cs_to = data2[i]
del data2[i]
else:
cs_to = 0 #frozenset([]) # deletion
if cs_from not in delta:
delta[cs_from] = {cs_to: [i]}
else:
if cs_to not in delta[cs_from]:
delta[cs_from][cs_to] = [i]
else:
delta[cs_from][cs_to].append(i)
buff = ''
for f in delta:
for t in delta[f]:
buff += '%d\t%d\t%d\n' % (f, t, len(delta[f][t]))
save_data(out, buff)
def write_vector(vector, path):
aux = [str(v) for v in vector]
aux = '\n'.join(aux)
save_data(path, aux)
if __name__ == '__main__':
"""create dataset.tsv. Matrix with charsets vs dates"""
cs = load_all_cs_file(all_charset_path()) #fronzenset:num
cs = {cs[c]:c for c in cs} #num:frozenset
cs = [str(cs[i]) for i in range(len(cs))]
ret = []
for date in DATES:
ti = time()
path = get_transformed_charset_path(date)
data = load_transformed_charset(path)
cur = [0]*len(cs)
for d in data:
cur[d[0]] = d[1]
cur = [str(c) for c in cur]
cur.insert(0, str(date))
ret.append(cur)
tf = time()
print(tf-ti, date)
cs.insert(0, 'charset')
ret.insert(0, cs)
ret = [list(_) for _ in zip(*ret)]
ret = ['\t'.join(r) for r in ret]
ret = '\n'.join(ret)
save_data(tsv_path(), ret)
def write_names(vector, path):
#no incluye colnames
aux = '\n'.join(vector[1:])
save_data(path, aux)