def compressed_case_mapping_lines(mappings):
values_per_line = 12
bytes_ = []
for t in mappings:
lzw.add_cp(bytes_, int(t[0], 16))
lzw.add_short(bytes_, t[1][0])
lzw.add_short(bytes_, t[1][1])
compressed_bytes = lzw.compress(bytes_)
print 'rewrote {} * 64 = {} bits as {} * 8 = {} bits'.format(len(mappings), len(mappings)*64, len(bytes_), len(bytes_)*8)
print 'compressed to {} * 16 = {} bits'.format(len(compressed_bytes), len(compressed_bytes) * 16)
return lzw.compressed_bytes_to_lines(compressed_bytes, values_per_line)
def compressed_case_mapping_to_lines(mappings):
values_per_line = 12
bytes_ = []
for t in mappings:
lzw.add_short(bytes_, t[0][0])
lzw.add_short(bytes_, t[0][1])
try:
x = case_conditions[t[1]] # TODO: Totally wrong! Just here for size eval.
except:
x = 0
lzw.add_short(bytes_, x)
compressed_bytes = lzw.compress(bytes_)
print 'rewrote {} * 48 = {} bits as {} * 8 = {} bits'.format(len(mappings), len(mappings)*48, len(bytes_), len(bytes_)*8)
print 'compressed to {} * 16 = {} bits'.format(len(compressed_bytes), len(compressed_bytes) * 16)
return lzw.compressed_bytes_to_lines(compressed_bytes, values_per_line)
if cp in cccs_dict:
ccc = cccs_dict[cp]
nfd_quick_check = 'quick_check::yes'
if cp in quick_check_maps['NFD']:
nfd_quick_check = quick_check_maps['NFD'][cp]
nfkd_quick_check = 'quick_check::yes'
if cp in quick_check_maps['NFKD']:
nfkd_quick_check = quick_check_maps['NFKD'][cp]
nfc_quick_check = 'quick_check::yes'
if cp in quick_check_maps['NFC']:
nfc_quick_check = quick_check_maps['NFC'][cp]
nfkc_quick_check = 'quick_check::yes'
if cp in quick_check_maps['NFKC']:
nfkc_quick_check = quick_check_maps['NFKC'][cp]
lzw.add_cp(prop_bytes_, cp)
lzw.add_short(prop_bytes_, canonical_decomp[0])
lzw.add_short(prop_bytes_, canonical_decomp[1])
lzw.add_short(prop_bytes_, compatible_decomp[0])
lzw.add_short(prop_bytes_, compatible_decomp[1])
lzw.add_byte(prop_bytes_, int(ccc))
lzw.add_byte(prop_bytes_, \
quick_checks_to_byte(nfd_quick_check, nfkd_quick_check))
lzw.add_byte(prop_bytes_, \
quick_checks_to_byte(nfc_quick_check, nfkc_quick_check))
value_per_line = 12
compressed_bytes = lzw.compress(prop_bytes_)
props_lines, num_shorts = lzw.compressed_bytes_to_lines(
compressed_bytes, value_per_line)
#print 'rewrote {} * 144 = {} bits as {} * 8 = {} bits'.format(len(all_cps), len(all_cps)*144, len(prop_bytes_), len(prop_bytes_)*8)
#print 'compressed to {} * 16 = {} bits'.format(len(compressed_bytes), len(compressed_bytes) * 16)
to_reorder_group(g[0], g[1], simple, lead_byte
in compressible_lead_bytes))
reorder_group_str = ' ' + ',\n '.join(reorder_group_strings) + ',\n'
values_per_line = 12
ce_bytes = []
for ce in collation_elements:
x = '0' + ''.join(ce[0])
if ce[0] != ('', ):
x += '0' * (4 - len(ce[0])) * 2
lzw.add_int(ce_bytes, int(x, 16))
x = '0' + ''.join(ce[1])
if ce[1] != ('', ):
x += '0' * (2 - len(ce[1])) * 2
lzw.add_short(ce_bytes, int(x, 16))
x = '0' + ''.join(ce[2])
if ce[2] != ('', ):
x += '0' * (2 - len(ce[2])) * 2
lzw.add_short(ce_bytes, int(x, 16))
compressed_ces = lzw.compress(ce_bytes)
def values_to_lines(values, value_type, values_per_chunk):
retval = ''
chunk_form = '''\
#ifdef _MSC_VER
{{
std::array<{0}, {1}> values {{{{
#endif
{2}
#ifdef _MSC_VER