def _summarize_blocks(start, limit, defined_cps, cp_to_scripts, all_scripts):
block = None
block_count = 0
defined_count = 0
script_counts = None
for cp in range(start, limit):
cp_block = unicode_data.block(cp)
if cp_block != block:
if block:
_summarize_block(
block, block_count, defined_count, script_counts)
block = cp_block
block_count = 0
defined_count = 0
script_counts = collections.defaultdict(int)
block_count += 1
is_defined = cp in defined_cps and cp not in _OMITTED
if not is_defined:
continue
defined_count += 1
scripts = _get_scripts(cp, cp_to_scripts)
for script in scripts:
script_counts[script] += 1
_summarize_block(block, block_count, defined_count, script_counts)
def _summarize_blocks(start, limit, defined_cps, cp_to_scripts, all_scripts):
block = None
block_count = 0
defined_count = 0
script_counts = None
for cp in range(start, limit):
cp_block = unicode_data.block(cp)
if cp_block != block:
if block:
_summarize_block(
block, block_count, defined_count, script_counts)
block = cp_block
block_count = 0
defined_count = 0
script_counts = collections.defaultdict(int)
block_count += 1
is_defined = cp in defined_cps and cp not in _OMITTED
if not is_defined:
continue
defined_count += 1
scripts = _get_scripts(cp, cp_to_scripts)
for script in scripts:
script_counts[script] += 1
_summarize_block(block, block_count, defined_count, script_counts)
def _list_blocks(start, limit, defined_cps, cp_to_scripts, all_scripts,
only_scripts, details):
start_cp = -1
defined_count = 0
block = None
showed_block = False
scripts = None
skip_empty = bool(only_scripts)
for cp in range(start, limit):
is_defined = cp in defined_cps
cp_block = unicode_data.block(cp)
cp_scripts = _get_scripts(cp, cp_to_scripts) if is_defined else None
if cp_block != block or (cp_scripts and scripts
and cp_scripts != scripts):
if block and block != 'No_Block':
if not (skip_empty and _is_empty_scripts(scripts)):
if not showed_block:
print('...') if block == 'No_Block' else block
showed_block = True
_list_range(start_cp, cp, defined_cps, defined_count,
scripts, all_scripts, only_scripts, details)
start_cp = cp
defined_count = 0
if cp_block != block:
block = cp_block
showed_block = False
scripts = None
if is_defined:
scripts = cp_scripts
defined_count += 1
if not (skip_empty and _is_empty_scripts(scripts)):
if not showed_block:
print('...') if block == 'No_Block' else block
_list_range(start_cp, limit, defined_cps, defined_count, scripts,
all_scripts, only_scripts, details)
def _print_detailed(cps, inverted_target=None):
last_block = None
for cp in sorted(cps):
block = unicode_data.block(cp)
if block != last_block:
print ' %s' % block
last_block = block
script = unicode_data.script(cp)
extensions = unicode_data.script_extensions(cp) - set([script])
if extensions:
extensions = ' (%s)' % ','.join(sorted(extensions))
else:
extensions = ''
if not inverted_target:
extra = ''
elif cp not in inverted_target:
extra = ' !missing'
else:
scripts = sorted(inverted_target[cp])
if len(scripts) > 3:
script_text = ', '.join(scripts[:3]) + '... ' + scripts[-1]
else:
script_text = ', '.join(scripts)
extra = ' (in %s)' % script_text
print ' %6s %4s %2s %3s %s%s%s' % (
'%04x' % cp,
script,
unicode_data.category(cp),
unicode_data.age(cp),
unicode_data.name(cp, ''),
extensions,
extra)
def _print_detailed(cps, inverted_target=None):
last_block = None
undefined_start = -1
undefined_end = -1
def show_undefined(start, end):
if start >= 0:
if end > start:
print(" %04x-%04x Zzzz <%d undefined>" %
(start, end, end - start - 1))
else:
print(" %04x Zzzz <1 undefined>" % start)
for cp in sorted(cps):
block = unicode_data.block(cp)
if block != last_block or (undefined_end > -1
and cp > undefined_end + 1):
show_undefined(undefined_start, undefined_end)
undefined_start, undefined_end = -1, -1
if block != last_block:
print(" %s" % block)
last_block = block
script = unicode_data.script(cp)
if script == "Zzzz":
if undefined_start >= 0:
undefined_end = cp
else:
undefined_start, undefined_end = cp, cp
continue
show_undefined(undefined_start, undefined_end)
undefined_start, undefined_end = -1, -1
extensions = unicode_data.script_extensions(cp) - {script}
if extensions:
extensions = " (script %s)" % ", ".join(sorted(extensions))
else:
extensions = ""
if not inverted_target:
extra = ""
elif cp not in inverted_target:
extra = " !missing"
else:
scripts = sorted(inverted_target[cp])
if len(scripts) > 3:
script_text = ", ".join(scripts[:3]) + "... " + scripts[-1]
else:
script_text = ", ".join(scripts)
extra = " (font %s)" % script_text
print(" %6s %4s %2s %3s %s%s%s" % (
"%04x" % cp,
script,
unicode_data.category(cp),
unicode_data.age(cp),
unicode_data.name(cp, ""),
extensions,
extra,
))
show_undefined(undefined_start, undefined_end)
def show_cps_by_block(cps):
print('%d missing codepoints' % len(cps))
block = None
for cp in sorted(cps):
new_block = unicode_data.block(cp)
if new_block != block:
print('# %s' % new_block)
block = new_block
print('%5s %s' % ('%04x' % cp, unicode_data.name(cp)))
def show_cps_by_block(cps):
print("%d missing codepoints" % len(cps))
block = None
for cp in sorted(cps):
new_block = unicode_data.block(cp)
if new_block != block:
print("# %s" % new_block)
block = new_block
print("%5s %s" % ("%04x" % cp, unicode_data.name(cp)))
def show_cps_by_block(cps):
print '%d missing codepoints' % len(cps)
block = None
for cp in sorted(cps):
new_block = unicode_data.block(cp)
if new_block != block:
print '# %s' % new_block
block = new_block
print '%5s %s' % ('%04x' % cp, unicode_data.name(cp))
def _print_detailed(cps, inverted_target=None):
last_block = None
undefined_start = -1
undefined_end = -1
def show_undefined(start, end):
if start >= 0:
if end > start:
print ' %04x-%04x Zzzz <%d undefined>' % (
start, end, end - start - 1)
else:
print ' %04x Zzzz <1 undefined>' % start
for cp in sorted(cps):
block = unicode_data.block(cp)
if block != last_block or (undefined_end > -1 and cp > undefined_end + 1):
show_undefined(undefined_start, undefined_end)
undefined_start, undefined_end = -1, -1
if block != last_block:
print ' %s' % block
last_block = block
script = unicode_data.script(cp)
if script == 'Zzzz':
if undefined_start >= 0:
undefined_end = cp
else:
undefined_start, undefined_end = cp, cp
continue
show_undefined(undefined_start, undefined_end)
undefined_start, undefined_end = -1, -1
extensions = unicode_data.script_extensions(cp) - set([script])
if extensions:
extensions = ' (%s)' % ','.join(sorted(extensions))
else:
extensions = ''
if not inverted_target:
extra = ''
elif cp not in inverted_target:
extra = ' !missing'
else:
scripts = sorted(inverted_target[cp])
if len(scripts) > 3:
script_text = ', '.join(scripts[:3]) + '... ' + scripts[-1]
else:
script_text = ', '.join(scripts)
extra = ' (in %s)' % script_text
print ' %6s %4s %2s %3s %s%s%s' % (
'%04x' % cp,
script,
unicode_data.category(cp),
unicode_data.age(cp),
unicode_data.name(cp, ''),
extensions,
extra)
show_undefined(undefined_start, undefined_end)
def _list_blocks(
start, limit, defined_cps, cp_to_scripts, all_scripts, only_scripts,
details):
start_cp = -1
defined_count = 0
block = None
showed_block = False
scripts = None
skip_empty = bool(only_scripts)
for cp in range(start, limit):
is_defined = cp in defined_cps
cp_block = unicode_data.block(cp)
cp_scripts = _get_scripts(cp, cp_to_scripts) if is_defined else None
if cp_block != block or (
cp_scripts and scripts and cp_scripts != scripts):
if block and block != 'No_Block':
if not (skip_empty and _is_empty_scripts(scripts)):
if not showed_block:
print '...' if block == 'No_Block' else block
showed_block = True
_list_range(
start_cp, cp, defined_cps, defined_count, scripts, all_scripts,
only_scripts, details)
start_cp = cp
defined_count = 0
if cp_block != block:
block = cp_block
showed_block = False
scripts = None
if is_defined:
scripts = cp_scripts
defined_count += 1
if not (skip_empty and _is_empty_scripts(scripts)):
if not showed_block:
print '...' if block == 'No_Block' else block
_list_range(
start_cp, limit, defined_cps, defined_count, scripts, all_scripts,
only_scripts, details)
def test_block(self):
"""Tests the block() method."""
self.assertEqual('Emoticons', unicode_data.block(0x1F600))
def test_block(self):
"""Tests the block() method."""
self.assertEqual('Emoticons', unicode_data.block(0x1F600))
