def report_compare(compare_result, detailed=True):
compare, base_cmap_data, target_cmap_data = compare_result
base_map = cmap_data.create_map_from_table(base_cmap_data.table)
target_map = cmap_data.create_map_from_table(target_cmap_data.table)
base_title = title_from_metadata(base_cmap_data.meta)
target_title = title_from_metadata(target_cmap_data.meta)
print 'base: %s' % base_title
print 'target: %s' % target_title
for script in sorted(compare):
added, removed = compare[script]
if added or removed:
name = base_map[script].name
print '%s # %s' % (script, name)
if added:
print ' added (%d): %s' % (
len(added), lint_config.write_int_ranges(added))
if detailed:
for cp in sorted(added):
print ' %6s %s' % (
'%04x' % cp, unicode_data.name(cp, ''))
if removed:
print ' removed (%d): %s' % (
len(removed), lint_config.write_int_ranges(removed))
if detailed:
for cp in sorted(removed):
print ' %6s %s' % (
'%04x' % cp, unicode_data.name(cp, ''))
def compare_cmap_data(base_cmap_data, target_cmap_data, scripts, cps,
except_scripts, except_cps, no_additions, no_removals):
result = {}
base_map = cmap_data.create_map_from_table(base_cmap_data.table)
target_map = cmap_data.create_map_from_table(target_cmap_data.table)
for script in sorted(base_map):
if except_scripts and script in except_scripts:
continue
if scripts and script not in scripts:
continue
if script in target_map:
name = base_map[script].name
base_cps = lint_config.parse_int_ranges(base_map[script].ranges)
target_cps = lint_config.parse_int_ranges(target_map[script].ranges)
if cps:
base_cps &= cps
target_cps &= cps
if except_cps:
base_cps -= except_cps
target_cps -= except_cps
if base_cps != target_cps:
added = None if no_additions else target_cps - base_cps
removed = None if no_removals else base_cps - target_cps
else:
added, removed = None, None
result[script] = (added, removed)
return result
def compare_cmap_data(base_cmap_data, target_cmap_data, scripts, cps,
except_scripts, except_cps, no_additions, no_removals):
result = {}
base_map = cmap_data.create_map_from_table(base_cmap_data.table)
target_map = cmap_data.create_map_from_table(target_cmap_data.table)
for script in sorted(base_map):
if except_scripts and script in except_scripts:
continue
if scripts and script not in scripts:
continue
if script in target_map:
name = base_map[script].name
base_cps = lint_config.parse_int_ranges(base_map[script].ranges)
target_cps = lint_config.parse_int_ranges(target_map[script].ranges)
added, removed = compare_cmaps(
base_cps, target_cps, cps, except_cps, no_additions, no_removals)
base_xcps, target_xcps = (set(), set())
if int(getattr(base_map[script], 'xcount', -1)) != -1:
base_xcps = lint_config.parse_int_ranges(base_map[script].xranges)
if int(getattr(target_map[script], 'xcount', -1)) != -1:
target_xcps = lint_config.parse_int_ranges(target_map[script].xranges)
xadded, xremoved = compare_cmaps(
base_xcps, target_xcps, cps, except_cps, no_additions, no_removals)
result[script] = added, removed, xadded, xremoved
return result
def report_compare(compare_result, detailed=True):
compare, base_cmap_data, target_cmap_data = compare_result
base_map = cmap_data.create_map_from_table(base_cmap_data.table)
target_map = cmap_data.create_map_from_table(target_cmap_data.table)
inverted_target = collections.defaultdict(set)
for script, row in target_map.iteritems():
cps = tool_utils.parse_int_ranges(row.ranges)
for cp in cps:
inverted_target[cp].add(script)
base_title = title_from_metadata(base_cmap_data.meta)
target_title = title_from_metadata(target_cmap_data.meta)
print 'base: %s' % base_title
print 'target: %s' % target_title
for script in sorted(compare):
added, removed = compare[script]
if added or removed:
name = base_map[script].name
print '%s # %s' % (script, name)
if added:
print ' added (%d): %s' % (
len(added), lint_config.write_int_ranges(added))
if detailed:
_print_detailed(added)
if removed:
print ' removed (%d): %s' % (
len(removed), lint_config.write_int_ranges(removed))
if detailed:
_print_detailed(removed, inverted_target)
def csv_from_cmap_data(data, scripts, exclude_scripts):
script_to_rowdata = cmap_data.create_map_from_table(data.table)
cols = []
max_lines = 0
num_cells = 0
for script in sorted(
script_to_rowdata, key=lambda s: _script_to_name(s).lower()):
if scripts and script not in scripts:
continue
if script in exclude_scripts:
continue
col = [
'"%s"' % _script_to_name(script)
]
rd = script_to_rowdata[script]
cps = tool_utils.parse_int_ranges(rd.ranges)
num_cells += len(cps)
col.extend('U+%04X' % cp for cp in sorted(
tool_utils.parse_int_ranges(rd.ranges)))
cols.append(col)
max_lines = max(max_lines, len(col))
num_cols = len(cols)
num_cells += num_cols # headers are not empty
all_cells = num_cols * max_lines
fmt = 'Columns: %d\nRows: %d\nNon-empty cells: %d\nCells: %d'
print >> sys.stderr, fmt % (num_cols, max_lines, num_cells, all_cells)
cmap_lines = []
cmap_lines.append(','.join(col[0] for col in cols))
for i in range(1, max_lines):
cmap_lines.append(','.join(col[i] if i < len(col) else '' for col in cols))
return '\n'.join(cmap_lines)
def compare_cmap_data(
base_cmap_data, target_cmap_data, scripts, except_scripts, opts=None):
result = {}
base_map = cmap_data.create_map_from_table(base_cmap_data.table)
target_map = cmap_data.create_map_from_table(target_cmap_data.table)
for script in sorted(base_map):
if except_scripts and script in except_scripts:
continue
if scripts and script not in scripts:
continue
if script in target_map:
base_data = base_map[script]
target_data = target_map[script]
result[script] = compare_data(base_data, target_data, opts)
return result
def compare_interscript_data(base_cmap_data, scripts, opts=None):
result = {}
base_map = cmap_data.create_map_from_table(base_cmap_data.table)
for i in range(len(scripts) - 1):
base_script = scripts[i]
for j in range(i + 1, len(scripts)):
target_script = scripts[j]
result[base_script, target_script] = compare_interscript(
base_map, base_script, target_script, opts)
return result
def report_compare(compare_result, detailed=True):
compare, base_cmap_data, target_cmap_data = compare_result
base_map = cmap_data.create_map_from_table(base_cmap_data.table)
target_map = cmap_data.create_map_from_table(target_cmap_data.table)
inverted_target = collections.defaultdict(set)
for script, row in target_map.iteritems():
cps = tool_utils.parse_int_ranges(row.ranges)
for cp in cps:
inverted_target[cp].add(script)
base_title = title_from_metadata(base_cmap_data.meta)
target_title = title_from_metadata(target_cmap_data.meta)
print 'base: %s' % base_title
print 'target: %s' % target_title
for script in sorted(compare):
added, removed, xadded, xremoved = compare[script]
label = '%s # %s' % (script, base_map[script].name)
report_cmap_compare(
label, added, removed, xadded, xremoved, inverted_target, detailed)
def report_interscript_compare(compare_result, detailed=True):
compare, base_cmap_data = compare_result
base_map = cmap_data.create_map_from_table(base_cmap_data.table)
title = title_from_metadata(base_cmap_data.meta)
print 'data: %s' % title
for t in sorted(compare):
added, removed, xadded, xremoved = compare[t]
base_script, target_script = t
label = 'base: %s (%s), target: %s (%s)' % (
base_script, base_map[base_script].name, target_script,
base_map[target_script].name)
report_cmap_compare(
label, added, removed, xadded, xremoved, None, detailed)
def csv_from_cmap_data(data, scripts, exclude_scripts):
script_to_rowdata = cmap_data.create_map_from_table(data.table)
cols = []
max_lines = 0
num_cells = 0
for script in sorted(
script_to_rowdata, key=lambda s: _script_to_name(s).lower()):
if scripts and script not in scripts:
continue
if exclude_scripts and script in exclude_scripts:
continue
rd = script_to_rowdata[script]
star = int(getattr(rd, 'xcount', -1)) != -1
col = [
'"%s%s"' % (_script_to_name(script), '*' if star else '')
]
cps = tool_utils.parse_int_ranges(rd.ranges)
xranges = getattr(rd, 'xranges', None)
if xranges != None:
xcps = frozenset(tool_utils.parse_int_ranges(xranges))
cps |= xcps
else:
xcps = frozenset()
num_cells += len(cps)
col.extend(
'%04X%s' % (cp, '*' if cp in xcps else '')
for cp in sorted(cps))
cols.append(col)
max_lines = max(max_lines, len(col))
num_cols = len(cols)
num_cells += num_cols # headers are not empty
all_cells = num_cols * max_lines
fmt = 'Columns: %d\nRows: %d\nNon-empty cells: %d\nCells: %d'
print >> sys.stderr, fmt % (num_cols, max_lines, num_cells, all_cells)
cmap_lines = []
cmap_lines.append(','.join(col[0] for col in cols))
for i in range(1, max_lines):
cmap_lines.append(','.join(col[i] if i < len(col) else '' for col in cols))
return '\n'.join(cmap_lines)
def csv_from_cmap_data(data, scripts, exclude_scripts):
script_to_rowdata = cmap_data.create_map_from_table(data.table)
cols = []
max_lines = 0
num_cells = 0
for script in sorted(script_to_rowdata,
key=lambda s: _script_to_name(s).lower()):
if scripts and script not in scripts:
continue
if exclude_scripts and script in exclude_scripts:
continue
rd = script_to_rowdata[script]
star = int(getattr(rd, 'xcount', -1)) != -1
col = ['"%s%s"' % (_script_to_name(script), '*' if star else '')]
cps = tool_utils.parse_int_ranges(rd.ranges)
xranges = getattr(rd, 'xranges', None)
if xranges != None:
xcps = frozenset(tool_utils.parse_int_ranges(xranges))
cps |= xcps
else:
xcps = frozenset()
num_cells += len(cps)
col.extend('U+%04X%s' % (cp, '*' if cp in xcps else '')
for cp in sorted(cps))
cols.append(col)
max_lines = max(max_lines, len(col))
num_cols = len(cols)
num_cells += num_cols # headers are not empty
all_cells = num_cols * max_lines
fmt = 'Columns: %d\nRows: %d\nNon-empty cells: %d\nCells: %d'
print >> sys.stderr, fmt % (num_cols, max_lines, num_cells, all_cells)
cmap_lines = []
cmap_lines.append(','.join(col[0] for col in cols))
for i in range(1, max_lines):
cmap_lines.append(','.join(col[i] if i < len(col) else ''
for col in cols))
return '\n'.join(cmap_lines)