def __init__(self, chars_file, corpus_dir=None, length=None, random=False):
self.corpus_dir = corpus_dir
self.length = length
self.corpus = []
self.random = random
self.chars_file = chars_file
self.charsets = load_chars(chars_file)
self.load()
def restore_exist_labels(label_path):
# 如果目标目录存在 labels.txt 则向该目录中追加图片
start_index = 0
if os.path.exists(label_path):
start_index = len(utils.load_chars(label_path))
print('增量生成图片 %s. 从 %d 开始' % (flags.save_dir, start_index))
else:
print('生成图片 %s' % flags.save_dir)
return start_index
def restore_exist_labels(label_path):
# 如果目标目录存在 labels.txt 则向该目录中追加图片
start_index = 0
if os.path.exists(label_path):
start_index = len(utils.load_chars(label_path))
print('Generate more text images in %s. Start index %d' % (flags.save_dir, start_index))
else:
print('Generate text images in %s' % flags.save_dir)
return start_index
def __init__(self, chars_file, corpus_dir=None, length=None):
self.corpus_dir = corpus_dir
self.length = length
self.corpus = [] # 语料库-字符串列表
self.chars_file = chars_file
self.charsets = load_chars(chars_file)
self.load()
def __init__(self, chars_file, corpus_dir=None, length=None):
self.corpus_dir = corpus_dir
self.length = length
self.charsets = load_chars(chars_file)
if not isinstance(self, RandomCorpus):
print("Loading corpus from: " + self.corpus_dir)
self.load()
def __init__(self, chars_file):
self.chars = ''.join(load_chars(chars_file))
# char_set_length + ctc_blank
self.num_classes = len(self.chars) + 1
self.encode_maps = {}
self.decode_maps = {}
self.create_encode_decode_maps(self.chars)
print('Load chars file: %s num_classes: %d + 1(CTC Black)' % (chars_file, self.num_classes - 1))
def __init__(self,
chars_file,
corpus_dir=None,
length=None,
max_length=None,
distribute_file=None):
self.corpus_dir = corpus_dir
self.length = length
self.corpus = []
self.chars_file = chars_file
self.charsets = load_chars(chars_file)
self.max_length = max_length
self.distribute_file = distribute_file
self.load()
def __init__(self, chars_file, corpus_dir=None, length=None):
self.corpus_dir = corpus_dir
self.length = length
self.corpus = []
self.chars_file = chars_file
self.charsets = load_chars(chars_file)
if not isinstance(self, RandomCorpus):
print("Loading corpus from: " + self.corpus_dir)
self.corpus_path = glob.glob(self.corpus_dir + '/**/*.txt',
recursive=True)
if len(self.corpus_path) == 0:
print("Corpus not found.")
exit(-1)
self.load()
def get_unsupported_chars(fonts, chars_file):
"""
Get fonts unsupported chars by loads/saves font supported chars from cache file
:param fonts:
:param chars_file:
:return: dict
key -> font_path
value -> font unsupported chars
"""
charset = load_chars(chars_file)
charset = ''.join(charset)
fonts_chars = get_fonts_chars(fonts, chars_file)
fonts_unsupported_chars = {}
for font_path, chars in fonts_chars.items():
unsupported_chars = list(filter(lambda x: x not in chars, charset))
fonts_unsupported_chars[font_path] = unsupported_chars
return fonts_unsupported_chars
def get_fonts_chars(fonts, chars_file):
"""
loads/saves font supported chars from cache file
:param fonts: list of font path. e.g ['./data/fonts/msyh.ttc']
:param chars_file: arg from parse_args
:return: dict
key -> font_path
value -> font supported chars
"""
out = {}
cache_dir = os.path.abspath(
os.path.join(os.path.dirname(__file__), '../', '.caches'))
if not os.path.exists(cache_dir):
os.makedirs(cache_dir)
chars = load_chars(chars_file)
chars = ''.join(chars)
for font_path in fonts:
string = ''.join([font_path, chars])
file_md5 = md5(string)
cache_file_path = os.path.join(cache_dir, file_md5)
if not os.path.exists(cache_file_path):
ttf = load_font(font_path)
_, supported_chars = check_font_chars(ttf, chars)
print('Save font(%s) supported chars(%d) to cache' %
(font_path, len(supported_chars)))
with open(cache_file_path, 'wb') as f:
pickle.dump(supported_chars, f, pickle.HIGHEST_PROTOCOL)
else:
with open(cache_file_path, 'rb') as f:
supported_chars = pickle.load(f)
print('Load font(%s) supported chars(%d) from cache' %
(font_path, len(supported_chars)))
out[font_path] = supported_chars
return out
type=str,
default='./data/chars/char_std_5990.txt')
parser.add_argument(
'--font_dir',
type=str,
default='/Users/guoxiaolu/work/code/text_renderer/data/fonts/chn')
parser.add_argument(
'--delete',
action="store_true",
default=False,
help=
'whether or not to delete font which not full support the chars_file')
args, _ = parser.parse_known_args()
charset = load_chars(args.chars_file)
font_paths = glob.glob(args.font_dir + '/*.*')
fonts = {}
for p in font_paths:
ttf = load_font(p)
fonts[p] = ttf
useful_fonts = []
for k, v in fonts.items():
try:
unsupported_chars, _ = check_font_chars(v, charset)
print("font: %s ,chars unsupported: %d" %
(k, len(unsupported_chars)))
if len(unsupported_chars) < 500:
print(unsupported_chars)
