breeze/ChatTTS-ui
1
0
Fork 0
mirror of https://github.com/jianchang512/ChatTTS-ui.git synced 2026-01-12 00:06:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #168 from zuyu/regex-fix

Browse Source 
Fixed regex SyntaxWarnings
This commit is contained in:
okmyworld
2024-06-30 17:16:55 +08:00
committed by GitHub
parent bc2a3e9539 3e08da6beb
commit ab4c6e1aa2
3 changed files with 41 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

48
uilib/utils.py
View File

@@ -31,7 +31,7 @@ def get_parameter(request, param, default, cast_type):
# 数字转为英文读法
def num_to_english(num):
num_str = str(num)
# English representations for numbers 0-9
english_digits = ["zero", "one", "two", "three", "four", "five", "six", "seven", "eight", "nine"]
@@ -41,29 +41,29 @@ def num_to_english(num):
need_and = False # Indicates whether 'and' needs to be added
part = [] # Stores each group of 4 digits
is_first_part = True # Indicates if it is the first part for not adding 'and' at the beginning
# Split the number into 3-digit groups
while num_str:
part.append(num_str[-3:])
num_str = num_str[:-3]
part.reverse()
for i, p in enumerate(part):
p_str = ""
digit_len = len(p)
if int(p) == 0 and i < len(part) - 1:
continue
hundreds_digit = int(p) // 100 if digit_len == 3 else None
tens_digit = int(p) % 100 if digit_len >= 2 else int(p[0] if digit_len == 1 else p[1])
# Process hundreds
if hundreds_digit is not None and hundreds_digit != 0:
p_str += english_digits[hundreds_digit] + " hundred"
if tens_digit != 0:
p_str += " and "
# Process tens and ones
if 10 < tens_digit < 20: # Teens exception
teen_map = {
@@ -79,17 +79,17 @@ def num_to_english(num):
p_str += tens_map[tens_val] + (" " + english_digits[ones_val] if ones_val != 0 else "")
elif tens_digit != 0 and tens_val < 2: # When tens_digit is in [1, 9]
p_str += english_digits[tens_digit]
if p_str and not is_first_part and need_and:
result += " and "
result += p_str
if i < len(part) - 1 and int(p) != 0:
result += " " + big_units[len(part) - i - 1] + ", "
is_first_part = False
if int(p) != 0:
need_and = True
return result.capitalize()
@@ -120,8 +120,8 @@ def num2text(text):
text = re.sub(r'((?:\d+\.)?\d+)\s*/\s*(\d+)', fraction_to_words, text)
# 取出数字 number_list= [('1000200030004000.123', '1000200030004000', '123'), ('23425', '23425', '')]
number_list=re.findall('((\d+)(?:\.(\d+))?%?)',text)
if len(number_list)>0:
number_list=re.findall(r'((\d+)(?:\.(\d+))?%?)', text)
if len(number_list)>0:
#dc= ('1000200030004000.123', '1000200030004000', '123','')
for m,dc in enumerate(number_list):
if len(dc[1])>16:
@@ -133,14 +133,14 @@ def num2text(text):
int_text=f' the pronunciation of {int_text}'
text=text.replace(dc[0],int_text)
return text.replace('1',' one ').replace('2',' two ').replace('3',' three ').replace('4',' four ').replace('5',' five ').replace('6',' six ').replace('7','seven').replace('8',' eight ').replace('9',' nine ').replace('0',' zero ').replace('=',' equals ')
# 中英文数字转换为文字，特殊符号处理
def split_text(text_list):
tx = TextNormalizer()
haserror=False
result=[]
@@ -177,12 +177,12 @@ def split_text_by_punctuation(text):
min_length = 150
punctuation_marks = "。？！，、；：”’》」』）】…—"
english_punctuation = ".?!,:;)}…"
# 结果列表
result = []
# 起始位置
pos = 0
# 遍历文本中的每个字符
text_length=len(text)
for i, char in enumerate(text):
@@ -196,11 +196,11 @@ def split_text_by_punctuation(text):
# 更新起始位置到当前标点的下一个字符
pos = i+1
#print(f'{pos=},{len(text)=}')
# 如果剩余文本长度超过120或没有更多标点符号可以进行分割，将剩余的文本作为一个分段添加到结果列表
if pos < len(text):
result.append(text[pos:])
return result
@@ -223,17 +223,17 @@ def ClearWav(directory):
print(f"文件删除错误 {file_path}, 报错信息: {e}")
return False, str(e)
return True, "所有wav文件已被删除."
# 保存音色
# 保存音色
# 参考 https://github.com/craii/ChatTTS_WebUI/blob/main/utils.py
def save_speaker(name, tensor):
def save_speaker(name, tensor):
try:
df = pd.DataFrame({"speaker": [float(i) for i in tensor]})
df.to_csv(f"{SPEAKER_DIR}/{name}.csv", index=False, header=False)
except Exception as e:
print(e)
# 加载音色
# 参考 https://github.com/craii/ChatTTS_WebUI/blob/main/utils.py
def load_speaker(name):
@@ -311,4 +311,4 @@ def modelscope_status():
return False
except Exception as e:
return False
return True
return True

3
uilib/zh_normalization/quantifier.py
View File

@@ -58,7 +58,6 @@ def replace_temperature(match) -> str:
def replace_measure(sentence) -> str:
for q_notation in measure_dict:
if q_notation in sentence and re.search(f'\d{q_notation}',sentence):
if q_notation in sentence and re.search(r'\d{q_notation}', sentence):
sentence = sentence.replace(q_notation, measure_dict[q_notation])
return sentence

33
uilib/zh_normalization/text_normlization.py
View File

@@ -60,7 +60,7 @@ class TextNormalizer():
text (str): The input text.
Returns:
List[str]: Sentences.
character_map = {
"：": "，",
"；": "，",
@@ -105,8 +105,8 @@ class TextNormalizer():
return sentences
def _post_replace(self, sentence: str) -> str:
#sentence = sentence.replace('/', '每')
sentence = sentence.replace('~', '至')
sentence = sentence.replace('～', '至')
@@ -146,8 +146,8 @@ class TextNormalizer():
sentence = sentence.replace('ψ', '普赛').replace('Ψ', '普赛')
sentence = sentence.replace('ω', '欧米伽').replace('Ω', '欧米伽')
sentence = sentence.replace('+', '加')
# re filter special characters, have one more character "-" than line 68
sentence = re.sub(r'[-——《》【】<=>{}()（）#&@“”^|…\\]', '', sentence)
return sentence
@@ -161,12 +161,12 @@ class TextNormalizer():
result = ""
zero_flag = False # 标记是否需要加'零'
part = [] # 存储每4位的数字
# 将数字按每4位分组
while num_str:
part.append(num_str[-4:])
num_str = num_str[:-4]
for i in range(len(part)):
part_str = ""
part_zero_flag = False
@@ -184,21 +184,21 @@ class TextNormalizer():
part_str = part_str[:-1] # 去除尾部的'零'
if part_str:
zero_flag = True
if i > 0 and not set(part[i]) <= {'0'}: # 如果当前部分不全是0，则加上相应的大单位
result = part_str + big_units[i] + result
else:
result = part_str + result
# 处理输入为0的情况或者去掉开头的零
result = result.lstrip(chinese_digits[0])
if not result:
return chinese_digits[0]
return result
def normalize_sentence(self, sentence: str) -> str:
# basic character conversions
# add
sentence = re.sub(r'(\d+)\s*[\*xX]\s*(\d+)', r'\1 乘 \2', sentence,re.I)
@@ -207,12 +207,12 @@ class TextNormalizer():
sentence = re.sub(r'(0\d+)\-(\d{3,})', r'\1杠\2', sentence,re.I)
sentence = sentence.replace('=', '等于')
sentence = sentence.replace('÷','除以')
#sentence = re.sub(r'(\d+)\s*\-', r'\1 减', sentence)
sentence = re.sub(r'((?:\d+\.)?\d+)\s*/\s*(\d+)', r'\2分之\1', sentence)
# 取出数字 number_list= [('1000200030004000.123', '1000200030004000', '123'), ('23425', '23425', '')]
number_list=re.findall('((\d+)(?:\.(\d+))?%?)',sentence)
number_list=re.findall(r'((\d+)(?:\.(\d+))?%?)', sentence)
numtext=['零','一','二','三','四','五','六','七','八','九']
if len(number_list)>0:
#dc= ('1000200030004000.123', '1000200030004000', '123','')
@@ -227,8 +227,7 @@ class TextNormalizer():
if dc[0][-1]=='%':
int_text=f'百分之{int_text}'
sentence=sentence.replace(dc[0],int_text)
sentence = tranditional_to_simplified(sentence)
sentence = sentence.translate(F2H_ASCII_LETTERS).translate(
F2H_DIGITS).translate(F2H_SPACE)
@@ -258,7 +257,7 @@ class TextNormalizer():
sentence = RE_DEFAULT_NUM.sub(replace_default_num, sentence)
sentence = RE_NUMBER.sub(replace_number, sentence)
sentence = self._post_replace(sentence)
sentence = sentence.replace('[一break]','[1break]')
return sentence