208 lines
7.3 KiB
Python
208 lines
7.3 KiB
Python
import re
|
|
import os
|
|
import math
|
|
from enum import Enum
|
|
|
|
|
|
class ScriptType(str, Enum):
|
|
GENERIC = "generic"
|
|
SCN = "scn"
|
|
XML = "xml"
|
|
|
|
def __str__(self):
|
|
return self.value
|
|
|
|
@staticmethod
|
|
def from_string(s: str):
|
|
for stype in ScriptType:
|
|
if stype.value == s.lower():
|
|
return stype
|
|
raise ValueError(f"Unknown ScriptType: {s}")
|
|
|
|
def is_scn(self):
|
|
return self == ScriptType.SCN
|
|
|
|
def is_xml(self):
|
|
return self == ScriptType.XML
|
|
|
|
|
|
ELLIPSIS = re.compile(r'\.{3,}')
|
|
ELLIPSIS_ZH = re.compile(r'…+')
|
|
HYPHEN_ZH = re.compile(r'—+')
|
|
HALF_FULL_MAP = (
|
|
(',', '，'),
|
|
('.', '。'),
|
|
('!', '！'),
|
|
('?', '？'),
|
|
('~', '～'),
|
|
('(', '（'),
|
|
(')', '）'),
|
|
(':', '：'),
|
|
(';', '；'),
|
|
('-', '—'),
|
|
('ー', '—'),
|
|
('&', '＆'),
|
|
('%', '％'),
|
|
('・', '·'),
|
|
)
|
|
MIXED_RULE1 = re.compile(r'[，。、]+([…—～]+)')
|
|
MIXED_RULE2 = re.compile(r'([…—～]+)[，。、]+')
|
|
MIXED_RULE3 = re.compile(r'([…—～])[…—～]+')
|
|
MIXED_RULE4 = re.compile(r'([，。、]+)）$')
|
|
# 平假名
|
|
HIRAGANA = re.compile(r'[\u3040-\u309F]+')
|
|
# 片假名
|
|
KATAKANA = re.compile(r'[\u30A0-\u30FF]+')
|
|
# 狗屎wamsoft在scn里拉的指令
|
|
SCN_SPECIAL = re.compile(r'(%(?:[^%;]*;|r)|#[0-9a-fA-F]{6,8};)')
|
|
MUL_QUOTE_RULE = re.compile(r'"(.*)"')
|
|
SINGLE_QUOTE_RULE = re.compile(r"'(.*)'")
|
|
XML_TAG_SPECIAL = re.compile(r'(<[^>]+>)')
|
|
CJK_LEFT_QUOTE = ["「", "『"]
|
|
CJK_RIGHT_QUOTE = ["」", "』"]
|
|
NEED_REPLACED_LEFT_QUOTE = ["“", "‘"]
|
|
NEED_REPLACED_RIGHT_QUOTE = ["”", "’"]
|
|
|
|
|
|
def process_line(data: str) -> str:
|
|
# 使用省略号替换连续的点
|
|
data = ELLIPSIS.sub(lambda m: '…' * math.ceil(len(m.group(0)) / 3), data)
|
|
for half, full in HALF_FULL_MAP:
|
|
data = data.replace(half, full)
|
|
#，。、和…—~混用
|
|
data = MIXED_RULE1.sub(lambda m: m.group(1), data)
|
|
data = MIXED_RULE2.sub(lambda m: m.group(1), data)
|
|
# 混用时…—~只保留一个
|
|
data = MIXED_RULE3.sub(lambda m: m.group(1) * len(m.group(0)), data)
|
|
# 省略号和破折号统一只留一对
|
|
data = ELLIPSIS_ZH.sub('……', data)
|
|
data = HYPHEN_ZH.sub('——', data)
|
|
# 删除所有平假名
|
|
data = HIRAGANA.sub('', data)
|
|
# 删除所有片假名
|
|
data = KATAKANA.sub('', data)
|
|
data = data.replace('！？', '？！')
|
|
data = data.replace('。」', '」')
|
|
# 去除）前多余的标点
|
|
data = MIXED_RULE4.sub('）', data)
|
|
while MUL_QUOTE_RULE.search(data):
|
|
data = MUL_QUOTE_RULE.sub(lambda m: '「' + m.group(1) + '」', data)
|
|
while SINGLE_QUOTE_RULE.search(data):
|
|
data = SINGLE_QUOTE_RULE.sub(lambda m: '『' + m.group(1) + '』', data)
|
|
return data
|
|
|
|
|
|
def process_scn_line(data: str) -> str:
|
|
parts = SCN_SPECIAL.split(data)
|
|
for i in range(len(parts)):
|
|
if i % 2 == 0:
|
|
parts[i] = process_line(parts[i])
|
|
# 特殊处理 。%command;」
|
|
if i > 0 and parts[i].startswith('」'):
|
|
while parts[i - 2].endswith('。'):
|
|
parts[i - 2] = parts[i - 2][:-1]
|
|
return ''.join(parts)
|
|
|
|
|
|
def process_xml_line(data: str) -> str:
|
|
parts = XML_TAG_SPECIAL.split(data)
|
|
for i in range(len(parts)):
|
|
if i % 2 == 0:
|
|
parts[i] = process_line(parts[i])
|
|
if i > 0 and parts[i].startswith('」'):
|
|
while parts[i - 2].endswith('。'):
|
|
parts[i - 2] = parts[i - 2][:-1]
|
|
return ''.join(parts)
|
|
|
|
|
|
def replace_quote_str(s) -> str:
|
|
depth = 0
|
|
re = ""
|
|
for c in s:
|
|
if c in CJK_LEFT_QUOTE:
|
|
re += CJK_LEFT_QUOTE[depth % 2]
|
|
depth += 1
|
|
elif c in CJK_RIGHT_QUOTE:
|
|
depth -= 1
|
|
re += CJK_RIGHT_QUOTE[depth % 2]
|
|
elif c in NEED_REPLACED_LEFT_QUOTE:
|
|
re += CJK_LEFT_QUOTE[depth % 2]
|
|
depth += 1
|
|
elif c in NEED_REPLACED_RIGHT_QUOTE:
|
|
depth -= 1
|
|
re += CJK_RIGHT_QUOTE[depth % 2]
|
|
else:
|
|
re += c
|
|
return re
|
|
|
|
|
|
def process_m3t(m3t_path, type=ScriptType.GENERIC, llm_only=False, replace_quotes=True):
|
|
lines = []
|
|
line_num = 0
|
|
for line in open(m3t_path, "r", encoding="utf-8"):
|
|
line = line.strip()
|
|
line = line.replace('\u200b', '') # 移除零宽空格
|
|
if line_num == 0:
|
|
line = line.lstrip("\ufeff") # 移除可能的BOM
|
|
line_num += 1
|
|
if (not llm_only and line.startswith("●")) or line.startswith("△"):
|
|
data = line[1:].strip()
|
|
if type.is_scn():
|
|
ndata = process_scn_line(data)
|
|
elif type.is_xml():
|
|
ndata = process_xml_line(data)
|
|
else:
|
|
ndata = process_line(data)
|
|
if replace_quotes:
|
|
ndata = replace_quote_str(ndata)
|
|
if data != ndata:
|
|
print(f"{m3t_path}:{line_num}: {data} -> {ndata}")
|
|
line = line[0] + " " + ndata
|
|
lines.append(line)
|
|
return lines
|
|
|
|
|
|
def save_lines(m3t_path, lines):
|
|
with open(m3t_path, "w", encoding="utf-8") as f:
|
|
for line in lines:
|
|
f.write(line + "\n")
|
|
|
|
|
|
def process_m3t_file(m3t_path, target_path=None, dry_run=False, type=ScriptType.GENERIC, llm_only=False, replace_quotes=True):
|
|
lines = process_m3t(m3t_path, type, llm_only, replace_quotes)
|
|
if not dry_run:
|
|
if target_path is None:
|
|
save_lines(m3t_path, lines)
|
|
else:
|
|
os.makedirs(os.path.dirname(target_path), exist_ok=True)
|
|
save_lines(target_path, lines)
|
|
|
|
|
|
def recursive_process_m3t(m3t_path, target_path=None, dry_run=False, type=ScriptType.GENERIC, llm_only=False, ext=".txt", replace_quotes=True):
|
|
for root, dirs, files in os.walk(m3t_path):
|
|
for file in files:
|
|
if not file.lower().endswith(ext):
|
|
continue
|
|
full_path = os.path.join(root, file)
|
|
output_path = None
|
|
if target_path is not None:
|
|
output_path = os.path.join(target_path, file)
|
|
process_m3t_file(full_path, output_path, dry_run, type, llm_only, replace_quotes)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
from argparse import ArgumentParser
|
|
parser = ArgumentParser(description="Post-process M3T files.")
|
|
parser.add_argument("m3t_path", help="Path to M3T file/directory.")
|
|
parser.add_argument("target_path", nargs="?", default=None, help="Path to save processed M3T file/directory. If not specified, overwrite original files.")
|
|
parser.add_argument("-d", "--dry-run", action="store_true", help="Only show changes without saving.")
|
|
parser.add_argument("--type", type=ScriptType.from_string, choices=list(ScriptType), help="Process lines with special rules.")
|
|
parser.add_argument("--llm-only", action="store_true", help="Only process LLM lines.")
|
|
parser.add_argument("--ext", default=".txt", help="File extension to process in directory mode.")
|
|
parser.add_argument("--no-replace-quotes", action="store_true", help="Do not replace quotes.")
|
|
args = parser.parse_args()
|
|
if os.path.isdir(args.m3t_path):
|
|
recursive_process_m3t(args.m3t_path, args.target_path, args.dry_run, args.type, args.llm_only, args.ext, not args.no_replace_quotes)
|
|
else:
|
|
process_m3t_file(args.m3t_path, args.target_path, args.dry_run, args.type, args.llm_only, not args.no_replace_quotes)
|