412 lines
16 KiB
Python
412 lines
16 KiB
Python
import re
|
||
|
||
import pandas as pd
|
||
from opencc import OpenCC
|
||
|
||
from era_data import dynasty_alias, KNOWN_ERAS
|
||
|
||
cc = OpenCC('t2s')
|
||
|
||
|
||
# === 中文数字转阿拉伯数字 ===
|
||
def chinese_to_num(text):
|
||
text = text.replace("卅", "三十").replace("卌", "四十")
|
||
if text in ["元", "元年"]:
|
||
return 1
|
||
|
||
chinese_numerals = {
|
||
'零': 0, '〇': 0, '○': 0,
|
||
'一': 1, '二': 2, '三': 3, '四': 4, '五': 5, '六': 6, '七': 7, '八': 8, '九': 9,
|
||
'十': 10, '百': 100, '千': 1000,
|
||
'壹': 1, '貳': 2, '叁': 3, '肆': 4, '伍': 5, '陸': 6, '柒': 7, '捌': 8, '玖': 9, '拾': 10,
|
||
'卄': 20, '廿': 20, '卅': 30, '卌': 40,
|
||
}
|
||
|
||
if text in chinese_numerals:
|
||
return chinese_numerals[text]
|
||
|
||
if '十' in text:
|
||
parts = text.split('十')
|
||
total = 0
|
||
if parts[0] == '':
|
||
total += 10
|
||
else:
|
||
total += chinese_numerals.get(parts[0], 0) * 10
|
||
if len(parts) > 1 and parts[1]:
|
||
total += chinese_numerals.get(parts[1], 0)
|
||
return total
|
||
|
||
if '廿' in text or '卄' in text:
|
||
return 20 + (chinese_numerals.get(text[-1], 0) if len(text) > 1 else 0)
|
||
if '卅' in text:
|
||
return 30 + (chinese_numerals.get(text[-1], 0) if len(text) > 1 else 0)
|
||
|
||
# 混合数字（如"二零〇一"）
|
||
total = 0
|
||
for ch in text:
|
||
total = total * 10 + chinese_numerals.get(ch, int(ch) if ch.isdigit() else 0)
|
||
return total
|
||
|
||
|
||
# 朝代别名见 era_data.py
|
||
|
||
|
||
def normalize_dynasty_name(dynasty):
|
||
if not dynasty:
|
||
return dynasty
|
||
# 先将繁体转简体
|
||
simplified = cc.convert(dynasty)
|
||
# 查别名表（简体）
|
||
for prefix, true_name in dynasty_alias.items():
|
||
if simplified.startswith(prefix):
|
||
return true_name
|
||
# 直接匹配朝代字典（简体）
|
||
for known_dynasty in era_dict.keys():
|
||
if simplified.startswith(known_dynasty) or known_dynasty.startswith(simplified):
|
||
return known_dynasty
|
||
return simplified
|
||
|
||
|
||
# === 年号表 ===
|
||
era_df = pd.read_csv("中国年号.csv")
|
||
era_df["年号"] = era_df["年号"].map(lambda x: cc.convert(str(x)))
|
||
era_df["所属朝代"] = era_df["所属朝代"].map(lambda x: cc.convert(str(x)))
|
||
|
||
era_dict = {}
|
||
for _, row in era_df.iterrows():
|
||
dynasty, era, start = row["所属朝代"], row["年号"], int(row["公元纪年"])
|
||
era_dict.setdefault(dynasty, {})[era] = start
|
||
|
||
|
||
# === 干支纪年 ===
|
||
_tiangan = '甲乙丙丁戊己庚辛壬癸'
|
||
_dizhi = '子丑寅卯辰巳午未申酉戌亥'
|
||
_ganzhi_to_offset = {}
|
||
for i in range(60):
|
||
tg = _tiangan[i % 10]
|
||
dz = _dizhi[i % 12]
|
||
_ganzhi_to_offset[tg + dz] = i # 0-based: 甲子=0
|
||
|
||
def ganzhi_to_era_year(ganzhi: str) -> int:
|
||
"""干支转年数（1-based）：甲子=1, 乙丑=2, ..., 癸亥=60"""
|
||
return _ganzhi_to_offset.get(ganzhi, -1) + 1
|
||
|
||
|
||
def gregorian_to_ganzhi(year: int) -> str:
|
||
"""按项目约定将公元年转换为干支（公元4年为甲子）。"""
|
||
offset = (year - 4) % 60
|
||
return _tiangan[offset % 10] + _dizhi[offset % 12]
|
||
|
||
|
||
_gz_pair = '[' + _tiangan + '][' + _dizhi + ']'
|
||
|
||
# === 数字转中文 ===
|
||
def arabic_to_chinese_year(num: int) -> str:
|
||
digits = "〇一二三四五六七八九"
|
||
if num < 0:
|
||
return '-' + ''.join(digits[int(d)] for d in str(abs(num)))
|
||
return ''.join(digits[int(d)] for d in str(num))
|
||
|
||
|
||
# === 古代年号正则 ===
|
||
# 从 era_dict 构建朝代匹配组（按长度降序排列，优先匹配长朝代名如"北宋"而非"北"）
|
||
# 额外添加常见但不在CSV中的朝代简称（如"汉"）
|
||
_extra_dynasty_names = ['汉', '晋', '宋', '周']
|
||
_dynasty_names = sorted(
|
||
list(era_dict.keys()) + [d for d in _extra_dynasty_names if d not in era_dict],
|
||
key=len, reverse=True
|
||
)
|
||
_dynasty_alt = '|'.join(re.escape(d) for d in _dynasty_names)
|
||
# 常见朝代前缀（大、皇等），用于匹配"大唐""大清"等
|
||
_dynasty_prefix = '大皇前后胡'
|
||
# 非元年数字字符（排除元，避免歧义：建元元年中第一个元是年号的一部分）
|
||
_year_digits = '〇零一二三四五六七八九十廿卄卅卌'
|
||
|
||
# 已知年号白名单见 era_data.py
|
||
|
||
# 预编译已知年号正则片段（按长度降序，优先匹配长年号如"太平兴国"而非"太平"）
|
||
_known_era_alt = '|'.join(sorted(KNOWN_ERAS, key=len, reverse=True))
|
||
_known_era_group = r'(?P<era>' + _known_era_alt + r')'
|
||
_yd = '[' + _year_digits + ']'
|
||
|
||
def _build_pattern(with_dynasty=True):
|
||
"""构建年号正则：已知年号优先 + 通用兜底（仅限有朝代前缀时）"""
|
||
if with_dynasty:
|
||
# 朝代前缀模式：加(?<![一-鿿])防止"元"在"开元"中被当朝代
|
||
dynasty_part = r'(?P<dynasty>(?:(?<![一-鿿])[' + _dynasty_prefix + r'](?:' + _dynasty_alt + r'))|(?:(?<![一-鿿])' + _dynasty_alt + r'))'
|
||
boundary = ''
|
||
else:
|
||
dynasty_part = r'(?P<dynasty>)'
|
||
# 无朝代模式不加边界（允许年号跟在其他汉字后面，如"永乐大帝在永乐三年"）
|
||
boundary = ''
|
||
|
||
if with_dynasty:
|
||
# 有朝代前缀：年号后直接跟年份数字，无需额外约束
|
||
era_year = _known_era_group + r'(?P<year>' + _yd + r'+)年'
|
||
era_yuan = _known_era_group + r'元年'
|
||
patterns = [
|
||
re.compile(dynasty_part + boundary + era_year),
|
||
re.compile(dynasty_part + boundary + era_yuan),
|
||
# 通用兜底（2-8字）
|
||
re.compile(dynasty_part + boundary + r'(?P<era>[一-鿿]{2,8})' + r'(?P<year>' + _yd + r'+)年'),
|
||
re.compile(dynasty_part + boundary + r'(?P<era>[一-鿿]{2,8})元年'),
|
||
]
|
||
else:
|
||
# 无朝代前缀：年号后必须紧跟年份数字（前瞻断言），防止"永乐大帝"误匹配
|
||
_year_ahead = r'(?=' + _yd + r'+年|元年)'
|
||
era_year = _known_era_group + _year_ahead + r'(?P<year>' + _yd + r'+)年'
|
||
era_yuan = _known_era_group + _year_ahead + r'元年'
|
||
patterns = [
|
||
re.compile(dynasty_part + boundary + era_year),
|
||
re.compile(dynasty_part + boundary + era_yuan),
|
||
]
|
||
return patterns
|
||
|
||
# 无朝代模式只匹配已知年号（避免误匹配如"国"被当年号）
|
||
_patterns_no_dynasty = _build_pattern(with_dynasty=False)
|
||
# 有朝代模式：已知年号 + 通用兜底
|
||
_patterns_dynasty = _build_pattern(with_dynasty=True)
|
||
|
||
|
||
# === 主函数 ===
|
||
def extract_era_years(text):
|
||
# 去除标记用的方括号（用户标注用，非原文内容）
|
||
text = text.replace('[', '').replace(']', '')
|
||
original_text = text
|
||
simplified_text = cc.convert(text)
|
||
results = []
|
||
|
||
# === 1. 民国 ===
|
||
minguo_pattern = re.compile(r'民[国國]\s*([〇零一二三四五六七八九十百廿卄卅卌元\d]+)\s*年')
|
||
for m in minguo_pattern.finditer(simplified_text):
|
||
year_text = m.group(1)
|
||
year_num = chinese_to_num(year_text)
|
||
start, end = m.start(), m.end()
|
||
raw_fan = original_text[start:end]
|
||
results.append({
|
||
"原文": raw_fan,
|
||
"原文(简体)": m.group(0),
|
||
"朝代": "民国",
|
||
"年号": "民国纪年",
|
||
"年数": year_num,
|
||
"公元": 1911 + year_num,
|
||
"公元中文": f"{arabic_to_chinese_year(1911 + year_num)}年",
|
||
"位置": {"起始": start, "结束": end},
|
||
"类型": "民国纪年",
|
||
})
|
||
|
||
# 1b. 民国+干支纪年（如"民国丁卯""民国甲戌"）
|
||
minguo_gz = re.compile(r'民[国國]\s*(?P<ganzhi>' + _gz_pair + r')')
|
||
for m in minguo_gz.finditer(simplified_text):
|
||
ganzhi = m.group("ganzhi")
|
||
gz_offset = _ganzhi_to_offset.get(ganzhi, -1)
|
||
if gz_offset < 0:
|
||
continue
|
||
# 民国元年=1912, base_gz = (1912-4)%60 = 8 (己酉)
|
||
base_gz = (1912 - 4) % 60
|
||
year_in_era = ((gz_offset - base_gz) % 60) + 1
|
||
gregorian = 1912 + year_in_era - 1
|
||
start, end = m.start(), m.end()
|
||
raw_fan = original_text[start:end]
|
||
results.append({
|
||
"原文": raw_fan,
|
||
"原文(简体)": m.group(0),
|
||
"朝代": "民国",
|
||
"年号": "民国纪年",
|
||
"年数": year_in_era,
|
||
"干支": ganzhi,
|
||
"公元": gregorian,
|
||
"公元中文": f"{arabic_to_chinese_year(gregorian)}年",
|
||
"位置": {"起始": start, "结束": end},
|
||
"类型": "民国纪年（干支）",
|
||
})
|
||
|
||
# === 2. 公元（允许无"年"字）===
|
||
gongyuan_pattern = re.compile(
|
||
r'(公元)?\s*([〇零○O一二三四五六七八九\d]{3,4})\s*(?:（.*?）)?\s*(年)?'
|
||
)
|
||
for m in gongyuan_pattern.finditer(simplified_text):
|
||
year_text = m.group(2)
|
||
year_num = chinese_to_num(year_text)
|
||
if not (1000 <= year_num <= 2100):
|
||
continue
|
||
start, end = m.start(), m.end()
|
||
raw_fan = original_text[start:end]
|
||
era_type = "公元纪年（中华人民共和国）" if year_num >= 1949 else "公元纪年（近代/其他）"
|
||
results.append({
|
||
"原文": raw_fan,
|
||
"原文(简体)": m.group(0),
|
||
"类型": era_type,
|
||
"公元": year_num,
|
||
"公元中文": f"{arabic_to_chinese_year(year_num)}年",
|
||
"位置": {"起始": start, "结束": end},
|
||
})
|
||
|
||
# === 3. 干支纪年（如"清雍正乙巳十一月十二午时生"）===
|
||
covered = set()
|
||
def _process_ganzhi_match(m):
|
||
"""处理干支纪年匹配"""
|
||
raw_dynasty = m.group("dynasty") if "dynasty" in m.groupdict() else ""
|
||
dynasty = normalize_dynasty_name(raw_dynasty or "")
|
||
era = cc.convert(m.group("era"))
|
||
ganzhi = m.group("ganzhi")
|
||
if ganzhi not in _ganzhi_to_offset:
|
||
return
|
||
|
||
# 查找朝代和基准年
|
||
base = None
|
||
if dynasty:
|
||
for d, eras in era_dict.items():
|
||
if dynasty in d or d in dynasty:
|
||
if era in eras:
|
||
base = eras[era]
|
||
dynasty = d
|
||
break
|
||
if not base:
|
||
for d, eras in era_dict.items():
|
||
if era in eras:
|
||
base = eras[era]
|
||
dynasty = d
|
||
break
|
||
if not base:
|
||
return
|
||
|
||
# 干支60年循环：计算基准年的干支位置，再求偏移
|
||
base_gz = (base - 4) % 60 # 公元4年=甲子(0)，所以偏移=4
|
||
gz_offset = _ganzhi_to_offset.get(ganzhi, 0) # 0-based
|
||
year_in_era = ((gz_offset - base_gz) % 60) + 1 # 1-based年数
|
||
gregorian = base + year_in_era - 1
|
||
start, end = m.start(), m.end()
|
||
if any(pos in covered for pos in range(start, end)):
|
||
return
|
||
covered.update(range(start, end))
|
||
raw_fan = original_text[start:end]
|
||
results.append({
|
||
"原文": raw_fan,
|
||
"原文(简体)": m.group(0),
|
||
"朝代": dynasty,
|
||
"年号": era,
|
||
"年数": year_in_era,
|
||
"干支": ganzhi,
|
||
"公元": gregorian,
|
||
"公元中文": f"{arabic_to_chinese_year(gregorian)}年",
|
||
"位置": {"起始": start, "结束": end},
|
||
"类型": "干支纪年",
|
||
})
|
||
|
||
# 模式A：有朝代前缀（如"清雍正乙巳"）
|
||
# 单字年号加前瞻(?![地支])，防止单字年号"丙"吞掉天干"丙午"中的"丙"
|
||
_gz_era_group = r'(?P<era>' + '|'.join(
|
||
(re.escape(e) + (r'(?![子丑寅卯辰巳午未申酉戌亥])' if len(e) == 1 else ''))
|
||
for e in sorted(KNOWN_ERAS, key=len, reverse=True)
|
||
) + r')'
|
||
ganzhi_dynasty = re.compile(
|
||
r'(?P<dynasty>(?:(?<![一-鿿])[' + _dynasty_prefix + r'](?:' + _dynasty_alt + r'))|(?:(?<![一-鿿])' + _dynasty_alt + r'))'
|
||
+ _gz_era_group
|
||
+ r'(?P<ganzhi>' + _gz_pair + r')(?:年)?'
|
||
)
|
||
for m in ganzhi_dynasty.finditer(simplified_text):
|
||
_process_ganzhi_match(m)
|
||
|
||
# 模式B：无朝代前缀（如"嘉庆癸亥"），年号≥2字符避免天干误匹配
|
||
_long_era_names = [e for e in KNOWN_ERAS if len(e) >= 2]
|
||
_long_era_alt = '|'.join(sorted(_long_era_names, key=len, reverse=True))
|
||
ganzhi_no_dynasty = re.compile(
|
||
r'(?P<dynasty>)'
|
||
+ r'(?P<era>' + _long_era_alt + r')'
|
||
+ r'(?P<ganzhi>' + _gz_pair + r')(?:年)?'
|
||
)
|
||
for m in ganzhi_no_dynasty.finditer(simplified_text):
|
||
_process_ganzhi_match(m)
|
||
|
||
# === 4. 古代年号 ===
|
||
|
||
def _process_era_match(m, year_override=None):
|
||
"""处理一个年号正则匹配结果"""
|
||
raw_dynasty = m.group("dynasty") if "dynasty" in m.groupdict() else ""
|
||
dynasty = normalize_dynasty_name(raw_dynasty or "")
|
||
era = cc.convert(m.group("era"))
|
||
|
||
# 过滤误匹配：年号是单个数字字符（如"民国二十六年"中的"二"）
|
||
if len(era) == 1 and era in '零一二三四五六七八九十廿卄卅卌':
|
||
return
|
||
# 过滤误匹配：年号全是数字汉字（如"元一九五二年"中的"一九五"）
|
||
if era and all(ch in '零一二三四五六七八九十〇' for ch in era):
|
||
return
|
||
year_text = year_override if year_override is not None else m.group("year")
|
||
year_num = chinese_to_num(year_text)
|
||
base = None
|
||
|
||
if dynasty:
|
||
for d, eras in era_dict.items():
|
||
if dynasty in d or d in dynasty:
|
||
if era in eras:
|
||
base = eras[era]
|
||
dynasty = d
|
||
break
|
||
else:
|
||
candidates = []
|
||
for d, eras in era_dict.items():
|
||
if era in eras:
|
||
candidates.append((d, eras[era]))
|
||
if len(candidates) == 1:
|
||
dynasty, base = candidates[0]
|
||
elif len(candidates) > 1:
|
||
dynasty, base = candidates[0]
|
||
|
||
start, end = m.start(), m.end()
|
||
raw_fan = original_text[start:end]
|
||
covered.update(range(start, end))
|
||
if base:
|
||
gregorian = base + year_num - 1
|
||
results.append({
|
||
"原文": raw_fan,
|
||
"原文(简体)": m.group(0),
|
||
"朝代": dynasty,
|
||
"年号": era,
|
||
"年数": year_num,
|
||
"公元": gregorian,
|
||
"公元中文": f"{arabic_to_chinese_year(gregorian)}年",
|
||
"位置": {"起始": start, "结束": end},
|
||
"类型": "古代纪年",
|
||
})
|
||
else:
|
||
results.append({
|
||
"原文": raw_fan,
|
||
"原文(简体)": m.group(0),
|
||
"朝代": dynasty,
|
||
"年号": era,
|
||
"年数": year_num,
|
||
"位置": {"起始": start, "结束": end},
|
||
"类型": "古代纪年（未匹配）",
|
||
})
|
||
|
||
# 收集所有匹配，对于重叠区间保留覆盖范围更大的（如"金大定元年"优先于"大定元年"）
|
||
all_matches = []
|
||
for pat in _patterns_no_dynasty:
|
||
for m in pat.finditer(simplified_text):
|
||
has_year = "year" in m.groupdict() and m.group("year") is not None
|
||
all_matches.append((m, None if has_year else "元", False))
|
||
for pat in _patterns_dynasty:
|
||
for m in pat.finditer(simplified_text):
|
||
has_year = "year" in m.groupdict() and m.group("year") is not None
|
||
all_matches.append((m, None if has_year else "元", True))
|
||
|
||
# 按覆盖范围降序排列（长的优先），同长度时朝代模式优先
|
||
all_matches.sort(key=lambda x: (x[0].end() - x[0].start(), x[2]), reverse=True)
|
||
|
||
for m, year_override, _is_dynasty in all_matches:
|
||
if not any(pos in covered for pos in range(m.start(), m.end())):
|
||
_process_era_match(m, year_override=year_override)
|
||
|
||
for result in results:
|
||
if result.get("公元") is not None:
|
||
result.setdefault("干支", gregorian_to_ganzhi(result["公元"]))
|
||
|
||
return results
|
||
|
||
|
||
def convert_text(text):
|
||
results = extract_era_years(text)
|
||
return results
|