多维度日期处理

parser.py

@@ -178,6 +178,13 @@ def gregorian_to_ganzhi(year: int) -> str:
 
 _gz_pair = '[' + _tiangan + '][' + _dizhi + ']'
 
+# 干支纪年中的年号组（已知年号按长度降序；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_ALL, key=len, reverse=True)
+) + r')'
+
 # === 数字转中文 ===
 def arabic_to_chinese_year(num: int) -> str:
     digits = "〇一二三四五六七八九"
@@ -249,6 +256,73 @@ _patterns_no_dynasty = _build_pattern(with_dynasty=False)
 _patterns_dynasty = _build_pattern(with_dynasty=True)
 
 
+# === 农历月日（干支纪年后的"十一月十四""闰六月廿一日"） ===
+_month_special = {"正": 1, "冬": 11, "腊": 12}
+
+def _parse_month(text):
+    """月文本转数字：正月=1、冬月=11、腊月=12，其余走中文数字（十一月→11）"""
+    return _month_special.get(text, chinese_to_num(text))
+
+def _parse_day(text):
+    """日文本转数字：剥"初"前缀（初八→8、初十→10），廿三→23"""
+    if text.startswith("初"):
+        text = text[1:]
+    return chinese_to_num(text)
+
+_month_day_re = re.compile(
+    r'(?P<leap>闰)?(?P<month>[正冬腊〇零一二三四五六七八九十廿卄卅卌]{1,3})月'
+    r'(?:(?P<day>[初〇零一二三四五六七八九十廿卄卅卌]{1,3})(?:日)?)?'
+)
+# 嵌入干支正则的可选月日片段（整体捕获，事后用 _month_day_re 解析）
+_md_body = r'(?:闰)?[正冬腊〇零一二三四五六七八九十廿卄卅卌]{1,3}月' \
+           r'(?:[初〇零一二三四五六七八九十廿卄卅卌]{1,3}(?:日)?)?'
+# 族谱常见的生死动词：月日与（公元对照）之间可能隔着"生/殁"等单字
+_md_verb = r'[生死殁卒薨殓]'
+
+def _attach_month_day(result, m):
+    """把匹配中的月日组（md1/md2/md，括号前/后位置）并入结果字段"""
+    md = None
+    for name in ("md1", "md2", "md"):
+        if name in m.groupdict() and m.group(name):
+            md = m.group(name)
+            break
+    if not md:
+        return
+    mm = _month_day_re.match(md)
+    if not mm:
+        return
+    result["月"] = _parse_month(mm.group("month"))
+    if mm.group("day"):
+        result["日"] = _parse_day(mm.group("day"))
+    result["月日"] = md
+
+
+# === 2c. 干支+（公元对照）（如"清同治壬申（1872）十一月十四生"） ===
+# 干支纪年旁直接给出公元年份：合并为一条结果、对照校验干支、紧跟的月日并入字段。
+# 置于公元段之前（见 extract_era_years），避免"（1872）"被单独摘出成公元纪年。
+_dynasty_opt = (r'(?P<dynasty>(?:(?<![一-鿿])[' + _dynasty_prefix + r'](?:' + _dynasty_alt + r'))'
+                r'|(?:(?<![一-鿿])' + _dynasty_alt + r'))?')
+# 括号中的公元年份：阿拉伯/全角/中文数字 3-4 位，可带"年"字
+_gz_gy_paren = r'[（(]\s*(?P<gy_year>[〇零○O一二三四五六七八九\d０-９]{3,4})\s*年?[)）]'
+_ganzhi_gongyuan_re = re.compile(
+    _dynasty_opt
+    + _gz_era_group
+    + r'(?P<ganzhi>' + _gz_pair + r')(?:年)?'
+    + r'(?P<md1>' + _md_body + r')?'
+    + r'(?:' + _md_verb + r')?'
+    + _gz_gy_paren
+    + r'(?P<md2>' + _md_body + r')?'
+)
+# 裸干支+（对照）：无年号可锚定（如"壬申（1872）"），公元取对照值
+_ganzhi_bare_gongyuan_re = re.compile(
+    r'(?P<ganzhi>' + _gz_pair + r')(?:年)?'
+    + r'(?P<md1>' + _md_body + r')?'
+    + r'(?:' + _md_verb + r')?'
+    + _gz_gy_paren
+    + r'(?P<md2>' + _md_body + r')?'
+)
+
+
 # === 主函数 ===
 def extract_era_years(text):
     # 标注符号（[]、【】）不参与匹配，但保留在文本中——位置偏移以原始文本为准
@@ -378,6 +452,74 @@ def extract_era_years(text):
                 result["干支不符"] = True
         results.append(result)
 
+    # === 2c. 干支+（公元对照）（如"清同治壬申（1872）十一月十四生"） ===
+    # 干支纪年旁直接给出公元对照：合并为一条结果，用对照校验干支，紧跟的月日并入字段。
+    # 置于公元段之前，避免"（1872）"被单独摘出成公元纪年（同 2b 的合并优先策略）。
+    def _process_ganzhi_gongyuan(m):
+        raw_dynasty = m.group("dynasty") if "dynasty" in m.groupdict() else ""
+        dynasty = normalize_dynasty_name(raw_dynasty or "")
+        era = cc.convert(m.group("era")) if "era" in m.groupdict() and m.group("era") else None
+        ganzhi = m.group("ganzhi")
+        gy_year = chinese_to_num(m.group("gy_year"))
+        if ganzhi not in _ganzhi_to_offset or not (1000 <= gy_year <= 2100):
+            return
+        start, end = m.start(), m.end()
+        if any(pos in covered for pos in range(start, end)):
+            return
+        # 查找年号基准年（与干支纪年段同逻辑：先精确朝代，后全量候选取首个并标注歧义）
+        base = None
+        multi = []
+        if dynasty and era:
+            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 base is None and era:
+            for d, eras in era_dict.items():
+                if era in eras:
+                    multi.append((d, eras[era]))
+            if multi:
+                dynasty, base = multi[0]
+        covered.update(range(start, end))
+        result = {
+            "原文": original_text[start:end],
+            "原文(简体)": m.group(0),
+            "干支": ganzhi,
+            "公元": gy_year,
+            "公元中文": f"{arabic_to_chinese_year(gy_year)}年",
+            "位置": {"起始": start, "结束": end},
+            "类型": "干支纪年",
+        }
+        if base is not None:
+            result["朝代"] = dynasty
+            result["年号"] = era
+            # 干支60年循环推算（同干支纪年段公式）；与对照不一致时标注，公元仍取对照值
+            base_gz = (base - 4) % 60
+            year_in_era = ((_ganzhi_to_offset[ganzhi] - base_gz) % 60) + 1
+            result["年数"] = year_in_era
+            computed = base + year_in_era - 1
+            if computed != gy_year:
+                result["干支不符"] = True
+                result["干支推算"] = computed
+            if len(multi) > 1:
+                result["多候选"] = True
+                result["候选"] = [list(c) for c in multi]
+            span = era_span.get(dynasty, {}).get(era)
+            if span is not None and gy_year > base + span - 1:
+                result["超出使用期"] = True
+        elif gregorian_to_ganzhi(gy_year) != ganzhi:
+            # 裸干支+对照：无年号可锚定，直接用公元年份的干支校验（同公元段约定）
+            result["干支不符"] = True
+        _attach_month_day(result, m)
+        results.append(result)
+
+    for m in _ganzhi_gongyuan_re.finditer(simplified_text):
+        _process_ganzhi_gongyuan(m)
+    for m in _ganzhi_bare_gongyuan_re.finditer(simplified_text):
+        _process_ganzhi_gongyuan(m)
+
     # === 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\d０-９]{3,4})\s*(?:（.*?）)?\s*(年)?'
@@ -480,18 +622,16 @@ def extract_era_years(text):
         span = era_span.get(dynasty, {}).get(era)
         if span is not None and gregorian > base + span - 1:
             result["超出使用期"] = True
+        _attach_month_day(result, m)
         results.append(result)
 
-    # 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_ALL, key=len, reverse=True)
-    ) + r')'
+    # 模式A：有朝代前缀（如"清雍正乙巳"）；年号组 _gz_era_group 见模块级定义
     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)?'
+        # 紧跟的农历月日并入同一单位（如"清雍正乙巳十一月十二午时生"）
+        + r'(?P<md>' + _md_body + r')?'
     )
     for m in ganzhi_dynasty.finditer(simplified_text):
         _process_ganzhi_match(m)
@@ -503,6 +643,7 @@ def extract_era_years(text):
         r'(?P<dynasty>)'
         + r'(?P<era>' + _long_era_alt + r')'
         + r'(?P<ganzhi>' + _gz_pair + r')(?:年)?'
+        + r'(?P<md>' + _md_body + r')?'
     )
     for m in ganzhi_no_dynasty.finditer(simplified_text):
         _process_ganzhi_match(m)