Update app.py

app.py

@@ -2,114 +2,108 @@ import gradio as gr
 import json
 from collections import defaultdict
 
-def build_family_groups():
-    """
-    构建家庭：每个家庭由 (丈夫, 妻子, [子女]) 组成
-    策略：
-      1. 所有记录：{name -> {father: ..., info: ...}}
-      2. 找出所有“子女”：他们的 father 是 A
-      3. 找出所有“妻子”：她们的 father 是 A，但她们自己也是母亲（有子女）
-      4. 丈夫 A 必须是男性（通过是否有同名作为 father 判断）
-    """
+def load_families_from_jsonl():
     people = {}  # name -> record
-    children_of = defaultdict(list)  # father_name -> [(child_name, info)]
+    children_by_parents = defaultdict(list)  # (father, mother) -> [child]
 
-    # 第一步：加载所有人
     try:
         with open('family_data.jsonl', 'r', encoding='utf-8') as f:
             for line in f:
                 line = line.strip()
-                if not line: continue
+                if not line:
+                    continue
                 try:
                     rec = json.loads(line)
                     name = rec.get('n')
-                    if not name: continue
-                    people[name] = {
-                        'f': rec.get('f'),
-                        'info': rec.get('info', '')
-                    }
-                    father = rec.get('f')
-                    if father and father != "无":
-                        children_of[father].append((name, rec.get('info', '')))
+                    if not name:
+                        continue
+                    people[name] = rec
                 except:
                     continue
     except Exception as e:
-        return [{"error": str(e)}]
+        return {"__ERROR__": f"无法读取 family_data.jsonl: {e}"}
 
-    # 第二步：找出所有“为人父母”的人（即他们的名字在 children_of 中作为 key 出现）
-    is_parent = set(children_of.keys())
+    # 第二步：按父母分组子女
+    for name, rec in people.items():
+        father = rec.get('f')
+        mother = rec.get('m')
+        # 只有同时有父亲和母亲的才归入家庭（可根据需要放宽）
+        if father and mother and father != "无" and mother != "无":
+            children_by_parents[(father, mother)].append((name, rec.get('info', '')))
 
-    # 第三步：构建家庭
+    # 第三步：构建家庭列表
     families = []
+    seen_parents = set()
 
-    # 遍历所有可能的丈夫（即所有有子女的人）
-    for husband in sorted(is_parent):
-        if husband == "无" or husband not in people:
+    for (father, mother), children in children_by_parents.items():
+        if (father, mother) in seen_parents:
             continue
+        seen_parents.add((father, mother))
 
-        # 获取所有挂在他名下的人（包括子女和可能的妻子）
-        related = children_of[husband]
-        spouse = None
-        children = []
-
-        for name, info in related:
-            # 如果这个人自己也是父母（有子女），则很可能是妻子（儿媳/配偶）
-            if name in is_parent:
-                # 优先作为配偶
-                if spouse is None:
-                    spouse = name
-                # 如果已有配偶，后续的可能是儿媳等，暂不处理
-            else:
-                children.append((name, info))
+        # 获取父母的完整信息
+        father_info = people.get(father, {}).get('info', '')
+        mother_info = people.get(mother, {}).get('info', '')
 
-        # 去重子女
-        children = list(set(children))
-        children.sort(key=lambda x: x[0])  # 按名字排序
+        # 去重并排序子女（按 info 中的年份或名字）
+        unique_children = list(set(children))
+        # 尝试按出生年份排序
+        def extract_year(info):
+            import re
+            match = re.search(r'(19\d{2}|20\d{2}|18\d{2})', str(info))
+            return int(match.group(1)) if match else 9999
+        unique_children.sort(key=lambda x: extract_year(x[1]))
 
-        # 只有当有子女或有配偶时才展示
-        if children or spouse:
-            families.append({
-                "husband": husband,
-                "wife": spouse or "（请补充配偶）",
-                "children": children
-            })
+        families.append({
+            "father": father,
+            "father_info": father_info,
+            "mother": mother,
+            "mother_info": mother_info,
+            "children": unique_children
+        })
 
+    # 按父亲名字排序
+    families.sort(key=lambda x: x["father"])
     return families
 
-def format_output():
-    families = build_family_groups()
+def format_family_output():
+    families = load_families_from_jsonl()
+    
+    if isinstance(families, dict) and "__ERROR__" in families:
+        return families["__ERROR__"]
     
-    if not families or (len(families) == 1 and "error" in families[0]):
-        error_msg = families[0].get("error", "未知错误")
-        return f"❌ 加��失败：{error_msg}"
+    if not families:
+        return "未找到任何家庭数据。请检查 family_data.jsonl 格式是否正确。"
 
-    lines = ["# 冯氏家族关系网家网成员表\n"]
+    lines = ["# 冯氏家族关系网 · 家庭成员表\n"]
     
     for fam in families:
-        h = fam["husband"]
-        w = fam["wife"]
-        kids = fam["children"]
+        f_name = fam["father"]
+        m_name = fam["mother"]
+        f_info = fam["father_info"]
+        m_info = fam["mother_info"]
+        children = fam["children"]
         
-        lines.append(f"### {h} 家庭")
-        lines.append(f"- **父亲**：{h}")
-        lines.append(f"- **母亲**：{w}")
+        # 家庭标题
+        lines.append(f"## {f_name} & {m_name} 家庭")
+        lines.append(f"- **父亲**：{f_name}（{f_info}）")
+        lines.append(f"- **母亲**：{m_name}（{m_info}）")
         lines.append("- **子女**：")
-        if kids:
-            for name, info in kids:
-                lines.append(f"  - {name} — {info}")
+        if children:
+            for child_name, child_info in children:
+                lines.append(f"  - {child_name} — {child_info}")
         else:
-            lines.append("  - （暂无子女记录）")
-        lines.append("")
-    
+            lines.append("  - （无子女记录）")
+        lines.append("")  # 空行分隔
+
     return "\n".join(lines)
 
 # Gradio 应用
 demo = gr.Interface(
-    fn=format_output,
+    fn=format_family_output,
     inputs=None,
-    outputs=gr.Markdown(label="冯氏家族关系网"),
-    title="冯氏家族关系网家网成员表",
-    description="点击 **Generate** 查看所有以男性户主为核心的家庭"
+    outputs=gr.Markdown(label="家庭成员表"),
+    title="冯氏家族关系网 · 家庭成员表",
+    description="点击 **Generate** 查看所有以父母为单位的家庭（数据来自 family_data.jsonl）"
 )
 
 if __name__ == "__main__":