如何合并重复列并堆叠对应值？（DataFrame字典数据处理）

我有一个从DataFrame转换来的字典数据如下：

{0: ['RevitCategory', 'Door'],
 1: ['DesignModelID', 'ModelA_Rev1'],
 2: ['DesignObjectID', 'ModelA_Rev1_Object1'],
 3: ['TypeName', 'ARC_DOR_INTERNAL'],
 4: ['Function', 'Internal'],
 5: ['Uniclass2015_Ss', 'Ss_25_30_20_25 : Doorset systems'],
 6: ['IfcExportAs', 'IfcDoorType.DOOR'],
 7: ['NRM1Classification', '2.8.1.1 Internal doors'],
 8: ['RevitCategory', 'Wall'],
 9: ['DesignModelID', 'ModelA_Rev1'],
 10: ['DesignObjectID', 'ModelA_Rev1_Object2'],
 11: ['TypeName', 'ARC_WALL_PARTITION_TYPE1'],
 12: ['Function', 'Internal'],
 13: ['Uniclass2015_Ss', 'Ss_25_10_30_35 : Gypsum board partition systems'],
 14: ['IfcExportAs', 'IfcWallType.PARTITIONING'],
 15: ['NRM1Classification', '2.7.1.1 Internal walls'],
 16: ['Area', 5],
 17: ['RevitCategory', 'Wall'],
 18: ['DesignModelID', 'ModelA_Rev1'],
 19: ['DesignObjectID', 'ModelA_Rev1_Object3'],
 20: ['TypeName', 'ARC_WALL_PARTITION_TYPE1'],
 21: ['Function', 'Internal'],
 22: ['Uniclass2015_Ss', 'Ss_25_10_30_35 : Gypsum board partition systems'],
 23: ['IfcExportAs', 'IfcWallType.PARTITIONING'],
 24: ['NRM1Classification', '2.7.1.1 Internal walls'],
 25: ['Area', 5],
 26: ['RevitCategory', 'Wall'],
 27: ['DesignModelID', 'ModelA_Rev1'],
 28: ['DesignObjectID', 'ModelA_Rev1_Object4'],
 29: ['TypeName', 'ARC_WALL_PARTITION_TYPE1'],
 30: ['Function', 'Internal'],
 31: ['Uniclass2015_Ss', 'Ss_25_10_30_35 : Gypsum board partition systems'],
 32: ['IfcExportAs', 'IfcWallType.PARTITIONING'],
 33: ['NRM1Classification', '2.7.1.1 Internal walls'],
 34: ['Area', 5]
}

需要合并重复的列名，把对应的值堆叠成多行，得到类似如下的表格结构化输出：

RevitCategory  | DesignModelID  | DesignObjectID          |  etc. 
-----------------------------------------------------------------               
Door           | ModelA_Rev1    |  ModelA_Rev1_Object1    |
Wall           | ModelA_Rev1    |  ModelA_Rev1_Object2    |
Wall           | ModelA_Rev1    |  ModelA_Rev1_Object3    |
Wall           | ModelA_Rev1    |  ModelA_Rev1_Object4    |
-----------------------------------------------------------------

用Python的pandas库可以快速实现需求，步骤如下：

1. 解析字典并分组对象 #

把原字典中的[列名, 值]条目转换成键值对，再按完整对象分组：

import pandas as pd

# 原字典数据
data = {0: ['RevitCategory', 'Door'],
 1: ['DesignModelID', 'ModelA_Rev1'],
 2: ['DesignObjectID', 'ModelA_Rev1_Object1'],
 3: ['TypeName', 'ARC_DOR_INTERNAL'],
 4: ['Function', 'Internal'],
 5: ['Uniclass2015_Ss', 'Ss_25_30_20_25 : Doorset systems'],
 6: ['IfcExportAs', 'IfcDoorType.DOOR'],
 7: ['NRM1Classification', '2.8.1.1 Internal doors'],
 8: ['RevitCategory', 'Wall'],
 9: ['DesignModelID', 'ModelA_Rev1'],
 10: ['DesignObjectID', 'ModelA_Rev1_Object2'],
 11: ['TypeName', 'ARC_WALL_PARTITION_TYPE1'],
 12: ['Function', 'Internal'],
 13: ['Uniclass2015_Ss', 'Ss_25_10_30_35 : Gypsum board partition systems'],
 14: ['IfcExportAs', 'IfcWallType.PARTITIONING'],
 15: ['NRM1Classification', '2.7.1.1 Internal walls'],
 16: ['Area', 5],
 17: ['RevitCategory', 'Wall'],
 18: ['DesignModelID', 'ModelA_Rev1'],
 19: ['DesignObjectID', 'ModelA_Rev1_Object3'],
 20: ['TypeName', 'ARC_WALL_PARTITION_TYPE1'],
 21: ['Function', 'Internal'],
 22: ['Uniclass2015_Ss', 'Ss_25_10_30_35 : Gypsum board partition systems'],
 23: ['IfcExportAs', 'IfcWallType.PARTITIONING'],
 24: ['NRM1Classification', '2.7.1.1 Internal walls'],
 25: ['Area', 5],
 26: ['RevitCategory', 'Wall'],
 27: ['DesignModelID', 'ModelA_Rev1'],
 28: ['DesignObjectID', 'ModelA_Rev1_Object4'],
 29: ['TypeName', 'ARC_WALL_PARTITION_TYPE1'],
 30: ['Function', 'Internal'],
 31: ['Uniclass2015_Ss', 'Ss_25_10_30_35 : Gypsum board partition systems'],
 32: ['IfcExportAs', 'IfcWallType.PARTITIONING'],
 33: ['NRM1Classification', '2.7.1.1 Internal walls'],
 34: ['Area', 5]
}

# 自动按对象分组：以RevitCategory为每个对象的起始点
objects = []
current_obj = {}
for item in data.values():
    col_name, val = item
    current_obj[col_name] = val
    # 遇到对象的结尾标识时，保存当前对象并重置
    if col_name in ['NRM1Classification', 'Area']:
        objects.append(current_obj)
        current_obj = {}

# 转换为DataFrame
df = pd.DataFrame(objects)

2. 输出结构化表格 #

直接将DataFrame转换为Markdown格式表格：

print(df.to_markdown(index=False))

输出结果如下：

内容的提问来源于stack exchange，提问作者Jimmy3421