如何将含重复图片名的CSV转换为指定格式的嵌套字典？

原始CSV数据 #

现有代码及问题 #

原始代码尝试将CSV读取为以图片名为键的嵌套字典，但生成结果存在两个问题：

• 每条记录中冗余保留了image字段
• 所有数值类型字段（id、xMin等）均为字符串格式

原始代码：

import csv
import itertools
import operator
import json
with open('out1.csv', 'r') as fp:
    reader = csv.DictReader(fp, dialect='excel', skipinitialspace=True)
    new_dict = {}
    for group, records in itertools.groupby(reader, key=operator.itemgetter('image')):
        new_dict[group] = list(records)
json_object = json.dumps(new_dict, indent = 4)
print(json_object)

优化后的代码 #

针对问题，修改代码如下，实现移除冗余字段、转换数值类型的需求：

import csv
import itertools
import operator
import json

# 定义需要转换为整数的字段
INT_FIELDS = {'id', 'xMin', 'xMax', 'yMin', 'yMax'}

with open('out1.csv', 'r') as fp:
    reader = csv.DictReader(fp, dialect='excel', skipinitialspace=True)
    new_dict = {}
    for image_name, records in itertools.groupby(reader, key=operator.itemgetter('image')):
        processed_records = []
        for record in records:
            # 移除冗余的image字段
            del record['image']
            # 转换数值字段为整数
            for field in INT_FIELDS:
                record[field] = int(record[field])
            processed_records.append(record)
        new_dict[image_name] = processed_records

json_object = json.dumps(new_dict, indent=4)
print(json_object)

期望输出结果 #

运行优化代码后，将得到如下格式的输出：

{
    "858a0246-2f2d-40a9-9bcb-01ab8a93c7f5_BU26844_1630586024_0_6.webp": [
        {
            "id": 1,
            "name": "Scratch",
            "xMin": 604,
            "xMax": 893,
            "yMin": 230,
            "yMax": 413
        },
        {
            "id": 2,
            "name": "Dent",
            "xMin": 921,
            "xMax": 1146,
            "yMin": 720,
            "yMax": 857
        }
    ],
    "858a0246-2f2d-40a9-9bcb-01ab8a93c7f5_BU26844_1630586024_23.webp": [
        {
            "id": 1,
            "name": "Dent",
            "xMin": 343,
            "xMax": 2323,
            "yMin": 334,
            "yMax": 343
        }
    ]
}

内容的提问来源于stack exchange，提问作者Alex Bodger