Python将Pandas转换为嵌套的JSON

我不认为这是最干净的方法，但如果您需要快速修复，这个相当手动的方法是可行的:

# Set up the sample DataFrame
data = {'id': {0: 'A'},
        'col1': {0: 'B'},
        'address_id': {0: ['123', 'ABC']},
        'address_1': {0: ['Street 123', 'Street ABC']},
        'address_2': {0: ['Road 123', 'Road ABC']},
        'city': {0: ['Dallas', 'Houston']},
        'state': {0: ['Texas', 'Texas']},
        'addition_details': {0: ['XYZ', 'LMP']}}

df = pd.DataFrame(data)

接下来，我们手动处理转换

# Setup a dictionary and populate it with the easy values.
result = {'id': df.at[0, 'id'],
          'col1': df.at[0, 'col1'],
          'address': [],
          'criteria': []}

# Obtain the address entry:
for i in range(len(df.at[0, 'address_id'])):
    address_entry = {
        'address_id': df.at[0, 'address_id'][i],
        'address_1': df.at[0, 'address_1'][i],
        'address_2': df.at[0, 'address_2'][i],
        'city': df.at[0, 'city'][i],
        'state': df.at[0, 'state'][i]
    }
    result['address'].append(address_entry)


# Obtain the criteria entry:
for i in range(len(df.at[0, 'addition_details'])):
    criteria_entry = {
        'addition_details': df.at[0, 'addition_details'][i]
    }
    result['criteria'].append(criteria_entry)

# Convert the result to JSON
json_output = json.dumps(result, indent=2)
print(json_output)

与您的问题相关的是您原始的DataFrame的 struct .DataFrame struct (每列有多个条目，但只有一行)不常见，通常不推荐使用.根据这段代码的目的以及长期可维护性的目标，您可能需要重新考虑这种 struct . 使用更标准的 struct 化DataFrame，通常也更容易将其转换为其他格式.