Transpose CSV to JSON

• Based on some comments, some people are just pandas haters. But I like to use the tool that allows me to solve the problem in the easiest manner possible, and with the fewest lines of code.
  • In this case, without a doubt, that's pandas
  • An added benefit of using pandas, is the data can easily be clean, analyzed , and visualized, if needed.
  • Solutions at How to convert CSV file to multiline JSON? offer some basics, but won't help transform the csv into the required shape.
• Because of the expected output of the JSON file, this is a non-trivial question, which requires reshaping/grouping the data in the csv and is easily accomplished with pandas.DataFrame.groupby.
  • groupby 'h1' since the column values will be the dict outer keys
  • groupby returns a DataFrameGroupBy object that can be split into, i, the value used to create the group ('c1' and 'c2' in this case) and the associated dataframe group, g.
    • Use pandas.DataFrame.to_dict to convert the dataframe into a list of dictionaries.

import json
import pandas as pd

# read the file
df = pd.DataFrame('test.csv')

# display(df)
   h1  h2  h3
0  c1  n1  a1
1  c2  n2  a2
2  c1  n3  a3

# groupby and create dict
data_dict = dict()
for i, g in df.groupby('h1'):
    data_dict[i] = g.drop(columns=['h1']).to_dict(orient='records')

# print(data_dict)
{'c1': [{'h2': 'n1', 'h3': 'a1'}, {'h2': 'n3', 'h3': 'a3'}],
 'c2': [{'h2': 'n2', 'h3': 'a2'}]}

# save data_dict to a file as a JSON
with open('result.json', 'w') as fp:
    json.dump(data_dict, fp)

JSON file

{
    "c1": [{
            "h2": "n1",
            "h3": "a1"
        }, {
            "h2": "n3",
            "h3": "a3"
        }
    ],
    "c2": [{
            "h2": "n2",
            "h3": "a2"
        }
    ]
}