View the nested structure of a pandas DataFrame

view_nested.py

from pandas import DataFrame, concat, Series, isna
import json


def expand_subdict(df_col):
    """
    Expand a column of dictionaries into a Series
    Drop any all nan columns
    """
    # drop empty columns that are sometimes created.
    return df_col.apply(Series).dropna(axis=1, how='all')

 
def values_are_dict_like(col):
    """
    Check if a is a dictionary
    """
    # coming from json, the dict may be a string
    if col.dtype == 'object':
      
        # is it a string, if so the see if it looks like json
        if hasattr(col,'str') and col.str.contains(r'\{.*\}').any():
            return True
         
        # otherwise check if it is a dict objects
        elif col.apply(lambda x: isinstance(x, dict)).any():
            return True
        else:
            return False
    else:
        return False

def values_are_list_like(col):
    """
    Check if any element in a column is a list
    """
    return col.apply(lambda x: isinstance(x, list)).any()

def convertable_to_DataFrame(x):
    """
    Check if a value can be expanded to a dataframe
    """
    return values_are_dict_like(x) or values_are_list_like(x)
    
def infer_schema(df, schema = {}):
    """
    Recursively find the structure of  a 
    pandas DataFrame with nested dictionaries
    Github: @johnarban
    """
    # if already a series or has a list of subcolum
    if isinstance(df, Series) or values_are_list_like(df):
        return infer_schema(expand_subdict(df), schema)
    else:
        for col in df.columns:
            schema[col] = {}
            if convertable_to_DataFrame(df[col]):
                infer_schema(df[col], schema[col])
    
    return schema 

 # copy output of json.dumps(schema) into https://codebeautify.org/jsonviewer