Create a cross tab / pivot table in Python

crosstab.py

import itertools

def crosstab(rows, columns, col_idx_for_columns, lst_col_idx_for_rows, value_col_idx, fill_val, format_func=None):
    """Take col_idx_to_cross_tab and make its unique values be new columns at the end filled with
    value_col_idx values.  col idx arguments are 0 based.
    
    This is basically a simplified pivot table creator with the limitations that you can only have one field in the columns
    and there is no aggregation.
    
    Usage:
>>> columns = ['month','owner','dogs owned']
>>> data = [
[1, 'greg', 3],
[1, 'greg', 5],
[2, 'greg', 3],
[3, 'greg', 15],
[4, 'ken', 25],
[4, 'greg', 53],
[5, 'greg', 5],
]
>>> print crosstab(data, columns, col_idx_for_columns=0, lst_col_idx_for_rows=[1], value_col_idx=2, fill_val=0)
(['owner', 1, 2, 3, 4, 5], [['greg', 5, 3, 15, 53, 5], ['ken', 0, 0, 0, 25, 0]])
    """
    #find all unique values
    cross_tab_cols = sorted(list(set([row[col_idx_for_columns] for row in rows])))
    columns.pop(max(col_idx_for_columns,value_col_idx))
    columns.pop(min(col_idx_for_columns,value_col_idx))
    columns += cross_tab_cols

    def get_key(x):
        key = []
        for idx in lst_col_idx_for_rows:
            key.append(x[idx])
        return key

    rows = sorted(rows, key=get_key)
    grouped_rows = itertools.groupby(rows, get_key)

    results = []
    for key, rows in grouped_rows:
        rows = list(rows)
        new_row = list(rows[0]) + [fill_val]*len(cross_tab_cols)
        #remove columns:
        new_row.pop(max(col_idx_for_columns,value_col_idx))
        new_row.pop(min(col_idx_for_columns,value_col_idx))
        assert len(new_row) == len(columns)
        for row in rows:
            target_col = row[col_idx_for_columns]
            value = row[value_col_idx]
            new_row[columns.index(target_col)] = (format_func and format_func(value)) or value
        results.append(new_row)
    return columns, results