Pretty HTML formatting for cross tabulation pandas DataFrame, similar to Qualtrics

crosstab_to_html.py

from pandas.core.indexes.multi import MultiIndex


def _read_index(idx):
    if isinstance(idx, MultiIndex):
        return list(idx), idx.names, lambda t: t
    else:
        return [(e,) for e in idx], [idx.name], lambda t: t[0]


skip_row = object()
skip_column = object()


def crosstab_to_html(df):
    row_index, row_names, row_access = _read_index(df.index)
    column_index, column_names, column_access = _read_index(df.columns)

    # On the left, leave out 2 * number of row indexes
    left_padding = 2 * len(row_index[0])

    # On the top, leave out 2 * number of column indexes
    top_padding = 2 * len(column_index[0])

    cells = [
        [('', 'empty')] * (left_padding + len(column_index))
        for _ in range(top_padding + len(row_index))
    ]

    # Write column index labels
    for i, label in enumerate(column_names):
        cells[i * 2][left_padding] = label, 'idx'
        for j in range(1, len(column_index)):
            cells[i * 2][left_padding + j] = skip_column
    # Write column values
    prev = ()
    for col, idx in enumerate(column_index):
        for row, value in enumerate(idx):
            if idx[:row + 1] != prev[:row + 1]:
                cells[row * 2 + 1][left_padding + col] = value, 'label'
            else:
                cells[row * 2 + 1][left_padding + col] = skip_column
        prev = idx

    # Write row index labels
    for i, label in enumerate(row_names):
        cells[top_padding][i * 2] = label, 'idx'
        for j in range(1, len(row_index)):
            cells[top_padding + j][i * 2] = skip_row
    # Write row values
    prev = ()
    for row, idx in enumerate(row_index):
        for col, value in enumerate(idx):
            if idx[:col + 1] != prev[:col + 1]:
                cells[top_padding + row][col * 2 + 1] = value, 'label'
            else:
                cells[top_padding + row][col * 2 + 1] = skip_row
        prev = idx

    # Write values
    for r, row in enumerate(row_index):
        for c, col in enumerate(column_index):
            cells[top_padding + r][left_padding + c] = \
                df[column_access(col)][row_access(row)], 'value'

    # Apply spans
    cells_spans = []
    for r, row in enumerate(cells):
        new_row = []
        cells_spans.append(new_row)

        for c, value in enumerate(row):
            if value is skip_column:
                cells_spans[r][c - 1][1][0] += 1
                new_row.append([None, cells_spans[r][c - 1][1], [1]])
            elif value is skip_row:
                cells_spans[r - 1][c][2][0] += 1
                new_row.append([None, [1], cells_spans[r - 1][c][2]])
            else:
                new_row.append([value, [1], [1]])

    return ('''\
        <style type="text/css">
    td.crosstab-value {
      border: 1px solid black;
      text-align: right;
    }
    td.crosstab-empty {
      background: white;
    }
    td.crosstab-idx {
      border: 1px solid black;
      font-weight: bold;
      text-decoration: underline;
      text-align: center;
    }
    td.crosstab-label {
      border: 1px solid black;
      font-weight: bold;
      text-align: center;
    }
        </style>
    ''' +
         '<table>' +
         '\n'.join('<tr class="crosstab">' +
                   ''.join('<td class="crosstab-%s" colspan="%d" rowspan="%d">'
                           '%s</td>' % (cell[1], cs, rs, cell[0])
                           for cell, [cs], [rs] in row
                           if cell is not None) +
                   '</tr>'
                   for row in cells_spans) +
         '</table>')