PYSPARK：UnicodeEncodeError：＆＃39; latin-1＆＃39;编解码器不能对位置515-517中的字符进行编码：序数不在范围内（256）

Question

尝试使用PySpark显示我的结果数据框，并收到上述错误。

from pyspark.sql.functions import col, count, countDistinct, when, trim, isnull, concat_ws, concat, lit, substring, round, avg, max, min, length, lit, udf
from pyspark.sql.types import *
from pyspark import SparkContext
from pyspark.sql import SQLContext

# Get stats for all columns in a table
def get_table_stats(table, columns):
    udf_to_string = udf(lambda x: str(x), StringType())

    # Populate dataframes
    table_df = sqlContext.sql('SELECT ' + ','.join(columns) + ' FROM ' + table)

    # New df columns
    table_headers = ['_column', \
                    '_max', \
                    '_min', \
                    '_avg', \
                    '_max_length', \
                    '_min_length', \
                    '_avg_length']

    # Cycle through each column, obtain main stats and put in to dataframe
    for index, column in enumerate(table_df.columns):
        # Selecting relevant column value and also the length of those values.
        # The values must first be converted to strings to obtain the lengths, hence the use of the to_string UDF
        length_alias = column + '_lengths'

        col_df = table_df.select(column, length(udf_to_string(col(column))).alias(length_alias))

        # Relevant aggregates are determined, given aliases and stored as a single row dataframe
        aggs_df = col_df.agg(max(column).alias(table_headers[1]), \
                                min(column).alias(table_headers[2]), \
                                avg(column).alias(table_headers[3]), \
                                max(length_alias).alias(table_headers[4]), \
                                min(length_alias).alias(table_headers[5]), \
                                avg(length_alias).alias(table_headers[6]))

        # Add the column name as a column in our results row dataframe
        temp_raw_df = aggs_df.withColumn(table_headers[0], lit(column))

        # As we want to concatenate each row of column results to return a full dataframe of results, we must
        # ensure all values are of the same type, thus convert every value to a string
        temp_df = temp_raw_df.select([temp_raw_df['_column']] + [temp_raw_df[col_to_cast].cast(StringType()) for col_to_cast in temp_raw_df.columns[:-1]])

        # Update master_df after each column results are aggregated
        if index == 0:
            master_df = temp_df
        else:
            master_df = master_df.union(temp_df)

    return master_df

定义此代码并运行以下代码会给出如下所示的错误。

    >>> mydf = get_table_stats(table, ['index', 'name', 'age'])
    >>> mydf.show()
    UnicodeEncodeError: 'latin-1' codec can't encode characters in position 515-517: ordinal not in range(256)

任何人都知道Spark本身有问题吗？或者这是一个python问题？

我已经检查了我的iPython控制台中使用的编码，它是＆＃39; UTF-8＆＃39;。