Data source normalization

normalization.py

df = google.select(
    f.lit('google').alias('source'),
    f.col('id').alias('source_id'),
    f.col('name'), f.col('description'),
    f.col('manufacturer'),
    f.col('price')
  )\
  .union(
    amazon.select(
      f.lit('amazon').alias('source'),
        f.col('id').alias('source_id'),
        f.col('title').alias('name'),
        f.col('description'),
        f.col('manufacturer'),
        f.col('price')
       )
    )

def trim_to_null(c):
  return (
    f.lower(
      f.when(f.trim(f.col(c)) == '', None)
      .when(f.trim(f.col(c)) == 'null', None)
      .otherwise(f.trim(f.col(c)))
    )
  )

STRING_COLS = ['name', 'description', 'manufacturer']
for c in STRING_COLS:
  df = df.withColumn(c, f.lower(trim_to_null(c)))

STRING_NUM_COLS = ['price']
for c in STRING_NUM_COLS:
  df = df.withColumn(c, trim_to_null(c).cast('float'))
  
# hyphenated words and version numbers seems salient to product name
# treat them differently by concatenating
def replace_contiguous_special_char(c, replace_str=''):
  return (
    f.regexp_replace(c, "(?<=(\d|\w))(\.|-|\')(?=(\d|\w))", replace_str)
  )

def replace_special_char(c, replace_str=' '):
  return (
    f.regexp_replace(c, "[\W]", replace_str)
  )



processed_df = df.withColumn('name', replace_special_char('name'))\
  .withColumn('description', replace_special_char('description'))\
  .withColumn('manufacturer', replace_special_char('manufacturer'))
  
  
display(processed_df)

what is f here? I suspect

import pyspark.sql.functions as f

It seems that the function replace_contiguous_special_char isn't used anywhere in the Medium code. What purpose does it serve?