Expand hadoop path

expand_hadoop_path.py

import re

def expand_hadoop_path(path):
   brace_patterns = re.findall('{{[^}]+}}', path)
   brace_patterns_values = [(brace_pattern, brace_pattern.replace('{{','').replace('}}','').split(',')) for brace_pattern in brace_patterns]

   replacements = []

   for (brace_pattern, values) in brace_patterns_values:
      if replacements:
         replacements_count = len(replacements)
         for i in range(replacements_count):
            replacement = replacements[i]
            for value in values:
               if not replacement.has_key(brace_pattern):
                  replacement[brace_pattern] = value
               else:
                  new_replacement = dict(replacement)
                  new_replacement[brace_pattern] = value
                  replacements.append(new_replacement)
      else:
         for value in values:
            replacement = dict()
            replacement[brace_pattern] = value
            replacements.append(replacement)

   expanded_paths = []

   for replacement in replacements:
      expanded_path = str(path)
      for brace_pattern, value in replacement.items():
         expanded_path = expanded_path.replace(brace_pattern, value)
      expanded_paths.append(expanded_path)

   return expanded_paths

testing_data = {
   's3://bucket/{{highlq,lowlq}}/201804/{{bid,nobid}}/': ['s3://bucket/highlq/201804/bid/', 's3://bucket/lowlq/201804/bid/', 's3://bucket/highlq/201804/nobid/', 's3://bucket/lowlq/201804/nobid/'],
   's3://bucket/{{highlq,midlq,lowlq}}/201804/{{bid,nobid}}/': ['s3://bucket/highlq/201804/bid/', 's3://bucket/midlq/201804/bid/', 's3://bucket/lowlq/201804/bid/', 's3://bucket/highlq/201804/nobid/', 's3://bucket/midlq/201804/nobid/', 's3://bucket/lowlq/201804/nobid/'],
   's3://bucket/{{highlq,lowlq}}/2018{{03,04}}/{{bid,nobid}}/': ['s3://bucket/highlq/201803/bid/', 's3://bucket/lowlq/201803/bid/', 's3://bucket/highlq/201804/bid/', 's3://bucket/lowlq/201804/bid/', 's3://bucket/highlq/201803/nobid/', 's3://bucket/lowlq/201803/nobid/', 's3://bucket/highlq/201804/nobid/', 's3://bucket/lowlq/201804/nobid/']
}

for (path, expected) in testing_data.items():
   assert set(expand_hadoop_path(path)) == set(expected), "expand_hadoop_path failed on #{path}"