Home > Articles

Python 3 Data Structures

  • Print
  • + Share This

Communications director of the Python Software Foundation, Doug Hellman, covers covers two modules in Python 3 related to memory management.

Save 35% off the list price* of the related book or multi-format eBook (EPUB + MOBI + PDF) with discount code ARTICLE.
* See informit.com/terms

This chapter is from the book

Python includes several standard programming data structures, such as list, tuple, dict, and set, as part of its built-in types. Many applications do not require other structures, but when they do, the standard library provides powerful and well-tested versions that are ready to be used.

The enum (page 66) module provides an implementation of an enumeration type, with iteration and comparison capabilities. It can be used to create well-defined symbols for values, instead of using literal strings or integers.

The collections (page 75) module includes implementations of several data structures that extend those found in other modules. For example, Deque is a double-ended queue, which allows the addition or removal of items from either end. The defaultdict is a dictionary that responds with a default value if a key is missing, while OrderedDict remembers the sequence in which items are added to it. namedtuple extends the normal tuple to give each member item an attribute name in addition to a numeric index.

For large amounts of data, an array (page 98) may make more efficient use of memory than a list. Since the array is limited to a single data type, it can use a more compact memory representation than a general-purpose list. At the same time, array instances can be manipulated using many of the same methods as a list, so it may be possible to replace a list with an array in an application without a lot of other changes.

Sorting items in a sequence is a fundamental aspect of data manipulation. Python’s list includes a sort() method, but sometimes it is more efficient to maintain a list in sorted order without re-sorting it each time its contents are changed. The functions in heapq (page 103) modify the contents of a list while preserving the sort order of the list with low overhead.

Another option for building sorted lists or arrays is bisect (page 109). It uses a binary search to find the insertion point for new items, and is an alternative to repeatedly sorting a list that changes frequently.

Although the built-in list can simulate a queue using the insert() and pop() methods, it is not thread-safe. For true ordered communication between threads use the queue (page 111) module. multiprocessing (page 586) includes a version of a Queue that works between processes, making it easier to convert a multithreaded program to use processes instead.

struct (page 117) is useful for decoding data from another application, perhaps coming from a binary file or stream of data, into Python’s native types for easier manipulation.

This chapter covers two modules related to memory management. For highly interconnected data structures, such as graphs and trees, use weakref (page 121) to maintain references while still allowing the garbage collector to clean up objects after they are no longer needed. Use the functions in copy (page 130) for duplicating data structures and their contents, including making recursive copies with deepcopy().

Debugging data structures can be time consuming, especially when wading through printed output of large sequences or dictionaries. Use pprint (page 136) to create easy-to-read representations that can be printed to the console or written to a log file for easier debugging.

Finally, if the available types do not meet the requirements, subclass one of the native types and customize it, or build a new container type using one of the abstract base classes defined in collections (page 75) as a starting point.

2.1 enum: Enumeration Type

The enum module defines an enumeration type with iteration and comparison capabilities. It can be used to create well-defined symbols for values, instead of using literal integers or strings.

2.1.1 Creating Enumerations

A new enumeration is defined using the class syntax by subclassing Enum and adding class attributes describing the values.

Listing 2.1: enum_create.py

      import enum


      class BugStatus(enum.Enum):

          new = 7
          incomplete = 6
          invalid = 5
          wont_fix = 4
          in_progress = 3
          fix_committed = 2
          fix_released = 1


      print('\nMember name: {}'.format(BugStatus.wont_fix.name))
      print('Member value: {}'.format(BugStatus.wont_fix.value))

The members of the Enum are converted to instances as the class is parsed. Each instance has a name property corresponding to the member name and a value property corresponding to the value assigned to the name in the class definition.

      $ python3 enum_create.py


      Member name: wont_fix
      Member value: 4

2.1.2 Iteration

Iterating over the enum class produces the individual members of the enumeration.

Listing 2.2: enum_iterate.py

     import enum


     class BugStatus(enum.Enum):

         new = 7
         incomplete = 6
         invalid = 5
         wont_fix = 4
         in_progress = 3
         fix_committed = 2
         fix_released = 1


     for status in BugStatus:
         print('{:15} = {}'.format(status.name, status.value))

The members are produced in the order they are declared in the class definition. The names and values are not used to sort them in any way.

     $ python3 enum_iterate.py

     new             =   7
     incomplete      =   6
     invalid         =   5
     wont_fix        =   4
     in_progress     =   3
     fix_committed   =   2
     fix_released    =   1

2.1.3 Comparing Enums

Because enumeration members are not ordered, they support only comparison by identity and equality.

Listing 2.3: enum_comparison.py

     import enum


     class BugStatus(enum.Enum):

         new = 7
         incomplete = 6
         invalid = 5
         wont_fix = 4
         in_progress = 3
         fix_committed = 2
         fix_released = 1


     actual_state = BugStatus.wont_fix
     desired_state = BugStatus.fix_released

     print('Equality:',
           actual_state == desired_state,
           actual_state == BugStatus.wont_fix)
     print('Identity:',
           actual_state is desired_state,
           actual_state is BugStatus.wont_fix)
     print('Ordered by value:')
     try:
         print('\n'.join(' ' + s.name for s in sorted(BugStatus)))
     except TypeError as err:
         print(' Cannot sort: {}'.format(err))

The greater-than and less-than comparison operators raise TypeError exceptions.

     $ python3 enum_comparison.py

     Equality: False True
     Identity: False True
     Ordered by value:
       Cannot sort: unorderable types: BugStatus() < BugStatus()

Use the IntEnum class for enumerations where the members need to behave more like numbers—for example, to support comparisons.

Listing 2.4: enum_intenum.py

     import enum


     class BugStatus(enum.IntEnum):

         new = 7
         incomplete = 6
         invalid = 5
         wont_fix = 4
         in_progress = 3
         fix_committed = 2
         fix_released = 1
     print('Ordered by value:')
     print('\n'.join(' ' + s.name for s in sorted(BugStatus)))
     $ python3 enum_intenum.py

     Ordered by value:
       fix_released
       fix_committed
       in_progress
       wont_fix
       invalid
       incomplete
       new

2.1.4 Unique Enumeration Values

Enum members with the same value are tracked as alias references to the same member object. Aliases do not cause repeated values to be present in the iterator for the Enum.

Listing 2.5: enum_aliases.py

     import enum


     class BugStatus(enum.Enum):

         new = 7
         incomplete = 6
         invalid = 5
         wont_fix = 4
         in_progress = 3
         fix_committed = 2
         fix_released = 1
         
         by_design = 4
         closed = 1


     for status in BugStatus:
         print('{:15} = {}'.format(status.name, status.value))

     print('\nSame: by_design is wont_fix: ',
           BugStatus.by_design is BugStatus.wont_fix)
     print('Same: closed is fix_released: ',
           BugStatus.closed is BugStatus.fix_released)

Because by_design and closed are aliases for other members, they do not appear separately in the output when iterating over the Enum. The canonical name for a member is the first name attached to the value.

     $ python3 enum_aliases.py

     new             =   7
     incomplete      =   6
     invalid         =   5
     wont_fix        =   4
     in_progress     =   3
     fix_committed   =   2
     fix_released    =   1

     Same: by_design is wont_fix: True
     Same: closed is fix_released: True

To require all members to have unique values, add the @unique decorator to the Enum.

Listing 2.6: enum_unique_enforce.py

     import enum


     @enum.unique
     class BugStatus(enum.Enum):

         new = 7
         incomplete = 6
         invalid = 5
         wont_fix = 4
         in_progress = 3
         fix_committed = 2
         fix_released = 1
         
         # This will trigger an error with unique applied.
         by_design = 4
         closed = 1

Members with repeated values trigger a ValueError exception when the Enum class is being interpreted.

     $ python3 enum_unique_enforce.py

     Traceback (most recent call last):
       File "enum_unique_enforce.py", line 11, in <module>
         class BugStatus(enum.Enum):
       File ".../lib/python3.5/enum.py", line 573, in unique
         (enumeration, alias_details))
     ValueError: duplicate values found in <enum 'BugStatus'>:
     by_design -> wont_fix, closed -> fix_released

2.1.5 Creating Enumerations Programmatically

In some cases, it is more convenient to create enumerations programmatically, rather than hard-coding them in a class definition. For those situations, Enum also supports passing the member names and values to the class constructor.

Listing 2.7: enum_programmatic_create.py

     import enum


     BugStatus = enum.Enum(
         value='BugStatus',
         names=('fix_released fix_committed in_progress '
                'wont_fix invalid incomplete new'),
     )

     print('Member: {}'.format(BugStatus.new))

     print('\nAll members:')
     for status in BugStatus:
         print('{:15} = {}'.format(status.name, status.value))

The value argument is the name of the enumeration, which is used to build the representation of members. The names argument lists the members of the enumeration. When a single string is passed, it is split on whitespace and commas, and the resulting tokens are used as names for the members, which are automatically assigned values starting with 1.

     $ python3 enum_programmatic_create.py

     Member: BugStatus.new

     All members:
     fix_released    = 1
     fix_committed   = 2
     in_progress     = 3
     wont_fix        = 4
     invalid         = 5
     incomplete      = 6
     new             = 7

For more control over the values associated with members, the names string can be replaced with a sequence of two-part tuples or a dictionary mapping names to values.

Listing 2.8: enum_programmatic_mapping.py

     import enum


     BugStatus = enum.Enum(
         value='BugStatus',
         names=[
             ('new', 7),
             ('incomplete', 6),
             ('invalid', 5),
             ('wont_fix', 4),
             ('in_progress', 3),
             ('fix_committed', 2),
             ('fix_released', 1),
         ],
     )

     print('All members:')
     for status in BugStatus:
         print('{:15} = {}'.format(status.name, status.value))

In this example, a list of two-part tuples is given instead of a single string containing only the member names. This makes it possible to reconstruct the BugStatus enumeration with the members in the same order as the version defined in enum_create.py.

     $ python3 enum_programmatic_mapping.py

     All members:
     new             = 7
     incomplete      = 6
     invalid         = 5
     wont_fix        = 4
     in_progress     = 3
     fix_committed   = 2
     fix_released    = 1

2.1.6 Non-integer Member Values

Enum member values are not restricted to integers. In fact, any type of object can be associated with a member. If the value is a tuple, the members are passed as individual arguments to __init__().

Listing 2.9: enum_tuple_values.py

     import enum


     class BugStatus(enum.Enum):

         new = (7, ['incomplete',
                    'invalid',
                    'wont_fix',
                    'in_progress'])
         incomplete = (6, ['new', 'wont_fix'])
         invalid = (5, ['new'])
         wont_fix = (4, ['new'])
         in_progress = (3, ['new', 'fix_committed'])
         fix_committed = (2, ['in_progress', 'fix_released'])
         fix_released = (1, ['new'])
         
         def __init__(self, num, transitions):
             self.num = num
             self.transitions = transitions
         
         def can_transition(self, new_state):
             return new_state.name in self.transitions


     print('Name:', BugStatus.in_progress)
     print('Value:', BugStatus.in_progress.value)
     print('Custom attribute:', BugStatus.in_progress.transitions)
     print('Using attribute:',
           BugStatus.in_progress.can_transition(BugStatus.new))

In this example, each member value is a tuple containing the numerical ID (such as might be stored in a database) and a list of valid transitions away from the current state.

     $ python3 enum_tuple_values.py

     Name: BugStatus.in_progress
     Value: (3, ['new', 'fix_committed'])
     Custom attribute: ['new', 'fix_committed']
     Using attribute: True

For more complex cases, tuples might become unwieldy. Since member values can be any type of object, dictionaries can be used for cases where there are a lot of separate attributes to track for each enum value. Complex values are passed directly to __init__() as the only argument other than self.

Listing 2.10: enum_complex_values.py

import enum


class BugStatus(enum.Enum):

   new = {
       'num': 7,
       'transitions': [
           'incomplete',
           'invalid',
           'wont_fix',
           'in_progress',
       ],
   }
   incomplete = {
       'num': 6,
       'transitions': ['new', 'wont_fix'],
   }
   invalid = {
       'num': 5,
       'transitions': ['new'],
   }
   wont_fix = {
       'num': 4,
       'transitions': ['new'],
   }
   in_progress = {
       'num': 3,
       'transitions': ['new', 'fix_committed'],
   }
   fix_committed = {
       'num': 2,
       'transitions': ['in_progress', 'fix_released'],
   }
   fix_released = {
       'num': 1,
       'transitions': ['new'],
   }

    def __init__(self, vals):
        self.num = vals['num']
        self.transitions = vals['transitions']

    def can_transition(self, new_state):
        return new_state.name in self.transitions


print('Name:', BugStatus.in_progress)
print('Value:', BugStatus.in_progress.value)
print('Custom attribute:', BugStatus.in_progress.transitions)
print('Using attribute:',
      BugStatus.in_progress.can_transition(BugStatus.new))

This example expresses the same data as the previous example, using dictionaries rather than tuples.

      $ python3 enum_complex_values.py

      Name: BugStatus.in_progress
      Value: {'transitions': ['new', 'fix_committed'], 'num': 3}
      Custom attribute: ['new', 'fix_committed']
      Using attribute: True
  • + Share This
  • 🔖 Save To Your Account