------------------------------------------------------------------------------
-- --
-- GNAT RUN-TIME COMPONENTS --
-- --
-- G N A T . D Y N A M I C _ H T A B L E S --
-- --
-- S p e c --
-- --
-- Copyright (C) 1995-2019, AdaCore --
-- --
-- GNAT is free software; you can redistribute it and/or modify it under --
-- terms of the GNU General Public License as published by the Free Soft- --
-- ware Foundation; either version 3, or (at your option) any later ver- --
-- sion. GNAT is distributed in the hope that it will be useful, but WITH- --
-- OUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY --
-- or FITNESS FOR A PARTICULAR PURPOSE. --
-- --
-- As a special exception under Section 7 of GPL version 3, you are granted --
-- additional permissions described in the GCC Runtime Library Exception, --
-- version 3.1, as published by the Free Software Foundation. --
-- --
-- You should have received a copy of the GNU General Public License and --
-- a copy of the GCC Runtime Library Exception along with this program; --
-- see the files COPYING3 and COPYING.RUNTIME respectively. If not, see --
-- . --
-- --
-- GNAT was originally developed by the GNAT team at New York University. --
-- Extensive contributions were provided by Ada Core Technologies Inc. --
-- --
------------------------------------------------------------------------------
-- Hash table searching routines
-- This package contains two separate packages. The Simple_HTable package
-- provides a very simple abstraction that associates one element to one key
-- value and takes care of all allocations automatically using the heap. The
-- Static_HTable package provides a more complex interface that allows full
-- control over allocation.
-- This package provides a facility similar to that of GNAT.HTable, except
-- that this package declares types that can be used to define dynamic
-- instances of hash tables, while instantiations in GNAT.HTable creates a
-- single instance of the hash table.
-- Note that this interface should remain synchronized with those in
-- GNAT.HTable to keep as much coherency as possible between these two
-- related units.
pragma Compiler_Unit_Warning;
package GNAT.Dynamic_HTables is
function Hash_Two_Keys
(Left : Bucket_Range_Type;
Right : Bucket_Range_Type) return Bucket_Range_Type;
pragma Inline (Hash_Two_Keys);
-- Obtain the hash value of keys Left and Right
-------------------
-- Static_HTable --
-------------------
-- A low-level Hash-Table abstraction, not as easy to instantiate as
-- Simple_HTable. This mirrors the interface of GNAT.HTable.Static_HTable,
-- but does require dynamic allocation (since we allow multiple instances
-- of the table). The model is that each Element contains its own Key that
-- can be retrieved by Get_Key. Furthermore, Element provides a link that
-- can be used by the HTable for linking elements with same hash codes:
-- Element
-- +-------------------+
-- | Key |
-- +-------------------+
-- : other data :
-- +-------------------+
-- | Next Elmt |
-- +-------------------+
generic
type Header_Num is range <>;
-- An integer type indicating the number and range of hash headers
type Element (<>) is limited private;
-- The type of element to be stored
type Elmt_Ptr is private;
-- The type used to reference an element (will usually be an access
-- type, but could be some other form of type such as an integer type).
Null_Ptr : Elmt_Ptr;
-- The null value of the Elmt_Ptr type
with function Next (E : Elmt_Ptr) return Elmt_Ptr;
with procedure Set_Next (E : Elmt_Ptr; Next : Elmt_Ptr);
-- The type must provide an internal link for the sake of the
-- staticness of the HTable.
type Key is limited private;
with function Get_Key (E : Elmt_Ptr) return Key;
with function Hash (F : Key) return Header_Num;
with function Equal (F1 : Key; F2 : Key) return Boolean;
package Static_HTable is
type Instance is private;
Nil : constant Instance;
procedure Reset (T : in out Instance);
-- Resets the hash table by releasing all memory associated with it. The
-- hash table can safely be reused after this call. For the most common
-- case where Elmt_Ptr is an access type, and Null_Ptr is null, this is
-- only needed if the same table is reused in a new context. If Elmt_Ptr
-- is other than an access type, or Null_Ptr is other than null, then
-- Reset must be called before the first use of the hash table.
procedure Set (T : in out Instance; E : Elmt_Ptr);
-- Insert the element pointer in the HTable
function Get (T : Instance; K : Key) return Elmt_Ptr;
-- Returns the latest inserted element pointer with the given Key or
-- null if none.
procedure Remove (T : Instance; K : Key);
-- Removes the latest inserted element pointer associated with the given
-- key if any, does nothing if none.
function Get_First (T : Instance) return Elmt_Ptr;
-- Returns Null_Ptr if the Htable is empty, otherwise returns one
-- unspecified element. There is no guarantee that 2 calls to this
-- function will return the same element.
function Get_Next (T : Instance) return Elmt_Ptr;
-- Returns an unspecified element that has not been returned by the same
-- function since the last call to Get_First or Null_Ptr if there is no
-- such element or Get_First has never been called. If there is no call
-- to 'Set' in between Get_Next calls, all the elements of the Htable
-- will be traversed.
private
type Table_Type is array (Header_Num) of Elmt_Ptr;
type Instance_Data is record
Table : Table_Type;
Iterator_Index : Header_Num;
Iterator_Ptr : Elmt_Ptr;
Iterator_Started : Boolean := False;
end record;
type Instance is access all Instance_Data;
Nil : constant Instance := null;
end Static_HTable;
-------------------
-- Simple_HTable --
-------------------
-- A simple hash table abstraction, easy to instantiate, easy to use.
-- The table associates one element to one key with the procedure Set.
-- Get retrieves the Element stored for a given Key. The efficiency of
-- retrieval is function of the size of the Table parameterized by
-- Header_Num and the hashing function Hash.
generic
type Header_Num is range <>;
-- An integer type indicating the number and range of hash headers
type Element is private;
-- The type of element to be stored
No_Element : Element;
-- The object that is returned by Get when no element has been set for
-- a given key
type Key is private;
with function Hash (F : Key) return Header_Num;
with function Equal (F1 : Key; F2 : Key) return Boolean;
package Simple_HTable is
type Instance is private;
Nil : constant Instance;
type Key_Option (Present : Boolean := False) is record
case Present is
when True => K : Key;
when False => null;
end case;
end record;
procedure Set (T : in out Instance; K : Key; E : Element);
-- Associates an element with a given key. Overrides any previously
-- associated element.
procedure Reset (T : in out Instance);
-- Releases all memory associated with the table. The table can be
-- reused after this call (it is automatically allocated on the first
-- access to the table).
function Get (T : Instance; K : Key) return Element;
-- Returns the Element associated with a key or No_Element if the given
-- key has not associated element
procedure Remove (T : Instance; K : Key);
-- Removes the latest inserted element pointer associated with the given
-- key if any, does nothing if none.
function Get_First (T : Instance) return Element;
-- Returns No_Element if the Htable is empty, otherwise returns one
-- unspecified element. There is no guarantee that two calls to this
-- function will return the same element, if the Htable has been
-- modified between the two calls.
function Get_First_Key (T : Instance) return Key_Option;
-- Returns an option type giving an unspecified key. If the Htable
-- is empty, the discriminant will have field Present set to False,
-- otherwise its Present field is set to True and the field K contains
-- the key. There is no guarantee that two calls to this function will
-- return the same key, if the Htable has been modified between the two
-- calls.
function Get_Next (T : Instance) return Element;
-- Returns an unspecified element that has not been returned by the
-- same function since the last call to Get_First or No_Element if
-- there is no such element. If there is no call to 'Set' in between
-- Get_Next calls, all the elements of the Htable will be traversed.
-- To guarantee that all the elements of the Htable will be traversed,
-- no modification of the Htable (Set, Reset, Remove) should occur
-- between a call to Get_First and subsequent consecutive calls to
-- Get_Next, until one of these calls returns No_Element.
function Get_Next_Key (T : Instance) return Key_Option;
-- Same as Get_Next except that this returns an option type having field
-- Present set either to False if there no key never returned before by
-- either Get_First_Key or this very same function, or to True if there
-- is one, with the field K containing the key specified as before. The
-- same restrictions apply as Get_Next.
private
type Element_Wrapper;
type Elmt_Ptr is access all Element_Wrapper;
type Element_Wrapper is record
K : Key;
E : Element;
Next : Elmt_Ptr;
end record;
procedure Set_Next (E : Elmt_Ptr; Next : Elmt_Ptr);
function Next (E : Elmt_Ptr) return Elmt_Ptr;
function Get_Key (E : Elmt_Ptr) return Key;
package Tab is new Static_HTable
(Header_Num => Header_Num,
Element => Element_Wrapper,
Elmt_Ptr => Elmt_Ptr,
Null_Ptr => null,
Set_Next => Set_Next,
Next => Next,
Key => Key,
Get_Key => Get_Key,
Hash => Hash,
Equal => Equal);
type Instance is new Tab.Instance;
Nil : constant Instance := Instance (Tab.Nil);
end Simple_HTable;
-------------------------
-- Dynamic_Hash_Tables --
-------------------------
-- The following package offers a hash table abstraction with the following
-- characteristics:
--
-- * Dynamic resizing based on load factor
-- * Creation of multiple instances, of different sizes
-- * Iterable keys
--
-- This type of hash table is best used in scenarios where the size of the
-- key set is not known. The dynamic resizing aspect allows for performance
-- to remain within reasonable bounds as the size of the key set grows.
--
-- The following use pattern must be employed when operating this table:
--
-- Table : Dynamic_Hash_Table := Create ();
--
--
--
-- Destroy (Table);
--
-- The destruction of the table reclaims all storage occupied by it.
-- The following type denotes the multiplicative factor used in expansion
-- and compression of the hash table.
subtype Factor_Type is Bucket_Range_Type range 2 .. 100;
-- The following type denotes the threshold range used in expansion and
-- compression of the hash table.
subtype Threshold_Type is Long_Float range 0.0 .. Long_Float'Last;
generic
type Key_Type is private;
type Value_Type is private;
-- The types of the key-value pairs stored in the hash table
No_Value : Value_Type;
-- An indicator for a non-existent value
Expansion_Threshold : Threshold_Type;
Expansion_Factor : Factor_Type;
-- Once the load factor goes over Expansion_Threshold, the size of the
-- buckets is increased using the formula
--
-- New_Size = Old_Size * Expansion_Factor
--
-- An Expansion_Threshold of 1.5 and Expansion_Factor of 2 indicate that
-- the size of the buckets will be doubled once the load factor exceeds
-- 1.5.
Compression_Threshold : Threshold_Type;
Compression_Factor : Factor_Type;
-- Once the load factor drops below Compression_Threshold, the size of
-- the buckets is decreased using the formula
--
-- New_Size = Old_Size / Compression_Factor
--
-- A Compression_Threshold of 0.5 and Compression_Factor of 2 indicate
-- that the size of the buckets will be halved once the load factor
-- drops below 0.5.
with function "="
(Left : Key_Type;
Right : Key_Type) return Boolean;
with procedure Destroy_Value (Val : in out Value_Type);
-- Value destructor
with function Hash (Key : Key_Type) return Bucket_Range_Type;
-- Map an arbitrary key into the range of buckets
package Dynamic_Hash_Tables is
----------------------
-- Table operations --
----------------------
-- The following type denotes a hash table handle. Each instance must be
-- created using routine Create.
type Dynamic_Hash_Table is private;
Nil : constant Dynamic_Hash_Table;
function Contains
(T : Dynamic_Hash_Table;
Key : Key_Type) return Boolean;
-- Determine whether key Key exists in hash table T
function Create (Initial_Size : Positive) return Dynamic_Hash_Table;
-- Create a new table with bucket capacity Initial_Size. This routine
-- must be called at the start of a hash table's lifetime.
procedure Delete
(T : Dynamic_Hash_Table;
Key : Key_Type);
-- Delete the value which corresponds to key Key from hash table T. The
-- routine has no effect if the value is not present in the hash table.
-- This action will raise Iterated if the hash table has outstanding
-- iterators. If the load factor drops below Compression_Threshold, the
-- size of the buckets is decreased by Copression_Factor.
procedure Destroy (T : in out Dynamic_Hash_Table);
-- Destroy the contents of hash table T, rendering it unusable. This
-- routine must be called at the end of a hash table's lifetime. This
-- action will raise Iterated if the hash table has outstanding
-- iterators.
function Get
(T : Dynamic_Hash_Table;
Key : Key_Type) return Value_Type;
-- Obtain the value which corresponds to key Key from hash table T. If
-- the value does not exist, return No_Value.
function Is_Empty (T : Dynamic_Hash_Table) return Boolean;
-- Determine whether hash table T is empty
function Present (T : Dynamic_Hash_Table) return Boolean;
-- Determine whether hash table T exists
procedure Put
(T : Dynamic_Hash_Table;
Key : Key_Type;
Value : Value_Type);
-- Associate value Value with key Key in hash table T. If the table
-- already contains a mapping of the same key to a previous value, the
-- previous value is overwritten. This action will raise Iterated if
-- the hash table has outstanding iterators. If the load factor goes
-- over Expansion_Threshold, the size of the buckets is increased by
-- Expansion_Factor.
procedure Reset (T : Dynamic_Hash_Table);
-- Destroy the contents of hash table T, and reset it to its initial
-- created state. This action will raise Iterated if the hash table
-- has outstanding iterators.
function Size (T : Dynamic_Hash_Table) return Natural;
-- Obtain the number of key-value pairs in hash table T
-------------------------
-- Iterator operations --
-------------------------
-- The following type represents a key iterator. An iterator locks
-- all mutation operations, and unlocks them once it is exhausted.
-- The iterator must be used with the following pattern:
--
-- Iter := Iterate (My_Table);
-- while Has_Next (Iter) loop
-- Key := Next (Iter);
-- . . .
-- end loop;
--
-- It is possible to advance the iterator by using Next only, however
-- this risks raising Iterator_Exhausted.
type Iterator is private;
function Has_Next (Iter : Iterator) return Boolean;
-- Determine whether iterator Iter has more keys to examine. If the
-- iterator has been exhausted, restore all mutation functionality of
-- the associated hash table.
function Iterate (T : Dynamic_Hash_Table) return Iterator;
-- Obtain an iterator over the keys of hash table T. This action locks
-- all mutation functionality of the associated hash table.
procedure Next (Iter : in out Iterator; Key : out Key_Type);
-- Return the current key referenced by iterator Iter and advance to
-- the next available key. If the iterator has been exhausted and
-- further attempts are made to advance it, this routine restores
-- mutation functionality of the associated hash table, and then
-- raises Iterator_Exhausted.
private
-- The following type represents a doubly linked list node used to
-- store a key-value pair. There are several reasons to use a doubly
-- linked list:
--
-- * Most read and write operations utilize the same primitve
-- routines to locate, create, and delete a node, allowing for
-- greater degree of code sharing.
--
-- * Special cases are eliminated by maintaining a circular node
-- list with a dummy head (see type Bucket_Table).
--
-- A node is said to be "valid" if it is non-null, and does not refer to
-- the dummy head of some bucket.
type Node;
type Node_Ptr is access all Node;
type Node is record
Key : Key_Type;
Value : Value_Type := No_Value;
-- Key-value pair stored in a bucket
Prev : Node_Ptr := null;
Next : Node_Ptr := null;
end record;
-- The following type represents a bucket table. Each bucket contains a
-- circular doubly linked list of nodes with a dummy head. Initially,
-- the head does not refer to itself. This is intentional because it
-- improves the performance of creation, compression, and expansion by
-- avoiding a separate pass to link a head to itself. Several routines
-- ensure that the head is properly formed.
type Bucket_Table is array (Bucket_Range_Type range <>) of aliased Node;
type Bucket_Table_Ptr is access Bucket_Table;
-- The following type represents a hash table
type Dynamic_Hash_Table_Attributes is record
Buckets : Bucket_Table_Ptr := null;
-- Reference to the compressing / expanding buckets
Initial_Size : Bucket_Range_Type := 0;
-- The initial size of the buckets as specified at creation time
Iterators : Natural := 0;
-- Number of outstanding iterators
Pairs : Natural := 0;
-- Number of key-value pairs in the buckets
end record;
type Dynamic_Hash_Table is access Dynamic_Hash_Table_Attributes;
Nil : constant Dynamic_Hash_Table := null;
-- The following type represents a key iterator
type Iterator is record
Curr_Idx : Bucket_Range_Type := 0;
-- Index of the current bucket being examined. This index is always
-- kept within the range of the buckets.
Curr_Nod : Node_Ptr := null;
-- Reference to the current node being examined within the current
-- bucket. The invariant of the iterator requires that this field
-- always point to a valid node. A value of null indicates that the
-- iterator is exhausted.
Table : Dynamic_Hash_Table := null;
-- Reference to the associated hash table
end record;
end Dynamic_Hash_Tables;
end GNAT.Dynamic_HTables;