GB2345556A

GB2345556A - Device for searching sorted list of data

Info

Publication number: GB2345556A
Application number: GB9900123A
Authority: GB
Inventors: Kevin Hughes
Original assignee: Fujitsu Services Ltd
Current assignee: Fujitsu Services Ltd
Priority date: 1999-01-06
Filing date: 1999-01-06
Publication date: 2000-07-12
Anticipated expiration: 2019-01-06
Also published as: GB9900123D0; GB2345556B

Abstract

A double-linked sorted list has a start pointer, which can be positioned to point to any item in the list. A search mechanism starts at the item currently pointed to by the start pointer, and searches in the direction of increasing or decreasing item values according to whether the search value is less than or greater than the value of the item currently pointed to by the start pointer. The start pointer is repositioned to the last item accessed. This provides reduced search times.

Description

LIST SEARCHING Background to the Invention This invention relates to a mechanism for performing fast list searching in a computer.

A list generally consists of a number of data items, chained together by means of pointers. A list may be single-linked, which means that each item has a pointer to the next item in the list only. Alternatively, a list may be double-linked, in which case each item has pointers to both the next and preceding items in the list. Double linking has the advantage that it is easier to remove an item from the list, since both the adjacent items can be immediately located and their pointers adjusted accordingly.

Each list may have a head pointer, which points to the first item on the list. A tail pointer may also be provided, pointing to the last item on the list.

Lists may be unordered, or ordered. In an unordered list, the items are not arranged in any particular order, and new items are added simply by inserting them at the head or tail of the list. In an ordered list, the items are linked together in order of some particular data attribute. In this case, inserting a new item is more difficult, since it is necessary to find the correct place to insert the item.

It is necessary to provide some mechanism for searching a list to find an item having a value that matches a specified search value. Searching such lists can be very time-consuming. The time taken to search a list generally depends on whether the list is unordered or ordered.

In the case of an unordered list, assuming that the position of the searched-for item is random, on average it is necessary to search half the list before the required item is found. In other words, if the time required to examine each item is t, the average search time to locate an item in a list of size N is T=N*t/2.

To be sure that an unordered list does not contain a particular value, it is necessary to search through every item in the list.

In this case, the time required is F=N*t.

The time taken to insert a new item in an unordered list is unrelated to the size of the list, since the item is simply inserted at the head or tail.

In the case of an ordered list, the average search time T to locate an item in a list of size N is again T=N*t/2. The average time taken to determine that an ordered list does not contain a particular item is F=N*t/2, i. e. it is the same as the search time. The time required to insert a new item equals the time taken to search for the correct place, plus the time i required to insert the item. Thus, in this case the insert time is I=N*t/2 + i.

The object of the present invention is to provide an improved list searching mechanism.

Summary of the Invention According to the invention, apparatus for storing and retrieving data in a computer comprises: a) a double-linked sorted list: b) a start pointer, movable to point to any item in the list; and c) a search mechanism, for searching for an item having a specified search value, starting at the item currently pointed to by the start pointer, the search mechanism searching in the direction of increasing or decreasing item values according to whether the search value is less than or greater than the value of the item currently pointed to by the start pointer.

It can be seen that, if the start pointer is near the middle of the list, on average only one-quarter of the list needs to be searched to find a required item. The average search time is therefore reduced to T=N*tl4. The average time required to determine that the list does not contain an item with a specified value is the same, i. e. F= N*t/4. The average time required to insert a new item is I=N*tl4 + i.

Thus, the present invention represents a significant improvement over conventional list searching mechanisms (with the sole exception that inserting a new item in an unordered list is, of course, faster).

One embodiment of the invention will now be described by way of example with reference to the accompanying drawings.

Brief Description of the Drawings Figure 1 is a schematic diagram representing a list structure.

Figure 2 is a flow chart showing the operation of a search mechanism for searching the list.

Description of an Embodiment of the Invention Figure 1 shows a double-linked list 10, comprising a number of data items 11-15. Each item contains a forward pointer, pointing to the next item on the list, and a backward pointer, pointing to the preceding item on the list. At the head of the list 15, the forward pointer has a null value. Similarly, at the tail end 11 of the list, the backward pointer has a null value.

The items on list 10 are linked together in order of increasing values, with the tail item 11 having the smallest value and the head item 15 having the largest value A movable start pointer 16 can be positioned to point to any item in the list. Initially, it is assumed it is positioned somewhere near the middle of the list. As will be described, each time an item is accessed (inserted or found), the start pointer is repositioned to point to that item.

The list has a search mechanism 17, for searching for an item having a specified search value. Preferably, the search mechanism is implemented in software, running on conventional computer hardware. Alternatively, it could be implemented by special-purpose hardware.

Figure 2 shows the operation of the search mechanism.

(Step 20) The search mechanism first compares the search value with the value of the item currently pointed to by the start pointer 16. If these values are equal, the search mechanism returns"success", indicating that the required item has been found.

(Step 21) If the search value is less than the value of the iter currently pointed to by the start pointer, the search mechanism proceeds to search sequentially down the list, in the direction of decreasing item values, i. e. towards the tail item 11. If the required item is found, the search returns"success". If, however, the tail item 11 is reached without finding the required item the search returns"failure".

(Step 22) If on the other hand the search value is greater than the value of the item currently pointed to by the start pointer, the search mechanism proceeds to search sequentially up the list, in the direction of increasing item values, i. e. towards the head item 15. As before, if the required item is found, the search returns"success". If, however, the head item 15 is reached without finding the required item the search returns "failure".

(Step 23) If the required item is found, the start pointer 16 is repositioned to point to the found item.

The process for inserting a new item is similar to that for searching. If the value of the item to be inserted is less than the value of the item currently pointed to by the start pointer, the list is searched in the direction of decreasing item values, until an item is found with a value less than that of the item to be inserted, or the tail of the list is reached. Conversely, if the value of the item to be inserted is greater than the value of the item currently pointed to by the start pointer, the list is searched in the direction of increasing item values, until an item is found with a value greater than that of the item to be inserted, or the head of the list is reached. In either case, the start pointer is repositioned to point to the newly inserted item.

Repositioning the start pointer to the last item accessed (inserted or found) serves two purposes.

* First, it is a simple expedient, which tends to leave the start pointer, on average, at the centre of the list. This helps to minimise search times.

It has the additional benefit that, if a series of adjacent (or nearly adjacent) items are accessed together, the access time is further reduced, since items after the first in the series will be near to the start pointer.

Some possible modifications It will be appreciated that many modifications may be made to the system described above without departing from the scope of the present invention.

In one possible modification, the list may be configured as a circular list, such that the forward pointer of the head item 15 points to the tail item 11, and the backward pointer of the tail item 11 points to the head item 15. In this case, the search may proceed until either the required item is found, or the search arrives back at the start pointer.

Claims

CLAIMS 1. Apparatus for storing and retrieving data in a computer comprising: a) a double-linked sorted list: b) a start pointer, movable to point to any item in the list ; and c) a search mechanism, for searching for an item having a specified search value, starting at the item currently pointed to by the start pointer, the search mechanism searching in the direction of increasing or decreasing item values according to whether the search value is less than or greater than the value of the item currently pointed to by the start pointer.
2. Apparatus according to Claim 1 including means for repositioning the start pointer to the last item accessed.
3. A method for storing and retrieving data in a computer comprising : a) forming a double-linked sorted list: b) providing a start pointer, movable to point to any item in the list; and c) searching for an item having a specified search value, starting at the item currently pointed to by the start pointer, and proceeding in the direction of increasing or decreasing item values according to whether the search value is less than or greater than the value of the item currently pointed to by the start pointer.
4. A method according to Claim 3 including repositioning the start pointer to the last item accessed.
5. An information carrier, holding a program for implementing a method according to Claim 3 or 4.
6. A search mechanism substantially as hereinbefore described with reference to the accompanying drawings.