JPH0324617A - Data processing system - Google Patents

Abstract

PURPOSE:To reduce the memory using capacity of sort processing by providing the data processing system with a sorting means, a sorting work storing means and a merging means. CONSTITUTION:The sorting means divides a record group having key items and relative work address items and only key values and relative address values from respective records are stored and rearranged in a sorting table. The sorting work storing means successively stores the records of work files indicated by the relative address values in the sorting table in a sorting work file. The merging means sets up the records of the minimum key values in of respective groups the sorting work file in each merging table, compares recordes in the table and outputs the minimum key record to a work file. Since the head records of respective groups of the sorting work file formed by dividing a file to be sorted and rearranging only the key values and relative address values are merged, memory capacity to be processed can be sharply reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a data processing method that performs rearrangement or sorting in the field of information processing. [Prior Art] Conventionally, this kind of sorting processing has been performed by expanding all records of a file to be sorted into memory and then repeatedly comparing the sizes of adjacent data.

FIG. 8 is a flowchart showing the procedure of such conventional data processing, that is, sorting processing.

Referring to FIG. 8, in the conventional sorting process, ``''l# is first set to the value of subscript 1, and the value obtained by adding ``''11 to the value of subscript 1 is set to subscript J (step 2- 1.
2-2). Step f2-3 is pressed, and if the first KEY value in the sort table is as large as the J-th KEY value, the first sort data and the J-th sort data are exchanged (Step 2-4.2- 5, Step 2
-6).

As a result of the above processing, if "1#" is set to the value of subscript l, "'1" is added to the value of subscript 1, and if a value other than 11" is set to the value of subscript 1, Subtract "l#" from the value of subscript 1 (Step 2-9.2-10).

In step 7"2-3, if the first KEY value of the sort table is smaller than the Jth KEY value, the subscript l
``'l'' is added to the value of (step f2-7).

Repeat the processing from step 7°2-2 to step 7°2-101 until the value of subscript l matches the number of sort table data items (step 2-11), and when they match, end the sorting process. .

[Problem to be solved by the invention]

However, in the conventional sorting process described above, all records in the file to be sorted are expanded in memory, so if the file consists of a large number of records, a large amount of memory capacity is required. It had the disadvantage of being clumsy. The present invention has been made to overcome these conventional drawbacks, and an object of the present invention is to provide a data processing method that can reduce the amount of memory used in sorting processing.

[Means to solve the problem]

The data relaxation method of the present invention includes a sorting means that divides a record group having a KEY item and a relative link address item, stores only the KEY value and the relative address value in a sort table, and sorts them; a sort work storage means for sequentially storing records of a work file indicated by relative address values of a table in a sort work file; and a minimum KE within each group on the sort work file.
Set each record with Y value in the merge table, compare the records in the merge table, and find the minimum KEY.
and merging means for outputting the records of to a work file.

[Effect]

The sorting means divides a record group having a KEY item and a relative work address item, and stores only the KEY value and relative address value from each record in a sorting table for sorting. Store the records of the work file indicated by the relative address value in the sort cook file sequentially,
The merging means sets the record with the highest /J-KEY value in each group on the sorted work file to the merge table force, compares the records in the merge table, and merges the record with the lowest KEY into the work file. Strengthen.

[Example of ability]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of an embodiment of the present invention,
In this embodiment, there is a sorting means 1-1 that introduces only KEYj; Sort hook storage means 1 stored in sort work file
-2, and a merging means 1-3 for outputting a work file by merging the divided and sorted sorted work files.

Next, the procedure for data processing, that is, sorting processing, to be applied to such a configuration will be explained using the flowchart shown in FIG. >> In Figure 2, step 7'3-1 to 3-5
is the process applied to the sorting means 1-1, step f3
Steps -6 to 3-9 are the processes to be applied to the network storage means 1-2, and steps 7'3-10 to 3-15 are the processes to be applied to the merging means 1-3. Sorting means 1-1
tli, before performing the sorting process, first perform the initial settings for the sorting process. That is, the end SW used for sorting processing is cleared (step 3-1) and the input counter is cleared (step 3-2).

Next, in step 3-3, a sort table is created.

In other words, add @1" to the value of the input counter and read the l record of the work file. If the record can be read, the KEY value and relative work address value of the input record are added to the position indicated by the input counter in the sort table. If the record cannot be read because the file has ended, the end SW is set to "1", "1" is subtracted from the value of the input counter, and the next process is performed.

In step 3-4, conditions for the processing in step 3-3 are checked. If the input kakunta value is "t#" or the end SWo value is "'11", proceed to the next process, but if neither condition is satisfied, step 3
-Repeat process 3. That is, when t records in the work file have been processed or all records in the work file have been read, the next step f3-5 is processed. The sorting process in step f3-5 is performed using jl! The conventional procedure as explained using FIG. 2 is used.

After performing the processing applied to the sorting means 1-1 in this manner, the sort hook storage means 1-2 performs the processing of storing the sort work file. Sort work storage means 1-2
First, the output counter used for sort work file storage processing is cleared (step 3-6). Next, in step f3-7, the value of the output quanta is set to ``
1m is added, and records in the work file that have the same value as the relative work address value indicated by the output counter of the sort table are written into the sort work file. In this way, the process of writing to the sort cook file in step 3-7 is repeated for the boxes where the value of the input column and the value of the output column match (step 3-7).
8). Hereinafter, steps 3-2 to 3-8t are repeated until the end SW value becomes "l#" (step 3
-9). In this way, when the end SWo value becomes ''l#, the merging means 1-3 first sets the initial value for mark processing. That is, it clears the end SW (step 3-10).
). Next, in step f3-11, "" is added to the merge iDXO value.
Add 1', [Merge IDX * m7't - (
rev'L -1)) is stored in the input relative record number. where m is the number of records in the work file. If the value of the input relative record number is IA specified by the sort hook storage means (number of processed records)
If it is so large, rALL''''FF- (< 6) J is stored in the position indicated by merge IDX in the merge table. If the value of the input relative record number is as small as the number of processed records, the record with the minimum KEY value in the group in the sort work file indicated by merge iDx is read and stored in the position indicated by merge iDX in the merge table. Step 7'3-1 like this
The process in step 1 is repeated for each tube in which the value of merge iDX becomes larger than t (step 3-12). When the value of Marso iDX becomes larger than t, step f3
- Proceed to 13, and in step 3-13, the merge iDX in the merge table stores the merge record located at @l# in the output record area, and then the merge iDX stores the merge record located at @l# from "11" to "ml". Compare the merge record KEY value at 'L'1 and the output record area KEY value, and if the output record area KEY value is larger,
Each time, a mark record is stored in the value of the output record area. After the merge iDX repeats the above comparison process from "l" to "rut" 1, the KE in the output record area
If the value of Y is rALL''Fr(14)'', end SW
Set the o value to “1” and set the KEY of the output record area.
If the value of is not r ALL'''FF'' (16) J, write the data stored in the output record area to the work file and mark the next record from the group in the sort work file to which the data belongs. Store in a table. This step 3-13 mark processing is repeated until the end SWO value becomes "1" (step 3-14), and when the processing of step 7" 3-13 is completed, the rearrangement of the work files is completed. The following steps will be explained in detail using Figures 3 to 7. Figure 3 is a diagram showing work file ■. This work file WK consists of m records.
Each record is composed of an a-byte KEY item, a b-byte relative address item, and a C-byte data item.

FIG. 4 shows a case where the work file WK shown in FIG. 3 is divided into groups of t items and stored in a sort table BT on the memory. However, m) shall be IOOXA. When storing in the sort table, as shown in Fig. 5, only the KEY value and relative work address value of each record are used to store (aXbXA)-f, that is, ( ( KE
Y value: a pie + relative work? Dress value: b no 4 i})
Create a 79-byte long sort table BT with XI group records number t. Sort table 8T in memory created in this way
By repeatedly comparing the size of adjacent data, it is sorted in KEY order. Next, records of the work file indicated by the relative work address values are sequentially read from the first position of the rearranged sort table ST, and are sequentially stored in the sort work file river.

This process is repeated rv'L times in the final record box of the work file WK, that is, for m pieces of data, and is stored on the sorted work file ■, as shown in FIG. Next, as shown in FIG. 7, a merge table k M ? in memory is created in which Vt groups of t items stored in the sort work file pool are provided for each group, one record at a time. ((a+b+e) Xm/L/
Minimum m applied to one record in each group (odd)
Stores a record of values. The L records stored on the merge table MT are compared, and the record with the minimum KEY value among them is output to the work file WK. The group targeted for output stores the next record in the corresponding record area of the mark table MTO.

If there is no next record, KEY is r ALL″FF
(,6)"" is stored. The above process is performed as the KEY of all records on the merge table MT.
becomes r ALL''''''(14)' J! and repeat.

As an example, the KEY value a is 7 bytes, the relative address value b is 3 bytes, the data item C is 10 bytes, and the number of records is m.
Assuming that there are 10,000 items and the number of items divided into l groups t is 100 items, the memory capacity required for the conventional method is (a+b+e
) Xm (=1 0 0 0 0 0 0 bytes), whereas in this example, (a+b)
b + e) [Effects of the Invention] As explained above, the present invention is capable of merging each group-f first record of a sort work file created by dividing a file to be sorted and rearranging only KEY values and relative address values. This has the effect that the memory capacity used can be significantly reduced compared to the conventional method.

[Brief explanation of drawings]

FIG. 1 is a block diagram of one embodiment of the present invention, and FIG. 2 is a block diagram of an embodiment of the present invention.
A flowchart showing the flow of the sorting process applied to the configuration shown in the figure, Figure 3 is a diagram showing the record layout of the file to be sorted, Figures 4 and 5 are diagrams for explaining the sort table in memory, Figure 6 7 is a diagram for explaining the processing operation of the network storage means, FIG. 7 is a diagram for explaining the processing operation of the merging means, and FIG. 8 is a flowchart showing the procedure of conventional sorting processing. In FIG. 1, 1-1... sorting means, 1-2...
- Sort work storage means, l-3... Merge means.

Claims (1)

[Claims] A sorting means that divides a record group having a KEY item and a relative work address item, stores only the KEY value and relative address value in a sort table, and sorts them, and a work file record indicated by the relative address value of the sorted sort table. A sort work storage means that stores sequentially in a sort work file, and records with the minimum KEY value in each group on the sort work file are set in a merge table, and the records in the merge table are compared to find the record with the minimum KEY. Output to a work file. A data processing method comprising: merging means.