JPS595372A - File processing system - Google Patents

Abstract

PURPOSE:To extract easily a desired value record group from plural record sets which differ in attribute and size by reading a master and a slave file on the basis of a key file. CONSTITUTION:The pointer area of the key file contains two kinds of pointers, i.e. master and slave pointers and starts at the final block of the file toward the BOE. The master pointers are linked mutually corresponding to the least significant indexes, one to one; each of them controls one master record respectively and also controls a chain of slave records with the same key. Each of the slave pointers control one slave record respectively. The slave pointers with the same key are chained and each chain is linked to the master pointer with the key.

Description

[Detailed description of the invention]

[Technical Field of the Invention] The present invention provides two sequential files and a key file that stores index information of these sequential files,
This relates to a file processing system that accesses the above two sequential files using a key file. [Prior art and problems] This system is used at the terminal level.

[The files used include sequential organization files and index J-organization files. In indexed sequential files, one key is one
However, it is also possible to have one key correspond to multiple records. This is done by chaining a plurality of related records using pointers and associating this chain of records with one key. However, it is very difficult to store the attributes of chained records in a single indexed sequential file. There are 5 requests to obtain 5 files as quickly as possible, but conventional indexed sequential files cannot meet these 5 requests. [Object of the Invention] The present invention is based on the above consideration. It is an object of the present invention to provide a file processing system that makes it possible to easily obtain a desired group of related records from a set of records having different attributes and sizes. [Structure of the Invention] Therefore, the file processing system of the present invention includes a parent file having a sequential file structure having a plurality of parent records, an amm rough file constituent child file having a plurality of child records, and a parent pointer. It has a child pointer that points to a child record and an index section that selects the corresponding parent pointer based on the value of a key which is one data item of the parent record, and which selects child records related to the same parent record. a key file configured such that the child pointers pointing to the parent record form a chain and the parent pointer points to a parent record and points to a corresponding chain of child pointers;・This invention is characterized by being configured such that reading processing from the key file and the child file is performed with reference to the key file. Figure m1 is a conceptual diagram of a chain file in the present invention. Figure 2 is a diagram showing a specific example of a parent record and child record, Figure 3 is a diagram showing the format of a key file in a volume,
Figure 4 shows the format of the key file, Figure 5 shows the logical structure of the index of the key file, Figure 6 shows the format of the index block, and Figure 7 shows the format of the index block. In the related diagrams, Figure 8 shows the format of a pointer, Figure 9 shows the chain format of a parent pointer and child pointer, and Figure 10 shows an outline of an unused area. FIG. 1 is a diagram for explaining a deletion pointer management system, and FIG. 12 is a diagram showing an overview of access. FIG. 1 is a conceptual diagram of a chained file used in the present invention. In Figure 1, KF is a key file, MF is a parent file, and SF is a child file. FLB indicates a file label. The key file KF has a file label FLB, an index, a parent pointer, a child pointer, etc. The parent file MF is a sequential file and has multiple parent records. Each parent record is pointed to by a parent pointer. The child file SF is also a sequential file and has multiple child records. are doing. Child records are pointed to by child pointers. FIG. 2 is a diagram showing a specific example of a parent record and a child record. In this example, the parent record is 1wA customer specific information. The child records are various maintenance records, and the key is the customer number. Note that the arrow indicates a chain. Figure 3 shows the format of the key file KF within the volume. The file area is defined by the file label F.
LB, key index area. It consists of a pointer area and an unused area. The file label FLB occupies the first 256 bytes,
The key index area manages file control information and key indexes. Contains several levels of indexes for linking parent pointers. The size of one index block is 512 bytes. The pointer area is the parent pointer that points to the parent record. A child pointer pointing to a child record is stored. The unused area is reserved for adding a key index or pointer, and this area is dynamically managed using file labels.
The file format of the file is shown, and FIG. 5 shows the logical structure of the index. FIG. 6 is a diagram showing the format of a key index block. The key index area is composed of several levels of index blocks. The ID is an identifier indicating the type and level of the index. - An identifier of Xmalo and 0ma indicates that the index is the lowest level index. Indicates that there is an l-to-l correspondence with the parent pointer.
The Xmalo0ma identifier indicates that the index is not linked to the lowest level index, but to a lower level index block. FP is a front pointer, which stores the relative record address of the index block with the next largest key value among the index blocks at the same level. BP is a back pointer, which stores the relative record address of the index block with the next smallest key value within the index blocks at the same level. nn is the number of entries used. That is, it shows the number of entries currently registered in this index block. The index block has multiple entries, and each entry pointing to a parent pointer consists of a key part and an address part. The key of the data record in the index block or the largest key value in the index block is stored. In the address part of the entry. The address of the linked index block or the address of the parent pointer is stored. Address information held by an entry has a different structure depending on the type of index. If ID=Xma00ma, it has the storage location of the parent pointer to be linked, and therefore has the following configuration. X1nnnnLLmannrLnn indicates a block address, and 11 indicates an address within the block. Furthermore, in the case of an index block with ID=XmaOlma, it has the storage location of the index block to be linked, and thus has the following configuration. X1nnnn001 nnnrb indicates the block address 0 An example of the correlation between the index blocks tffy described above is 1
Shown in Figure 1E7. This example shows the case where the index level is 4. The pointer area consists of two types of pointers, parent pointers and child pointers, and B
Created towards OE (Bgysnzny Of Extgnt). The parent pointer is linked to the lowest level index in an l-to-l correspondence, and each manages one parent record. Also manages chains of child records with the same key. Figure 8 shows the structure of the parent pointer.Address 1 shows the record address of the parent record in the linked parent file, and address 2 shows the record address of the chain of linked child pointers. The pointer address of the first child pointer is shown, and address 3 shows the pointer address of the child pointer at the edge of the chain among the group of linked child pointers. Note that when there is no chain with child pointers, addresses 2 and 3 are
It is said to be a ma. The secondary bonnet also has a configuration as shown in FIG. Each child pointer manages one child record. Further, child pointers having the same key form a chain, and each chain is linked to a parent pointer having the same key. Address 1 points to the record address of the child record in the linked child file,
Address 2 indicates the pointer address of the next linked child pointer in the child pointer chain. However, at the final pointer of the chain,
It is set to o. Address 3 indicates the point address of the child pointer linked immediately before in the chain of child pointers. However, the start pointer of the chain is set to X00000. 9th
The figure shows a chain format of parent pointers and child pointers. FIG. 10 is a diagram showing an outline of the unused area. The key index area is from BOE to EO'FJ (E+
bd Qf gxtgnt), and the pointer area is extended from EOE to BOE. The unused area is located between the key interaction area and the pointer area, and is a reserved area for adding an index block or a pointer. This area is. Unused space changes dynamically as index blocks or pointers are added. It is managed by the unused first block address and unused first pointer address in the file label. FIG. 11 shows a deletion pointer management format. When a parent record is deleted, the corresponding parent pointer is connected to the chain for deleted parent pointers by addresses 2 and 3 in that pointer, and when a child record is deleted (the parent record is deleted) (If the child record linked to it is also deleted), connect the corresponding child pointer to the chain for deleted child pointers using addresses 2 and 3 in the pointer, like this: , parent pointer, and child pointer deletion pointers are managed by separate chains. Therefore, a pointer that once became a parent pointer in the pointer area. When used after deletion, it is always the parent pointer; the same applies to child pointers. The deletion parent pointer chain is the deletion parent pointer start address of the file label,
A34 and the deletion parent pointer end address F'35. Additionally, the delete child pointer is
It is managed by the deletion child pointer start address 36 and the deletion child pointer end address F" 37 of the label. Once the address l in the pointer is registered, it cannot be updated regardless of whether it is used or unused (deleted). In other words, managing the deletion pointer means managing the deletion/reuse of records in the parent file or child file. Figure 12 shows an overview of access. Parent record To read a child record, first read the index and parent pointer, then read the parent record based on the parent pointer.To read a child record, first read the child pointer, then read the child record based on the child pointer. The following similar processing is repeated. [Effects of the Invention] As is clear from the above explanation, according to the present invention, (a) A normal sequential file may be used as the data file. Unlike conventional indexed sequential files, there is no need to change the storage format such as providing an index section in the data file or adding pointer information to the data record. It is also possible to access it as a simple sequential file by accessing it using the following access method. ←) With one key file, the parent file and child files can be accessed in a chain. fj key・Because the key file has all the information for access, it is possible to access the same parent file and child file using multiple key files.・

[Brief explanation of the drawing]

FIG. 1 is a conceptual diagram of a chain file in the present invention. Figure 2 is a diagram showing a specific example of a parent record and child record, Figure 3 is a diagram showing the format of a key file in a volume,
Figure 4 shows the format of the key file. Figure 1M5 shows the logical structure of the key file index. Figure 6 shows the format of the index block. Figure 7 shows the relationship between the index blocks. , FIG. 8 is a diagram showing the format of a pointer, FIG. 9 is a diagram showing a chain format of a parent pointer and child pointer, and FIG. 10 is a diagram showing an outline of an unused area. Figure 11 is a diagram for explaining the deletion pointer management method, and Figure 12 is a diagram showing an overview of access.KF...Key file, MP...Parent file, S
F...child file, FLB...file label
Patent Applicant Fujitsu Ltd. Representative Patent Attorney Kyotani Yobu Ya 1 Zu Sai 2 Zu Ken 8 Zu Ke'? Diagram 10 Diagram 12

Claims (1)

[Scope of Claims] A parent file having a sequential file structure having a plurality of parent records, a child file having a sequential file structure having a plurality of child records, a parent pointer, a child pointer pointing to a child record, and a child pointer pointing to a child record, and a child file having a sequential file structure having a plurality of child records. Child pointers that have an index section that selects a corresponding parent pointer based on the value of a key, which is one data item, and that point to child records related to the same parent record form a chain, and the parent pointer a key file configured to point to a parent record and point to a corresponding chain of child pointers. A file processing system that uses the above key file to read parent records and child records from the above parent file and child files.