JPH04138565A - Connection processing system of plural files - Google Patents

Abstract

PURPOSE:To obtain a system capable of executing rapid processing and simplifying virtual file definition by providing this connection processing system of plural files with a connection retrieving means for referring to only a connection index and a connection index record whose number of pointers is equal to the total number of reference files constituting virtual files and executing connection retrieving processing. CONSTITUTION:A data base 1 stores a directory 2, a connection index record file 6, an index record file 7, and a file record file 8. A reference file definition information 3, connection index definition information 4 and virtual file definition file 5 are registered in the directory 2 and the connection retrieving means 9 receiving a user's request, accesses the data base 1. The means 9 refers to only the connection index record having the number of pointers equal to the total number of reference files constituting the virtual files to execute connection retrieving processing. Consequently, processing is extremely rapidly executed, the virtual file is defined only by one definition, operation is simplified, and resource using efficiency is improved.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a multiple file combination processing method in a database system of an electronic computer system.

[Conventional technology]

In a relational database system, when obtaining necessary information from multiple existing basic files, instead of creating a new basic file by combining multiple basic files, a virtual file is defined from the multiple basic files. However, this virtual file is often accessed in order to maintain consistency in data content between basic files and to prevent duplicate storage of data with the same content.

By the way, conventionally, the definition of a virtual file included the following items.

■Determining the basic file ■Determining the format ■Determining the access route ■Determining the relationship index ■Determining the format association Here, the format is the selection of necessary items from the basic file, taking into account the order, and the access route. is the order from which format to access a virtual file, a relational index is an index used for processing to combine multiple basic files, and a format association is an index used to combine multiple basic files, etc. This is a mode of synthesizing. Note that joining refers to sequentially extracting only file records that have the same index value (key value) in a predetermined item in the format of multiple basic files, and chaining refers to sequentially extracting only file records that have the same index value (key value) in the format of multiple basic files, and chaining means that the access route is the format of the first basic file. merging refers to the successive extraction of file records that have consecutive access routes and the same index value in a predetermined item of the format. This refers to extracting all file records in the order of index values of predetermined items. However, in the present invention, only merging is targeted.

FIG. 8 shows the configuration of a multiple file combination processing method in which a combination search is performed for virtual files defined from a plurality of basic files.

In FIG. 8, the database 11 includes directory 1.
2, an index record file 15, and a file record file 16 are stored, and the directory 12 stores basic file definition information 13 and virtual file definition information 14.
are registered, and the combined search means 17 accesses the database 11 upon receiving a request from a user.

FIG. 9 shows an example of the basic file definition information 13 and virtual file definition information 14 in the directory 12 in FIG. That is, as the basic file definition information 13, item A1. A2, A3. Definition information of basic file A consisting of A4, items Bl, B2. B3
Definition information of basic file B consisting of and item C1
, C2゜C3, C4, the definition information of basic file C, information to set index a for item A1 of basic file A, and item B of basic file B.
1, and information that index C is to be set for item C1 of basic file C. In addition, as virtual file definition information 14, the definition information of formats A, B, and C in which all items of basic files A, B, and C are selected, and formats (basic files) A, B, and C are combined as a format association. (J
), and the information (VABC-J”) that the information is to be synthesized by
The information includes information indicating that the access route is format A - format A - format C - format C.

FIG. 10 shows an example of the index record file 15 in FIG. 8. That is, index a has index records a1 to a6, index b has index records bl to b4, and index C has index records 01 to c5. Note that each index record has an index value and a pointer to the corresponding file record. Note that the index entry name is also registered in a catalog file (not shown).

Figure 11 shows file record file 1 in Figure 8.
This shows an example of No. 6. That is, basic file A
has file records A-1 to 8-6, basic file B has file records 81 to B-6, and basic file C has file records C-1 to C-6. Note that the file record of each basic file includes items corresponding to the basic file definition information 13 in the directory 12.

FIG. 12 shows the processing of the combination search means 17 in FIG. 8. In addition, in FIG. 12, H, K,
F, I, and G indicate predetermined processing. Also,
z, y, s are index records held during processing, z+, , + are index values of index records z, y, 0 is the current index record,
0゛ is the index value, n is the number of related indexes, X is the access route rank given in the process, m is the access route rank of the index record to be obtained,
p is the access route rank of the current index.

Hereinafter, the operation of the above conventional example will be explained in different cases.

(1) Processing to find the first record This processing is performed when the index value is the minimum and the access route rank is "1" belonging to the virtual file defined by the join.
This is the process of acquiring the file record of.

In this process, the current index record 0 is nu
ll, and processing H etc. targeting the current index record 0 is not executed.

Furthermore, the contents of processes F and G are as follows.

F: Process of finding the index record with the minimum index value from the index of the designated access route rank (X) G; Process of finding the index record that is equal to or greater than index value 2' from the index of the designated access route rank (X) ; m=n+1 First, through the processing from steps 81 to S15, the index record with the minimum index value is extracted from the index with the access route order of "1" and is held as the index record S. Note that if there is no corresponding index record, it is assumed that the corresponding file record does not exist, and the process ends via step S25.

Next, in the processing of steps 316 to S23, it is checked according to the access route order whether all indexes include an index record having the same index value as the index record already retrieved in step 311, and if this is not satisfied, the process is performed again. The acquisition and comparison of index records is repeated starting from the access route rank "1". Note that if the corresponding index record no longer exists, it is assumed that the corresponding file record does not exist, and the process ends via step S25.

With the above processing, among the index records that exist with the same index value in each index, the one with the smallest index value and the access route rank of "1"
The index record of is retrieved as S, and the file record it points to is referenced in step S24.

In the example of indexes shown in FIG. 10, index values r5J and #3J exist in common among indexes a, b, and c, and among these, index record a3 is the smallest index record. File record A-3 that is retrieved and points to
is referenced.

(2) Process for obtaining the final record This process is a process for obtaining the file record with the maximum index value and the last access route ranking that belongs to the virtual file defined by the join.

In this process, the current index record 0 is nu
ll, and processing H etc. targeting the current index record 0 is not executed.

Furthermore, the contents of processes F and G are as follows.

Process G for finding an index record with a large Fl index value from an index with a specified access route rank (X) ;m=O The operation is almost the same as the process for finding the first record described above, and the only difference is the setting of the access route rank m and the magnitude relationship of the index values.

In the index example shown in FIG. 10, index record c5 is the largest index record.
is retrieved, and among the file records pointed to, the later file record C-6 with the larger pointer value is retrieved.
is referenced.

(3) Process for obtaining the next record after the current record This process is a process for obtaining the file record located next to the record given as the current record from the file records belonging to the virtual file defined by the join.

Processing H, F, 1. G contains the following content:

H; Integer of the access route rank specified by the index record that is equal to the index value 0' and greater than the current pointer value.

Process F to obtain an index record larger than the index value 2° from the index of the specified access route rank ■; Process G to obtain the index record equal to the index value 2° from the index of the specified access route rank Processing 1 to obtain an index record that is equal to or larger than index value 2 from the index of the specified access route order 1<;m=p+1 First, by the processing of steps 31 to S3, a duplicate record is added to the current index record 0. If there is
It is taken out. For example, if index record b2 pointing to file record B-3 in FIG. 10 is the current index record, the same index record b2 pointing to file record B-4 is retrieved by process H.

If there is no duplicate record, the index record with the larger index value is retrieved for the index with the access route rank "1" through the processing of steps 84 to S15, and if the access route ranks m and x match, the index record is extracted. It is held as record S. In addition,
If the access route ranking of the current record is the highest, processing F is performed, otherwise processing I is performed.

This is because the search has already been performed up to the top level and there are no duplicate records, so there is no need to search for the same index value again.

Note that if there is no corresponding index record, it is assumed that the corresponding file record does not exist, and the process ends via step 325.

Next, through the processes of steps 316 to 323, it is checked according to the access route order whether all indexes include an index record having the same index value as the index record already retrieved in steps Sll and S12. If the condition is not satisfied, the acquisition and comparison of index records is repeated again starting from the access route rank "1". Note that if the corresponding index record no longer exists, it is assumed that the corresponding file record does not exist, and the process ends via step S25.

Through the above processing, the index record located next to the current index record is retrieved from among the index records that exist in each index with the same index value, and the file record pointed to by the index record is retrieved.

In the example of the index shown in FIG. 10, if the file records are read out in order starting from the first file record, the order will be as follows.

A-3→ B-3-B-4→ C-4→ A-6→B-6→ C-5→ C-6 (4) Process to obtain the record immediately before the current record This process is defined by joining This is a process of acquiring the file record located immediately before the record given as the current record from among the file records belonging to the virtual file given as the current record.

Processing H, F, 1. G contains the following content:

H; Process of obtaining an index record that is equal to the index value O° and smaller than the current pointer value from the index of the specified access route order F; Obtaining an index record that is smaller than the index value 2'' of the specified access route order Process I for obtaining an index record from an index; Process G for obtaining an index record equal to index value 2' from an index with a specified access route rank; Process G that obtains an index record equal to or smaller than index value 2' from an index with a specified access route rank m=p-1 The operation is almost the same as the process of finding the next record after the current record described above, and the only difference is the setting of the access route rank m and the magnitude relationship of the index values.

[Problem to be solved by the invention]

Although the conventional multiple file combination processing method was configured and operated as described above, it had the following drawbacks.

■To obtain file records that match the join conditions,
It is necessary to repeatedly match the index values of index records, and input/output to the index occurs at least as many times as the number of related indexes n, and at the maximum, the number of index records multiplied by the number of related indexes n. It requires input/output and takes a very long time to process.

■As the definition of a relational index is included in the definition of a virtual file in a directory, when defining multiple virtual files, even if the format of the basic files composing each virtual file is the same, the relational index may be different. Virtual files must be defined separately, which complicates preparation work, maintenance, and management, and the storage of duplicate information reduces database usage efficiency.

■Join while defining a virtual file in a directory,
It includes format relationships such as chaining and merging, and is statically defined as an executable virtual file, so even if the formats of the basic files that make up a virtual file are the same, if the format relationships are different, the virtual files will be created separately. must be defined and lacks flexibility. That is, chain merging is a process that is used in combination with join, and although it is desirable to be able to dynamically change the modal association, this cannot be done at the execution stage.

The present invention has been proposed in view of the above points, and its purpose is to provide a multiple file combination processing method that can perform high-speed processing and simplify virtual file definition. be.

[Means to solve the problem]

In order to achieve the above object, the present invention includes a virtual file that does not include format relationships and only defines the format of basic files as constituent elements, and a virtual file that is defined independently, with an index value, the number of pointers, an ID, and an index. Performs join search processing by referring only to a concatenated index consisting of multiple concatenated index records including pointers to records, and concatenated index records in which the number of pointers is equal to the total number of basic files constituting the virtual file. A combined search means is also provided.

[Effect]

In the multiple file combination processing method of the present invention, the combination search means performs the combination search process by referring only to the connected index records in which the number of pointers is equal to the total number of basic files constituting the virtual file.

〔Example〕

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of the multiple file combination processing method of the present invention.

In FIG. 1, a database 1 stores a directory 2, a linked index record file 6, an index record file 7, and a file record file 8, and a directory 2 stores basic file definition information 3, linked index definition information 4, and virtual files. Definition information 5
are registered, and the combined search means 9 accesses the database 1 upon receiving a request from a user.

FIG. 2 shows examples of basic file definition information 3, concatenated index definition information 4, and virtual file definition information 50 in directory 2 in FIG. That is, as basic file definition information 3, item A1. A2. A3.
Definition information of basic file A consisting of A4 and item B1. Definition information of basic file B consisting of B2B5, definition information of basic file C consisting of items CI, C2, C3, and C4, information to set index a for item A1 of basic file A, Information indicating that index b is to be set for item B1 of basic file B, and information that index C is to be set for item C1 of basic file C are included. In addition, as consolidated index definition information 4, indexes a, b
, c is selected as the relational index, and a
Information indicating that the access route rankings are set in the order of b and c is included. Furthermore, as virtual file definition information 5, definition information of formats A, B, and C in which all items of basic files A, B, and C are selected, and the virtual file is composed of formats (basic files) A, B, and C. Information that it will be done (V
ABC) are included. In this way, the virtual file definition also includes information about what should be used as a relational index.
It also does not include format-related information.

FIG. 3 shows an example of the linked index record file 6 in FIG. 1, and FIG. 4 shows the logical structure of the linked index records. That is, in the linked index, one linked index record is provided for each index value of the relational index defined in the directory, and the linked index record includes the index value, the number of pointers indicating the number of index records having the index value, and the linked index record. , a TD indicating the access route ranking, and one or more pointer sections having pointers to the corresponding index records. In the example of FIG. 3, the concatenated index record file 6 includes the concatenated index record 21
~ Contains z8. Note that the linked index record file 6 is updated when a new file record is added to the file record file 8, an existing file record is deleted, or the contents are updated.
Concatenated index records are added, deleted, and updated accordingly. This is similar to the index record file 7.

FIG. 5 shows an example of the index record file 7 in FIG. 1, and the contents are the same as in FIG. 1θ exemplified in the conventional example.

FIG. 6 shows an example of the file record file 8 in FIG.
It is similar to the figure.

FIG. 7 shows the processing of the combination search means 9 in FIG. 1. In addition, in Fig. 7, HK, G,
F indicates predetermined processing. In addition, 2 is a concatenated index record held during the process, 2' is its index value, S is a pointer to the index record held during the process, 0 is the current index record, 0 is its index value, and n is The number of related indexes, m is the access route ranking of the record to be acquired, and p is the access route ranking of the current index.

Hereinafter, the operation of the above embodiment will be explained in different cases.

(1) Process for finding the first record This process is performed when the formats of multiple basic files constituting a virtual file are combined under the condition of merging, and among the corresponding file records, the index value is the smallest and the access route rank is "1". ” is the process of acquiring the item.

In this process, the current index record 0 is nu
ll, and processing H etc. targeting the current index record 0 is not executed.

Furthermore, the contents of processes F, G, and K are as follows.

F; Process for finding the concatenated index record with the minimum index value from the concatenated index G; Index value 2
In the process of finding the concatenated index record with the next largest index value after ' from the concatenated index; m=1 First, since the current index record 0 is null, step SL is not executed, and since S is null, step S3 to move to.

In step S3, m is set to 1, and in step S4. S
Based on the determination in step S9, the process moves to step S9.

In step S9, a process F is performed to obtain a concatenated index record with the minimum index value from the concatenated index. Here, if there is no corresponding linked index record, it is assumed that the corresponding file record does not exist, and the process ends through steps S10 and S15.

If it is not null, it is determined in step Sll whether the value of the number of pointers of the retrieved linked index record Z is equal to the number of relational indexes n, and if not, the linked index record with the next largest index value after the index value 2° is determined. A process G for determining ? from the concatenated index is performed in step S12, and steps SIO and subsequent steps are repeated.

In step Sll, if the value of the number of pointers in the retrieved linked index record 2 is equal to the number of related indexes n, in step 313 the pointer in the pointer section whose ID of the linked index record 2 is the access route rank m is set to S. Then, in step 514, the file record pointed to by S is found, and the process ends.

In the example of the linked index record file 6 in FIG. 3, the linked index records are taken out sequentially starting from 21, and in the current example, the number of relational indexes is "3", so the linked index record with the number of pointers is "3". Record z5 is finally selected, and the index record a indicated by the pointer in the pointer section whose ID is rlJ
3, file record A-3 is obtained.

(2) Processing to obtain the final record This processing is performed when the formats of multiple basic files constituting a virtual file are combined with the condition of merging, and among the corresponding file records, the index value is the highest and the access route ranking is the last. This is the process of acquiring the .

In this process, the current index record 0 is nu
ll, and processing H etc. targeting the current index record O is not executed.

Furthermore, the contents of processes F and G are as follows.

F; Process for determining the concatenated index record with the maximum index value from the concatenated index G; Index value 2
In the process of finding the concatenated index record with the next smallest index value after ' from the concatenated index: m=n The operation is almost the same as the process of finding the first record described above, and the only difference is the magnitude relationship of the index values.

In the example of the concatenated index record file 6 in Figure 3, the concatenated index record z8 is retrieved first, and its pointer count matches the number of related indexes "3", so it is selected as is, and its ID is "3". Index record C indicated by a pointer in a certain pointer section
5, file record C-6 is obtained.

(3) Processing to obtain the next record of the current record This processing is performed by selecting the record given as the current record among the corresponding file records when the formats of multiple basic files that make up a virtual file are combined with the condition of merging. This is the process of acquiring the file record located next to .

Processes H, F, and G have the following contents.

H: Processing for obtaining a pointer larger than the current pointer value for the current index record ols F; Processing for obtaining a linked index record with an index value equal to index value 2゛ from the linked index G: The next largest index after index value 2゛Process K for obtaining a linked index record of a value from a linked index; m=p+1 First, in step S1, processing H for obtaining a pointer value larger than the current pointer value is performed for the current index record 0. This process is performed not on the linked index but on the index record of the index record file 7. If there is a corresponding index record, the value of the pointer is held as S, the file record pointed to by S is found in step 314, and the process ends.

If there is no duplicate record, in step S3, the access route rank m is set to the current access route rank p to r.
lJ is added, and if m becomes larger than n in step 5486, the process moves to step S5; otherwise, the process moves to step S9.

Note that in the current processing, m will never become smaller than "1".

If m is larger than n, set "1" to m in step S5.
'' is set, and processing G is executed to find a concatenated index record with the next largest index value after the index value 2'' of the current record from the concatenated index.

In other cases, in step S9, a process F is executed in which a linked index record having an index value equal to the index value 2' is obtained from the linked index.

This means that the fact that the result of adding "1" has increased the number of related indexes n means that processing has been completed up to the last access route rank for the same index value, so the current index record This is to omit the processing for the same index value.

Here, if there is no corresponding linked index record, it is assumed that the corresponding file record does not exist.
The process ends after steps SIO and S15.

If it is not null, it is determined in step Sll whether the value of the number of pointers of the retrieved linked index record 2 is equal to the number of relational indexes n, and if not, the linked index record with the next largest index value after index value 2 is determined. A process G for determining the value from the concatenated index is performed in step S12, and steps 310 and subsequent steps are repeated.

In step Sll, if the value of the number of pointers in the retrieved linked index record 2 is equal to the number of relational indexes n, in step 313, the pointer in the pointer section where rD of the linked index record 2 is the access route rank m is set to S. Then, in step 314, the file record pointed to by S is found, and the process ends.

In the examples of FIGS. 3 and 5, if the file records are read out in order starting from the first file record, the order will be as follows.

A-3-B-3 → B-4 → C-4 → A-6 → B-6 → C-5-C-6 (4) Processing to obtain the record immediately before the current record This process is performed using a virtual file. This process obtains the file record located immediately before the record given as the current record among the corresponding file records when the formats of the multiple basic files constituting the file are combined under the condition of merging.

Processes H, F, and G have the following contents.

H: Processing to obtain a pointer smaller than the current pointer value for the current index record 0 F; Processing to obtain a linked index record from the linked index with an index value equal to or smaller than index value 2° G: smaller than index value 2° The process of finding the linked index record of the index value from the linked index; m=p-1 The operation is almost the same as the process of finding the next record after the current record described above, and the setting of the access route rank m and the size of the index value The only difference is the relationship.

(5) Process to find a record equal to the specified index value In this process, the current index record 0 is nu
ll, and processing H etc. targeting the current index record 0 is not executed.

Further, processing F has the following contents.

F; In the process of finding a concatenated index record with an index value equal to the specified index value from the concatenated index; m-1 operation, steps Sl and 32 are not executed, and in step S3 rlJ is added to the access route rank m. is set,
Based on the determination in steps 34 and 36, the process moves to step S9, and processing F is executed to obtain a linked index record having an index value equal to the specified index value from the linked index. If there is no corresponding record, step S10. The process ends after S15. Furthermore, even if there is a corresponding record, if the number of pointers is not equal to the number of relational indexes n in step 311, processing G is not defined, and the processing ends via step 810.515.

(6) Process for obtaining records for each format This process is a process for obtaining file records of combined virtual files for each format.

In each process described above, as the access route rank m,
Actions are performed by providing the value of the ID of the specified format.

For example, in FIGS. 3 and 5, when format B is specified and all applicable records are retrieved, index record b2 with ID "2" is extracted from linked index record 25.28. b4 is obtained, and file records B-3, B-4, and B-6 are obtained by tracing them.

〔Effect of the invention〕

As explained above, in the multiple file merging processing method of the present invention, ■ To obtain file records that meet the merging conditions,
Since only those indexes whose pointer count is equal to the total number of basic files constituting the virtual file need to be extracted from the concatenated indexes, processing can be performed at extremely high speed.

■As the definition of a relational index is not included in the definition of virtual files in a directory, when defining multiple virtual files, if the format of the basic files that make up each virtual file is the same, the definition of the virtual files will be the same. It is sufficient to create the necessary linked indexes separately, which simplifies the work and improves the efficiency of resource use.

■Join while defining a virtual file in a directory,
Since format relationships such as chaining and merging are not included, if the format of the basic files that make up the virtual file and the relationship index are the same, processes such as chain merging can be performed using the same linked index, providing flexibility. increases significantly.

There are other effects.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the multiple file combination processing method of the present invention, FIG. 2 is a diagram showing an example of definition information in the directory in FIG. 1, and FIG. 3 is a concatenation index in FIG. 1. Figure 4 is a logical configuration diagram of a linked index record; Figure 5 is an example of the index record file in Figure 1; Figure 6 is a diagram of the file record file in Figure 1. Figure 7 is a flowchart showing the processing of the combined search means in Figure 1; Figure 8 is a configuration diagram showing a conventional example; Figure 9 is an example of definition information in the directory in Figure 8. Figure 10 is a diagram showing an example of the index record file in Figure 8, Figure 11 is a diagram showing an example of the file record file in Figure 8, and Figure 12 is the processing of the combined search means in Figure 8. It is a flowchart which shows. In the diagram, 1... Database 2... Directory 3... Basic file definition information 4... Consolidated index definition information 5...
・Virtual file definition information 6...Concatenated index record file 7・
...Index record file 8...
・File record file 9... Combined search means

Claims (2)

[Claims] (1) A virtual file that does not include format relationships and only defines the format of the basic file that is a component, and multiple files that are defined independently of the virtual file and that include an index value, number of pointers, ID, and a pointer to an index record. a concatenated index composed of concatenated index records of A multiple file combination processing method featuring the following. (2) The multiple file combination processing method according to claim 1, wherein the combination search means searches for each format by extracting only a pointer corresponding to a designated ID from the referenced linked index record.