JPH03154941A - Data base operation processing system - Google Patents

Abstract

PURPOSE:To compare the data having different types with each other and to register these data by adding an area to a definition information holding part to hold the data storing attribute and providing a step to a data base operation control part to update the corresponding area added to the definition information holding part. CONSTITUTION:A data base DB definition control part 40 is called from a request control part 30 when a request is received for the definition of the DB or its change. Then the part 40 updates a DB 60 and a definition informa tion holding part 70 based on the contents of a request. A DB operation control part 50 is called from the part 30 at acceptance of the requests for operations (addition, change, deletion) given to the data stored in the DB 60. Then the part 50 analyzes the requests of users based on the definition information held by the part 70 and operates the data stored in the DB 60. Then, at the part 70, the information on the data types 83 showing the column attributes, the sorting attributes 84 showing the meaning types, and the data unit 85 are addi tionally held in a column definition information control table 80. Thus, the comparison is attained among the data of different types and these data can be registered.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a database operation processing method, and more specifically, to a database management system or a database access system located between a database management system and a user (hereinafter referred to as rD
This relates to a processing method for operation requests (referred to as BJ).

[Conventional technology]

Conventional database management systems (hereinafter referred to as rDB-MSJ) require the definition of type attributes for data held in a DB, and perform comparisons between data and data registration operations based on the defined type attributes. was checking the validity of For example, IS 09075:
Information Processing S
systems D database L anguage
A relational database language (SQL) standardized by the International Organization for Standardization (ISO), as shown in SQL.
In the table statement that defines the DB, data types such as integer type, real number type, 9-character type, etc. are specified in the column definitions that make up the table to be defined.

Figure 4 shows the employee table (emp) and the department table (
An example of the definition of ``dept'' is shown below. 120 is an example of the definition of an employee table, in which the employee number (empno), employee name (name), department number (dno), salary (sa)
It consists of the information of l). Further, 121 is an example of a definition of a department table, which includes a department number (dno) and a department name (name).

Administrator employee number (mgrno) + budget amount (bu
dget). In addition, int, eger
indicates that the data type is an integer, and char(n
) indicates that the data type is character data with a maximum of 0 digits.

When manipulating tuples in a table, the columns to be compared and registered, and the data types between columns and constants, are compared, and only manipulations between character data and numeric data are permitted. In this case, the DBMS only uses data type information to compare data and check the validity of registration operations. Therefore, a comparison between the employee number and the manager's employee number results in an error because it is a comparison between numerical data and character data. On the other hand, even if the comparison between the employee number and the department number is semantically incorrect, the DBMS does not judge it as an error.

[Problem to be solved by the invention]

With the spread and expansion of networks, there is a movement to integrate DB systems that have been developed individually. There are three possible forms of DB system integration as shown in FIG. In the figure, DBx means each individually developed DB, and APx is
It means a user program.

FIG. 3(a) shows a plurality of D B (601, 602, .
...) are integrated by one DBMS 701. In addition, figure (b) shows the DB, which is the core of each DB system.
When configuring an integrated system by communicating between MSs (702, 703,...), the same figure (C)
I) A distributed DBMS (704, 706, etc.) that controls communication and coordination between BMs (705, 707,...
,...) to achieve integration. The selection criteria for these configurations depends on the functions provided by the DBMS, the heterogeneity of the DBMS, the characteristics of the application business, the mode of operation, and the like.

In the present invention, a DBMS or a group of DBMSs located between a user program and a DB are collectively referred to as a DB access system.

When constructing an integrated system as described above, handling of already created DBs becomes a problem. In other words, each DB
A large amount of data is stored in each. Here, if data is managed consistently between these systems, it is easy to integrate them. However, in general, DBs are not designed with integration in mind, so there are problems such as duplication of data, unconventional column names and column data types, and the reality is that integration is not easy. It is. Among these problems, the present invention aims to solve the problem of non-standardization of data types and data units.

As an example, consider the table configuration shown in FIG. Here, consider a query that asks, ``Obtain the name of an employee who receives a higher salary than the manager of the department he belongs to, and the name of that manager.'' When this arrangement is described in SQL, it becomes as shown in FIG. 5, 122. Since the administrator's employee number is defined as character data, comparing it with the employee number in the employee table will result in an error.

Information such as the employee number mentioned above can be found in the personnel database or payroll database.
etc. are stored in various DBs. Depending on the DB, some information may be stored as a character code, while others may be stored as a numerical code. In addition, regarding time and monetary data, there is a problem of unstandardized units. Under these circumstances, in order to use the conventional DBMS as is, you must either re-edit the DB to convert the data type, or change the DBMS.
There is also a way to deal with this problem on the application program side that uses it. However, this is not a practical solution because it requires a lot of effort and 9 hours of expense.

The present invention has been made in view of the above circumstances, and its purpose is to solve the above-mentioned problems in the conventional technology, and to solve the problem of unstandardized data types and data units by re-editing the DB. The object of the present invention is to provide a DB operation processing method that makes it possible to solve the problem without having to do so.

Another object of the present invention is to solve the problem that all comparisons and registrations of character data and numerical data are permitted. For example, in the example shown in FIG. 4, the employee's name and the department's name have the same data type, although the number of digits is different, and the comparison between them is possible even though it is semantically incorrect. As mentioned above, in separately developed DBs, even if the same column name is used, the meaning may be different. If the user does not understand the contents of the DB well, the user may request the wrong operation and may not be able to obtain the intended result.

In the present invention, a semantic type attribute (hereinafter referred to as
By providing this with a "classification attribute"), the above-mentioned problem is solved.

[Means for Solving the Problems] The above-mentioned object of the present invention is to include a DB and a definition information holding unit that holds definition information of data to be stored in the DB, accept requests from user programs, and process requests. a request control unit that distributes processing according to the content; a DB definition management unit that updates the DB and definition information storage unit according to the request in the case of a DB definition or definition change; and a DB definition management unit that handles operations (additions, changes, In a DB access system having a DB operation management unit that operates the data in the DB according to the request in the case of a request (deletion), when comparing data or registering data in the definition information holding unit, In order to judge the validity, add an area to hold the classification attributes of the data in step 2-, and receive the definition of the classification attributes of the data and its changes in the DB operation management section, and add the area to the corresponding area in the definition information storage section. D, characterized in that it includes a step of updating the
In the B operation processing method or the above-mentioned DB access system, an area for storing data units and conversion formulas between units is added to the definition information storage section, and an area for storing the data units and the conversion formula between the units is added to the DB definition management section. This is achieved by a DB operation processing method characterized by the step of accepting definitions of units and conversion formulas between units and changes thereto, and updating the corresponding area of the definition information holding unit in accordance with the request.

[Operation] In the DB operation processing method according to the present invention, a terminal user who uses a DB access system defines classification attributes and units when defining a DB. At this time, the same classification attribute is assigned to columns that can be compared. For example, the fourth
The definition of the table in the figure is as shown in FIG. That is, after the column definition, the classification attribute 132. Unit 133
Describe.

In order to enable comparison between the employee number in the employee table and the employee number of the manager in the department table, the same classification attribute (empcode) is set.

Furthermore, when setting classification attributes and units for an existing table, the definitions are changed as shown in FIG.

In FIG. 7, 140 to 143 indicate an example in which classification attributes are set for employee number 1 name 1 department number and salary in the employee table, respectively. Note that in 143, the unit is also set.

The units and their conversion formulas are defined, for example, as shown in FIG. +50 defines yen as a unit. Here, since the basic unit is the yen, no conversion formula is specified. On the other hand, 151.152 defines 1,000 yen (ki 1 oyen) and 10,000 yen (manyen). Define the unit before specifying the unit name in the column definition.

As will be described in detail later, the DB definition management section of the DB access system updates the column definition information management table and the unit information management table in accordance with the request when receiving the above-mentioned requested amount. In the case of table definition, add the definition information of the columns that make up the table to be defined to the column definition information management table.

In the case of changing the column definition, the definition information of the corresponding column in the column definition information management table is searched, and the specified definition contents are changed. In the case of unit definition, the unit definition information defined in the unit information management table is added.

When requesting a DB operation, the user issues the request in the same manner as before. At this time, if the columns have matching classification attributes,
Comparison conditions can be specified even if the data types are different.

For example, suppose that an inquiry as shown in FIG. 5 is issued. The DB operation management unit of the DB access system analyzes and executes the user's request. At this time, a comparative validity check of the condition part is performed based on the classification attribute and unit. During analysis, comparisons between columns with different classification attributes will result in an error.

In the case of the above example, the comparison between the employee number and the manager's employee number is allowed because they have the same classification attribute and the unit is not defined. Note that if the +5 classification attribute is not set or a constant is specified, it is assumed that the column has the same classification attribute and unit as the column to which the comparison or registration operation is performed. When explicitly specifying classification attributes and units, for example, write as follows.

Setting target (classification attribute 2 units) The above setting target section contains the column name and constant specification, followed by the classification attribute and unit specified in parentheses.

The DB operation management unit executes the requested operation after analyzing the requested operation. When performing an operation, when comparing data, type conversion is first performed to match the data types. When comparing character data and numeric data, convert the character data to match the data type of the numeric data. Note that when character data needs to be handled as numerical data, if the character data conforms to the numerical description format, the type is converted to numerical data. If it does not follow the numerical description format, type conversion cannot be performed and it will not be subject to subsequent processing. Then, in the case of numeric value 6 data, if a unit is defined, conversion is performed to match the unit.

Also, when performing a registration operation when adding or updating data, the registration operation is performed after type conversion and conversion, as in the case of comparison. Note that when registering numerical data as character data, type conversion is performed after conversion.

The classification attributes and units that are the results of four arithmetic operations on data with different classification attributes are treated as equivalent to the classification attributes and units that are to be compared and registered using the results. The same applies to the operation result of a built-in function specification.

In the case of four arithmetic operations on data of the same classification attribute, when adding or subtracting convertible data, the addition or subtraction is performed after unit matching of data values. At this time, the unit of the calculation result is the converted unit. The classification attributes and units of the results of multiplication and division operations are treated as equivalent to the classification attributes and units of the objects to be compared and registered using the results.

The above-described processing makes it possible to compare and register data of different data types and to compare and register data of different units without re-editing the DB. Requests can be issued without knowing the details of the definition information.

(Example) Hereinafter, an example of the present invention will be described in detail based on the drawings.

FIG. 1 is a diagram illustrating the overall configuration of an embodiment of the present invention. In the embodiment shown in FIG. 1, a DB access system 20 implements the present invention. The user accesses the DB access system 20 via the user program IO.
request processing.

The DB access system 20 is broadly divided into request control units 30. DB definition management section 40. It is composed of three parts: a DB operation management section 50; The request control unit 30 receives requests from users and distributes processing depending on the contents. DB
The definition management section 40 is sometimes called by the request control section 30 to accept requests for DB definitions or definition changes, and updates the DB 60 and definition information holding section 70 according to the contents of the request. The DB operation management unit 50 is sometimes called by the request control unit 30 to accept requests for operations (additions, changes, deletions) on data in the DB 18-60, and uses definition information in the definition information holding unit 70 to accept requests for operations (additions, changes, deletions) on data in the DB 18-60. After parsing the request,
Manipulate data in DB60. In addition to the data type 83, the definition information holding unit 70 includes column attributes such as a classification attribute 84 indicating a semantic type and a data unit 85.
Such information is additionally held in the column definition information management table 80. Also, as definition information regarding the unit of data, unit name 91. Basic unit 92. The conversion formula 93 is held in the unit information management table 90. In the embodiment, since a relational DB is assumed, management information is managed in a table format.

FIG. 2 is a diagram showing the minimum equipment configuration for implementing the present invention. The aforementioned DB access system 20 and user program 10, which are components of the present invention, operate on the computer 2. The secondary storage device 3 stores a DB 60 and a definition information holding section 70. The terminal 1 is used to start the user program IO and to interact with the user program 10. Although only one computer is shown in Figure 1+9-2, the above-mentioned components may be placed on multiple computers connected by a network.

Note that the DB access system 20 shown in FIG. 1 includes three forms for managing a plurality of DBs, as shown in FIG. 3 above.

Hereinafter, the processing contents of the DB access system 20 when receiving a DB definition, definition change, and DB operation request from a user will be explained in detail.

(1) Request Control Process First, the flow of the process of the above-mentioned request control section 30 is shown in FIG. The request control unit 30 first parses the content of the request from the user and generates an analysis tree (step 30
1). After that, check the contents of the request in step 30.
2) If the request is related to the DB definition,
The DB definition management unit 40 is called (step 303). D
If the request is related to B operation, the DB operation management section 50 is called (step 304).

For other requests, call the corresponding management department (
Step 305). Note that the above-mentioned other requests 20 are related to DB control, etc., but since they are not directly related to the present invention, their explanations are omitted here.

When the processing of each management section is completed and control is returned, control is returned to the calling user program IO.

FIG. 13 shows the processing flow of the syntax analysis 301 described above. In the syntactic analysis 301, while performing lexical analysis, for example,
An analysis tree as shown in FIG. 9 is created.

This analysis tree is an example of analyzing the inquiry shown in FIG. The analysis tree includes a request management block 200 that holds the overall structure of the request, a node that shows the details of the request (the part surrounded by an oval in the figure), and a storage part for names and constants (the part surrounded by a square in the figure). It consists of The request management block 200 includes a field 2 that holds the type of request.
01 and according to the sentence structure describing each type of request,
It consists of pointers (202 to 207) to the node list. The meaning of the pointer varies depending on the request.

As shown in FIG. 13, in syntax analysis 301, the first lexical is analyzed to check the request type, and a code indicating the content of the request (search, addition, change, deletion, table definition) is added to the request type of the request management block. etc.) (step 320). Then, according to the type of request, a parse tree is generated according to each request type while checking whether the syntax specifications are followed (steps 321 to 330).
).

For example, the processing flow for the search request 327 is
As shown in the figure. First, a projection item node indicating a column to be searched from the 5select clause is connected to the request management block (step 340). In the example of FIG. 9, a projection item node 218 and name storage areas 219 and 220 are created for x and name in the arrangement shown in FIG. 5, and the projection item pointer 203
Points from. Next, a table specification indicating the table to be searched is extracted from the from clause, a table specification node is generated, and it is stored in the request management block (step 341). In the example in FIG. 9, the table specification node 21 corresponds to the three tables specified in the from clause.
0 to 212 and storage areas 213 to 217 for their respective names.
are generated and connected by a pointer chain, and the table 1 fixed pointer 202 points to the first table specification node 210.

After that, the where clause, group by clause, ha
If the ving clause and orderby clause are specified, corresponding nodes are generated and connected to the fields pointing to the condition clause, group specification, group condition, and sort specification in the request management block (steps 342 to 345).
).

In the example of FIG. 9, only the where clause is specified, so the AND nodes 221, 229, comparison condition nodes 222, 230, 236, and column specification nodes 223, 224, 231, 232, 237 corresponding to the conditional expression are specified. ，
238 and name storage area 225-228.233-235
．． 239 to 242, connect them with a pointer chain, and create an AND node 22+ from the conditional clause pointer 204.
point.

Each node in the analysis tree has a node type 251. which indicates the contents of the node, as shown in FIG. 10 250. Left and right node pointers 252 for pointer chains between nodes.
253, fields 254 and 255 indicating the classification attribute and unit specified in the request, and a field 256 indicating a conversion formula used in subsequent processing.

In this embodiment, if a classification attribute or unit is specified in a constant specification or column specification specified in a projection item or condition, the specified classification attribute or unit is copied to the area of the analysis tree, Add processing to point from the corresponding node. Normally, the analysis process for constant specification and column specification is made into a common module, so you only need to make changes there. Analysis processing of additions, changes, and deletions (steps 328 to 3)
30), the above-mentioned changes are the same. When defining a table or changing a definition, add analysis processing when a classification attribute or unit is specified in a column definition. Also, add analysis processing for unit definitions and their change requests.

(2) Processing of the DB definition management unit 40 I) The flow of processing of the B definition management unit 40 is shown in FIG. The DB definition management unit 40 first, in step 401,
The request type 201 of the request management block 200 is checked, and the analysis tree and definition information holding unit 7 are created for each request type.
The requested process is executed while referring to 0. When the processing is completed, control is returned to the request control unit 304.

The processing for each request type is shown below.

(2-1) In the case of table definition In the case of table definition, it is first checked that the specified table is undefined and that the specified column names are unique (steps 402 and 403). Also, it is checked that the unit specified in the column definition exists in the unit information management table 90 (step 40).
4). After that, register the table definition information of the defined table in the table definition information management table in step 4.
05), the definition information of the column group is registered in the column definition information management table 80 (step 406). When registering column definition information, if classification attributes and units are specified, they are also registered.

For example, in the case of the employee table definition 130 in FIG. 6, column definition information is stored as shown in the column definition information management table 80 in FIG.

(2-2) When changing the table definition When changing the table definition, first check whether the specified table is already defined (step 41).
O). Next, the contents of the change request are checked (step 411). When adding a column, after checking the uniqueness of the column name, the column definition information of the column to be added is added to the column definition information management table 80 (step 41).
2). If classification attributes and units are specified, register them together with column names and data types. In the case of column definition update, after checking the existence of the specified column, the column definition information of the specified column in the column definition information management table 80 is updated (step 413). In this process, when a unit change is specified, it is checked whether the unit exists in the unit information management table 90, and if there is a change in the classification attribute or unit, it is checked in the column definition information management table 80. , is characterized by changing the classification attributes and units to be changed. Note that changing the unit of a column is to correct a definition error, and changing the unit does not change the data value.

(2-3) In the case of table deletion In the case of table deletion, after checking that the specified table has already been defined (step 420), all table definition information and column definition information of the specified table are deleted. Delete from the information holding unit (step 421,
422). In table deletion, there are no changes according to the present invention.

(2-4) In the case of a unit definition In the case of a unit definition, the new unit specified in the unit definition is
The unit specified in the conversion formula does not exist in the unit information management table 90 (step 430), and the unit specified in the conversion formula already exists in the unit information management table 90 (step 4
Check 31). After checking, the basic unit of the unit is set (step 432).

A unit without a conversion formula is itself a basic unit. If a conversion formula is specified, the unit definition information of the unit specified in the conversion formula is extracted and the basic unit is set as the basic unit of the defined unit. Finally, the definition information of the specified unit, such as the unit name and conversion formula, is registered in the unit information management table 90 (step 433).

(2-5) In case of unit definition change In the case of unit definition change, that is, conversion formula change, the specified unit must first exist in the unit information management table 90 (step 440), Then, it is checked whether the unit specified in the conversion formula already exists in the unit information management table 90 (step 441). After checking, set the basic unit (step 4)
42). When a conversion formula is specified, the conversion formula for the unit information of the specified unit in the unit information management table 90 is changed, the unit information of the unit specified in the conversion formula is extracted, and its basic unit is changed. , set as the basic unit of the unit to be changed. However, if the units in the conversion formula before and after the change are the same, the basic unit will not change. If no conversion formula is specified, the unit itself becomes the base unit.

Thereafter, necessary conversion formulas and basic units are changed for the specified unit definition information in the unit information management table 90 (step 443). Finally, update the definition information for the units affected by the definition change (step 4).
44). That is, when the basic unit of the unit whose definition is to be changed changes, the basic unit in the unit definition information for that unit is updated for the unit specified in the S conversion formula. This update is performed recursively.

(2-6) In the case of deleting a unit definition In the case of deleting a unit definition, first, it is checked whether the specified unit exists in the unit information management table 90 (step 450). Next, for the column whose unit is the unit to be deleted, the designation of the unit is deleted from the column definition information (step 451).

Furthermore, in step 452, the unit definition information of the unit that specifies the unit to be deleted as the unit in the conversion formula is deleted. This deletion process and the accompanying deletion of the unit specification in the column definition information are performed recursively. Finally, unit information management table 9
Delete the specified unit definition information in 0 (Step 4
53).

The processes (2-4) to (2-6) above are newly introduced by the present invention. In addition, if the process of changing or deleting the unit definition above involves updating other unit information,
One way to think about it is to create a specification that does not accept requests that are considered errors or requests that cause recursive updates.

The units defined above include "time": seconds2
Minutes, 9 hours, 9 days, 2 years, etc. "Weight": nigerum, kilogram, ton, ounce, momme, etc. "Length": meters, yards, dimensions, etc. can be considered. . If it is considered standard, it may be possible to equip it on the system side.

(3) Processing of the DB operation management unit 50 The flow of processing by the DB operation management section 50 is shown in FIG.

The D'B operation management section 50 first performs semantic analysis based on the analysis tree received from the request control section 30 while comparing it with the contents of the definition information holding section 70 (step 501). In this semantic analysis, table definition information and column definition information are retrieved from the definition information holding unit 70, and the validity of the request is checked, such as checking the existence of a specified table or column. Furthermore, as shown in Figure 1+, table management blocks 270, column management blocks 280, etc. that are required in subsequent processing are created, and the pointer chains 30 of the analysis tree are replaced. After the semantic analysis, optimization is performed to determine the access procedure to the DB, which is a characteristic of relational DBs (step 502). After optimization, a processing procedure for executing the determined access procedure is generated (step 503). Thereafter, the generated processing procedure is executed to realize the requested DB operation (step 504). When the processing is completed, control is returned to the request control unit 30.

Details of each process are shown below.

(3-1) Semantic analysis The flow of semantic analysis processing is shown in FIG. 17. In the semantic analysis, first, in step 510, the table definition information of the specified table is searched for while tracing the chain of table specification nodes, and the table management block 270 is created while checking for existence. At this time, the pointer to the table name of the table specification node is changed to point to the table management block. Note that this table management block 270 holds table definition information such as a table name 271. In addition, if a label is specified, 1
- If so, generate a label management block 275 and establish a pointer chain to the table management block 270 of the labeled table. The pointer to the label name of the table specification node is replaced with a pointer to the label management block 275. The label management block 275 includes a label name 276 and a pointer 2 to the corresponding table management block.
Holds 77th grade. Note that the table management block 270
generates only one for each table.

Next, in step 511, while tracing the chain of nodes such as projection items and conditional clauses, those with specified columns are extracted, the column definition information of the specified column is searched, and the column management block 280 is generated. The presence of the column is checked by doing this. If the column has a table name or label name, check whether the column exists in the corresponding table. If the column does not have a table name or label name, it is checked that the column exists in the table in which the table management block 270 was created, and that it does not exist in multiple tables. Furthermore, a pointer to the column name, table name, and label name of the node where the column is specified are respectively stored in the corresponding column management block 280. The table management block 270 and label management block 275 are reassigned to point to them. Only one column management block 280 is generated for each column, and includes column name 281, classification attribute 282 . It holds column definition information such as a unit 283 and management information such as a pointer 284 to the table management block 270 of the table to which the column belongs.

Finally, in step 512, a type compatibility check is performed. Conventionally, data type compatibility was checked between the column in which the value is registered and the value to be registered in the case of addition or change, and between the objects to be compared in the condition. In other words, comparison and registration between single character data and numerical data caused an error. In the present invention, compatibility is checked by classification attributes and units instead of data types. Even if the data is character data, if the content is in a numeric representation format, it will be treated as a numeric value, so no error will occur.

As a representation format for numerical values, in SQL, [numeric constant (n u
A format called meric 1itera+) is used. exact numeric 1ite
ral) and approximate numbers (approxima+, e number
An example of specifying ric 1 iteral) is shown below.

(True number) 123. -23.4. ．． 385 (approximate number), 42
8E10, 1.88E-10 The procedure for this compatibility check is shown below based on the processing flow of FIG.

First, the classification attributes and units of the corresponding columns and constants are set in the column specification nodes and constant specification nodes in the analysis tree (step 520). In the case of a column, the classification attribute and unit of the column are set from the corresponding column management block 280. If there is no classification attribute, set it to indeterminate. It is also undefined if there is no unit. When setting a unit, the basic unit and conversion formula of the unit are retrieved from the unit information management table 90 and set together in the column designation node. In the case of a constant, when parsing the constant, if there is no specification that sets the classification attribute or unit specified in the constant specification, each becomes undefined. When a unit is specified, the basic unit and conversion formula of the unit are searched from the unit information management table 90 and set. If the unit specified by the user is not in the unit information management table 90, an error is generated.

Next, if an arithmetic expression is specified, in step 521, classification attributes and units are set for the node corresponding to the arithmetic expression. The classification attribute of the arithmetic expression is made undefined.

If the arithmetic expression consists only of addition and subtraction, and the basic units of all terms in the arithmetic expression match, then the unit of the arithmetic expression is made the basic unit. In other cases, the unit of the arithmetic expression is undefined.

Finally, at step 522, the individual comparisons in the request.

Check validity of registration. Compare the classification attributes on the left and right sides of the node corresponding to the comparison and registration operation. It is valid if the classification attributes on both sides match, or if one of them is undefined. Otherwise, an error will occur. Next, compare the units. It is valid if the basic units of the units on both sides match, or if one of them is indefinite. Otherwise, an error will occur.

35 If the unit becomes undefined, the unit information (conversion formula, basic unit, etc.) of that node will be deleted.

In the inquiry example shown in FIG. 5, there are three comparison conditions, and the above-described processing is performed for each of them. The comparison condition for the department numbers is the same classification attribute (deptcode) and is valid. Comparison of administrator number and employee number shows that although the data type is different between integer and character string, they have the same classification attribute (em
pcode), so it is valid.

Salary is a column in the same table, so the classification attributes and units match and are valid. Therefore, this arrangement is judged to be appropriate.

(3-2) Optimization In the optimization step 502, candidates for access procedures are generated, the access cost of each candidate is estimated, and the access procedure with the minimum cost is selected. This process is described, for example, in the paper RelaLi by W, F., King, M.
onal Database System: Wh
are WeStand TodayJ(I F I
2.2 AccessPath 5 in P, +980)
Election (pp, 373-375).

Regarding this processing, optimization processing of the conversion formula is added. If conversion processing is required before comparison and registration, that is,
Add processing to convert the conversion formula as follows when the target units for comparison and registration are both non-determined.

■ <For comparison> If the conversion formula for one unit is a conversion formula from the other unit, use that conversion formula to match the units. If not, convert both conversion formulas to conversion formulas from the base unit, then compare the coefficients and create a conversion formula from the one with the smaller coefficient to the one with the larger coefficient,
Use this for unit matching.

■<For registration> Generate a conversion formula to convert the unit to be registered into the unit of the registration destination.

Additionally, in an arithmetic expression consisting only of addition and subtraction, when the basic units of all terms in the arithmetic expression match, a process is added to convert the conversion expression of each term into a conversion expression from the basic unit. In addition,
The addition of the above processing does not affect the order of conventional optimization processing, so it can be added anywhere within this optimization processing.

(3-3) Processing Procedure Generation Processing procedure generation step 503 generates a processing procedure that performs processing according to the access procedure determined by the double optimization. Regarding the present invention, the following procedure is generated while generating a processing procedure that follows the access procedure.

■Procedure for converting character data into numerical data when comparing character data with numerical data or registering it as numerical data.

■Procedure to convert numerical data to character data when registering numerical data as character data.

■Conversion procedure when there is a unit conversion formula.

Note that in the case of registration, if both a type conversion procedure and a conversion procedure are performed, the execution order of the procedures is set so that the conversion procedure is performed on numerical data.

(3-4) Execute processing procedure Execute the generated procedure. There are no process changes related to the present invention.

Note that in the above-described embodiment, object code that can be executed as a processing procedure is generated. Therefore, when executing a processing procedure, it is possible to execute the requested operation simply by passing control to the beginning of the generated processing procedure. Depending on the DBMS, there is a method in which a processing procedure at an intermediate language level is generated in the processing procedure generation step 503, and processing is performed interpretively while interpreting the processing procedure in the processing procedure execution step 504. In the case of such a method, since the data conversion procedures and conversion procedures described above are also performed interpretively, it is necessary to add processing to interpret and execute these procedures.

According to the embodiment described above, it is possible to manage classification attributes and units in the DB access system 20, and there is an effect that operations on data whose data types and units do not match are facilitated. Note that the above embodiment is shown as an example of the present invention, and the present invention is not limited thereto. For example, 9 It is also effective to implement the present invention by assigning only data classification attributes.

〔Effect of the invention〕

As explained in detail above, according to the present invention, comparison and registration between data of different data types becomes 1'+J possible without re-editing the DB, and the remarkable effect of improving usability for the user is achieved. There is.

Furthermore, since it becomes possible to compare and register data in different units without re-editing the DB, and the system automatically performs conversion between units, it has the effect of improving usability for the user.

In addition, by making it possible to compare and register data with different data types and units, multiple databases developed separately can be used.
In building distributed systems and integrated systems,
This has the effect of making it easier to maintain consistency between data, improving the flexibility and availability of the DB system, and making it easier to expand the scope of application.

[Brief explanation of the drawing]

Fig. 1 is a diagram showing the overall configuration of a DB system that is an embodiment of the present invention, Fig. 2 is a diagram showing the minimum equipment configuration of the system of the embodiment, and Fig. 3 is an operational form of the DB access system of the embodiment. FIG. 4 is a diagram showing an example of a conventional DB definition request, FIG. 5 is a diagram showing an example of a makeshift request, and FIGS. 6 to 8 are DB definitions and changes in the embodiment. and a diagram showing an example of a unit definition request, Figure 9 is a diagram showing an example of the structure of the parse tree after syntax analysis, Figure 1O is a diagram showing the configuration of nodes in the parse tree, and Figure 1+ is a diagram after semantic analysis. Figures 12 to 18 showing the structure of the analysis tree are flowcharts showing the flow of processing in the DB access system of the embodiment. l: terminal, 2: computer, 3: secondary storage device, 10 user program, 20: DB access system, 30: request control unit, 40: DB definition management unit, 50: DB operation management unit, 60: DB , 70: definition information holding unit, 80: column definition information management table, 90: unit information management table,
270: Table management block, 275 Label management block, 280: Column management block. 8 figure untt yen 51 ◇define unit kiJoyen-1α)
O＊yen""'y define unit man
yen -10*kiloyen sign square root,, 1-G) 4 so 41 L18. ) Symbol 154941 (19) Fig. 1 Fig.

Claims (1)

[Claims] 1. A request control unit that includes a database and a definition information holding unit that holds definition information of data to be stored in the database, receives requests from user programs, and distributes processing according to the content of the request. , a database definition management unit that updates the database and the definition information storage unit according to a request in the case of a database definition or definition change; In a database access system having a database operation management unit that operates data in the database, the definition information holding unit stores data for determining validity when comparing data or registering data. The method further comprises a step of adding an area for holding classification attributes, and receiving definitions of classification attributes of the data and changes thereto in the database operation management unit, and updating the corresponding areas in the definition information storage unit. Database operation processing method. 2. The database operation according to claim 1, characterized in that, in addition to each of the constituent requirements, the database operation management section is provided with a step of validating the requested operation based on the sameness of classification attributes of the data. Processing method. 3. In addition to each of the above configuration requirements, when comparing character data and numeric data that match the classification attributes of the data, or when registering character data as numeric data, character The database according to claim 2, further comprising a step of converting the data into numerical data according to its contents, and a step of converting the numerical data into a character data format when registering the numerical data as character data. Operation processing method. 4. In addition to each of the above configuration requirements, the database operation management unit includes a step of ignoring the sameness of the classification attributes of the operation target in the four arithmetic operations between data, and a step of ignoring the classification attributes of the operation result after the operation. Claim 2 further comprising a step of treating the classification attribute as being equivalent to the classification attribute.
The database operation processing method described. 5. In addition to each of the above configuration requirements, a step of allowing the database operation management unit to change the classification attribute specified by the user during the request, and regarding data with undefined classification attributes as equivalent to the classification attribute of the operation partner. 3. The database operation processing method according to claim 2, further comprising a step of processing the data. 6. A request control unit that includes a database and a definition information holding unit that holds definition information of data to be stored in the database, receives requests from user programs, and distributes processing according to the content of the request, and a definition or definition of the database. A database definition management unit that updates the database and definition information holding unit in accordance with a request in the case of a change, and a database definition management unit that updates the data in the database in accordance with the request in the case of an operation (addition, change, deletion) request for data in the database. In a database access system having a database operation management unit that performs operations, an area for storing units of data and conversion formulas between units is added to the definition information storage unit, and an area for storing units of data and conversion formulas between units is added to the definition information storage unit; 1. A database operation processing method, comprising the step of accepting definitions of units and conversion formulas between units and changes thereto, and updating corresponding areas of the definition information storage unit in accordance with requests. 7. The database according to claim 6, wherein, in addition to each of the constituent requirements, the database operation management section is provided with a step of validating the requested operation based on the possibility of conversion between units of the data. Operation processing method. 8. In addition to each of the above-mentioned configuration requirements, the definition information holding section holds an area that holds the basic unit of the defined unit, and the database definition management section stores the following information when defining or changing the definition of a unit: The step of finding the basic unit by recursively tracing the relationship between the units in the conversion formula and the conversion formula, and setting it in the corresponding area of the definition information holding unit, 7. The database operation processing method according to claim 6, further comprising the step of determining, when the units of the data do not match, the conversion possibility between the data based on the match of the basic units of the units of the data. 9. In addition to each of the above configuration requirements, when comparing character data and numerical data whose units can be converted, or registering character data as numerical data, character data must be Convert the numerical data according to the content, and if conversion is necessary, convert it to the unit of the numerical data of the operation partner according to the conversion formula, and when registering the numerical data as character data, convert the numerical data. If necessary, operate according to the conversion formula. 7. The database operation processing method according to claim 6, further comprising the step of converting the character data of the other party into units and then converting the data into a character data format. 10. In addition to the above-mentioned configuration requirements, a step of allowing the database operation management unit to change the unit specified by the user during the request, and processing data whose unit is undefined as being equivalent to the unit of the operation partner. 7. The database operation processing method according to claim 6, further comprising the step of: