JPS59502038A - Dynamic database representation - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention] 1 Dynamic database representation Technical field The present invention is used in computer-based databases, especially computer-based information retrieval systems. This paper relates to dynamically changing representations of database information.

Background of the invention Representing large amounts of information as records in a database that can be accessed by a computer It is often done. Such databases are typically magnetic disk systems. stored in a mass storage device such as a stem. In the prior art, certain Use the calculator bar to access and work with such information. in such storage media to take advantage of the specific computer software available. Each record was formatted and structured.

Database access is required to take advantage of and adapt to new environments. When the program must be changed to reflect the new data format. It is often necessary to reconfigure the database to For example, a data item The entire database must be reconfigured to reflect this change due to the addition or deletion of Sometimes it doesn't. If your database is very large, The size itself may be a limiting factor in improving database systems. sand In other words, once the database size reaches a certain scale, the data will be processed according to the external environment. Changing the format of data in the base becomes uneconomical. At this point database innovation has stopped, and database innovation has stopped, with static data in an external environment that does not change over time. It becomes an expression. Therefore, from this point on, the usefulness of the database will depend on the passage of time. This will result in a decline.

King Jinri!旌 According to one embodiment of the invention, a data representation method used external to the processing computer By separating the data representation used inside the processing computer from the These drawbacks are resolved. By separating these two data representations , it is possible to eliminate the limiting factor of database size in the development of information retrieval systems. can.

To be more specific, the basic fixation of records is first devised.

From this - membranous record definition, two distinct but logically equivalent Data representation is induced. The first data representation, called the external record, is database itself, but this is the computer that uses the database. External to the program. The second data representation is called an internal record, which is exists in computer programs that handle databases and therefore Sharing data structures and formats native to the program's own programming language use Internal data representation of information and any desired format required by the external RAM method -Used to adjust structural differences between nodes. Therefore, internal and external Code formats are independent of each other and can be used to meet specific requirements in the environment. You will be able to do it.

For example, external record formats are independent of any processing program; It can also be made to have meaning on its own.

Therefore, external data is defined as a pure hierarchical structure of alphanumeric character strings. In reading A and reading A, it is possible to format 6 as a pair of shaped data items and data values. I can do it. That is, each row of a record consists of the name of an information entity and its entity value. Ru. Rows are intended to properly reflect hierarchical relationships between data items . Internal data formats, on the other hand, are used to form processing programs. is formed in such a way that it takes advantage of all the expressive power of the programming language from which it is written.

In this configuration, the representation in the external database is as pure alphanumeric text that is generated. is generated and maintained using standard word processor techniques. In addition, data Deletion and modification of data items are performed using these word processing techniques. like this The internal data representation required to process information in a database is a processing program. dynamically generated from an external data representation by the RAM. Accessing external records The program on the computer that is requesting this also converted into a suitable form. Unique capabilities of the original programming language and the hardware used By taking advantage of the unique physical characteristics of the software system, you can When the request program is written in a different original programming language (NPL), can provide different internal representations of Therefore, completely new hardware This can be done without any changes to the system or database.

Brief description of the drawing Figure 1 shows a data hese W product search system using dual data representation according to the present invention. General flow chart; Figure 2 shows more of the offline preparations used to realize the system in Figure 1. Detailed block diagram; Figure 3 shows a more detailed explanation of the Rantaises used to implement the system in Figure 1. Flock diagram; Figure 4 is a flowchart of the GETREC routine used in Figure 3; Figure 5 shows a detailed flowchart of the P[JTREC routine used in Figure 3. For reference, Figure 1 shows how different data representations are used inside and outside the computer. 1 shows a general block diagram of a database system. This system is convenient For convenience, this is called the single-tier record interface technique (GRIT). Especially database 1 0 has a number of records, each of which can be accessed directly by a standard word processing system 11. It is represented by alphanumeric characters that can be accessed and edited. In other words, All alpha vents, numbers, line and space control characters are ASCII A standard file that can be directly displayed or printed without processing on a word processing terminal 11 such as expressed in code. Therefore, a new record is added to database 10. or modify existing records in the database 10, or Using a word processing terminal to delete records from base 10 Can be done.

The format of each record in database 10 (e.g., order of writing) ) is described in the record definition 12 in standard form.

In general these formats can be completely arbitrary, but the relationships between the data It is desirable that the information be arranged so that even beginners can understand it at a glance. For example, In code definition 12, each data field's label is entered into that field. is supplied along with the maximum length in characters of data to be used. These record specifications i1 2 is supplied to the record compiler 13 in a machine-readable form.

The compiler 13 takes the record definitions 912 and converts each of the records into the original program language. Structure definition of word (NPL) and initialized NFL data called record descriptor Structure 15 (called General Record Interface Processor, GRIP) It is a computer program.

GRIT record definitions are processed by the GRIT compiler 13. Each record Two files 14 and 1.5 are created for the code. On the other hand (file 1 4> contains the definition of the data structure and the other (file 15) contains the record descriptor. .

The data structure definition 14 is written in the original programming language, that is, NFL (for example, C language). Represents a RIT record representation. This definition covers the various forms of data structures needed for the application. used by application programmers to generate

Record descriptor 15 associates a name in the form of an ASCII string with a particular offset. A symbol table with sufficient information to fit into a given data structure. be. This descriptor uses the basic online I10 functions getrec() and putre c (), and this is used in G of internal programs and external files as described later. Used to perform mapping between RIT formats. Record descriptor is NFL , each of which has its own initialized instance. Has a common data structure shape used for record descriptors.

GRIT record definitions are written in a record description language. This is the intended summary Based on the simple description of the shape. For example, inf.

name　C40〕 addr log (2,) rm(6) has two fields “name” and “addr” (address omitted). Define a GRIT record called “1nfo” that includes: name is 40 sentences characters or less. addr has three fields “log”, “rm” and “e xt” and the maximum length of each is 2.6 and 4 characters.

The length or depth of the GRIT record definition (or deep indentation) is essentially There are no restrictions.

The formal description of the record definition language is the Hasokasnaur Format (BNF) grammar as follows: given to.

Record definition: field definition Field definition: User name object definition object definition @: [integer] Object definition: →Definition list← Definition list: Definition list: Definition list field definition Definition list: Definition list array definition Array definition: Username [integer] Object definition username is given by the user. must be consistent with NFL name syntax. Arrows - and - are one Indicates the increase/decrease in indentation due to tabs. Therefore, the phrase ``one definition list'' represents an indented list of definitions. An integer is an integer number, a feature In field definitions, it represents the maximum number of characters in the field; in array definitions, it represents the maximum number of characters in the field. Represents the number of objects in the array.

The grammar is interpreted as follows. Record definitions are field definitions. feel A code definition is a username followed by an object definition. Object definition is in parentheses is an integer in , or the intended definition list. There are 0 definition list or more field or array definitions. Array definition is username It is followed by an integer in parentheses, followed by an object definition.

Please note that the grammar is cursive. For example, A field definition contains an object definition, which contains a definition list, which contains field definitions, but this is the starting point. With this definition method, GRIT The general tree-like nature of record definitions emerges, where each field in turn It will consist of fields.

The GRIT compiler 13 implements the BNF grammar described above and writes the GRIT record definition. Forms a dendritic internal representation. This tree is then used by two subroutines. Ru. One generates file 14 containing the NFL structure definition and the other contains the record description. Launch a file 15 containing children and an initialized NPL structure. NFL compiler 16 therefore stores the record descriptor 15 directly in the computer's memory 12. Compile to the object code format 17 that is passed through the code.

Application program 18 is written using the same source code, but NFL structure definition 1 4 and is compiled by the NFL compiler 19. Compilers 16 and 19 for two compilation operations (record descriptor 15 and application program 18) It may be the same compiler used at different times. once off soextco When compiled into a program, record descriptor 17 from compiler 19 and off-shelf The application program for these modules is passed to the rotor program 20, which The file is stored in the internal memory 21 of the general-purpose digital computer. Memory 21 is 4 Section 22.23.24 and 25 as a memory map. This is illustrated in Figure 1. A user application program in the form of object code is It is stored in section 22. Record descriptors are stored in section 23 of memory 21. Stored. Two other programs reside in memory 21 and in section 24. The “GRTREC” program is in section 25, and the “PUTREC” program is in section 25. Contains grams. The GRTREC program 24 is external from the database 10. A program that has an internal format for database records that have a formant This is an input conversion function to be stored in the memory 21. PtJTR, EC program 25 transfers the internally formatted records in the program memory 21 to an external format. This is an output conversion function that copies the data to the database IO in the mat. GRTREC and PUTREC is the record specified to direct internal/external format conversion. This is a general program that uses the information in the code descriptor 23. Of course, each For each individual record definition 12, a unique record descriptor 23 exists. De Not only database IO but also record address and actually accessing the record. includes the physical motion required for I'm reading. For example, in a magnetic disk storage system, addresses are number, sector number, and track number so that the read head can read the disk pack. must be able to move to the correct trunk on the appropriate sector of the specified disk. Must be.

Both GRTREC and PUTREC collect three pieces of information each time they access database 10. information (parameters) is required. This is where the record is stored (P[JTR EC) or the memory 21 in which the record is to be stored (GRTREC) address is required. Each of programs 24 and 25 also includes an internal former record 1 self-descriptor 23 to indicate conversion between the formant and the external formant. Need a dress. Finally, programs 24 and 25 have records output between them. requires an address in the database IO to be populated.

All components and procedures illustrated above dashed line 27 are offline 11! Preparation i.e. the computer actually processes the data from the database IO. What is prepared when you have not done it before and can actually be done on a completely different computer? Please be careful. Furthermore, procedures above dashed line 27 are application programs. need only be run once for each new set of . These programs After being stored in the memory 21 in the form of objects, these programs according to a simple user request without compiling the object code again Executed an infinite number of times.

Therefore, the procedure below dashed line 27 is online, and therefore the application program System 18 is dynamically called and used from moment to moment in response to the specific needs of the application for which it is written. It will be done.

Other application programs similar to program 18 and other records similar to definition 12 data compiled at different times and by different people on the same computer. Note that the same database 10 can be used . In fact, these programs use completely different data structures and are completely different programs. can be written in any other language. In this case, compilers 168 and 19 are not used. The record compiler 13 is a compiler for the language that is written in the record definition 1. 2 into the appropriate record descriptor for the language used. Described later The GETREC program 24 and the PtJTREC program 25 are different as follows. Extend data fields with aborts and contacts to accommodate record definitions. Rui has a shortening function.

According to one feature of the invention, the format of the data records in database 10 is separate and completely independent from the representation of records in the computer's memory 21 , has become. Record interface programs 24 and 25 are data A translation method between the external record representation of Base 10 and the internal record representation of Memory 21A. Used as a canism. Records are converted on-the-go as necessary. record The code descriptor 23 contains all the information necessary to perform such a translation. Ru.

Records in database 10 are therefore often processed using a simple standard word processor. It has a format that allows for easy additions, deletions, and changes using the Ru. New fields can be added to records, and new records can be added to regular fields. added as a text. All you need is a new or modified record A new record definition 12 corresponding to is written, and a new record descriptor 15 is written. The structure definition 14 has been compiled and stored in the memory 21. day No structural changes are required in the database 10. Therefore, the entire database 10 is restructured. New applications of databases that require new data items without having to can be programmed. Therefore, the application of database is has various constraints that rewrite the entire database every time a record changes. It can grow smoothly without any restrictions.

Figure 2 is a more detailed diagram of the example offline program shown at the top of Figure 1. be. A simple but specific data structure is defined here; However, it specifies specific user operations. High level used in the example in Figure 2 The programming language of is "c" language, which is Prentice-Hall L, published in 978, B, W, Kernighan, M, Ritchie. It is described in detail in “The CP programming language”. It's being ignored.

Specifically, record definition ■2 is shown as a simple six-line list in Figure 2. ing. The first line starting from the left margin is named “1nfo” in Figure 2. Indicates a record identifier. The second line is indented from the margin. is the name of the first field of the record, i.e. Rahel, and in square brackets is Indicates the maximum number of characters allowed in the field. In Figure 2, the first The field gives Rahel “name” and its length can be up to 40 characters. Ru.

The third line of record definition 12 has Rahel “addr”, which is address represents. This does not have a field length, the field name "addr" is In reality, Figure 2 shows multiple subfields double indented from the margin. It shows. Therefore, the field "addr" has a maximum of 2 characters, 6 characters, It consists of subfields named “loc”, “,” and “ext” with four characters. ing. Rahel “loc” represents 1 location, which is the 2-sentence of region. (For example, MH stands for Murray Hill, New Jersey). Rahe "rm+" represents the room number, and "name" Contains the room number at the individual's location indicated by . Rahel “ext” is of course This is the telephone extension number of the individual indicated here.

Record definition 12 includes information related to a number of database records (e.g. , all employees of a particular company). Sara Gko Figure 2 Noshi code definition 12 definition format The set allows all fields of a record to be identified by name, i.e. Rahel. and indicates the maximum storage space required by each field in the record. lastly, Record definition 12 uses Rahel indentation to define the data field. To be able to express hierarchical relationships between Number of fields and indentation The level is infinite, so the definition format in Figure 2 can be used for almost all devices. database records.

Using this record definition, the GRIP program 13 creates a record descriptor 14 and A C language data structure definition 15 is generated. Record descriptor 14 is C compiler 1 9 repeats the information in the record definition 12 in a format suitable for compilation by Just give it back.

Record descriptor 14 contains literal field names and minimum and maximum fields. Contains limit values. Record descriptor 14 identifies fields by name and hierarchical position and is also used at runtime to indicate field size limits.

Data structure 15 specifies the specific program language to be used by the application programmer. is a definition of a general data structure written in words.

In the illustrated example, this is the C language, and the writing conventions that data structures follow are C It is a language habit. Of course, you can also select other programming languages; If the data structure 15 is AS I-C etc.) customs will be followed.

Block 18 represents a specific example of a user application program, also written in C. It shows. The functionality of the user program I8 is based on the data defined in box 12. It stamps the name and extension number of the individual represented in the record. data record Other more complex procedures involving deleting, adding, or changing code may also be considered. .

The first line of box 18 is a comment that is not executed, and "/*...*/" is This is a comment mark in C language. This comment indicates that this procedure will not be recorded in the database. Procedures for printing names and extensions of all individuals with codes, i.e. telephone directory Indicates that this is a procedure for creating a sting.

The second line of box 18 indicates that this program has database structure definition 15 (1nf o, h”)).

Data structure definition 15 correctly interprets data records recovered from the database is made available by program 18 to save, remember, and restore. There must be. The third line secures a storage area called “buf”. ``1nfo'' record is temporarily stored during processing. .

The fourth line of program 18 is the beginning of the program body and is complete with the rest of the lines. form a complete program. The fifth line forms a loop and uses this width to create a database file. The procedure "getrec" is called for each record in the file. “get The three parameters in the parentheses of “rec” are 1) temporary for the record (buf); address of memory, 2) record descriptor for record type “1nfo” address, 3) the name of the file from which records are to be returned (stdin ). The print next line command provides formatting information for printing each line. The last two lines specify the location of the buffer where the data to be printed is stored. do.

The user's source code 18 and record descriptor 14 are passed to the compiler program 19. Therefore, it is compiled as an executable object code program. , out”. When the program is executed, In some cases, the user program 20 of a and out is an internal memory of a general-purpose digital computer. control of the computer is transferred to position "a, out".

All procedures described in connection with Figure 2 can be performed “offline”. Leave it optional. In other words, these procedures require actual telephone directory listing. It can be done long before it becomes. Object code 22 requires a phone book later. Executed when needed. In fact, information in the database can be moved, extensions added, etc. Since it changes continuously due to additions, retirements, and promotions, the object code 22 It is executed many times, perhaps periodically.

FIG. 3 shows the process that occurs each time the object code 22 in FIG. 2 is executed. It shows the nature of the The record descriptor corresponds exactly to the record descriptor in Figure 2. However, this time it is in the form of object code instead of source code. Ru. In other words, the information content of descriptor 14 is an internal binary format suitable for direct search by a computer. expressed in the form

A computer program called “getrec” is also a computer program in object form. It's in. The database 10 has at least one record type “1nfo”. ” and exists in an external storage medium such as a magnetic tape.The storage space 31 is The internal memory of the computer is used as a temporary storage location for records from the database 10. Part of Molly. Storage space 31 is identified by Rahel "buf".

Procedure 18 for stamping the name and extension number can also be written as object code on a computer. exists in memory. Optionally, procedure 18 uses the “getrec” program. 24 (control is transferred to this).

The program 24 reads the external record “1nfo” from the database 10 and records the record. This is stored in the “buf” storage space 31 according to the format information of the code descriptor 30. Convert to Party A's internal record. As shown in the storage space 31, the file “1n The value (content) of the data field from "fo" is stored in the buffer as multiple character codes. It is stored in the memory 31. The unused reserved storage space of buffer 31 is The field is filled with "end of field" characters (10), and the end of the field is also filled with "fill" characters (10). "End of field" character is entered.The end of field character is placed anywhere in the field as a value. Any character or character string that does not occur can be used.

Print routine 18 uses record descriptor 30 to retrieve the desired value and Take the desired field values (name and ext) from the file 31 and output them. to the agent medium 32. The medium 32 may include, for example, a printed copy of a telephone directory. It is a standard printer from which pages are retrieved.

The systems in Figures 2 and 3 contain internal and external formats for database records. - Notice that the mats are separated. External record of database 10 Code has a format and is easily produced by standard word processing systems. Uses a storage method that allows the user to create and edit information. Internal record of buffer memory 31 The code, on the other hand, is in the form best suited for computer processing, and It has a format predetermined by the programming language. Program “gri” p" 13 (Figure 1), "getrec" 24 and "putrec" 25 (Figure 1) This allows the internal record format and external record format to be selected independently. De The database can grow smoothly without rewriting most of the processing software. Significant changes to processing software without changing database format can do.

FIG. 4 shows a program flow diagram of the procedure "getrec". get In response to the request 40 to run the program of rec, the program of getrec is executed. First, in box 41, the buffer 31 (FIG. 3) is cleared. in box 42 The database record with the symbol name "1nfo" is searched, and the database record is searched. is translated into the physical address of that record in O.

In box 43, the external records of database 10 are obtained in box 42. is searched using the specified physical address. This information is finally stored in buffer 31. processed before being sent. For example, in box 44, the data in the external record is fields in the data have no corresponding names in the record descriptor 30 (Figure 3). If so, store it in the internal buffer 31 or retrieve it from the internal buffer. There would be no mechanism used for this.

Correspondingly, in box 45, if the record field in descriptor 3o field is specified and the corresponding field does not exist in the external record. then the space reserved for the value of this field in buffer 31 is blank. It becomes a field of research. When I try to print this value, I end up printing blank spaces. Become.

When the field value of the external record is longer than that specified by descriptor 30 , the field is typed to the length specified by descriptor 30 in box 46. Be cut. (As mentioned earlier, short field values are end-of-field data Filled by Milita. ) box 47, if record descriptor 30 is a single value field, and the external record is a multi-value field (multiple values). field), multiple field values are combined into a single value. Naively concatenated to the allowable length of the field descriptor. On the contrary, If descriptor 3o has a multivalued field and the external record has a field with many values, box 48, each of a number of subfields. Repeat one field value. Finally, in box 49, the result The internal record obtained is stored in the buffer register 31. At this time, Execution of the getrec procedure is complete and control is transferred to the getrec procedure at box 50. The program is returned to the program from which it was originally called.

Data the new version of the user program by steps 44 to 48 of the flowchart in FIG. You will be able to use the old version of the base and vice versa. Therefore , create a new user program that requires additional record fields. All you have to do is recompile it with the record descriptor. old user program If the system continues to run, unused information will be discarded. This is it and then recode the existing program each time a database record is added. This allows database systems to grow smoothly without having to be re-engineered.

Table 1 shows the pseudo code implementation of "getrec".

Table 1 clear buffer call getfield() getfield() get name from fileand l1m1t’in buffe rif field is elementarycopy value fr om file 1nto bufferobserving 11m1t 1se fc) r eBc-h 5ubfieldcall getfield() eturn Figure 5 shows a computer program that sends data records back to the database after processing. 3 shows a flowchart for the program "putree"; In Figure 3, the data There is no change, so the version already stored in the database remains unchanged. putrec was not used. In this sense, records from database 10 Searching for codes is non-destructive.

The request 50 to put a record into the database is the same as the getrec procedure. It has three parameters. Internal buffer position (buf”), same external record name 1 nfo")'8 and the same name in the file where the record is stored 5tdin") It is. In box 51, if there is a blank field in buffer 31, , there is no need to remember a blank field, so this is discarded. bock The step 52 uses the value in the buffer 31 and the information in the record descriptor 3o to change the format of the department record into a form suitable for storage in database 1o, Add field descriptors from literals in record descriptors. in box 53 converts the symbolic address of a database record to a physical address. Bo In box 54, the physical address of database 1o obtained in box 53 is Records are stored. In box 55, the procedure "pu tree" is completed, Control is returned to the calling program.

Table 2 shows the pseudocode implementation of the putrec program.

Table 2 if rec desc is a gro, up node1se if field is non null (in the buffer) write field name (s) to files neces sary copy value from program buffer to fil e eturn FIG, / FIG. 2 FIG. 4 FIG. 5 UTREC international search report

Claims (1)

[Claims] ], convert the database records into a first external format in the data storage medium. and a means of expressing The database is stored in the internal storage medium of the digital computer in a second different format. A means to represent each of the records and only when one of the records is processed. means for translating the database record between the first and second formats; A database system comprising: 2. In the database system according to claim 1, the first The matte should contain only alphanumeric characters, symbols, and control characters for word processing. A database system characterized by. 3. In the database system according to claim 1, the second format -mat is the data structure of the programming language to be used in the database system. A database system that is characterized by being compatible with 4. In the database system according to claim 1, the translation means comprises: Furthermore, the length of the data field can be shortened by truncation or filling of blanks. A database system characterized by including means for decompressing. 5. In the database system according to claim 1, the translation means Furthermore, a different number of subfields is specified for the internal and external formats. A means of repeating or splicing data field values when A database system comprising: 6. In the database system according to claim 1, the database system further includes cough data. A database system characterized by including means for using database records. . 7. Data format in a format suitable for use by word processing equipment. Stores database records in storage media and supports data structures in programming languages. database records in the internal memory of a digital computer in a corresponding format. records associated with the processing of the one record by the digital computer. Only if there is a difference between the data storage medium format and the internal storage format. of a database record, comprising the step of translating a record. Usage. 8. In the method according to claim 7, the translation step further comprises: The data field length of one of the formats. data field to match the corresponding but shorter field length of the other side of the data field. shorten the length, Set the data field length of one of the formats to the corresponding data field length of the other format. stretches the data field length to fit the longer data field length. Ru A method of using a database record characterized by including stages. 9. In the method according to claim 7, the translation step further comprises: Data field values that occur uniquely in one of the formats are Repeat to match other multi-data field values, Concatenate multiple data field values in one of the formats to the other format. match the flower field position of the cut. A method of using a database record characterized by including stages. 10. The method according to claim 7, in which data in the database record is the step of adding, deleting or changing data field values. How to use database records.