JPH03218522A - Automatic generation system for program specification - Google Patents

Abstract

PURPOSE:To automatically generate a specification whose abstraction degree is high and which is easy to understand from a program by taking out a condition which satisfies the prescribed condition from plural branching sentences taken out from the program, obtaining similarity between the processings of the branching sentences and generating the specification of table structure. CONSTITUTION:The branching sentence whose condition satisfies the prescribed condition is take out among plural branching sentences which a branching sentence taking out part 1 extracts by syntactic analysis for the program. A similarity calculation part 2 collects the branching sentences whose name of variables are same kinds from the taken out branching sentences. Then, similar ity between the processings of respective sentences is calculated and a table generation part 3 generates the specification of table structure where a processing is substituted with an identifier based on calculated similarity and the description of the processing is simplified. Thus, the specification whose abstraction degree is high and which is easy to understand can automatically be generated.

Description

[Detailed Description of the Invention] [Summary] Regarding an automatic program specification creation method that automatically generates a worksheet from a program, a predetermined condition (
For example, create a table-structured specification that simplifies the description by extracting items that match the condition (all variables = all values and variable names are the same) and replacing them with identifiers based on the similarity between the processing of these branch statements. The purpose is to automatically generate specifications that have a high level of abstraction and are easy to understand.For multiple branching statements extracted from a program, the conditions in the branching statements are set to predetermined conditions {for example, all variables = value, variable name are all the same};
a similarity calculation unit that calculates the similarity of processing between each sentence by grouping together branch sentences with the same type of variable name (for example, the same type of judgment code) from among the branch sentences extracted by the branch sentence extraction unit; , and a table creation unit that creates a table structure specification in which processing is replaced with an identifier (for example, error processing) based on the similarity calculated by this similarity calculation unit, and the specification created by this table creation unit. Configure it to output a document.

INDUSTRIAL APPLICATION FIELD The present invention relates to an automatic program specification creation method for automatically generating specifications from a program. Software 1! During capability expansion/maintenance work, system design specifications are frequently referenced, and in large-scale systems, there are many of them, and the time required to search for specifications is enormous.
For this purpose, it is possible to improve the search process by computerizing specifications, but many of the existing specifications are created by hand. Simply inputting this information into a computer by hand would result in high work costs, and there would be problems with its accuracy. Therefore, a method for automatically generating specifications from existing programs is desired. [Problems to be solved by conventional technology and inventions] Conventional specifications were generated by converting variable names and keywords in the program into Japanese, and by showing control structures in diagrams. However, this basically only shows a one-to-one correspondence with the program, and has the drawback that it is not possible to show the program's functions at a sufficient level of abstraction.
The present invention analyzes multiple branching statements extracted from a program under predetermined conditions (for example, all variables = value, all variable names are the same).
Based on the similarity between the processing of these branch statements, we create a specification with a table structure that simplifies the description by replacing it with the ll identifier, and creates a specification that is highly abstract and easy to understand. The purpose is to automatically generate documents.

[Means to solve intelligence problems]

In FIG. 1, which explains means for solving the problem with reference to FIG. For example, g4 extracts the variables that match the condition (all variables = value, all variable names are the same).

The similarity degree calculation unit 2 collects the branch sentences with the same type of variable name (for example, the same type in the category of judgment code) from among the branch sentences extracted by the branch sentence extraction unit 1, and calculates the similarity between the processing of each of these grouped sentences. This is to calculate the degree. The table creation unit 3 creates a specification with a table structure in which the process is replaced with an identifier and the description of the process is simplified based on the 11th degree calculated by the similarity calculation unit 2. [Operation] As shown in FIG. 1, the present invention provides for a plurality of branching statements extracted by the branching statement extraction unit l by parsing the program.
The similarity calculation unit 2 extracts branching sentences whose conditions match a predetermined condition (for example, all variables = values, all variable names are the same), and the similarity calculation unit 2 extracts branching sentences whose variable names are the same (for example, judgment We collect code (of the same type in a category called code) and calculate the degree of similarity between the processing of each of these summarized sentences.
Based on the calculated similarity, the table creation unit 3 replaces the process with an identifier to create a table structure specification that simplifies the description of the process. Therefore, from multiple branching statements extracted from the program, those that meet the predetermined condition Cfj4, all variables = value, variable names are all the same] are extracted, and based on the i similarity between the processing of these branching statements, By creating a table structure specification that simplifies the description by replacing it with identifiers, the level of abstraction is high.
It becomes possible to automatically generate specifications that are easy to understand. [Embodiment] Next, the configuration and operation of an embodiment of the present invention will be explained in detail using FIGS. 3 to 10 in accordance with the order shown in the flowchart of FIG. 1. In Figure 1, ■ extracts a sentence from the intermediate representation. This extracts sentences from the 4th ryJ intermediate representation generated by parsing the input program example in Figure 3 (sentence 1, sentence 2, -
・Take out sentences in units of sentences). ■ determines whether it is a branching sentence or not. This will determine whether the sentence extracted in ■ is a branching sentence. For example, it is determined whether or not the sentence l extracted from the beginning of the intermediate expression in Figure 4 is a branching sentence.Here, it is determined that it is a branching sentence because it is parsed as "sentence 1 type branching sentence゜゜.If YES, , do ■.If NO, when [phase] is NO, extract the next sentence with O. ■ determines whether all the conditions are in the form of variable = value. Determine whether all conditions in the statement determined to be a branching statement are in the form variable = value. YES
In the case of (for example, condition 1.11 condition 1.2 in sentence L in Figure 4)
, Condition 1.3 &,: Tweet? respectively l[OZANoC
l=O, KOZANQCI[=L, KOZANOCI
[= QT anal 2, variable (IOZANOcK) = value (0, 1, 011Ell) and YES)
, [exist]. If the answer is NO, press ■ to extract the next sentence. ■ determines whether all variables are the same. This determines whether all the variables of the condition in the sentence are the same. If YES (if YES because all variables become [OZANOC[ for condition l-1, conditions 1, 2, and conditions 1.3 in sentence l in Figure 4), perform ■. If NO, [phase]
If the answer is No, press ■ to extract the next sentence. ■ converts the intermediate representation. This is @YES, 6Y
Convert the parts corresponding to ES and OYES. g4
For example, convert the area enclosed by the rectangle in Figure 4 as shown in the area enclosed by the rectangle in Figure 5. [Phase] determines whether the intermediate representation file is free. This involves, for example, extracting a sentence from the file storing the intermediate representation shown in Figure 4 and determining whether it is empty or not. If YES (empty), perform ■. If NO, extract the next sentence from the intermediate expression using ■, and repeat steps after ■. Otherwise, replace the intermediate representation file with the converted one.
As a result, for example, the part surrounded by a rectangle in the intermediate representation in Figure 4 is replaced as shown by the part surrounded by a rectangle in the intermediate representation in Figure 5. Through the above processing, for example, from the intermediate expression generated by the syntactic analysis in Figure 4, we can obtain a branching statement (@YES), a condition in which all variables = values (■YES), and all variables being the same ([
The branch statements, conditions, variables, and values that satisfy the predetermined condition (YES), which are enclosed in the rectangle in FIG. 4, are converted and replaced as enclosed in the rectangle in FIG.

At tm-th, @ retrieves successive transformed intermediate representations from the intermediate representation file. This extracts only the converted sentences from the intermediate representation after replacing them with the converted sentences in ■, and excludes unconverted sentences. [Similar] determines whether the variable names in the sentences are of the same type. This is, for example, the intermediate representation after performing the transformation in Figure 5, where the variable name of statement 1 = KOZANOCK', and the variable name of statement 2 = ``ZENZA''.
It is determined whether the variable name "RIRITυCl1" in NOCr1 statement 3 is of the same type or not, and here, these three variables are the account number determination code, the front balance determination code, and the interest rate code determination code by referring to the semantic information in Figure 7. , are of the same type in the category of determination codes, and determined as YES. In the case of YES,
Do [stocks]. If NO, it is not the same type and ends. @Pa, create the original form of the table. This is because sentences 1, 2, and 3 in the intermediate expression in FIG. Create the original form of the table summarizing the information as shown in the figure. [Phase] calculates the similarity between the processes in the table in order to simplify the description of the processes after branching. For example, the degree of similarity between processing 1.2 and processing 163 in Fig. 6 is calculated as shown in Fig. 9 (described later). [Phase] determines whether the frequency is greater than or equal to a predetermined value. If YES, perform 0. If NO, perform [phase]. @ gives a common play symbol, registers it, and simplifies the table.
For the rectangular table in the upper part of Figure 6, a common identifier such as "error processing" is given to processes for which the degree of similarity is determined to be greater than a predetermined threshold using @, and the table is simplified as shown in Figure 10. do. This common part is registered as a similar pattern in the semantic information, and is registered as shown in Figure B in order to simplify the description of other processes. [Phase] assigns an identifier to each process, enters the identifier in the table, and describes the process as a supplementary description. By the above processing from @ to [phase], the description of the processing in the table in the rectangular column at the top of Figure 6 is simplified, and the table (specification in tabular form) shown in Figure 10 is created. Next, each of Figures 2 to 10 will be explained. FIG. 2 shows an overall configuration diagram of the present invention. In FIG. 2, a program 11 is a program for creating a tabular specification, and is, for example, the program shown in FIG. 3. The syntactic analysis unit 12 takes in the program 11, parses it, and converts it into an intermediate representation. For example, it converts the program shown in FIG. 3 into the intermediate representation shown in FIG. The intermediate representation 13 is an intermediate representation converted by the syntax M capture unit 12. This intermediate representation 13 is stored as a file. The converter 14 converts the intermediate representation 13 into a tabular specification 17. The semantic information 15 is semantic information regarding the data name,
For example, as shown in Figure 7, the data name “KOzANO(J
" is the information that means the account number determination code. The similar pattern detection unit 16 detects whether or not the processes are similar (■, 6 YES/No in FIG. 1). Next, The operation will be explained. (1) The syntactic analysis unit 12 sequentially takes in, for example, the program shown in FIG. For example, the intermediate expression shown in Figure 4 is taken in sentence by sentence, the similar pattern detection unit 16 detects similar processing while referring to the semantic information Ill5, and the branched sentence is converted into a tabular specification (Figure 1O). Figure 3 shows an example input program.
1i by L! This is a part of the input check program for bank deposit processing. FIG. 4 shows an example of an intermediate representation generated by the syntactic analysis of the present invention. This is the third
This is an example of an intermediate representation generated by parsing the illustrated example of a program. Part of the relationship between the program in Figure 3 and the intermediate representation in Figure 4 is shown below.

! 11 EVALtlATE Upper type Branching statement (2) HEN O CONTINtlE↓ Condition 1.1 KOZAIfOC[=O Processing 1-1
1, (3) KaHE? 1 1? IOVE 10001
To ERRI'! SGNOPIOVE l[OZA
IIIoID No To EI? l? InNOperf
orm ERROR-SECTION↓ Condition 1.2 KOZANOCK=1 Processing 1.2 Statement 1.2.1 Type Assignment statement ERRfISGNO:=lOOIL Statement 1.2.2 Type Assignment statement Efl? IDNO:=KOZllNOInNO statement 1.2
．． 3 types Procedure call statement PEI? FOl? PrER20R-SECτION(4
) The same applies hereafter, so it will be omitted. FIG. 5 shows an example of intermediate expression obtained by converting the branch statement of the present invention. This is the branching statement enclosed in a rectangle in Figure 4, one condition, one variable name, and one variable name! Regarding l, Figure 1 [stock] YES, OYES, ■
Since it corresponds to YES, it is converted and replaced as shown in the rectangular area in Figure 5. Figure 6 shows an example of an intermediate representation in which branching statements of the present invention are merged. This is the variable ([QZILNOC[, ZENZAffC[,
RII? 1rLlcI [) 6 It is determined that the determination code is the same species by Noriharu (Fig. 1 OYES),
The branch sentences of sentence 1, sentence 2, and sentence 3 are merged as shown. FIG. 7 shows the semantic information (Japanese name) used for conversion according to the present invention. This represents the meaning information in Japanese of variable names used in branching statements, etc. Here, the same type of judgment code in Japanese is shown in the diagram K.
OZANOCK, ZENZANCK, flflTU
(There is J. Fig. 8 shows an example of similar pattern registration according to the present invention. This is an identifier "
Regarding error handling, it is registered to be used in similar processing as shown in the diagram. FIG. 9 shows an example of similarity calculation according to the present invention. For example, the similarity degree of process 1.2 and process 1.3 in FIG. 6 is expressed as shown in the figure, and is determined as shown in the figure. Below processing 1, processing 2
This section provides a general explanation of the calculation of similarity between 0. 1! The degree is initially set to 0.

Compare the first sentence of process l and the first sentence of process 2 in order. Calculate the similarity between sentences as follows. If the sentence types match, add 1 to the similarity. Next, the similarity is calculated for each sentence type as follows.

Assignment statement: If the variables on the left side match, add 2 to the variable. Add. If the variables on the right side are of the same type, then Add 1 to the variable. If the expressions on the right side match, add 2. Otherwise, if the right-hand side is only a variable, then If the variables are of the same type, add l. Procedure call statement: With arguments: If the procedure names match, add 2. Furthermore, (number of matching arguments x 2 + homogeneous number of arguments) ÷ number of arguments.

If there are no arguments: If the procedure names match, add 4.

Branching statement: If the conditions match, add 4. Otherwise, if the condition is in the form of one variable value, and the variable matches, add 2. Otherwise, in the condition variable = value format, if the variables are of the same type, add 1. A repeating statement that calculates the similarity for each process after branching and adds it: If the conditions match, add 4. Otherwise, in the condition variable = value format, if the variable matches, add 2. Otherwise, if the condition variable = value format and the variables are of the same type! Add. About repeated individual processing! ! Calculate the degree and add it. If there is a sentence with a degree of similarity of 2 or more, and it is the i-th sentence in Process 2, then the j-th sentence in Process 1 and the i+j-1 sentence in Process 2 are compared. If there is no sentence with a degree of 111th rank of 2 or more, the next sentence in process L and each sentence in process 2 are compared in order, and the same thing as above is performed. After calculating the overall similarity, it is divided into Process 1 and Process 2.
Divide by the sum of the number of 0 sentences and use it as the degree of similarity between the two processes. FIG. 10 shows an example of a specification sheet output by the present invention.
This is done by calculating the similarity in Figure 9 for the processing of the branch sentences merged using the judgment code in the rectangular column in the upper part of Figure 6, and replacing the processing with the similarity with a common identifier "error processing" for processes whose similarity is higher than a predetermined threshold. This is a simplified description.
For processes whose degree of similarity is less than the predetermined darkness value, an identifier is assigned to each process and a supplementary description is written as shown in the figure. [Effects of the Invention] As explained above, according to the present invention, branch statements that meet a predetermined condition (for example, all variables = value, all variable names are the same) are extracted from a plurality of branch statements extracted from a program, and these Since we have adopted a configuration that creates a tabular business report based on the similarity between the processing of branch statements, it is possible to automatically generate a highly abstract and easy-to-understand specification from a program, making maintenance easier. It is possible to improve work efficiency.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram of the configuration and operation of one embodiment of the present invention, FIG. 2 is an overall configuration diagram of the present invention, FIG. 3 is an example of an input program, and FIG.
The figure shows an example of an intermediate expression generated by the syntactic analysis of the present invention, Figure 5 is an example of an intermediate expression generated by converting the branch statement of the present invention, and Figure 6 shows an example of an intermediate expression generated by the syntax analysis of the present invention.
The figure shows an example of intermediate representation by merging the branch sentences of the present invention, Figure 7 shows the semantic information (Japanese name) used for the conversion according to the present invention, and Figure 8
The figure shows an example of similar pattern registration according to the present invention, Fig. 9 shows an example of similarity calculation according to the present invention, and Fig. 10 shows an example of a specification outputted according to the present invention. In the figure, 1 is a branch sentence extraction unit, 2 is a similarity calculation unit, 3 is a table creation unit, 11 is a program, l2 is a syntax analysis unit, 13 is an intermediate representation, l4 is a conversion unit, 15 is semantic information, and 16 is A similar pattern detection unit, 17 represents a specification. Overall configuration diagram of the present invention Figure 2 Example of a specification output by the present invention Part 10 Part subjected to figure-head conversion Example of an intermediate expression in which the branching statement of the present invention is converted Figure 5 Merging the branching statement of the present invention Figure 6 Example of intermediate representation

Claims (1)

[Claims] In a program specification automatic creation method that automatically generates a specification from a program, for a plurality of branch statements extracted from a program, a condition in the branch statement is set to a predetermined condition (for example, all variables = value, variable name are all the same).
), and a similarity method that calculates the similarity of processing between each sentence by grouping together branch sentences with the same type of variable name (for example, the same type in the category of judgment code) from among the branch sentences extracted by this branch sentence extraction unit (1). a table creation unit (3) that creates a table structure specification in which processes are replaced with identifiers (for example, error handling) based on the similarity calculated by the similarity calculation unit (2); An automatic program specification creation method, characterized in that the table creation unit (3) is configured to output the specifications created by the table creation unit (3).