JP2012212281A - Range of influence setting device for program and range of influence specification method using the same and extraction method for program with ripple of influence - Google Patents

Abstract

PROBLEM TO BE SOLVED: To mechanically set the range of the ripple of influence without extracting any class or method with no ripple of influence, and to achieve the improvement of the investigating accuracy of the ripple of influence and the improvement of working efficiency in the case of modifying a largely scaled system.SOLUTION: The source code of a program is subject to syntax analysis so as to be analyzed to an abstract syntax tree, and instruction sentences forming the abstract syntax tree are retrieved from instruction terms, and the connection of the instruction sentences is formed in accordance with the dependency of the instruction sentences from the retrieved instruction sentences, and a dependency graph is formed on the basis of the instruction sentences about the source code and the connection of the instruction sentences so that the range of the ripple of influence having dependency can be set.

Description

  The present invention relates to an influence range setting device for a program, an influence range specifying method using the same, and a program extraction method having influence influence.

  When renovating a large-scale system, it is necessary to improve the accuracy and work efficiency of the investigation to influence other programs as the program is renovated.

  Conventionally, for the influence spread investigation, the influence ripple identification method centering on "method call relation" has been used.

  Patent Document 1 describes that items to be compared are defined and compared for each hierarchical structure of modules, files, functions, and source codes, which are logical structures in design.

  In Patent Document 2, conventionally, an analysis rule is defined, and the impact analysis of the change is performed based on the rule. In this method, it takes time to create the rule. It is described to store an influence analysis chart in which all possible verification methods are described.

JP 2006-277127 A JP 2009-157505 A

  In the conventional method of determining the influence spread by the method call relationship, there is a problem that a large number of classes / methods that are not affected are extracted, and it is necessary to specify a hierarchy that traces the call relationship.

  Further, the method for storing the influence analysis chart has a problem that it takes a lot of time to construct the influence analysis chart.

  In view of this point, the present invention does not extract classes and methods that do not have an influence spread, and can set the range of the influence spread mechanically. The purpose is to improve the work efficiency.

In the present invention, the source code of a program is parsed and analyzed into an abstract syntax tree, and the statements that form the abstract syntax tree are retrieved,
A connection of command statements is formed from the retrieved command statements according to the dependency relationship of the command statements,
Provided is an influence spread range setting device for a program characterized in that a dependency relation graph based on a sentence and a connection between instruction sentences for a source code is formed, and an influence spread range in dependency relation is set.

The present invention parses the program source code into an abstract syntax tree,
The statements that form the abstract syntax tree are retrieved,
A connection of command statements is formed from the retrieved command statements according to the dependency relationship of the command statements,
For multiple source codes, a sequence of retrieved statements is formed,
When the searched command statements of a plurality of source codes are the same, the command statement is a command statement shared by each source code,
A dependency graph including a combination of the shared instruction sentence and the connection of the instruction sentence including a combination of the instruction sentence and the connection of the instruction sentence for a plurality of source codes is formed and shared, and the influence spreading range in the dependency relation An influence spread range setting device for a program, characterized in that is set.

The present invention also provides a method for identifying an influence range by an influence spread range setting device of the program,
When a base statement is specified on the dependency graph, the influence spread range is specified by tracing the dependency from the base statement. Provide a specific method.

The present invention also provides a method for extracting a program having an influence spread by the influence spread range setting device of the program,
When a command statement as a base point is specified on the dependency relationship graph, a program having a connection between the command statement and the command statement is traced as a program having an influence ripple effect by tracing the dependency relationship from the command statement as the base point. The present invention provides a program extraction method having an influence effect, characterized by extracting.

  In the present invention, source code is input, but here it is used in a concept including byte code. Byte code means an executable program described by instructions defined so as not to depend on a specific OS or hardware, and can be executed by a computer and source code that is a design drawing written by a human. It is in the middle of native code and can be handled as source code here.

  According to the present invention, as described above, the dependency relationship between the statements is given by the syntax analysis of the source code, and the influence spread range is set and specified, so that the class / method having no influence is extracted. Therefore, the range of influence spread can be set mechanically from the command statement, and when revising a large-scale system, the accuracy of the impact investigation can be improved and the work efficiency can be improved.

The figure which shows the structure of the Example of this invention with a block. The figure which shows the structure of the engine employ | adopted by a present Example. The figure which shows the flow of a process of a present Example. The figure which shows the example of a syntax analysis. The figure which shows the example of assignment | providing of the dependency relation between command statements. The figure which shows the dependence relationship graph by the setting of an influence influence range. The figure which added the class and method relation to FIG. The figure which shows the specific example of the connection of the command sentence shown in FIG. 6, and a command sentence. The figure which shows the example of the influence propagation investigation which followed the dependence relationship. The figure which shows the result of an influence spread range. The figure which shows the result of the influence spread by a prior art example. The figure which shows the result of the influence spread by a prior art example. The figure which shows the result of the influence spread by a prior art example. The figure which shows the result of the influence spread by a present Example.

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

FIG. 1 is a block diagram showing the configuration of a program influence range setting apparatus according to an embodiment of the present invention.
In FIG. 1, a program influence range setting device 100 includes an input unit 1, an arithmetic processing unit 2, an output unit 3, and a database 4.

  A program 5 to be repaired and a command statement 6 to be a base point are input to the input means 1 from the outside in the case of a large-scale system.

  The arithmetic processing means 2 includes the following processing means (functions).

・ Syntax analysis of source code (ie grammar analysis)
・ Extract syntax tree analysis ・ Command statement search ・ Formation of command statement connection ・ Extraction of shared command statement ・ Dependence by command statement ・ Formation of graph ・ Setting of influence range ・ Identification of influence range Output means 3
-Tracing the dependency relationship-The set influence range-The specified influence range is output and displayed on the screen 8 of the screen display device 7.

  The database 4 stores a dependency relationship graph (that is, data) based on a command sentence generated by arithmetic processing by the arithmetic processing means 2.

The arithmetic processing means 2 will be described.
The source code syntax analysis means parses the source code of the program.
The abstract syntax tree analyzing means analyzes the abstract syntax tree as a hierarchical structure based on the syntax analysis.

  The imperative sentence search means searches for an imperative sentence that forms an abstract syntax tree. Dependencies are analyzed for all nodes (lines, command statements) expressed in the abstract syntax tree, and command statements are searched.

  The command statement connection creating means forms a command statement link from the retrieved command statements according to the dependency relationship of the command statements. A sequence of statements is formed for a plurality of source codes.

  The shared command sentence extraction means extracts the command text as a command text shared by the respective source codes when they are the same.

  Dependency graph formation means by command statements is a graph of command dependency graphs for multiple source codes including combinations of shared command statements and command statements connected by dependency relationships with other programs. Form.

  The influence spread range setting means sets the influence spread range in the dependency relationship based on the dependency relationship graph.

  The influence influence range specifying means specifies the influence influence range based on the set influence influence range.

FIG. 2 shows the configuration of an engine using this embodiment.
The engine is composed of a language dependent part and a language independent part. (1) For syntax analysis, various plug-ins, for example, an A analysis plug-in, a B analysis plug-in, and a C analysis plug-in are used. One plug-in may be used.

  As shown in FIG. 2, syntax analysis is performed, the extracted syntax tree is analyzed, and (2) dependency relationship analysis is performed. A dependency is formed by this analysis, and then (3) an influence spread analysis is performed. This analysis generates an influence ripple result.

FIG. 3 shows the flow of processing.
(1) Grammar analysis is performed on the source code (S1). You may make it complement the type information etc. which are insufficient only with a source code with supplement information, such as a byte code. Here, “complement” means supplementary information for supplementing information that is insufficient only by the source code. For example, it refers to class and jar file information in Java (registered trademark) language.
(1) An abstract syntax tree is formed based on syntax analysis (S2), and (2) dependency analysis is performed (S3). Based on the analysis result, a dependency relationship is formed (S4), and (3) influence spread analysis is performed based on the base point of influence (S5). An influence spread result is generated based on the analysis result (S6).

  FIG. 4 shows that the syntax of the source code is analyzed by the source code syntax analysis means and the abstract syntax tree analysis means, and the abstract syntax tree is analyzed and formed. Here, in order to approach the pursuit with human eyes, the meaning of each statement in the source code is analyzed, and the result is converted into an abstract syntax tree. Analyzes the abstract syntax tree and retrieves imperative sentences from imperative terms.

  FIG. 5 shows analyzing the connection between statements.

  For example, a method call statement leads to a method declaration at the call destination. In this way, not only the method call but also the connection is given to all command statements such as the use and declaration of variables. This is called dependency.

  Dependency is used as a path to follow the influence spread.

FIG. 6 shows setting of the influence spread range and identification of the influence spread range.
FIG. 6 shows a series of command statement connections in the source code starting from the command statement 1 and the command statement 2.

  The source code starting from the command statement 1 and the source code starting from the command statement 2 indicate that the command statement 5 is the same as the command statement shared by the source code if the command statement 5 is the same.

  The command statement 5 is connected to the command statement 7 and the command statement 8, and the command statement 7 is connected to the command statement 9.

  As shown in the figure, the imperative sentence 5, the imperative sentence 7, and the imperative sentence 9 form a connection of the imperative sentences shared by the respective source codes in relation to other programs.

  A dependency graph based on the command statements for a plurality of source codes including a combination of the shared command statement and the command statement connected by the dependency relationship is formed, and the influence spread range is set. The influence spread is investigated by following the dependency between statements. Now, the source code for the statement 2 in the source code that is the starting point is used as the source code of the base point. The influence spread range set by specifying the command statement 2 as the base point and following the dependency relationship of the command statement as the base point is specified.

FIG. 7 shows the relationship with classes and methods in the state shown in FIG.
In FIG. 7, the command statement belongs to a class or a method.

  Therefore, the class or method to which the user belongs can be specified by tracing the statement. For example, the statement 1 belongs to the method α of class A.

FIG. 8 shows the connection between the command statement shown in FIG. 6 and the command statement as a specific example.
In FIG. 8, the influence propagation route is shown for each instruction sentence of the source code, and the instruction sentence subject to the influence influence is automatically extracted from the instruction terms.

FIG. 9 shows an example of influence spread investigation that traces the dependency relationship.
FIG. 9 shows an example of specifying the influence range of the name property defined in the UseInfo class. In this example, the range that needs to be modified or tested as the value of the name property changes is specified. For example, this is a case where a value conventionally expressed with 3 digits is corrected to be expressed with 10 digits.

The description of FIG. 9 is as follows.
Investigate the impact on the change of the name property defined in the field of UserInfo class.

[Influence of UserInfo class]
1. The name property is used in the getName () and setName (String) methods.
Since the name property is returned by the return statement of the getName () method, the influence is affected to the caller of the getName () method.
⇒To makeKey (Date, String) method of Utility class.
2. In the setName (String) method, the value of the argument of the setName (String) method is overwritten in the name property, so there is no effect on the change of the name property, and no further ripple occurs.

[Effect of the Utility class]
1. Since it affects the value of userInfo.getName (), it affects the string variable that stores the calculation result using that value.
2. Since the string variable is passed to the encryption.encrypt (String) method, the effect affects the encryptedString variable and the encrypted (String) method.
⇒To encryption (String) method of Encryption class.
3. Since the encryptedString variable becomes the return value of the makeKey (Date, String) method, the influence affects the caller of the makeKey (Date, String) method.
⇒To logic () method of Logic class.

[Influence of the Encryption class]
1. Since it affects the string argument of the encrypt (String) method, it affects the argument of the statement "process (string)" using that value.
2. Since it affects the string variable of the argument of process (String) method, it affects the "return string" statement that uses that value. Since the return statement is affected, the return value of the process (String) method is affected, so the influence is spread to the caller of the process (String) method.
⇒To the encryption (String) method (3) of the Encryption class.
3. Since the encryptedString variable is returned by the return statement, the influence is spread to the caller of the encrypt (String) method.
⇒To makeKey (Date, String) method of Utility class.

[Influence of Logic class]
Since it affects the return value of makeKey (Date, String) of Utility class, it affects the key variable that sets the value.
Since the key variable is not used after that, the influence spread ends here.

  FIG. 10 shows the result of the influence influence range investigation according to the present embodiment by comparison with the influence influence investigation by the calling relation. In the influence spread investigation method based on the call relation, a large number of classes / methods having no influence are extracted, and it is necessary for a human to judge and narrow down how far to go. On the other hand, in the present embodiment, the result of the influence spread investigation is a necessary and sufficient result, and therefore it is not necessary to determine how far to go.

According to the present embodiment, by parsing the grammar for each imperative sentence,
・ Ability to analyze dependencies between statements ・ A feature of analyzing the flow of processing can be obtained. As a result, it is possible to trace the imperative statement, and for example, it is possible to solve the problem of the influence spread investigation centering on “call relation”. Then, as shown in FIGS. 6 and 7, the end point of the influence can be mechanically determined. Even in the case of a spaghetti code, a necessary command statement relationship can be extracted.

  FIG. 11 to FIG. 14 specifically show the comparison of the result of the influence spread range investigation according to this embodiment with the influence spread investigation by the calling relationship.

  FIG. 11 to FIG. 13 show the results of the influence spread investigation by the conventional call relationship. Details are described in these figures.

  FIG. 14 shows the results of an influence spread investigation according to this example.

  Therefore, according to the present embodiment, a large number of unaffected classes and methods are not extracted from these comparisons, and the influence spillover range is specified by following the dependency relationship between the imperative sentence and imperative sentence. Can do.

  In this way, this embodiment can improve the system of the influence spread investigation that is performed at the time of renovation of a large-scale system, and can improve the work efficiency.

  DESCRIPTION OF SYMBOLS 1 ... Input means, 2 ... Arithmetic processing means, 3 ... Output means, 4 ... Database, 5 ... Program, 6 ... Command sentence used as a base point, 7 ... Screen display apparatus, 8 ... Screen, 100 ... Setting of influence influence range of program apparatus.

Claims (4)

The program source code is parsed and analyzed into an abstract syntax tree, and the statements that form the abstract syntax tree are retrieved,
A connection of command statements is formed from the retrieved command statements according to the dependency relationship of the command statements,
An influence spread range setting device for a program characterized in that a dependency relation graph based on a statement and a connection between instruction sentences for a source code is formed, and an influence spread range in the dependency relation is set. The program source code is parsed into an abstract syntax tree,
The imperative sentence forming the abstract syntax tree is searched from imperative sentence terms,
A connection of command statements is formed from the retrieved command statements according to the dependency relationship of the command statements,
For multiple source codes, a sequence of retrieved statements is formed,
When the searched command statements of a plurality of source codes are the same, the command statement is a command statement shared by each source code,
A dependency graph including a combination of the shared instruction sentence and the connection of the instruction sentence including a combination of the instruction sentence and the connection of the instruction sentence for a plurality of source codes is formed and shared, and the influence spreading range in the dependency relation An influence spread range setting device for a program, characterized in that is set. In the method for identifying the influence range by the influence influence range setting device of the program according to claim 1 or 2,
When a base statement is specified on the dependency graph, the influence spread range is specified by tracing the dependency from the base statement. Identification method. In the extraction method of the program with the influence influence by the influence influence range setting apparatus of the program described in Claim 1 or 2,
When a command statement as a base point is specified on the dependency relationship graph, a program having a connection between the command statement and the command statement is traced as a program having an influence ripple effect by tracing the dependency relationship from the command statement as the base point. A program extraction method with influence influence characterized by extracting.

Images