JP2010049585A - Automatic program generating apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an automatic program generating apparatus which eliminates the need for a user to explicitly set a rule change repetitively and also enables an automatic flexible rule change to automatically generate a program more smoothly and efficiently. <P>SOLUTION: The automatic program generating apparatus has a learning means for learning a user's rule change and rule addition and automatically achieves the rule change from the learned result so that the user enables the rule change without setting the rule change explicitly. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an automatic program generation device that supports program generation from a software model description.

  In recent years, in the development of large-scale and complicated software, the product has become highly functional, so the software has become complicated, and an increase in development man-hours has become a problem.

Therefore, by designing the software design with a model such as UML (Unified Modeling Language), software reusability is improved, and coding code is automatically generated from the model to improve software development efficiency and ensure quality. There is an automatic program generation device that performs (For example, Patent Document 1)
JP 2007-280138 A

  In order for a user to convert from a model to a program code, it is necessary to define a rule as to what model is converted into what code.

  In a conventional automatic program generation device, a rule that is initially set in the automatic program generation device is used as a rule for converting a model into a program code. When using a rule different from the initial setting, it is necessary for the user to set in detail each rule to be used for the automatic program generation device. In addition, when it is desired to newly set a rule that is not prepared by the automatic program generation device, it is necessary to explicitly perform each time a new rule is registered, which is troublesome.

  Therefore, the present invention has been made paying attention to the unresolved problems of the prior art, the rule change is automatically performed without the user being aware of the rule setting, and the user's trouble of rule setting The purpose is to provide an easy-to-use automatic program generator.

  In order to achieve the above object, in the present invention, the automatic program generation device learns the user's rule change and rule addition and automatically learns the user without changing the rule. It is possible to change the generation rule.

  Thus, the user can automatically change the rule without explicitly setting the rule change, and provides an automatic program generation device capable of automatically generating a program more smoothly and efficiently.

  According to the present invention, a user can automatically change a rule without having to explicitly change or add a rule every time, and an automatic program generation device capable of automatically generating a program more smoothly and efficiently can be provided. It becomes.

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

  A first embodiment of the present invention will be described.

  FIG. 1 shows the components of the automatic program generation device 100.

  The automatic program generation apparatus 100 includes an input unit 101 for a user to input a model and a code, a model generation unit 102 for generating a model by an input from the user, a code generation unit 103 for generating a code from the generated model, An output means 107 for displaying the model and code displayed to the user, a rule database (indicated as rule DB in FIG. 1) 105 storing rules used for code generation, and a rule to be applied are selected from the rule database 105 An application rule selection unit 104, a rule learning unit 106 that analyzes and learns the tendency of application rules and additional rules from user operations, and an output unit 107 that displays models and codes are included.

  Next, the hardware configuration of each function will be described.

  The input unit 101 includes a keyboard, buttons, and the like, and can be input by a user. In addition, an input from an external device such as a microphone or a camera may be received.

  The model creation means 102, the code generation means 103, the application rule selection means 104, the rule learning means 106, and the rule database 15 are recorded by a recording device. For example, it is composed of ROM (Read Only Memory), RAM (Random Access Memory), etc., and stores all data and software handled by the automatic program generation device. Usually, it is configured inside the housing, but it may be realized by a removable memory card or a storage device connected from the outside.

  The means for executing each function, such as the model creation means 102, the code generation means 103, the applied rule selection means 104, and the rule learning means 106, reads the program stored in the storage device and refers to it by the CPU. Operation, processing, and execution.

  The output means 107 is composed of, for example, a display device such as an LCD (Liquid Crystal Display) or an organic EL, and is generated by a character or operation input from the input means, a model created by the model creation means, or a code generation means. Displayed code.

  In this embodiment, the software development support apparatus is described assuming a PC (Personal Computer) having an input unit such as a mouse and a keyboard. However, if the program can be input and displayed, a mobile phone, PHS, etc. (Personal Handy-phone System), PDA (Personal Digital Assistant), etc., it does not matter.

  The operation will be described in detail.

  FIG. 2 is a flowchart of a process when a code is automatically generated from a model in this embodiment.

  When the process is started (S201), the input unit accepts an input from the user (S202). The model creation means creates a model by an operation from the input means by the user (S203). The created model and the result of the operation are displayed to the user by the output means (S204).

  After the model is created, if the user instructs automatic code generation from the input means (S205), the applied rule selection means is selected from the rule list in the rule database storing the rules that can be used by the automatic program generation device. Which rule is actually applied to generate the code is selected from the rule learning unit based on the analysis result and transmitted to the code generation unit (S206). The code generation unit converts the model created by the model creation unit into a code using the application rule acquired from the application rule selection unit, and generates a code (S207).

  The generated code is displayed to the user by the output means (S208), and the process ends (S209).

  FIG. 3 is a flowchart of processing in the rule learning means.

  When the processing is started (S301), the rule learning unit monitors the change of the application rule by the user from the input unit, and detects the change of the application rule by the user (S302). The rule learning means has a learning means database therein and accumulates detected changes in the database (S303). Then, it learns the rule changes accumulated in the database for learning means (S304), analyzes the tendency of the user's usage rules, and then decides which rule to apply when the user generates code from the model Then, the process is terminated (S306).

  When the application rule selection unit described in S206 of FIG. 2 determines which rule is actually applied from the rule list in the rule database to generate the code, the rule learning unit analyzes and determines The rule to be used is designated to the application rule selection means. Thereby, the application rule selection means can select the setting to be used next without an explicit setting from the user.

  FIG. 4 is an example in which the rule learning means has learned the rules.

  This is an example 401 in which a model called class A having an int type variable name is created by the model creation means.

  In the automatic program generation device that does not have rule learning means or before learning rules, in the initial state, if the user does not explicitly set the rule change, the code automatically generated by the code generation means is the definition of class A and the variable Assume that the code 402 is automatically generated using a rule that generates a definition of name, a getName function for acquiring a variable name, and a setName function for setting the variable name.

  In the automatic program generation device with rule learning means, when the user frequently performs an operation to select a rule that does not automatically generate the name acquisition function getName, the user next generates code from the model. The rule learning means learns that a rule that does not automatically generate getName should be taken, and the rule that does not automatically generate getName is selected by the application rule selecting means and specified to the code generating means. The code 403 that does not generate the code of the getName function even if the rule change is not a child is automatically generated.

  As shown in FIG. 5, the rule learning unit 106 includes a code correction detection unit 501 and a learning unit 502. The code correction detection unit 501 monitors the user's code correction to the code generated by the code generation unit, and if there is a code correction, extracts the code before the correction and the code after the correction, and corrects the code If the part is a part automatically generated by the code generation means, it is detected which rule in the rule database has been changed to which rule, and the rule is stored in the learning database. The learning means 502 uses a learning database as data for learning.

  FIG. 6 shows an example in which the rule learning means has learned this rule with this configuration. This is an example 601 in which class A and variable name are defined by the model creation means.

  At this time, the code automatically generated by the code generation means of the program automatic generation apparatus that does not have the rule learning means or the program automatic generation apparatus before learning is the class A definition, the definition of the variable name, and the getName for acquiring the variable name. It is assumed that a rule for generating a setName function for setting a function and a variable name is acquired from the application rule selecting unit and a code 602 is automatically generated.

  When the user deletes the code of the getName function for the automatically generated code, the code correction detection means detects that the correction has been performed, extracts the code difference after the code correction before the code correction, Recognizing that the getName function is a place where the code is automatically generated, the fact that the rule for automatically generating the getName function in the rule database has been changed to a rule that does not automatically generate is accumulated in the learning means database and learning is performed.

  After learning, the next time the user generates code from the model, the rule that does not automatically generate getName is selected by the application rule selection means, and the code for the getName function is automatically generated for specification to the code generation means. A code 603 that is not to be generated is generated.

  Alternatively, the rule learning unit 106 includes a rule selection / extraction unit 701 and a learning unit 702 as shown in FIG. The rule selection / extraction means 701 monitors the application rules set by the user with respect to the application rule selection means of the automatic program generation device when automatically generating code from the model, acquires the applied rules, and learns In the database. The learning means 702 uses a learning database as data for learning.

  FIG. 8 shows an example in which the rule learning means learns the rule with this configuration. This is an example 801 in which the class A and the variable name are defined by the model creation means.

  At this time, the code automatically generated by the code generation means of the program automatic generation apparatus that does not have the rule learning means or the program automatic generation apparatus before learning is the class A definition, the definition of the variable name, and the getName for acquiring the variable name. It is assumed that a code 802 is automatically generated by acquiring a rule for generating a setName function for setting a function and a variable name from the application rule selection unit.

  If the user explicitly selects a rule that does not automatically generate the getName function for the application rule selection unit of the automatic program generation device before the code generation unit automatically generates the code, the code generation unit selects the application rule. The code of getName function is not automatically generated.

  If the rule learning means learns that the user frequently selects a rule that does not automatically generate the getName function, then after learning, the applied rule selection means automatically generates getName without the user explicitly changing the rule. Since the rule to be not selected is selected and specified in the code generation means, the code 803 that does not generate the code of the getName function is automatically generated.

  Of course, data may be acquired in any way as long as it is possible to learn application rules of the user, even if the configuration and method described above are not used.

  In this embodiment, the rule learning means learns and selects the next rule from the rule list in the rule database prepared by the automatic program generation device. For whatever reason, the rule is selected. You can select the rule depending on the content (Figure 9).

  When the process is started (S901), the rule learning unit monitors and detects a change in the applied rule by the user from the input unit (S902). Next, the detected changes are accumulated in the learning database (S903).

  Then, the rule change accumulated in the learning means database is learned, and the change factor is analyzed and learned from the tendency of the change rule of the user (S904). Next, when the user generates a code from the model, which rule is applied for what reason is determined (S905), and the process is terminated (S906).

  For example, when the user changes the rule so far, if the same rule change is made for all the items in the same file, the user's rule change wants to change all the rules in the file. Learning a certain thing and selecting a rule reflecting the rule change in the application range.

  Also, the rule learning means automatically adds a rule when the user adds a rule that is not in the rule list of the rule database of the automatic program generation device, includes it as a learning target, and selects a newly added rule as an application rule. It does not matter (Figure 10).

  When the process is started (S1001), the rule learning unit monitors the change of the application rule by the user from the input unit, and detects the addition of the application rule by the user (S1002). The rule learning means has a database for learning means inside, and accumulates the detected new rules in the database (S1003).

  Also, the detected new rule is registered in the rule database (S1004). Then, learn the rule changes accumulated in the database for learning means (S1005), analyze the tendency of the user's usage rules, and then select which rule to apply when the user generates code from the model (S1006), and the process ends (S1007).

  FIG. 11 is an example in which rules are learned.

  This is an example 1101 in which a model that class A has three objects of class B is created by the model creation means.

  In the program automatic generation device that does not have rule learning means or before learning the rules, in the initial state, if the user does not explicitly set the rule change, the code automatically generated by the code generation means is the class A definition, 1102 suppose that three lines of code for generating the substance of class B are generated in class A, one line.

  In the automatic program generation device with rule learning means, when the user generates the same class multiple times by one class, the class generation by for statement is performed without writing the class generation code line by line. If the user learns that the code is frequently modified by applying rules that are not in the rule database, the next time the user generates code from the model, the code is generated 1103 using the for statement.

  In this embodiment, the automatic program generation apparatus 100 includes an input unit 101, a model generation unit 102, a code generation unit 103, an output unit 107, a rule database 105, an applied rule selection unit 104, a rule learning unit 106, and an output unit 107. However, if the rule learning unit 106 learns from the rules in the rule database 105 and the application rule selection unit 104 selects a rule, the input unit 101, the model creation unit 102, and the code generation The means equivalent to the means 103 and the output means 107 may have any configuration.

  Further, a model input from the outside by the input unit may be used without the model creation unit.

  In a conventional automatic program generation device, a rule that is initially set in the automatic program generation device is used as a rule for converting a model into a program code. When a rule different from the initial setting is used, it is necessary for the user to make detailed settings for each type of rule for the automatic program generation device.

  However, according to the above-described embodiment of the present invention, the user can automatically and flexibly change the rule without explicitly setting the rule change repeatedly, and can automatically generate the program more smoothly and efficiently. A generation device can be provided.

1 is a configuration diagram of an automatic program generation device according to a first embodiment of the present invention. FIG. 4 is a diagram illustrating a flowchart of processing when converting from a model to a code in the automatic program generation device according to the first exemplary embodiment of the present invention. FIG. 5 is a diagram showing a flowchart of processing of rule learning means of the automatic program generation device in the first exemplary embodiment of the present invention. FIG. 5 is a diagram showing a learning example of rule learning means of the automatic program generation device in the first exemplary embodiment of the present invention. FIG. 3 is a diagram showing a configuration diagram of rule learning means of the automatic program generation device in the first exemplary embodiment of the present invention. FIG. 5 is a diagram showing a learning example of rule learning means of the automatic program generation device in the first exemplary embodiment of the present invention. FIG. 3 is a diagram showing a configuration diagram of rule learning means of the automatic program generation device in the first exemplary embodiment of the present invention. FIG. 5 is a diagram showing a learning example of rule learning means of the automatic program generation device in the first exemplary embodiment of the present invention. FIG. 5 is a diagram showing a flowchart of processing of rule learning means of the automatic program generation device in the first exemplary embodiment of the present invention. FIG. 5 is a diagram showing a flowchart of processing of rule learning means of the automatic program generation device in the first exemplary embodiment of the present invention. FIG. 5 is a diagram showing a learning example of rule learning means of the automatic program generation device in the first exemplary embodiment of the present invention.

Explanation of symbols

100: automatic program generation apparatus 101: input means 102: model creation means 103: code generation means 104: applied rule selection means 105: rule database 106: rule learning means 107: output means 401: model example 402: generated code 403: learning Generated code 501 after: Code correction detecting means 502: Learning means 601: Model example 602: Generated code 603: Generated generated code 701: Rule selection extracting means 702: Learning means 801: Model example 802: Generated code 803: Learning Generated code 1001: Model example 1002: Generated code 1003: Generated code after learning

Claims (8)

An input means for inputting a model or source code;
Code generation means for generating source code from the model;
A rule database in which rules referred to by the code generation means are stored;
Application rule selection means for selecting a rule to be applied by the code generation means from the rule database;
In an automatic program generation device having
A rule learning means for detecting and learning the change of the applied rule from the creation and change of the model is provided,
The program automatic generation apparatus, wherein the rule learning unit sets a learning result in the application rule selection unit. In the automatic program generation device according to claim 1,
A program generation apparatus comprising model creation means for creating and changing a model by input to the input means. The automatic program generation device according to claim 1 or 2,
An automatic program generation apparatus comprising an output means for displaying a model created by the model creation means or a source code generated by the code generation means. In the automatic program generation device according to claims 1 to 3,
The rule learning means is
A learning database that stores the learned content;
Code modification detection means for monitoring code modifications and detecting applied rules changed from code modifications;
Learning means for learning the application rule,
The code correction detection unit detects the changed application rule from the rule database when the code generation unit corrects the code generated by the code generation unit;
The automatic learning apparatus according to claim 1, wherein the learning means stores a learning result from the detected application rule in the learning database. In the automatic program generation device according to claims 1 to 3,
The rule learning means is
A learning database that stores the learned content;
Rule selection extraction means for extracting application rules set by the application rule selection means;
Learning means for learning the application rule,
The rule selection extraction unit monitors the application rule set in the application rule selection unit, extracts the set application rule,
The automatic learning apparatus characterized in that the learning means stores a learning result from the extracted application rule in the learning database. The automatic program generation device according to any one of claims 1 to 5,
The rule learning means is
An automatic program generation device characterized by learning and selecting a changed application rule from a rule list in the rule database. The automatic program generation device according to any one of claims 1 to 5,
The rule learning means is
An automatic program generation device characterized by detecting a changed rule from a rule list of the rule database and learning a changed factor. The automatic program generation device according to any one of claims 1 to 5,
The rule learning means is
When a user applies a rule that is not in the rule list of the rule database, the application rule is added to the rule database,
A program automatic generation device characterized by learning the added rule.

Images