JP6906155B2 - Automatic software generator capable of reverse engineering - Google Patents

Description

The present invention relates to reverse engineering of software engineering and automatic software generation, and more particularly to automatic software generation for reusing software.

In the past, reverse engineering was "extracting the reusable source code of the upstream process from the source code of the downstream process of software development", and automatic software generation was "from the reusable design information of the upstream process of software development". It was completely different by "automatically generating the source code of the downstream process".

In reverse engineering, the source code of the downstream process is made into a parse tree (syntax tree), and the computer is made to perform analysis based on it. The parser that generates this parse tree is used every time the language environment of the source code changes. Every time I hand-coded the parser for that language as a program for a particular language environment. Therefore, there is a problem that the development cost of the parser becomes enormous when trying to support multiple languages.

In software automatic generation, the source code of the target language environment is automatically generated from the design information of the upstream process, but every time the language environment is different, the generator of that language is used as the program of a specific language environment. It was created by hand coding as. Therefore, there is a problem that the development cost of the generator becomes enormous when trying to support multiple languages.

In reverse engineering, there is no established method for extracting and using reusable source code, and only humans can understand it using support tools and reuse it manually.

With software automatic generation, semi-automatic generation such as skeleton generation was possible, but when trying to fully automatic generation, the data that was supposed to be abstract design information of the upstream process became a program in a specific programming language. Therefore, there was a problem that it was meaningless to generate it completely automatically.

Japanese Unexamined Patent Publication No. 2001-296996

Japanese Unexamined Patent Publication No. 2000-148501

An object of the present invention is to realize a graph database (a database based on bidirectional graph structure data, which is neither a relational DB nor an object-oriented DB nor an XML-DB) for storing reusable pieces of software, and has been separate in the past. Reverse engineering and automatic software generation are connected through this graph database through the following three tasks.

The first task is to realize the reverse engineering parser and software automatic generation generator, which used to create programs for each language, at the same time by simply giving a grammar definition file that describes the grammar. It is the realization of a device that can be used.

The second task is to convert the perspective tree to graph structure data so that the perspective tree can be handled by the graph database, and to convert the reverse, just by giving a tree graph conversion rule file that describes the same conversion rule. It is the realization of a device that will be realized at the same time.

The third problem is to provide means for performing various analyzes and various processes on the bidirectional graph structure data.

In various analyzes, for example, in the future, creating a call graph, creating a flowchart, creating a data flow graph, creating a state transition diagram, estimating the performance of SQL statement execution in advance, etc. It is to be able to do it.

In the future as an example in various processing,
Automatically put or delete or to Rukoto the log code for debugging,
Automatically remove unused parts of the program,
To provide a means to extract how to use the library ,
Automatically generate necessary and sufficient test specifications,
Etc. should be possible.

A feature of the present invention is that it comprises three main devices.

The first "general-purpose parser & general-purpose generator" device has a parser function that turns the source code of a language into a parse tree and the language by simply giving a "grammar definition file" that describes the context-free grammar of a certain language environment. Simultaneously realizes a generator function that automatically generates source code from the perspective tree of.

The second "for tree graph conversion" extended Prolog device is
A function to convert "Lisp binary tree perspective tree" that can be traced only in one direction into "environment-dependent model bidirectional graph structure data" that can be traced in both directions,
The ability to do the opposite conversion,
Just prepare the same "tree graph conversion rule file" and it will be realized at the same time.

The third "for graph data manipulation" extended Prolog device represents a program
"Environment-dependent model bidirectional graph structure data" and
Various analyzes and various processes are performed using a reusable program for "environment-independent standard model bidirectional graph structure data".

As for the overall flow, the "general-purpose parser & general-purpose generator" device is
Given a "grammar definition file" for a language and the source code for that language
Generate a "Lisp binary tree perspective tree" and
The "for tree graph conversion" extended Prolog device
With the "tree graph conversion rule file" of the language environment
Given that "Lisp binary tree perspective tree" ,
Converts a "Lisp binary tree perspective tree" that can be traced only in one direction into "environment-dependent model bidirectional graph structure data" that can be traced in both directions.

"For tree graph conversion" Converted with the extended Prolog device
"Environment-dependent model bidirectional graph structure data" depends on the language environment,
For "graph data manipulation" extended Prolog device,
Given the "standard model program file" and this "environment-dependent model bidirectional graph structure data" ,
Convert this to "environment-independent standard model bidirectional graph structure data".

In addition, the extended Prolog device for "graph data manipulation" is
For "environment-independent standard model bidirectional graph structure data"
Program analysis and processing are performed using "various analysis programs" and "various processing programs" prepared in advance.

The "for tree graph conversion" extended Prolog device
If you give "environment-independent standard model bidirectional graph structure data" after various processing and "tree graph conversion rule file for standard model" of any language,
Generate a "Lisp binary tree perspective tree" for that language.

The "general purpose parser & general purpose generator" device was generated by the "grammar definition file" and "tree graph conversion" extended Prolog device for that language.
If you give "Lisp Binary Perth Tree",
Automatically generate source code for that language.

Example 1 (Migration of an existing system to a different language environment)

In general, running computer software source code written in one language in a different language environment has been very time-consuming to port.

The reason is that when another person analyzes or modifies a system created manually by a person, it costs a huge amount of money in the phase of first manually understanding the system specifications, implementation, and actual operating conditions. If you had to read the source code and decipher it, the cost would be even higher.

However, if the "reverse engineerable software automatic generator" of the present invention matures and the various files shown in FIG. 1 of the invention are enriched, the transfer of computer software programs to different language environments can be done manually and at a high cost. You will be able to do it in a short period of time without multiplying.

Example 2 (Modification or expansion of a large-scale system)

It is common for software once created to be significantly modified or expanded in an existing system environment.

At this time, if the external specifications of the system are known but no one has implemented the program, when modifying or extending the program, it is necessary to understand the internal specifications and the source code before modifying or extending the program. difficult. At present, it takes a lot of time to understand this internal specification / source code.

If the "reverse engineering software that can automatically generate device" is mature,
The source code goes through "Lisp binary tree perspective tree" and "environment-dependent graph structure data",
Finally, if it can be made into "environment-independent model bidirectional graph structure data",
People no longer have to read and understand the internal specifications / source code directly.
Human work is the maintenance of various files shown in FIG. 1 of the invention.
"Various analysis programs" and "various machining programs" for analyzing / processing "environment-independent standard model bidirectional graph structure data" will be enhanced including addition.

If the various file groups shown in FIG. 1 of the invention, especially the "various analysis program files" group and the "various processing program files" group, are sufficiently enriched, the main software development work of a person is from "program understanding and coding" to "graph". The work content changes to a dialogue with the "database".

It is a block diagram which shows the structural example of the embodiment of the "automatic generator of software capable of reverse engineering" of this invention. This is an example of a "grammar definition file" for a grammar with addition, subtraction, multiplication, division, substitution, and control statements. This is an example of the source code of the file automatically generated by the “grammar definition file” in FIG. This is an example of a "grammar definition file" of XML simple grammar. This is an excerpt of the source code example of the file automatically generated by the "grammar definition file" in FIG. This is an excerpt from the example of "Tree graph conversion rule file". This is an example of expressing an "object-oriented program" with "environment-independent standard model bidirectional graph structure data". This is the first excerpt of the source code example of the "Standard Modeling Program File" in the Altair language. This is the second excerpt of the source code example of the "Standard Modeling Program File" in the Altair language.

Example 1 (Migration of an existing system to a different language environment)

A first embodiment of the reverse engineerable soft air automatic generator according to the present invention will be described in detail with reference to the accompanying drawings. References to FIGS. 1 to 9 show embodiments of the present invention.

FIG. 1 is a block diagram showing a configuration example of a reverse engineerable soft air automatic generation medium of the present embodiment.

First, in FIG. 1, the "grammar definition file" group (5), the "tree graph conversion rule" group (6), the "standard modeling program file" group (7) of the existing system language environment, and the migration destination system language environment If there is no "grammar definition file" group (5) and "standard model tree graph conversion rule file" group (10), create them. An example of each file group is shown later.

Next, the "source code file of the existing system language environment" group (4) of the existing system language environment is used in the "grammar definition file" group (5) of the existing system language environment and the "general-purpose parser & general-purpose generator" device (1). Convert to the internal data of "Lisp binary tree perspective tree". 2 and 4 show specific examples of the “grammar definition file”.

The grammar definition file consists of three parts: an indent / line feed / spacing description part when used as a generator, a context-free grammar syntax definition part, and a context-free grammar lexical definition part.

の基本として用いている。BNF is often used as a method of describing context-free grammars, but BNF has to describe repetitions recursively, which is not good for readability. In the grammar definition file, 0 or more

It is used as the basis of.

Next, the internal data of the "Lisp binary tree perspective tree" that can be traced only in one direction is used in the "tree graph conversion rule file" group (6) and the "tree graph conversion" extended prolog device (2) in the existing system environment. Convert to internal data of "environment-dependent model bidirectional graph structure data" that can be traced bidirectionally with. FIG. 6 shows an excerpt of an example of the “tree graph conversion rule file”.

A "tree graph conversion rule file" is a set of a set of a name of a grammar unit and a description of a conversion rule between a perspective tree of the grammar unit and graph structure data. Grammar units are hard to read and write with only parentheses, but grammar units are automatically generated from the grammar definition file, so you don't have to worry about it at all. The "tree graph conversion rule file" must be created if it does not exist, but the skeleton equivalent is automatically generated from the grammar definition file. In FIG. 6, the part with the description “[-/ □□ /-> ○○]” is the part where Prolog is extended so that graph data can be handled.

Next, the internal data of this "environment-dependent model bidirectional graph structure data" is used in the "standard modeling program file" group (7) of the existing system language environment and the "environment" for "graph data manipulation" extended prolog device (3). Convert to internal data of "independent standard model bidirectional graph structure data".
In Fig. 7, "a program that adds 1 to 100 and outputs the result"
An example expressed by "environment-independent standard model bidirectional graph structure data" is shown.
8 and 9 show excerpts from an example of a "standard model program file" in the language Altair.

Figure 7 shows a computer software program.
Is indicated by "environment-independent standard model interactive graph structure data",
The graph of the graph structure data is a directed graph,
Ark can be traced in both directions on a computer.
In the arc of graphs of computer software programs represented by bidirectional graph structure data,
There are two types of arcs, inclusion and association.
The graph of inclusion only is
It represents the syntax tree as a result of parsing the program.
association is
As a result of semantic analysis of the program after parsing the program,
It is an arc that connects the parts determined to be the same symbol (same class name / same variable name / same function name, etc.).

8 and 9 are examples of a graph data manipulation program given to the “graph data manipulation” extended Prolog device, but the extended portion of Prolog is mainly “before: = {...}, after: =. It is the part where {...} and common: = {...} are paired. " In the graph structure data, before: = {If the graph currently of interest matches the graph pattern written below, after: = {By creating / modifying the graph pattern written below. , Graph structure data can be created / modified. common: = {In the following, conditions can be written in logical predicates, enabling detailed creation and modification of graph structure data.

Next, the internal data of this "environment-independent standard model bidirectional graph structure data" is used in the "tree graph conversion rule for standard model" group (10) and the "tree graph conversion" extended program device of the migration destination system language environment. In (2), it is converted into the internal data of "Parse tree of Lisp binary tree". The method of writing the "tree graph conversion rule for standard model" group (10) is the same as that of the "tree graph conversion rule file" group (6).

Next, this "Lisp binary tree perspective tree" is used in the "grammar definition file" group (5) of the migration destination system language environment and the "general purpose parser & general purpose generator" device (1) to "target language environment" of the migration destination system language environment. Source code file group (11) is automatically generated.
FIG. 3 shows an example of the source code automatically generated by the “grammar definition file” of FIG.
FIG. 5 shows an example of the source code automatically generated by the “grammar definition file” of FIG.

Example 2 (Modification or expansion of a large-scale system)

A second embodiment of the reverse engineerable soft air automatic generator according to the present invention will be described in detail with reference to the accompanying drawings. The second embodiment of the present invention is almost the same as the first embodiment except for the parts shown and described.

FIG. 1 is a block diagram showing a configuration example of a reverse engineerable soft air automatic generation medium of the present embodiment.

First, in Fig. 1, the "grammar definition file" group (5), "tree graph conversion rule file" group (6), "standard modeling program file" group (7), and "tree graph conversion for standard model" of the existing system environment If there is no "rule file" group (10), create it. This is the same as that of the first embodiment of the invention, except that the “grammar definition file” of the target language environment is the same as the “grammar definition file” of the existing system language environment.

Next, the "source code file of the existing system language environment" group (4) of the existing system language environment is used in the "grammar definition file" group (5) of the existing system language environment and the "general-purpose parser & general-purpose generator" device (1). Convert to the internal data of "Lisp binary tree perspective tree". This is the same as the embodiment of the first embodiment of the invention.

Next, the internal data of the "Lisp binary tree perspective tree" that can be traced only in one direction is used in the "tree graph conversion rule file" group (6) and the "tree graph conversion" extended prolog device (2) in the existing system environment. Convert to internal data of "environment-dependent model bidirectional graph structure data" that can be traced bidirectionally with. This is the same as the embodiment of the first embodiment of the invention.

Next, the internal data of this "environment-dependent model bidirectional graph structure data" is used in the "standard modeling program file" group (7) of the existing system language environment and the "environment" for "graph data manipulation" extended prolog device (3). Convert to internal data of "independent standard model bidirectional graph structure data". This is the same as the embodiment of the first embodiment of the invention.

Next, the internal data of this "environment-independent standard model graph structure data" is inquired / inserted / deleted / updated by the "graph data manipulation" extended Prolog device (3) to modify the program. Since making this modification is similar to executing an RDB (Relational Data Base) SELECT statement / INSERT statement / DELETE statement / UPDATE statement, the "for graph data manipulation" extended Prolog device of FIG. 1 of the present invention can be used. Will be referred to as GDB (Graph Data Base).

Software modifications can be made in the programming language most familiar to engineers using the reverse engineerable software automatic generator of the present invention. If you are a technician who can do C ++, the engineer will automatically generate the "source code file of the existing system language environment" (4) written in C ++, and the technician will make the correction in C ++ and reverse engineer it. Make corrections by reflecting it in the internal data of "Environment-independent standard model bidirectional graph structure data". If you are a technician who can do Java, the engineer will automatically generate the "source code file of the existing system language environment" (4) written in Java, and the technician will make corrections in Java and reverse engineer it. Make corrections by reflecting it in the internal data of "Environment-independent standard model bidirectional graph structure data".

Next, the internal data of this modified "environment-independent standard model bidirectional graph structure data" is used in the "tree graph conversion rule for standard model" group (10) and the extended prolog device (2) for "tree graph conversion". Converts to the internal data of "Lisp binary tree perspective tree". This is the same as the embodiment of the first embodiment of the invention.

Next, the internal data of this "Lisp binary tree perspective tree" is used in the "grammar definition file" group (5) and the "general-purpose parser & general-purpose generator" device (1) in the "source file of the existing system language environment" group (4). Is automatically generated. This is the same as that of the first embodiment of the invention, except that the “grammar definition file” of the target language environment is the same as the “grammar definition file” of the existing system language environment.

Computer system software development sites have come to be called 3K (difficult, dirty, ugly) and are becoming less popular. Many people are forced to work overtime due to the approaching delivery date. Looking back on this situation, it seems that the work that can be done by the machine is still done manually, and the work itself is complicated. It can be expected that the productivity and quality of computer software at the computer system software development site can be significantly improved by making most of the parts manually performed by programmers into IT and automation by computers.

1 General-purpose parser & general-purpose generator device 2 Extended Prolog device for tree graph conversion 3 Extended Prolog device for graph data operation 4 Source code file group of existing system language environment 5 Grammar definition file group 6 Tree graph conversion rule file group 7 Standard modeling program File group 8 Various analysis program file group 9 Various processing program file group 10 Tree graph conversion rule file group for standard model 11 Source code file group of target language environment 12 Various analysis result group 13 Various product group 14 Graph database

Claims (4)

"Lisp binary tree perspective tree" that can be traced only in one direction, which is internal data, using a group of "existing system environment-dependent source code files" written in the language of "grammar definition file" that describes the context-free grammar of a specific language. And the parser function to generate
"Target system environment-dependent source code files" or "target system environment-dependent source code files" written in the language described in the grammar definition file using the "Lisp binary tree perspective tree" and "grammar definition file" that can be traced only in one direction, which is internal data. It consists of a generator function that generates "existing system environment-dependent source code files".
Just by giving the same "grammar definition file" for the parser function and the generator function,
A "general-purpose parser & general-purpose generator" device that is characterized by being realized at the same time. Converting to "environment-dependent model bidirectional graph structure data" that can be traced in both directions by inputting "Lisp binary tree perspective tree" that can be traced only in one direction generated from the source code and grammar definition file, and
Converting "environment-dependent model bidirectional graph structure data" that can be traced in both directions to "Lisp binary tree perspective tree" that can be traced only in one direction to make the source code in the grammar definition file,
Function to convert "Lisp binary tree perspective tree" that can be traced only in one direction to "environment-dependent model bidirectional graph structure data" that can be traced in both directions using the "tree graph conversion rule file" group for mutual conversion of When,
The ability to do the opposite conversion,
An extended Prolog device for "tree graph conversion" that can be realized at the same time just by giving the same "tree graph conversion rule file". It has a function to convert "environment-dependent model bidirectional graph structure data" to "environment-independent standard model bidirectional graph structure data" using the "standard modeling program file" group.
At least one "analysis program file" in the "analysis program file" group that "creates a call graph", "creates a flowchart", "creates a data flow graph", and "creates a state transition diagram" " ,
"Automatically insert and delete debug log code""Automatically delete unused parts of the program""Provide a means to extract how to use the library""Necessary and sufficient It is possible to analyze and process "environment-independent standard model bidirectional graph structure data" using at least one "machining program file" in the "machining program file" group that "automatically generates various test specifications". An extended program device for "graph data manipulation" characterized by. The "general-purpose parser & general-purpose generator" device according to claim 1,
The "tree graph conversion" extended Prolog device according to claim 2 and
It is composed of the extended Prolog device for "graph data manipulation" according to claim 3.
Convert the "source code file" group according to claim 1 of the existing system language environment to the "Lisp binary tree perspective tree" according to claim 1 which can be traced only in one direction by the above-mentioned "general-purpose parser & general-purpose generator" device. death,
The above-mentioned "Lisp binary tree perspective tree" is converted into the "environment-dependent model bidirectional graph structure data" according to claim 2, which can be traced in both directions by the above-mentioned "tree graph conversion" extended Prolog device, and the above-mentioned The "environmentally dependent model bidirectional graph structure data" is converted into the "environment-independent standard model bidirectional graph structure data" according to claim 3 by the above-mentioned "graph data manipulation" extended Prolog apparatus.
The above-mentioned "environment-independent standard model bidirectional graph structure data" is analyzed by the above-mentioned "for graph data manipulation" extended Prolog apparatus using the "analysis program file" group according to claim 3 .
The above-mentioned "environment-independent standard model bidirectional graph structure data" is processed by the above-mentioned "for graph data manipulation" extended Prolog apparatus using the "machining program file" group according to claim 3 .
Was converted into its said "environment-independent standard model interactive graph structure data" of the "for translation tree graph" not traced only unidirectionally extended Prolog system said "Lisp2 quadtree parse tree" after processing,
As the apparatus for automatically generating the of a "source code file" group existing systems language environment or target language environment from the "Lisp2 quadtree parse tree" in the above "universal parser & generic Generator" devices.

Images