JP2001005655A - Device and method for aiding development of application generator - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the efficiency in the development of an application generator by making describable a program piece to be described in conversion rule definition to be the specifications of generator operation similarly to an output program piece to be outputted from a generator and making automatically generaable an abstract syntax tree data type definition expressing a syntax analyzer and generator input specifications from grammar specifications to be the generator operation specifications. SOLUTION: The application generator development aiding device 10 is provided with a means 101 for generating the data structure of a term analyzer, a syntax analyzer and an abstract syntax tree from a term analysis specification and a grammar specification to be the operation specifications of the generator and generating an input specification abstract syntax tree expressing the syntax analytical result of an input specification for regulating the operation specifications of a generator output program by using the data structure of the abstract syntax tree, a means 102 for converting output program piece description in conversion rule definition to be the operation specifications of the generator into a conversion rule and a means 103 for executing the abstract syntax tree of the input specification on the basis of the conversion rule outputted from the means 102 and outputting the execution result as a generator output program.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a program development support apparatus and a program development support method, and more particularly, to an application used to support the development of an application generator for automatically generating a program from operation specifications of the program. The present invention relates to a generator development support device and an application generator development support method.

[0002]

2. Description of the Related Art In support of development of a program operating on a computer, a method using an application generator that automatically generates a program that satisfies the operation specification from a description of the operation specification of the program is widely adopted.

[0003] The application generator broadly includes a compiler. Usually the compiler is C
/ Generates an assembler language that can be directly executed on computer hardware from a general-purpose high-level language that does not specify a field, such as the C ++ language. The C / C ++ language is widely used as a high-level language that is input to the compiler as a language independent of a specific operating system or hardware.

[0004] At present, such high-level languages have been widely used, and application generators have been widely used as a means for more efficiently developing application programs in specific fields. The application generator can automatically generate a program described in a high-level language that satisfies the operation specifications by inputting the operation specification description of a program appropriate for a specific field in which the program generated by the generator is used.

The application generator only has to describe the operation specifications of a program limited to a specific field, and therefore has the advantage that the operation specifications can be described in a simpler method than in a general-purpose high-level language.

A typical application generator is yacc, for example. yacc is an application generator for the field of parsing. yacc can automatically generate a parser by describing the syntax of the input syntax and the operation for generating an abstract syntax tree as the operation specifications of the program. For more information on the operation of yacc, please refer to "Compiler I / II Principles, Techniques and Tools" by Aiho / Cessie / Ulman.

Although yacc is an application generator limited to a specific field called a parser, in order to efficiently construct an application generator for programs in various fields, a system for supporting the development of the application generator is required. Or a support method is needed.

In recent years, a domain-oriented language (Domain Specifi
c Language) is used (see p. 1083 of the document "Encyclopedia Electronic Information and Communication Handbook (Ohmsha, published on November 30, 1998)").

This domain-oriented language provides an application generator developer with an environment for supporting application generator construction in order to develop a general-purpose application generator that does not specify a field.

Therefore, an application generator developer can efficiently construct application generators in various fields by developing an application generator using a domain-oriented language.

[0011] Here, as an example of an application generator construction support system using a domain-oriented language, "KHEPERA" (USENIX The C
on-frence on Domain-Specific Languages Proceedings
243) ”). In the following description, a development support device using a domain-oriented language is called DSL.
It is called a development support device.

FIG. 28 is a configuration block diagram of an application generator using a domain-oriented language. In FIG. 28, the lexical analysis specification 601 and the syntax analysis specification 60
2 is determined by determining a specification description expression specific to a specific field. The lexical analyzer specification 601 is provided to the lexical analyzer generator 603 to generate a lexical analyzer program 605. In addition, the syntax analysis specification 602 is
04 to generate a parser program 606.
The lexical analyzer generator 603 includes lex (lexical a
nalyzer generator) is widely used, and the parser generator 604 includes yac (yet another control char).
t) is widely used.

The lexical analyzer generator 603 and the lexical analyzer program 605 and the parser program 606 generated from the parser generator 604 are used to generate the lexical specifications and the abstract syntax specifications generated from the syntax specifications. In the lexical specification and the syntax specification, the operation of outputting this abstract syntax is described as the operation when performing lexical analysis and syntax analysis.

In the description of these operations and in the method of expressing an abstract syntax tree, a data type using a type constructor which enables strict type definition, which is often found in functional languages, is actively used. It is widely known that it can improve the maintainability.

An abstract syntax tree definition 607 is a definition of an abstract syntax tree of an input specification obtained by parsing an input specification by the lexical analyzer program 605 and the parser program 606 described above and obtaining an output thereof.

In the example of FIG. 28, the DSL development support device 61
0 outputs a conversion execution device program 611 for repeatedly applying the conversion rule 609 to the abstract syntax tree definition 607. The conversion rule 609 performs pattern matching on the abstract syntax tree, and rewrites the adapted abstract syntax tree into a different abstract syntax tree in accordance with the rules described in the conversion rule 609. This is a program that converts to an abstract syntax tree.

The DSL development support device 610 generates a text shaping device program 612 from the text shaping rules 608. The text shaping device program 612 shapes the given abstract syntax based on the text shaping rules corresponding to each abstract syntax. That is, the abstract syntax tree of the output program is shaped into an output program by the text shaping device program 612 and output.

The text shaping technique is described in the document "Blashek
By "User-adaptable Prettyprinting" SOFTWARE PRACTICE
AND EXEPERIENCE, VOL19 (7), pp. 687-702 "is known.

The DSL development support device 610 outputs a library program 613. This library program 613 is used for the application generator 61.
5 is a program necessary for execution. Compiler 614
Creates an executable application generator 615 by compiling the library program 613, the lexical analyzer program 605, the syntax analyzer program 606, the conversion execution device program 611, and the text shaping device program 612. .

[0020]

The above-described conventional configuration has the following problems. First, since the fragments of the output program appearing in the conversion rules are represented by the abstract syntax tree, if the conversion rules are described in the abstract syntax, the output program generated by the generator must be described directly in the conversion rules. Can not. For this reason, it is difficult to understand the structure of the program output from the generator, and the development efficiency and maintainability of the generator deteriorate.

As a second problem, since a text formatting rule cannot be embedded in a fragment of an output program appearing in a conversion rule, the syntax of a programming language to be output when an abstract syntax tree of the output program is created. Must be considered in detail, each syntax element must be strictly defined, and the corresponding text formatting rules must be defined.
Therefore, the development work of the generator is complicated.

A third problem is that a developer must describe a procedure for converting a result of parsing into an internal expression. The reason is that, in the conventional method, there is no means for automatically generating a conversion procedure to the internal representation, and the method is not based on an expression method using a type constructor that enables strict type definition.

The fourth problem is that the developer must describe the expression format of the abstract syntax tree. The reason is that the conventional method has no means for automatically generating the format of the internal representation, and is not based on a representation method using a type constructor that enables strict type definition.

In order to solve the first and second problems, an object of the present invention is a means capable of describing a fragment of an output program directly output in a conversion rule, and a means for embedding a text shaping rule at the time of output. An object of the present invention is to provide an application generator development support device and an application generator development support method including:

Another object of the present invention is to generate a data structure definition representing an internal expression for automatically expressing an abstract syntax tree from the specification of a parser in order to solve the third and fourth problems. Another object of the present invention is to provide an application generator development support device and an application generator development support method including means for automatically generating an abstract syntax tree from a grammar specification.

[0026]

According to a first aspect of the present invention, there is provided an application generator development support apparatus comprising: a lexical analyzer, a parser, and an abstract syntax tree based on a lexical analysis specification and a grammatical specification given as an operation specification of a generator. A syntax analysis means for generating an input specification abstract syntax tree expressing a syntax analysis result of an input specification defining an operation specification of a generator output program by using a data structure of the abstract syntax tree; A conversion rule processing unit for converting an output program fragment description in a conversion rule definition given as an operation specification into a conversion rule; and executing the input specification abstract syntax tree based on the conversion rule from the conversion rule processing unit. And a conversion rule executing means for outputting as a generator output program.

According to the first aspect of the present invention, the parsing means generates the lexical analyzer, the parser and the abstract syntax tree data type based on the generator operation specification, and the conversion rule processing means based on the generator input specification. From the output program fragment description, generate a conversion rule for outputting the output program fragment description as a generator output program, and execute the input specification abstract syntax tree obtained by parsing the generator input specification based on the conversion rule by the conversion rule execution means. Output the generator output program.

Therefore, when an operation specification of a program suitable for a certain field is input, a program satisfying the operation specification can be automatically generated. At this time, the parsing means can efficiently create a parser, and the conversion rule processing means can take measures for improving the conversion rule development efficiency and maintainability.

According to a second aspect of the present invention, in the application generator development supporting apparatus according to the first aspect, the conversion rule definition includes a generator conversion rule, the program fragment description, and a text formatting description. Wherein the conversion rule processing means extracts the output program fragment description in the conversion rule definition, storage means for storing a text formatting rule corresponding to the text formatting description, Means for describing, in the extracted output program fragment description, an interpretation result based on the text formatting rules extracted from the storage means when the output program fragment description includes a text formatting description, and an output program in which the interpretation result is described Output an output program equivalent to the fragment description Characterized in that it comprises a means for generating a conversion rule.

According to the second aspect of the present invention, the output program fragment description extracted from the conversion rule definition is converted into a conversion rule for outputting the output program fragment description as a generator output program. When described, interpret the text formatting description,
From the result of interpreting the output program fragment description and the text shaping rules, the output program fragment description is converted into a conversion rule for outputting the output program fragment description as a generator output program.

Therefore, since the fragment description of the program in the conversion rule definition can be described in the same manner as the generator output program as the output result, the efficiency of developing the conversion rule and the maintainability can be improved.

According to a third aspect of the present invention, there is provided the application generator development supporting apparatus according to the first or second aspect, wherein the parsing means includes: As a means for generating a data structure of an abstract syntax tree for expressing an abstract syntax tree obtained as a result of parsing, as a means for extracting a non-terminal symbol information from the storage unit and the grammar specification, Means for determining and storing information on the type-named non-terminal symbol in the storage unit, extracting node information from the grammar specification, selecting a node that needs to be incorporated in a syntax tree, and storing the selected node information in the storage unit Means for storing the nonterminal symbol determined from the storage unit and the selected node information, and Means for determining the type of each node, storing the determined node type in the storage unit, a value constructor associated with the rule of the grammar specification, and the type of the determined non-terminal symbol extracted from the storage unit Means for generating a data type of the non-terminal symbol based on the name and the type of the determined node; and, based on the grammar specification, the data type of the generated non-terminal symbol, and the selected node information extracted from the storage unit. Means for generating an operation for constructing an abstract syntax tree; means for generating a parser based on the grammar specification, the data type of the generated non-terminal symbol, and the operation for constructing the generated abstract syntax tree; Means for outputting an abstract syntax tree data type based on the data type of the non-terminal symbol.

According to the third aspect of the present invention, the storage unit stores, in order, the type name of the non-terminal symbol, the node that needs to be incorporated in the syntax tree, and the type of the node. , A nonterminal symbol data type, an operation for constructing an abstract syntax tree, a parser, and an abstract syntax tree data type.

Therefore, just by inputting the grammar specification,
It can automatically output a data structure for expressing the abstract syntax tree of the input specification according to the grammar, and a parse tree for outputting the data structure, and represents an internal representation for expressing the abstract syntax tree Data structure definitions can be automatically generated.

According to a fourth aspect of the present invention, there is provided an application generator development supporting method, wherein a lexical analyzer, a parser, and a data structure of an abstract syntax tree are generated from a lexical analysis specification and a grammatical specification given as operation specifications of the generator. Generating an input specification abstract syntax tree in which the syntax analysis result of the input specification that defines the operation specification of the generator output program is represented using a data structure of the abstract syntax tree; and a conversion rule definition given as the operation specification of the generator. Converting the output program fragment description therein into a conversion rule, executing the input specification abstract syntax tree based on the conversion rule from the conversion rule processing means, and outputting an execution result as a generator output program. It is characterized by.

According to a fifth aspect of the present invention, in the application generator development supporting method according to the first aspect, the conversion rule definition includes a generator conversion rule, the program fragment description, and a text formatting description. Wherein the step of converting to the conversion rule includes a step of extracting an output program fragment description in the conversion rule definition, and a step of storing when the extracted output program fragment description includes a text formatting description. A procedure for describing an interpretation result by the text shaping rule extracted from the means in the extracted output program fragment description, and a procedure for converting the output program fragment description in which the interpretation result is described into a conversion rule for outputting an output program equivalent to the output program fragment description. It is characterized by having.

According to a sixth aspect of the present invention, there is provided the application generator development supporting method according to the first or second aspect, wherein the grammar specification includes a parser and a parsing result. As a procedure for generating a data structure of the abstract syntax tree for expressing the obtained abstract syntax tree, the method extracts the information of the non-terminal symbols from the grammar specification, determines the type name of each non-terminal symbol, and determines the type name determined non-terminal symbol. Storing the node information from the grammar specification, selecting a node that needs to be included in the syntax tree, and storing the selected node information in the storage unit; and storing the selected node information in the storage unit. From the determined nonterminal symbol type and the selected node information, determine the type of each node based on them, and determine The procedure for storing the type of the node in the storage unit, the value constructor associated with the rule of the grammar specification, the type name of the determined non-terminal symbol extracted from the storage unit, and the type of the determined node. Generating a data type of a non-terminal symbol based on the grammar specification, a data type of the generated non-terminal symbol, and an operation of constructing an abstract syntax tree based on the selected node information extracted from the storage unit; A procedure for generating a parser based on the grammar specification, the data type of the generated non-terminal symbol, and an operation of constructing the generated abstract syntax tree; and an abstract syntax based on the data type of the generated non-terminal symbol. Outputting a tree data type.

[0038]

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

FIG. 1 is a configuration block diagram of an application generator development support device according to an embodiment of the present invention. In FIG. 1, the application generator development support device 10 includes a generator operation specification 20 for providing an operation specification of the application generator development support device 10 and a generator input specification for providing an operation specification of a program output by the application generator development support device 10. 30 to generate and output a generator output program 40.

The application generator development support device 10 includes a syntax analysis unit 101, a conversion rule processing unit 10
2. It has three means of conversion rule execution means 103.

The lexical analyzer 101 includes a lexical analyzer generator 101a, a lexical analyzer 101b generated by the lexical analyzer generator 101a, a parser and an abstract parse tree data type generator 101c, a lexical analyzer and Parser 101d generated by abstract syntax tree data type generator 101c
And an abstract syntax data type 101e and an input specification abstract syntax tree 101f.

The conversion rule processing means 102 includes a generator operation specification reading means 102a, an output program fragment description processing means 102b, and a text shaping description interpretation means 10
2c, a text formatting rule storage unit 102d, an output program fragment description storage unit 102e, and a conversion rule generation unit 1.
02f and a conversion rule storage unit 102g.

The conversion rule executing means 103 includes a conversion rule compiling means 103a and an execution code storage section 103.
b and the conversion executing means 103c.

The generator operation specification 20 includes a syntax rule definition 201 and a conversion rule definition 202.

The syntax rule definition 201 includes a lexical analysis specification 201a and a grammar specification 2 related to the lexical definition of the specification description input to the application generator development support apparatus 10.
01b.

The conversion rule definition 202 includes a generator conversion rule 202a, an output program fragment description 202b, and a text shaping description 202c related to the operation specification description of the application generator development support apparatus 10.

Here, each processing means of the application generator development support apparatus 10 is a computer (CP
U), and the storage unit is configured by a storage device such as a semiconductor memory or a magnetic disk device. Then, the generator operation specification 20 and the generator input specification 30 are stored in the storage device.

Hereinafter, the operation of the application generator development support apparatus of the present embodiment will be described in detail with reference to FIGS. Note that FIG.
5 is an operation flowchart of a part of the conversion rule processing unit 102 of FIG. FIG. 3 is an operation flowchart of a part of the syntax analysis unit 101 and the conversion rule execution unit 103.

FIG. 4 is a diagram describing a description grammar of the generator operation specification 20. FIG. 5 shows a lexical analysis specification 201.
FIG. 4 is a diagram describing a description grammar of FIG. FIG. 6 is a diagram of a specific description of the description grammar of the lexical analysis specification 201a. FIG.
Is a diagram describing the grammar description of the grammar specification 201b.
FIG. 8 is a diagram of a specific description of the description grammar of the grammar specification 201b.

FIG. 9 is a diagram describing the description grammar of the generator conversion rule 202a. FIG. 10 is a diagram showing a specific description of the description grammar of the generator conversion rule 202a. FIG. 11 is a diagram describing the description grammar of the output program fragment description 202b. FIG. 12 is a diagram describing a pattern in the description grammar of the output program fragment description 202b.

FIG. 13 shows an output program fragment description 202.
FIG. 13 is a diagram showing a specific description of a description grammar of b and a text shaping description 202c. FIG. 14 shows a text format description 20
It is the figure which showed the example of 2c. FIG. 15 is a diagram illustrating an example of a text shaping rule stored in the text shaping rule storage unit 102d.

FIG. 16 shows an output program fragment description 202.
It is a figure which shows the specific description of b. FIG. 17 is a diagram showing an output program generated from the description of FIG. FIG. 18 is a definition of a conversion rule (a conversion rule that outputs an output program equivalent to the extracted output program fragment description) generated by the conversion rule generation unit 102f.

First, the operation of a part of the syntax analysis means 101 and the conversion rule processing means 102 will be described with reference to the operation flowchart of FIG. In FIG. 2, S2 to S5 are operations of the syntax analysis unit 101, and S6 to S14 are operations of the conversion rule processing unit 102.

In S1, the generator operation specification reading means 102a reads the generator operation specification 20 from the storage device and, in accordance with the content of the description, generates a lexical analyzer generator 101a, a syntax analyzer, and an abstract syntax tree data type generator 101b. To the corresponding units of the conversion rule storage unit 102g and the output program fragment description processing unit 102b (S
2, S4, S6, S8).

The generator operation specification 20 is described in a grammar as shown in FIG. In FIG. 4, the specification type is an identifier for distinguishing the lexical analysis specification 201a, the grammar specification 201b, the generator conversion rule 202a, the output program fragment description 202b, and the text formatting description 202c, which constitute the generator operation specification 20. In this way, by adding a name as an identifier for each specification type, reference from other specification definitions becomes possible.

In S2, if the read generator operation specification 20 is the lexical analysis specification 201a, the generator operation specification reading means 102a passes the lexical analysis specification 201a to the lexical analyzer generation means 101a and starts up.

The lexical analysis specification 201a is described by a grammar as shown in FIG. In the lexical analysis specification 201a,
The definition of the token used in the syntax analysis described later is described using a regular expression. For example, in FIG. 6, a token IDENTIFIER representing an identifier is described using a regular expression, and calc_token
The example defined by the name is shown.

The lexical analyzer generating means 101a executes lexical analyzer generating processing based on the lexical analysis specification 201a to generate the lexical analyzer 101b (S3).

The operation of the lexical analysis generating means and the lexical analyzer is described in "Aiho / Cessy / Ulman," Compiler I / II Principles, Techniques and Tools "Science, Inc."
Familiar with.

In S4, when the generator operation specification 101 that has been read is the grammar specification 201b, the generator operation specification reading means 102a passes it to the syntax analyzer and the abstract syntax tree data generation means 101c and starts up.

The grammar specification 201b is described by a grammar as shown in FIG. The grammar specification 201b is described in the same format as a grammar specification (see FIG. 21) handled by a parser generator (FIG. 19) described later. FIG. 8 shows a specific description example of the grammar specification. In the description example of FIG. 8, a description grammar of a grammar specification for performing a simple addition / subtraction operation is defined.
Defined under the name alc_grammer.

The syntax analyzer and the abstract syntax tree data generating means 101c execute this grammar specification 201b as described later.
A parser and abstract syntax tree data generation processing are executed based on
1e is generated (S5).

In S6, if the read generator operation specification 20 is the generator conversion rule 202a, the generator operation specification reading means 102a converts the generator conversion rule 202a into the conversion rule compilation means 1023.
a into an internal representation that can be processed by the
2g (S7).

The generator conversion rule 202a is described by a grammar as shown in FIG. In this embodiment, the description method of the generator conversion rule 202a is described by a description method conforming to the ML. Regarding ML, refer to the document "ML for the Working Programmer" by LC Paulson.
Cambridge University Press ”.

In FIG. 10, the syntax analyzer 101d is called based on the lexical analysis specifications and grammar specifications described as shown in FIGS. 5 to 7 as concrete definitions of the generator conversion rules. It describes the operation of binding the parse tree to the variable spec and generating the output program by calling the function generate.

The storage method of the generator conversion rules described in this manner is described in “Compiling by Andrew W. Appel”
with Continuations "Cambridge University Press".

In S8, if the read generator operation specification 20 is the output program fragment description 202b, the generator operation specification reading means 102a passes it to the output program fragment description processing means 102b and starts up.

The output program fragment description 202b describes a part of the program output from the generator. Two types of descriptions can be performed on the output program fragment description 202b. One is a description of the output program itself. In this case, the text image of the output program may be directly described. The other is a description of a text shaping description for expanding other program fragments based on the information described in the abstract syntax tree given from the generator input specification 30.
This text formatting description includes indentation and line feed characters and text formatting rules. Note that these two types of descriptions can be described together in an output program fragment.

The output program fragment description 202b is
As shown in FIG. 11, it is described as a sequence of patterns. The pattern is described as a sequence of arguments and program fragments, as shown in FIG. FIG. 13 shows ma of the C language.
Describes a program fragment that outputs the in function.

A text shaping description to be described when describing another program fragment in a program fragment can be described, for example, as shown in FIG. That is, in the example of FIG.
Specifies that the output program fragment specified by the name switch_out is to be embedded in the insert part when the main function is expanded.

A variable passed as an argument of the pattern in the description of the output program fragment, that is, the branch in FIG.
In the part represented by the variable es, the input specification abstract syntax tree 101f obtained by applying the abstract syntax tree data type 101e to the result of parsing the generator input specification 30 by the syntax analyzer 101d is converted by a conversion rule. The result is passed.

Therefore, a generator output program 40 conforming to the generator input specification 30 can be obtained by embedding the variable branches passed as an argument of the pattern in the output program fragment.

In S9, the output program fragment description processing means 102b transmits the received output program fragment description 20
It is checked whether or not there is a text formatting description 202c in 2b. If there is no text shaping description 202c, the output program fragment description processing means 102b stores the received output program fragment description 202b in the output program fragment description storage unit 102e (S10).

On the other hand, output program fragment description processing means 1
02b is the received output program fragment description 202b
If there is a text formatting description 202c therein, it is extracted and the contents of the text formatting description 202c are checked (S11).
First, processing for indentation and line feed characters of the text formatting description 202c is performed (S12), and then processing of the conversion rules is entrusted to the text formatting description interpreting means 102c (S1).
3-S15).

That is, output program fragment description processing means 1
02b extracts indentation and line feed characters from the text formatting description 202c, internally stores the indentation and line feed position information (S12), and stores the indentation and line feed characters in the output program fragment description storage unit 102e. It is stored (S10).

Then, the output program fragment description processing means 102b passes the text shaping description 202c, which has completed the processing for indentation and line feed characters, to the text shaping description interpreting means 102c. Thereby, the text shaping description interpreting means 102c performs the processing of S13 to S15.

FIG. 14 shows a list of text formatting descriptions that can be described in this embodiment. The text shaping rule storage unit 102d stores a text shaping rule describing a text shaping method. FIG. 15 shows an example.

The text shaping description 202c is stored in the text shaping rule storage unit 10c.
Used when searching for 2d. Therefore, when a new text shaping description is required, it is possible to freely add a text shaping rule corresponding to the text shaping description by registering the text shaping rule in the text shaping rule storage unit 102d.

The text shaping description interpreting means 102c searches the text shaping rule storage unit 1025d for a text shaping rule as shown in FIG. 15 (S13), and binds the value of the text shaping description 202c to each argument of the text shaping rule. To interpret the text formatting description (S
14) Generate an output program fragment description 202b to which information for performing text shaping is added based on the interpretation result and the indentation and line feed position information internally stored in S12 (S12).
15) and passes it to the conversion rule generating means 102f (S1)
6).

The conversion rule generating means 102f stores the output program fragment description, the characters for indentation and line feed, and the information for performing the text shaping obtained from the text shaping description interpreting means 102c stored in the output program fragment description storage unit 102e. Is generated from the output program fragment description 202b to which is added (S16), and the generated conversion rule is stored in the conversion rule storage unit 102g (S7).

As described in the section of the prior art, the conversion rule referred to here performs pattern matching between the abstract syntax tree given as input and the abstract syntax tree defined by the rule, and the pattern matching succeeds. In this case, the program rewrites the abstract syntax tree that is the output defined in the rule.

FIG. 16 is an example of a program fragment description described using a text formatting description. In this example, FIG.
The program fragment description switch_out referenced using the insert format description shown in 3 is defined. In the program fragment description switch_out, a C language switch statement is described.
A description for expanding a C language case statement for each element of the argument branches is provided inside a switch statement.

Description of the output program fragment of FIG. 13, FIG.
When a list of character strings such as ["1", "2"] is input to the output program fragment description and the generator input specification 30 is executed, the output in consideration of indentation and the like as shown in FIG. The program fragment description is output as a generator output program 40.

FIG. 18 is a description example of a conversion rule generated from the output program fragment description shown in FIG. In this example, a conversion rule is shown in which a program fragment description without a text formatting description is directly output as a character string in an output program. Also, a conversion rule is described so that a program fragment description having a text formatting description is described in an output program together with indentation and line feed position information described in the text formatting rule definition.

Then, in S16, the above S1 to S16
Are executed for all the generator operation specifications 20. As a result, the syntax analysis unit 101 generates a lexical analyzer 101b, a syntax analyzer 101d, and an abstract syntax tree data type 101e, and the conversion rule processing unit 101 stores the generator conversion rule 202a in the conversion rule storage unit 102g. Conversion rule and output program fragment description 20
The conversion rule of 2b is stored one by one.

Next, a part of the syntax analysis means 101 and the operation of the conversion rule execution means 103 will be described with reference to FIGS. In FIG. 3, S20 to S23 correspond to the operation of the syntax analysis unit 101, and S24 to S29 correspond to the operation of the conversion rule execution unit 103.

In FIG. 1 and FIG. 3, syntax analysis means 101
Then, the generator input specification 30 is read (S2
0), the lexical analyzer 101b uses the generator input specification 30
(S21), and the syntax analyzer 101d analyzes the syntax of the generator input specification 30 using the lexical analysis result (S22). And the input specification 3 that was parsed
The abstract syntax tree 0 is represented by an input specification abstract syntax tree 101f which is an internal representation using the abstract syntax tree data type 101d (S23).

Next, in the conversion rule executing means 103, the conversion rule compiling means 103a outputs the input specification abstract syntax tree 1
01f is compiled into an execution code (S24). Further, the conversion rule compiling means 103a searches the conversion rule storage unit 101g for a conversion rule required for conversion, and converts the conversion rule into an executable code (S25, S26).

Then, the conversion rule compiling means 103
a is an execution code storage unit that stores the execution code corresponding to the generator input specification 30 obtained in S24 and the execution code corresponding to the generator conversion rule 202a and the change rule generated by the change rule generation unit 102f obtained in S26. 1
03b (S27).

The conversion rule compiling means 103a is described in "Compilin by Andrew W. Appel" mentioned above.
g with Continuations "Cambridge University Press"
Familiar with.

Next, the conversion execution means 103c converts the generator input specification 30 using the execution code stored in the execution code storage section 103b, and outputs the conversion result as a generator output program 40 (S28,
S29).

Next, FIG. 19 is a block diagram showing the configuration of a parser generator according to the embodiment of the present invention. The parser generating apparatus includes a parser and an abstract parser tree data type generating unit 101c and a parser 101 shown in FIG.
d, corresponding to the entire abstract syntax tree data type 101e.

Referring to FIG. 19, this parser generating apparatus is a data processing apparatus 50 operated under program control.
1 and a storage device 502 for storing information, and are configured to automatically output an abstract syntax tree data type 504 and a parser 505 from an input grammar specification 503.

The data processing device 501 includes a grammar description reading unit 5011, a non-terminal symbol type name determining unit 5012, an element selecting unit 5013, and each node type determining unit 5014.
Non-terminal symbol data type generation means 5015, syntax tree construction operation generation means 5016, and syntax analyzer generation means 501
7 and an abstract syntax tree data type output unit 5018. These means are realized by a computer (CPU).

The storage device 502 includes a grammar information storage unit 502
1 and an abstract syntax tree data type storage unit 5022.
The storage device 502 is a semiconductor memory, a magnetic disk device, or the like. The grammar information storage unit 5021 stores the internal representation of the grammar description, the type name of the non-terminal symbol, the selected node and its type, and the like. The abstract syntax tree data type storage unit 5022 stores an abstract syntax tree data type.

Next, the operation of the parser generator of the present embodiment configured as described above will be described in detail with reference to FIGS. FIG. 20 is an operation flowchart of the parser generator. FIG. 21 is a diagram showing a specific example of a description method of a grammar rule. FIG. 22 is a diagram showing types of syntax trees that can be described in the grammar rules. FIG. 23 is a diagram illustrating an example of a grammar description. FIG.
FIG. 24 is a diagram showing an example of an input described in the grammar of FIG. FIG. 25 is a diagram showing an example in which the input of FIG. 24 is represented by an internal expression. FIG. 26 is a diagram showing a data type generated from the internal representation of FIG. FIG. 27 is a diagram in which an operation is added to the internal representation of FIG.

First, the grammar specification 503 will be described.
The grammar description to be used in the present embodiment is BN described below.
It is described by F grammar (hereinafter, referred to as “BNF1”).
BNF1 is composed of a set of rules, as shown in FIG. A rule is one of the nonterminal symbols defined in the rule.
And one or more branches. Each branch consists of a value constructor name and a sequence of zero or more nodes.

Of these, the value constructor name described at the head of each branch is a BNF1 specific element used in the method of the present embodiment (hereinafter referred to as “the present method”). That is, the value constructor name is used as a value constructor name when automatically generating a data type for the abstract syntax tree.

The node column is an element that is always possessed in a general BNF grammar. This column of nodes indicates what grammatical element the nonterminal symbol represented by the rule is expanded to. Each node is 1 to 7 in the table of FIG.
Any of the seven listed below.

Of these, one token and two non-terminal symbols are elements that always exist in a general BNF grammar.
A token of 1 means a token which is the minimum unit for parsing. A non-terminal symbol of 2 means a non-terminal symbol defined by another rule. Nodes 3, 4, and 5 are
An element that exists depending on the BNF grammar, and indicates whether the target node is repeated or whether that node appears.

The delimiter in the nodes 6 and 7 means that when the node is repeated two or more times, a node corresponding to the delimiter is interposed between the elements. The notation of the delimiter in the nodes 6 and 7 is an element unique to the present method.

Note that, in the following description, :: = |
This method can be applied even if symbols such as <> [] * + are replaced with other symbols.

In the BNF1 grammar used in the present method, each of the nodes 3 to 7 in FIG.
Although not NF, this technique can be applied even if the use of this list expression is partially or entirely restricted.

In the initial state, the grammar information storage unit 5021 does not store information on the grammar itself, but stores only information on the lexical analysis stage before the syntax analysis. The information on the lexical analysis stage includes the types of tokens such as identifiers, character strings, and integers, and the types of those tokens.

The operation will be described below with reference to the operation flowchart of FIG. In S31, the grammar description given as the grammar specification 503 is supplied to the grammar description reading unit 5011. This grammar description is, for example, as shown in FIGS. FIG. 23 is a grammar description example based on the grammar shown in FIG. FIG. 24 is an input example described in the grammar of FIG.

The grammar description reading means 5011 has an internal representation (FIG. 2) which can handle this grammar description in the data processing device 501.
5) and stored in the grammar information storage unit 5021. The internal representation at this time does not include the type name, the label of selection or non-selection in the information of each node, and the type information among the information included in FIG. These elements are
It is derived in the following steps. Regarding this internal representation,
Any form can be used as long as it can be handled by the data processing device.

In S32, the non-terminal symbol type name determining means 50
12 extracts a non-terminal symbol name from each rule of the internal representation of the grammar (FIG. 25) stored in the grammar information storage unit 5021, attaches a character string "Type" to the non-terminal symbol name, and sets it as a type name of the rule. , Stored in the grammar information storage unit 5021. At this point, the type name information in FIG. 25 has been added. Here, the name of the type of each non-terminal symbol may be anything as long as it is unique for each non-terminal symbol. As an example, the name of the non-terminal symbol itself may be used.

In S33, the element selecting means 5013 uses the internal expression (FIG. 2) stored in the grammar information storage unit 5021.
The information about the node is extracted from 5), a node that needs to be incorporated into the syntax tree (that is, a node that needs to be retained after the syntax analysis) is selected, and information on the selected node is stored in the grammar information storage unit 5021. I do.

Here, a node that needs to be incorporated into the syntax tree is a node whose contents are variable. The node whose content is variable is, for example, a non-terminal symbol or a character string token. In the example of FIG. 23 currently being handled, Arg (a non-terminal symbol), INTEGER (a token representing an integer), and the like correspond thereto.

A node whose content is not variable is a token whose content does not change, such as a keyword, a parenthesis, or a token node that exists for grammatical delimitation. In the example of FIG.
+ "And"{", and so on.

Then, among the internal expressions stored in the grammatical information storage unit 5021, nodes determined to be necessary are labeled as selection, and unnecessary nodes are labeled as non-selection. At this point, the selection / non-selection information in FIG. 25 has been added. However, the method of the present invention can be applied even if the criteria of “necessary nodes” here are changed. For example, all nodes may be required.

In S34, each node type determining means 5014
Extracts the type name of each non-terminal symbol and information on the selected node from the grammar information storage unit 5021, determines the type of each selected node from the information, and stores the information in the grammar information storage unit 5021. .

This decision is made based on the type of each node. In FIG. 22, when the target node is one token, the type of the token itself is used. As an example, a token representing an integer is an integer type, and a token representing a character string is a string type. In the case of the non-terminal symbol of 2, it is defined as a type having the name of the non-terminal symbol added in S32.

Although the entity of the type is not defined at this time, it is defined in a later procedure, so that it is sufficient that correspondence is established by name. In the case of 3, 4, 5, 6, and 7, it is a list type of the types of the included nodes. According to this, the types of all the selected nodes are determined.

For example, in FIG. 25, the token INTEGER of the node 1 of the branch 1 of the rule 3 is an int type because it represents an integer. In the case of a non-terminal symbol, the type has the type name of the rule that defines the non-terminal symbol. For example, the node 2 of the branch 1 or 2 of the rule 1 is a non-terminal symbol Ar
g, and referring to Rule 2 that defines Arg, the type name is ArgType.

Therefore, the type is ArgType. In the case of a node enclosed by '['']', the same non-terminal symbol is a list type with the type name of the rule defining the non-terminal symbol.

For example, node 2 of branch 1 of rule 2
Is [AtExpr; ","] +, so the type is AtE
xpType list.

In the same manner, the type of the selected node is assigned, and the information is stored in the internal expression (FIG. 25) of the grammar information storage unit 5021. At this point, the type information attached to each node in FIG. 25 has been added.

In S35, the non-terminal symbol data type generation unit 5015 extracts information of a value constructor associated with the rule of the internal expression of the grammar (FIG. 25) from the grammatical information storage unit 5021, and further extracts the type name of each non-terminal symbol. And the information of the type of each node is extracted, and the data type of the non-terminal symbol defined by the rule is generated based on the information.

This data type is a type having a data type name as the name assigned by the non-terminal symbol type name determining means 5012 and having a value constructor taken out. The type associated with each value constructor is S
At 34, the type of the node determined by each node type determining means 5014 is a set.

In other words, in the data type to be generated, the type name becomes the name of the data type, the value constructor name of each branch becomes the value constructor of the data type, and is selected in each branch. The type of the node is the type attached to its value constructor.

The data type of the non-terminal symbol thus generated is the data type of the abstract syntax tree. The generated data type is stored in the abstract syntax tree data type storage unit 5022. In the case of the present example, the generated data type is as shown in FIG.

In S36, the abstract syntax tree construction operation generating means 5016 stores the internal expression of the grammar and the information of the selected node from the grammatical information storage unit 5021 and the data type of the abstract syntax tree from the abstract syntax tree data type storage unit 5022. , Respectively, and generate an operation for constructing an abstract syntax tree.

Since the data type of the non-terminal symbol defined by the present method reflects the type of each node, the operation of constructing the abstract syntax tree to be generated is performed by combining the contents of each node as a set and applying it to the value constructor. This is an operation to embed the corresponding content in that part.

For example, when the type of a node is a list type, a list of elements corresponding to that part is used as a node value. In the example currently being dealt with, the value of the branch to be generated by the operation defined here is as shown in FIG. However, in each branch in FIG.
The content of each node (number) is $ (number).

In S37, the parser generating means 5017
Extracts the grammar information from the grammar information storage unit 5021 and the data type of the abstract syntax tree from the abstract syntax tree data type storage unit 5022, respectively.
An operation for constructing an abstract syntax tree is received from 016, and a parser 505 is generated from the information.

The method of generating a parser based on the grammar, the data type of the abstract syntax tree, and the operation of constructing the abstract syntax tree is the same as the conventional method.

In S38, the parser generating means 5017
Output the generated parser 505.

In S39, the abstract syntax tree data type output unit 5018 outputs the abstract syntax tree data type storage unit 5022.
, And extracts the data structure of the abstract syntax tree as an abstract syntax tree data type 504.

As a result, in the syntax analysis of the generator input specification 30 in FIG. 1, the syntax analyzer 505 (10
1d) and the abstract syntax tree data type 504 (101
e) Input specification abstract syntax tree 101f which is an internal representation
Is automatically generated.

Note that the abstract syntax tree data type 504 generated in the present embodiment targets ML which is a typical language that realizes a data type having a strict type system having a polymorphic type. It is. The literature on ML has been introduced earlier.

[0132]

As described in detail above, according to the present invention,
The program fragment described in the conversion rule definition, which is the generator operation specification, can be described in the same way as the output program fragment output by the generator, and the parser and generator input specification can be automatically converted from the grammar specification, which is the generator operation specification. Since the abstract syntax tree data type definition to be represented can be generated, the application generator can be efficiently developed.

[Brief description of the drawings]

FIG. 1 is a configuration block diagram of an application generator development support device according to an embodiment of the present invention.

FIG. 2 is an operation flowchart of a part of a syntax analysis unit and a conversion rule processing unit.

FIG. 3 is an operation flowchart of a part of the syntax analysis unit and a conversion rule execution unit.

FIG. 4 is a diagram describing a description grammar of a generator operation specification.

FIG. 5 is a diagram describing a description grammar of a lexical analysis specification.

FIG. 6 is a diagram of a specific description of a description grammar of a lexical analysis specification.

FIG. 7 is a diagram describing a description grammar of a grammar specification.

FIG. 8 is a diagram of a specific description of a description grammar of a grammar specification.

FIG. 9 is a diagram describing a description grammar of a generator conversion rule.

FIG. 10 is a diagram showing a specific description of a description grammar of a generator conversion rule.

FIG. 11 is a diagram describing a description grammar of an output program fragment description.

FIG. 12 is a diagram describing a pattern in a description grammar of an output program fragment description.

FIG. 13 is a diagram showing a specific description of a description grammar and a text formatting description of an output program fragment description.

FIG. 14 is a diagram showing an example of a text formatting description.

FIG. 15 is a diagram illustrating an example of a text shaping rule stored in a text shaping rule storage unit.

FIG. 16 is a diagram of a specific description of an output program fragment description described using a text formatting description.

FIG. 17 is a diagram of an output program generated from the description of FIG. 16;

FIG. 18 is a definition of a conversion rule (a conversion rule generated from an extracted output program fragment description and enabling the output program fragment description to be output as a generator output program) generated by the conversion rule generation means.

FIG. 19 is a configuration block diagram of a parser generation device of the embodiment.

FIG. 20 is an operation flowchart of the parser generator.

FIG. 21 is a diagram showing a specific example of a description method of a grammar rule.

FIG. 22 is a diagram showing types of syntax trees that can be described in a grammar rule.

FIG. 23 is a diagram illustrating an example of a grammar description.

FIG. 24 is a diagram showing an example of an input described in the grammar of FIG. 24;

FIG. 25 is a diagram showing an example in which the input of FIG. 25 is represented by an internal expression.

FIG. 26 is a diagram showing a data type generated from the internal representation of FIG. 26.

FIG. 27 is a diagram in which an operation is added to the internal representation of FIG. 26;

FIG. 28 is a configuration block diagram of a conventional application generator development support device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Application generator development support apparatus 20 Generator operation specification 30 Generator input specification 40 Generator output program 101 Syntax analyzer 101a Lexical analyzer generator 101b Lexical analyzer 101c Parser and abstracter data type generator 101d, 504 Abstract syntax tree data Type 101e, 505 Syntax analyzer 101f Input specification abstract syntax tree 102 Conversion rule processing means 102a Generator operation specification reading means 102b Output program fragment processing means 102c Text format description interpreter 102d Text format rule storage unit 102e Output program fragment description storage unit 102f Conversion rule generation unit 102g Conversion rule storage unit 103 Conversion rule execution unit 103a Conversion rule compilation unit 103b Execution code storage unit 103c Modification Execution means 201 Syntax rule definition 201a Lexical analysis specification 201b, 503 Grammar specification 202 Conversion rule definition 202a Generator conversion rule 202b Output program fragment description 202c Text formatting description 501 Data processing device 502 Storage device 5011 Grammar description reading means 5012 Nonterminal symbol name determining means 5013 element selection means 5014 each node type determination means 5015 non-terminal symbol data type generation means 5016 abstract syntax tree construction operation generation means 5017 syntax analyzer generation means 5018 abstract syntax tree data type output means 5021 grammar information storage unit 5022 abstract syntax tree data type Memory

Claims (6)

[Claims] 1. A lexical analyzer, a parser, and a data structure of an abstract syntax tree are generated from a lexical analysis specification and a grammar specification given as an operation specification of a generator, and an input specification that defines an operation specification of a generator output program is generated. Input specification expressing the syntax analysis result using the data structure of the abstract syntax tree Syntax analysis means for generating an abstract syntax tree, and converting the output program fragment description in the conversion rule definition given as the operation specification of the generator into a conversion rule An application generator comprising: a conversion rule processing unit; and a conversion rule execution unit that executes the input specification abstract syntax tree based on the conversion rule from the conversion rule processing unit and outputs an execution result as a generator output program. Development support equipment. 2. The application generator development support device according to claim 1, wherein the conversion rule definition comprises a combination of a generator conversion rule, the program fragment description, and a text formatting description. Means for extracting the output program fragment description in the conversion rule definition; storage means for storing a text formatting rule corresponding to the text formatting description; and when the extracted output program fragment description includes a text formatting description Means for describing, in the extracted output program fragment description, the result of interpretation based on the text formatting rules extracted from the storage means, and generating a conversion rule for outputting an output program equivalent to the output program fragment description in which the interpretation result is described. Characterized by comprising means for performing Generator development support equipment. 3. The application generator development support device according to claim 1, wherein the syntax analysis unit is configured to express a syntax analyzer and an abstract syntax tree obtained as a result of the syntax analysis from a grammar specification. As a means for generating a data structure of an abstract syntax tree of: a storage unit, extracting information of non-terminal symbols from the grammar specification, determining a type name of each non-terminal symbol, and storing the information of the type-name determined non-terminal symbol in the storage unit. Means for storing the node information from the grammar specification, selecting a node that needs to be incorporated into the syntax tree, and storing the selected node information in the storage unit; and The type name of the non-terminal symbol and the selected node information are taken out, the type of each node is determined based on them, and the determined node type is stored in the storage unit. Means for generating a data type of a non-terminal symbol based on a value constructor associated with a rule of the grammar specification, a type name of the determined non-terminal symbol retrieved from the storage unit, and a type of the determined node. Means for generating an operation for constructing an abstract syntax tree based on the grammar specification, the data type of the generated non-terminal symbol, and the selected node information retrieved from the storage unit, and the grammar specification and the generated non-terminal Means for generating a parser based on the data type of the symbol and the operation of constructing the generated abstract syntax tree; and means for outputting an abstract syntax tree data type based on the data type of the generated non-terminal symbol Application generator development support device characterized by the following. 4. A lexical analyzer, a parser, and a data structure of an abstract syntax tree are generated from a lexical analysis specification and a grammatical specification given as an operation specification of a generator, and an input specification that defines an operation specification of a generator output program is generated. A step of generating an input specification abstract syntax tree expressing a syntax analysis result using a data structure of the abstract syntax tree; and a step of converting an output program fragment description in a conversion rule definition given as an operation specification of a generator into a conversion rule. Executing the input specification abstract syntax tree based on a conversion rule from the conversion rule processing means, and outputting an execution result as a generator output program. 5. The application generator development supporting method according to claim 1, wherein the conversion rule definition comprises a combination of a generator conversion rule, the program fragment description, and a text formatting description. Extracting the output program fragment description in the conversion rule definition; and, when the extracted output program fragment description includes a text formatting description, extracting the interpretation result by the text formatting rule extracted from the storage unit. An application generator development support method, comprising: a procedure to be described in an output program fragment description; and a procedure to convert to a conversion rule that outputs an output program equivalent to the output program fragment description in which the interpretation result is described. 6. The application generator development supporting method according to claim 1, wherein a syntax analyzer and an abstract syntax tree for expressing an abstract syntax tree obtained as a result of the syntax analysis are obtained from the grammar specification. A procedure for extracting the non-terminal symbol information from the grammar specification, determining a type name of each non-terminal symbol, and storing the information on the type name determined non-terminal symbol in a storage unit as a procedure for generating a data structure; and A procedure for extracting the node information from the node, selecting a node that needs to be incorporated in the syntax tree, and storing the selected node information in the storage unit; and a type name of the non-terminal symbol determined from the storage unit and the selected A step of extracting node information, determining a type of each node based on the node information, and storing the determined node type in the storage unit; A procedure for generating a data type of a non-terminal symbol based on an associated value constructor, the type name of the determined non-terminal symbol extracted from the storage unit, and the determined node type; and the grammar specification and the generated non-terminal Generating an operation for constructing an abstract syntax tree based on a data type of a symbol and the selected node information extracted from the storage unit; a grammar specification, a data type of the generated non-terminal symbol, and the generated abstract An application generator development method comprising: a step of generating a parser based on an operation of constructing a syntax tree; and a step of outputting an abstract syntax tree data type based on the data type of the generated non-terminal symbol. .