JPH02133824A - Translator - Google Patents

Abstract

PURPOSE:To cope with plural rule systems by preparing plural syntax analysis parts in response to different grammatical rules and switching these analysis parts with the requests of a meaning analysis part. CONSTITUTION:A control part 20 refers to a table 26 and calls a syntax analysis part 21. The part 21 calls a word/phrase analyzing part 22, and this part 22 inputs an input text one by one character and divides the text 5 into plural tokens. These tokens are inputted to the part 21 for the production of a syntax tree. A meaning analysis part 25 extracts the information on the header part of the text 5 out of the syntax tree. Then the part 25 sets the address of a function corresponding to a syntax analysis part 23 to the table 26. Thus the part 20 refers to the table 26 to call the part 23. The part 23 extracts the information on a code part in the same procedure. In such a manner, even a text having plural grammatical rules can be converted by a single path and the converting time is shortened.

Description

[Detailed Description of the Invention] <Industrial Application Field> The present invention relates to an improvement in a translator that extracts desired information by transverse text analysis of regular input text and converts it into another text. be.

<Prior Art> A translator is used to convert an input text into another text, and a compiler is also considered to be a type of translator. The configuration of such a translator is shown in FIG. 5. In FIG. 5, the translator 1 is composed of a lexical analyzer 2, a syntactic analyzer 3, and a semantic analyzer 4.

The input text 5 is first decomposed into a plurality of tokens by the lexical analysis unit 2, and then a syntax tree indicating the relationship between these tokens is generated by the lateral sentence analysis t163. This generated syntax tree is input to the semantic analysis section 4 to extract necessary information or to convert it into other text. 1 lexical analysis unit 2 that can obtain the output text 6 in this way
The syntax analysis unit 3 differs depending on the rule system of the input text 1, and generates and uses the optimal one. That is, these are composed of a plurality of functions, and the lexical analysis section 2 and the syntactic analysis section 3 are generated by defining these plural intervals from the token definition and rule definition defined by the user.

<Problems to be Solved by the Invention> However, such translators are only intended for input text having one type of initial syntax. Therefore, in order to convert an input text consisting of multiple rule systems, it is necessary to combine the rule systems to obtain a new rule system and generate a lexical analyzer and a syntactic analyzer based on this new rule system, or to convert an input text consisting of multiple rule systems. They had to use translators and input the input text into these translators one after another to convert it. However, the former method has the problem that the rule system becomes complicated and there are cases where it is impossible to combine the rules into one rule system. Furthermore, the latter method has the problem of taking a long processing time.

<Object of the invention> The object of the invention is to provide a translator that can support multiple rule systems.

Means for Solving the Problems> In order to solve the above problems, the present invention includes a lexical analysis unit that decomposes an input text into tokens, a syntactic analysis unit that determines a syntax tree indicating the relationship between the decomposed tokens, This is a translator that uses this syntax tree to extract necessary information in a semantic analysis section and performs conversion processing, and has a plurality of syntax analysis sections corresponding to different grammar rules. A syntactic analysis unit that is suitable for the input text is selected from among the syntactic analysis units, and the token is input to the selected syntactic analysis unit.

<Embodiment> FIG. 1 shows the configuration of a translator according to the present invention. This example is a translator that extracts various information from text written in C language. This text consists of a header section in which an explanation of the module's functions, etc. is written in a C language comment format, and a code section written based on the C language grammar. The header section and the code section have different grammar rules. In Figure 1, 2
Reference numeral 1 denotes a syntax analysis unit, which is called by the control unit 20 that controls the entire translator. 22 is a lexical analysis section, which is called from the syntactic analysis section 21.

23 is a syntax analysis unit, which is called by the control unit 20. 24 is a lexical analysis section, which is called from the syntactic analysis section 23. 25 is a semantic analysis section, which is called by the control section 20. 26 is a table, and a syntax analysis unit 21.2
The addresses of the functions that make up 3 are stored. This table 26 is rewritten by the semantic analysis section 25 and referred to by the control section 20. The input text 5 is input to a lexical analyzer 22.24. #I sentence analysis section 21 and lexical analysis section 22 correspond to the header section of input text 5, and syntax analysis section 23 and lexical analysis section 24 correspond to the code section.

Next, the operation of this embodiment will be explained based on FIG. 2.

In FIG. 2, 27 is an input text, and a header section and a code section appear in this order.

28 is the translator shown in FIG. First, the address of the function corresponding to the syntax analysis section 21 is stored in the table 26. The control unit 20 refers to this table,
Call the syntax analysis unit 21.

The syntactic analysis unit 21 calls the lexical analysis unit 22, which inputs the input text 27 character by character and compares it with the tokens held therein to divide the input text into a plurality of tokens.

This token is input to the syntax analysis unit 21, and a syntax tree is generated. The syntax tree generated by the syntax analysis section 21 is sent to the semantic analysis section 25. The semantic analysis unit 25 extracts necessary information from the syntax tree and outputs it.

29 represents information extracted in this way.

In this way, information on the header portion of the input text 27 is extracted. When the sentence 30 in the header section is input to the semantic analysis section 25, the semantic analysis section 25 recognizes that the header section has ended, and sets the address of the function corresponding to the syntax analysis section 23 in the table 26. At this stage, the control unit 20 refers to the table 26 and calls the syntax analysis unit 23. From then on, the syntax analysis unit 23 extracts information about the code portion using the same procedure. The fact that the sentence 30 is the end of the header section is set in advance in the semantic analysis section 25. The control section 20 refers to the table 26 for each sentence and determines which syntax analysis section to call.

Next, the operation of the translator will be explained in more detail based on the flowchart of FIG. First, the control unit 20 takes out the address of the function from the table 26 and sets it. As a result, either the parsing unit 21 or 23 is determined. In this explanation, it is assumed that syntactic analysis unit 2] is selected. Next, the syntax analysis section 21 calls the lexical analysis section 22, and the lexical analysis section 22 inputs one character of the input text.

At this stage, when the end of the input text file is reached, the process ends. If it is not the end of the file, the lexical analyzer 2
2 matches the tokens contained therein with the input characters, and if they match, outputs them as tokens to the syntax analysis unit 21. If it does not match, enter one more character and perform the same process. The syntax analysis section 21 generates a syntax tree from the input tokens, checks for syntax errors, and outputs the syntax tree to the semantic analysis section 25. In this case, if the following token is required due to grammar, r g e
Input the next token from "t, J or rnext" and continue the syntax tree generation process. When a syntax error occurs, input up to the if& of the sentence and output an error message. 11 If there are no sentence errors, the syntax tree for one sentence is semantically analyzed by the semantic analysis unit 25 and necessary information is extracted.
Also, it is determined whether there is a need to switch the syntactic analysis section, that is, whether it is the end of the ladder section. If it is the terminal part, the address of the function of the other parsing part is set in the table 26. That is, semantic analysis is performed in units of one sentence, and it is determined whether or not it is necessary to switch the syntax analysis section. If switching is necessary, the address of the function of the corresponding syntax analysis section is set in the table 26. Also, at the beginning of the next sentence, the table 26 is referred to to determine which parser to select. Therefore, even if sentences belonging to different grammar rules are mixed in the input text, they can be automatically and continuously converted.

Next, the operation of the syntax analysis units 21 and 23 will be explained based on FIG. FIG. 4(A) shows a part of the grammar of the syntactic analysis unit, and is composed of nodes of "rlist" and reXprJ. rlist" node is reXprJ and "\rz
(line feed code), and the reXprJ node consists of six rules. [expr −+−expr] indicates that two expr and three elements of ten are arranged in Ili, and the following three rules have the same meaning. [
ID] represents an alphabetic character, and [NUM] represents a token composed of numbers. Token "a" is input first. This token is rlist, 1 node to reXp
rJ node, [Recognized by ID code and reXprJ
The zero-order token "+" returned to ``+'' also goes through the same route,
It is checked against the rule of the reXprJ node and recognized as the first rule.

Since the next token "b" is known to be reXprJ from the rules, it is recognized as [ID].

Furthermore, the next token is the newline code "\n", so rl
ist” node. Following these processes,
A syntax tree showing the relationship between each token is generated as shown in (B). Although the actual syntax analysis unit is more complicated than this example, it generates a syntax tree using a similar procedure.

In the embodiment shown in FIG. 1, the syntactic analysis units 21 and 23 call different lexical analysis units, but if the lexical analysis routines are the same, they can be used in common.

Further, in this embodiment, the case of the C language has been described, but the same structure can be applied to the case of other languages.

The input text may be composed of multiple languages.

<Effects of the Invention> As described above in detail based on the embodiments, the translator of the present invention has a plurality of syntactic analysis units corresponding to different grammar rules, and the translator of the invention The syntax analysis section can now be switched. For that reason,
Even text having multiple grammar rules can be converted in one bus, which has the effect of shortening conversion time.

Additionally, since multiple grammars can be handled at the same time, the grammar itself can be modularized, making it easier to respond to expansions and modifications.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the configuration of a translator according to the present invention,
Figures 2 and 3 are diagrams for explaining its operation, Figure 4 is a diagram for explaining the operation of the syntax analysis section, and Figure 5 is a diagram for explaining the configuration of a conventional translator. . 20...Control unit, 21.23...Syntax analysis unit, 22
゜24...Lexical analysis unit, 25...Semantic analysis unit, 26
···table. Figure (A) Procedure Ne... Correct: C) (Form) 1. Indication of the case 2. Name of the invention 3. Name of the address of the person making the amendment in relation to the case 4. Address of the agent -287444 Translator

Claims (1)

[Claims] A lexical analyzer that reads input text character by character and decomposes it into tokens, and multiple syntaxes corresponding to different grammar rules that receive the tokens decomposed by this lexical analyzer and generate a syntax tree that shows the relationships between these tokens. an analysis section, a semantic analysis section that receives the syntax tree generated by the syntax analysis section, extracts necessary information, and performs conversion processing; and a control section that controls the lexical analysis section, syntactic analysis section, and semantic analysis section. A translator characterized in that the control section switches and controls the syntactic analysis section according to a request from the semantic analysis section.