com.ph.pr.parser
Class ASTBuilder

java.lang.Object
  com.ph.pr.parser.ASTBuilder

All Implemented Interfaces:

goldengine.java.GPMessageConstants

public class ASTBuilder
extends java.lang.Object
implements goldengine.java.GPMessageConstants

A singleton class providing capabilities for creating a class hierarchy for an AST for a supplied grammar, and creating the tree of AST objects, each one corresponding to a non-terminal, upon parsing. This is based on GOLD Parsing System and its Java engine .

Author:

" Greg Golberg "

See Also:

parse(ClassLoader, String, String), generateClasses(String, String, String, String)

Field Summary

static java.lang.String	AST_SUFFIX
private java.util.Map	ast2Comments
private boolean	caseInsensitive
private java.util.Map	childMap
private ASTNode	curAst Current AST node (eventually this is returned by parse(ClassLoader, String, String)).
private java.lang.String	curComment
static java.lang.String	GOLD_NON_TERMINAL_END
static java.lang.String	GOLD_NON_TERMINAL_START
static java.lang.String	GOLD_PROD_SYMB
private java.lang.String	grammar
private boolean	grammarLoaded
private java.lang.String	input
private static ASTBuilder	instance The singleton instance
private java.io.PrintWriter	out
private goldengine.java.GOLDParser	parser
private java.util.Map	prodMap Map of namesof rule non-terminalsto an ArrayList of productions, as Strings.
private java.util.Map	red2Ast Map of Reductionto corresponding ASTNode.
private java.util.Map	terminalMap

Fields inherited from interface goldengine.java.GPMessageConstants

gpMsgAccept, gpMsgCommentError, gpMsgCommentLineRead, gpMsgInternalError, gpMsgLexicalError, gpMsgNotLoadedError, gpMsgReduction, gpMsgSyntaxError, gpMsgTokenRead

Constructor Summary

private

ASTBuilder()

Method Summary

void	generateClasses(java.lang.String root, java.lang.String pkg, java.lang.String visitorPkg, java.lang.String visitorName) Generate the classes implementing ASTNodefor each non-terminal.
java.util.Collection	getAllAsts()
java.util.Map	getAst2Comments()
ASTNode	getStart()
java.util.Map	getTerminalMap()
private boolean	isNonTerminal(java.lang.String s)
static void	main(java.lang.String[] argv)
private java.lang.String	nonTerminalName(java.lang.String s)
ASTNode	parse(java.lang.ClassLoader loader, java.lang.String pkg, java.lang.String input) Parse the input.
static java.lang.String	pkg2Dir(java.lang.String pkg)
private void	processReduction(java.lang.ClassLoader myLoader, java.lang.String pkg, goldengine.java.Reduction reduction)
private void	readRules() Read the GOLD's .cgt file (provided in #setGrammar(String)) and initialize childMapand prodMap.
void	setGrammar(java.lang.String g, boolean force) Loads the grammar.
static ASTBuilder	singleton()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

AST_SUFFIX

public static final java.lang.String AST_SUFFIX

See Also:

Constant Field Values

instance

private static ASTBuilder instance

The singleton instance

GOLD_NON_TERMINAL_START

public static final java.lang.String GOLD_NON_TERMINAL_START

See Also:

Constant Field Values

GOLD_NON_TERMINAL_END

public static final java.lang.String GOLD_NON_TERMINAL_END

See Also:

Constant Field Values

GOLD_PROD_SYMB

public static final java.lang.String GOLD_PROD_SYMB

See Also:

Constant Field Values

curComment

private java.lang.String curComment

childMap

private final java.util.Map childMap

prodMap

private final java.util.Map prodMap

Map of namesof rule non-terminalsto an ArrayList of productions, as Strings.

out

private java.io.PrintWriter out

grammar

private java.lang.String grammar

input

private java.lang.String input

parser

private final goldengine.java.GOLDParser parser

grammarLoaded

private boolean grammarLoaded

red2Ast

private java.util.Map red2Ast

Map of Reductionto corresponding ASTNode.

ast2Comments

private java.util.Map ast2Comments

curAst

private ASTNode curAst

Current AST node (eventually this is returned by parse(ClassLoader, String, String)).

terminalMap

private java.util.Map terminalMap

caseInsensitive

private boolean caseInsensitive

Constructor Detail

ASTBuilder

private ASTBuilder()

Method Detail

singleton

public static ASTBuilder singleton()

Returns:

the singleton ASTBuilder.

nonTerminalName

private java.lang.String nonTerminalName(java.lang.String s)

isNonTerminal

private boolean isNonTerminal(java.lang.String s)

readRules

private void readRules()

Read the GOLD's .cgt file (provided in #setGrammar(String)) and initialize childMapand prodMap.

setGrammar

public void setGrammar(java.lang.String g,
                       boolean force)
                throws goldengine.java.ParserException

Loads the grammar.

Parameters:

g - path to GOLD compiled grammar (.cgt file).

force - if true, the grammar is reloaded, otherwise, if the grammar points to the same one loaded previously, it is ignored.

Throws:

goldengine.java.ParserException - whenever {GOLDParser#loadCompiledGrammar} does.

See Also:

GOLDParser.loadCompiledGrammar(String)

generateClasses

public void generateClasses(java.lang.String root,
                            java.lang.String pkg,
                            java.lang.String visitorPkg,
                            java.lang.String visitorName)
                     throws goldengine.java.ParserException,
                            java.io.IOException

Generate the classes implementing ASTNodefor each non-terminal. They will be named by the name of the non-terminal with a suffix AST_SUFFIXand structure as explained below. The parse(ClassLoader, String, String)method will instantiate and populate appropriate objects from this hierarchy during parsing, using setters as described here.

Thus, for the following production

 
  
   
    
     
      
       
        
         
          
           
            
             
              <ext_statement> ::= 
                        SUB tkName <formal_list_opt> tkEol <statement_list> END SUB tkEol
                |       DECLARE FUNCTION tkName LIB tkString <formal_list_opt> <as_opt> tkEol
                |
              
             
            
           
          
         
        
       
      
     
    
   
  
 

a ext_statement_AST class will be generated, containing the following:

• For each production, a public static final String field, named PROD_n (where n is 1, 2, 3, ...). In our example, there will be 3 such fields, PROD_1 will be set to "SUB tkName tkEol END SUB tkEol", PROD_3 to "" (empty string), and what PROD_2 will be is left as an exercise.

• A private String field production (and an accessor, ASTNode.getProduction(), which will be set by parse(ClassLoader, String, String)to one of the PROD_n constants, as above.

• For each non-terminal appearing on the right-hand side, a private field of that type, with prefix inst_ and suffix-- since there will be a class generated for that non-terminal, in turn, and accessors (getters and setters). Depending on the actual reduction, not all such fields, obviously, will be set. In our example, thus, one of the fields will be private formal_list_opt_AST inst_formal_list_opt (and its accessors will be set_formal_list_opt_AST(ASTNode) and get_formal_list_opt_AST(). Another such field will be inst_as_opt_AST.

  It's easy to see that if an instance of ext_statement_AST is created as a result of first reduction ( SUB tkName..., or, in other words, when getProduction() returns a String that's equal to PROD_1) the inst_as_opt_AST will not be set; while inst_formal_list_opt will be set in cases of both PROD_1 and PROD_2.

  Notice that:

  1. The _inst prefix not part of the accessor name.
  2. The setter takes ASTNode, rather than that particular type (and downcasts inside). This is to make my life easier.

• A private List terminals field (with a getTerminals() accessor) that will return the values of all terminals in the current reduction, in order. Thus, in or example, in case of PROD_2, this field will contain 6 elements:
  1. "DECLARE"
  2. "FUNCTION",
  3. Whatever the tkName token was
  4. "LIB"
  5. tkString token was
  6. tkEol token was

In addition to the ASTNode hierarchy, a skeleton for a class extending ReflectiveVisitorwill be generated. It will provide a visit method for every ASTNode subclass that was generated. Using our example above, one such method will be:

 public void visit(ext_statement_AST node) {
     // ext_statement ::= SUB tkName <formal_list_opt> tkEol
     // <statement_list> END SUB tkEol
     if (node.getProduction().equals(ext_statement_AST.PROD_1)) {
         String curStr = "";
 
         // TERMINAL: [SUB]
         curStr += (String) node.getTerminals().get(0);
         // TERMINAL: [tkName]
         curStr += (String) node.getTerminals().get(1);
         node.get_formal_list_opt_AST().accept(this);
 
         // TODO
         curStr = this.code;
         this.code = "";
 
         // TERMINAL: [tkEol]
         curStr += (String) node.getTerminals().get(2);
         node.get_statement_list_AST().accept(this);
 
         // TODO
         curStr = this.code;
         this.code = "";
 
         // TERMINAL: [END]
         curStr += (String) node.getTerminals().get(3);
 
         // TERMINAL: [SUB]
         curStr += (String) node.getTerminals().get(4);
 
         // TERMINAL: [tkEol]
         curStr += (String) node.getTerminals().get(5);
         return;
     }
 
     // ext_statement ::= DECLARE FUNCTION tkName LIB tkString
     // <formal_list_opt> <as_opt> tkEol
     if (node.getProduction().equals(ext_statement_AST.PROD_2)) {
         String curStr = "";
 
         // TERMINAL: [DECLARE]
         curStr += (String) node.getTerminals().get(0);
 
         // TERMINAL: [FUNCTION]
         curStr += (String) node.getTerminals().get(1);
 
         // TERMINAL: [tkName]
         curStr += (String) node.getTerminals().get(2);
 
         // TERMINAL: [LIB]
         curStr += (String) node.getTerminals().get(3);
 
         // TERMINAL: [tkString]
         curStr += (String) node.getTerminals().get(4);
 
         node.get_formal_list_opt_AST().accept(this);
 
         // TODO
         curStr = this.code;
         this.code = "";
         node.get_as_opt_AST().accept(this);
 
         // TODO
         curStr = this.code;
         this.code = "";
         // TERMINAL: [tkEol]
         curStr += (String) node.getTerminals().get(5);
         return;
     }
 
     // ext_statement ::=
     if (node.getProduction().equals(ext_statement_AST.PROD_3)) {
         String curStr = "";
         return;
     }
 }
 

This visitor can then be modified.

TODO

1. If rhs is a single non-terminal, just visit that one, without touching this.code
2. If the production is empty, do nothing
3. BUG: how to handle same names in a rhs? E.g.: ::= FOR TO ? For now, introduce Nodes list.
4. Generate rudimentary Javadoc with the code.

Parameters:

root - path where the pkg containing generated classes is to be created

pkg - package into which to place generated classes; if it exists, its contents will be deleted prior to the generation.

visitorPkg - package into which a subclass of ReflectiveVisitoris generated -- different from pkg in that it's something to be modified/customized by a developer, rather than to be used as generated.

visitorName - a name of the subclass of ReflectiveVisitorto be generated.

Throws:

goldengine.java.ParserException - if an error occurs

java.io.IOException

pkg2Dir

public static final java.lang.String pkg2Dir(java.lang.String pkg)

getAllAsts

public java.util.Collection getAllAsts()

parse

public ASTNode parse(java.lang.ClassLoader loader,
                     java.lang.String pkg,
                     java.lang.String input)
              throws goldengine.java.ParserException

Parse the input.

Parameters:

loader - ClassLoader to use to load classes implementing ASTNode that were generated by #generateClasses(String, String)(useful if not running in an IDE, etc.); if null, then Thread.getContextClassLoader()of the Thread.currentThread()is used.

pkg - package where the classes implementing ASTNode that were generated by #generateClasses(String, String) reside.

input - input to parse.

Returns:

the ASTNode corresponding to the last ( <start> production).

Throws:

goldengine.java.ParserException - if there were any errors in parsing

getAst2Comments

public java.util.Map getAst2Comments()

main

public static void main(java.lang.String[] argv)

Parameters:

argv - In the following order:

1. E.g.,

getTerminalMap

public java.util.Map getTerminalMap()

processReduction

private void processReduction(java.lang.ClassLoader myLoader,
                              java.lang.String pkg,
                              goldengine.java.Reduction reduction)
                       throws goldengine.java.ParserException

Throws:

goldengine.java.ParserException

getStart

public ASTNode getStart()