From: Mike D. Lowis <mike@mdlowis.com>
Date: Tue, 29 May 2012 18:49:00 +0000 (-0400)
Subject: Updated parser to support all core forms except quote
X-Git-Url: https://git.mdlowis.com/?a=commitdiff_plain;h=7ea1e0f83312fe3ea569831f8788e6efd998479a;p=archive%2Fdlang.git

Updated parser to support all core forms except quote
---

diff --git a/example.dl b/example.dl
index bf1ee18..6411a0e 100644
--- a/example.dl
+++ b/example.dl
@@ -15,12 +15,34 @@ foo()
 foo(1)
 foo(1 2)
 foo(1 2 3)
-foo(1,2,3)
-(foo . $bar)(1, 2, 3)
+(foo . $bar)(1 2 3)
 
 # Definition and assignment
-define foo 5
-set! foo 6
+define foo 5;
+set! foo 6;
+
+# If statement
+if conditional
+    if_branch
+#else
+    else_branch
+end
+
+if conditional
+    if_branch
+end
+
+# Begin block
+begin end
+
+begin
+    foo()
+end
+
+begin
+    foo()
+    bar()
+end
 
 # Lambda expressions
 lambda () end
@@ -44,45 +66,39 @@ lambda (a b c)
     foo(a b c)
 end
 
-# Begin block
-begin end
+# Macro expressions
+macro () end
+macro (a) end
+macro (a b) end
+macro (a b c) end
 
-begin
-    foo()
+macro ()
+    foo(a)
 end
 
-begin
-    foo()
-    bar()
+macro (a)
+    foo(a)
 end
 
-# If statement
-if conditional
-    if_branch
-else
-    else_branch
+macro (a b)
+    foo(a b)
 end
 
-if conditional
-    if_branch
+macro (a b c)
+    foo(a b c)
 end
 
-## Define an infix operator (No Precedence)
-#infix =
-#
-## Define dummy keywords (Whitespace)
-#keyword ;
-#
-#foo = (1 + 1) ;
-
-# Macros
-macro let ( := = ) ;
-    ( a := b )
-        define a b
+# Quoting Expressions
+#quote( (1 + 1) )
 
-    ( a = b )
-        set! a b
-end
+# Syntactic Extensions
+#syntax let ( := = ) ;
+#    ( a := b )
+#        define a b ;
+#
+#    ( a = b )
+#        set! a b ;
+#end
 
-let foo := "bar" ;
-let foo = 5 ;
+#let foo := "bar" ;
+#let foo = 5 ;
diff --git a/source/dllexer/dllexer.cpp b/source/dllexer/dllexer.cpp
index 98aebb3..7eaa420 100644
--- a/source/dllexer/dllexer.cpp
+++ b/source/dllexer/dllexer.cpp
@@ -97,13 +97,6 @@ Token DLLexer::next(void)
             Symbol(ret);
         }
 
-        // Consume a comma
-        else if(lookahead(1) == ',')
-        {
-            ret = Token( COMMA, ",", line, column );
-            consume();
-        }
-
         // Consume parentheses
         else if (lookahead(1) == '(')
         {
@@ -127,12 +120,13 @@ Token DLLexer::next(void)
 
             Id(ret);
         }
+    }
 
-        if( !escaped && (ret.text().compare( terminator_string ) == 0) )
-        {
-            ret.type( TERM );
-        }
+    if( !escaped && (ret.text().compare( terminator_string ) == 0) )
+    {
+        ret.type( TERM );
     }
+    cout << terminator_string << " " << escaped << " " << ret.type() << " " << ret.text() << "|"<< endl;
 
     return ret;
 }
diff --git a/source/dllexer/dllexer.h b/source/dllexer/dllexer.h
index 5b7a953..391e5fb 100644
--- a/source/dllexer/dllexer.h
+++ b/source/dllexer/dllexer.h
@@ -6,30 +6,34 @@
 
 typedef enum TokenTypes
 {
-    // Symbols
+    // Core Forms
     PROGRAM    = 0,
     DEFINE     = 1,
     ASSIGN     = 2,
-    LAMBDA     = 3,
+    IF         = 3,
     BEGIN      = 4,
-    IF         = 5,
-    APPLY      = 6,
-    ID_LIST    = 7,
-    EXP_LIST   = 8,
-    MACRO      = 9,
-    TRANSFORM  = 10,
-    MACRO_APP  = 11,
-    LPAR       = 12,
-    RPAR       = 13,
-    COMMA      = 14,
-    TERM       = 15,
+    QUOTE      = 5,
+    LAMBDA     = 6,
+    MACRO      = 7,
+    SYNTAX     = 8,
+
+    // Punctuation and Symbols
+    LPAR       = 10,
+    RPAR       = 11,
+    TERM       = 12,
+
+    // Actions and Virtual nodes
+    APPLY      = 20,
+    EXPAND     = 21,
+    TRANSFORM  = 22,
+    LIST       = 23,
 
     // Datatypes
-    ID         = 16,
-    NUM        = 17,
-    CHAR       = 18,
-    STRING     = 19,
-    SYMBOL     = 20,
+    ID         = 30,
+    NUM        = 31,
+    CHAR       = 32,
+    STRING     = 33,
+    SYMBOL     = 34,
 } eTokenTypes;
 
 typedef struct {
diff --git a/source/dlparser/dlparser.cpp b/source/dlparser/dlparser.cpp
index 25a1b24..63bef51 100644
--- a/source/dlparser/dlparser.cpp
+++ b/source/dlparser/dlparser.cpp
@@ -9,10 +9,12 @@ DLParser::DLParser() : BTParser()
 {
     core_forms["define"] = DEFINE;
     core_forms["set!"]   = ASSIGN;
-    core_forms["lambda"] = LAMBDA;
-    core_forms["begin"]  = BEGIN;
     core_forms["if"]     = IF;
+    core_forms["begin"]  = BEGIN;
+    core_forms["quote"]  = QUOTE;
+    core_forms["lambda"] = LAMBDA;
     core_forms["macro"]  = MACRO;
+    core_forms["syntax"] = SYNTAX;
 }
 
 DLParser::~DLParser()
@@ -54,15 +56,17 @@ AST* DLParser::Expression(void)
     AST* ret = NULL;
 
     // Expression := CoreForm
-    //             | FuncApp
     //             | BasicExp
+    //             | FuncApp
     //
     // CoreForm := 'define' ID Expression TERM
     //           | 'set!' ID Expression TERM
-    //           | 'lambda' IdList ExpList? TERM
     //           | 'begin' ExpList* TERM
+    //           | 'quote' ExpList* TERM
     //           | 'if' Expression Expression 'else' Expression? TERM
-    //           | 'macro' ID IdList ID ExpList TERM
+    //           | 'lambda' IdList ExpList? TERM
+    //           | 'macro' IdList ExpList? TERM
+    //           | 'syntax' ID IdList ID ExpList TERM
     //
     // FuncApp := BasicExp '(' ParamList ')'
     //
@@ -90,10 +94,14 @@ AST* DLParser::Expression(void)
     else
     {
         ret = BasicExp();
-        //if ( speculate_ParamList() )
-        //{
-        //    ret + ParamList()
-        //}
+
+        // Traditional Function Application
+        if( lookaheadType(1) == LPAR )
+        {
+            match(LPAR);
+            ret = new AST(APPLY, 2, ret, ExpList(RPAR));
+            match(RPAR);
+        }
     }
 
     // Register any new macros and expand any existing macros
@@ -106,38 +114,48 @@ AST* DLParser::Expression(void)
 AST* DLParser::CoreForm(void)
 {
     AST* ret = NULL;
+    std::string term = ((DLLexer*)lexer)->terminator();
     eTokenTypes form_id = getCoreFormId();
     consume(); // Throw away the form name (we don't need it anymore)
     switch( form_id )
     {
         case DEFINE:
         case ASSIGN:
+            ((DLLexer*)lexer)->terminator(";");
             ret = new AST( lookaheadToken(1) );
             match(ID);
             ret = new AST(form_id, 2, ret, Expression());
             break;
 
-        case LAMBDA:
-            ret = new AST(LAMBDA, 2, IdList(), ExpList(TERM));
-            match(TERM);
-            break;
-
         case BEGIN:
+            ((DLLexer*)lexer)->terminator("end");
             ret = new AST(BEGIN, 1, ExpList(TERM));
-            match(TERM);
             break;
 
         case IF:
+            ((DLLexer*)lexer)->terminator("end");
             ret = new AST(IF, 2, Expression(), Expression());
             if(lookaheadType(1) != TERM)
             {
                 ret->addChild( Expression() );
             }
-            match(TERM);
             break;
 
+        //case QUOTE:
+        //    match(LPAR);
+        //    ret = new AST(QUOTE, 1, Expression());
+        //    ((DLLexer*)lexer)->terminator(")");
+        //    break;
+
+        case LAMBDA:
         case MACRO:
-            ret = new AST(MACRO);
+            ((DLLexer*)lexer)->terminator("end");
+            ret = new AST(form_id, 2, IdList(), ExpList(TERM));
+            break;
+
+        case SYNTAX:
+            ((DLLexer*)lexer)->terminator("end");
+            ret = new AST(SYNTAX);
 
             // Get the macro name
             ret->addChild( new AST( lookaheadToken(1) ) );
@@ -158,13 +176,15 @@ AST* DLParser::CoreForm(void)
                 transform->addChild( Expression() );
                 ret->addChild( transform );
             }
-            match(TERM);
             break;
 
+        case QUOTE:
         default:
             throw Exception( lookaheadToken(1) );
             break;
     }
+    match(TERM);
+    ((DLLexer*)lexer)->terminator( term );
     return ret;
 }
 
@@ -185,7 +205,7 @@ AST* DLParser::BasicExp(void)
         // Register the new terminator
 
         // Consume the name
-        ret = new AST( MACRO_APP, 1, new AST( lookaheadToken(1) ));
+        ret = new AST( EXPAND, 1, new AST( lookaheadToken(1) ));
         consume();
 
         // Consume the expressions
@@ -198,16 +218,6 @@ AST* DLParser::BasicExp(void)
         // Reset the terminator to its old value
     }
 
-    //// Traditional Function Application
-    //else if( (lookaheadType(1) == ID) && (lookaheadType(2) == LPAR) )
-    //{
-    //    ret = new AST( lookaheadToken(1) );
-    //    consume();
-    //    match(LPAR);
-    //    ret = new AST(APPLY, 2, ret, ExpList(RPAR));
-    //    match(RPAR);
-    //}
-
     // Infix Function Application
     else if( lookaheadType(1) == LPAR )
     {
@@ -222,7 +232,7 @@ AST* DLParser::BasicExp(void)
         operand2 = Expression();
         match(RPAR);
 
-        ret = new AST( APPLY, 2, operation, new AST(EXP_LIST, 2, operand1, operand2) );
+        ret = new AST( APPLY, 2, operation, new AST(LIST, 2, operand1, operand2) );
     }
 
     // Literal
@@ -255,23 +265,9 @@ AST* DLParser::Literal(void)
     return ret;
 }
 
-AST* DLParser::ParamList(void)
-{
-    AST* ret = new AST(EXP_LIST);
-    match(LPAR);
-    ret->addChild( Expression() );
-    if( COMMA == lookaheadType(1) )
-    {
-        match(COMMA);
-        ret->addChild( Expression() );
-    }
-    match(RPAR);
-    return ret;
-}
-
 AST* DLParser::ExpList(eTokenTypes term)
 {
-    AST* ret = new AST(EXP_LIST);
+    AST* ret = new AST(LIST);
     while(term != lookaheadType(1))
     {
         ret->addChild( Expression() );
@@ -281,7 +277,7 @@ AST* DLParser::ExpList(eTokenTypes term)
 
 AST* DLParser::IdList(void)
 {
-    AST* ret = new AST(ID_LIST);
+    AST* ret = new AST(LIST);
     match(LPAR);
     while(ID == lookaheadType(1))
     {
diff --git a/source/dlparser/dlparser.h b/source/dlparser/dlparser.h
index d2aa201..4ffb962 100644
--- a/source/dlparser/dlparser.h
+++ b/source/dlparser/dlparser.h
@@ -27,7 +27,6 @@ class DLParser : public BTParser
         AST* FuncApp(void);
         AST* BasicExp(void);
         AST* Literal(void);
-        AST* ParamList(void);
         AST* ExpList(eTokenTypes term);
         AST* IdList(void);
 };
diff --git a/source/visitors/macroprocessor.cpp b/source/visitors/macroprocessor.cpp
index ae631ba..4e1d325 100644
--- a/source/visitors/macroprocessor.cpp
+++ b/source/visitors/macroprocessor.cpp
@@ -14,7 +14,8 @@ void MacroProcessor::afterVisit(AST* cur, int depth)
 
 void MacroProcessor::beforeChildren(AST* cur, int depth)
 {
-    if (cur->type() == MACRO_APP)
+    // If we reached a syntax use then expand it
+    if (cur->type() == EXPAND)
     {
         //expandMacro(cur);
     }
@@ -22,7 +23,8 @@ void MacroProcessor::beforeChildren(AST* cur, int depth)
 
 void MacroProcessor::afterChildren(AST* cur, int depth)
 {
-    if (cur->type() == MACRO)
+    // If we have a new syntax definition then register it
+    if (cur->type() == SYNTAX)
     {
         std::string name = (*(cur->children()->begin()))->text();
         Macro* macro = new Macro();