some work on arrays and reading from stdin

7 files changed, 162 insertions, 43 deletions

diff --git a/minie/DESIGN b/minie/DESIGN
index 5a6d15d..beb46de 100644
--- a/minie/DESIGN
+++ b/minie/DESIGN
@@ -44,7 +44,7 @@ Steps:
 - N: new language we want to have a compiler for
 - N', N'': subset languages of language N with reduced features
 
-Step 1: Build a translator from N'' -> O'' written in O, O', O''
+Step 1: Build a translator (a transpiler) from N'' -> O'' written in O, O', O''
 
 We also use the O-toolchain for building all artifacts (compiler, assembler, linker).
 Try to build minimal subsets of N and use as little features of O for
diff --git a/minie/README b/minie/README
index fe0ebc5..e7ea35d 100644
--- a/minie/README
+++ b/minie/README
@@ -1,14 +1,21 @@
-gcc -g -O0 -Wall -pedantic -std=c89 -o e2c e2c.c
-./e2c < test1.e 
-./e2c < test2.e 
-./e2c < test3.e 
-./e2c < test4.e 
-
-./e2c < test4.e > test4.c && gcc -o test4 test4.c && ./test4
-
-./e2c < ec.e
-./e2c < ec.e > ec.c && gcc -o ec ec.c
-./ec < test1.e > test1.asm
-nasm -o test1.bin -f bin test1.asm
-ndisasm -b32 test1.bin
-../crenshaw/emul test1.bin
+Transpiler
+----------
+
+e2c.c : written in C89, translates E code to C89 code. Compile to
+        a running binary with any C toolchain
+
+Stage 0 E compiler
+------------------
+
+ec.e : stage 0 E compiler
+
+testX.e : test cases for stage 0 compiler
+
+Test cases for the language E
+-----------------------------
+
+test1: empty module
+test2: syntax error, no module definition
+test3: module definition lacking a module name
+test4: general expressions and variables
+test5: basic system input/output, import of pseudo module 'system'
diff --git a/minie/build.sh b/minie/build.sh
new file mode 100755
index 0000000..3dc1339
--- /dev/null
+++ b/minie/build.sh
@@ -0,0 +1,17 @@
+#!/bin/bash
+
+gcc -g -O0 -Wall -pedantic -std=c89 -o e2c e2c.c
+./e2c < test1.e
+./e2c < test2.e
+./e2c < test3.e
+./e2c < test4.e
+./e2c < test5.e
+
+./e2c < test5.e > test5.c && gcc -g -o test5 test5.c && ./test5
+
+./e2c < ec.e
+./e2c < ec.e > ec.c && gcc -o ec ec.c
+./ec < test1.e > test1.asm
+nasm -o test1.bin -f bin test1.asm
+ndisasm -b32 test1.bin
+../crenshaw/emul test1.bin
diff --git a/minie/e2c.c b/minie/e2c.c
index f1ba44b..438dbe4 100644
--- a/minie/e2c.c
+++ b/minie/e2c.c
@@ -5,7 +5,7 @@
 #include <assert.h>
 
 enum {
-	MAX_IDENT_LEN = 16,
+	MAX_IDENT_LEN = 24,
 	MAX_NUMBER_LEN = 9,
 	MAX_ERRMSG_LEN = 64,
 	MAX_STRING_LEN = 64
@@ -21,6 +21,7 @@ static char moduleName[MAX_IDENT_LEN+1];
 static char varName[MAX_IDENT_LEN+1];
 static char typeName[MAX_IDENT_LEN+1];
 static char procName[MAX_IDENT_LEN+1];
+static int isArray = 0;
 
 typedef enum {
 	S_char = 0,
@@ -36,6 +37,8 @@ typedef enum {
 	S_if,
 	S_do,
 	S_else,
+	S_array,
+	S_of,
 	S_dot,
 	S_comma,
 	S_semicolon,
@@ -47,6 +50,8 @@ typedef enum {
 	S_slash,
 	S_lparen,
 	S_rparen,
+	S_lbracket,
+	S_rbracket,
 	S_eof
 } Symbol;
 
@@ -64,6 +69,8 @@ char *symname[S_eof+1] = {
 	"if",
 	"do",
 	"else",
+	"array",
+	"of",
 	".",
 	",",
 	";",
@@ -75,6 +82,8 @@ char *symname[S_eof+1] = {
 	"/",
 	"(",
 	")",
+	"[",
+	"]",
 	"eof"
 };
 
@@ -141,6 +150,7 @@ static int isCharacter( int c )
 		case ' ':
 		case '[':
 		case ']':
+		case ':':
 		/* TODO: allow more characters as we go along */
 			return 1;
 		default:
@@ -235,9 +245,14 @@ static Symbol getSym( void )
 		case '7':
 		case '8':
 		case '9':
-		case 'a':
 			number( );
 			return S_number;	
+		case 'a':
+			identifier( );
+			if( strcmp( ident, "array" ) == 0 ) {
+				return S_array;
+			}
+			return S_ident;
 		case 'b':
 			identifier( );
 			if( strcmp( ident, "begin" ) == 0 ) {
@@ -282,7 +297,14 @@ static Symbol getSym( void )
 			}
 			return S_ident;
 		case 'n':
+			identifier( );
+			return S_ident;
 		case 'o':
+			identifier( );
+			if( strcmp( ident, "of" ) == 0 ) {
+				return S_of;
+			}
+			return S_ident;			
 		case 'p':
 			identifier( );
 			if( strcmp( ident, "procedure" ) == 0 ) {
@@ -369,6 +391,12 @@ static Symbol getSym( void )
 		case ')':
 			look = getChar( );
 			return S_rparen;
+		case '[':
+			look = getChar( );
+			return S_lbracket;
+		case ']':
+			look = getChar( );
+			return S_rbracket;
 		case EOF:
 			return S_eof;
 		default:
@@ -588,17 +616,31 @@ static void statement( void )
 		qualident( );
 		if( sym == S_lparen ) {
 			/* TODO: add internal function maps as symbols and procedures */
-			if( strcmp( varName, "system.writeln" ) == 0 ) {
-				emit( "printf( " );
+			if( strcmp( varName, "system.writestring" ) == 0 ||
+			    strcmp( varName, "system.writeline" ) == 0 ) {
+				emit( "printf( \"%%s\", " );
+			} else if( strcmp( varName, "system.writeinteger" ) == 0 ) {
+				emit( "printf( \"%%d\", " );
 			} else if( strcmp( varName, "system.halt" ) == 0 ) {
 				emit( "exit( " );
+			} else if( strcmp( varName, "system.readline" ) == 0 ) {
+				emit( "{ size_t _n = %d; fgets( (char *)&", num );
+			} else if( strcmp( varName, "length" ) == 0 ) {
+				emit( "255" );
 			} else {
 				emit( "%s( ", varName );
 			}
 			parameterList( );
 			/* TODO: procedure call */
-			emitLn( ");" );
-			if( strcmp( varName, "system.writeln" ) == 0 ) {
+			if( strcmp( varName, "system.readline" ) == 0 ) {
+				/* TODO: check if parameter is an array of char,
+				 get length of the defined array and put it into the parameter
+				 of getline */
+				emitLn( ", _n, stdin ); }" );
+			} else {
+				emitLn( ");" );
+			}
+			if( strcmp( varName, "system.writeline" ) == 0 ) {
 				emitLn( "puts( \"\" );" );
 			}
 		} else {
@@ -626,26 +668,58 @@ static void statementBlock( void )
 	Expect( S_end );
 }
 
+static void type( void );
+
+static void simpleType( void )
+{
+	/* TODO: handle build-in types in init, register them in
+	   type table */
+	if( strcmp( ident, "integer" ) == 0 ) {
+		strcpy( typeName, "signed int" );
+		sym = getSym( );
+	} else if( strcmp( ident, "boolean" ) == 0 ) {
+		/* C89, no bool */
+		strcpy( typeName, "unsigned char" );
+		sym = getSym( );
+	} else if( strcmp( ident, "char" ) == 0 ) {
+		/* TODO: Unicode, for now ASCII */
+		strcpy( typeName, "unsigned char" );
+		sym = getSym( );
+	} else if( strcmp( ident, "byte" ) == 0 ) {
+		/* C89, no bool */
+		strcpy( typeName, "unsigned char" );
+		sym = getSym( );			
+	} else {
+		Abort( "Unknown type '%s'", ident );
+	}
+}
+
+static void arrayType( void )
+{
+	sym = getSym( );
+	Expect( S_lbracket );
+	number( );
+	sym = getSym( );
+	Expect( S_rbracket );
+	/* array size is in num */
+	if( sym == S_of ) {
+		sym = getSym( );
+		type( );
+		/* type is in ident */
+	} else {
+		Abort( "of' expected in array definition" );			
+	}
+}
+
 static void type( void )
 {
 	identifier( );
 	if( sym == S_ident ) {
-		/* TODO: handle build-in types in init, register them in
-		   type table */
-		if( strcmp( ident, "integer" ) == 0 ) {
-			strcpy( typeName, "signed int" );
-			sym = getSym( );
-		} else if( strcmp( ident, "bool" ) == 0 ) {
-			/* C89, no bool */
-			strcpy( typeName, "unsigned char" );
-			sym = getSym( );
-		} else if( strcmp( ident, "byte" ) == 0 ) {
-			/* C89, no bool */
-			strcpy( typeName, "unsigned char" );
-			sym = getSym( );			
-		} else {
-			Abort( "Unknown type '%s'", ident );
-		}
+		simpleType( );
+		isArray = 0;
+	} else if( sym == S_array ) {
+		arrayType( );
+		isArray = 1;
 	}
 }
 
@@ -655,7 +729,11 @@ static void variableDeclaration( void )
 	sym = getSym( );
 	Expect( S_colon );
 	type( );
-	emitLn( "static %s %s;", typeName, varName );
+	if( isArray ) {
+		emitLn( "static %s %s[%d];", typeName, varName, num );
+	} else {
+		emitLn( "static %s %s;", typeName, varName );
+	}
 }
 
 static void variableBlock( void )
diff --git a/minie/ec.e b/minie/ec.e
index b51559b..272cc13 100644
--- a/minie/ec.e
+++ b/minie/ec.e
@@ -4,14 +4,14 @@ import system;
 
 procedure prologue;
 begin
-	system.writeln( "[bits 32]" );
-	system.writeln( "cpu 486" );
-	system.writeln( "org 0x1000000" );
+	system.writeline( "[bits 32]" );
+	system.writeline( "cpu 486" );
+	system.writeline( "org 0x1000000" );
 end
 
 procedure epilogue;
 begin
-	system.writeln( "hlt" );
+	system.writeline( "hlt" );
 end
 
 begin
diff --git a/minie/test4.e b/minie/test4.e
index 0ad93e0..2dc3722 100644
--- a/minie/test4.e
+++ b/minie/test4.e
@@ -4,7 +4,7 @@ var
 	x : integer;
 	y : integer;
 	z : integer;
-	b : bool;
+	b : boolean;
 	c : byte;
 	res : integer;
 	
diff --git a/minie/test5.e b/minie/test5.e
new file mode 100644
index 0000000..1b4e991
--- /dev/null
+++ b/minie/test5.e
@@ -0,0 +1,17 @@
+module test4;
+
+import system;
+
+var
+	s : array[64] of char;
+	len : integer;
+	
+begin
+	system.readline( s );
+	len := len( s );
+	system.writestring( "String: " );
+	system.writeline( s );
+	system.writestring( "Length: " );
+	system.writeinteger( len );
+	system.writeline( "" )
+end