Merge pull request #27 from PaytonTurnage/master

Implementation for Preprocessor Defines

SYNTAX

@@ -78,9 +78,17 @@ Labels must be specified at the beginning of a line or on their own line.
 // 4. PREPROCESSOR ///////////////////////////////
 //////////////////////////////////////////////////
 
-TVM has a preprocessor that allows other source files to be included with the directive "%include filename.vm"
+TVM's preprocessor works similarly to many C compilers and uses the prefix "%".
 
-During preprocessing, the entire included source file is inserted in place of the %include directive in memory.
+	// I. Include //
+
+		%include filename
+		Pastes all of the contents of filename into the source code before interpretting it.
+
+	// II. Define //
+
+		%define identifier value
+		Define a constant so that all instances of the string "identifier" will be replaced by "value".
 
 //////////////////////////////////////////////////
 // 5. INSTRUCTION LISTING ////////////////////////

include/tvm/tvm_htab.h

@@ -8,6 +8,7 @@ typedef struct tvm_htab_node_s
 {
 	char *key;
 	int value;
+	void *valptr;
 	struct tvm_htab_node_s *next;
 } tvm_htab_node_t;
 
@@ -22,6 +23,8 @@ tvm_htab_t* htab_create();
 void htab_destroy(tvm_htab_t *htab);
 
 int htab_add(tvm_htab_t *htab, const char *key, int value);
+int htab_add_str(tvm_htab_t *htab, const char *key, const void *valptr, int len);
 int htab_find(tvm_htab_t *htab, const char *key);
+char *htab_find_str(tvm_htab_t *htab, const char *key);
 
 #endif

include/tvm/tvm_lexer.h

@@ -4,6 +4,8 @@
 #define MAX_ARGS 2
 #define MAX_TOKENS 4
 
+#include "tvm_htab.h"
+
 typedef struct tvm_lexer_s
 {
 	char **source_lines;
@@ -14,6 +16,6 @@ tvm_lexer_t *lexer_create();
 void lexer_destroy(tvm_lexer_t *l);
 
 /* Tokenize the character array "source" into lines and tokens */
-void lex(tvm_lexer_t *lexer, char *source);
+void lex(tvm_lexer_t *lexer, char *source, tvm_htab_t *defines);
 
 #endif

include/tvm/tvm_preprocessor.h

@@ -1,6 +1,8 @@
 #ifndef TVM_PREPROCESSOR_H_
 #define TVM_PREPROCESSOR_H_
 
-int tvm_preprocess(char *src, int *src_len);
+#include "tvm_htab.h"
+
+int tvm_preprocess(char *src, int *src_len, tvm_htab_t *defines);
 
 #endif

include/tvm/tvm_program.h

@@ -20,6 +20,8 @@ typedef struct tvm_program_s
 	int **values;
 	int num_values;
 
+	tvm_htab_t *defines;
+
 	tvm_htab_t *label_htab;
 
 } tvm_program_t;

libtvm/tvm_htab.c

@@ -25,6 +25,8 @@ void htab_destroy(tvm_htab_t *htab)
 		while(node)
 		{
 			next = node->next;
+			if(node->valptr)
+				free(node->valptr);
 			free(node->key);
 			free(node);
 			node = next;
@@ -65,7 +67,13 @@ static void htab_rehash(tvm_htab_t *orig, unsigned int size)
 		while(node)
 		{
 			next = node->next;
-			htab_add(new, node->key, node->value);
+			if (node->valptr)
+			{
+				htab_add_str(new, node->key, node->valptr, strlen(node->valptr) + 1);
+				free(node->valptr);
+			}
+			else
+				htab_add(new, node->key, node->value);
 			free(node->key);
 			free(node);
 			node = next;
@@ -113,7 +121,26 @@ int htab_add(tvm_htab_t *htab, const char *k, int v)
 	if((float)++htab->num_nodes / htab->size > HTAB_LOAD_FACTOR)
 		htab_rehash(htab, htab->num_nodes * 2);
 
-	return 0;
+	return hash;
+}
+
+int htab_add_str(tvm_htab_t *htab, const char *key, const void *valptr, int len)
+{
+	int hash = htab_add(htab, key, 0);
+	int found = 0;
+	tvm_htab_node_t *node = htab->nodes[hash];
+
+	while (node && !found)
+	{
+		if (!strcmp(node->key, key))
+			found = 1;
+		else
+			node = node->next;
+	}
+
+	node->valptr = calloc(len, sizeof(char));
+	memcpy(node->valptr, valptr, len);
+	return hash;
 }
 
 int htab_find(tvm_htab_t *htab, const char *key)
@@ -131,3 +158,17 @@ int htab_find(tvm_htab_t *htab, const char *key)
 	return -1;
 }
 
+char *htab_find_str(tvm_htab_t *htab, const char *key)
+{
+	int hash = htab_hash(key, htab->size);
+	tvm_htab_node_t *node = htab->nodes[hash];
+
+	while(node)
+	{
+		if(!strcmp(node->key, key))
+			return node->valptr;
+		node = node->next;
+	}
+
+	return NULL;
+}

libtvm/tvm_lexer.c

@@ -26,7 +26,7 @@ void lexer_destroy(tvm_lexer_t *lexer)
 	free(lexer);
 }
 
-void lex(tvm_lexer_t *lexer, char *source)
+void lex(tvm_lexer_t *lexer, char *source, tvm_htab_t *defines)
 {
 	int i, j;
 	char *pToken, *pLine = strtok(source, "\n");
@@ -60,12 +60,16 @@ void lex(tvm_lexer_t *lexer, char *source)
 
 		for(j = 0; (pToken && j < MAX_TOKENS); j++)
 		{
-			lexer->tokens[i][j] = (char *)calloc(1, (strlen(pToken) + 1));
-			strcpy(lexer->tokens[i][j], pToken);
+			char *token = htab_find_str(defines, pToken);
+			token = token ? token : pToken;
+
+			lexer->tokens[i][j] = (char *)calloc(1, (strlen(token) + 1));
+			strcpy(lexer->tokens[i][j], token);
 
 			pToken = strtok(NULL, " \t,");
 		}
 	}
 
 	lexer->tokens[i] = NULL;
+	htab_destroy(defines);
 }

libtvm/tvm_preprocessor.c

@@ -3,7 +3,7 @@
 
 #include <string.h>
 
-int tvm_preprocess(char *src, int *src_len)
+int tvm_preprocess(char *src, int *src_len, tvm_htab_t *defines)
 {
 	char* pp_directive_delimiter = NULL;
 	if((pp_directive_delimiter = strstr(src, "%include")))
@@ -22,7 +22,7 @@ int tvm_preprocess(char *src, int *src_len)
 		if(!pFile)
 		{
 			printf("Unable to open file \"%s\"\n", filename);
-			return 0;
+			return -1;
 		}
 
 		free(temp_str);
@@ -48,6 +48,63 @@ int tvm_preprocess(char *src, int *src_len)
 		*src_len = strlen(src);
 		return 1;
 	}
+	else if((pp_directive_delimiter = strstr(src, "%define ")))
+	{
+		char *begin = pp_directive_delimiter;
+		char *end = strchr(begin, '\n');
+
+		if(!end) return 0;
+
+		int offset = strlen("%define ");
+
+		if(begin + offset >= end)
+		{
+			printf("Define missing arguments.\n");
+			return -1;
+		}
+
+		int length = (end - (begin + offset));
+		char tempstr[length + 1];
+		memset(tempstr, 0, length + 1);
+		memcpy(tempstr, begin + offset, length);
+
+		char *keystr = tempstr;
+		char *valstr = strchr(tempstr, ' ');
+
+		/* If there is a value, seperate the key and value
+		   with a null character. */
+		if(valstr)
+		{
+			*valstr = 0;
+			valstr += 1;
+		}
+
+		if(!keystr || !valstr)
+		{
+			printf("Define missing arguments.\n");
+			return -1;
+		}
+
+		if(htab_find(defines, keystr) < 0)
+			htab_add_str(defines, keystr, valstr, strlen(valstr) + 1);
+		else
+		{
+			printf("Multiple definitions for %s.\n", keystr);
+			return -1;
+		}
+
+		/* Remove the define line so it is not processed again. */
+		size_t new_length = *src_len - (end - begin);
+		size_t first_block_len = begin - src;
+		size_t second_block_len = (src + *src_len) - end;
+
+		memmove(&src[first_block_len], end, second_block_len);
+
+		src = realloc(src, sizeof(char) * new_length);
+		*src_len = new_length;
+
+		return 1;
+	}
 
 	return 0;
 

libtvm/tvm_program.c

@@ -8,6 +8,7 @@ tvm_program_t *program_create()
 {
 	tvm_program_t *p = (tvm_program_t *)calloc(1, sizeof(tvm_program_t));
 	p->label_htab = htab_create();
+	p->defines = htab_create();
 
 	return p;
 }
@@ -54,10 +55,15 @@ pi_interpret:
 	tvm_fcopy(source, source_length, pFile);
 	fclose(pFile);
 
-	while(tvm_preprocess(source, &source_length));
+	int err = 0;
+	while((err = tvm_preprocess(source, &source_length, p->defines)) > 0);
+
+	/* The preprocessor encountered a problem. */
+	if (err < 0)
+		return 1;
 
 	tvm_lexer_t *lexer_ctx = lexer_create();
-	lex(lexer_ctx, source);
+	lex(lexer_ctx, source, p->defines);
 	free(source);
 
 	if(parse_labels(p, (const char ***)lexer_ctx->tokens) != 0) return 1;

programs/tinyvm/preprocessor/define.vm

@@ -0,0 +1,20 @@
+%define ONE 1
+%define ZERO 0
+
+start:
+	mov eax, ONE
+	mov ebx, ZERO
+
+loop:	add eax, ebx
+	add ebx, eax
+
+	prn eax
+	prn ebx
+
+	cmp eax, ZERO
+	jl end
+
+	cmp ebx, ZERO
+	jg loop
+
+end: