shithub: scc

Download patch

ref: 3674aab1851afa91ffb0715055c0cf1d0ccc0058
parent: 7681aff4af1f56f6a0ed6a9451cc632f14bffbdd
author: Roberto E. Vargas Caballero <[email protected]>
date: Fri Aug 8 09:47:43 EDT 2014 

Rewrite parser of cc2 to handle lines

This new parser read full lines and don't store state in the own
parser. Instead of that it read line by line and check when a line
is correct using a variable that indicate if we are in the body of
a function


--- a/cc2/cc2.h
+++ b/cc2/cc2.h
@@ -33,6 +33,8 @@
 	ETYPERR,       /* incorrect type in expression */
 	EEXPBAL,       /* expression not balanced */
 	ESYNTAX,       /* syntax error */
+	ELNLINE,       /* line too long */
+	EFERROR,       /* error reading from file:%s*/
 	ENUMERR
 };
 
@@ -40,6 +42,7 @@
 #define REGISTER 'R'
 #define STATIC 'S'
 #define CONST '#'
+#define LABEL 'L'
 
 extern void error(unsigned nerror, ...);
-extern void genaddable(Node *np);
+extern void genaddable(Node *list[]);
--- a/cc2/cgen.c
+++ b/cc2/cgen.c
@@ -12,8 +12,8 @@
  *     STATIC => 12
  *     CONST => 20
  */
-void
-genaddable(Node *np)
+static void
+xaddable(Node *np)
 {
 	if (!np)
 		return;
@@ -34,4 +34,13 @@
 		np->addable = 20;
 		break;
 	}
+}
+
+void
+genaddable(Node *list[])
+{
+	Node *np;
+
+	while ((np = *list++) != NULL)
+		xaddable(np);
 }
--- a/cc2/parser.c
+++ b/cc2/parser.c
@@ -1,7 +1,9 @@
 
-#include <ctype.h>
+#include <errno.h>
 #include <stdint.h>
 #include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
 
 #include <cc.h>
 #include <sizes.h>
@@ -15,8 +17,9 @@
 #define NR_NODEPOOL 128
 #define NR_EXPRESSIONS 64
 
+char funbody;
 static Node *stack[NR_STACKSIZ], **stackp = stack;
-static Node *listexp[NR_EXPRESSIONS], **listp = &listexp[0];
+static Node *listexp[NR_EXPRESSIONS], **listp = listexp;
 static Node nodepool[NR_NODEPOOL], *newp = nodepool;
 
 static Symbol *localtbl;
@@ -23,12 +26,11 @@
 static Symbol *globaltbl;
 
 static Symbol *
-local(void)
+local(char *num)
 {
-	unsigned i;
+	unsigned i = atoi(num+1);
 	static unsigned nr;
 
-	scanf("%u", &i);
 	if (i >= NR_INT_IDENT)
 		error(EINTNUM);
 	if (i >= nr)
@@ -37,12 +39,11 @@
 }
 
 static Symbol *
-global(void)
+global(char *num)
 {
-	unsigned i;
+	unsigned i = atoi(num+1);
 	static unsigned nr;
 
-	scanf("%u", &i);
 	if (i >= NR_EXT_IDENT)
 		error(EEXTNUM);
 	if (i >= nr)
@@ -83,13 +84,11 @@
 }
 
 static char
-gettype(void)
+gettype(char *type)
 {
-	char t;
-
-	switch (t = getchar()) {
+	switch (type[0]) {
 	case L_INT16: case L_INT8:
-		return t;
+		return type[0];
 	default:
 		error(ETYPERR);
 	}
@@ -96,26 +95,27 @@
 }
 
 static void
-variable(char op)
+variable(char *token)
 {
 	Symbol *sym;
+	char op;
 	Node *np = newnode();
 
-	switch (op) {
+	switch (token[0]) {
 	case 'A':
-		sym = local();
+		sym = local(token);
 		op = AUTO;
 		break;
 	case 'R':
-		sym = local();
+		sym = local(token);
 		op = REGISTER;
 		break;
 	case 'T':
-		sym = local();
+		sym = local(token);
 		op = STATIC;
 		break;
 	case 'G':
-		sym = global();
+		sym = global(token);
 		op = STATIC;
 		break;
 	}
@@ -128,7 +128,7 @@
 }
 
 static void
-immediate(char op)
+immediate(char *token)
 {
 	Node *np = newnode();
 
@@ -135,30 +135,31 @@
 	np->op = CONST;
 	/* TODO: deal with constant non integer */
 	np->type = L_INT;
-	scanf("%d", &np->u.imm);
+	np->u.imm = atoi(token+1);
 	np->left = np->right = NULL;
 	push(np);
 }
 
 static void
-operator(char op)
+operator(char *token)
 {
 	Node *np = newnode();
 
 	np->left = pop();
 	np->right = pop();
-	np->type = gettype();
-	np->op = op;
+	np->type = gettype(token+1);
+	np->op = token[0];
 	push(np);
 }
 
 static void
-label(char op)
+label(char *token)
 {
 	Node *np = newnode();
 
 	np->left = np->right = NULL;
-	np->op = 'L';
+	np->op = LABEL;
+	np->u.sym = local(token);
 	push(np);
 }
 
@@ -171,7 +172,7 @@
 	return np;
 }
 
-static void (*optbl[])(char) = {
+static void (*optbl[])(char *) = {
 	['+'] = operator,
 	['-'] = operator,
 	['*'] = operator,
@@ -185,107 +186,118 @@
 };
 
 static void
-expression(void)
+expression(char *token)
 {
-	int c;
-	void (*fun)(char);
+	void (*fun)(char *);
 
 	do {
-		if (!isprint(c = getchar()))
+		if ((fun = optbl[token[0]]) == NULL)
 			error(ESYNTAX);
-		if ((fun = optbl[c]) == NULL)
-			error(ESYNTAX);
-		(*fun)(c);
-	} while ((c = getchar()) == '\t');
+		(*fun)(token);
+	} while ((token = strtok(NULL, "\t")) != NULL);
 
-	if (c != '\n')
-		error(ESYNTAX);
 	link(getroot());
 }
 
 static void
-declaration(char sclass, char islocal)
+declaration(char *token)
 {
-	Symbol *sym = (islocal) ? local() : global();
+	char class = token[0];
+	Symbol *sym;
 
-	getchar(); /* skip tab */
-	sym->u.v.storage = sclass;
-	sym->u.v.type = gettype();
-	if (getchar() != '\n')
-		error(ESYNTAX);
+	sym = (class == 'G') ? global(token) : local(token);
+
+	sym->u.v.storage = class;
+	sym->u.v.type = gettype(strtok(NULL, "\t"));
 }
 
 static void
-chop(void)
+deflabel(char *token)
 {
-	int c;
+	Symbol *sym;
 
-	while ((c = getchar()) != EOF && c != '\n')
-		/* nothing */;
+	sym = local(token);
+	sym->u.l.addr = listp - listexp;
 }
 
 static void
-deflabel(void)
+function(char *token)
 {
-	Symbol *sym = local();
-
-	sym->u.l.addr = listp - listexp;
-	chop();
+	funbody = 1;
+	listp = listexp;
+	newp = nodepool;
 }
 
 static void
-function(void)
+endfunction(char *token)
 {
+	funbody = 0;
+	listp = NULL;
+	genaddable(listexp);
+}
+
+void
+parse(void)
+{
+	void (*fun)(char *tok);
+	uint8_t len;
 	int c;
-	char name[IDENTSIZ + 1];
+	char line[200];
 
-	scanf("%" STR(IDENTSIZ) "s", name);
-	chop();
 	for (;;) {
 		switch (c = getchar()) {
-		case '\t':
-			expression();
-			break;
 		case 'L':
-			deflabel();
+			if (!funbody)
+				goto syntax_error;
+			fun = deflabel;
 			break;
+		case '\t':
+			if (!funbody)
+				goto syntax_error;
+			fun = expression;
+			break;
 		case 'S':
 			/* struct */
 			break;
 		case 'T': case 'A': case 'R':
-			declaration(c, 1);
+			if (!funbody)
+				goto syntax_error;
+			fun = declaration;
 			break;
-		case '}':
-			chop();
-			return;
-		default:
-			error(ESYNTAX);
-			break;
-		}
-	}
-}
-
-void
-parse(void)
-{
-	int c;
-	void genaddr(Node *np);
-
-	for (;;) {
-		switch (c = getchar()) {
 		case 'G':
-			declaration(c, 0);
+			if (funbody)
+				goto syntax_error;
+			fun = declaration;
 			break;
 		case 'X':
-			function();
-			genaddable(listexp[0]);
+			if (funbody)
+				goto syntax_error;
+			fun = function;
 			break;
-		case EOF:
-			return;
+		case '}':
+			fun = endfunction;
 			break;
+		case EOF:
+			goto found_eof;
 		default:
-			error(ESYNTAX);
+			goto syntax_error;
 		}
+
+		ungetc(c, stdin);
+		if (!fgets(line, sizeof(line), stdin))
+			break;
+		len = strlen(line);
+		if (line[len-1] != '\n')
+			error(ELNLINE);
+		line[len-1] = '\0';
+		(*fun)(strtok(line, "\t"));
 	}
-}
 
+found_eof:
+	if (ferror(stdin))
+		error(EFERROR, strerror(errno));
+	return;
+
+syntax_error:
+	error(ESYNTAX);
+}