patch up unifdef to recognize and remove "#if defined _LIBC && something"

blocks too (and similar)

2 files changed, 160 insertions, 35 deletions

diff --git a/extra/scripts/unifdef.c b/extra/scripts/unifdef.c
index 552025e72..87402c0ce 100644
--- a/extra/scripts/unifdef.c
+++ b/extra/scripts/unifdef.c
@@ -296,6 +296,7 @@ main(int argc, char *argv[])
 		input = stdin;
 	}
 	process();
+	debug("bug at line %d", __LINE__);
 	abort(); /* bug */
 }
 
@@ -438,8 +439,10 @@ done(void)
 static void
 ignoreoff(void)
 {
-	if (depth == 0)
+	if (depth == 0) {
+		debug("bug at line %d", __LINE__);
 		abort(); /* bug */
+	}
 	ignoring[depth] = ignoring[depth-1];
 }
 static void
@@ -471,8 +474,10 @@ nest(void)
 static void
 unnest(void)
 {
-	if (depth == 0)
+	if (depth == 0) {
+		debug("bug at line %d", __LINE__);
 		abort(); /* bug */
+	}
 	depth -= 1;
 }
 static void
@@ -596,8 +601,11 @@ getline(void)
 				linestate = LS_DIRTY;
 		}
 		/* skipcomment should have changed the state */
-		if (linestate == LS_HASH)
-			abort(); /* bug */
+// Hmm, happens sometimes on valid files
+//		if (linestate == LS_HASH) {
+//			debug("bug at line %d", __LINE__);
+//			abort(); /* bug */
+//		}
 	}
 	if (linestate == LS_DIRTY) {
 		while (*cp != '\0')
@@ -649,24 +657,31 @@ static const struct ops {
 	eval_fn *inner;
 	struct op {
 		const char *str;
+		int short_circuit_val;
 		int (*fn)(int, int);
 	} op[5];
 } eval_ops[] = {
-	{ eval_table, { { "||", op_or } } },
-	{ eval_table, { { "&&", op_and } } },
-	{ eval_table, { { "==", op_eq },
-			{ "!=", op_ne } } },
-	{ eval_unary, { { "<=", op_le },
-			{ ">=", op_ge },
-			{ "<", op_lt },
-			{ ">", op_gt } } }
+	{ eval_table, { { "||", 1, op_or } } },
+	{ eval_table, { { "&&", 0, op_and } } },
+	{ eval_table, { { "==", -1, op_eq },
+			{ "!=", -1, op_ne } } },
+	{ eval_unary, { { "<=", -1, op_le },
+			{ ">=", -1, op_ge },
+			{ "<", -1, op_lt },
+			{ ">", -1, op_gt } } }
 };
 
 /*
- * Function for evaluating the innermost parts of expressions,
- * viz. !expr (expr) defined(symbol) symbol number
+ * Function for evaluating the innermost parts of expressions, viz.
+ * "!expr", "(expr)", "defined(symbol)", "defined symbol", "symbol", "number".
  * We reset the keepthis flag when we find a non-constant subexpression.
  */
+// TODO: we use LT_IF both as "I don't know whether it's false or true"
+// (example: "#if defined FOO") and when we see syntax error
+// (example: "#if (1 || 2" - no closing paren!), but this is wrong.
+// Binary && and || need to distinguish these cases in order to handle this:
+// "#if defined KNOWN_UNDEFINED && FOO" - discard
+// "#if defined KNOWN_UNDEFINED && (syntax_error_here" - do not discard!
 static Linetype
 eval_unary(const struct ops *ops, int *valp, const char **cpp)
 {
@@ -678,39 +693,60 @@ eval_unary(const struct ops *ops, int *valp, const char **cpp)
 	if (*cp == '!') {
 		debug("eval%d !", ops - eval_ops);
 		cp++;
-		if (eval_unary(ops, valp, &cp) == LT_IF)
+		if (eval_unary(ops, valp, &cp) == LT_IF) {
+			*cpp = cp;
 			return (LT_IF);
+		}
 		*valp = !*valp;
+
 	} else if (*cp == '(') {
+		Linetype expr_res;
+
 		cp++;
-		debug("eval%d (", ops - eval_ops);
-		if (eval_table(eval_ops, valp, &cp) == LT_IF)
-			return (LT_IF);
+		debug("eval%d (%s", ops - eval_ops, cp);
+		expr_res = eval_table(eval_ops, valp, &cp);
 		cp = skipcomment(cp);
+		*cpp = cp;
 		if (*cp++ != ')')
 			return (LT_IF);
+		*cpp = cp;
+		if (expr_res == LT_IF)
+			return (LT_IF);
+
 	} else if (isdigit((unsigned char)*cp)) {
 		debug("eval%d number", ops - eval_ops);
 		*valp = strtol(cp, &ep, 0);
 		cp = skipsym(cp);
+
 	} else if (strncmp(cp, "defined", 7) == 0 && endsym(cp[7])) {
+		bool parens;
+
 		cp = skipcomment(cp+7);
-		debug("eval%d defined", ops - eval_ops);
-		if (*cp++ != '(')
-			return (LT_IF);
-		cp = skipcomment(cp);
+		debug("eval%d defined '%s'", ops - eval_ops, cp);
+		parens = (*cp == '(');
+		if (parens)
+			cp = skipcomment(cp+1);
 		sym = findsym(cp);
-		if (sym < 0)
-			return (LT_IF);
-		*valp = (value[sym] != NULL);
 		cp = skipsym(cp);
 		cp = skipcomment(cp);
-		if (*cp++ != ')')
+		if (parens) {
+			if (*cp != ')')
+				return (LT_IF);
+			cp = skipcomment(cp+1);
+		}
+		*cpp = cp;
+		if (sym < 0) {
+			debug("sym not found, returning LT_IF");
 			return (LT_IF);
+		}
+		*valp = (value[sym] != NULL);
 		keepthis = false;
+
 	} else if (!endsym(*cp)) {
 		debug("eval%d symbol", ops - eval_ops);
 		sym = findsym(cp);
+		cp = skipsym(cp);
+		*cpp = cp;
 		if (sym < 0)
 			return (LT_IF);
 		if (value[sym] == NULL)
@@ -720,8 +756,8 @@ eval_unary(const struct ops *ops, int *valp, const char **cpp)
 			if (*ep != '\0' || ep == value[sym])
 				return (LT_IF);
 		}
-		cp = skipsym(cp);
 		keepthis = false;
+
 	} else {
 		debug("eval%d bad expr", ops - eval_ops);
 		return (LT_IF);
@@ -738,15 +774,18 @@ eval_unary(const struct ops *ops, int *valp, const char **cpp)
 static Linetype
 eval_table(const struct ops *ops, int *valp, const char **cpp)
 {
+	Linetype left_side;
 	const struct op *op;
 	const char *cp;
 	int val;
 
-	debug("eval%d", ops - eval_ops);
+	debug("eval%d '%s'", ops - eval_ops, *cpp);
+	left_side = ops->inner(ops+1, valp, cpp);
 	cp = *cpp;
-	if (ops->inner(ops+1, valp, &cp) == LT_IF)
-		return (LT_IF);
+
 	for (;;) {
+		Linetype right_side;
+
 		cp = skipcomment(cp);
 		for (op = ops->op; op->str != NULL; op++)
 			if (strncmp(cp, op->str, strlen(op->str)) == 0)
@@ -754,14 +793,37 @@ eval_table(const struct ops *ops, int *valp, const char **cpp)
 		if (op->str == NULL)
 			break;
 		cp += strlen(op->str);
-		debug("eval%d %s", ops - eval_ops, op->str);
-		if (ops->inner(ops+1, &val, &cp) == LT_IF)
+		debug("eval%d '%s'", ops - eval_ops, op->str);
+		right_side = ops->inner(ops+1, &val, &cp);
+
+		/* If short_circuit_val is 0 or 1, we can ignore
+		 * right side if left size is known, and its value
+		 * (i.e., *valp) is 0 or !0, respectively */
+		if (left_side != LT_IF && op->short_circuit_val == !!*valp) {
+			debug("op->short_circuit_val:%d *valp:%d cp:'%s'",
+					op->short_circuit_val, *valp, cp);
+			*valp = !!*valp;
+			break;
+		}
+		/* Same for the right side */
+		if (right_side != LT_IF && op->short_circuit_val == !!val) {
+			debug("op->short_circuit_val:%d val:%d cp:'%s'",
+					op->short_circuit_val, val, cp);
+			left_side = right_side;
+			*valp = !!val;
+			break;
+		}
+
+		if (left_side == LT_IF || right_side == LT_IF)
 			return (LT_IF);
 		*valp = op->fn(*valp, val);
+		left_side = right_side;
 	}
 
 	*cpp = cp;
-	debug("eval%d = %d", ops - eval_ops, *valp);
+	debug("eval%d = %d LT_IF:%d", ops - eval_ops, *valp, (left_side == LT_IF));
+	if (left_side == LT_IF)
+		return (LT_IF);
 	return (*valp ? LT_TRUE : LT_FALSE);
 }
 
@@ -779,7 +841,7 @@ ifeval(const char **cpp)
 	debug("eval %s", *cpp);
 	keepthis = killconsts ? false : true;
 	ret = eval_table(eval_ops, &val, cpp);
-	debug("eval = %d", val);
+	debug("val:%d ret:%d keepthis:%d", val, ret, keepthis);
 	return (keepthis ? LT_IF : ret);
 }
 
@@ -827,8 +889,9 @@ skipcomment(const char *cp)
 				cp += 1;
 			} else if (strchr(" \t", *cp) != NULL) {
 				cp += 1;
-			} else
+			} else {
 				return (cp);
+			}
 			continue;
 		case CXX_COMMENT:
 			if (strncmp(cp, "\n", 1) == 0) {
@@ -886,6 +949,7 @@ skipcomment(const char *cp)
 				incomment = C_COMMENT;
 			continue;
 		default:
+			debug("bug at line %d", __LINE__);
 			abort(); /* bug */
 		}
 	return (cp);
diff --git a/extra/scripts/unifdef.test b/extra/scripts/unifdef.test
new file mode 100644
index 000000000..7cd4f148e
--- /dev/null
+++ b/extra/scripts/unifdef.test
@@ -0,0 +1,61 @@
+Run me through unifdef -UA
+*** Nothing should be visible here:
+#if defined A && defined B
+hello world
+#endif
+#if defined A && B
+hello world
+#endif
+#if defined A && 1
+hello world
+#endif
+#if defined A && (1 > 0)
+hello world
+#endif
+#if defined B && defined A
+hello world
+#endif
+#if B && defined A
+hello world
+#endif
+#if 1 && defined A
+hello world
+#endif
+#if (1 > 0) && defined A
+hello world
+#endif
+
+*** Everything should be visible here, but #if/#endif removed:
+#if defined B || !defined A
+hello world 1
+#endif
+#if !defined A || defined B
+hello world 2 (last)
+#endif
+
+*** This should be unchanged (#if/#endif not removed):
+#if defined A || defined B
+I am here 1
+#endif
+#if defined B || defined A
+I am here 2
+#endif
+I am here 3
+#if !defined FOO && !defined BAR \
+    && !defined BAZ
+# error "I am here 4"
+#endif
+I am here 5
+#if (!defined FOO \
+     && (defined BAR || defined BAZ \
+         || defined XYZ))
+I am here 6
+#endif
+I am here 7
+#if !defined FOO \
+    && defined BAR
+I am here 8
+#endif
+I am here 9 (last)
+
+*** End