quote: move nasm_unquote_cstr() to quote.c; disallow control characters

Move nasm_unquote_cstr() to quote.c so we can use it in other places.
Provide nasm_unquote_pp() as a minimal wrapper for the preprocessor
itself.

Disallow non-ASCII characters except tab in the output.

Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>

3 files changed, 52 insertions, 18 deletions

diff --git a/asm/preproc.c b/asm/preproc.c
index 8e1e6369..11d521fb 100644
--- a/asm/preproc.c
+++ b/asm/preproc.c
@@ -479,20 +479,11 @@ static Token *delete_Token(Token * t);
 #define tok_isnt_(x,v)  ((x) && ((x)->type!=TOK_OTHER || strcmp((x)->text,(v))))
 
 /*
- * nasm_unquote with error if the string contains NUL characters.
- * If the string contains NUL characters, issue an error and return
- * the C len, i.e. truncate at the NUL.
+ * Wrapper for nasm_unquote_cstr()
  */
-static size_t nasm_unquote_cstr(char *qstr, enum preproc_token directive)
+static void nasm_unquote_pp(char *qstr, enum preproc_token directive)
 {
-    size_t len = nasm_unquote(qstr, NULL);
-    size_t clen = strlen(qstr);
-
-    if (len != clen)
-        nasm_error(ERR_NONFATAL, "NUL character in `%s' directive",
-              pp_directives[directive]);
-
-    return clen;
+    nasm_unquote_cstr(qstr, pp_directives[directive]);
 }
 
 /*
@@ -1819,7 +1810,7 @@ static bool if_condition(Token * tline, enum preproc_token ct)
             if (tline->type == TOK_PREPROC_ID)
                 p += 2;         /* Skip leading %! */
             if (*p == '\'' || *p == '\"' || *p == '`')
-                nasm_unquote_cstr(p, ct);
+                nasm_unquote_pp(p, ct);
             if (getenv(p))
                 j = true;
             tline = tline->next;
@@ -2600,7 +2591,7 @@ static int do_directive(Token *tline, char **output)
                   "trailing garbage after `%%depend' ignored");
         p = t->text;
         if (t->type != TOK_INTERNAL_STRING)
-            nasm_unquote_cstr(p, i);
+            nasm_unquote_pp(p, i);
         nasm_add_string_to_strlist(dephead, p);
         free_tlist(origline);
         return DIRECTIVE_FOUND;
@@ -2620,7 +2611,7 @@ static int do_directive(Token *tline, char **output)
                   "trailing garbage after `%%include' ignored");
         p = t->text;
         if (t->type != TOK_INTERNAL_STRING)
-            nasm_unquote_cstr(p, i);
+            nasm_unquote_pp(p, i);
         inc = nasm_malloc(sizeof(Include));
         inc->next = istk;
         inc->conds = NULL;
@@ -2662,7 +2653,7 @@ static int do_directive(Token *tline, char **output)
             nasm_error(ERR_WARNING|ERR_PASS1,
                   "trailing garbage after `%%use' ignored");
         if (tline->type == TOK_STRING)
-            nasm_unquote_cstr(tline->text, i);
+            nasm_unquote_pp(tline->text, i);
         use_pkg = nasm_stdmac_find_package(tline->text);
         if (!use_pkg)
             nasm_error(ERR_NONFATAL, "unknown `%%use' package: %s", tline->text);
@@ -3321,7 +3312,7 @@ issue_error:
          * are stored with the token stream reversed, so we have to
          * reverse the output of tokenize().
          */
-        nasm_unquote_cstr(t->text, i);
+        nasm_unquote_pp(t->text, i);
         macro_start = reverse_tokens(tokenize(t->text));
 
         /*
diff --git a/asm/quote.c b/asm/quote.c
index 75a93726..8e64dc4a 100644
--- a/asm/quote.c
+++ b/asm/quote.c
@@ -41,6 +41,7 @@
 
 #include "nasmlib.h"
 #include "quote.h"
+#include "error.h"
 
 char *nasm_quote(const char *str, size_t len)
 {
@@ -255,7 +256,6 @@ size_t nasm_unquote(char *str, char **ep)
 	    p++;
 	    *q++ = c;
 	}
-	*q = '\0';
 	break;
 
     case '`':
@@ -417,6 +417,8 @@ size_t nasm_unquote(char *str, char **ep)
 
     if (ep)
 	*ep = p;
+
+    *q = '\0';
     return q-str;
 }
 
@@ -477,3 +479,37 @@ char *nasm_skip_string(char *str)
 	return str;		/* Not a string... */
     }
 }
+
+/*
+ * nasm_unquote with error if the string is malformed or contains
+ * control characters.
+ *
+ * Returns a pointer to the first input character after the string;
+ * qstr will be a valid C string on return.
+ */
+char *nasm_unquote_cstr(char *qstr, const char *where)
+{
+    char *ep;
+    size_t len = nasm_unquote(qstr, &ep);
+    size_t clen;
+
+    if (!*ep)
+        nasm_error(ERR_NONFATAL, "invalid quoted string");
+    else
+        ep++;                   /* Skip closing quote */
+
+    for (clen = 0; qstr[clen] >= ' ' || qstr[clen] == '\t'; clen++)
+        ;
+
+    qstr[clen] = '\0';          /* Truncate string if necessary */
+
+    if (len != clen) {
+        if (where) {
+            nasm_error(ERR_NONFATAL, "control character in `%s' directive", where);
+        } else {
+            nasm_error(ERR_NONFATAL, "invalid control character in string");
+        }
+    }
+
+    return ep;
+}
diff --git a/asm/quote.h b/asm/quote.h
index 2d8ce87b..cfa7cb99 100644
--- a/asm/quote.h
+++ b/asm/quote.h
@@ -36,9 +36,16 @@
 
 #include "compiler.h"
 
+/* Is it the leading character in a quoted string? */
+static inline bool isquote(char c)
+{
+    return c == '\'' || c == '\"' || c == '`';
+}
+
 char *nasm_quote(const char *str, size_t len);
 size_t nasm_unquote(char *str, char **endptr);
 char *nasm_skip_string(char *str);
+char *nasm_unquote_cstr(char *str, const char *where);
 
 #endif /* NASM_QUOTE_H */