tempcommitting WARN_UNKNOWN_CONTROL_SEQUENCE and handling of backslashes in strings

diff --git a/lexer.c b/lexer.c
index 3f1aa76c0f5c312273a1d17a8c33fe06996e8627..bfa7443f3eb09fa13e1b104e33f67562a84f7b88 100644 (file)
--- a/lexer.c
+++ b/lexer.c
@@ -24,6 +24,25 @@ void lexerror(lex_file *lex, const char *fmt, ...)
        printf("\n");
 }
 
+void lexwarn(lex_file *lex, int warn, const char *fmt, ...)
+{
+       va_list ap;
+
+       if (!OPTS_WARN(warn))
+           return;
+
+       if (lex)
+               printf("warning %s:%lu: ", lex->name, (unsigned long)lex->sline);
+       else
+               printf("warning: ");
+
+       va_start(ap, fmt);
+       vprintf(fmt, ap);
+       va_end(ap);
+
+       printf("\n");
+}
+
 token* token_new()
 {
        token *tok = (token*)mem_a(sizeof(token));
@@ -333,10 +352,6 @@ static int GMQCC_WARN lex_finish_string(lex_file *lex, int quote)
                if (ch == quote)
                        return TOKEN_STRINGCONST;
 
-               if (!lex_tokench(lex, ch))
-                       return (lex->tok->ttype = TOKEN_FATAL);
-
-               /* as lexer we only care about \" to not terminate the string prematurely */
                if (ch == '\\') {
                        ch = lex_getch(lex);
                        if (ch == EOF) {
@@ -344,10 +359,28 @@ static int GMQCC_WARN lex_finish_string(lex_file *lex, int quote)
                                lex_ungetch(lex, EOF); /* next token to be TOKEN_EOF */
                                return (lex->tok->ttype = TOKEN_ERROR);
                        }
-                       /* so we just add the next character no matter what it actually is */
+
+            switch (ch) {
+            case '\\': break;
+            case 'a':  ch = '\a'; break;
+            case 'b':  ch = '\b'; break;
+            case 'r':  ch = '\r'; break;
+            case 'n':  ch = '\n'; break;
+            case 't':  ch = '\t'; break;
+            case 'f':  ch = '\f'; break;
+            case 'v':  ch = '\v'; break;
+            default:
+                lexwarn(lex, WARN_UNKNOWN_CONTROL_SEQUENCE, "unrecognized control sequence: \\%c", ch);
+                           /* so we just add the character plus backslash no matter what it actually is */
+                           if (!lex_tokench(lex, '\\'))
+                                   return (lex->tok->ttype = TOKEN_FATAL);
+            }
+            /* add the character finally */
                        if (!lex_tokench(lex, ch))
                                return (lex->tok->ttype = TOKEN_FATAL);
                }
+               else if (!lex_tokench(lex, ch))
+                       return (lex->tok->ttype = TOKEN_FATAL);
        }
        lexerror(lex, "unexpected end of file within string constant");
        lex_ungetch(lex, EOF); /* next token to be TOKEN_EOF */

diff --git a/main.c b/main.c
index 9cbf026dd75fa5c1a697fa5579dc1e82b1f699b0..7edb2b5d641ef5152352bb581995e09ab3bc4c42 100644 (file)
--- a/main.c
+++ b/main.c
@@ -308,6 +308,22 @@ static bool options_parse(int argc, char **argv) {
     return true;
 }
 
+static void options_setflag(uint32_t *flags, size_t idx, bool on)
+{
+    longbit lb = LONGBIT(idx);
+#if 0
+    if (on)
+        flags[lb.idx] |= (1<<(lb.bit));
+    else
+        flags[lb.idx] &= ~(1<<(lb.bit));
+#else
+    if (on)
+        flags[0] |= (1<<(lb));
+    else
+        flags[0] &= ~(1<<(lb));
+#endif
+}
+
 bool parser_init();
 bool parser_compile(const char *filename);
 bool parser_finish(const char *output);
@@ -317,6 +333,9 @@ int main(int argc, char **argv) {
     size_t itr;
     app_name = argv[0];
 
+    /* default options / warn flags */
+    options_setflag(opts_warn, WARN_UNKNOWN_CONTROL_SEQUENCE, true);
+
     if (!options_parse(argc, argv)) {
         return usage();
     }

diff --git a/warns.def b/warns.def
index 6238e504bbc60d703f4e7b5791bcbe04d59037fc..e7ebda70aa6c2d004fada92c806c20a8599f3189 100644 (file)
--- a/warns.def
+++ b/warns.def
@@ -3,3 +3,4 @@
 #endif
 
 GMQCC_DEFINE_FLAG(UNUSED_VARIABLE)
+GMQCC_DEFINE_FLAG(UNKNOWN_CONTROL_SEQUENCE)