Fix hexadecimal unescaping & test case

Author: ChAoSUnItY

src/defs.h

@@ -10,6 +10,8 @@
 
 /* definitions */
 
+#define DEBUG_BUILD false
+
 /* Common macro functions */
 #define is_whitespace(c) (c == ' ' || c == '\t')
 #define is_newline(c) (c == '\r' || c == '\n')
@@ -22,7 +24,6 @@
 /* Limitations */
 #define MAX_TOKEN_LEN 256
 #define MAX_ID_LEN 64
-#define MAX_ESCAPED_CHAR_LEN 5
 #define MAX_LINE_LEN 256
 #define MAX_VAR_LEN 32
 #define MAX_TYPE_LEN 32

src/globals.c

@@ -703,7 +703,7 @@ int unescape_string(const char *input, char *output, int output_size)
                 i++;
                 break;
             case 'e':
-                output[j++] = 23;
+                output[j++] = 27;
                 i++;
                 break;
             case 'n':
@@ -746,10 +746,12 @@ int unescape_string(const char *input, char *output, int output_size)
                     return -1;
 
                 int value = 0;
+                int count = 0;
 
-                while (is_hex(input[i])) {
-                    value = value * 16 + hex_digit_value(input[i]);
+                while (is_hex(input[i]) && count < 2) {
+                    value = (value << 4) + hex_digit_value(input[i]);
                     i++;
+                    count++;
                 }
 
                 output[j++] = (char) value;
@@ -1363,6 +1365,7 @@ void global_release(void)
     hashmap_free(CONSTANTS_MAP);
 }
 
+#if DEBUG_BUILD
 void dbg_token(token_t *token)
 {
     char *name;
@@ -1637,6 +1640,7 @@ void dbg_token(token_t *token)
                token->location.column);
     }
 }
+#endif
 
 /* Reports an error without specifying a position */
 void fatal(char *msg)

src/parser.c

@@ -1124,26 +1124,24 @@ void read_parameter_list_decl(func_t *func, bool anon)
 
 void read_literal_param(block_t *parent, basic_block_t *bb)
 {
-    char literal[MAX_TOKEN_LEN];
-    char unescaped[MAX_TOKEN_LEN];
-    char combined[MAX_TOKEN_LEN];
+    char literal[MAX_TOKEN_LEN], unescaped[MAX_TOKEN_LEN],
+        combined[MAX_LINE_LEN];
     int combined_len = 0;
 
     /* Read first string literal */
     lex_ident(T_string, literal);
-    unescape_string(literal, unescaped, MAX_TOKEN_LEN);
-    strcpy(combined, unescaped);
-    combined_len = strlen(unescaped);
+    unescape_string(literal, combined, MAX_LINE_LEN);
+    combined_len = strlen(combined);
 
     /* Check for adjacent string literals and concatenate them */
     while (lex_peek(T_string, NULL)) {
         lex_ident(T_string, literal);
-        unescape_string(literal, unescaped, MAX_TOKEN_LEN);
+        unescape_string(literal, unescaped, MAX_LINE_LEN - combined_len);
         int unescaped_len = strlen(unescaped);
-        if (combined_len + unescaped_len >= MAX_TOKEN_LEN - 1)
+        if (combined_len + unescaped_len >= MAX_LINE_LEN - 1)
             error("Concatenated string literal too long");
 
-        strcpy(combined + combined_len, literal);
+        strcpy(combined + combined_len, unescaped);
         combined_len += unescaped_len;
     }
 

src/riscv-codegen.c

@@ -99,14 +99,15 @@ void update_elf_offset(ph2_ir_t *ph2_ir)
         else
             elf_offset += 4;
         return;
-    case OP_sign_ext:
+    case OP_sign_ext: {
         /* Decode source size from upper 16 bits */
         int source_size = (ph2_ir->src1 >> 16) & 0xFFFF;
         if (source_size == 2)
             elf_offset += 8; /* short extension: 2 instructions */
         else
             elf_offset += 12; /* byte extension: 3 instructions */
         return;
+    }
     case OP_cast:
         elf_offset += 4;
         return;
@@ -443,7 +444,7 @@ void emit_ph2_ir(ph2_ir_t *ph2_ir)
             fatal("Unsupported truncation operation with invalid target size");
         }
         return;
-    case OP_sign_ext:
+    case OP_sign_ext: {
         /* Decode size information:
          * Lower 16 bits: target size
          * Upper 16 bits: source size
@@ -468,6 +469,7 @@ void emit_ph2_ir(ph2_ir_t *ph2_ir)
             emit(__srai(rd, rd, shift_amount));
         }
         return;
+    }
     case OP_cast:
         /* Generic cast operation - for now, just move the value */
         emit(__addi(rd, rs1, 0));

tests/driver.sh

@@ -5342,14 +5342,14 @@ EOF
 # String literal and escape coverage (additional)
 try_output 0 "AZ" << 'EOF'
 int main() {
-    printf("%s", "\\x41Z"); /* hex escape then normal char */
+    printf("%s", "\x41Z"); /* hex escape then normal char */
     return 0;
 }
 EOF
 
 try_output 0 "AZ" << 'EOF'
 int main() {
-    printf("%s", "A\\132"); /* octal escape for 'Z' */
+    printf("%s", "A\132"); /* octal escape for 'Z' */
     return 0;
 }
 EOF