Merge pull request #4572 from jepler/re-compilecode-overflow

extmod/re1.5: Check and report byte overflow errors in _compilecode.

extmod/re1.5/compilecode.c

@@ -8,9 +8,9 @@
     ((code ? memmove(code + at + num, code + at, pc - at) : 0), pc += num)
 #define REL(at, to) (to - at - 2)
 #define EMIT(at, byte) (code ? (code[at] = byte) : (at))
+#define EMIT_CHECKED(at, byte) (_emit_checked(at, code, byte, &err))
 #define PC (prog->bytelen)
 
-
 static char unescape(char c) {
     switch (c) {
         case 'a':
@@ -33,9 +33,17 @@ static char unescape(char c) {
 }
 
 
+static void _emit_checked(int at, char *code, int val, bool *err) {
+    *err |= val != (int8_t)val;
+    if (code) {
+        code[at] = val;
+    }
+}
+
 static const char *_compilecode(const char *re, ByteProg *prog, int sizecode)
 {
     char *code = sizecode ? NULL : prog->insts;
+    bool err = false;
     int start = PC;
     int term = PC;
     int alt_label = 0;
@@ -96,7 +104,7 @@ static const char *_compilecode(const char *re, ByteProg *prog, int sizecode)
                     EMIT(PC++, *re);
                 }
             }
-            EMIT(term + 1, cnt);
+            EMIT_CHECKED(term + 1, cnt);
             break;
         }
         case '(': {
@@ -107,7 +115,7 @@ static const char *_compilecode(const char *re, ByteProg *prog, int sizecode)
             if (capture) {
                 sub = ++prog->sub;
                 EMIT(PC++, Save);
-                EMIT(PC++, 2 * sub);
+                EMIT_CHECKED(PC++, 2 * sub);
                 prog->len++;
             } else {
                     re += 2;
@@ -118,7 +126,7 @@ static const char *_compilecode(const char *re, ByteProg *prog, int sizecode)
 
             if (capture) {
                 EMIT(PC++, Save);
-                EMIT(PC++, 2 * sub + 1);
+                EMIT_CHECKED(PC++, 2 * sub + 1);
                 prog->len++;
             }
 
@@ -133,7 +141,7 @@ static const char *_compilecode(const char *re, ByteProg *prog, int sizecode)
             } else {
                 EMIT(term, Split);
             }
-            EMIT(term + 1, REL(term, PC));
+            EMIT_CHECKED(term + 1, REL(term, PC));
             prog->len++;
             term = PC;
             break;
@@ -141,7 +149,7 @@ static const char *_compilecode(const char *re, ByteProg *prog, int sizecode)
             if (PC == term) return NULL; // nothing to repeat
             INSERT_CODE(term, 2, PC);
             EMIT(PC, Jmp);
-            EMIT(PC + 1, REL(PC, term));
+            EMIT_CHECKED(PC + 1, REL(PC, term));
             PC += 2;
             if (re[1] == '?') {
                 EMIT(term, RSplit);
@@ -149,7 +157,7 @@ static const char *_compilecode(const char *re, ByteProg *prog, int sizecode)
             } else {
                 EMIT(term, Split);
             }
-            EMIT(term + 1, REL(term, PC));
+            EMIT_CHECKED(term + 1, REL(term, PC));
             prog->len += 2;
             term = PC;
             break;
@@ -161,20 +169,20 @@ static const char *_compilecode(const char *re, ByteProg *prog, int sizecode)
             } else {
                 EMIT(PC, RSplit);
             }
-            EMIT(PC + 1, REL(PC, term));
+            EMIT_CHECKED(PC + 1, REL(PC, term));
             PC += 2;
             prog->len++;
             term = PC;
             break;
         case '|':
             if (alt_label) {
-                EMIT(alt_label, REL(alt_label, PC) + 1);
+                EMIT_CHECKED(alt_label, REL(alt_label, PC) + 1);
             }
             INSERT_CODE(start, 2, PC);
             EMIT(PC++, Jmp);
             alt_label = PC++;
             EMIT(start, Split);
-            EMIT(start + 1, REL(start, PC));
+            EMIT_CHECKED(start + 1, REL(start, PC));
             prog->len += 2;
             term = PC;
             break;
@@ -192,9 +200,9 @@ static const char *_compilecode(const char *re, ByteProg *prog, int sizecode)
     }
 
     if (alt_label) {
-        EMIT(alt_label, REL(alt_label, PC) + 1);
+        EMIT_CHECKED(alt_label, REL(alt_label, PC) + 1);
     }
-    return re;
+    return err ? NULL : re;
 }
 
 int re1_5_sizecode(const char *re)

tests/extmod/ure_limit.py

@@ -0,0 +1,34 @@
+# Test overflow in ure.compile output code.
+
+try:
+    import ure as re
+except ImportError:
+    print("SKIP")
+    raise SystemExit
+
+
+def test_re(r):
+    try:
+        re.compile(r)
+    except:
+        print("Error")
+
+
+# too many chars in []
+test_re("[" + "a" * 256 + "]")
+
+# too many groups
+test_re("(a)" * 256)
+
+# jump too big for ?
+test_re("(" + "a" * 62 + ")?")
+
+# jump too big for *
+test_re("(" + "a" * 60 + ".)*")
+test_re("(" + "a" * 60 + "..)*")
+
+# jump too big for +
+test_re("(" + "a" * 62 + ")+")
+
+# jump too big for |
+test_re("b" * 63 + "|a")

tests/extmod/ure_limit.py.exp

@@ -0,0 +1,7 @@
+Error
+Error
+Error
+Error
+Error
+Error
+Error