Avoid calling std::memcmp with nullptr

[vitalybuka]

Summary

UBSAN complains that this is undefined behavior.

We can assume that empty substring (N==1) always satisfy conditions. So
std::memcmp will be called only only for N > 1 and Str.size() > 0.

[zturner]

include/llvm/ADT/StringSwitch.h

76 ↗

(On Diff #77903)

I don't think this second assert is valid. Consider this code:

char dir[MAX_PATH];
sprintf(dir, "%s/%s", parent, child);
auto X = StringSwitch<int>(MyPath).Case(dir, 1);

There is no guarantee what is the value of S[N-1]

[ruiu]

Do you mean that if it is called with "", such as StartsWith("", foo), S becomes a nullptr?

[zturner]

In D26646#595108, @ruiu wrote:

Do you mean that if it is called with "", such as StartsWith("", foo), S becomes a nullptr?

Were you asking me? My comment is just saying that string literal is not the only way to invoke this overload. You can invoke it with a char array too, and strlen(char_array) is not guaranteed to be the same as sizeof(char_array) - 1.

[ruiu]

No, I was asking Vitaly. The commit message says that "Avoid calling std::memcmp with nullptr", but I don't understand how can a nullptr be passed to memcmp.

[vitalybuka]

If Str is empty then Str.data() will be nullptr

[vitalybuka]

Removed assert(S[N-1] == '\0')

llvm/trunk/include/llvm/ADT/StringSwitch.h

public:
   }
 
   ~StringSwitch() = default;
 
   // Case-sensitive case matchers
   template<unsigned N>
   LLVM_ATTRIBUTE_ALWAYS_INLINE
   StringSwitch& Case(const char (&S)[N], const T& Value) {
+    assert(N);
     if (!Result && N-1 == Str.size() &&
-        (std::memcmp(S, Str.data(), N-1) == 0)) {
+        (N == 1 || std::memcmp(S, Str.data(), N-1) == 0)) {
       Result = &Value;
     }
     return *this;
   }
 
   template<unsigned N>
   LLVM_ATTRIBUTE_ALWAYS_INLINE
   StringSwitch& EndsWith(const char (&S)[N], const T &Value) {
+    assert(N);
     if (!Result && Str.size() >= N-1 &&
-        std::memcmp(S, Str.data() + Str.size() + 1 - N, N-1) == 0) {
+        (N == 1 || std::memcmp(S, Str.data() + Str.size() + 1 - N, N-1) == 0)) {
       Result = &Value;
     }
     return *this;
   }
 
   template<unsigned N>
   LLVM_ATTRIBUTE_ALWAYS_INLINE
   StringSwitch& StartsWith(const char (&S)[N], const T &Value) {
+    assert(N);
     if (!Result && Str.size() >= N-1 &&
-        std::memcmp(S, Str.data(), N-1) == 0) {
+        (N == 1 || std::memcmp(S, Str.data(), N-1) == 0)) {
       Result = &Value;
     }
     return *this;
   }
 
   template<unsigned N0, unsigned N1>
   LLVM_ATTRIBUTE_ALWAYS_INLINE
   StringSwitch &Cases(const char (&S0)[N0], const char (&S1)[N1],