Modify pad function to handle invalid unicode string for padString (facebookincubator#10549)

Summary:
Pull Request resolved: facebookincubator#10549

Fuzzer caught this bug:

https://www.internalfb.com/chronos/job_instance/silver/3377708311833077/simple-logs

Reviewed By: kevinwilfong

Differential Revision: D60181544

fbshipit-source-id: 2b8857db6d2217db10e772a8d5f7ae887450b1a3

Author: Daniel Hunte

velox/functions/lib/string/StringImpl.h

@@ -593,6 +593,8 @@ FOLLY_ALWAYS_INLINE void pad(
       "pad size must be in the range [0..{})",
       padMaxSize);
   VELOX_USER_CHECK(padString.size() > 0, "padString must not be empty");
+  int64_t padStringCharLength = length<isAscii>(padString);
+  VELOX_USER_CHECK(padStringCharLength > 0, "padString must be a valid string");
 
   int64_t stringCharLength = length<isAscii>(string);
   // If string has at most size characters, truncate it if necessary
@@ -608,7 +610,6 @@ FOLLY_ALWAYS_INLINE void pad(
     return;
   }
 
-  int64_t padStringCharLength = length<isAscii>(padString);
   // How many characters do we need to add to string.
   int64_t fullPaddingCharLength = size - stringCharLength;
   // How many full copies of padString need to be added.

velox/functions/lib/string/tests/StringImplTest.cpp

@@ -652,10 +652,18 @@ TEST_F(StringImplTest, pad) {
                              const std::string& padString) {
     core::StringWriter output;
 
-    EXPECT_THROW(
-        (facebook::velox::functions::stringImpl::pad<true, true>(
-            output, StringView(string), size, StringView(padString))),
-        VeloxUserError);
+    bool padStringIsAscii = isAscii(padString.c_str(), padString.size());
+    if (padStringIsAscii) {
+      EXPECT_THROW(
+          (facebook::velox::functions::stringImpl::pad<true, true>(
+              output, StringView(string), size, StringView(padString))),
+          VeloxUserError);
+    } else {
+      EXPECT_THROW(
+          (facebook::velox::functions::stringImpl::pad<true, false>(
+              output, StringView(string), size, StringView(padString))),
+          VeloxUserError);
+    }
   };
 
   // ASCII string with various values for size and padString
@@ -712,6 +720,9 @@ TEST_F(StringImplTest, pad) {
   runTest(
       "abcd\xff \xff ef", 11, "0", "0abcd\xff \xff ef", "abcd\xff \xff ef0");
   runTest("abcd\xff ef", 6, "0", "abcd\xff ", "abcd\xff ");
+  // Testcase for when padString is a sequence of unicode continuation bytes
+  // for which effective length is 0.
+  runTestUserError(/*string=*/"\u4FE1", /*size=*/6, /*padString=*/"\xBF\xBF");
 }
 
 // Make sure that utf8proc_codepoint returns invalid codepoint (-1) for