[tf.data] Changed parsing logic for CsvDataset for better performance and correctness

PiperOrigin-RevId: 198812512

2 files changed, 28 insertions, 0 deletions

diff --git a/tensorflow/core/lib/strings/numbers.cc b/tensorflow/core/lib/strings/numbers.cc
index 987e4fe733..f18c6dc709 100644
--- a/tensorflow/core/lib/strings/numbers.cc
+++ b/tensorflow/core/lib/strings/numbers.cc
@@ -345,6 +345,19 @@ bool safe_strtof(const char* str, float* value) {
   return processed_characters_count > 0;
 }
 
+bool safe_strtof(StringPiece str, float* value) {
+  int processed_characters_count = -1;
+  auto len = str.size();
+
+  // If string length exceeds buffer size or int max, fail.
+  if (len >= kFastToBufferSize) return false;
+  if (len > std::numeric_limits<int>::max()) return false;
+
+  *value = StringToFloatConverter().StringToFloat(
+      str.data(), static_cast<int>(len), &processed_characters_count);
+  return processed_characters_count > 0;
+}
+
 bool safe_strtod(const char* str, double* value) {
   int processed_characters_count = -1;
   auto len = str_util::Strnlen(str, kFastToBufferSize);
@@ -359,6 +372,19 @@ bool safe_strtod(const char* str, double* value) {
   return processed_characters_count > 0;
 }
 
+bool safe_strtod(StringPiece str, double* value) {
+  int processed_characters_count = -1;
+  auto len = str.size();
+
+  // If string length exceeds buffer size or int max, fail.
+  if (len >= kFastToBufferSize) return false;
+  if (len > std::numeric_limits<int>::max()) return false;
+
+  *value = StringToFloatConverter().StringToDouble(
+      str.data(), static_cast<int>(len), &processed_characters_count);
+  return processed_characters_count > 0;
+}
+
 size_t FloatToBuffer(float value, char* buffer) {
   // FLT_DIG is 6 for IEEE-754 floats, which are used on almost all
   // platforms these days.  Just in case some system exists where FLT_DIG
diff --git a/tensorflow/core/lib/strings/numbers.h b/tensorflow/core/lib/strings/numbers.h
index 9cb56415cb..f62584dedb 100644
--- a/tensorflow/core/lib/strings/numbers.h
+++ b/tensorflow/core/lib/strings/numbers.h
@@ -116,12 +116,14 @@ bool safe_strtou64(StringPiece str, uint64* value);
 // Values may be rounded on over- and underflow.
 // Returns false on invalid input or if `strlen(value) >= kFastToBufferSize`.
 bool safe_strtof(const char* str, float* value);
+bool safe_strtof(StringPiece str, float* value);
 
 // Convert strings to double precision floating point values.
 // Leading and trailing spaces are allowed.
 // Values may be rounded on over- and underflow.
 // Returns false on invalid input or if `strlen(value) >= kFastToBufferSize`.
 bool safe_strtod(const char* str, double* value);
+bool safe_strtod(StringPiece str, double* value);
 
 inline bool ProtoParseNumeric(StringPiece s, int32* value) {
   return safe_strto32(s, value);