ARROW-3846: minor fixes

cpp/src/gandiva/date_utils.h

@@ -60,19 +60,31 @@ class GANDIVA_EXPORT DateUtils {
 namespace internal {
 
 /// \brief Returns seconds since the UNIX epoch
-static inline bool ParseTimestamp(const char* buf, const char* format, int64_t* out) {
+static inline bool ParseTimestamp(const char* buf, const char* format,
+                                  bool ignoreTimeInDay, int64_t* out) {
 #if defined(_MSC_VER)
   static std::locale lc_all(setlocale(LC_ALL, NULLPTR));
   std::istringstream stream(buf);
   stream.imbue(lc_all);
 
-  date::sys_seconds secs;
-  stream >> date::parse(format, secs);
-  if (stream.fail()) {
-    return false;
+  // TODO: date::parse fails parsing when the hour value is 0.
+  // eg.1886-12-01 00:00:00
+  date::sys_seconds seconds;
+  if (ignoreTimeInDay) {
+    date::sys_days days;
+    stream >> date::parse(format, days);
+    if (stream.fail()) {
+      return false;
+    }
+    seconds = days;
+  } else {
+    stream >> date::parse(format, seconds);
+    if (stream.fail()) {
+      return false;
+    }
   }
-
-  *out = secs.time_since_epoch().count();
+  auto seconds_in_epoch = seconds.time_since_epoch().count();
+  *out = seconds_in_epoch;
   return true;
 #else
   struct tm result;
@@ -81,9 +93,13 @@ static inline bool ParseTimestamp(const char* buf, const char* format, int64_t*
     return false;
   }
   // ignore the time part
-  auto days = date::sys_days(date::year(result.tm_year + 1900) / (result.tm_mon + 1) /
-                             result.tm_mday);
-  *out = days.time_since_epoch().count();
+  date::sys_seconds secs = date::sys_days(date::year(result.tm_year + 1900) /
+                                          (result.tm_mon + 1) / result.tm_mday);
+  if (!ignoreTimeInDay) {
+    secs += (std::chrono::hours(result.tm_hour) + std::chrono::minutes(result.tm_min) +
+             std::chrono::seconds(result.tm_sec));
+  }
+  *out = secs.time_since_epoch().count();
   return true;
 #endif
 }

cpp/src/gandiva/precompiled/testing.h

@@ -30,7 +30,7 @@ namespace gandiva {
 
 timestamp StringToTimestamp(const char* buf) {
   int64_t out = 0;
-  DCHECK(internal::ParseTimestamp(buf, "%Y-%m-%d %H:%M:%S", &out));
+  DCHECK(internal::ParseTimestamp(buf, "%Y-%m-%d %H:%M:%S", false, &out));
   return out * 1000;
 }
 

cpp/src/gandiva/tests/projector_test.cc

@@ -86,10 +86,14 @@ TEST_F(TestProjector, TestProjectCache) {
   EXPECT_EQ(cached_projector, projector);
 
   // if configuration is different, should return a new projector.
+
+  // build a new path by replacing the first '/' with '//'
+  std::string alt_path(GANDIVA_BYTE_COMPILE_FILE_PATH);
+  auto pos = alt_path.find('/', 0);
+  EXPECT_NE(pos, std::string::npos);
+  alt_path.replace(pos, 1, "//");
   auto other_configuration =
-      ConfigurationBuilder()
-          .set_byte_code_file_path("/" + std::string(GANDIVA_BYTE_COMPILE_FILE_PATH))
-          .build();
+      ConfigurationBuilder().set_byte_code_file_path(alt_path).build();
   std::shared_ptr<Projector> should_be_new_projector2;
   status = Projector::Make(schema, {sum_expr, sub_expr}, other_configuration,
                            &should_be_new_projector2);

cpp/src/gandiva/to_date_holder.cc

@@ -82,12 +82,11 @@ int64_t ToDateHolder::operator()(ExecutionContext* context, const std::string& d
   // 1. processes date that do not match the format.
   // 2. does not process time in format +08:00 (or) id.
   int64_t seconds_since_epoch = 0;
-  if (!internal::ParseTimestamp(data.c_str(), pattern_.c_str(), &seconds_since_epoch)) {
+  if (!internal::ParseTimestamp(data.c_str(), pattern_.c_str(), true,
+                                &seconds_since_epoch)) {
     return_error(context, data);
     return 0;
   }
-  // Drop any time component
-  seconds_since_epoch -= seconds_since_epoch % 86400;
 
   *out_valid = true;
   return seconds_since_epoch * 1000;

cpp/src/gandiva/to_date_holder_test.cc

@@ -62,6 +62,18 @@ TEST_F(TestToDateHolder, TestSimpleDateTime) {
       to_date(&execution_context_, "1986-12-01 01:01:01 +0800", true, &out_valid);
   EXPECT_EQ(millis_since_epoch, 533779200000);
 
+#if 0
+  // TODO : this fails parsing with date::parse and strptime on linux
+  millis_since_epoch =
+      to_date(&execution_context_, "1886-12-01 00:00:00", true, &out_valid);
+  EXPECT_EQ(out_valid, true);
+  EXPECT_EQ(millis_since_epoch, -2621894400000);
+#endif
+
+  millis_since_epoch =
+      to_date(&execution_context_, "1886-12-01 01:01:01", true, &out_valid);
+  EXPECT_EQ(millis_since_epoch, -2621894400000);
+
   millis_since_epoch =
       to_date(&execution_context_, "1986-12-11 01:30:00", true, &out_valid);
   EXPECT_EQ(millis_since_epoch, 534643200000);