Still adjusting double -> (U)INT64 conversion limits

src/RParquetOutFile.cpp

@@ -1081,11 +1081,11 @@ void RParquetOutFile::write_double_int64(std::ostream &file, SEXP col,
     }
     if (is_signed) {
       // smallest & largest double that can be put into an int64_t
-      double min = -9223372036854775295.0, max = 9223372036854775295.0;
+      double min = -9223372036854774900.0, max = 9223372036854774900.0;
       for (uint64_t i = from; i < until; i++) {
         double val = REAL(col)[i];
         if (R_IsNA(val)) continue;
-        const char *w = val < min ? "small" : (val > max ? "large" : "");
+        const char *w = val <= min ? "small" : (val >= max ? "large" : "");
         if (w[0]) {
           r_call([&] {
             Rf_errorcall(
@@ -1110,7 +1110,7 @@ void RParquetOutFile::write_double_int64(std::ostream &file, SEXP col,
       has_minmax_value[idx] = has_minmax_value[idx] || has_min;
     } else {
       // largest double that can be put into an uint64_t
-      double max = 18446744073709550591.0;
+      double max = 18446744073709550592.0;
       uint64_t min_value = 0, max_value = 0;
       bool has_min = false, has_max = false;
       bool minmax = write_minmax_values && is_minmax_supported[idx];
@@ -1121,7 +1121,7 @@ void RParquetOutFile::write_double_int64(std::ostream &file, SEXP col,
       for (uint64_t i = from; i < until; i++) {
         double val = REAL(col)[i];
         if (R_IsNA(val)) continue;
-        if (val > max) {
+        if (val >= max) {
           r_call([&] {
             Rf_errorcall(
               nanoparquet_call,

tests/testthat/_snaps/write-parquet-3.md

@@ -1590,15 +1590,15 @@
       write_parquet(d, tmp, schema = parquet_schema("INT_64"))
     Condition
       Error in `write_parquet()`:
-      ! Integer value too large for INT with bit width 64: 9223372036854775808.000000 at column 1, row 2.
+      ! Integer value too large for INT with bit width 64: 922337203685477xxxx.000000 at column 1, row 2.
 
 ---
 
     Code
       write_parquet(d, tmp, schema = parquet_schema("INT_64"))
     Condition
       Error in `write_parquet()`:
-      ! Integer value too small for INT with bit width 64: -9223372036854775808.000000 at column 1, row 2.
+      ! Integer value too small for INT with bit width 64: -922337203685477xxxx.000000 at column 1, row 2.
 
 ---
 
@@ -1629,7 +1629,7 @@
       write_parquet(d, tmp, schema = parquet_schema("UINT_64"))
     Condition
       Error in `write_parquet()`:
-      ! Integer value too large for unsigned INT with bit width 64: 18446744073709551616.000000 at column 1, row 2.
+      ! Integer value too large for unsigned INT with bit width 64: 18446744073709551616.000000 at column 1, row 1.
 
 ---
 

tests/testthat/helper.R

@@ -39,3 +39,7 @@ test_write <- function(d, schema = NULL, encoding = NULL) {
     as.data.frame(read_parquet(tmp))
   })
 }
+
+redact_maxint64 <- function(x) {
+  gsub("922337203685477[0-9][0-9][0-9][0-9]", "922337203685477xxxx", x)
+}

tests/testthat/test-write-parquet-3.R

@@ -564,14 +564,14 @@ test_that("double to INT(64, *)", {
     as.data.frame(read_parquet(tmp))
   })
 
-  d <- data.frame(d = c(9223372036854775295, 9223372036854775296))
+  d <- data.frame(d = c(9223372036854774899, 9223372036854774900))
   expect_snapshot(error = TRUE, {
     write_parquet(d, tmp, schema = parquet_schema("INT_64"))
-  })
-  d <- data.frame(d = -c(9223372036854775295, 9223372036854775296))
+  }, transform = redact_maxint64)
+  d <- data.frame(d = -c(9223372036854774899, 9223372036854774900))
   expect_snapshot(error = TRUE, {
     write_parquet(d, tmp, schema = parquet_schema("INT_64"))
-  })
+  }, transform = redact_maxint64)
 
   d <- data.frame(d = as.double(c(0:5, NA)))
   write_parquet(d, tmp, schema = parquet_schema("UINT_64"))