Merge branch 'master' into fix-3339

Author: Asa-Henry

NEWS.md

@@ -50,7 +50,9 @@
 
 9. `fread()` no longer replaces a literal header column name `"NA"` with an auto-generated `Vn` name when `na.strings` includes `"NA"`, [#5124](https://github.com/Rdatatable/data.table/issues/5124). Data rows still continue to parse `"NA"` as missing. Thanks @Mashin6 for the report and @shrektan for the fix.
 
-10. `fread()` would not give a warning when every second line of input was empty, [#3339](https://github.com/Rdatatable/data.table/issues/3339). Now, a warning message 'The rows in this file appear to be separated by blank lines.' is given and suggests to set `blank.lines.skip` to `TRUE`. Thanks to @Henrik-P for the report and @Asa-Henry for the fix.
+10. `fread()` no longer misreads dates with negative years, [#7704](https://github.com/Rdatatable/data.table/issues/7704). Thanks to @kevinushey for the report and @aitap for the fix.
+
+11. `fread()` would not give a warning when every second line of input was empty, [#3339](https://github.com/Rdatatable/data.table/issues/3339). Now, a warning message 'The rows in this file appear to be separated by blank lines.' is given and suggests to set `blank.lines.skip` to `TRUE`. Thanks to @Henrik-P for the report and @Asa-Henry for the fix.
 
 ### Notes
 

inst/tests/tests.Rraw

@@ -21587,3 +21587,9 @@ close(con)
 file.create(f <- tempfile())
 test(2367.6, fread(file(f)), data.table(), warning="Connection has size 0.")
 unlink(f)
+
+# negative years caused UB in leap year calculation, #7704
+x = fread("x\n-1-01-01")$x
+test(2368.1, year(x), -1L)
+test(2368.2, month(x), 1L)
+test(2368.3, mday(x),  1L)

src/fread.c

@@ -1087,9 +1087,13 @@ static void parse_iso8601_date_core(const char **pch, int32_t *target)
   if (day == NA_INT32 || day < 1 || (day > (isLeapYear ? leapYearDays[month - 1] : normYearDays[month - 1])))
     return;
 
+  int32_t cycle_year = year % 400;
+  if (cycle_year < 0) cycle_year += 400;
+  int32_t cycle = (year - cycle_year) / 400;
+
   *target =
-    (year / 400 - 4) * cumDaysCycleYears[400] + // days to beginning of 400-year cycle
-    cumDaysCycleYears[year % 400] + // days to beginning of year within 400-year cycle
+    (cycle - 4) * cumDaysCycleYears[400] + // days to beginning of 400-year cycle
+    cumDaysCycleYears[cycle_year] + // days to beginning of year within 400-year cycle
     (isLeapYear ? cumDaysCycleMonthsLeap[month - 1] : cumDaysCycleMonthsNorm[month - 1]) + // days to beginning of month within year
     day - 1; // day within month (subtract 1: 1970-01-01 -> 0)