|
15 | 15 | */ |
16 | 16 | #pragma once |
17 | 17 |
|
| 18 | +#include <boost/algorithm/string/case_conv.hpp> |
18 | 19 | #include <velox/type/Timestamp.h> |
19 | 20 | #include "velox/core/QueryConfig.h" |
| 21 | +#include "velox/expression/ComplexViewTypes.h" |
20 | 22 | #include "velox/external/date/date.h" |
21 | 23 | #include "velox/external/date/iso_week.h" |
22 | 24 | #include "velox/functions/Macros.h" |
| 25 | +#include "velox/functions/lib/DateTimeFormatter.h" |
23 | 26 | #include "velox/type/tz/TimeZoneMap.h" |
24 | 27 |
|
25 | 28 | namespace facebook::velox::functions { |
@@ -123,4 +126,205 @@ struct InitSessionTimezone { |
123 | 126 | timeZone_ = getTimeZoneFromConfig(config); |
124 | 127 | } |
125 | 128 | }; |
| 129 | + |
| 130 | +/// Converts string as date time unit. Throws for invalid input string. |
| 131 | +/// |
| 132 | +/// @param unitString The input string to represent date time unit. |
| 133 | +/// @param throwIfInvalid Whether to throw an exception for invalid input |
| 134 | +/// string. |
| 135 | +/// @param allowMicro Whether to allow microsecond. |
| 136 | +/// @param allowAbbreviated Whether to allow abbreviated unit string. |
| 137 | +FOLLY_ALWAYS_INLINE std::optional<DateTimeUnit> fromDateTimeUnitString( |
| 138 | + const StringView& unitString, |
| 139 | + bool throwIfInvalid, |
| 140 | + bool allowMicro = false, |
| 141 | + bool allowAbbreviated = false) { |
| 142 | + const auto unit = boost::algorithm::to_lower_copy(unitString.str()); |
| 143 | + |
| 144 | + if (unit == "microsecond" && allowMicro) { |
| 145 | + return DateTimeUnit::kMicrosecond; |
| 146 | + } |
| 147 | + if (unit == "millisecond") { |
| 148 | + return DateTimeUnit::kMillisecond; |
| 149 | + } |
| 150 | + if (unit == "second") { |
| 151 | + return DateTimeUnit::kSecond; |
| 152 | + } |
| 153 | + if (unit == "minute") { |
| 154 | + return DateTimeUnit::kMinute; |
| 155 | + } |
| 156 | + if (unit == "hour") { |
| 157 | + return DateTimeUnit::kHour; |
| 158 | + } |
| 159 | + if (unit == "day") { |
| 160 | + return DateTimeUnit::kDay; |
| 161 | + } |
| 162 | + if (unit == "week") { |
| 163 | + return DateTimeUnit::kWeek; |
| 164 | + } |
| 165 | + if (unit == "month") { |
| 166 | + return DateTimeUnit::kMonth; |
| 167 | + } |
| 168 | + if (unit == "quarter") { |
| 169 | + return DateTimeUnit::kQuarter; |
| 170 | + } |
| 171 | + if (unit == "year") { |
| 172 | + return DateTimeUnit::kYear; |
| 173 | + } |
| 174 | + if (allowAbbreviated) { |
| 175 | + if (unit == "dd") { |
| 176 | + return DateTimeUnit::kDay; |
| 177 | + } |
| 178 | + if (unit == "mon" || unit == "mm") { |
| 179 | + return DateTimeUnit::kMonth; |
| 180 | + } |
| 181 | + if (unit == "yyyy" || unit == "yy") { |
| 182 | + return DateTimeUnit::kYear; |
| 183 | + } |
| 184 | + } |
| 185 | + if (throwIfInvalid) { |
| 186 | + VELOX_UNSUPPORTED("Unsupported datetime unit: {}", unitString); |
| 187 | + } |
| 188 | + return std::nullopt; |
| 189 | +} |
| 190 | + |
| 191 | +/// Adjusts the given date time object to the start of the specified date time |
| 192 | +/// unit (e.g., year, quarter, month, week, day, hour, minute). |
| 193 | +FOLLY_ALWAYS_INLINE void adjustDateTime( |
| 194 | + std::tm& dateTime, |
| 195 | + const DateTimeUnit& unit) { |
| 196 | + switch (unit) { |
| 197 | + case DateTimeUnit::kYear: |
| 198 | + dateTime.tm_mon = 0; |
| 199 | + dateTime.tm_yday = 0; |
| 200 | + FMT_FALLTHROUGH; |
| 201 | + case DateTimeUnit::kQuarter: |
| 202 | + dateTime.tm_mon = dateTime.tm_mon / 3 * 3; |
| 203 | + FMT_FALLTHROUGH; |
| 204 | + case DateTimeUnit::kMonth: |
| 205 | + dateTime.tm_mday = 1; |
| 206 | + dateTime.tm_hour = 0; |
| 207 | + dateTime.tm_min = 0; |
| 208 | + dateTime.tm_sec = 0; |
| 209 | + break; |
| 210 | + case DateTimeUnit::kWeek: |
| 211 | + // Subtract the truncation. |
| 212 | + dateTime.tm_mday -= dateTime.tm_wday == 0 ? 6 : dateTime.tm_wday - 1; |
| 213 | + // Setting the day of the week to Monday. |
| 214 | + dateTime.tm_wday = 1; |
| 215 | + |
| 216 | + // If the adjusted day of the month falls in the previous month |
| 217 | + // Move to the previous month. |
| 218 | + if (dateTime.tm_mday < 1) { |
| 219 | + dateTime.tm_mon -= 1; |
| 220 | + |
| 221 | + // If the adjusted month falls in the previous year |
| 222 | + // Set to December and Move to the previous year. |
| 223 | + if (dateTime.tm_mon < 0) { |
| 224 | + dateTime.tm_mon = 11; |
| 225 | + dateTime.tm_year -= 1; |
| 226 | + } |
| 227 | + |
| 228 | + // Calculate the correct day of the month based on the number of days |
| 229 | + // in the adjusted month. |
| 230 | + static const int daysInMonth[] = { |
| 231 | + 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31}; |
| 232 | + int daysInPrevMonth = daysInMonth[dateTime.tm_mon]; |
| 233 | + |
| 234 | + // Adjust for leap year if February. |
| 235 | + if (dateTime.tm_mon == 1 && (dateTime.tm_year + 1900) % 4 == 0 && |
| 236 | + ((dateTime.tm_year + 1900) % 100 != 0 || |
| 237 | + (dateTime.tm_year + 1900) % 400 == 0)) { |
| 238 | + daysInPrevMonth = 29; |
| 239 | + } |
| 240 | + // Set to the correct day in the previous month. |
| 241 | + dateTime.tm_mday += daysInPrevMonth; |
| 242 | + } |
| 243 | + dateTime.tm_hour = 0; |
| 244 | + dateTime.tm_min = 0; |
| 245 | + dateTime.tm_sec = 0; |
| 246 | + break; |
| 247 | + case DateTimeUnit::kDay: |
| 248 | + dateTime.tm_hour = 0; |
| 249 | + FMT_FALLTHROUGH; |
| 250 | + case DateTimeUnit::kHour: |
| 251 | + dateTime.tm_min = 0; |
| 252 | + FMT_FALLTHROUGH; |
| 253 | + case DateTimeUnit::kMinute: |
| 254 | + dateTime.tm_sec = 0; |
| 255 | + break; |
| 256 | + default: |
| 257 | + VELOX_UNREACHABLE(); |
| 258 | + } |
| 259 | +} |
| 260 | + |
| 261 | +/// Returns timestamp with seconds adjusted to the nearest lower multiple of the |
| 262 | +/// specified interval. If the given seconds is negative and not an exact |
| 263 | +/// multiple of the interval, it adjusts further down. |
| 264 | +FOLLY_ALWAYS_INLINE Timestamp |
| 265 | +adjustEpoch(int64_t seconds, int64_t intervalSeconds) { |
| 266 | + int64_t s = seconds / intervalSeconds; |
| 267 | + if (seconds < 0 && seconds % intervalSeconds) { |
| 268 | + s = s - 1; |
| 269 | + } |
| 270 | + int64_t truncatedSeconds = s * intervalSeconds; |
| 271 | + return Timestamp(truncatedSeconds, 0); |
| 272 | +} |
| 273 | + |
| 274 | +// Returns timestamp truncated to the specified unit. |
| 275 | +FOLLY_ALWAYS_INLINE Timestamp truncateTimestamp( |
| 276 | + const Timestamp& timestamp, |
| 277 | + DateTimeUnit unit, |
| 278 | + const tz::TimeZone* timeZone) { |
| 279 | + Timestamp result; |
| 280 | + switch (unit) { |
| 281 | + // For seconds ,millisecond, microsecond we just truncate the nanoseconds |
| 282 | + // part of the timestamp; no timezone conversion required. |
| 283 | + case DateTimeUnit::kMicrosecond: |
| 284 | + return Timestamp( |
| 285 | + timestamp.getSeconds(), timestamp.getNanos() / 1000 * 1000); |
| 286 | + |
| 287 | + case DateTimeUnit::kMillisecond: |
| 288 | + return Timestamp( |
| 289 | + timestamp.getSeconds(), timestamp.getNanos() / 1000000 * 1000000); |
| 290 | + |
| 291 | + case DateTimeUnit::kSecond: |
| 292 | + return Timestamp(timestamp.getSeconds(), 0); |
| 293 | + |
| 294 | + // Same for minutes; timezones and daylight savings time are at least in |
| 295 | + // the granularity of 30 mins, so we can just truncate the epoch directly. |
| 296 | + case DateTimeUnit::kMinute: |
| 297 | + return adjustEpoch(timestamp.getSeconds(), 60); |
| 298 | + |
| 299 | + // Hour truncation has to handle the corner case of daylight savings time |
| 300 | + // boundaries. Since conversions from local timezone to UTC may be |
| 301 | + // ambiguous, we need to be carefull about the roundtrip of converting to |
| 302 | + // local time and back. So what we do is to calculate the truncation delta |
| 303 | + // in UTC, then applying it to the input timestamp. |
| 304 | + case DateTimeUnit::kHour: { |
| 305 | + auto epochToAdjust = getSeconds(timestamp, timeZone); |
| 306 | + auto secondsDelta = |
| 307 | + epochToAdjust - adjustEpoch(epochToAdjust, 60 * 60).getSeconds(); |
| 308 | + return Timestamp(timestamp.getSeconds() - secondsDelta, 0); |
| 309 | + } |
| 310 | + |
| 311 | + // For the truncations below, we may first need to convert to the local |
| 312 | + // timestamp, truncate, then convert back to GMT. |
| 313 | + case DateTimeUnit::kDay: |
| 314 | + result = adjustEpoch(getSeconds(timestamp, timeZone), 24 * 60 * 60); |
| 315 | + break; |
| 316 | + |
| 317 | + default: |
| 318 | + auto dateTime = getDateTime(timestamp, timeZone); |
| 319 | + adjustDateTime(dateTime, unit); |
| 320 | + result = Timestamp(Timestamp::calendarUtcToEpoch(dateTime), 0); |
| 321 | + break; |
| 322 | + } |
| 323 | + |
| 324 | + if (timeZone != nullptr) { |
| 325 | + result.toGMT(*timeZone); |
| 326 | + } |
| 327 | + return result; |
| 328 | +} |
| 329 | + |
126 | 330 | } // namespace facebook::velox::functions |
0 commit comments