讀取/寫入CSV文件時出現日期問題

Question

將在IST時區中具有datetime列的.csv文件加載到RStudio時，會將其轉換為UTC語言環境（-5：30）。 因此，我的子集比較檢查沒有給出確切的結果。 例如。

原始字符串，如csv文件中所示：

> coilid$StartTime[1:5]
[1] "06/01/2017 00:29:10" "06/01/2017 01:19:19" "06/01/2017 02:09:31" "06/01/2017 03:05:08"
[5] "06/01/2017 03:48:51"

使用lubridate加載：

lub_date = mdy_hms(coilid$StartTime[1:5])
> lub_date
[1] "2017-06-01 00:29:10 UTC" "2017-06-01 01:19:19 UTC" "2017-06-01 02:09:31 UTC" "2017-06-01 03:05:08 UTC"
[5] "2017-06-01 03:48:51 UTC"

字符串應被視為IST日期。 默認為UTC

> lub_date[1] > '2017-06-01 05:58:00'
[1] TRUE
> lub_date[1] > '2017-06-01 05:59:59'
[1] FALSE

因此增加了tz

lub_date_ist = mdy_hms(coilid$StartTime[1:5], tz = 'Asia/Calcutta')
> lub_date_ist
[1] "2017-06-01 00:29:10 IST" "2017-06-01 01:19:19 IST" "2017-06-01 02:09:31 IST" "2017-06-01 03:05:08 IST"
[5] "2017-06-01 03:48:51 IST"
> lub_date_ist[1] > '2017-06-01 00:29:59'
[1] FALSE
> lub_date_ist[1] > '2017-06-01 00:29:00'
[1] TRUE

現在的問題是，當我將它另存為data.table到csv文件中並再次讀取時，數據會損壞。

fwrite(data.table(lub_date_ist), file = paste0(path_loc, 'xxx.csv'))
fread(file = paste0(path_loc, 'xxx.csv'))
              V1 V2  V3
1: 2017-05-31T18 59 10Z
2: 2017-05-31T19 49 19Z
3: 2017-05-31T20 39 31Z
4: 2017-05-31T21 35 08Z
5: 2017-05-31T22 18 51Z

Answer 1

想法是以UTC格式在file/database中write/maintain日期/時間。 以locale格式存儲日期/時間既困難又容易出錯。 特別是在應用DST時區。

實現它的步驟非常簡單。 即將寫入文件之前，將日期/時間轉換為UTC格式。 同樣，以UTC格式讀取它，然后將其轉換為語言環境時區。 with_tz的使用對於將日期/時間轉換為另一個時區非常重要。

請允許我用OP中的數據進行演示。

library(lubridate)
coild <-  read.table(text = 'Sl Time
1 "06/01/2017 00:29:10" 
2 "06/01/2017 01:19:19"
3 "06/01/2017 02:09:31"
4 "06/01/2017 03:05:08"
5  "06/01/2017 03:48:51"', header = T, stringsAsFactors = F)

# The time is in IST format
coild$Time <- mdy_hms(coild$Time, tz="Asia/Calcutta")
[1] "2017-06-01 00:29:10 IST" "2017-06-01 01:19:19 IST" "2017-06-01 02:09:31 IST" 
[4] "2017-06-01 03:05:08 IST" "2017-06-01 03:48:51 IST"

# Convert it to "UTC" just before writing to file
fwrite(data.table(with_tz(coild$Time, tzone = "UTC")), file = 'xxx.csv')

# Read and convert it date/time in UTC format
DT <- fread(file = 'xxx.csv', sep = " ", colClasses=c("POSIXt"))
# Date is in UTC format
readTime <- parse_date_time(DT$V1, "%Y-%m-%d %H:%M:%S%z")

# Change date/time in IST 
with_tz(readTime, tzone = "Asia/Calcutta")

[1] "2017-06-01 00:29:10 IST" "2017-06-01 01:19:19 IST" "2017-06-01 02:09:31 IST"
[4] "2017-06-01 03:05:08 IST" "2017-06-01 03:48:51 IST"