Data.table: fread mungkin gagal mengurai file yang valid saat dec = ','

Dibuat pada 13 Apr 2018 · 3Komentar · Sumber: Rdatatable/data.table

> fread('a,b,c,d\n1e1,1e2,1e3,"4,0001"\n1,2,3,4\n', dec=',')
       a     b     c      d
   <num> <num> <num>  <num>
1:    10   100  1000 4.0001
Warning message:
In fread("a,b,c,d\n1e1,1e2,1e3,\"4,0001\"\n1,2,3,4\n", dec = ",") :
  Discarded single-line footer: <<1,2,3,4>>

Ini terjadi karena float parser dengan rakus mengonsumsi 1,2 sebagai token tunggal, sedangkan tanpa tanda kutip, ia harus diurai sebagai 2 bidang terpisah.

Selain itu, bagian "Detail" dalam dokumentasi memiliki informasi berikut (yang sudah lama tidak berlaku lagi):

'fread' menggunakan fungsi C 'strtod' untuk membaca data numerik; misalnya,
'1.23' atau '1,23'. 'strtod' mengambil pemisah desimal ('.' atau
',' biasanya) dari lokal sesi R daripada sebagai
argumen yang diteruskan ke fungsi 'strtod'. Maka untuk
'fread (..., dec = ",")' to work, 'fread' mengubah ini (dan hanya ini)
Lokal sesi R untuk sementara ke lokal yang menyediakan
pemisah desimal yang diinginkan.

bug fread

Sumber

st-pasha

👍1

Semua 3 komentar

Paket luar biasa! Terima kasih.
Secara kebetulan, saya mengalami perilaku aneh yang mungkin terkait dengan ketakutan kemarin.
Berikut adalah contoh sederhana (berasal dari kasus nyata menggunakan fread lapply pada banyak file).

library(data.table)

DT = data.table(A = rep("20,1", 1e4))
fwrite(DT, "DT.csv", quote = FALSE)

classA = character(1e3)

for (i in seq_along(classA)) {

  DT = fread("DT.csv", sep = ";", dec = ",", colClasses = "numeric")
  classA[i] = DT[, class(A)]
}

table(classA)

Ini memberi saya:

Warning message:
In fread("DT.csv", sep = ";", dec = ",", colClasses = "numeric",  :
  Bumped column 1 to type character on data row 387, field contains '20,1'. Coercing previously read values in this column from logical, integer or numeric back to character which may not be lossless; e.g., if '00' and '000' occurred before they will now be just '0', and there may be inconsistencies with treatment of ',,' and ',NA,' too (if they occurred in this column before the bump). If this matters please rerun and set 'colClasses' to 'character' for this column. Please note that column type detection uses a sample of 1,000 rows (100 rows at 10 points) so hopefully this message should be very rare. If reporting to datatable-help, please rerun and include the output from verbose=TRUE.

table(classA)
# classA
# character   numeric 
#        1       999 
which(classA == "character")
# [1] 9
`````

So, sometimes, the column is read as character with the spotted row index being different for different runs. And the ```which``` indicates it is more likely to happen in the first iterations. I don't get the randomness and the fact that colClasses is ignored (I set colClasses after reading fread doc...).
And... I did not manage to reproduce the error when setting verbose = TRUE... Also, the bug was observed using RStudio, not reproduced in the R console...
Sorry if I missed or misunderstood something...

```r
sessionInfo()
R version 3.3.2 (2016-10-31)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 7 x64 (build 7601) Service Pack 1

locale:
[1] LC_COLLATE=English_United States.1252  LC_CTYPE=English_United States.1252   
[3] LC_MONETARY=English_United States.1252 LC_NUMERIC=C                          
[5] LC_TIME=English_United States.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] data.table_1.10.4-3

loaded via a namespace (and not attached):
[1] tools_3.3.2 yaml_2.1.18