Integer+Alpha codes parse as numeric

[barryrowlingson]

I have a CSV column with area codes that are of the form "13T". They are guessed as numeric, and the trailing char is lost:

> parse_guess(c("13T","13T","21R"))
[1] 13 13 21

Unless there's a zero:

> parse_guess(c("13T","13T","0F"))
[1] "13T" "13T" "0F" 
> parse_guess(c("13T","13T","00F"))
[1] "13T" "13T" "00F"

I'd prefer these ("13T" etc) to be parsed as character. I can't think of a context where this pattern would be numeric. Except in special cases, such as issue #316, where he has 21N, 13.5E etc for coordinates. He gets numerics (but I'm not sure he wants them) except in a column that has a "0N" which returns as character...

more than one trailing alphabetical seems to be enough to trigger character guessing:

> parse_guess(c("13T","13T","10N"))
[1] 13 13 10
> parse_guess(c("13T","13T","10NN"))
[1] "13T"  "13T"  "10NN"

Seems a bit inconsistent...

readr_0.2.2

[benmarwick]

I've run into this problem also. Here's a reproducible example:

df <- data.frame(one = letters[1:10],
                 two = 1:10,
                 three = sapply(1:10, function(i) paste0(i, "L")))
df
##    one two three
## 1    a   1    1L
## 2    b   2    2L
## 3    c   3    3L
## 4    d   4    4L
## 5    e   5    5L
## 6    f   6    6L
## 7    g   7    7L
## 8    h   8    8L
## 9    i   9    9L
## 10   j  10   10L
str(df)
## 'data.frame':    10 obs. of  3 variables:
##  $ one  : Factor w/ 10 levels "a","b","c","d",..: 1 2 3 4 5 6 7 8 9 10
##  $ two  : int  1 2 3 4 5 6 7 8 9 10
##  $ three: Factor w/ 10 levels "10L","1L","2L",..: 2 3 4 5 6 7 8 9 10 1

If I use base::read.csv I see the "L" in the data still:

write.csv(df, "df.csv")
df_in <- read.csv("df.csv")
str(df_in)
## 'data.frame':    10 obs. of  4 variables:
##  $ X    : int  1 2 3 4 5 6 7 8 9 10
##  $ one  : Factor w/ 10 levels "a","b","c","d",..: 1 2 3 4 5 6 7 8 9 10
##  $ two  : int  1 2 3 4 5 6 7 8 9 10
##  $ three: Factor w/ 10 levels "10L","1L","2L",..: 2 3 4 5 6 7 8 9 10 1

But when I use readr::read_csv, it's coerced to numeric and I lose the "L" from my data, which is quite unexpected:

library(readr)
## Warning: package 'readr' was built under R version 3.2.3
df_in_ <- read_csv("df.csv")
str(df_in_)
## Classes 'tbl_df', 'tbl' and 'data.frame':    10 obs. of  4 variables:
##  $      : int  1 2 3 4 5 6 7 8 9 10
##  $ one  : chr  "a" "b" "c" "d" ...
##  $ two  : int  1 2 3 4 5 6 7 8 9 10
##  $ three: num  1 2 3 4 5 6 7 8 9 10

sessionInfo()
R version 3.2.2 (2015-08-14)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 7 x64 (build 7601) Service Pack 1

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

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

other attached packages:
[1] readr_0.2.2   plyr_1.8.3    Bchron_4.1.2  inline_0.3.14 knitr_1.11   
[6] ktc11_0.1    

loaded via a namespace (and not attached):
 [1] Rcpp_0.12.1      highr_0.5.1      tools_3.2.2      digest_0.6.8    
 [5] mclust_5.1       G2Sd_2.1.5       gtable_0.1.2     lattice_0.20-33 
 [9] shiny_0.12.2     DBI_0.3.1        yaml_2.1.13      parallel_3.2.2  
[13] proto_0.3-10     rJava_0.9-7      coda_0.18-1      stringr_1.0.0   
[17] dplyr_0.4.3      xlsxjars_0.6.1   grid_3.2.2       hdrcde_3.1      
[21] ellipse_0.3-8    R6_2.1.1         rmarkdown_0.9.2  reshape2_1.4.1  
[25] ggplot2_1.0.1    magrittr_1.5     scales_0.3.0     htmltools_0.2.6 
[29] MASS_7.3-43      assertthat_0.1   mime_0.4         xtable_1.8-0    
[33] colorspace_1.2-6 httpuv_1.3.3     xlsx_0.5.7       stringi_1.0-1   
[37] munsell_0.4.2 

[hadley]

Yeah, this is a duplicate of #316