Passing ... logicals as T or F fails through dfm(x, ...)

[kbenoit]

This needs to be more robust, since too many people (sadly) use the (sadly) available shortcuts T and F instead of TRUE and FALSE:

> dfm("This: contains punctuation.", remove_punct = TRUE)
Document-feature matrix of: 1 document, 3 features (0% sparse).
1 x 3 sparse Matrix of class "dfmSparse"
       features
docs    this contains punctuation
  text1    1        1           1
> dfm("This: contains punctuation.", remove_punct = T)
 Error in eval(expr, envir, enclos) : 'nthcdr' needs a list to CDR down 

[koheiw]

It seems that both T and F are converted to a string '..1' when passed through ellipsis to fun2. We cannot do much for this problem. The best solution is to stop translating the old arguments to new...

fun1 <- function(a, b, ...) {
    args <- as.list(match.call())
    cat("fun1 args = \n")
    print(args)
    fun2(b, ...)
}

fun2 <- function(b, ...){
    args <- as.list(match.call())
    cat("fun2 args = \n")
    print(args)
    cat("b =", b , "\n")
}

fun1(b = TRUE, c = TRUE) # OK
fun1(b = FALSE, c = FALSE) # OK

fun1(b = T, c = T)
# fun1 args = 
#     [[1]]
# fun1
# 
# $b
# T
# 
# $c
# T
# 
# fun2 args = 
#     [[1]]
# fun2
# 
# $b
# b
# 
# $c
# ..1
# 
# b = TRUE 

fun1(b = F, c = F)
# fun1 args = 
#     [[1]]
# fun1
# 
# $b
# F
# 
# $c
# F
# 
# fun2 args = 
#     [[1]]
# fun2
# 
# $b
# b
# 
# $c
# ..1
# 
# b = FALSE 

[trinker]

I believe this is related. If not tell me and I'll file an issue elsewhere:

txt <- c("This is software testing: looking for (word) pairs the dog! peas and carrots",  
         "This [is] a software testing again. For. the dog want the book peas and carrots ",
         "Here: this is more Software Testing, want the book looking again for word pairs. peas and carrots ")

x <- !TRUE

quanteda::dfm(txt, stem = FALSE, verbose = FALSE, remove_numbers = FALSE, remove_punct = TRUE)
quanteda::dfm(txt, stem = FALSE, verbose = FALSE, remove_numbers = FALSE, remove_punct = FALSE)
quanteda::dfm(txt, stem = FALSE, verbose = FALSE, remove_numbers = FALSE, remove_punct =!TRUE)
quanteda::dfm(txt, stem = FALSE, verbose = FALSE, remove_numbers = FALSE, remove_punct =x)

The first two dtm calls work. The later two throw the error:

 Error in eval(a, enclos = parent.frame()) : 
  the ... list does not contain 2 elements 

This is particularly problematic when quanteda is being used by another package and needs to pass these arguments in by reference.

[ArthurSpirling]

Earlier

dfm("This: contains punctuation.", remove_punct = TRUE)
returned

Error in qatd_cpp_tokens_select(x, types, features_id, 2, padding) : expecting a single value

for me. This was using the github version of quanteda. But the latest CRAN version seems ok.

[kbenoit]

Odd... works fine for me:

> sessionInfo()
R version 3.4.0 (2017-04-21)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS Sierra 10.12.4

Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.4/Resources/lib/libRlapack.dylib

locale:
[1] en_GB.UTF-8/en_GB.UTF-8/en_GB.UTF-8/C/en_GB.UTF-8/en_GB.UTF-8

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

other attached packages:
[1] quanteda_0.9.9-61

loaded via a namespace (and not attached):
 [1] Rcpp_0.12.10        lattice_0.20-35     grid_3.4.0          plyr_1.8.4         
 [5] gtable_0.2.0        scales_0.4.1        RcppParallel_4.3.20 ggplot2_2.2.1      
 [9] stringi_1.1.5       lazyeval_0.2.0      data.table_1.10.4   Matrix_1.2-9       
[13] fastmatch_1.1-0     tools_3.4.0         munsell_0.4.3       compiler_3.4.0     
[17] colorspace_1.3-2    tibble_1.3.0       

> dfm("This: contains punctuation.", remove_punct = TRUE)
Document-feature matrix of: 1 document, 3 features (0% sparse).
1 x 3 sparse Matrix of class "dfmSparse"
       features
docs    this contains punctuation
  text1    1        1           1

[kbenoit]

I can fix this, but it will involve rewriting the internal deprecate_argument() in tokens(). I'll get working on this asap.

While personally I believe punishment is warranted for using T and F 😉, @trinker's issue is clearly a problem that needs to be solved.