Unexpected parsing behavior of readr (returns 0 rows)

[seandavi]

This is a report related to seandavi/GEOquery#78. Here, I am comparing the parsing behavior of read.table and read_tsv. In particular, read_tsv returns a tibble <0x0>. Interestingly, if one replaces the GSE14308 with GSE14309, I get the expected behavior.

I have not tried this with a prior readr version yet, but I can if that is helpful.

destfile = 'GSE14308_series_matrix.txt.gz'
f = curl::curl_download("ftp://ftp.ncbi.nlm.nih.gov/geo/series/GSE14nnn/GSE14308/matrix/GSE14308_series_matrix.txt.gz",
                        destfile = destfile)

z1 = readr::read_tsv(destfile,
                 skip=65, col_names=TRUE, comment = '!')
#> Parsed with column specification:
#> cols()
dim(z1)
#> [1] 0 0

z2 = read.table(gzfile(destfile),
                     skip=65, header=TRUE, comment = '!')
dim(z2)
#> [1] 45101    13
sessionInfo()
#> R version 3.5.1 (2018-07-02)
#> Platform: x86_64-apple-darwin15.6.0 (64-bit)
#> Running under: macOS Sierra 10.12.6
#> 
#> Matrix products: default
#> BLAS: /Library/Frameworks/R.framework/Versions/3.5/Resources/lib/libRblas.0.dylib
#> LAPACK: /Library/Frameworks/R.framework/Versions/3.5/Resources/lib/libRlapack.dylib
#> 
#> locale:
#> [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
#> 
#> attached base packages:
#> [1] stats     graphics  grDevices utils     datasets  methods   base     
#> 
#> loaded via a namespace (and not attached):
#>  [1] Rcpp_1.0.0      crayon_1.3.4    digest_0.6.18   rprojroot_1.3-2
#>  [5] R6_2.3.0        backports_1.1.2 magrittr_1.5    evaluate_0.12  
#>  [9] pillar_1.3.0    rlang_0.3.0.1   stringi_1.2.4   curl_3.2       
#> [13] rmarkdown_1.10  tools_3.5.1     stringr_1.3.1   readr_1.3.0    
#> [17] hms_0.4.2       yaml_2.2.0      compiler_3.5.1  pkgconfig_2.0.2
#> [21] htmltools_0.3.6 knitr_1.20      tibble_1.4.2

^{Created on 2018-12-13 by the reprex package (v0.2.1)}

[jimhester]

Thanks, this should now be fixed. There was a bug in how quoted strings were handled during the skip parsing. The 'GSE14308_series_matrix' file had

!Series_contributor "John,J,O'Shea"

in line 14, and because the single quote in O'Shea was treated the same as the double quote this meant the quotes were mismatched, and the rest of the data was essentially ignored.

Should now be fixed however, thank you for opening the issue!

[seandavi]

Thanks, @jimhester, for the quick diagnosis and fix. Will this be coming out in a bug-fix release anytime soon? I'd rather not rely on having folks install from github when using GEOquery, but that decision will be driven by timing of release.

[assaron]

@jimhester thanks for the quick fix. I'm also interested in this getting into a release. Are there any estimates for this?

[stufield]

I also ran into this issue. @jimhester seems to have quickly tracked it down to terminating single quotes during skip lines. Just for future reference in case someone else runs across this with v1.3.0 installed, here's a simplified reprex of the issue:

library(readr)
packageVersion("readr")            # broken in 1.3.0; fixed in 1.3.1
txt <- "\t\t\tTarget\t5'-thymine\n\t\t\tNew\tUnknown"  # note the single quote (5')
read_lines(txt)                        # works as expected; both lines
read_lines(txt, n_max = 1)             # works as expected; 1st line
read_lines(txt, n_max = 1, skip = 0)   # works as expected; 1st line
read_lines(txt, n_max = 1, skip = 1)   # unexpected empty character(0); 2nd line
txt <- "\t\t\tTarget\t5'-'thymine\n\t\t\tNew\tUnknown"  # note closing quote
read_lines(txt, n_max = 1, skip = 1)   # works as expected with the closing single quote

Thanks @jimhester for getting a bug fix version 1.3.1 out to CRAN so quickly!

[lock]

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/