[Rd] BUG: HMTL-based help.search() on vignettes may generate error (with PATCH)

[Henrik Bengtsson]

UPDATE: This has been fixed in R-devel (rev 73170) by Kurt H.  I
checked with him and he'll also try to get it into the upcoming R
3.4.2 (ETA 2017-09-28).

/Henrik

On Tue, Aug 29, 2017 at 4:19 PM, Henrik Bengtsson
<henrik.bengtsson at gmail.com> wrote:
> REPRODUCIBLE EXAMPLE:
>
> With the R.rsp package installed, the following search, which gives a hit:
>
>     options(help_type = "html")
>     help.search("rsp")
>
> generates:
>
>     Error in if (nchar(Outfile)) Outfile else File :
>       argument is not interpretable as logical
>
> in the browser (e.g. http://127.0.0.1:30410/doc/html/Search?results=1).
>
> Another example is help.search("tm") with jsonlite installed.
>
>
> TROUBLESHOOTING:
>
> This occurs because utils:::merge_vignette_index() uses:
>
>     base[, "Name"] <- sub("\\.[^.]*$", "", basename(vDB$File))
>     base[, "Topic"] <- base[, "Name"]
>
> which assumes that the 'Name' (and hence the 'Topic') can be inferred
> from the basename of the vignette source file by dropping the filename
> extension.  This assumption was valid in R (< 3.0.0), but with the
> introduction of generic vignette engines, we may now have multiple
> ("nested") file-name extensions on vignette source files.  For
> instance, vignette source file 'vignette.tex.rsp' outputs
> 'vignette.pdf'.  Another example is 'vignette.pdf.asis' that outputs
> 'vignette.pdf'.
>
> Now, the assumption on a single file-name extension is still valid on
> the vignette output product file, where we can only have file name
> extensions *.pdf and *.html.  Because of this, a solution is to use
> 'basename(vDB$PDF)' instead of 'basename(vDB$File)'.  This is also
> what tools:::httpd() uses internally and the difference between the
> two approach is what causes the error in help.search() above.
>
>
> PATCH:
>
> A patch (against R-devel) is:
>
> svn diff src/library/utils/R/help.search.R
> Index: src/library/utils/R/help.search.R
> ===================================================================
> --- src/library/utils/R/help.search.R (revision 73159)
> +++ src/library/utils/R/help.search.R (working copy)
> @@ -43,7 +43,7 @@
>   base[, "LibPath"] <- path
>   id <- as.character(1:nrow(vDB) + NROW(hDB[[1L]]))
>   base[, "ID"] <- id
> - base[, "Name"] <- sub("\\.[^.]*$", "", basename(vDB$File))
> + base[, "Name"] <- tools::file_path_sans_ext(basename(vDB$PDF))
>   base[, "Topic"] <- base[, "Name"]
>   base[, "Title"] <- vDB$Title
>   base[, "Type"] <- "vignette"
>
> I have verified that the above patch solves the problem.  I thought it
> is ok to use tools::file_path_sans_ext() since other functions in
> 'utils' already do so.
>
> I'm happy to submit this one via https://bugs.r-project.org/bugzilla3/ as well.
>
>
> APPENDIX:
>
>> sessionInfo()
> R version 3.4.1 (2017-06-30)
> Platform: x86_64-pc-linux-gnu (64-bit)
> Running under: Ubuntu 16.04.3 LTS
>
> Matrix products: default
> BLAS: /usr/lib/atlas-base/atlas/libblas.so.3.0
> LAPACK: /usr/lib/atlas-base/atlas/liblapack.so.3.0
>
> locale:
>  [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C
>  [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8
>  [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8
>  [7] LC_PAPER=en_US.UTF-8       LC_NAME=C
>  [9] LC_ADDRESS=C               LC_TELEPHONE=C
> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
>
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base
>
> loaded via a namespace (and not attached):
> [1] compiler_3.4.1
>
> /Henrik