Whisker seems to fail when there is a period in the name of the variable. Considering that R variables quite typically have periods in the name, this is problematic.

> whisker.render("{{user.name}}", list(user.name=342))
[1] ""
> whisker.render("{{username}}", list(username=342))
[1] "342"

If there is a way around this please let me know. I imagine it's a quick fix. Also if this is an intentional design choice that'd be good to know too. Thanks!

Do you happen to have a simple example to illustrate how to use rowSplit()? It would be a big help. Dank u vriendelijk!

Hi @edwindj. As per our chat, I'm happy to take the lead on getting a release ready. It would be nice to do a few housekeeping tasks during this process, e.g. tag releases, update README & badges, etc.. To that end, maybe you could add me as a collaborator with write or admin permissions. That way I could do things, cc'ing or tagging you for awareness, but not requiring you to be involved in each step.

I was using the pattern sometimes used in the tidyverse where you name your variables something short with a period at the beginning like .v1 or .v2. This doesn't work in whisker for some reason. The variable just gets replaced by an empty string.

See your example (slightly altered) below. When I change name to .name it no longer renders:

template <- 
  'Hello {{.name}}
You have just won ${{value}}!
{{#in_ca}}
Well, ${{taxed_value}}, after taxes.
{{/in_ca}}
'

data <- list( .name = "Chris"
              , value = 10000
              , taxed_value = 10000 - (10000 * 0.4)
              , in_ca = TRUE
)

text <- whisker.render(template, data)
cat(text)
## Hello 
## You have just won $10000!
## Well, $6000, after taxes.

Dear @edwindj
I posted a question on the devtools github when I ran into a problem there and @wch pointed out that there was something funny going on with the whisker package and the devtools package. My original question is posted on Stackoverflow. Do you have any insights to share?

It would be nice to have a function that generates a data frame of keys for a given template for programmatic generation of input data. I'm thinking something along the lines of:

template <- 
    "{{#dat}}
    count: {{count}}, spray: {{spray}}\n
    {{/dat}}" 
whisker::getKeys(template)
#>   rawkey first last   key type
#> 1   #dat     1    8   dat    #
#> 2  count    20   28 count     
#> 3  spray    38   46 spray     
#> 4   /dat    49   56   dat    /

FWIW, the functionality in {whisker} already exists, it just needs to be exposed:

template <- 
    "{{#dat}}
    count: {{count}}, spray: {{spray}}\n
    {{/dat}}" 

f <- whisker:::parseTemplate(template)
get("key", envir = environment(f))
#>   rawkey first last   key type
#> 1   #dat     1    8   dat    #
#> 2  count    20   28 count     
#> 3  spray    38   46 spray     
#> 4   /dat    49   56   dat    /

^{Created on 2020-09-10 by the reprex package (v0.3.0)}

Total newbie here (R and whisker).
A good addition to the homepage would be an example of taking a template that is a file, and creating an output html file.
Thanks for considering

Thanks for a great package. R really needs a good templating engine.

If the names of hash$people are unknown, there appears to be now way of accessing the data in it's children.

library(whisker)

hash <- list(
  people = list(
    a = list(
      first = "john",
      last = "johnsson"
    ),
    b = list(
      first = "peter",
      last = "petersson"
    ),
    c = list(
      first = "steve",
      last = "stevesson"
    )
  ),
  stuff = c(1:5)
)

template <- "
{{#people}}
<h1>Hello {{first}}</h1>
{{/people}}
"

result <- whisker::whisker.render(template = template, data = hash)
print(HTML(result))

Link to stackoverflow.

If I have the following data:

data <- list(users = list(list(name = "bob"), list(name = "alice")))

I might want to know if the list is empty or not. For example, this would be my desired template:

template <- "{{#users.length}}Users: {{#users}}{{name}} {{/users}}{{/users.length}}{{^users.length}}No users{{/users.length}}"

It uses .length as a conditional that is true if there are any elements and false if it's empty.

This is not an official standard mustache feature, but it is implemented in some languages. See here for details.

How can I get that work?

R> dd = list(a = '>> blim')
R> tt = "{{a}} blam blum"
R> whisker.render(tt,dd)
### [1] ">> blim blam blum"

I want

">> blim blam blum"

Thanks, Michel

I'm hoping to use whisker for a markdown file where treatment of spaces and newlines is critical for, for example, specifying new paragraphs. I think the treatment of newlines may be inconsistent when using whisker.

For example, consider this template and calls to whisker.render:

template <- 
"{{#var}}
var is true
{{/var}}
{{^var}}
var is not true
{{/var}}"
whisker.render(template, list(var=TRUE))
whisker.render(template, list(var=FALSE))

I would expect the output to be either "var is true\n" or "var is not true\n" depending on the value of var. However, in the second case, the output has an extra newline at the start: "\nvar is not true\n".

There is an inconsistency in the whitespace generated by these two templates:

str1 <- 'header
{{#query}}
holla
{{/query}}
footer'
str2 <- 'header
{{#obj.query}}
holla
{{/obj.query}}
footer'

(the only difference is that the section variable is within a list obj in the second version).

This version gives what I would expect:

library(whisker)
whisker.render(str1, list(query=TRUE))  # "header\nholla\nfooter"
whisker.render(str1, list(query=FALSE)) # "header\nfooter"

While the second version generates extra newlines:

library(whisker)
whisker.render(str2, list(obj=list(query=TRUE)))  # "header\n\nholla\n\nfooter"
whisker.render(str2, list(obj=list(query=FALSE))) # "header\n\nfooter"

This does not affect the version from CRAN (0.3-2).

Hey,

could you consider pushing 0.3-3 to CRAN?

It would be so nice to have strict = FALSE available by default without any editing in johnmyleswhite/ProjectTemplate. See sql.reader.R#L336.

Pretty please? ;)

Best regards,
crubb

I think it would be great to have the ability to turn on an option which checked to see if parameters existed, and printed a warning / died if missing, e.g. something like

 template <- "Hello {{place}}!"
 whisker.render(template, checkvars=TRUE)

produces a warning, e.g.

  Warning: variable "place" undefined in ....

Due to the possibility of nested tags, this is very hard to check externally, so something like this is hard to verify without parsing the structure of the mustache file and comparing to the structure of the R environment:

 {{#repo}}
    <b>{{name}}</b>
 {{/repo}}

This is an issue as minor typos can lead to missing variables and hard to detect errors in rendered templates.

See also discussion here: http://stackoverflow.com/questions/16595621/error-safe-templating-with-brew-whisker

I did not know about the usage of triple mustaches for unescaping. I struggled with it a bit until I found:
http://stackoverflow.com/questions/20323351/how-to-disable-html-characters-escaping-in-whisker-render-in-r

Which answered my question. It would be nice if this was more prominent in the documentation such as that for the package or in whisker.render

Templates don' t handle names containing .

a.test <- 'World'
template <- template <- "Hello {{a.test}}!"
whisker.render(template)
## Error in value[[key]] : subscript out of bounds
whisker.render(template, list(a.test = a.test))
## [1] "Hello !"

Hello,

Thank you for your work on this package!

I noticed there are a few references to mustache.github.com in the README and the About section, but it appears Github would prefer them to be mustache.github.io

❯ whisker.render(c("xxx", "foobar"), list())
[1] "xxx\nfoobar"

❯ whisker.render(c("", "foobar"), list())
[1] ""

I suppose I can remove the "" and add one or more \ns to the beginning of the first non-empty string in the character vector as a workaround. But still, this is a bug imo, unless I am missing something.

Are there any plans to implement lambdas?

I've been trying to get whisker working with LaTeX (which uses a lot of braces)

At first I thought it wouldn't be an issue since there's proper nesting:

> template <- 
+ 
+ 'Hello {{{{name}}}}
+ You have just won ${{value}}!
+ 
+ '
> text <- whisker.render(template, data)
> cat(text)
Hello 
You have just won $10000!

So I thought, fine, I'll just switch the delimeters:

> template <- 
+ '
+ {{=<% %>=}}
+ Hello {{<%name%>}}
+ You have just won ${{<%value%>}}!
+ 
+ '
> text <- whisker.render(template, data)
> cat(text)

Hello ~~~~~~~~~~~~~Chris{{
You have just won $}}10000~~~~~~~~~~~~~!

Which is a bit weird? Apologies if I'm just doing something incorrectly.

I am new to whisker but my assumption with a template is that its main purpose is to be used when iterating over a vector or list. (If you just want output of length 1 you'd just write the output directly).

But I can't work out how to feed a set of lists (eg a tibble) as the environment for whisker - the documentation isn't clear enough for me. Here's a reprex of what I'm trying. I've used purrr::pmap to show what I want, and then tried to achieve the equivalent with whisker.

suppressPackageStartupMessages(library(dplyr))
suppressPackageStartupMessages(library(stringr))
suppressPackageStartupMessages(library(purrr))
suppressPackageStartupMessages(library(whisker))

df <- tibble(
  name = c("Adam", "Barbara", "Chris"),
  home = c("Aylesbury", "Blackburn", "Coxwold"),
  url = c("adamsapples.co.uk", "barbsberries.co.uk", "coxwoldcherries.co.uk")
)

# with purrr and stringr
purrr::pmap_chr(df, ~ stringr::str_glue(
  "<li><a href=\"{..3}\">{..1} from {..2}</a></li>")) %>% 
  str_c(collapse = "\n") %>% 
  str_c("<ul>\n", ., "\n</ul>")
#> [1] "<ul>\n<li><a href=\"adamsapples.co.uk\">Adam from Aylesbury</a></li>\n<li><a href=\"barbsberries.co.uk\">Barbara from Blackburn</a></li>\n<li><a href=\"coxwoldcherries.co.uk\">Chris from Coxwold</a></li>\n</ul>"
# writeLines(output, "output.html")

# with whisker
templ <- "<li>
<a href=\"{{url}}\">{{name}} from {{home}}</a>
</li>"

# expected this to work but doesn't
whisker::whisker.render(templ, df)
#> [1] "<li>\n<a href=\"adamsapples.co.uk,barbsberries.co.uk,coxwoldcherries.co.uk\">Adam,Barbara,Chris from Aylesbury,Blackburn,Coxwold</a>\n</li>"

# trying with pmap instead - gives error
purrr::pmap_chr(df, ~ whisker::whisker.render(templ, .))
#> Error in value[[key]]: subscript out of bounds

^{Created on 2020-03-20 by the reprex package (v0.3.0)}

I wanted to bring this to your attention. We recently submitted a new release of a package, rslurm, to CRAN that has a strong dependency of whisker (>=0.3) (in the Imports field of the DESCRIPTION file). Unfortunately our package fails CRAN checks for the experimental r-devel build on Windows (r-devel-windows-x86_64-gcc10-UCRT). It looks like this is because whisker depends on markdown which isn't available on that version of R: https://www.r-project.org/nosvn/R.check/r-devel-windows-x86_64-gcc10-UCRT/whisker-00check.html

Anyway, we really appreciate the work you put into this package and it would be great if we could get this resolved so that we can continue depending on whisker in the future. Please let me know if there is anything I can do.

Excellent work with Whisker. I use it with all my R packages. I was wondering if you had any plans to push the latest version of whisker to CRAN. Especially, the fix for empty templates is critical to the functionality of several R packages that I am authoring.

I am also curious about implementation of lambdas in Whisker. If you have already done some work, that would be useful, and I can take a crack at more implementation.

While whisker supports sections, it does not appear to support sections with non-empty lists as described in https://mustache.github.io/mustache.5.html:

Template:

{{#repo}}
  <b>{{name}}</b>
{{/repo}}

Hash:

{
  "repo": [
    { "name": "resque" },
    { "name": "hub" },
    { "name": "rip" }
  ]
}

Output:

<b>resque</b>
<b>hub</b>
<b>rip</b>

Using JSON

This closely follows the mustache documentation by usingjsonlite to convert the hash to an R object. The R object is a list with a data frame of 3 rows.

library(jsonlite)
library(whisker)

tmpl <- 
  "{{#repo}}
  <b>{{name}}</b>
  {{/repo}}"

hash <- 
  '{
    "repo": [
      { "name": "resque" },
      { "name": "hub" },
      { "name": "rip" }
    ]
  }'

data <- fromJSON(hash)  # list of data.frame not list of list


whisker.render(tmpl,data)   

This does not conform to the specification is not correct as in collapses the name column rather than returning three results. According to the specification, I would expect this result:

c("<b>resque</b>","<b>hub</b>","<b>rip</b>")

Using list-of-lists

Using a list of lists does not work either:

data <- list( repo = list( name="resque", name="hub", name="rip") )
whisker.render(tmpl,data) 
# [1] "<b>resque</b>"

I'm exploring how to set up a nested loop. Here's my template and call to whisker.render:

temp_string <- "
{{#i}}
# {{.}}
{{#j}}
## {{.}}
key = paste({{../}}, {{.}})
{{/j}}
{{/i}}"

template_contents <- strsplit(whisker::whisker.render(temp_string,
                                                      data = list(j = 1:2,
                                                                  i = letters[1:3])), "\n")[[1]]

Which results in:

# a
## 1
key = paste(, 1)
## 2
key = paste(, 2)
# b
## 1
key = paste(, 1)
## 2
key = paste(, 2)
# c
## 1
key = paste(, 1)
## 2
key = paste(, 2)

ALMOST what I need! Is there any way to get the {{../}} to pull the "parent" i?

I would like to use mustache templates with latex. But the trouble is that latex makes extensive use of { and } which interferes with mustache syntax. I read in the mustache documentation that it is possible to customize delimiters. Does whisker already support it, or can it be made to support it?

Hi.

I am just wondering if there are plans to continue the development of whisker? It is a handy package!

I would be happy to help push version 0.3.3 to CRAN, for instance.

Thank you.

Anders.