Use R6 representation

[zkamvar]

I've added an R6 class object with four methods: new, write, reset, and add_md.

I've also added test fixtures and snapshot tests (which are super fun!) (I also may have been on the rOpenSci allcaps when I was writing some of the tests...)

this PR will fix #25, fix #27, and address #28

[maelle]

This is awesome, thank you! I left a few comments.

Another question that I have is how this should be documented in the README. What approach would we recommend first, R6 or not? In any case at the very top of the README where there's a list of use cases you could add "add an arbitrary new Markdown element like a table".

[zkamvar]

This is awesome, thank you! I left a few comments.

Another question that I have is how this should be documented in the README. What approach would we recommend first, R6 or not? In any case at the very top of the README where there's a list of use cases you could add "add an arbitrary new Markdown element like a table".

I think we can recommend the R6 approach first since it's arguably more straightforward given that we are defining the namespace prefix in the object. So the first block would turn into this:

# From Markdown to XML
path <- system.file("extdata", "example1.md", package = "tinkr")
y <- yarn$new(path)

library("magrittr")
# transform level 3 headers into level 1 headers
body <- y$body
ns <- y$ns
body %>%
  xml2::xml_find_all(xpath = './/md:heading[@level="3"]', ns) -> headers3

xml2::xml_set_attr(headers3, "level", 1)

y$body <- body

# Back to Markdown
y$write("newmd.md")
file.edit("newmd")

[zkamvar]

Still to do:

• README
• test coverage 🙈

[maelle]

Reg the example I think adding R6 is fine but we might lose some folks if using "->" for assignment 😉

I wonder whether we should add a short list of things that people might enjoy learning about for maximal enjoyment of tinkr

• XPath
• XSLT (not crucial if they are ok with the default one)
• R6 (although reading the docs of tinkr should help)

[zkamvar]

Reg the example I think adding R6 is fine but we might lose some folks if using "->" for assignment 😉

come to think of it, for this particular example, we don't even need to use any assignment:

y$body %>%
  xml2::xml_find_all(xpath = './/md:heading[@level="3"]', y$ns) %>%
  xml2::xml_set_attr("level", 1)

y$write("newmd.md")

I wonder whether we should add a short list of things that people might enjoy learning about for maximal enjoyment of tinkr

* XPath

* XSLT (not crucial if they are ok with the default one)

* R6 (although reading the docs of tinkr should help)

YES! Your original link to the XPath page really helped me learn, so I think these extras would also help. It may also be worthwhile thinking about writing up some documentation about how the XSLT stylesheet in {tinkr} works broadly (e.g. that it uses the commonmark namespace and imports the commonmark stylesheet).

I also realized that it's possible to manipulate a stylesheet with {xml2} so that users can change things like bullet point shape

[maelle]

I opened a few issues

[zkamvar]

This is very strange.... I'm able to get the tests to pass on my machine, even with a clean version of the git repo.

[zkamvar]

I think this can be merged that the basic elements of the R6 class have been addressed. I think any other modifications should be new PRs.

It's also worthwhile to note that this merge will not change any existing workflows.