Recover R source from a knitted R Markdown / Quarto HTML file

Extracts every input-code chunk pandoc/knitr rendered into the document (<pre class="r"><code>...</code></pre> or <div class="sourceCode r">), in document order, and writes them — separated by a blank line, matching how code_read() already treats whitespace between statements — as a sibling .R file. This is the HTML analogue of .smcl_export_syntax() / .spv_export_syntax(): the ORIGINAL analysis code sits verbatim inside the rendered document (HTML-entity-escaped only, never altered), so it is exactly recoverable, not reconstructed or guessed.

Usage

.html_export_r_source(html_path, code_dir_name = "code")

Arguments

html_path

path to a local .html file already confirmed (by .html_sniff_kind()) to be "rmd"

code_dir_name

the sibling folder to write into, same convention as .spv_export_syntax() / .smcl_export_syntax() (default "code"). A file of the user's own gets this folder in the temporary folder instead, so the user's folder is not changed (see .recovered_code_dir()).

Value

the path to the written .R file, or NA_character_ if the document could not be read or contained no recoverable R chunk

Details

Printed console OUTPUT (## -prefixed lines pandoc/knitr interleave after each chunk) is deliberately NOT included: this produces a re-runnable .R script for code_check/reproducibility_check (the modules in inst/modules/), and a ##-prefixed line is a comment in R syntax anyway, so leaving it out changes nothing a parse would see — it is dropped here only to keep the recovered file to actual code.