extractCached: Extract cached objects

In LTLA/rebook: Re-using Content in Bioconductor Books

Extract cached objects

Description

Extract specific R objects from the knitr cache of a previously compiled Rmarkdown file (the “donor”) so that it can be used in the compilation process of another Rmarkdown file (the “acceptor”).

Usage

extractCached(path, chunk, objects, envir = parent.frame(1), link.text = NULL)

Arguments

path	String containing the path to the donor Rmarkdown file.
chunk	String containing the name of the requested chunk.
objects	Character vector containing variable names for one or more objects to be extracted.
envir	Environment where the loaded objects should be stored. Defaults to the environment in which this function is called.
link.text	String containing an Rmarkdown-formatted link to the donor file, to be inserted in the collapsible element's title. If NULL, we attempt to construct this automatically from path using rmd2id. If NA, no link text is inserted.

Details

Each R object is extracted in its state at the requested chunk and inserted into envir. Note that the object does not have to be generated or even referenced in the requested chunk, provided it was generated in a previous named chunk.

The parser in this function is rather limited, so the donor Rmarkdown file is subject to several constraints:

• All chunks involved in generating the requested objects (indirectly or otherwise) should be named.

• All named chunks should be executed; eval=FALSE is not respected.

• All relevant code occurs within triple backticks, i.e., any inline code should be read-only.

Unnamed chunks are allowed but cannot be referenced and will not be used for searching for objects. Chunks with names starting with unref- are considered to be the same as unnamed chunks and will be ignored; this is useful for figure-generating chunks that need to be referenced inside the donor report. In general, neither of these should be used for code that might affect variables in the named chunks, i.e., code in unnamed chunks should be “read-only” with respect to variables in the named chunks.

Obviously, this entire process assumes that donor report has already been compiled with cache=TRUE. If not, extractCached will compile it (and thus generate the cache) using compileChapter. A report-specific lock is applied during this process to avoid problems with concurrent compilation.

Value

Variables with names objects are created in envir. A markdown chunk (wrapped in a collapsible element) is printed that contains all commands needed to generate those objects, based on the code in the named chunks of the donor Rmarkdown file.

Author(s)

Aaron Lun

See Also

setupHTML and chapterPreamble, to set up the code for the collapsible element.

compileChapter, to compile a Rmarkdown report to generate the cache.

Examples

# Mocking up an Rmarkdown report.
donor <- tempfile(fileext=".Rmd")
write(file=donor, "```{r some-monsters}
destoroyah <- 1
mecha.king.ghidorah <- 2
```
                                                                
```{r more-monsters}
space.godzilla <- 3
```

```{r}
msg <- 'I am not referenced.'
```

```{r unref-figure}
plot(1, 1, main='I am also not referenced.')
```

```{r even-more-monsters}
megalon <- 4
```")

# Extracting stuff from it in another report.
acceptor <- tempfile(fileext=".Rmd")
dpb <- deparse(basename(donor))
write(file=acceptor, sprintf("```{r, echo=FALSE, results='asis'}
chapterPreamble()
```
                                                                
```{r, results='asis', echo=FALSE}
extractCached(%s, chunk='more-monsters', 
   objects=c('space.godzilla', 'destoroyah'))
```

```{r}
space.godzilla * destoroyah
```

```{r, results='asis', echo=FALSE}
extractCached(%s, chunk='even-more-monsters', 
   objects=c('megalon', 'mecha.king.ghidorah'))
```

```{r}
mecha.king.ghidorah * megalon
```", dpb, dpb))

rmarkdown::render(acceptor)
if (interactive()) browseURL(sub(".Rmd$", ".html", acceptor))

LTLA/rebook documentation built on June 5, 2023, 6:24 p.m.