CODEWITHSUNDEEP
rarchitecturestructurediagram
Emil Bille
Emil Bille

Reputation: 11

Is there a way to create a code architecture diagram, that gives an overview over R scripts that source each other?

I have alot of different scripts in R that sources one another with source(). Im looking for a way to create an overview diagram, that links each script visually, so i can easily see the "source hierarchy" of my code.

The result could look something like:

enter image description here

I hope there is a solution, that doesnt require a software license.

Hope it makes sence! :)

Upvotes: 0

Views: 1091

Answers (2)

G. Grothendieck
G. Grothendieck

Reputation: 270075

For purposes of example change directory to an empty directory and run the code in the Note at the end to create some sample .R files.

In the first two lines of the code below we set the files variable to be a character vector containing the paths to the R files of interest. We also set st to the path to the main source file. Here it is a.R but it can be changed appropriately.

The code first inserts the line contained in variable insert at the beginning of each such file.

Then it instruments source using the trace command shown so that each time source is run a log record is produced. We then source the top level R file.

Finally we read in the log and use the igraph package to produce a tree of source files. (Any other package that can produce suitable graphics could be used instead.)

# change the next two lines of code appropriately.
# Settings shown are for the files generated in the Note at the end
# assuming they are in the current directory and no other R files are.
files <- Sys.glob("*.R")
st <- "a.R"

# inserts indicated line at top of each file unless already inserted
insert <- "this.file <- normalizePath(sys.frames()[[1]]$ofile)"
for(f in files) 
  inp <- readLines(f)
  ok <- !any(grepl(insert, inp, fixed = TRUE)) # TRUE if insert not in f
  if (ok) writeLines(c(insert, input), f)
}

# instrument source and run to produce log file
if (file.exists("log")) file.remove("log")
this.file <- "root"
trace(source, quote(cat("parent:", basename(this.file), 
  "file:", file, "\n", file = "log", append = TRUE)))
source(st)  # assuming a.R is the top level program
untrace(source)

# read log and display graph
DF <- read.table("log")[c(2, 4)]

library(igraph)
g <- graph.data.frame(DF)
plot(g, layout = layout_as_tree(g))

For example, if we have the files generated in the Note at the end then the code above generates this diagram:

screenshot

Note

cat('
source("b.R")
source("c.R")
', file = "a.R")

cat("\n", file = "b.R")

cat("\n", file = "C.R")

Upvotes: 0

Licantropo
Licantropo

Reputation: 11

I can suggest you use Knime. it has the kind of diagram you are looking for. It has some scripts already wrote to clean, visualize data and write output and has integration with R and Python.

https://docs.knime.com/?category=integrations&release=2019-12 https://www.knime.com/

Good luck.

Upvotes: 1

Related Questions