build

#!/usr/bin/env Rscript

# config #######################################################################
src = "src"
dest = "docs"


## urls for manual pages
# - rdocumentation
urlMlrFunctions = "http://www.rdocumentation.org/packages/mlr/functions/"
urlContribPackages = urlBasePackages = "http://www.rdocumentation.org/packages/"
docs = "functions/"
ext = ".html"

# - ianhowson (does not have base packages)
# urlMlrFunctions = "http://rpackages.ianhowson.com/cran/mlr/man/"
# urlContribPackages = "http://rpackages.ianhowson.com/cran/"
# urlBasePackages = "http://www.inside-r.org/r-doc/"
# docs = "man/"
# ext = ".html"
# baseR = c("base", "compiler", "datasets", "graphics", "grDevices", "grid", "methods", "parallel", "splines", "stats", "stats4","tcltk", "tools", "utils")

# - inside-r (does have contributed packages and base/recommended packages in different places)
# urlMlrFunctions = "http://www.inside-r.org/packages/cran/mlr/docs/"
# urlContribPackages = "http://www.inside-r.org/packages/cran/"
# urlBasePackages = "http://www.inside-r.org/r-doc/"
# docs = "docs/"
# ext = ""
# baseR = c("base", "boot", "class", "cluster", "codetools", "compiler", "datasets", "foreign", "graphics", "grDevices", "grid", "KernSmooth", "lattice", "MASS", "Matrix", "methods", "mgcv", "nlme", "nnet", "parallel", "rpart", "spatial", "splines", "stats", "stats4", "survival", "tcltk", "tools", "utils")


# macros to link to manual pages:
# - [NAME](&PACKAGE::FUNCTION)
# - [NAME](&FUNCTION) (mlr documentation)
# - [&PACKAGE::FUNCTION]
# - [&FUNCTION] (mlr documentation)
# - [NAME](%PACKAGE)
# - [%PACKAGE]

macros = list(
  list(pattern = "\\[(.+?)\\]\\(&([A-Za-z0-9.]+?)::([A-Za-z0-9.-_]+)\\)", replacement = paste("[\\1](", urlContribPackages, "\\2/", docs, "\\3", ext, ")", sep = "")),
  list(pattern = "\\[(.+?)\\]\\(&([A-Za-z0-9.-]+?)\\)", replacement = paste("[\\1](", urlMlrFunctions, "\\2", ext, ")", sep = "")),
  list(pattern = "\\[&([A-Za-z0-9.]+?)::([A-Za-z0-9.-_]+?)\\]", replacement = paste("[\\1::\\2](", urlContribPackages, "\\1/", docs, "\\2", ext, ")", sep = "")),
  list(pattern = "\\[&([A-Za-z0-9.-]+?)\\]", replacement = paste("[\\1](", urlMlrFunctions, "\\1", ext, ")", sep = "")),
  list(pattern = "\\[(.+?)\\]\\(%([A-Za-z0-9.-]+?)\\)", replacement = paste("[\\1](", urlContribPackages, "\\2/)", sep = "")),
  list(pattern = "\\[%([A-Za-z0-9.-]+?)\\]", replacement = paste("[\\1](", urlContribPackages, "\\1/)", sep = ""))
)

if (urlContribPackages != urlBasePackages) {
  # macros to link to docs:
  # - [NAME](&PACKAGE::FUNCTION)
  # - [&PACKAGE::FUNCTION]
  # - [NAME](%PACKAGE)
  # - [%PACKAGE]
  macrosRecom = c(
    lapply(baseR, function(p) list(pattern = paste("\\[(.+?)\\]\\(&(", p, ")::([A-Za-z0-9.-_]+)\\)", sep = ""), replacement = paste("[\\1](", urlBasePackages, "\\2/\\3)", sep = ""))),
    lapply(baseR, function(p) list(pattern = paste("\\[&(", p, ")::([A-Za-z0-9.-_]+?)\\]", sep = ""), replacement = paste("[\\1::\\2](", urlBasePackages, "\\1/\\2)", sep = ""))),
    lapply(baseR, function(p) list(pattern = paste("\\[(.+?)\\]\\(%(", p, ")\\)", sep = ""), replacement = paste("[\\1](", urlBasePackages, "\\2/)", sep = ""))),
    lapply(baseR, function(p) list(pattern = paste("\\[%(", p, ")\\]", sep = ""), replacement = paste("[\\1](", urlBasePackages, "\\1/)", sep = "")))
  )
  macros = c(macrosRecom, macros)
}


# helper functions #############################################################
# generate useful exit status
exit = function(status, msg, ...) {
  if (!missing(msg))
    cat(sprintf(msg, ...), "\n\n", sep = "")
  quit(save = "no", status = status)
}

# set seed depending on file name
setSeed = function(x, algo = "crc32") {
  hash = digest(x, algo = algo)
  set.seed(strtoi(sprintf("0x%s", substr(hash, 1L, 7L))))
}

# worker function
knitIt = function(f, dev) {
  messagef("Knitting file '%s' ...", f)
  lines = readLines(file.path(src, f))

  # apply replacement macros
  for (macro in macros)
    lines = str_replace_all(lines, macro$pattern, macro$replacement)

  # should we append the full source code at the end?
  # to opt out add this to the yaml front matter
  # params:
  #  full.code: FALSE
  knit.params = knit_params(lines)
  if (!isFALSE(knit.params$full.code$value) && dev != "pdf") {
    lines = c(
      lines, "\n", 
      "## Complete code listing", "\n",
      "The above code without the output is given below:", "\n",
      "```{r, purl=FALSE, echo = FALSE, comment=''}",
      paste0("cat(readLines('./",file.path("full_code_src", str_replace(f, "\\.Rmd$", "\\.R")),"'),fill = 2)"),
      "```")
  }

  cache.path = file.path("cache", f, "")
  opts_chunk$set(cache = TRUE, cache.path = cache.path, fig.path = file.path("figure", dev, str_replace(f, "\\.Rmd", ""), ""), dev = dev, error = FALSE, comment = "#>", collapse = TRUE)
  setSeed(f)

  # FIXME: remove if mkdocs handles this, currently mkdocs puts yaml front matter in the output.
  yaml.lines = which(lines == "---")
  if (length(yaml.lines == 2)) {
    lines = lines[-1 * seq(from = yaml.lines[1], to = yaml.lines[2])]
  }
  
  knit(
    text = lines,
    output = file.path(dev, dest, str_replace(f, "\\.Rmd$", "\\.md")),
    quiet = TRUE
  )
}

# helper function to extract the source code
purlIt = function(f, dest) {
  messagef("Purling file '%s' ...", f)
  setSeed(f)
  purl(
    input = file.path(src, f),
    output = file.path(dest, str_replace(f, "\\.Rmd$", "\\.R")),
    quiet = TRUE,
    documentation = 0
  )
}

# start #######################################################################
message("Loading required packages ...")
suppressPackageStartupMessages({
  # namespace conflicts
  library(glmnet)   # auc
  library(ROCR)

  # required to build
  library(methods)
  library(parallel)
  library(digest)
  library(pander)
  library(BBmisc)
  library(knitr)
  library(caret) # we get problems if this shadows "mlr::train"
  suppressWarnings(library(rgl)) # suppress "no X" warning on travis
  library(mlr)
  library(stringr)
  library(backports)
})

# turn warnings to errors, we don't want to miss them
options(warn = 2L)
# number of spaces to indent the code is 2 (for chunks with tidy=TRUE)
opts_chunk$set(tidy.opts = list(indent = 2, width.cutoff = 80))

# create output directories for pdf and svg devices
if (!isDirectory("svg") | !isDirectory(file.path("svg", "docs")))
  dir.create(file.path("svg", "docs"), recursive = TRUE)
if (!isDirectory("pdf") | !isDirectory(file.path("pdf", "docs")))
  dir.create(file.path("pdf", "docs"), recursive = TRUE)
if (!isDirectory("full_code_src"))
  dir.create(file.path("full_code_src"))

## Handle mlr Versions ####

# if we have a new mlr version move old devel files into version directory
mlr.version = as.character(packageVersion("mlr"))
link = Sys.readlink(mlr.version)
if (is.na(link)) {
  # find the directory that linked to devel before and copy devel content to it.
  old.version = names(which(sapply(list.files(".", pattern = "\\d\\..*", include.dirs = TRUE), Sys.readlink) == "devel"))
  file.remove(old.version)
  dir.create(old.version)
  system(sprintf("cp -r devel/* %s", old.version))
  # this old.version is probably the latest release so change this link to
  file.remove("release")
  file.symlink(old.version, "release")
  # No directory or link for the new version exists so we create one
  file.symlink("devel", mlr.version)
  # Take care of github on Travis
  if (identical(Sys.getenv("TRAVIS"), "true")) {
    system(sprintf("git add %s", paste(mlr.version, old.version, "devel", "release")))
  }
} else if (link == "") {
  # The directory exists but is not a link
  warning("You are building the tutorial with an outdated mlr version! Please use the github version!")
} else if (link != "devel") {
  stopf("The directory %s should link to devel and not to %s", mlr.version, link)
}

## Do the building ####

# do lapply for now. with mclapply it seems unclear how to stop best on error. this was 100% broken before
message("Building purl version...")
dummy = lapply(list.files(src, pattern = "\\.Rmd$"), purlIt, "full_code_src")
message("Building PDF version...")
dummy = lapply(list.files(src, pattern = "\\.Rmd$"), knitIt, "pdf")
ok = system3("./build-pdf.py")
if (ok$exit.code == 0L) {
  # copy pdf to devel
  ok = file.copy("mlr-tutorial-build.pdf", file.path("devel", "mlr-tutorial.pdf"), overwrite = TRUE)
  if (!ok) {
    stop("copy pdf failed.")
  }
  # embed images in HTML
  opts_knit$set(upload.fun = image_uri)
  message("Building HTML version...")
  dummy = lapply(list.files(src, pattern = "\\.Rmd$"), knitIt, "svg")
  ok = system3("mkdocs", "build")
  message("Build successful!")
  message("Now zip everything!")
  setwd("devel/html/")
  suppressAll(zip("../mlr_tutorial.zip", files = "."))
  setwd("../../")
  exit(0L)
}

messagef("Build failed!")
exit(ok$exit.code)

# vim: set ft=r: