Add encoding support.

R/config.R

@@ -25,7 +25,24 @@ slidifyDefaults <- function(){list(
 get_config <- function(cfile = 'config.yml'){
   config = slidifyDefaults()
   if (file.exists(cfile)){
-    config = modifyList(config, yaml::yaml.load_file(cfile))
+    # @kohske
+    # yaml only accepts UTF8 (probably)
+    # so here,
+    # 1. read config in .input.enc
+    # 2. convert it into UTF8
+    # 3. load yaml
+    # 4. back the resutls into native.enc
+    txt = read_file(cfile)
+    txt = enc2utf8(txt)
+    config = modifyList(config, yaml.load(txt))
+    config = rapply(config, function(x) {
+      if (is.character(x)) {
+        Encoding(x) <- "UTF-8"
+        enc2native(x)
+      } else {
+        x
+      }
+    }, how = "replace")
   }
   return(config)
 }

R/parse.R

@@ -14,14 +14,16 @@ parse_page <- function(postFile, knit_deck = TRUE, envir){
     opts_chunk$set(fig.path = "assets/fig/", cache.path = '.cache/', cache = TRUE)
     outputFile <- gsub(".[r|R]md", ".md", inputFile)
     deckFile <- ifelse(knit_deck, 
-      knit(inputFile, outputFile, envir = envir), inputFile)
+      knit(inputFile, outputFile, envir = envir, encoding = .input.enc), inputFile)
     post <- deckFile %|% parse_deck
     post$file = postFile
     post$filename = tools:::file_path_sans_ext(inputFile)
     if (!is.null(post$date)) {
       post$date = as.Date(post$date, '%Y-%m-%d')
     }
-    post$link = gsub("*.Rmd", ".html", post$file)
+    # @kohske
+    # shouldn't be like this?
+    post$link = gsub("\\.Rmd$", ".html", post$file, fixed = TRUE)
     post$raw = read_file(inputFile)
     # saveRDS(post, file = "_payload.rds")
   })

R/process.R

@@ -5,9 +5,35 @@
 #' @keywords internal
 #' @noRd
 to_deck <- function(doc){
+  # @kohske
+  # read_file cares encoding and return native.enc char.
+  # txt should also be native.enc.
   txt = str_split_fixed(read_file(doc), '\n---', 2)
-  meta = yaml.load(gsub("^---\n+", '', txt[1]))
+
+  # @kohske
+  # Here txt is native.enc. Probably yaml.load accepts only UTF8 (and ascii?).
+  # In MBCS locale, gsub/sub sometimes returns char marked as UTF8
+  # (when fixed = FALSE), but sometimes not (if there is no match).
+  # So just in case, convert it into utf8 before yaml.load.
+  meta = yaml.load(enc2utf8(gsub("^---\n+", '', txt[1])))
+  # Then mark meta as UTF8 (because yaml.load doesn't this)
+  # and convert it to native.enc.
+  meta = rapply(meta, function(x) {
+    if (is.character(x)) {
+      Encoding(x) <- "UTF-8"
+      enc2native(x)
+    } else {
+      x
+    }
+  }, how = "replace")
+
+  # custom config also care encoding
+  # Note that if custom config is MBCS, it MUST be same encoding
+  # as input Rmd.
   cfile = ifelse(is.null(meta$config), 'config.yml', meta$config)
+
+  # Now all texts are native.enc
+
   deck = modifyList(get_config(cfile), c(meta, slides = txt[2]))
   deck$standalone = ifelse(deck$mode == "standalone", TRUE, FALSE)
   return(deck)

R/render.R

@@ -70,15 +70,36 @@ render_page <- function(page, payload, return_page = FALSE, save_payload = FALSE
     if (save_payload){
       save(layout, payload, partials, file = "payload.RData")
     }
-    cat(whisker.render(layout, payload, partials = partials), file = outputFile)
-
+
+    # @kohske
+    # output file should be UTF-8 definietely (right?),
+    # because almost all libraries put meta tag indicating utf-8 charaset in HTML.
+    # So here the encoding is hard-coded.
+    # In future, it may be better to accept output encoding.
+    outp = whisker.render(layout, payload, partials = partials) # native.enc
+
     # create standalone deck if page mode is standalone
     if (page$mode == 'standalone'){
-      outputFile = make_standalone(page, outputFile)
+      outp = make_standalone(page, outp)
     }
+
+    con <- file(outputFile, "w", encoding = "UTF8")
+    cat(outp, file = outputFile)
+    writeLines(outp, con)
+    close(con)
+
+    # @kohske
+    # why not make standalone before the file?
+    # I changed these order, but if there is reason
+    # please check it.
+    # create standalone deck if page mode is standalone
+
+    #if (page$mode == 'standalone'){
+    #  outputFile = make_standalone(page, outputFile)
+    #}
 
     # Extract R Code from Page if purl = TRUE
-    if (page$purl %?=% TRUE) purl(page$file)
+    if (page$purl %?=% TRUE) purl(page$file, encoding = .input.enc)
   })
   if (return_page){ return(page) }
 }

R/slidify.R

@@ -4,19 +4,39 @@
 #' @param knit_deck whether the file needs to be run through knit
 #' @param return_page should the function return the payload
 #' @param save_payload should the payload be saved to the slide directory
+#' @param encoding the encoding of the input file; see \code{\link{file}}
 slidify <- pagify <- function(inputFile, knit_deck = TRUE, 
-  return_page = FALSE, save_payload = FALSE, envir = parent.frame()){
-
+  return_page = FALSE, save_payload = FALSE, envir = parent.frame(),
+  encoding = getOption('encoding')){
+
+  # @kohske
+  # To make changes as small as possible,
+  # I use here a global variable, instead of
+  # passing function params.
+  .input.enc <<- encoding
+
   ## REMOVE LINES AFTER KNITR IS UPDATED ------
   options('knitr.in.progress' = TRUE)
   on.exit(options('knitr.in.progress' = FALSE))
   ## -------------------------------------------
 
   .SLIDIFY_ENV <<- new.env()
+
+  # @kohske
+  # I have no idea what 'site.yml' is. It is possible to be MultiByte char or only ascii?
+  # If there is possibly MB char, we need to fix this.
   site = ifelse(file.exists('site.yml'), yaml.load_file('site.yml'), list())
+
+  # @kohse
+  # there are changes inside parse_page to care encoding
   page = parse_page(inputFile, knit_deck, envir = envir)
-
+
+  # @kohske
+  # What's this?
   page = modifyList(page, as.list(.SLIDIFY_ENV))
+
+  # @kohske
+  # render_page is changes so that output is always UTF8
   render_page(page, payload = list(site = site), return_page, save_payload)
 }
 
@@ -80,3 +100,9 @@ check_slidifyLibraries <- function(){
   }
   return(invisible())
 }
+
+#' Encoding of input file
+#'
+#' @keywords internal
+#' @noRd
+.input.enc <- NULL

R/utils.R

@@ -149,8 +149,20 @@ minify_css <- function(css_file){
 #' @return string with document contents
 #' @keywords internal
 #' @noRd
-read_file <- function(doc, ...){
-  paste(readLines(doc, ...), collapse = '\n')
+read_file <- function(doc, encoding = NULL, ...){
+  # @kohske
+  # read all files in specified encoding, and then convert it
+  # into native.enc.
+  # So, if input Rmd/md is CP932, all libraries etc needs to be
+  # written with CP932. Maybe we need more fine control,
+  # but useally libraries are written in ascii,
+  # so there is unlikely a problem.
+  if (is.null(encoding)) encoding = .input.enc
+  con <- file(doc, "r", encoding = encoding)
+  text <- paste(readLines(con, ...), collapse = '\n')
+  text <- enc2native(text) # this may be unnesessary...(?)
+  close(con)
+  return(text)
 }
 
 #' Capture patterns matched by regular expression
@@ -240,23 +252,33 @@ mgsub <- function(myrepl, mystring){
   Reduce(gsub_, myrepl, init = mystring, right = T) 
 }
 
+# @kohske
+# I changed this function so that it doesn't read and write file.
+# see also render_page()
+
 #' Create a standalone version of an HTML File
 #' 
 #' It works by embedding all images, switching links to use Slidify's googlecode
 #' repository and inlining all user assets.
 #' 
 #' @param deck parsed deck
-#' @param html_in html file with library files linked locally
+#' @param html output html text (native.enc)
 #' @noRd
 #' @keywords internal
-make_standalone <- function(deck, html_in){
+make_standalone <- function(deck, html){
   lib_cdn = paste0(deck$lib_cdn %||% 'http://slidifylibraries2.googlecode.com/git/inst/libraries', '/')
   lib_url = paste0(deck$url$lib, '/')
-  html = read_file(html_in, warn = FALSE) %|% markdown:::.b64EncodeImages
-  html = gsub(lib_url, lib_cdn, html)
+
+  html = html %|% markdown:::.b64EncodeImages
+
+  # @kohske
+  # shouldn't be fixed=TRUE?
+  html = gsub(lib_url, lib_cdn, html, fixed = TRUE)
   # html_out = sprintf('%s.html', basename(getwd()))
-  cat(html, file = html_in)
-  return(html_in)
+
+  # @kohske
+  # need not write a file, but return text
+  return(html)
 }