Building a website with blogdown
Having seen a few nice looking websites which use the Hugo academic theme, I thought I’d have a go at revamping my website. I’m not going to write a whole new post about how to do this - there are plenty of these online. I just thought I’d point out some resources that I found useful.
I mainly worked from Alistair Bailey’s really detailed guide to
building a website using blogdown in R. This was pretty comprehensive and I had a site I was happy with up and running very quickly. Two key things I needed from elsewhere were (1) how to upload the site to my existing GitHub pages address; and (2) how to get the cite buttons on publications that I’d seen on other sites.
Linking to GitHub pages
The main issue I had with building the site was linking it to my existing GitHub pages site. The guide I followed provided instructions on linking to Netlify. For anyone also wanting to link to GitHub pages, Tyler Clavelle explains how to do so with a two repository approach.
Cite button
To get the cite button next to the publication (along with PDF Code etc. buttons), there needs to be a separate .bib file for each publication in the static/files/citations directory. This file needs to have the same name as the .md file for the publication. I have modified the code by
Lorenzo Busetto (which creates the required .md files for each publication) to also output .bib files to static/files/citations:
bibtex_2academic <- function(bibfile,
abstract = TRUE,
overwrite = FALSE) {
require(RefManageR)
require(dplyr)
require(stringr)
require(anytime)
# Import the bibtex file and convert to data.frame
mypubs <-
ReadBib(bibfile, check = "warn",
.Encoding = "UTF-8") %>%
as.data.frame()
# assign "categories" to the different types of
# publications
mypubs <- mypubs %>%
dplyr::mutate(
pubtype = dplyr::case_when(
bibtype == "Article" ~ "2",
bibtype == "Article in Press" ~ "2",
bibtype == "InProceedings" ~ "1",
bibtype == "Proceedings" ~ "1",
bibtype == "Conference" ~ "1",
bibtype == "Conference Paper" ~ "1",
bibtype == "MastersThesis" ~ "3",
bibtype == "PhdThesis" ~ "3",
bibtype == "Manual" ~ "4",
bibtype == "TechReport" ~ "4",
bibtype == "Book" ~ "5",
bibtype == "InCollection" ~ "6",
bibtype == "InBook" ~ "6",
bibtype == "Misc" ~ "0",
TRUE ~ "0"
)
)
# function to create the filename
create_filename <- function(x, ext) {
filename <- paste(
x[["year"]],
x[["title"]] %>%
str_replace_all(fixed(" "), "_") %>%
str_remove_all(fixed(":")) %>%
str_sub(1, 20) %>%
paste0(ext),
sep = "_"
)
}
# create a function which populates the md template
# based on the info
# about a publication
create_md <- function(x) {
# define a date and create filename by appending date
# and start of title
if (!is.na(x[["year"]])) {
x[["date"]] <- paste0(x[["year"]], "-01-01")
} else {
x[["date"]] <- "2999-01-01"
}
filename <- create_filename(x, ".md")
# start writing
if (!file.exists(file.path("content/publication", filename)) |
overwrite) {
fileConn <- file.path("content/publication", filename)
write("+++", fileConn)
# Title and date
write(paste0("title = \"", x[["title"]], "\""),
fileConn,
append = T)
write(paste0("date = \"", anydate(x[["date"]]), "\""),
fileConn,
append = T)
# Authors. Comma separated list, e.g. `["Bob Smith",
# "David Jones"]`.
auth_hugo <-
str_replace_all(x["author"], " and ", "\", \"")
auth_hugo <-
stringi::stri_trans_general(auth_hugo, "latin-ascii")
write(paste0("authors = [\"", auth_hugo, "\"]"),
fileConn,
append = T)
# Publication type. Legend:
# 0 = Uncategorized, 1 = Conference paper,
# 2 = Journal article
# 3 = Manuscript, 4 = Report, 5 = Book, 6 = Book
# section
write(paste0("publication_types = [\"", x[["pubtype"]],
"\"]"),
fileConn,
append = T)
# Publication details: journal, volume, issue,
# page numbers and doi link
publication <- x[["journal"]]
if (!is.na(x[["volume"]]))
publication <- paste0(publication,
", (", x[["volume"]], ")")
if (!is.na(x[["pages"]]))
publication <- paste0(publication,
", _pp. ", x[["pages"]], "_")
if (!is.na(x[["doi"]]))
publication <- paste0(publication,
", ",
paste0("https://doi.org/",
x[["doi"]]))
write(paste0("publication = \"", publication, "\""),
fileConn,
append = T)
write(paste0("publication_short = \"",
publication,
"\""),
fileConn,
append = T)
# Abstract and optional shortened version.
if (abstract) {
write(paste0("abstract = \"", x[["abstract"]], "\""),
fileConn,
append = T)
} else {
write("abstract = \"\"",
fileConn,
append = T)
}
write(paste0("abstract_short = \"", "\""),
fileConn,
append = T)
# other possible fields are kept empty. They can be
# customized later by
# editing the created md
write("image_preview = \"\"",
fileConn,
append = T)
write("selected = false", fileConn, append = T)
write("projects = []", fileConn, append = T)
write("tags = []", fileConn, append = T)
#links
write("url_pdf = \"\"", fileConn, append = T)
write("url_preprint = \"\"",
fileConn,
append = T)
write("url_code = \"\"", fileConn, append = T)
write("url_dataset = \"\"",
fileConn,
append = T)
write("url_project = \"\"",
fileConn,
append = T)
write("url_slides = \"\"", fileConn, append = T)
write("url_video = \"\"", fileConn, append = T)
write("url_poster = \"\"", fileConn, append = T)
write("url_source = \"\"", fileConn, append = T)
#other stuff
write("math = true", fileConn, append = T)
write("highlight = true", fileConn, append = T)
# Featured image
write("[header]", fileConn, append = T)
write("image = \"\"", fileConn, append = T)
write("caption = \"\"", fileConn, append = T)
write("+++", fileConn, append = T)
}
}
# apply the "create_md" function over the
# publications list to generate
# the different "md" files.
apply(
mypubs,
FUN = function(x)
create_md(x),
MARGIN = 1
)
# Added section to output .bib files to get the cite button
lapply(
1:nrow(mypubs),
FUN = function(x) {
# define a date and create filename by appending date
# and start of title
filename = create_filename(mypubs[x,], ".bib")
bib <- as.BibEntry(mypubs[x,])
WriteBib(bib, paste0("static/files/citations/", filename))
}
)
}
Final thoughts and extra resources
To make tweaks to my site, I mostly just found it useful to have a bit of a look at other sites that have been built with blogdown. Often the source code is discoverable (for example for a <username>.github.io site, the code will be at github.com/<username>/<username>.github.io). There are lots of resources out there about using blogdown to build sites. Here are some examples:
- The blogdown package handbook
- Up and running with blogdown, by Alison Presmanes Hill: This guide also provides links to resources on using Git, RStudio and Terminal.
- Mikey Harper explains the why of migrating to blogdown
- …and many more out there.
Happy website building!
