Small script to build a bibtex file from a collection of PDF files

pdfs_to_bib.R

library("pdftools")

# load files
files <- dir(pattern = "pdf$")

# setup metadata
year <- ifelse(grepl("\\d{4}", files), regmatches(files, regexpr("(?<=, )\\d{4}", files, perl = TRUE)), "")
journal <- unlist(lapply(files, function(x) {
    if (grepl("(?<=\\().+(?=, \\d{4})", x, perl = TRUE)) {
        regmatches(x, regexpr("(?<=\\().*?(?=, \\d{4})", x, perl = TRUE))
    } else {
        ""
    }
}))
authors <- gsub("( \\(.+,? ?[[:digit:]]?\\))|(\\)?\\.pdf)", "", files)
key <- unlist(lapply(authors, function(x) {
    x <- strsplit(x, ", ")[[1]]
    if (length(x) > 3) {
        paste0(head(x, 1), "etal", collapse = "")
    } else {
        paste0(head(x, 3), collapse = "")
    }
}))

title <- lapply(files, function(x) {
    title <- pdftools::pdf_info(x)$keys$Title
    if (!is.null(title)) {
        return(title)
    } else {
        return("")
    }
})

# write to disk
cat(
paste0("@Article{", paste0(key, year), ",\n",
"author = {", gsub(", ", " and ", authors, fixed = TRUE), "},\n",
"title = {", title, "},\n",
"year = {", year, "},\n",
"journal = {", journal, "},\n",
"file = {:", files, ":PDF}\n",
"}", collapse = "\n\n"), "\n", 
file = "tmp.bib")