fetch_bibtex

[ben-domingue]

Hi Ben! I spent some today to write the BibTex gen function in R. Perhaps we can share this with Mike and include this as part of the irw library. It works like follows:

fetch_bibtex(
  data_index_file = "IRW Data Dictionary - data index.csv",
  dataset_without_doi_file = "BibTex_Manual.csv",
  dataset_names = c("16_personalityfactors", "enem_2020_1mil_mt"),
  output_bib_file = "output.bib"
)

It reads the BibTex from the 2 files that I attached below, which are the same ones that will be used to render the docs on IRW website.

[ben-domingue]

library(dplyr)
library(readr)
library(glue)
library(httr)

fetch_bibtex <- function(data_index_file, dataset_without_doi_file, dataset_names, output_bib_file) {
  
  data_index <- read_csv(data_index_file, na = character()) %>%
    select(dataset = `Filename`, doi = DOI) %>%
    mutate(dataset = gsub(".R.ata$", "", dataset))
  
  dataset_without_doi <- read_csv(dataset_without_doi_file, na = character()) %>%
    select(dataset = Filename, BibTex)
  
  data_index <- data_index %>%
    left_join(dataset_without_doi, by = "dataset") %>%
    mutate(doi_or_bibtex = if_else(is.na(doi) | doi == "", BibTex, doi)) %>%
    select(dataset, doi_or_bibtex)
  
  filtered_data <- data_index %>%
    filter(dataset %in% dataset_names)
  bibtex_entries <- filtered_data %>%
    rowwise() %>%
    mutate(bibtex = {
      if (!is.na(doi_or_bibtex) && grepl("^@", doi_or_bibtex)) {
        # Directly use the BibTeX if it's already provided
        doi_or_bibtex
      } else if (!is.na(doi_or_bibtex)) {
        # Fetch BibTeX using DOI
        response <- tryCatch({
          GET(glue("https://doi.org/{doi_or_bibtex}"), add_headers(Accept = "application/x-bibtex"))
        }, error = function(e) NULL)
        
        if (!is.null(response) && status_code(response) == 200) {
          content(response, as = "text", encoding = "UTF-8")
        } else {
          glue("# Error fetching BibTeX for DOI: {doi_or_bibtex}")
        }
      } else {
        glue("# No BibTeX or DOI available for dataset: {dataset}")
      }
    }) %>%
    pull(bibtex)
  
  writeLines(bibtex_entries, output_bib_file)
  cat(glue("BibTeX entries written to {output_bib_file}\n"))
}

# Example
fetch_bibtex(
  data_index_file = "IRW Data Dictionary - data index.csv",
  dataset_without_doi_file = "BibTex_Manual.csv",
  dataset_names = c("16_personalityfactors", "enem_2020_1mil_mt"),
  output_bib_file = "output.bib"
)

[KingArthur0205]

Attaching the used CSV files:
IRW Data Dictionary - data index.csv
BibTex_Manual.csv