Emacs Lisp Functions for Easy BibTeX Recording

2022-07-09 • 4 min read

Here’s an Emacs Lisp function for getting a BibTeX entry from a URL (like those URLs you get when clicking “cite” and copying the “BibTeX” link on a Google Scholar search result):

(defun get-bibtex-from-url (url)
  "Get a BibTeX entry from URL and store it in my `references.bib'."
  (interactive "MBibTeX URL: ")
  (write-bibtex-entry (get-bibtex-url-contents url)))

It’s very simple: interactive allows the user to enter a URL, get-bibtex-url-contents fetches the contents of that URL and write-bibtex-entry appends the contents to my references.bib file.

Now a somewhat less simple function. This one takes a Digital Object Identifier (DOI), fetches the corresponding BibTeX entry from the Crossref API and, again, appends the entry to references.bib. (I adapted this from Pavel Iosad’s original.) The returned entry contains the DOI (as a full URL) with an ASCII-encoded slash – the %2F there – so the function also replaces that with a decoded slash.

(defun get-bibtex-from-doi (doi)
  "Get a BibTeX entry from the DOI and store it in my `references.bib'."
  (interactive "MDOI: ")
  (let* ((url-mime-accept-string "text/bibliography;style=bibtex")
         (base-url
          "https://api.crossref.org/works/%s/transform/application/x-bibtex")
         (doi-resource (s-replace "http://dx.doi.org/" "" doi))
         (url (format base-url doi-resource))
         (response-body (get-bibtex-url-contents url))
         (bibtex-entry (s-replace "%2F" "/" response-body)))
    (write-bibtex-entry bibtex-entry)))

Okay, how do get-bibtex-url-contents and write-bibtex-entry work? Let’s look at the first one first. The with-temp-buffer macro takes one or more functions, creates a temporary buffer and evaluates the functions as if the temporary buffer were the current buffer. The url-insert-file-contents function takes a URL, calls it and puts the response body in a buffer, which it returns. So now get-bibtex-url-contents has the response body in the current (temporary) buffer. It then takes the value of that buffer (using buffer-string) and stores it in the result variable, which it returns.

(defun get-bibtex-url-contents (url)
  "Fetch the contents of URL and return the result as a string."
  (with-temp-buffer
    (url-insert-file-contents url)
    (setq result (buffer-string)))
  result)

The write-bibtex-entry function takes a BibTeX entry (as a string) and adds it to the end of a references.bib file. It too uses the with-temp-buffer macro to create a temporary buffer, into which it pastes the current contents of the references.bib file. It goes to the end of the file using goto-char and point-max and inserts the BibTeX entry there. Then it goes to the beginning of the newly inserted entry (BibTeX entries begin with @) and cleans and formats the entry. Finally, it takes the whole content of the temporary buffer and writes it to references.bib. Done! (It also refreshes the Citar cache so that I can immediately find the new entry when inserting citations; this step can be replaced or removed if you’re using helm-bibtex/ivy-bibtex or some other alternative to Citar.)

(defun write-bibtex-entry (bibtex-entry)
  "Write BIBTEX-ENTRY (a string) to my `references.bib' file."
  (let ((reference-filepath "/path/to/references.bib"))
    (with-temp-buffer
      (insert-file-contents reference-filepath)
      (goto-char (point-max))
      (insert "\n" (s-trim bibtex-entry) "\n")
      (search-backward "@")
      (bibtex-fill-entry)
      (bibtex-clean-entry)
      (write-region (point-min) (point-max) reference-filepath))
    (citar-refresh)))

I wrote these functions because, having started using Emacs to manage my references back in February (you might’ve noticed the new References section at the end of some posts), I wanted to reduce the friction of adding a paper or book to my bibliography. That’s the blessing and the curse of Emacs: you get endless extensibility, but you also get endless extensibility. It’s kind of a trap, which is why I don’t recommend it for other people. But it’s also fun, and powerful.

Will this save me time in the long run? I’ve added 116 citations to my references.bib so far; over a period of five months that makes roughly 0.77 citations per day. Suppose it would take me ten seconds to find the reference, copy the BibTeX entry, open up references.bib, paste the entry and refresh Citar without these functions. Suppose it does take me three seconds to do the same with these functions. Suppose I spent roughly three hours implementing these functions (they were passable pretty early on but I spent some time fixing a bug). To make up my deficit I need to add 3 × 60 × 60 ÷ (10 - 3) = 1,543 entries. At the current rate, that will take me only 1,543 ÷ 0.77 = 2,004 days or five and a half years!

There are some more advantages in doing this sort of thing:

← Return to home page ↑ Return to top