informit.citations

This is a very messy citation exporter for Informit.org. For some strange reason, despite being run by an Australian university, informit does not support exporting AGLC-compliant citations.

Usage

1. Start with a list of Informit DOIs - perhaps you were saving articles while doing your research
2. Run injest.js to build the API call. injest.js expects a good.links file with a list of the DOIs you want to export. You can use other tools (awk/sed) to build the request.
3. Retrieve an API response from the URL generated by the previous step with, e.g. weget -O results.json <url>
4. Run index.js - this will parse the information in results.json and output a 95% AGLC4-complaint list of references in markdown format.
5. Run index.js bib to format the reference list as an AGLC bibliography by reversing the first and last names of the first-listed author and removing the full stop at the end of references.
6. Optional: convert Markdown to Word Doc with: [pandoc](https://pandoc.org/) -s --ascii -o aglc.doc aglc.md

Caution: manual review is required. This script tries to comply with AGLC4, but I cannot guarantee perfect coverage. Sometimes there are AGLC4 edgecases which I have not considered (such as journal articles spanning multiple years) and sometimes there are deficiencies in the API response, such as missing delimiters in container-title or missing keys.

Methodology

Old Method

The old method relied on scraping the HTML from the Informit website to build citation information and save it in a JSON dictionary. A second script would then convert the dictionary into Markdown. The markdown could be converted to doc/rtf with pandoc.

Update 2022-01-09

The web scraping method produces inconsistent results. Some pages just do not render the tags I rely upon for citation information. Fortunately, Informit exposes an API which they themselves use to export citations. It supports batch jobs, too!

The URL takes the form of:

https://data.informit.org/action/exportCiteProcCitation?dois=DOI_LIST_HERE&targetFile=custom-refWorks&format=text

The DOI list is comma-separated and must include the Informit prefix 10.3316/.

The workflow is to:

1. Get the DOIs of the files we want to cite with, say, ls -1 /path/to/saved/articles > good.links (assuming files were downloaded in the first place and filenames weren't changed)
2. Build the API call - note that local filenames won't have the 10.3316 prefix
3. Fire the call and retrieve the JSON - at the moment this needs to be done in wget as got does not handle all the redirects and cookies that the API requires.
4. Process the JSON citations and write them to Markdown
   1. Bonus: write separate bibliography and footnote files
   2. Bonus: process.exec pandoc conversion

API Response Spec

The API returns a JSON response. Two top-level keys are of relevance:

• items, which is an array of citation objects; and
• exportedDoiLength which is self-explanatory and may be redundant if you just use items.length.

A citation object looks like:

{
	"<doi>": {
		// "<citation-key>:<citation:value>
		// ...
	}
}

Citation fields vary between articles. These are the most relevant:

• author, which is an array of author objects, each containing a family and given key, like:

"author": [
  {
    "family": "COOPER",
    "given": "RE"
  }
]

• issued.date-parts[0] - the year the article was published. The date is presented as an array, so just look at the first element
• container-title this can be ambiguous: it contains either the name of the journal only (if separate keys exist for volume, issue and/or page), or journal name + volume and issue + page range
• volume, issue and title- are self explanatory.
• page is the page-range for the article, where the start and end pages are separated by a dodgy unicode en dash or hyphen.

one-liners

Download a file straight from the abstract page:

In the browser console:

let source = window.location.href; let parsed = source.split('/'); parsed.splice(4, 0, "pdf"); let dest = parsed.join('/'); dest += '?download=true'; location.replace(dest)

License

Code licensed Apache 2. Documentation licensed CC-BY-SA.

This project uses node.js, got and lodash.

Known issues/improvements

• Building the API call and retrieving the API response currently requires two separate steps/scripts. While the API does not require credentials, it nevertheless sets and checks for certain cookies and headers that got does not handle by default. I don't have the time to look into it for now, so just use wget.
• I'm not using async code 100% effectively