Canary tracks the master branch on GitHub. Its version numbers follow the format 0.0.0-canary-GIT_COMMIT_SHA, so version ranges in your package.json are not meaningful. But you can install the latest one with:

yarn add @citeproc-rs/wasm@canary
# alternatively, a specific commit
yarn add @citeproc-rs/wasm@0.0.0-canary-COMMIT_SHA

If you use NPM, replace yarn add with npm install.

Including in your project

For Node.js, simply import the package as normal. Typescript definitions are provided, though parts of the API that cannot have auto-generated type definitions are alluded to in doc comments with an accompanying type you can import.

// Node.js
const { Driver } = require("@citeproc-rs/wasm");

Using Webpack

When loading on the web, for technical reasons and because the compiled WebAssembly is large, you must load the package asynchronously. Webpack comes with the ability to import packages asynchronously like so:

// Webpack
import("@citeproc-rs/wasm")
    .then(go)
    .catch(console.error);

function go(wasm) {
    const { Driver } = wasm;
    // use Driver
}

When you do this, your code will trigger a download (and streaming parse) of the binary, and when that is complete, your go function will be called. The download can of course be cached if your web server is set up correctly, making the whole process very quick.

You can use the regular-import Driver as a TypeScript type anywhere, just don't use it to call .new().

Note the caveats in around Microsoft Edge's TextEncoder/TextDecoder support in the wasm-bindgen tutorial.

import { Driver } from "@citeproc-rs/wasm";

function doSomethingWithDriver(driver: Driver) {
    // ...
}

Importing it in a script tag (`web` target)

To directly import it without a bundler in a (modern) web browser with ES modules support, the procedure is different. You must:

Make the _web subdirectory of the published NPM package available in a content directory on your webserver, or use a CDN like unpkg.
Include a <script type="module"> tag in your page's <body>, like so:

<script type="module">
    import init, { Driver } from './path/to/_web/citeproc_rs_wasm.js';
    async function run() {
        await init();
        // use Driver
    }
    run()
</script>

Importing it in a script tag (`no-modules` target)

This replicates the wasm-bindgen guide entry, noting the caveats. You will, similarly to the web target, need to make the contents of the _no_modules subdirectory of the published NPM package available on a webserver or via a CDN.

<html>
  <head>
    <meta content="text/html;charset=utf-8" http-equiv="Content-Type"/>
  </head>
  <body>
    <!-- Include the JS generated by `wasm-pack build` -->
    <script src='path/to/@citeproc-rs/wasm/_no_modules/citeproc_rs_wasm.js'></script>

    <script>
      // Like with the `--target web` output the exports are immediately
      // available but they won't work until we initialize the module. Unlike
      // `--target web`, however, the globals are all stored on a
      // `wasm_bindgen` global. The global itself is the initialization
      // function and then the properties of the global are all the exported
      // functions.
      //
      // Note that the name `wasm_bindgen` will at some point be configurable with the
      // `--no-modules-global` CLI flag (https://github.com/rustwasm/wasm-pack/issues/729)
      const { Driver } = wasm_bindgen;

      async function run() {
        // Note the _bg.wasm ending
        await wasm_bindgen('path/to/@citeproc-rs/wasm/_no_modules/citeproc_rs_wasm_bg.wasm');

        // Use Driver
      }

      run();

    </script>
  </body>
</html>

Usage

Overview

The basic pattern of interactive use is:

Create a driver instance with your style
Edit the references or the citation clusters as you please
Call driver.batchedUpdates()
Apply the updates to your document (e.g. GUI)
Go to step 2 when a user makes a change

Step three is the important one. Each time you edit a cluster or a reference, it is common for only one or two visible modifications to result. Therefore, the driver only gives you those clusters or bibliography entries that have changed, or have been caused to change by an edit elsewhere. You can submit any number of edits between each call.

The API also allows for non-interactive use. See below.

1. Creating a driver instance

First, create a driver. Note that for now, you must also call .free() on the Driver when you are finished with it to deallocate its memory, but there is a TC39 proposal in the implementation phase that will make this unnecessary.

A driver needs an XML style string, a fetcher (below), and an output format (one of "html", "rtf" or "plain").

let driver = Driver.new(cslStyleTextAsXML, fetcher, "html");
// ... use the driver ...
driver.free()

The library parses and validates the CSL style input. Any validation errors are reported, with line/column positions, the text at that location, a descriptive and useful message (only in English at the moment) and sometimes even a hint for how to fix it. This is thrown as an error, which you can catch in a try {} catch (e) {} block.

Fetcher

There are hundreds of locales, and the locales you need change depending on the references that are active in your document, so the procedure for retrieving one is asynchronous to allow for fetching one over HTTP. There's not much more to it than this:

class Fetcher {
    async fetchLocale(lang) {
        return fetch("https://some-cdn-with-locales.com/locales-${lang}.xml")
            .then(res => res.text());

        // or just
        // return "<locale> ... </locale>";
        // return LOCALES_PRELOADED[lang];

        // or if you don't support locales other than the bundled en-US!
        // return null;
    }
}

let fetcher = new Fetcher(); // Pass to Driver.new()

Unless you don't have async syntax, in which case, return a Promise directly, e.g. return Promise.resolve("<locale> ... </locale>").

2. Edit the references or the citation clusters

References

You can insert a reference like so. This is a CSL-JSON object.

driver.insertReference({ id: "citekey", type: "book", title: "Title" });
driver.insertReferences([ ... many references ... ]);
driver.resetReferences([ ... deletes any others ... ]);
driver.removeReference("citekey");

When you do insert a reference, it may have locale information in it. This should be done after updating the references, so any new locales can be fetched.

// May call your Fetcher instance.
await driver.fetchAll();

Citation Clusters and their Cites

A document consists of a series of clusters, each with a series of cites. Each cluster has an id, which is any integer except zero.

// initClusters is like booting up an existing document and getting up to speed
driver.initClusters([
    { id: 1, cites: [ {id: "citekey"} ] },
    { id: 2, cites: [ {id: "citekey", locator: "56", label: "page" } ] },
]);
// Update or insert any one of them like so
driver.insertCluster({ id: 1, cites: [ { id: "updated_citekey" } ] });
driver.insertCluster({ id: 3, cites: [ { id: "new_cluster_here" } ] });

These clusters do not contain position information, so reordering is a separate procedure. Without calling setClusterOrder, the driver considers the document to be empty.

So, setClusterOrder expresses the ordering of the clusters within the document. Each one in the document should appear in this list. You can skip note numbers, which means there were non-citing footnotes in between. Omitting note means it's an in-text reference. Note numbers must be monotonic, but you can have more than one cluster in the same footnote.

driver.setClusterOrder([ { id: 1, note: 1 }, { id: 2, note: 4 } ]);

You will notice that if an interactive user cuts and pastes a paragraph containing citation clusters, the whole reordering operation can be expressed in two calls, one after the cut (with some clusters omitted) and one after the paste (with those same clusters placed somewhere else). No calls to insertCluster need be made.

Uncited items

Sometimes a user wishes to include references in the bibliography even though they are not mentioned in a citation anywhere in the document.

driver.includeUncited("None"); // Default
driver.includeUncited("All");
driver.includeUncited({ Specific: ["citekeyA", "citekeyB"] });

The "All" is based on which references your driver knows about. If you have this set to "All", simply calling driver.insertReference() with a new reference ID will result in an entry being added to the bibliography. Entries in Specific mode do not have to exist when they are provided here; they can be, for instance, the citekeys of collection of references in a reference library which are subsequently provided in full to the driver, at which point they appear in the bibliography, but not items from elsewhere in the library.

3. Call `driver.batchedUpdates()` and apply the diff

This gets you a diff to apply to your document UI. It includes both clusters that have changed, and bibliography entries that have changed.

// Get the diff since last time batchedUpdates, fullRender or drain was called.
let diff = driver.batchedUpdates();

// apply cluster changes to the UI.
for (let changedCluster of diff.clusters) {
    let [id, html] = changedCluster;
    myDocument.updateCluster(id, html);
}

// Null? No change to the bibliography.
if (diff.bibliography != null) {
    let bib = diff.bibliography;
    // Save the entries that have actually changed
    for (let key of Object.keys(bib.updatedEntries)) {
        let rendered = bib.updatedEntries[key];
        myDocument.updateBibEntry(key, rendered);
    }
    // entryIds is the full list of entries in the bibliography.
    // If a citekey isn't in there, it should be removed.
    // It is non-null when it has changed.
    if (bib.entryIds != null) {
        myDocument.setBibliographyOrder(bib.entryIds);
    }
}

Note, for some intuition, if you call batchedUpdates() again immediately, the diff will be empty.

Bibliographies

Beyond the interactive batchedUpdates method, there are two functions for producing a bibliography statically.

// returns BibliographyMeta, with information about how a library consumer should
// lay out the bibliography. There is a similar API in citeproc-js.
let meta = driver.bibliographyMeta();

// This is an array of BibEntry
let bibliography = driver.makeBibliography();
for (let entry of bibliography) {
    console.log(entry.id, entry.value);
}

Preview citation clusters

Sometimes, a user wants to see how a cluster will look while they are editing it, before confirming the change.

let cites = [ { id: "citekey", locator: "45" }, { ... } ];
let positions = [ ... before, { id: 0, note: 34 }, ... after ];
let preview = driver.previewCitationCluster(cites, positions, "html");

The positions array is exactly like a call to setClusterOrder, except exactly one of the positions has an id of 0. This could either:

Replace an existing cluster's position, and preview a cluster replacement; or
Represent the position a cluster is hypothetically inserted.

If you passed only one position, it would be like previewing an operation like "delete the entire document and replace it with this one cluster". That would mean you would never see "ibid" in a preview. So for maximum utility, assemble the positions array as you would a call to setClusterOrder with exactly the operation you're previewing applied.

Non-Interactive use, or re-hydrating a previously created document

If you are working non-interactively, or re-hydrating a previously created document for interactive use, you may want to do one pass over all the clusters in the document, so that each cluster and bibliography entry reflects the correct value.

// Get the clusters from your document (example)
let allNotes = myDocument.footnotes.map(fn => {
    return { cluster: getCluster(fn), number: fn.number }
});

// Re-hydrate the entire document based on the reference library and your
// document's clusters
driver.resetReferences(myDocument.allReferences);
driver.initClusters(allNotes.map(fn => fn.cluster));
driver.setClusterOrder(allNotes.map(fn => { id: fn.cluster.id, note: fn.number }));

// Render every cluster and bibliography item.
// It then drains the update queue, leaving the diff empty for the next edit.
// see the FullRender typescript type
let render = driver.fullRender();

// Write out the rendered clusters into the doc
for (let fn of allNotes) {
    fn.renderedHtml = render.allClusters[fn.cluster.id];
}

// Write out the bibliography entries as well
let allBibKeys = render.bibEntries.map(entry => entry.id);
for (let bibEntry of render.bibEntries) {
    myDocument.bibliographyMap[entry.id] = entry.value;
}

// Update your (example) UI
updateUserInterface(allNotes, myDocument, whatever);

FAQs

What is @citeproc-rs/wasm?

Is @citeproc-rs/wasm popular?

Is @citeproc-rs/wasm well maintained?

Package last updated on 19 Nov 2020

Did you know?

Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.

Install

@citeproc-rs/wasm

Installation / Release channels