NodeJS implementation (written in TypeScript) and HTTP micro-services (Express middleware) for https://github.com/readium/architecture/tree/master/streamer
-
https://nodejs.org NodeJS >= 8, NPM >= 5 (check with command line
node --version
andnpm --version
) - OPTIONAL: https://yarnpkg.com Yarn >= 1.0 (check with command line
yarn --version
)
https://github.com/readium/r2-streamer-js
There is no github.io site for this project (no gh-pages branch).
https://www.npmjs.com/package/r2-streamer-js
Command line install:
npm install r2-streamer-js
OR
yarn add r2-streamer-js
...or manually add in your package.json
:
"dependencies": {
"r2-streamer-js": "latest"
}
The JavaScript code distributed in the NPM package is usable as-is (no transpilation required), as it is automatically-generated from the TypeScript source.
Several ECMAScript flavours are provided out-of-the-box: ES5, ES6-2015, ES7-2016, ES8-2017:
https://unpkg.com/r2-streamer-js/dist/
(alternatively, GitHub mirror with semantic-versioning release tags: https://github.com/edrlab/r2-streamer-js-dist/tree/develop/dist/ )
The JavaScript code is not bundled, and it uses require()
statement for imports (NodeJS style).
More information about NodeJS compatibility:
Note that web-browser Javascript is currently not supported (only NodeJS runtimes).
The type definitions (aka "typings") are included as *.d.ts
files in ./node_modules/r2-streamer-js/dist/**
, so this package can be used directly in a TypeScript project.
Example usage:
// from the index file
import { Server } from "r2-streamer-js";
// ES5 import (assuming node_modules/r2-streamer-js/):
import { Server } from "r2-streamer-js/dist/es5/src/http/server";
// ... or alternatively using a convenient path alias in the TypeScript config (+ WebPack etc.):
import { Server } from "@r2-streamer-js/http/server";
https://david-dm.org/readium/r2-streamer-js
A package-lock.json is provided (modern NPM replacement for npm-shrinkwrap.json
).
A yarn.lock file is currently not provided at the root of the source tree.
https://travis-ci.org/readium/r2-streamer-js
TravisCI builds are triggered automatically at every Git "push" in the develop
branch.
A test server app (not production-ready) is automatically deployed at every Git "push" in the develop
branch:
NPM package (latest published):
https://unpkg.com/r2-streamer-js/dist/gitrev.json
Alternatively, GitHub mirror with semantic-versioning release tags:
https://raw.githack.com/edrlab/r2-streamer-js-dist/develop/dist/gitrev.json
Latest deployment:
https://streamer.edrlab.org/version
Command line steps (NPM, but similar with YARN):
cd r2-streamer-js
-
git status
(please ensure there are no local changes, especially inpackage-lock.json
and the dependency versions inpackage.json
) -
rm -rf node_modules
(to start from a clean slate) -
npm install
, or alternativelynpm ci
(both commands initialize thenode_modules
tree of package dependencies, based on the strictpackage-lock.json
definition) -
npm run build:all
(invoke the main build script: clean, lint, compile) -
ls dist
(that's the build output which gets published as NPM package) -
npm run server-debug -- PATH_TO_EPUB_OR_DIR " -1"
(ES8-2017 dist, path is relative or absolute, -1 means no limits for HTTP header prefetch Links) - or:
npm run start -- 99
(ES6-2015 dist, default./misc/epubs
folder, the 99 value overrides the default maximum number of HTTP header prefetch Links)
// ES5 import (assuming node_modules/r2-streamer-js/):
import { Server } from "r2-streamer-js/dist/es5/src/http/server";
// ... or alternatively using a convenient path alias in the TypeScript config (+ WebPack etc.):
import { Server } from "@r2-streamer-js/http/server";
// Constructor parameter is optional:
// disableDecryption: true
// disableOPDS
// disableReaders: true
// disableRemotePubUrl: true to deactivate
const server = new Server({
disableDecryption: false, // deactivates the decryption of encrypted resources (Readium LCP).
disableOPDS: true, // deactivates the HTTP routes for the OPDS "micro services" (browser, converter)
disableReaders: true, // deactivates the built-in "readers" for ReadiumWebPubManifest (HTTP static host / route).
disableRemotePubUrl: true, // deactivates the HTTP route for loading a remote publication.
maxPrefetchLinks: 5, // Link HTTP header, with rel = prefetch, see server.ts MAX_PREFETCH_LINKS (default = 10)
});
// First parameter: port number, zero means default (3000),
// unless specified via the environment variable `PORT` (process.env.PORT).
// Tip: the NPM package `portfinder` can be used to automatically find an available port number.
const url = await server.start(3000, false);
// Second constructor parameter: if true, HTTPS instead of HTTP, using a randomly-generated self-signed certificate.
// Also validates encrypted HTTP header during request-request cycles, so should only be used in runtime
// contexts where the client side has access to the private encryption key (i.e. Electron app, see r2-navigator-js)
console.log(server.isSecured()); // false
// A client app that is capable of setting HTTP headers for every request originating from content webviews
// can obtain the special encrypted header using this function:
// (as used internally by the Electron-based `r2-navigator-js` component to secure the transport layer)
const nameValuePair = server.getSecureHTTPHeader(url + "/PATH_TO_RESOURCE");
console.log(nameValuePair.name);
console.log(nameValuePair.value);
// http://127.0.0.1:3000
// Note that ports 80 and 443 (HTTPS) are always implicit (ommitted).
console.log(url);
// `serverInfo.urlScheme` ("http")
// `serverInfo.urlHost` ("127.0.0.1")
// `serverInfo.urlPort` (3000)
console.log(server.serverInfo());
// Calls `uncachePublications()` (see below)
server.stop();
console.log(server.isStarted()); // false
To serve a /robots.txt
file that completely disables search robots:
// Call this before `server.start()`
server.preventRobots();
To add custom HTTP routes:
// Call these before `server.start()`.
// They are equivalent to `app.use()` and `app.get()`, where `app` is the underlying Express instance:
server.expressUse("/static-files", express.static("/path/to/files", {
dotfiles: "ignore",
etag: true,
fallthrough: false,
immutable: true,
index: false,
maxAge: "1d",
redirect: false,
}));
server.expressGet(["/hello.html"], (req: express.Request, res: express.Response) => {
// Optionally, to add permissive CORS headers to the HTTP response
server.setResponseCORS(res);
res.status(200).send("<html><body>Hello</body></html>");
});
To register publications references (local filesystem paths) inside the internal server state (which is used to create the OPDS2 feed, see below):
// This can be called before or after `server.start()`:
// the returned array contains URL routes to the ReadiumWebPubManifests,
// e.g. `/pub/ID/manifest.json`, where `ID` is the base64 encoding of the registered path.
// Note that the returned base64 URL path components are already URI-encoded (escaped).
// (`=` and `/` are typically problematic edge-cases)
const publicationURLs = server.addPublications(["/path/to/book.epub"]);
// ...then:
const publicationPaths = server.getPublications(); // ["/path/to/book.epub"]
// ...and (calls `uncachePublication()`, see below):
const publicationURLs = server.removePublications(["/path/to/book.epub"]);
To get the OPDS2 feed for the currently registered publications:
// This launches a potentially time-consuming Node process that scans (loads) each registered Publication,
// and stores the generated OPDS2 feed inside a temporary filesystem location.
// So this returns `undefined` at the first call, and the client must invoke the function again later.
// Note that both `addPublications()` and `removePublications()` clear the OPDS2 feed entirely,
// requiring its subsequent re-generation (full scan of registered publication paths).
// (poor design, but at this stage really just an OPDS2 demo without real use-case)
const opds2 = server.publicationsOPDS();
To actually load+parse a publication reference (local filesystem path) into a ReadiumWebPubManifest Publication instance, stored in the server's state:
// The Publication object model is defined in `r2-shared-js`
const publication = await server.loadOrGetCachedPublication("/path/to/book.epub");
// The above is basically a lazy-loader that checks the cache before loading+parsing a publication,
// equivalent to:
const publication = server.cachedPublication("/path/to/book.epub");
if (!publication) {
publication = ...; // load and parse "/path/to/book.epub"
server.cachePublication("/path/to/book.epub", publication);
}
console.log(server.isPublicationCached("/path/to/book.epub")); // true
// see also:
// (calls `publication.freeDestroy()` to cleanup allocated objects in the Publication,
// particularly the file handle to the underlying zip/EPUB/CBZ file)
server.uncachePublication("/path/to/book.epub");
server.uncachePublications();
Note that HTTP/remote publications URLs can be loaded into the server's cache
and subsequently served by the streamer without prior registration via addPublications()
.
However, publications from the local filesytem will only be served when registered,
even if they are cached (in other words, the HTTP route is disabled when the publication is non-registered).