OverDrive is great and distributes DRM-free MP3s instead of some fragile DRM-ridden format, which is awesome. Way to go, Rakuten / OverDrive, fight the man!

Their "OverDrive Media Console" application on Mac OS X is pretty simple... so simple, I'm like, why have an app?

So I wrote a shell script, overdrive.sh, which takes one or more .odm files (which are just XML), and downloads the audio content files locally, just like the app.

The following will install the main script to your /usr/local/bin folder and mark it executable:

curl https://chbrown.github.io/overdrive/overdrive.sh > /usr/local/bin/overdrive
chmod +x /usr/local/bin/overdrive

Assuming /usr/local/bin is on your PATH, you can now run overdrive --help to show all the options.

This script is tested (i.e., developed and used) on macOS with bash 4.4. It depends on the following (potentially non-standard?) executables being available on your PATH:

• curl
• uuid
• xmlstarlet
• iconv
• openssl
• base64
• tidy

Download an OverDrive loan file from your library or wherever. I'll assume that yours is called Novel.odm.

When you run overdrive download Novel.odm, the script first checks if there is already a Novel.odm.license file alongside Novel.odm. If that file already exists, the script will not request a new license, since the OverDrive server will only grant one license per .odm loan. If not, it will request the license from the OverDrive server and write it to a new Novel.odm.license file, by performing the following actions:

1. Extract the AcquisitionUrl and MediaID from the Novel.odm file.
2. Compute a Base64-encoded SHA-1 hash from a few |-separated values and a suffix of OVERDRIVE*MEDIA*CONSOLE, but backwards. (Thanks to https://github.com/jvolkening/gloc for somehow figuring out how to construct that hash!)
3. Using those values and a randomly generated ClientID GUID, it submits a request to the OverDrive server to get the full license for this book, which is an XML file that has a <License> element at the root. That element contains a long Base64-encoded <Signature>, which is subsequently used to request the content files.

Now, license in hand, the script downloads the audio content files by taking the following steps:

4. Extract the Title and Author values from the CDATA content nested in Novel.odm.
5. For each of the parts of the book listed in Novel.odm, make a request to another OverDrive endpoint, which will validate the request and redirect to the actual MP3 file on their CDN, and save the result into a folder in the current directory, named like Author - Title/Title-Part0N.mp3.

The OverDrive format makes "returning" a loan extremely simple. All you have to do is request the URL specified by the <EarlyReturnURL> element in the loan file. The return command does exactly that, e.g.:

overdrive return Novel.odm

If you have trouble getting the script to run successfully, add the --verbose flag and retry, e.g.:

overdrive download Novel.odm --verbose

This will call set -x to turn bash's xtrace option on, which causes a trace of all commands to be printed to standard error, prefixed with one or more + signs.

If that doesn't help you debug the problem, open an issue, including the full debug output (optimally as a gist), and I'll try to help you out.

The most up-to-date ID3(v2) tagging tool I've found is mutagen, which is used and maintained by the Quod Libet audio player project. It was the only tagger that let me set the Genre tag to an arbitrary string (like iTunes).

Unfortunately, it appears that iTunes always loads .mp3s as "Music" despite the Genre (there doesn't seem to be any way to import .mp3s into iTunes as "Audiobooks", you have to open iTunes, "Get Info" for the intended songs, and set the "Media Kind" to "Audiobook" manually.

Copyright © 2017–2018 Christopher Brown. MIT Licensed.