SoylentNews

martyb [soylentnews.org] writes:

Over the years I have viewed many a video on YouTube. I quickly noticed an "ID" string that appeared in each video URL. Here's an example: https://www.youtube.com/watch?v=ShvnDSgjfXw [youtube.com] -- see that string "ShvnDSgjfXw"? What characters are permitted? How long is it?

Along the way, I came upon an amazingly useful utility: youtube-dl [youtube-dl.org]. I accidentally discovered that it will happily download a YouTube video given just the Video ID. (Don't let the name of the utility mislead you; it seems to work fine with Instagram, Twitter, Sound Cloud... it's amazing!)

Now with my curiosity suitably piqued, I started a genuine search for what the parameters were that defined a valid YouTube Video ID. This question [stackexchange.com] on "Web Applications Stack Exchange" was most helpful. Especially this response [stackexchange.com].

It appears that the Video ID (and the Channel ID) are modified base64 encodings of 64-bit (and 128-bit) integers. The primary change is that the base64 encoding produces two characters that are verboten in URLs. A generated "/" is replaced with "-" and a generated "+" is replaced with a "_".

There is no official documentation claiming that the ID lengths are guaranteed to always be 11 or 22 characters long, but empirical evidence suggests that is the current, de-facto standard.

There is even mention of " the maximally-constrained regular expression (RegEx) for the videoId" being:

[0-9A-Za-z_-]{10}[048AEIMQUYcgkosw]

Things get even more interesting if you are using Windows. Under NTFS, file names default to be case-preserving, but case-insensitive. Say I create a file called "Foo.txt" and then get a directory listing. Sure enough, I see: "Foo.txt" displayed. The fun comes if I do "DIR foo.txt" or "DIR FOO.TXT" or any other variation... they all find the same file: "Foo.txt"; this is counter to Unix where filenames are case-sensitive and each of those variations would be treated as separate and distinct files. Though it is possible to make an NTFS volume case-sensitive, it is not for the faint of heart [nicklowe.org]!

One could, therefore, reverse-engineer the integer that produced the Video ID and use that in addition (or for the adventuresome: instead of) the Video ID.

The whole discussion was well-worth the read and highly recommended for anyone who would like more information on where it came from and how it came about.

Original Submission