Create transcripts of youtube, and perhaps other online services
See it working at www.transcriptsearch.com.es
The main idea of this set of scripts is to offer a search engine for youtube transcription. We are agnostics about the spider; in this simple case we just offer a sitemap to google custom engine, and let it to do the crawl. You could also use Yacy, or perhaps a python based crawler and search engine.
The scripts are made to work with apache WSGIa module. You must provide executions for /id, /comic and / :
WSGIScriptAlias /id /cript/video.wsgi
WSGIScriptAlias /comic /cript/comic.wsgi
WSGIScriptAlias / /cript/index.wsgi
The main worry for installation is to be sure that you have all the needed modules for python and the webserver. If you do implementations using other libraries (eg, getting rid of httplib2), please honor the AGPLv3, it is so easy as branching it here in github :-D
You must also provide a configfile with your youtube API key and a comma separated list of proxies. That, or edit the proxy code out.
If you plan to use this idea as a service, you should consider Youtube terms and conditions. Particularly, under 5.3, you should not cache the contents of a crawl, so you should disable the cache feature in httplib2 for the "video.wsgi" page or purge it frequently. Surely the license of google about youtube contents is different, so you can consider to allow them to keep the cache. You can also consider to "redirect 30x" directly to youtube.com, but googlebot does not like extensive use of redirections, so better do not.
Neither the video.wsgi script nor the comic.wsgi are replaying the video, so most of the TOS issues about independent players do not apply. But comic.wsgi can be considered a rework of the original work; so you are in safer grounds if you only allow comic.wsgi for videos with Creative Commons license. The videos under general youtube license also allow for reworking, but the wording of the license is more convolved and it only applies while the video still exists in youtube.
You can also consider to offer the original Youtube Player in the result pages, but I am not sure if in such case you go away from the search aid approach and you wander towards a typical youtube scraper clone.
In any case, do not take my words as legal advice; it is simply a guide for you to understand that there are issues always when you plan an internet search aid. You must consider your local rules, and you can also consider Google's BCP: as they are the owners both of Youtube and of the Dominant Search Engine in the web, their commom practices are surely the most valid practices.