This is a grab bag of scripts I used for combing thru a digital text & finding songs. Here's a long-winded description of the whole thing.
These are alternatingly bash, node and Go scripts. The Go scripts in particular have some dependencies, which should be fetch-able using
go get ./....
What follows are the steps in the order I followed them.
cat big-input-file.txt | ./find_songs.sh > 1-possible-songs.txt
Takes the grep result and excludes anything that's not mostly capitalized (crude "songfulness" heuristic).
cat 1-possible-songs.txt | ./mostly_capitals.js > 2-mostly-capitals.txt
Takes the filtered grep result, greps again with line numbers in the source text, and (with the following snippet) sorts/unique-ifies things so you end up with a unique list of mmatches in the order they appeared in the text. It also includes the originally matched song name in brackets (it's used in the next step).
cat 2-mostly-capitals.txt | ./in_context.sh | sort -t: -n -k1 > 3-in-context.txt
This takes the ordered matches and, for each one, interactively prompts you to select a nearby artist-like name. The result is piped out to stdout.
cat 3-in-context.txt | go run digfornames/main.go > 4-songs-with-artists.txt
This takes a list of songs with artists, searches YouTube for them, and runs a little server which interactively presents the results and prompts you to select the one you want. The result is piped out to stdout.
cat 4-songs-with-artists.txt | go run findyoutube/main.gog > 5-youtube-urls.txt
This isn't part of
elvis-tools, but say you had a list of YouTube URL's, you could use this command to convert them to mp3's:
youtube-dl --batch-file 5-youtube-urls.txt --extract-audio --audio-format mp3 --output "%(autonumber)s%(title)s.%(ext)s
And furthermore, you can combine them into one giant mp3 with this:
cat *.mp3 > elvis.mp3