Lyrics Filler with Tonal Concerns (especially tonal languages). In TypeScript.
Source Lyrics (String)
`- Verses (Separated by $ (Or by 2 newlines))
`- Sentences (Separated by space, comma (,) or by newlines)
`- Phrases (Separated by /)
Example
今天我/寒夜裡看雪飄過 <- This is a whole sentence, while music breaks at /
懷著冷卻了的心窩漂遠方
風雨裏追趕
霧裡分不清影蹤
天空海闊你與我/可會變 <- This is a whole sentence, while the melody pauses at /
（誰沒在變）
...
Which is parsed into a string in the form ___/_______,_____,_____,_____,_____/____$...
Can use - for AND, | for OR phrases.
No prefix <- Dictionary with pure vocabs
: <- dictionary alias, that decomposes into aliases or no prefix
^ <- Sentence splitter
$ <- Verses splitter
# <- Song
Not needed at the moment:
, <- Phrase splitter
/ <- Tempo splitter (there should be a break between vocabs, but this doesn't break the phrase (like between 如果 & 可以))
Say, a song can be filled by the Story : "Start THEN Development THEN Twist THEN End".
Which, only Development can consume more than 1 verse.
So the corresponding constraint (conceptually, not syntactically) would be: #Story = $Start-#Development-$Twist-$End.
Which #Development should be decomposed as $Development|$Development-#Development (1 or more than 1 $Development).
Let's dig deeper with $Start. Assume it is Name THEN Background, which only Background can span across sentences.
Thus $Start = ^Name-$Background, which $Background = ^Background|^Background-$Background.
Then for sentence ^Background, it consists of several phrases (to be separated by commas). Let's just call them ^phrases (as if we are treating an incomplete sentence as sentence itself), which are in the form of :subject-:verb-:object which connects directly to dictionary subject, etc..