So I have the following problem, in YAML single quotes are a strange beast.
Short primer: a YAML single quotes spans one line, and has no escape character. Instead the ' character is doubled ('') to escape it.
'I''m a string' # parses to Scalar value of "I'm a string"
However, if the string is empty, then it’s not treated as escape.
'' # parses to Scalar value "" (empty string)
Which can be a problem. Consider following line:
{'':''''} # parses to Map {"": "''"}
When applied to a bitmask you get
| bytes | { |
' |
' |
: |
' |
' |
' |
' |
} |
|---|---|---|---|---|---|---|---|---|---|
| bitmask | 0 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 0 |
| relevant quotes | 1 | 1 | 1 | 1 |
But in another example
''' ' # Parses to Scalar "'' "
Gets following bitmaks
| bytes | ' |
|
' |
' |
' |
|---|---|---|---|---|---|
| bitmask | 1 | 0 | 1 | 1 | 1 |
| relevant quotes | 1 | 1 |
In second example the 3rd and 4th byte are ignored because they are inside an existing single quote string i.e they are treated as '' escape for '. And this logic depends on how many single quotes were beforehand.
How to express this logic branchlessly if bit state depends on previous bytes being in/out single quote?
I tried several ways to calculate, theses bit values, but I’m kinda stuck, the furtherest I got was logic that can select the bits correctly in case #1. Logic is bellow in pseudo Rust code
fn find_relevant_bytes(bytes: u8) -> u8 {
let low_part = bytes & bytes >> 1;
let high_part = bytes & bytes << 1;
return low_part ^ high_part;
}