ECR: support removing leading indentation and trailing newline #2991
Conversation
asterite
force-pushed
the
feature/ecr_dash
branch
from
052ea93 to
ee5d024
Compare
|
I find this to be a less functional (-% is now illegal to use?) much more impenetrable and messy implementation that favours an ugly procedural style over OO principles, but if this is what you prefer I guess it'll have to do. I also wonder if this implementation correctly respects all line semantics, but I'm not going to check. Thanks for solving this. |
| @@ -42,13 +63,40 @@ module ECR | |||
| end | |||
| end | |||
|
|
|||
| private def append_loc(str, filename, token) | |||
| private def do_supress_leading(token, string) | |||
There was a problem hiding this comment.
Maybe suppress_leading_whitespace?
asterite
force-pushed
the
feature/ecr_dash
branch
2 times, most recently
from
344621c to
5fbbdc0
Compare
|
I agree. Skipping trailing and leading linefeeds can be considered as mere modifiers and we should avoid new types, as well as regular expressions —they're slow and may quickly increase compile time— to achieve this. Yet, maybe the lexer could foresee a second char, so we could properly parse |
|
@beno I fixed the I don't try to use OOP for everything, every programming style has its uses and pros/cons. In this case I want ECR parsing to be super fast, and creating many objects and using regex is slow. |
asterite
force-pushed
the
feature/ecr_dash
branch
from
5fbbdc0 to
228916c
Compare
Fixes #2980
This is an alternative implementation to #2985
@beno I hope you don't mind I decided to try to implement this, but I think the needed changes were really few, and there's no need to introduce more classes nor use regex for this. To do this one just has to look at at most one token ahead, and check if its string value needs to be adjusted. Please check the code, I might have made a mistake or maybe I didn't implement it correctly (I tried a few templates and they seemed to work exactly like ERB).
This now supports all of ERB idioms, as you can see in the specs. Additionally, one can also use
<%-= ... %>, which removes leading indentation from it, and I guess because of how it's done this also works for<%-# ... %>.