Rewrite parser generator #84
Conversation
heckad
force-pushed
the
fix-zero-or-more-problem-parsing
branch
from
220a632
to
5471713
Compare
| @final | ||
| class NotParsedMarker(Enum): | ||
| marker = 0 | ||
|
|
||
|
|
||
| not_parsed = NotParsedMarker.marker |
There was a problem hiding this comment.
You can just use object() and it'll give the same effect
Also @final doesn't do a lot, It's better to just use Literal[not_parsed] for annotations
| | ENDMARKER { None } | ||
| | ENDMARKER { [] } |
There was a problem hiding this comment.
Why this change was made?
There was a problem hiding this comment.
None is now a valid result but we expect a list to be returned there.
| return not ok | ||
| return not_parsed if ok is not not_parsed else True |
There was a problem hiding this comment.
bool check is more than enough here and in positive_lookahead
There was a problem hiding this comment.
No, False now a correct parsing result. For mark incorrect parsing result we always should use not_parsed.
I don't use any another idea. I just improved my last attempt. Now |
|
@MatthieuDartiailh, please look at pull request. |
heckad
force-pushed
the
fix-zero-or-more-problem-parsing
branch
2 times, most recently
from
48d9637
to
5471713
Compare
lysnikolaou
force-pushed
the
main
branch
2 times, most recently
from
5aae014
to
3cefc85
Compare
What does this pull request do?
This pull request changes what we consider successful parsing. Now we have the
not_parsedmarker (the name can be changed) which mark that rule can't parse input. This solves the problem with empty collections being interpreted as parsing failure and also allows you to drag None up. Unfortunately this is a very big change. Parsers generated by the previous version will not work. However, this solves all problems.