Add known upper limit to capture search. #217
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| self.captures_nfa( | ||
| MatchNfaType::Auto, slots, text, s) | ||
| MatchNfaType::Auto, slots, &text[..e], s) |
There was a problem hiding this comment.
Should this change also be applied above in the MatchType::Literal case?
There was a problem hiding this comment.
(And below in the DfaAnchoredReverse case?)
There was a problem hiding this comment.
Good catch on the literal case. For anchored reverse it won't help because if there's a match, it's guaranteed to end at the input (since the case only happens when the regex is anchored to the end of the text).
The DFA will report the end location of a match, so we should pass that along to capture detection. In theory, the DFA and the NFA report the same match locations, so this upper bound shouldn't be necessary---the NFA should quit once it finds the right match. It turns out though bounding the text has two important ramifications: 1. It will enable the backtracking engine to be used more often. In particular, the backtracking engine can only be used on small inputs and this change decreases the size of the input by only considering the match. 2. The backtracking engine must start every search by zeroing memory that is proportional to the size of the input. If the input is smaller, then this runs more quickly. We are also careful to bound the match to one additional "character" past the end of the match, so that lookahead operators work correctly.
BurntSushi
force-pushed
the
fix-capture-perf
branch
from
7084f14
to
49e8df5
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The DFA will report the end location of a match, so we should pass that
along to capture detection. In theory, the DFA and the NFA report the
same match locations, so this upper bound shouldn't be necessary---the
NFA should quit once it finds the right match. It turns out though
bounding the text has two important ramifications:
particular, the backtracking engine can only be used on small inputs and
this change decreases the size of the input by only considering the
match.
that is proportional to the size of the input. If the input is smaller,
then this runs more quickly.
We are also careful to bound the match to one additional "character"
past the end of the match, so that lookahead operators work correctly.
See also: #215.