labels: cache parsed regex in FastRegexMatcher

[dimitarvdimitrov]

Summary

Caches the parsed regex during FastRegexMatcher initialization instead of re-parsing the regex each time SingleMatchCost() is called.

Since the tree is optimized in stringMatcherFromRegexp now the cost of some matchers has changed because we do the estimation on the optimized tree instead of on the original tree.

---

Note

Caches the parsed regex in FastRegexMatcher and uses it for cost estimation (with sane fallbacks), updating test expectations accordingly.

• Labels / Matching:
  • Cache parsed regex in FastRegexMatcher via new field parsedRe; set during initialization in newFastRegexMatcherWithoutCache() and reused in SingleMatchCost() to avoid re-parsing.
  • Refine FastRegexMatcher.SingleMatchCost() to fall back to estimatedStringEqualityCost on parse failure and cap with max(estimatedStringEqualityCost, costEstimate(parsed)).
• Tests:
  • Update expected costs in model/labels/cost_test.go for several =~ and !~ cases to reflect estimation on the optimized parse tree.
• Internals:
  • Wire parsed tree from stringMatcherFromRegexp() into the matcher (m.parsedRe = parsed).

^{Written by Cursor Bugbot for commit a963ec3. This will update automatically on new commits. Configure here.}

[Copilot]

Pull request overview

This PR optimizes regex cost estimation by caching the computed cost during FastRegexMatcher initialization, eliminating redundant regex parsing on every SingleMatchCost() call. The cost is now calculated once after the regex tree has been optimized by stringMatcherFromRegexp, resulting in more accurate cost estimates.

Key changes:

• Added singleMatchCost field to FastRegexMatcher to cache the estimated cost computed during initialization
• Refactored SingleMatchCost() method to check optimization paths first (setMatches, map-based matchers, prefix), then fall back to the cached cost
• Improved documentation for costEstimate() function with proper function comment and formatted TODO

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
model/labels/regexp.go	Added singleMatchCost field to cache cost estimation; computed after regex tree optimization in newFastRegexMatcherWithoutCache()
model/labels/cost.go	Refactored SingleMatchCost() to use cached cost and removed duplicate optimization checks; improved costEstimate() documentation
model/labels/cost_test.go	Updated test expectations to reflect new cost estimates based on optimized regex trees

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[chencs]

Nice find!

Dismissing until CI is fixed

[bboreham]

Nitpicking the title: you're caching the parsed data structure not the cost.

[dimitarvdimitrov]

ugh, i didn't update the title again after fixing the PR