mirror of
https://github.com/hanzoai/traces.git
synced 2026-10-30 18:20:41 +00:00
76c2c85b97
* [Generator] Add DRAIN span_name sanitizer * Fix linter * Fix formatting * Update default manifest * Fix build error in update-fixtures * Remove bad test * Remove misleading comment * Add per-tenant override for drain mode * Do a machines job * Address PR feedback * Address PR review: rename drain_mode config and improve sanitizer - Rename `drain_mode` to `span_name_sanitization` across all config, legacy config, interfaces, and overrides for better user-facing clarity - Add explicit default value "disabled" in RegisterFlagsAndApplyDefaults - Add ValidateSpanNameSanitization function in validation/fields.go - Rename metrics to remove DRAIN reference: `drain_spans_compacted_total` -> `spans_sanitized_total` `post_drain_demand_estimate` -> `post_sanitization_demand_estimate` - Pass configured staleDuration to NewDrainSanitizer instead of hardcoded 15min, consistent with other registry components - Extract "span_name" magic string to labelSpanName constant - Update all tests with new names and signatures * Update manifest * Use empty string as disabled * Update CHANGELOG.md Co-authored-by: Suraj Nath <[email protected]> * Update manifest --------- Co-authored-by: Suraj Nath <[email protected]>
44 lines
1.2 KiB
Go
44 lines
1.2 KiB
Go
package drain
|
|
|
|
import "unicode"
|
|
|
|
type IsDataHeuristic func(token string) bool
|
|
|
|
// hexWithAtLeastOneDigit checks if the token contains at least one digit and is hex-like or guid-like.
|
|
func hexWithAtLeastOneDigit(token string) bool {
|
|
atLeastOneDigit := false
|
|
for _, c := range token {
|
|
if unicode.IsNumber(c) {
|
|
atLeastOneDigit = true
|
|
continue
|
|
}
|
|
if unicode.Is(unicode.Hex_Digit, c) || unicode.IsPunct(c) {
|
|
continue
|
|
}
|
|
// Non-hex, non-punct found.
|
|
return false
|
|
}
|
|
// All hex-like or guid-like values
|
|
// But still require at least one digit.
|
|
return atLeastOneDigit
|
|
}
|
|
|
|
// significantNumbers checks if the token is comprised of at least 25% numbers.
|
|
func significantNumbers(token string) bool {
|
|
numberCount := 0
|
|
for _, c := range token {
|
|
if unicode.IsNumber(c) {
|
|
numberCount++
|
|
}
|
|
}
|
|
return numberCount > max(len(token)/4, 1)
|
|
}
|
|
|
|
// defaultIsDataHeuristic is the default implementation of the IsDataHeuristic
|
|
// interface. It checks if the token is likely data based on a set of heuristics.
|
|
// This is the second main point of customization for the DRAIN algorithm, and is
|
|
// largely domain-specific.
|
|
func defaultIsDataHeuristic(token string) bool {
|
|
return hexWithAtLeastOneDigit(token) || significantNumbers(token)
|
|
}
|