perf(prometheus): optimize duplicate metric splitting to O(N) complexity

[kongfei605]

No description provided.

[Copilot]

Pull request overview

This PR optimizes the Prometheus parser's metric splitting algorithm to reduce time complexity from O(N*M) to O(N) where N is the buffer size and M is the number of metric chunks. The optimization replaces repeated bytes.Index calls with a single-pass scan using bytes.IndexByte to quickly locate potential delimiters.

• Replaces dual bytes.Index calls per iteration with efficient bytes.IndexByte scanning
• Adds logic to skip current delimiter when positioned at one, ensuring forward progress
• Improves performance for large Prometheus metric payloads with many duplicate metrics

---

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

[Copilot]

The variable name 'p' conflicts with the Parser receiver variable in the method signature. While this is scoped within the for loop, it reduces readability and could be confusing. Consider using a more descriptive name like 'hashPos' or 'pos' instead.

Suggested change

	p := bytes.IndexByte(rest[scanOffset:], '#')
	if p == -1 {
	idx = -1
	break
	}
	// Potential delimiter found at relative index (scanOffset + p)
	candidate := rest[scanOffset+p:]
	// Verify if it is indeed a header
	if bytes.HasPrefix(candidate, helpHeader) || bytes.HasPrefix(candidate, typeHeader) {
	idx = scanOffset + p
	break
	}
	// False alarm (e.g. '#' inside a label or comment), continue scanning
	scanOffset += p + 1
	hashPos := bytes.IndexByte(rest[scanOffset:], '#')
	if hashPos == -1 {
	idx = -1
	break
	}
	// Potential delimiter found at relative index (scanOffset + hashPos)
	candidate := rest[scanOffset+hashPos:]
	// Verify if it is indeed a header
	if bytes.HasPrefix(candidate, helpHeader) || bytes.HasPrefix(candidate, typeHeader) {
	idx = scanOffset + hashPos
	break
	}
	// False alarm (e.g. '#' inside a label or comment), continue scanning
	scanOffset += hashPos + 1