where-simd-helps/analysis/pkg/parse/parse.go

// Package parse reads pqc-bench .out files produced by the SLURM harness.
//
// Each file contains a SLURM prolog header followed by 1–N "loop spin" blocks.
// Each spin block reports one median+average pair per benchmarked operation.
package parse

import (
	"bufio"
	"fmt"
	"os"
	"strconv"
	"strings"
)

// Meta holds the SLURM prolog metadata extracted from the file header.
type Meta struct {
	JobID        string
	JobName      string
	Node         string
	StartedAt    string
	Directory    string
	// Explicit fields emitted by submit.sh for reliable downstream parsing.
	BenchVariant string
	BenchParam   string
	BenchNSpins  string
}

// Measurement is a single operation's reported statistics for one loop spin.
type Measurement struct {
	Median  int64
	Average int64
}

// Run holds everything parsed from one .out file.
type Run struct {
	File string
	Meta Meta
	// Spins[i] maps operation name → measurement for loop spin i+1.
	Spins []map[string]Measurement
}

// ParseFile reads a single .out file and returns a Run.
func ParseFile(path string) (*Run, error) {
	f, err := os.Open(path)
	if err != nil {
		return nil, err
	}
	defer f.Close()

	run := &Run{File: path}
	scanner := bufio.NewScanner(f)
	// Default buffer size is 64KB; lines are short so this is fine.

	var currentSpin map[string]Measurement
	var currentOp string
	var pendingMedian int64
	inSpin := false

	for scanner.Scan() {
		line := strings.TrimSpace(scanner.Text())

		// SLURM prolog lines start with ##
		if strings.HasPrefix(line, "##") {
			parsePrologLine(line, &run.Meta)
			continue
		}

		// New loop spin
		if strings.HasPrefix(line, "Loop spin:") {
			if inSpin && currentSpin != nil {
				run.Spins = append(run.Spins, currentSpin)
			}
			currentSpin = make(map[string]Measurement)
			currentOp = ""
			inSpin = true
			continue
		}

		if !inSpin {
			continue
		}

		// Operation name line ends with ':'
		if strings.HasSuffix(line, ":") && !strings.HasPrefix(line, "median") && !strings.HasPrefix(line, "average") {
			currentOp = strings.TrimSuffix(line, ":")
			currentOp = strings.TrimSpace(currentOp)
			continue
		}

		if currentOp == "" {
			continue
		}

		if strings.HasPrefix(line, "median:") {
			v, err := parseCycles(line)
			if err != nil {
				return nil, fmt.Errorf("%s: %w", path, err)
			}
			pendingMedian = v
			continue
		}

		if strings.HasPrefix(line, "average:") {
			avg, err := parseCycles(line)
			if err != nil {
				return nil, fmt.Errorf("%s: %w", path, err)
			}
			currentSpin[currentOp] = Measurement{Median: pendingMedian, Average: avg}
			currentOp = ""
			pendingMedian = 0
			continue
		}
	}

	if err := scanner.Err(); err != nil {
		return nil, fmt.Errorf("%s: %w", path, err)
	}

	// Flush last spin
	if inSpin && currentSpin != nil {
		run.Spins = append(run.Spins, currentSpin)
	}

	return run, nil
}

// parseCycles extracts the integer from lines like "median: 25194 cycles/ticks".
func parseCycles(line string) (int64, error) {
	// Format: "<label>: <N> cycles/ticks"
	parts := strings.Fields(line)
	if len(parts) < 2 {
		return 0, fmt.Errorf("unexpected line format: %q", line)
	}
	return strconv.ParseInt(parts[1], 10, 64)
}

// parsePrologLine extracts key/value pairs from SLURM header lines.
func parsePrologLine(line string, meta *Meta) {
	// Lines look like: "##    Job ID : 11233228"
	// Strip leading "##" and optional decoration lines ("####...")
	trimmed := strings.TrimLeft(line, "#")
	trimmed = strings.TrimSpace(trimmed)
	key, val, ok := strings.Cut(trimmed, ":")
	if !ok {
		return
	}
	key = strings.TrimSpace(key)
	val = strings.TrimSpace(val)

	switch key {
	case "Job ID":
		meta.JobID = val
	case "Job Name":
		meta.JobName = val
	case "Nodelist":
		meta.Node = val
	case "Job Started":
		meta.StartedAt = val
	case "Directory":
		meta.Directory = val
	case "BENCH_VARIANT":
		meta.BenchVariant = val
	case "BENCH_PARAM":
		meta.BenchParam = val
	case "BENCH_NSPINS":
		meta.BenchNSpins = val
	}
}

// InferVariant returns the benchmark variant for a run.
//
// Priority:
//  1. Explicit BENCH_VARIANT metadata emitted by submit.sh (most reliable).
//  2. The path segment immediately following "kyber/" in the SLURM Directory
//     field (works for old-style runs that ran from inside the kyber tree).
//  3. "unknown" if neither is available.
func InferVariant(meta Meta) string {
	if meta.BenchVariant != "" {
		return meta.BenchVariant
	}
	const marker = "kyber/"
	idx := strings.LastIndex(meta.Directory, marker)
	if idx < 0 {
		return "unknown"
	}
	rest := meta.Directory[idx+len(marker):]
	variant, _, _ := strings.Cut(rest, "/")
	return variant
}