gadfly/cmd/gadfly/consolidate.go

package main

import (
	"fmt"
	"strings"
)

// verdict is the severity a specialist (or the whole review) lands on.
type verdict int

const (
	verdictUnknown  verdict = iota // couldn't parse; treated below Minor for display, not blocking
	verdictClean                   // "No material issues found"
	verdictMinor                   // "Minor issues"
	verdictBlocking                // "Blocking issues found"
)

func (v verdict) label() string {
	switch v {
	case verdictClean:
		return "No material issues found"
	case verdictMinor:
		return "Minor issues"
	case verdictBlocking:
		return "Blocking issues found"
	default:
		return "Reviewed"
	}
}

// parseVerdict extracts a specialist's self-reported verdict from its output.
// The base prompt tells each lens to lead with one of the three phrases.
func parseVerdict(out string) verdict {
	l := strings.ToLower(out)
	// Match the EARLIEST-appearing verdict phrase (the lead verdict), and match
	// leniently — models write "Blocking issues found", "**Blocking issues**",
	// "blocking issue", etc. (A strict "blocking issues found" check let a
	// gpt-oss section that said "**Blocking issues**" fall through to unknown,
	// so the overall header wrongly read "No material issues found".)
	best, bestIdx := verdictUnknown, -1
	for _, c := range []struct {
		phrase string
		v      verdict
	}{
		{"blocking issue", verdictBlocking},
		{"minor issue", verdictMinor},
		{"no material issue", verdictClean},
	} {
		if i := strings.Index(l, c.phrase); i >= 0 && (bestIdx == -1 || i < bestIdx) {
			best, bestIdx = c.v, i
		}
	}
	return best
}

// specialistResult pairs a specialist with its rendered review and verdict.
type specialistResult struct {
	spec    Specialist
	out     string
	verdict verdict
	errored bool // the review pass failed (timeout/model error) — not a clean result
}

// worstVerdict returns the most severe verdict across results. The optional
// "improvements" lens never escalates the overall verdict (it's advisory).
func worstVerdict(results []specialistResult) verdict {
	worst := verdictClean
	for _, r := range results {
		if r.spec.Name == "improvements" {
			continue
		}
		if r.verdict > worst {
			worst = r.verdict
		}
	}
	return worst
}

// renderConsolidated assembles the single comment body: an overall verdict line
// followed by one verbatim section per specialist. run.sh wraps this with the
// "🪰 Gadfly review — <model>" header and the advisory footer.
func renderConsolidated(results []specialistResult) string {
	errored := 0
	for _, r := range results {
		if r.errored {
			errored++
		}
	}

	headline := "Verdict: " + worstVerdict(results).label()
	if len(results) > 0 && errored == len(results) {
		// Every lens errored — do NOT report this as "clean".
		headline = "Review incomplete — all lenses errored"
	} else if errored > 0 {
		headline += fmt.Sprintf(" · ⚠️ %d/%d lens(es) errored", errored, len(results))
	}

	var b strings.Builder
	fmt.Fprintf(&b, "**%s** — %d reviewers: %s\n",
		headline, len(results), strings.Join(specialistNames(results), ", "))

	for _, r := range results {
		body := strings.TrimSpace(r.out)
		if body == "" {
			body = "_(no output)_"
		}
		summary := r.verdict.label()
		if r.errored {
			summary = "⚠️ could not complete"
		}
		fmt.Fprintf(&b, "\n<details><summary><b>%s</b> — %s</summary>\n\n%s\n\n</details>\n",
			r.spec.Title, summary, body)
	}
	return strings.TrimRight(b.String(), "\n")
}

func specialistNames(results []specialistResult) []string {
	names := make([]string, 0, len(results))
	for _, r := range results {
		names = append(names, r.spec.Name)
	}
	return names
}