ultisuite-backend/internal/contacts/discovery/signature.go

package discovery

import (
	"html"
	"regexp"
	"strings"
)

var (
	sigDelimiterRe    = regexp.MustCompile(`(?m)^--\s*$`)
	forwardedHeaderRe = regexp.MustCompile(`(?i)(?:^|\n)(?:-{5,}\s*)?(?:forwarded message|message transféré|message transmis|-----Original Message-----)`)
	quotedReplyStartRe = regexp.MustCompile(`(?is)(?:^|\n)\s*(?:` +
		`Le\s+.{4,200}?\s+a\s+écrit\s*:` +
		`|On\s+.{4,200}?\s+wrote\s*:` +
		`|Am\s+.{4,200}?\s+schrieb\s*:` +
		`|El\s+.{4,200}?\s+escribió\s*:` +
		`|Il\s+.{4,200}?\s+ha\s+scritto\s*:` +
		`|-----Original Message-----` +
		`|\n_{5,}` +
		`)`)
	replyHeaderBlockRe = regexp.MustCompile(`(?is)\n\s*(?:De|From)\s*:\s*.+\n\s*(?:Envoyé|Sent|Date)\s*:`)
	replyHeaderInSigRe = regexp.MustCompile(`(?i)(?:\bwrote\s*:|\ba\s+écrit\s*:|-----Original|Envoyé\s*:|Sent\s*:|schrieb\s*:|escribió\s*:|ha\s+scritto\s*:)`)
	phoneInSigRe       = regexp.MustCompile(`(?:\+?\d[\d\s().-]{7,}\d)`)
	emailInSigRe       = regexp.MustCompile(`(?i)[a-z0-9._%+\-]+@[a-z0-9.\-]+\.[a-z]{2,}`)
	titleKeywordsRe    = regexp.MustCompile(`(?i)(?:directeur|director|manager|ceo|cto|engineer|consultant|developer|president|founder|co-founder|chef|responsable)`)
)

func stripHTMLTags(s string) string {
	if len(s) > 8000 {
		s = s[len(s)-8000:]
	}
	s = regexp.MustCompile(`(?is)<style[^>]*>.*?</style>`).ReplaceAllString(s, "")
	s = regexp.MustCompile(`(?is)<script[^>]*>.*?</script>`).ReplaceAllString(s, "")
	s = regexp.MustCompile(`(?i)<br\s*/?>`).ReplaceAllString(s, "\n")
	s = regexp.MustCompile(`(?i)</p>`).ReplaceAllString(s, "\n")
	s = regexp.MustCompile(`<[^>]+>`).ReplaceAllString(s, "")
	return html.UnescapeString(s)
}

func stripQuotedPrefixLines(raw string) string {
	lines := strings.Split(raw, "\n")
	cut := len(lines)
	for i, line := range lines {
		if i < 2 {
			continue
		}
		trimmed := strings.TrimSpace(line)
		if strings.HasPrefix(trimmed, ">") {
			cut = i
			break
		}
	}
	if cut < len(lines) {
		raw = strings.TrimSpace(strings.Join(lines[:cut], "\n"))
	}
	return raw
}

func stripQuotedReplyContent(raw string) string {
	if loc := forwardedHeaderRe.FindStringIndex(raw); loc != nil {
		raw = raw[:loc[0]]
	}
	if loc := quotedReplyStartRe.FindStringIndex(raw); loc != nil && loc[0] >= 15 {
		raw = raw[:loc[0]]
	}
	if loc := replyHeaderBlockRe.FindStringIndex(raw); loc != nil && loc[0] >= 15 {
		raw = raw[:loc[0]]
	}
	return stripQuotedPrefixLines(raw)
}

func emailsInText(s string) []string {
	found := emailInSigRe.FindAllString(strings.ToLower(s), -1)
	if len(found) == 0 {
		return nil
	}
	seen := map[string]struct{}{}
	var out []string
	for _, e := range found {
		e = strings.TrimSpace(e)
		if e == "" {
			continue
		}
		if _, ok := seen[e]; ok {
			continue
		}
		seen[e] = struct{}{}
		out = append(out, e)
	}
	return out
}

func signatureMatchesSender(candidate, senderEmail, displayName string) bool {
	if replyHeaderInSigRe.MatchString(candidate) {
		return false
	}

	sender := strings.ToLower(strings.TrimSpace(senderEmail))
	if sender == "" {
		return true
	}

	for _, e := range emailsInText(candidate) {
		if e != sender {
			return false
		}
	}

	if displayName != "" {
		nameTokens := strings.Fields(strings.ToLower(displayName))
		if len(nameTokens) >= 2 {
			candidateLower := strings.ToLower(candidate)
			matches := 0
			for _, tok := range nameTokens {
				if len(tok) < 2 {
					continue
				}
				if strings.Contains(candidateLower, tok) {
					matches++
				}
			}
			if matches == 0 && !phoneInSigRe.MatchString(candidate) && len(emailsInText(candidate)) == 0 {
				return false
			}
		}
	}

	return true
}

func extractSignature(bodyText, bodyHTML, senderEmail, displayName string) (text string, confidence float64) {
	raw := strings.TrimSpace(bodyText)
	if len(raw) > 8000 {
		raw = raw[len(raw)-8000:]
	}
	if raw == "" && bodyHTML != "" {
		raw = stripHTMLTags(bodyHTML)
	}
	if raw == "" {
		return "", 0
	}

	raw = stripQuotedReplyContent(raw)
	raw = strings.TrimSpace(raw)
	if raw == "" {
		return "", 0
	}

	parts := sigDelimiterRe.Split(raw, -1)
	candidate := strings.TrimSpace(raw)
	confidence = 0.3

	if len(parts) > 1 {
		candidate = strings.TrimSpace(parts[len(parts)-1])
		confidence = 0.7
	} else {
		lines := strings.Split(raw, "\n")
		if len(lines) > 4 {
			start := len(lines) - 8
			if start < 0 {
				start = 0
			}
			tail := strings.Join(lines[start:], "\n")
			if phoneInSigRe.MatchString(tail) || emailInSigRe.MatchString(tail) || titleKeywordsRe.MatchString(tail) {
				candidate = strings.TrimSpace(tail)
				confidence = 0.55
			}
		}
	}

	candidate = strings.TrimSpace(candidate)
	if len(candidate) < 10 || len(candidate) > 2000 {
		return "", 0
	}

	if !signatureMatchesSender(candidate, senderEmail, displayName) {
		return "", 0
	}

	senderLower := strings.ToLower(strings.TrimSpace(senderEmail))
	if senderLower != "" {
		local := strings.Split(senderLower, "@")[0]
		if local != "" && !strings.Contains(strings.ToLower(candidate), local) {
			if !emailInSigRe.MatchString(candidate) && !phoneInSigRe.MatchString(candidate) {
				confidence *= 0.6
			}
		}
	}

	if confidence < 0.35 {
		return "", 0
	}
	return candidate, confidence
}

func detectForwardedAddresses(bodyText, bodyHTML string) []string {
	raw := bodyText
	if raw == "" && bodyHTML != "" {
		raw = stripHTMLTags(bodyHTML)
	}
	if raw == "" {
		return nil
	}

	var out []string
	seen := map[string]struct{}{}

	fromLineRe := regexp.MustCompile(`(?im)^(?:from|de|expéditeur)\s*:\s*(?:[^<\n]*<)?([a-z0-9._%+\-]+@[a-z0-9.\-]+\.[a-z]{2,})>?`)
	for _, m := range fromLineRe.FindAllStringSubmatch(raw, -1) {
		if len(m) > 1 {
			email := strings.ToLower(strings.TrimSpace(m[1]))
			if _, ok := seen[email]; !ok && email != "" {
				seen[email] = struct{}{}
				out = append(out, email)
			}
		}
	}
	return out
}