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

package discovery

import (
	"context"
	"encoding/json"
	"strings"
)

func (s *Service) inferMissingCompanies(ctx context.Context, externalUserID, ncUserID, bookID string) {
	domainCompanies := s.loadDomainCompanyHints(ctx, externalUserID, ncUserID, bookID)

	rows, err := s.db.Query(ctx, `
		SELECT p.id::text, p.primary_email, COALESCE(p.enriched_data, '{}'::jsonb)
		FROM contact_discovered_profiles p
		JOIN users u ON p.user_id = u.id
		WHERE u.external_id = $1
		  AND p.status = 'suggested'
		  AND NOT p.is_mailing_list
		  AND NOT p.is_disposable
		  AND NOT p.is_spam_heavy` + noReplyProfilesSQL + `
	`, externalUserID)
	if err != nil {
		return
	}
	defer rows.Close()

	for rows.Next() {
		var id, email string
		var enrichedJSON []byte
		if err := rows.Scan(&id, &email, &enrichedJSON); err != nil {
			continue
		}
		domain := emailDomain(email)
		if isConsumerMailDomain(domain) {
			continue
		}
		var data EnrichedContactData
		_ = json.Unmarshal(enrichedJSON, &data)
		if strings.TrimSpace(data.Company) != "" {
			continue
		}
		company, ok := domainCompanies[domain]
		if !ok || strings.TrimSpace(company) == "" {
			continue
		}
		data.Company = company
		payload, _ := json.Marshal(data)
		_, _ = s.db.Exec(ctx, `
			UPDATE contact_discovered_profiles
			SET enriched_data = $3::jsonb,
			    enrichment_status = CASE
			        WHEN enrichment_status IN ('skipped', 'failed', 'pending') THEN 'enriched'
			        ELSE enrichment_status
			    END,
			    updated_at = NOW()
			WHERE id = $1::uuid AND user_id = (SELECT id FROM users WHERE external_id = $2)
		`, id, externalUserID, string(payload))
	}
}

func (s *Service) loadDomainCompanyHints(ctx context.Context, externalUserID, ncUserID, bookID string) map[string]string {
	hints := map[string]int{}
	domainBest := map[string]string{}

	rows, err := s.db.Query(ctx, `
		SELECT lower(split_part(p.primary_email, '@', 2)) AS domain,
		       trim(both '"' from (p.enriched_data->>'company')) AS company,
		       COUNT(*)::int AS cnt
		FROM contact_discovered_profiles p
		JOIN users u ON p.user_id = u.id
		WHERE u.external_id = $1
		  AND p.enriched_data->>'company' IS NOT NULL
		  AND trim(p.enriched_data->>'company') != ''
		GROUP BY 1, 2
		HAVING COUNT(*) >= 2
	`, externalUserID)
	if err == nil {
		defer rows.Close()
		for rows.Next() {
			var domain, company string
			var cnt int
			if err := rows.Scan(&domain, &company, &cnt); err != nil {
				continue
			}
			if isConsumerMailDomain(domain) {
				continue
			}
			if cnt > hints[domain] {
				hints[domain] = cnt
				domainBest[domain] = company
			}
		}
	}

	if s.nc != nil && ncUserID != "" {
		contacts := s.loadNCContacts(ctx, ncUserID, bookID)
		orgByDomain := map[string]map[string]int{}
		for _, c := range contacts {
			org := strings.TrimSpace(c.Org)
			if org == "" {
				continue
			}
			domain := emailDomain(c.Email)
			if isConsumerMailDomain(domain) {
				continue
			}
			if orgByDomain[domain] == nil {
				orgByDomain[domain] = map[string]int{}
			}
			orgByDomain[domain][org]++
		}
		for domain, orgs := range orgByDomain {
			bestOrg := ""
			bestCnt := 0
			for org, cnt := range orgs {
				if cnt > bestCnt {
					bestCnt = cnt
					bestOrg = org
				}
			}
			if bestCnt < 2 {
				continue
			}
			batchOrg := domainBest[domain]
			batchCnt := hints[domain]
			if batchCnt >= 2 && !companyNamesMatch(batchOrg, bestOrg) {
				delete(domainBest, domain)
				continue
			}
			if batchCnt == 0 || bestCnt >= batchCnt {
				domainBest[domain] = bestOrg
			}
		}
	}

	return domainBest
}

func companyNamesMatch(a, b string) bool {
	a = normalizeCompanyName(a)
	b = normalizeCompanyName(b)
	if a == "" || b == "" {
		return false
	}
	return a == b
}

func normalizeCompanyName(s string) string {
	return strings.ToLower(strings.TrimSpace(s))
}