gohorsejobs/backend/internal/utils/document_validator.go

package utils

import (
	"regexp"
	"strings"
)

// DocumentValidator provides flexible document validation for global use
type DocumentValidator struct {
	// Country code to use for validation (empty = accept all formats)
	CountryCode string
}

// NewDocumentValidator creates a new validator
func NewDocumentValidator(countryCode string) *DocumentValidator {
	return &DocumentValidator{CountryCode: strings.ToUpper(countryCode)}
}

// ValidationResult represents the result of document validation
type ValidationResult struct {
	Valid   bool
	Message string
	Clean   string // Cleaned document number
}

// ValidateDocument validates a document based on country
// For a global portal, this supports multiple formats
func (v *DocumentValidator) ValidateDocument(doc string, docType string) ValidationResult {
	// Remove all non-alphanumeric characters for cleaning
	clean := regexp.MustCompile(`[^a-zA-Z0-9]`).ReplaceAllString(doc, "")

	if clean == "" {
		return ValidationResult{Valid: true, Message: "Documento opcional não fornecido", Clean: ""}
	}

	switch v.CountryCode {
	case "BR":
		return v.validateBrazil(clean, docType)
	case "JP":
		return v.validateJapan(clean, docType)
	case "US":
		return v.validateUSA(clean, docType)
	default:
		// For global/unknown countries, accept any alphanumeric
		if len(clean) < 5 || len(clean) > 30 {
			return ValidationResult{Valid: false, Message: "Documento deve ter entre 5 e 30 caracteres", Clean: clean}
		}
		return ValidationResult{Valid: true, Message: "Documento aceito", Clean: clean}
	}
}

// validateBrazil validates Brazilian documents (CNPJ/CPF)
func (v *DocumentValidator) validateBrazil(doc string, docType string) ValidationResult {
	switch strings.ToUpper(docType) {
	case "CNPJ":
		if len(doc) != 14 {
			return ValidationResult{Valid: false, Message: "CNPJ deve ter 14 dígitos", Clean: doc}
		}
		if !validateCNPJ(doc) {
			return ValidationResult{Valid: false, Message: "CNPJ inválido", Clean: doc}
		}
		return ValidationResult{Valid: true, Message: "CNPJ válido", Clean: doc}
	case "CPF":
		if len(doc) != 11 {
			return ValidationResult{Valid: false, Message: "CPF deve ter 11 dígitos", Clean: doc}
		}
		if !validateCPF(doc) {
			return ValidationResult{Valid: false, Message: "CPF inválido", Clean: doc}
		}
		return ValidationResult{Valid: true, Message: "CPF válido", Clean: doc}
	default:
		// Unknown Brazilian document type, accept if reasonable length
		if len(doc) < 11 || len(doc) > 14 {
			return ValidationResult{Valid: false, Message: "Documento brasileiro deve ter entre 11 e 14 dígitos", Clean: doc}
		}
		return ValidationResult{Valid: true, Message: "Documento aceito", Clean: doc}
	}
}

// validateCNPJ validates Brazilian CNPJ using checksum algorithm
func validateCNPJ(cnpj string) bool {
	if len(cnpj) != 14 {
		return false
	}
	// Check for known invalid patterns
	if cnpj == "00000000000000" || cnpj == "11111111111111" || cnpj == "22222222222222" ||
		cnpj == "33333333333333" || cnpj == "44444444444444" || cnpj == "55555555555555" ||
		cnpj == "66666666666666" || cnpj == "77777777777777" || cnpj == "88888888888888" ||
		cnpj == "99999999999999" {
		return false
	}

	// Calculate first check digit
	weights1 := []int{5, 4, 3, 2, 9, 8, 7, 6, 5, 4, 3, 2}
	sum := 0
	for i, w := range weights1 {
		sum += int(cnpj[i]-'0') * w
	}
	remainder := sum % 11
	checkDigit1 := 0
	if remainder >= 2 {
		checkDigit1 = 11 - remainder
	}
	if int(cnpj[12]-'0') != checkDigit1 {
		return false
	}

	// Calculate second check digit
	weights2 := []int{6, 5, 4, 3, 2, 9, 8, 7, 6, 5, 4, 3, 2}
	sum = 0
	for i, w := range weights2 {
		sum += int(cnpj[i]-'0') * w
	}
	remainder = sum % 11
	checkDigit2 := 0
	if remainder >= 2 {
		checkDigit2 = 11 - remainder
	}
	return int(cnpj[13]-'0') == checkDigit2
}

// validateCPF validates Brazilian CPF using checksum algorithm
func validateCPF(cpf string) bool {
	if len(cpf) != 11 {
		return false
	}
	// Check for known invalid patterns
	if cpf == "00000000000" || cpf == "11111111111" || cpf == "22222222222" ||
		cpf == "33333333333" || cpf == "44444444444" || cpf == "55555555555" ||
		cpf == "66666666666" || cpf == "77777777777" || cpf == "88888888888" ||
		cpf == "99999999999" {
		return false
	}

	// Calculate first check digit
	sum := 0
	for i := 0; i < 9; i++ {
		sum += int(cpf[i]-'0') * (10 - i)
	}
	remainder := sum % 11
	checkDigit1 := 0
	if remainder >= 2 {
		checkDigit1 = 11 - remainder
	}
	if int(cpf[9]-'0') != checkDigit1 {
		return false
	}

	// Calculate second check digit
	sum = 0
	for i := 0; i < 10; i++ {
		sum += int(cpf[i]-'0') * (11 - i)
	}
	remainder = sum % 11
	checkDigit2 := 0
	if remainder >= 2 {
		checkDigit2 = 11 - remainder
	}
	return int(cpf[10]-'0') == checkDigit2
}

// validateJapan validates Japanese corporate numbers
func (v *DocumentValidator) validateJapan(doc string, docType string) ValidationResult {
	// Japanese Corporate Number (法人番号) is 13 digits
	if len(doc) == 13 {
		return ValidationResult{Valid: true, Message: "法人番号 válido", Clean: doc}
	}
	// Accept other formats loosely
	if len(doc) >= 5 && len(doc) <= 20 {
		return ValidationResult{Valid: true, Message: "Documento aceito", Clean: doc}
	}
	return ValidationResult{Valid: false, Message: "Documento japonês inválido", Clean: doc}
}

// validateUSA validates US documents (EIN)
func (v *DocumentValidator) validateUSA(doc string, docType string) ValidationResult {
	// EIN is 9 digits
	if strings.ToUpper(docType) == "EIN" {
		if len(doc) != 9 {
			return ValidationResult{Valid: false, Message: "EIN must be 9 digits", Clean: doc}
		}
		return ValidationResult{Valid: true, Message: "EIN válido", Clean: doc}
	}
	// Accept other formats loosely
	if len(doc) >= 5 && len(doc) <= 20 {
		return ValidationResult{Valid: true, Message: "Document accepted", Clean: doc}
	}
	return ValidationResult{Valid: false, Message: "Invalid US document", Clean: doc}
}