feat/SSM-34: introduce lpc parser sanitizer & validator

service-app/internal/parser/lpc_parser/lpc_parser.go

@@ -0,0 +1,11 @@
+package lpc_parser
+
+import (
+	"github.com/ministryofjustice/opg-scanning/internal/parser"
+	"github.com/ministryofjustice/opg-scanning/internal/types/lpc_types"
+)
+
+func Parse(data []byte) (interface{}, error) {
+	doc := &lpc_types.LPCDocument{}
+	return parser.DocumentParser(data, doc)
+}

service-app/internal/parser/lpc_parser/lpc_sanitizer.go

@@ -0,0 +1,38 @@
+package lpc_parser
+
+import (
+	"fmt"
+
+	"github.com/ministryofjustice/opg-scanning/internal/parser"
+	"github.com/ministryofjustice/opg-scanning/internal/types/lpc_types"
+)
+
+type Sanitizer struct {
+	doc           *lpc_types.LPCDocument
+	baseSanitizer *parser.BaseSanitizer
+}
+
+func NewSanitizer() *Sanitizer {
+	return &Sanitizer{
+		doc: &lpc_types.LPCDocument{},
+	}
+}
+
+func (v *Sanitizer) Setup(doc interface{}) error {
+	if doc == nil {
+		return fmt.Errorf("document is nil")
+	}
+
+	v.doc = doc.(*lpc_types.LPCDocument)
+	v.baseSanitizer = parser.NewBaseSanitizer(v.doc)
+
+	return nil
+}
+
+func (s *Sanitizer) Sanitize() (interface{}, error) {
+	if err := s.baseSanitizer.SanitizeStruct(s.doc); err != nil {
+		return nil, err
+	}
+
+	return s.doc, nil
+}

service-app/internal/parser/lpc_parser/lpc_validator.go

@@ -0,0 +1,47 @@
+package lpc_parser
+
+import (
+	"fmt"
+
+	"github.com/ministryofjustice/opg-scanning/internal/parser"
+	"github.com/ministryofjustice/opg-scanning/internal/types/lpc_types"
+)
+
+type Validator struct {
+	doc           *lpc_types.LPCDocument
+	baseValidator *parser.BaseValidator
+}
+
+func NewValidator() *Validator {
+	return &Validator{
+		doc: &lpc_types.LPCDocument{},
+	}
+}
+
+func (v *Validator) Setup(doc interface{}) error {
+	if doc == nil {
+		return fmt.Errorf("document is nil")
+	}
+
+	v.doc = doc.(*lpc_types.LPCDocument)
+	v.baseValidator = parser.NewBaseValidator(v.doc)
+
+	return nil
+}
+
+func (v *Validator) Validate() error {
+	// Common witness validations
+	v.baseValidator.WitnessSignatureFullNameAddressValidator("Page10", "Section9")
+
+	// Section validations
+	v.baseValidator.ValidateSignatureDate("Page11", "Section10", "")
+
+	// TODO add more validation for LPC
+
+	// Return errors if any
+	if messages := v.baseValidator.GetValidatorErrorMessages(); len(messages) > 0 {
+		return fmt.Errorf("failed to validate LP1H document: %v", messages)
+	}
+
+	return nil
+}

service-app/internal/parser/lpc_parser/lpc_validator_test.go

@@ -0,0 +1,71 @@
+package lpc_parser
+
+import (
+	"regexp"
+	"testing"
+
+	"github.com/ministryofjustice/opg-scanning/internal/parser"
+	"github.com/ministryofjustice/opg-scanning/internal/util"
+	"github.com/stretchr/testify/require"
+)
+
+func TestValidLPCXML(t *testing.T) {
+	validator := getLPCValidator(t, "LPC-valid.xml")
+	err := validator.Validate()
+	require.NoError(t, err, "Expected no errors for valid LPC XML")
+}
+
+func TestInvalidLPCXML(t *testing.T) {
+	validator := getLPCValidator(t, "LPC-invalid.xml")
+	err := validator.Validate()
+	require.Error(t, err, "Expected validation errors for invalid LPC XML, but got none")
+
+	messages := validator.(*Validator).baseValidator.GetValidatorErrorMessages()
+
+	expectedErrMsgs := []string{
+		"(?i)found only 1 attorney block; 2 required",     // minOccurs=2 but only 1 present
+		"(?i)missing Donor element in ContinuationSheet1", // required <Donor> is missing
+		"(?i)invalid element BURNN instead of BURN",       // if there is a typo in the element name
+		"(?i)missing PhysicalPage element",                // required element not found
+	}
+
+	t.Log("Actual messages from validation:")
+	t.Log(messages)
+
+	for _, pattern := range expectedErrMsgs {
+		regex, compErr := regexp.Compile(pattern)
+		require.NoError(t, compErr, "Failed to compile regex for pattern: %s", pattern)
+
+		found := false
+		for _, msg := range messages {
+			if regex.MatchString(msg) {
+				found = true
+				break
+			}
+		}
+
+		require.True(t, found, "Expected error message pattern not found: %s", pattern)
+	}
+}
+
+func TestInvalidLPCDateOrderXML(t *testing.T) {
+	validator := getLPCValidator(t, "LPC-invalid-dates.xml")
+	err := validator.Validate()
+	require.Error(t, err, "Expected validation errors due to date ordering but got none")
+
+	messages := validator.(*Validator).baseValidator.GetValidatorErrorMessages()
+
+	found := util.Contains(messages, "all form dates must be before the earliest applicant signature date")
+	require.True(t, found, "Expected date ordering validation error not found in messages")
+}
+
+func getLPCValidator(t *testing.T, fileName string) parser.CommonValidator {
+	xml := util.LoadXMLFileTesting(t, "../../../xml/"+fileName)
+
+	doc, err := Parse(xml)
+	require.NoError(t, err, "Failed to parse %s", fileName)
+
+	validator := NewValidator()
+	validator.Setup(doc)
+	return validator
+}

service-app/xml/LPC-invalid.xml

@@ -0,0 +1,31 @@
+<LPC xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+     xsi:noNamespaceSchemaLocation="example.xsd">
+    <Page1>
+        <ContinuationSheet1>
+            <!-- Only ONE Attorney here, which violates minOccurs="2" -->
+            <Attorney>
+                <Attorney>true</Attorney>
+                <ReplacementAttorney>false</ReplacementAttorney>
+                <PersonToNotify>false</PersonToNotify>
+                <Title>Mr</Title>
+                <FirstName>Harry</FirstName>
+                <LastName>Potter</LastName>
+                <DOB>1980-07-31</DOB>
+                <Address>
+                    <Address1>4 Privet Drive</Address1>
+                    <Address2>Little Whinging</Address2>
+                    <Address3>Surrey</Address3>
+                    <Postcode>HP1 1HP</Postcode>
+                </Address>
+                <Email>[email protected]</Email>
+            </Attorney>
+
+            <!-- Donor is missing entirely. The XSD requires <Donor> here. -->
+        </ContinuationSheet1>
+
+        <!-- BURN is present but spelled incorrectly: "BURNN" (typo) -->
+        <BURNN>Typo in element name</BURNN>
+
+        <!-- Missing <PhysicalPage> altogether -->
+    </Page1>
+</LPC>

service-app/xml/LPC-valid.xml

@@ -0,0 +1,48 @@
+<LPC xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+     xsi:noNamespaceSchemaLocation="example.xsd">
+    <Page1>
+        <ContinuationSheet1>
+            <Attorney>
+                <Attorney>true</Attorney>
+                <ReplacementAttorney>false</ReplacementAttorney>
+                <PersonToNotify>false</PersonToNotify>
+                <Title>Mr</Title>
+                <FirstName>John</FirstName>
+                <LastName>Doe</LastName>
+                <DOB>1970-01-01</DOB>
+                <Address>
+                    <Address1>1 Example Road</Address1>
+                    <Address2>Example Town</Address2>
+                    <Address3>Example County</Address3>
+                    <Postcode>EX4 MPL</Postcode>
+                </Address>
+                <Email>[email protected]</Email>
+            </Attorney>
+            <Attorney>
+                <Attorney>false</Attorney>
+                <ReplacementAttorney>true</ReplacementAttorney>
+                <PersonToNotify>false</PersonToNotify>
+                <Title>Mrs</Title>
+                <FirstName>Mary</FirstName>
+                <LastName>Smith</LastName>
+                <DOB>1975-05-05</DOB>
+                <Address>
+                    <Address1>2 Example Avenue</Address1>
+                    <Address2>Example Town</Address2>
+                    <Address3>Example County</Address3>
+                    <Postcode>EX4 MPL</Postcode>
+                </Address>
+                <Email>[email protected]</Email>
+            </Attorney>
+
+            <!-- Required Donor block -->
+            <Donor>
+                <FullName>Jane Donor</FullName>
+                <Signature>true</Signature>
+                <Date>2025-01-29</Date>
+            </Donor>
+        </ContinuationSheet1>
+        <BURN>Some text here</BURN>
+        <PhysicalPage>1</PhysicalPage>
+    </Page1>
+</LPC>