Merge pull request #6 from viafintech/feature/extend-email-options

Extend email anonymizer to allow anonymizing domains
viafintech · Jul 7, 2023 · 1c483c0 · 1c483c0
2 parents 322c7db + 4a3a49e
commit 1c483c0
Show file tree

Hide file tree

Showing 2 changed files with 98 additions and 17 deletions.
diff --git a/postgres/email_anonymizer.go b/postgres/email_anonymizer.go
@@ -14,14 +14,41 @@ import (
 //
 // would be anonymized by a 15 character string based on the md5 hash of the local part.
 type EmailAnonymizer struct {
-  mailLocalPartLength int
+  mailAnonymizedPartLength int
+  mailAnonymizeDomainPart  bool
+}
+
+// EmailAnonymizerOption is the function type for passing options
+// to the EmailAnonymizer during initialization.
+type EmailAnonymizerOption func(*EmailAnonymizer)
+
+// EmailAnonymizedPartLengthOption allows setting the target length for the local part
+// of the anonymization rule
+func EmailAnonymizedPartLengthOption(length int) EmailAnonymizerOption {
+  return func(anonymizer *EmailAnonymizer) {
+    anonymizer.mailAnonymizedPartLength = length
+  }
+}
+
+// EmailAnonymizeDomainPartOption allows configuring whether the domain part should
+// also explicitly be anonymized
+func EmailAnonymizeDomainPartOption(anonymizeDomainPart bool) EmailAnonymizerOption {
+  return func(anonymizer *EmailAnonymizer) {
+    anonymizer.mailAnonymizeDomainPart = anonymizeDomainPart
+  }
 }
 
 // NewEmailAnonymizer initializes a new EmailAnonymizer object
-func NewEmailAnonymizer() *EmailAnonymizer {
-  return &EmailAnonymizer{
-    mailLocalPartLength: 15,
+func NewEmailAnonymizer(options ...EmailAnonymizerOption) *EmailAnonymizer {
+  anonymizer := &EmailAnonymizer{
+    mailAnonymizedPartLength: 15,
+  }
+
+  for _, option := range options {
+    option(anonymizer)
   }
+
+  return anonymizer
 }
 
 // Build returns the partial query holding the logic to overwrite
@@ -38,16 +65,28 @@ func (a *EmailAnonymizer) Build(tableName, columnName string) string {
           ) ||
           '@'::text
         ) ||
-        split_part(
-          (%[1]s)::text,
-          '@'::text,
-          2
-        )
+        %[3]s
       )
     )::CHARACTER VARYING
     ELSE %[1]s
     END`,
     gotidus.FullColumnName(tableName, columnName),
-    a.mailLocalPartLength,
+    a.mailAnonymizedPartLength,
+    a.domainPart(tableName, columnName),
+  )
+}
+
+func (a *EmailAnonymizer) domainPart(tableName, columnName string) string {
+  if a.mailAnonymizeDomainPart {
+    return fmt.Sprintf(
+      `("left"(md5(split_part((%[1]s)::text, '@'::text, 2)::text), %[2]d) || '.com')`,
+      gotidus.FullColumnName(tableName, columnName),
+      a.mailAnonymizedPartLength,
+    )
+  }
+
+  return fmt.Sprintf(
+    `split_part((%s)::text, '@'::text, 2)`,
+    gotidus.FullColumnName(tableName, columnName),
   )
 }
diff --git a/postgres/email_anonymizer_test.go b/postgres/email_anonymizer_test.go
@@ -11,7 +11,8 @@ func TestEmailAnonymizerBuild(t *testing.T) {
 	columnName := "bar"
 
 	cases := []struct {
-		title string
+		title   string
+		options []EmailAnonymizerOption
 
 		expectedString string
 	}{
@@ -28,11 +29,52 @@ func TestEmailAnonymizerBuild(t *testing.T) {
           ) ||
           '@'::text
         ) ||
-        split_part(
-          (foo.bar)::text,
-          '@'::text,
-          2
-        )
+        split_part((foo.bar)::text, '@'::text, 2)
+      )
+    )::CHARACTER VARYING
+    ELSE foo.bar
+    END`,
+		},
+		{
+			title: "email overlay with different length",
+			options: []EmailAnonymizerOption{
+				EmailAnonymizedPartLengthOption(10),
+			},
+
+			expectedString: `CASE WHEN ((foo.bar)::TEXT ~~ '%@%'::TEXT)
+    THEN (
+      (
+        (
+          "left"(
+            md5((foo.bar)::text),
+            10
+          ) ||
+          '@'::text
+        ) ||
+        split_part((foo.bar)::text, '@'::text, 2)
+      )
+    )::CHARACTER VARYING
+    ELSE foo.bar
+    END`,
+		},
+		{
+			title: "email overlay with anonymized domain part length",
+			options: []EmailAnonymizerOption{
+				EmailAnonymizedPartLengthOption(10),
+				EmailAnonymizeDomainPartOption(true),
+			},
+
+			expectedString: `CASE WHEN ((foo.bar)::TEXT ~~ '%@%'::TEXT)
+    THEN (
+      (
+        (
+          "left"(
+            md5((foo.bar)::text),
+            10
+          ) ||
+          '@'::text
+        ) ||
+        ("left"(md5(split_part((foo.bar)::text, '@'::text, 2)::text), 10) || '.com')
       )
     )::CHARACTER VARYING
     ELSE foo.bar
@@ -43,7 +85,7 @@ func TestEmailAnonymizerBuild(t *testing.T) {
 	for _, c := range cases {
 		t.Run(c.title, func(t *testing.T) {
 
-			anonymizer := NewEmailAnonymizer()
+			anonymizer := NewEmailAnonymizer(c.options...)
 
 			testutils.CompareStrings(
 				anonymizer.Build(tableName, columnName),