Auto-fix nested CDATA tags in the note content

encoding/enex/enex.go

@@ -7,6 +7,7 @@ import (
 	"fmt"
 	"io"
 	"regexp"
+	"strings"
 )
 
 type (
@@ -123,7 +124,13 @@ type StreamDecoder struct {
 }
 
 func NewStreamDecoder(r io.Reader) (*StreamDecoder, error) {
-	d := xml.NewDecoder(r)
+	buf := bytes.Buffer{}
+	if _, err := buf.ReadFrom(r); err != nil {
+		return nil, err
+	}
+	clean := removeNestedCDATA(buf.String())
+
+	d := xml.NewDecoder(strings.NewReader(clean))
 	d.Strict = false
 
 	for {
@@ -202,3 +209,26 @@ func decodeRecognition(n *Note) error {
 
 	return nil
 }
+
+var reCDATA = regexp.MustCompile(`<!\[CDATA\[(.*?)\]\]>`)
+
+// removeNestedCDATA tags in the note content
+//
+// Nested CDATA tags are not allowed by XML specification
+// but Evernote puts them anyway, causing "Unexpected EOF" errors during decoding
+func removeNestedCDATA(input string) string {
+	output := reCDATA.ReplaceAllStringFunc(input, func(match string) string {
+		submatch := reCDATA.FindStringSubmatch(match)
+		if len(submatch) > 1 {
+			return submatch[1]
+		}
+		return match
+	})
+
+	// Recursively remove nested CDATA tags
+	if output != input {
+		return removeNestedCDATA(output)
+	}
+
+	return output
+}

encoding/enex/enex_test.go

@@ -153,6 +153,22 @@ func TestStreamDecodeWrongFile(t *testing.T) {
 		t.Errorf("Expected error, got = %v", err)
 	}
 }
+func TestStreamDecodeAutofixCDATA(t *testing.T) {
+	enexContent, err := os.Open("testdata/cdata.issue.enex")
+	if err != nil {
+		t.Fatal(err)
+	}
+	d, err := enex.NewStreamDecoder(enexContent)
+	if err != nil {
+
+		t.Errorf("Error while Decoding = %v", err)
+	}
+	var got enex.Note
+	err = d.Next(&got)
+	if err != nil {
+		t.Error(err)
+	}
+}
 
 func readFile(filename string) []byte {
 	file, err := os.ReadFile(filename)

encoding/enex/testdata/cdata.issue.enex

@@ -0,0 +1,30 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE en-export SYSTEM "http://xml.evernote.com/pub/evernote-export4.dtd">
+<en-export export-date="20230121T235625Z" application="Evernote" version="10.19.2">
+  <note>
+    <title>Test note with nested CDATA tags</title>
+    <created>20230121T235548Z</created>
+    <updated>20230121T235625Z</updated>
+    <tag>test data</tag>
+    <tag>data</tag>
+    <note-attributes>
+      <author>me</author>
+    </note-attributes>
+    <content>
+      <![CDATA[<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<!DOCTYPE en-note SYSTEM "http://xml.evernote.com/pub/enml2.dtd">
+<en-note>
+<div>email</div>
+<![CDATA[>]]>
+<div><b>Title</b><span><br/></span></div>
+<div>Date<br/></div>
+<div>text</div>]]
+<![CDATA[>]]>
+<div><br></div>
+<div lang="EN-US">
+</div>
+<div><br/></div>
+</en-note>      ]]>
+    </content>
+  </note>
+  </en-export>