Problem with comments in XML files

I have an m-file that reads an XML file to get inputs. I'm running into a problem with comments in the XML file. Simply put, if I put a comment inside the tag but before the value, then I get an empty string for the value. Comments elsewhere don't cause problems

Here's a toy example of my m-file:

function userInputStruct = readToyXmlInput()

toyXmlInputFile = 'toyXmlInput.xml';

xDoc = xmlread(toyXmlInputFile);

userInput = xDoc.getElementsByTagName('userInput');

userInputStruct = struct();  % Initialize

thisUserInput = userInput.item(0);

thisInput_01 = thisUserInput.getElementsByTagName('input_01');
thisInput_01Element = thisInput_01.item(0);
temp01 = thisInput_01Element.getFirstChild.getData;
temp02 = string(temp01);
input_01 = strip(temp02);

userInputStruct.input_01 = input_01;

Here's a toy example of my XML file:

<toyXmlInput>
    <userInput>
        <!-- A comment here is fine. -->
        <input_01>
            Dummy input #1  <!-- A comment here is fine. -->
            <!-- A comment here is fine. -->
        </input_01>
    </userInput>
</toyXmlInput>

Here's a toy example of my XML file with a comment that causes a problem.

<toyXmlInput>
    <userInput>
        <input_01>
            <!-- A comment here causes problems -->
            Dummy input #1
        </input_01>
    </userInput>
</toyXmlInput>

What am I doing wrong? What can I do differently so that comments after the tag but before the value don't cause problems?

Worst-case, I'll just document in the m-file and the XML file that there's a limit on where comments can go. However, I'd prefer a more elegant solution.

Answer 1

[Stephen23]

The issue you're encountering is related to how XML Document Object Model (DOM) parsing handles different types of nodes within an element. When you use getFirstChild, you're getting the very first child node, which could be a text node, comment node, or element node. In your problematic case, the first child is a comment node, not the text content you want.

Here's what's happening:

1. In your working XML, the first child of <input_01> is a text node (even if it's just whitespace before "Dummy input #1")
2. In your problematic XML, the first child is a comment node, so getFirstChild.getData returns empty

XML Node Types:

• Node Type 1: Element node (like <input_01>)
• Node Type 3: Text node (the actual text content)
• Node Type 4: CDATA section
• Node Type 8: Comment node (what's causing your problem)

Solution:

Use getTextContent() - this is the simplest and most elegant solution. Replace your current line:

temp01 = thisInput_01Element.getFirstChild.getData;

With:

temp01 = thisInput_01Element.getTextContent;

The getTextContent() method automatically:

• Concatenates all text content within the element
• Ignores comments, processing instructions, and other non-text nodes
• Handles nested elements properly

The getTextContent() approach will work with both of your XML examples and is the standard way to extract text content from XML elements while ignoring comments.

[Bob]

Thanks! I obviously didn't know about getTextContent(). That method's doing exactly what I want and need.

Answer 2

[Steven Lord]

Do you need the comments to be imported? If not, and if upgrading to release R2020b or later is an option, consider using readstruct to read the contents of the XML file into a struct array. I've attached xmlfile1.txt and xmlfile2.txt as your two toy XML files. [I can't attach the files as .xml.] But I copy them into XML files with the first two lines.

copyfile xmlfile1.txt xmlfile1.xml
copyfile xmlfile2.txt xmlfile2.xml

The comments are not present in the data1 and data2 struct arrays, but the data is present.

data1 = readstruct('xmlfile1.xml')

data1 = struct with fields:
    userInput: [1×1 struct]

data1.userInput

ans = struct with fields:
    input_01: "Dummy input #1"

data2 = readstruct('xmlfile2.xml')

data2 = struct with fields:
    userInput: [1×1 struct]

data2.userInput

ans = struct with fields:
    input_01: "Dummy input #1"

[Bob]

No, I don't need to import the comments. The comments are basically to give users instructions about the inputs.

Updating to R2020b or later is coming but the time frame is out of my control. If and when that happens, I'll keep your suggestion of using readstruct.

Is there a man page or help document or something about reading XML files? All I could find were some old Mathworks blog posts. Those blog posts were what I was working from, and why I didn't know about getTextContent or the various XML node types.

[Stephen23]

https://www.mathworks.com/help/matlab/xml-documents.html

Use the Contents browser on the left hand side to navigate the documentation.

[Bob]

Thanks again!