validate gives a error.table.bad_field_read error

[msbentley]

🐛 Describe the bug

validate 2.4.0 reading a browse product which previously passed validation gives this error:

  FAIL: file:/Users/mbentley/Downloads/TEST_V1/bc_mpo_mag/browse_calibrated/near_earth_commissioning/sc/ob/scf/mag_cal_sc_browse_ob_s2_scf_00000_20181024.xml
      ERROR  [error.table.bad_field_read]   line 0: Cannot invoke "gov.nasa.arc.pds.xml.generated.ObjectLength.getValue()" because the return value of "gov.nasa.arc.pds.xml.generated.EncodedByteStream.getObjectLength()" is null

📜 To Reproduce

Steps to reproduce the behavior:

1. validate the attached product

🕵️ Expected behavior

Product passes validation

📚 Version of Software Used

validate 2.4.0

🩺 Test Data / Additional context

mag_cal_sc_browse_example.zip

🏞Screenshots

🖥 System Info

• OS: macOS
• Browser N/A
• Version 12.5.1

---

🦄 Related requirements

⚙️ Engineering Details

[msbentley]

I also get the same results with validate 3.0.0

[jimmie]

PR out for review and approval - needed for validate release.

[nutjob4life]

Well, even with the newly-released, non-SNAPSHOT pds4-jparser version 2.3.0, this error still occurs.

But I'm out of time for today so I'll have to investigate tomorrow.

[nutjob4life]

Okay, here's an (interesting?) observation. If I add the indicated line to @msbentley's mag_cal_sc_browse_ob_s4_scf_00000_20181024.xml file …

<Encoded_Image>
    <offset unit="byte">0</offset>
    <object_length unit="byte">12345</object_length> <!-- 👈 this line -->
    <encoding_standard_id>PNG</encoding_standard_id>
</Encoded_Image>

… then validate works just fine.

So I'm suspecting that <object_length> is somehow being marked as "required" when it isn't—or, possibly, it is now required 🤷‍♀️

[nutjob4life]

In the file Label.java at line 666–669, we have:

    if (stream instanceof EncodedByteStream) {
      EncodedByteStream ebs = (EncodedByteStream) stream;
      size = ebs.getObjectLength().getValue().longValueExact(); // 👈 There's your problem
      offset = ebs.getOffset().getValue().longValueExact();

The getObjectLength() is returning null, so we can't call getValue() on it.

However, this code hasn't changed in … well, in a while! So the assumption it's made that it would never be null was incorrect for some time.

Looking at the src/build/resources/schema files, though, I don't see any place that has <object_length> as required (except in <Parsable_Byte_Stream>s where minOccurs="1").

So something else has changed. Maybe the "object length" had a default value before?

[miguelp1986]

Test results: https://cae-testrail.jpl.nasa.gov/testrail/index.php?/tests/view/4212017