Canonical path is not processed correctly ('/../ in path)

[c-suh]

Checked for duplicates

Yes - I've already checked

🐛 Describe the bug

When I ran Validate on the 2 submitted packages from https://github.com/NASA-PDS/operations/issues/373, it failed. Using the vo mawd package as an example, here is the error:

ERROR [error.table.field_value_data_type_mismatch] data object 1, record 11, field 2: Value does not match its data type 'ASCII_File_SpecificationName': The directory spec '/document/../catalog/' does not match the pattern '/?([A-Za-z0-9][A-Za-z0-9-][A-Za-z0-9]/?|[A-Za-z0-9][^-_]/?)'

Looking at the offending transfer manifest tab file, there are paths such as /document/../catalog/dataset.cat.

Looking at the sip tab file, this corresponds to https://pds-atmospheres.nmsu.edu/PDS/data/vo_3001/document/../catalog/dataset.cat.

🕵️ Expected behavior

I expected Deep Archive to massage these URLs into a format acceptable by PDS.

📜 To Reproduce

1. Run Validate v3.3.1 on the NSSDCA Delivery Package from https://github.com/NASA-PDS/operations/issues/373
2. Read the Validate report

🖥 Environment Info

?

📚 Version of Software Used

?

🩺 Test Data / Additional context

NSSDCA delivery package: https://github.com/NASA-PDS/operations/files/10960033/pdsatm_pack9_20230313.tar.gz

Validate reports:

• vo_irtm_v1.0_20230313-validate.txt
• vo_mawd_v1.0_20230313-validate.txt

🦄 Related requirements

No response

⚙️ Engineering Details

No response

[nutjob4life]

@c-suh @jordanpadams is it possible to get the original files from the Atmosphere node (as well as the command they ran) that produced the incorrect SIP? I'd like to first make sure I can reproduce this by running pds-deep-archive against those files and making sure I see /../ in the output.

[jordanpadams]

@nutjob4life if you mean the archive data itself, you can download that from here: https://pds-atmospheres.nmsu.edu/PDS/data/vo_3002/

[nutjob4life]

Thanks @jordanpadams.

FYI for my own reference, here's a command I came up with to mirror an HTTPD index listing and all its files:

wget \
    --cut-dirs=2 \
    --execute robots=off \
    --mirror \
    --no-check-certificate \
    --no-host-directories \
    --no-parent \
    --quiet \
    --recursive \
    --reject='index.html*' \
    --relative \
    --timestamping \
    https://pds-atmospheres.nmsu.edu/PDS/data/vo_3002/

[nutjob4life]

@c-suh @jordanpadams okay so in the source data as provided by the Atmospheres node in data/vo_3002/document/dataset.xml, there's this:

        <Document_Edition>
            <edition_name>ASCII Text</edition_name>
            <language>English</language>
            <files>1</files>
            <Document_File>
                <file_name>../catalog/dataset.cat</file_name> <!--  👈notice -->
                <local_identifier>vo_irtm_dataset_cat</local_identifier>
                <document_standard_id>7-Bit ASCII Text</document_standard_id>
            </Document_File>
        </Document_Edition>

Before I fix this, my first question is: is this legal in a PDS label?

[jordanpadams]

@nutjob4life checking with some folks but I am 95% certain that is invalid. Validate Tool does not catch it, but I don't think that means it should be allowed.

[tloubrieu-jpl]

@c-suh do you still have the package with which this error was originally raised ? for I&T @gxtchen would use that to validate that the bug is fixed. Thanks

[tloubrieu-jpl]

@nutjob4life ☝️

[nutjob4life]

@tloubrieu-jpl @c-suh https://jpl.slack.com/archives/GCGR1R3A4/p1680820436039979

[c-suh]

@tloubrieu-jpl there's a link in the description under "Test Data/Additional Context" to the package which showed this error. However, just in case they've replaced it, here is the one I had saved locally: pdsatm_pack9_20230313.tar.gz.