Shared strings table not found in recent XLSX file

[jeff-emanuel]

I'm using version v0.10.4, and I get the error shown below reading data from xlsx file written from the Excel for Mac version 16.103.1 (25111624). I see in code for sst_load! that it checks for the shared string table by (line 76-77 in v0.10.4 and line 109-110 in the current 9a01450 master):

    relationship_type = "http://schemas.openxmlformats.org/officeDocument/2006/relationships/sharedStrings"
    if has_relationship_by_type(workbook, relationship_type)

But my xlsx file has relationship type "http://purl.oclc.org/ooxml/officeDocument/relationships/sharedStrings" instead.

ERROR: Shared Strings Table not found for this workbook.
Stacktrace:
[1] error(s::String)
@ Base ./error.jl:35
[2] sst_load!(workbook::XLSX.Workbook)
@ XLSX ~/.julia/packages/XLSX/vzApD/src/sst.jl:95
[3] sst_unformatted_string
@ ~/.julia/packages/XLSX/vzApD/src/sst.jl:132 [inlined]
[4] sst_unformatted_string
@ ~/.julia/packages/XLSX/vzApD/src/sst.jl:143 [inlined]
[5] sst_unformatted_string
@ ~/.julia/packages/XLSX/vzApD/src/sst.jl:144 [inlined]
[6] sst_unformatted_string
@ ~/.julia/packages/XLSX/vzApD/src/sst.jl:145 [inlined]
[7] getdata(ws::XLSX.Worksheet, cell::XLSX.Cell)
@ XLSX ~/.julia/packages/XLSX/vzApD/src/cell.jl:181
[8] getdata
@ ~/.julia/packages/XLSX/vzApD/src/worksheet.jl:91 [inlined]
[9] getdata(xl::XLSX.XLSXFile, ref::XLSX.SheetCellRef)
@ XLSX ~/.julia/packages/XLSX/vzApD/src/workbook.jl:109
[10] getdata(xl::XLSX.XLSXFile, s::String)
@ XLSX ~/.julia/packages/XLSX/vzApD/src/workbook.jl:124
[11] top-level scope
@ REPL[124]:1

[TimG1964]

Here is what ChatGPT says about this:

http://schemas.openxmlformats.org/officeDocument/2006/relationships/styles is the Transitional namespace defined by ECMA-376 and used in most Office documents for backward compatibility, while http://purl.oclc.org/ooxml/officeDocument/relationships/styles is the Strict namespace defined by ISO/IEC 29500. Both point to the same kind of part (styles.xml), but they belong to different conformance classes of OOXML.

📘 Background on OOXML Namespaces
Transitional OOXML (2006 schemas):

URI: http://schemas.openxmlformats.org/...

Originates from the ECMA-376 standard (2006).

Allows legacy constructs for compatibility with older Office versions.

Widely used in Microsoft Office by default because it ensures older documents remain valid.

Strict OOXML (purl.oclc.org URIs):

URI: http://purl.oclc.org/ooxml/...

Defined in ISO/IEC 29500 (the international standard).

Removes deprecated or legacy constructs, enforcing stricter compliance.

Intended for long-term archival and interoperability across different implementations.

Less commonly used in practice, since Office defaults to Transitional mode unless explicitly saved as Strict.

🔗 Styles Relationship in Both
Transitional: http://schemas.openxmlformats.org/officeDocument/2006/relationships/styles → Used in .xlsx files saved in the default (Transitional) format.

Strict: http://purl.oclc.org/ooxml/officeDocument/relationships/styles → Used in .xlsx files saved in the Strict OOXML format.

Both URIs identify the relationship to the styles part (styles.xml), but the namespace signals which conformance class the document follows.

⚡ Why the Difference Matters
Compatibility: Transitional ensures older Office versions can open files. Strict may break compatibility with older software that doesn’t support ISO namespaces.

Validation: Strict documents validate against ISO schemas, while Transitional documents validate against ECMA schemas.

Archival: Strict is preferred for long-term preservation because it avoids deprecated constructs.

✅ Summary
The two URIs are not functionally different in what they point to (the styles part), but they signal different OOXML conformance classes:

schemas.openxmlformats.org → Transitional (ECMA-376, backward-compatible)

purl.oclc.org/ooxml → Strict (ISO/IEC 29500, standards-compliant)

The Readme for XLSX.jl says

This package follows the EMCA-376 to parse and generate XLSX files

I don't know how much work would be involved to make XLSX.jl support the ISO Schemas or if it is viable for it to support both. Are you able to switch Excel locally to use EMCA-376, or is this a new Mac implementation? I have not seen it in my Windows version of Excel (365 - latest).

⚡ Extra note: Microsoft Excel by default saves in Transitional format unless you explicitly choose “Strict Open XML Spreadsheet (.xlsx)”* in the Save As dialog.

[TimG1964]

A very superficial exploration suggests it may not be too hard to read a "Strict" xlsx file into XLSX.jl by converting in to a "transitional" file on the fly as it is read. Conversion seems mainly to involve changing the XML namespace references and changing the way dates are read. There may be more to this (there usually is!), but it is perhaps not huge.

This may be worth exploring after v0.11.0 is released. However, I'm uncertain of the demand. If anyone has a view either way, it would be good to know.

In the mean time, the obvious work-around seems to be to open Strict files in Excel and then save them again in the default format.

[jeff-emanuel]

Thanks a lot for the research. I'll check whether I can save the spreadsheet as non-strict later today. It didn't notice an option about that when I saved it, only that the .xlsx choice did say strict. The company IT forces many softwares to be at the latest version, so reverting is impossible. If I can't save it as non-strict, I'll investigate changing the namespaces.

[jeff-emanuel]

I can save the spreadsheet as non-strict. When I do that, the shared strings table is found and I can get the data from cells with shared strings. So this issue and 329 can be considered to be enhancement requests. I'm still experiencing problems with #330.