Fido identifying some XLSX, PPTX, and DOCX as fido-fmt/{x}

[ross-spencer]

Dev Effort

1D

Description

Via @sromkey the MS-Office Open XML files in this Archivematica test data zip are being identified as fido-fmt/{x} in Fido:

• sample-files

ross-spencer@artefactual:~/git/artefactual-labs/am/src/archivematica-sampledata/SampleTransfers/OfficeDocsExtracted/objects$ fido *
FIDO v1.3.12 (formats-v94.xml, container-signature-20180920.xml, format_extensions.xml)"
OK,14,fido-fmt/189.ppt,"Microsoft Office Open XML - Powerpoint","Microsoft Office Open XML - Powerpoint",47215,"MS-OfficeOpenXML-samples/samplepptx.pptx","None","signature"
OK,10,fido-fmt/189.word,"Microsoft Office Open XML - Word","Microsoft Office Open XML - Word",14860,"MS-OfficeOpenXML-samples/sampledocx.docx","None","signature"
OK,11,fido-fmt/189.xl,"Microsoft Office Open XML - Excel","Microsoft Office Open XML - Excel",12050,"MS-OfficeOpenXML-samples/samplexlsx.xlsx","None","signature"
FIDO: Processed      9 files in 343.28 msec, 26 files/sec

If the fido-fmt{x} entries are removed as per here: #36 (comment) then the closest match seems to be generic OOXML:

ross-spencer@artefactual:~/Desktop/temp/ndsa/office-samples-and-skeletons/samples$ fido *
FIDO v1.3.12 (formats-v94.xml, container-signature-20180920.xml, format_extensions.xml)
OK,150,fmt/189,"Microsoft Office Open XML","Microsoft Office Open XML",14860,"sampledocx.docx","None","signature"
OK,8,fmt/189,"Microsoft Office Open XML","Microsoft Office Open XML",47215,"samplepptx.pptx","None","signature"
OK,9,fmt/189,"Microsoft Office Open XML","Microsoft Office Open XML",12050,"samplexlsx.xlsx","None","signature"
FIDO: Processed      3 files in 206.92 msec, 14 files/sec

Unfortunately the Skeleton Suite looks like it won't help debug here as the extracted samples (three per puid) all identify correctly.

• skeleton-suite-v94

I have extracted the samples and the skeleton files here for easy access.

• office-samples-and-skeletons.zip

NB. Also noted by Sarah is that Siegfried will identify the formats correctly:

ross-spencer@artefactual:~/git/artefactual-labs/am/src/archivematica-sampledata/SampleTransfers/OfficeDocsExtracted/objects$ sf *
---
siegfried   : 1.7.11
scandate    : 2019-02-24T12:22:11+01:00
signature   : default.sig
created     : 2019-02-16T11:10:03+01:00
identifiers : 
  - name    : 'pronom'
    details : 'DROID_SignatureFile_V94.xml; container-signature-20180917.xml'
---
filename : 'MS-OfficeOpenXML-samples/sampledocx.docx'
filesize : 14860
modified : 2007-08-14T23:29:00+02:00
errors   : 
matches  :
  - ns      : 'pronom'
    id      : 'fmt/412'
    format  : 'Microsoft Word for Windows'
    version : '2007 onwards'
    mime    : 'application/vnd.openxmlformats-officedocument.wordprocessingml.document'
    basis   : 'extension match docx; container name [Content_Types].xml with byte match at 460, 94 (signature 1/3)'
    warning : 
---
filename : 'MS-OfficeOpenXML-samples/samplepptx.pptx'
filesize : 47215
modified : 2007-08-14T23:51:16+02:00
errors   : 
matches  :
  - ns      : 'pronom'
    id      : 'fmt/215'
    format  : 'Microsoft Powerpoint for Windows'
    version : '2007 onwards'
    mime    : 'application/vnd.openxmlformats-officedocument.presentationml.presentation'
    basis   : 'extension match pptx; container name [Content_Types].xml with byte match at 2326, 96 (signature 1/3)'
    warning : 
---
filename : 'MS-OfficeOpenXML-samples/samplexlsx.xlsx'
filesize : 12050
modified : 2007-08-14T23:50:24+02:00
errors   : 
matches  :
  - ns      : 'pronom'
    id      : 'fmt/214'
    format  : 'Microsoft Excel for Windows'
    version : '2007 onwards'
    mime    : 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet'
    basis   : 'extension match xlsx; container name [Content_Types].xml with byte match at 676, 88 (signature 1/3)'
    warning : 
---

[ablwr]

If this is an error for these handful of files but not other files of its type, is the outcome better that FIDO should return the generic Microsoft OOXML with a standard PRONOM fmt/189 ID, rather than the custom fido-fmt ID?

Asking because I get the same results in master but don't necessarily have the bandwidth to fully investigate and change and test a larger solution for these Microsoft files, but I can remove the custom fido-fmts which will produce fmt/189 results (better for preservation..?)

[replaceafill]

@carlwilson I investigated this using commit 6211d66 of the rc/1.6 branch using signature versions FIDO v1.4.1 (formats-v97.xml, container-signature-20200121.xml, format_extensions.xml) and the office-samples-and-skeletons.zip file shared by Ross.

From what I can see for the files in the office-samples-and-skeletons/samples directory fido finds three signatures. For example for the samplexlsx.xlsx file in it the match_formats method initially gets a list similar to:

 [('x-fmt/263', 'ZIP format'), ('fmt/189', 'Microsoft Office Open XML'), ('fido-fmt/189.xl', 'Microsoft Office Open XML - Excel')]

Then the priority logic determines that ('fido-fmt/189.xl', 'Microsoft Office Open XML - Excel') from the format_extensions.xml file is the best match.

The difference with the files in the office-samples-and-skeletons/skeleton directory is that only one signature is found and that makes fido to detect the formats using the container signature file container-signature-20200121.xml instead. For example for the fmt-214-container-signature-id-2030.xlsx file in it the match_formats method gets a list similar to:

[('x-fmt/263', 'ZIP format')]

From it a container type ZIP is determined getting the format from the [Content_Types].xml file contained in the xlsx file.

Do you have any advice on how to proceed with this?