wrong detection of Docx file as Zip if first entry is not `word/`

[piranna]

I have a Docx file that for some reason, first entry is [trash], so when looking for a word/ instead (or any other of the entries of OOXML files), it fails.`

Entries of the culprit file:

[trash]/0000.dat
[trash]/0002.dat
[trash]/0001.dat
word/document.xml
word/footnotes.xml
word/endnotes.xml
word/theme/theme1.xml
word/settings.xml
word/webSettings.xml
word/fontTable.xml
word/styles.xml
[trash]/0003.dat
docProps/app.xml
customXml/item1.xml
customXml/itemProps1.xml
customXml/item2.xml
customXml/itemProps2.xml
customXml/_rels/item1.xml.rels
[trash]/0004.dat
customXml/_rels/item2.xml.rels
word/_rels/document.xml.rels
customXml/item3.xml
customXml/itemProps3.xml
docProps/custom.xml
customXml/_rels/item3.xml.rels
_rels/.rels
docProps/core.xml
[Content_Types].xml

Actual value checked at ZippedDocumentBase.compare_bytes() for offset 0x1E:

bytearray(b'[tras')
bytearray(b'[tra')
bytearray(b'[tr')
bytearray(b'[trash]/0000.dat\xff\xff\xff')
bytearray(b'[trash]/000')
bytearray(b'[trash]/')

file` command detects the file as DocX correctly. Not sure if we should open the zip file an iterate on its entries, or a directa approach without openning it would be possible.