pdf metadata export

[devils_add]

Hi,
I just noticed, but when Calibre exports pdf files it disregards pdf conventions. For example in in author it will write "John Doe & Jane Doe" instead of John Doe; Jane Doe;
Also, it will duplicate entries from those that already exist in the pdf. This confuses me, because if Calibre can read those entries to import data, why does it need to create it's own copy of metadata entries in the file, adding extra weight to it.
In addition, why does it need to add custom field metadata to the file if custom filed is left blank.

[kovidgoyal]

calibre does not duplicate entries. It creates entries in the /Info dictionary and in the XMP metadata block, using the standard Dublin Core metadata namespace, because both are needed for different PDF reading apps.

As for author conventions, calibre uses it's own author conventions. There is no universal multi author convention for author metadata in PDF info keys. In XMP metadata multiple authors are represented using multiple an rdf sequence, which is the correct way to represent multiple entries in XMP.

I suggest you spend some time educating yourself about PDF metadata.

[devils_add]

Sorry, I was talking about "Dublin Core Properties", where separate authors are separated by ';' and not by '&', and the reason I asked was that Calibre was writing into those properties using '&' separator.

[kovidgoyal]

calibre does not write multiple authors separated by & it writes them using an rdf:Sequence. That is with multiple <dc:author> elements.

[BetterRed]

I suspect your you're both right. When it comes to metadata PDF's are like images, they have multiple metadata blocks, often containing similar data but not necessarily the same data (eg a jpeg can have at least three Comment 'properties' - all of them different)

I just converted an epub to PDF. First attachment is from exiftool (GUI), second is from Tracker XChange and the third is from Nitro Pro

Note the inconsistencies - I tend to set more credence by what exiftool 'says'

BR