generated namespace prefix is invalid

[bertsky]

With 1.0.0b12, I now get documents from to_xml that begin like this:

<pc:PcGts xmlns="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15">

Thus, all elements use the pc prefix, but this is not declared in the root as xmlns:pc=....

IMO, this regression is a combination of c0c1ea6 and the earlier af5098d – first to_xml was rewritten to not include the pc prefix anymore (but at that time it was also not used in the generateDS call), then the generateDS call changed (but to_xml was not changed back accordingly).

[bertsky]

Oh, I just found your fix 22a4624

[bertsky]

(Could we separate this out of #268, and merge it right away?)

[n00blet]

@kba hi, can you please fix this ?

[kba]

Fixed in 1.0.0b15

[bertsky]

Thanks!

[dstoekl]

what is the purpose of the pc prefix? e.g. pc:pcgts? It does not validate against the schema.

[cneud]

what is the purpose of the pc prefix? e.g. pc:pcgts? It does not validate against the schema.

pc stands for page content within PAGE-XML afaict. The PAGE schema also has various other parts, e.g. relating to evaluation of binarization etc.

[dstoekl]

thx. but why would you prefix every single namespace with it? I haven't seen this on other pagexmls and it does not validate against the schema given.

[kba]

thx. but why would you prefix every single namespace with it? I haven't seen this on other pagexmls and it does not validate against the schema given.

Why would this not validate against the PAGE schema? Our PAGE-XML define xmlns:pc=xmlns="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15" so this should be valid. If we produce invalid XML, please open a new issue and share the setup that led to the faulty PAGE-XML.

[dstoekl]

ok. never mind.