Luigi Toscano ha scritto: > Dag ha scritto: >> Dag skrev den 2020-03-26 09:57: >> >>> I have adjusted the structure so now xml2po extracts all texts, at least. >>> >>> Haven't tested if xml2po works. >>> >>> Also noted there are problems with keycode/keycap, but let's see if my >>> changes is enough to generate translated docs. >> >> After further investigation it seems xml2po cannot handle docs that are split >> into several files (despite man page indicates so). > >> Docbook operates by including files effectively creating one document out of >> all the files, while xml2po operates on each file individually. This means >> that entities etc will not be available to xml2po as you cannot have >> <!DOCTYPE >> tag in included files. > > We do have documents split into multiple files. See for example kstars and > krusader. Please follow their structure. > > >> >> A see two ways to fix this: >> 1) Make xml2po understand included files. >> 2) Have the whole document in one file. >> >> A don't think 1) will happen soon, so I will go with 2). > > I suggest to revert the change 2), or it will break all existing translations > and it will make the work on it very complicated. Again, we have examples of > documentation spanning multiple files. >
In fact, the last changes worked and all strings seem to be extracted: https://websvn.kde.org/?view=revision&revision=1565268 The Ukrainian documentation is complete already (thanks Yuri) and it can be correctly generated. I'm going to revert the last change which moves all the files together. -- Luigi
