Jean-Marc Lasgouttes wrote:
> Angus> * Copies any BibTeX databases to the temp directory, mangling
> Angus> their names in the process. This ensures that any "files with
> Angus> spaces" that cause BibTeX to fail can be processed
> Angus> successfully.
> I am OK with the principle, but the code looks a bit weird to me.
That's why I post it, so that some fresh eyes can tell me I'm making a mess
of things ;-)
> Why is this databases needed, whereas the entries could be generated on
> the fly?
What happens if there are no databases? Is "\bibliography{}" OK? Maybe it's
OK, but it's certainly lazy and ugly. Try 2:
Separator const separator(",");
Tokenizer const tokens(getContents(), separator);
Tokenizer::const_iterator const begin = tokens.begin();
Tokenizer::const_iterator const end = tokens.end();
ostringstream dbs;
for (Tokenizer::const_iterator it = begin; it != end; ++it) {
string database = ...
if (it != begin)
dbs << ',';
dbs << database;
}
string const dbs_str = dbs.str();
int nnewlines = 1;
if (!dbs_str.empty()) {
++nnewlines;
os << "\\bibliography{" << dbs_str << "}\n";
}
return nnewlines;
}
If you're happy with this, I'll make the change, commit it to the 1.3.x
tree and forward-port it to 1.4.x. AFAIK, that's all squashable bugs
squashed for 1.3.x. If you know different, please keep it to yourself ;-)
--
Angus
