On Jun 4, 2011 22:19 "Niels Thykier" <niels@thykier.net> wrote: > [...] > Hey, > > Tracker bug for stuff like #398658. > > ~Niels > > [...] > As far as I can tell, writing a full (and correct) parser for bash is not going to be trivial. But we may be able improve the accuracy of our current ad-hoc parser by adding some "Text::ParseWords" to the mix. A fair simple perl snippet (attached) seems to be fairly accurate in its tokenization of private/refresh-debhelper-data. I have noticed two issues so far: * Contents of heredocs (<<EOF ... EOF) are tokenized and probably shouldn't be. * ";" are not split into a separate token if there is no space between it and the previous token. (e.g. "if [ ... ]; then") ~Niels
Attachment:
test.pl
Description: Perl program
Attachment:
refresh-debhelper-data.tok
Description: Binary data