Hi All,
I have recently written an R scanner and parser for the Obversive
project. I wrote it using re2c and lemon in order to make sure that it
doesn't require any external libraries. It currently only has 9
conflicts which I can't quite figure out how to resolve and I'm sure it
doesn't parse exactly correct. One thing I can't quite fix is the
ambiguous use of newlines in R. I know that R currently uses a "work
around" where there is an empty production called cr and this is used
to tell the scanner when to start ignoring newlines. I'll most likely
end up doing the same thing.
I'm sharing this with the R dev team to get some input into it's
correctness. I'm curious about general opinions, usefulness, etc. The
main purpose of this parser is to give users better feedback on R code
before they submit it to R for processing. We're not trying explicitly
to replicate R's parsing behavior, but more trying to enforce a
particular style. This is because we generate R code from GUI
interfaces, so the user isn't writing code, only reviewing it for
errors.
The main advantages that this parser has over the current R parser is
that it is damn fast and doesn't require external libraries. My own
informal tests on my machine shows that it can parse 25M of R code in
about 40 seconds, including error detection, reporting, recovery, and
logging of the entire parsing internal state. It is also designed to
have no memory leaks, which is admittedly really easy when all you do
is parse.
Anyway, send me your thoughts on this. Thanks.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: parser.y
Type: application/octet-stream
Size: 5650 bytes
Desc: not available
Url :
https://www.stat.math.ethz.ch/pipermail/r-devel/attachments/20030424/25fc7e79/parser.exe
-------------- next part --------------
A non-text attachment was scrubbed...
Name: scanner.re
Type: application/octet-stream
Size: 5143 bytes
Desc: not available
Url :
https://www.stat.math.ethz.ch/pipermail/r-devel/attachments/20030424/25fc7e79/scanner.exe
