On 3/14/2014 2:51 PM, Peter Mawhorter wrote:
First of all, hi everyone, I'm new to this list.
Welcome.
I'm a grad student who's worked on and off with Python on various projects for 8ish years now. I recently wanted to construct a parser for another programing language in Python and was dissapointed that Python doesn't have a built-in module for building parsers, which seems like a common-enough task. There are plenty of different 3rd-party parsing libraries available, specialized in lots of different ways (see e.g., [1]). I happened to pick one that seemed suitable for my needs but didn't turn out to support the recursive structures that I needed to parse. Rather than pick a different one I just built my own parser generator module, and used that to build my parser: problem solved. It would have been much nicer if there were a fully-featured builtin parser generator module in Python, however, and the purpose of this email is to test the waters a bit: is this something that other people in the Python community would be interested in? I imagine the route to providing a built-in parser generator module would be to first canvass the community to figure out what third-party libraries they use, and then contact the developers of some of the top libraries to see if they'd be happy integrating as a built-in module. At that point someone would need to work to integrate the chosen third-party library as a built-in module (ideally with its developers).
I think the idea has been raised before, but I am not sure which list (this one, pydev, or python-ideas).
My first reaction, as a core developer, is that the stdlib is, if anything, too large. It is already not as well-maintained as we would like.
My second is that parser generation is an application, not a library. A parser generator is used by running it with an input specification, not by importing it and using specific functions and classes.
From what I've looked at PyParsing and PLY seem to be standout parser generators for Python, PyParsing has a bit more Pythonic syntax from what I've seen. One important issue would be speed though: an implementation mostly written in C for low-level parsing tasks would probably be much preferrable to one written in pure Python, since a builtin module should be geared towards efficiency, but I don't actually know exactly how that would work (I've both extended and embedded Python with/in C before, but I'm not sure how that kind of project relates to writing a built-in module in C).
Something written in Python can be run with any implementation of Python. Something written in C tends to be tied to CPython,
Sorry if this is a bit rambly, but I'm interested in feedback from the community on this idea: is a builtin parser generator module desirable? If so, would integrating PyParsing as a builtin module be a good solution? What 3rd-party parsing module do you think would serve best for this purpose?
[1] http://nedbatchelder.com/text/python-parsers.html
Perhaps something like this should be in the wiki, if not already. -- Terry Jan Reedy -- https://mail.python.org/mailman/listinfo/python-list
