CFeedParser is a feed parser (yeah!) written in C with Python bindings. It is heavily inspired by the excellent Universal Feed Parser and, in fact, share some code with it. Differences with Universal Feed Parser are :
- far less features :
CFeedParser only support the direct extraction of most used elements (title, link, id, summary, content and author). It does not extract other elements (like copyright, source, generator,...), nor it extract elements attributes (like language or content type), does not export metadata (like encoding or feed format), know nothing about fancy things like microformats, and does not “sanitize” the HTML content.
CFeedParser support all formats supported by Universal Feed Parser (I reused a great part of Universal Feed Parser unit tests), as long as they are well-formed.
- no support for illformed feeds :
CFeedParser does not try to fix feeds that are not real XML.
- fast :
CFeedParser is amazingly fast. If fact, that’s the reason it was written — Universal Feed Parser was too slow for one on my usage. For example, let’s try to parse all my feeds (35) with CParserFeed, then its Python bindings, and then with Universal Feed Parser.
$ time ./cfeedparser-test feeds/* > /dev/null 0,21s user 0,01s system 94% cpu 0,239 total
$ time ./cfeedparser.py feeds/* > /dev/null 0,76s user 0,05s system 95% cpu 0,844 total
$ time ./feedparser.py feeds/* > /dev/null 8,42s user 0,04s system 99% cpu 8,525 total
Yes, the Python bindings are 10 times faster than Universal Feed Parser, and CFeedParser 40 times faster !
libxml-2.0 and glib-2.0. That’s all.
The Python bindings are designed to be a subset of Universal Feed Parser. There is three ways to use them:
- Standalone: you have the same features and performances than CFeedParser.
- With Universal Feed Parser date parsing and encoding detection
capabilites: you have reduced performances (~ 2 times slower), but the
dates are parsed, and it support more encodings. To enable this mode,
just put fp_date.py and fp_encoding.py in the same directory that
cfeedparser.py and then use
cfeedparser.parsefunction. - With Universal Feed Parser: in addition with the preceding mode, you can add Universal Feed Parser in your path, and all ill-formed feeds will automatically be handled by Universal Feed Parser (you have an overhead only for ill-formed feeds)
Python 3 is supported, but only in the standalone mode.
With the go tool, you can just add
import "github.com/sloonz/cfeedparser"
to your imports, and then use feedparser.ParseFile and feedparser.ParseURL