[Spambayes-checkins] spambayes/spambayes message.py,1.17,1.18

[Skip Montanaro]

    Tim> Compiled the crlf regex, and added a couple comments

Tim,

I doubt compiling the regex helps.  Running timeit.py suggests this is the
case:

    % timeit.py -s "data = 'line 0\r\nline 1\nline 2\rline 3\r\r\nline 4\n' ; import re ; CRLF='\r\n'" "re.sub(r'\r\n|\n|\r', CRLF, data)"
    10000 loops, best of 3: 38.4 usec per loop
    % timeit.py -s "data = 'line 0\r\nline 1\nline 2\rline 3\r\r\nline 4\n' ; import re ; CRLF='\r\n' ; CRLFRE = re.compile(r'\r\n|\n|\r')" "re.sub(CRLFRE, CRLF, data)"
    10000 loops, best of 3: 38.5 usec per loop

I ran both several times.  The above times are the best I came up with.

In fact, looking at the code for sre.sub and sre._compile suggests that
using strings should be slightly faster than using compiled regular
expressions, because strings are cached with their compiled re's, whereas
compiled re's aren't.

Skip