[Tutor] string.count in Windows vs UNIX

[Tim Peters]

[UNIX Guru]
> I've been dabbling with Python for a bit, and use the following
> script-excerpt to go through a large file checking for specific text. On
> UNIX it finds the correct number of occurances (6665 - double-checked
> with grep -e "Subject: Results:" mail.file | wc -l) but when run on
> Windows (2K/XP) it stops finding, consistently,  after 4195 occurances.
> ...

> mailfile = open('mail.file', 'r')

Use 'rb' instead.  Python makes the same distinction between text-mode and
binary-mode files as C makes, since Python file objects are just a thin
wrapper around C stdio streams (FILE*).  As a UNIX Guru <wink>, you're used
to systems where text- and binary-mode files act identically.  They don't on
Windows, and some non-printable characters in Windows text-mode files have
meta-meanings (chiefly that for first occurrence of chr(26) acts as an EOF
marker in files opened in text mode on Windows).