diff options
author | Fred Drake <fdrake@acm.org> | 2000-02-03 15:26:16 (GMT) |
---|---|---|
committer | Fred Drake <fdrake@acm.org> | 2000-02-03 15:26:16 (GMT) |
commit | d89ffebd85e782f584e7d815b7567eb834c79450 (patch) | |
tree | 93108efe1fa2293dde56fc521899c18837fadfd7 /Lib/lib-old/cmp.py | |
parent | ff484183de25b477db56e7fe04f9b60f6558787a (diff) | |
download | cpython-d89ffebd85e782f584e7d815b7567eb834c79450.zip cpython-d89ffebd85e782f584e7d815b7567eb834c79450.tar.gz cpython-d89ffebd85e782f584e7d815b7567eb834c79450.tar.bz2 |
These modules are now declared obsolete.
Diffstat (limited to 'Lib/lib-old/cmp.py')
-rw-r--r-- | Lib/lib-old/cmp.py | 63 |
1 files changed, 63 insertions, 0 deletions
diff --git a/Lib/lib-old/cmp.py b/Lib/lib-old/cmp.py new file mode 100644 index 0000000..943d8bd --- /dev/null +++ b/Lib/lib-old/cmp.py @@ -0,0 +1,63 @@ +"""Efficiently compare files, boolean outcome only (equal / not equal). + +Tricks (used in this order): + - Files with identical type, size & mtime are assumed to be clones + - Files with different type or size cannot be identical + - We keep a cache of outcomes of earlier comparisons + - We don't fork a process to run 'cmp' but read the files ourselves +""" + +import os + +cache = {} + +def cmp(f1, f2, shallow=1): + """Compare two files, use the cache if possible. + Return 1 for identical files, 0 for different. + Raise exceptions if either file could not be statted, read, etc.""" + s1, s2 = sig(os.stat(f1)), sig(os.stat(f2)) + if s1[0] <> 8 or s2[0] <> 8: + # Either is a not a plain file -- always report as different + return 0 + if shallow and s1 == s2: + # type, size & mtime match -- report same + return 1 + if s1[:2] <> s2[:2]: # Types or sizes differ, don't bother + # types or sizes differ -- report different + return 0 + # same type and size -- look in the cache + key = (f1, f2) + try: + cs1, cs2, outcome = cache[key] + # cache hit + if s1 == cs1 and s2 == cs2: + # cached signatures match + return outcome + # stale cached signature(s) + except KeyError: + # cache miss + pass + # really compare + outcome = do_cmp(f1, f2) + cache[key] = s1, s2, outcome + return outcome + +def sig(st): + """Return signature (i.e., type, size, mtime) from raw stat data + 0-5: st_mode, st_ino, st_dev, st_nlink, st_uid, st_gid + 6-9: st_size, st_atime, st_mtime, st_ctime""" + type = st[0] / 4096 + size = st[6] + mtime = st[8] + return type, size, mtime + +def do_cmp(f1, f2): + """Compare two files, really.""" + bufsize = 8*1024 # Could be tuned + fp1 = open(f1, 'rb') + fp2 = open(f2, 'rb') + while 1: + b1 = fp1.read(bufsize) + b2 = fp2.read(bufsize) + if b1 <> b2: return 0 + if not b1: return 1 |