byext.py

#! /usr/bin/env python

"""Show file statistics by extension."""

from __future__ importprint_function

importos
importsys

classStats:

def__init__(self):
self.stats= {}

defstatargs(self, args):
forarginargs:
ifos.path.isdir(arg):
self.statdir(arg)
elifos.path.isfile(arg):
self.statfile(arg)
else:
sys.stderr.write("Can't find %s\n"%arg)
self.addstats("<???>", "unknown", 1)

defstatdir(self, dir):
self.addstats("<dir>", "dirs", 1)
try:
names=sorted(os.listdir(dir))
exceptos.erroraserr:
sys.stderr.write("Can't list %s: %s\n"% (dir, err))
self.addstats("<dir>", "unlistable", 1)
return
fornameinnames:
ifname.startswith(".#"):
continue# Skip CVS temp files
ifname.endswith("~"):
continue# Skip Emacs backup files
full=os.path.join(dir, name)
ifos.path.islink(full):
self.addstats("<lnk>", "links", 1)
elifos.path.isdir(full):
self.statdir(full)
else:
self.statfile(full)

defstatfile(self, filename):
head, ext=os.path.splitext(filename)
head, base=os.path.split(filename)
ifext==base:
ext=""# E.g. .cvsignore is deemed not to have an extension
ext=os.path.normcase(ext)
ifnotext:
ext="<none>"
self.addstats(ext, "files", 1)
try:
f=open(filename, "rb")
exceptIOErroraserr:
sys.stderr.write("Can't open %s: %s\n"% (filename, err))
self.addstats(ext, "unopenable", 1)
return
data=f.read()
f.close()
self.addstats(ext, "bytes", len(data))
ifb'\0'indata:
self.addstats(ext, "binary", 1)
return
ifnotdata:
self.addstats(ext, "empty", 1)
#self.addstats(ext, "chars", len(data))
lines=data.splitlines()
self.addstats(ext, "lines", len(lines))
dellines
words=data.split()
self.addstats(ext, "words", len(words))

defaddstats(self, ext, key, n):
d=self.stats.setdefault(ext, {})
d[key] =d.get(key, 0) +n

defreport(self):
exts=sorted(self.stats.keys())
# Get the column keys
columns= {}
forextinexts:
columns.update(self.stats[ext])
cols=sorted(columns.keys())
colwidth= {}
colwidth["ext"] =max([len(ext) forextinexts])
minwidth=6
self.stats["TOTAL"] = {}
forcolincols:
total=0
cw=max(minwidth, len(col))
forextinexts:
value=self.stats[ext].get(col)
ifvalueisNone:
w=0
else:
w=len("%d"%value)
total+=value
cw=max(cw, w)
cw=max(cw, len(str(total)))
colwidth[col] =cw
self.stats["TOTAL"][col] =total
exts.append("TOTAL")
forextinexts:
self.stats[ext]["ext"] =ext
cols.insert(0, "ext")
defprintheader():
forcolincols:
print("%*s"% (colwidth[col], col), end=" ")
print()
printheader()
forextinexts:
forcolincols:
value=self.stats[ext].get(col, "")
print("%*s"% (colwidth[col], value), end=" ")
print()
printheader() # Another header at the bottom

defmain():
args=sys.argv[1:]
ifnotargs:
args= [os.curdir]
s=Stats()
s.statargs(args)
s.report()

if__name__=="__main__":
main()