4
List python source files.
6
There are three functions to check whether a file is a Python source, listed
7
here with increasing complexity:
9
- has_python_ext() checks whether a file name ends in '.py[w]'.
10
- look_like_python() checks whether the file is not binary and either has
11
the '.py[w]' extension or the first line contains the word 'python'.
12
- can_be_compiled() checks whether the file can be compiled by compile().
14
The file also must be of appropriate size - not bigger than a megabyte.
16
walk_python_files() recursively lists all Python files under the given directories.
18
__author__ = "Oleg Broytmann, Georg Brandl"
20
__all__ = ["has_python_ext", "looks_like_python", "can_be_compiled", "walk_python_files"]
25
binary_re = re.compile('[\x00-\x08\x0E-\x1F\x7F]')
35
size = os.stat(fullpath).st_size
36
except OSError as err: # Permission denied - ignore the file
37
print_debug("%s: permission denied: %s" % (fullpath, err))
40
if size > 1024*1024: # too big
41
print_debug("%s: the file is too big: %d bytes" % (fullpath, size))
45
return open(fullpath, 'rU')
46
except IOError as err: # Access denied, or a special file - ignore it
47
print_debug("%s: access denied: %s" % (fullpath, err))
50
def has_python_ext(fullpath):
51
return fullpath.endswith(".py") or fullpath.endswith(".pyw")
53
def looks_like_python(fullpath):
54
infile = _open(fullpath)
58
line = infile.readline()
61
if binary_re.search(line):
62
# file appears to be binary
63
print_debug("%s: appears to be binary" % fullpath)
66
if fullpath.endswith(".py") or fullpath.endswith(".pyw"):
68
elif "python" in line:
69
# disguised Python script (e.g. CGI)
74
def can_be_compiled(fullpath):
75
infile = _open(fullpath)
83
compile(code, fullpath, "exec")
84
except Exception as err:
85
print_debug("%s: cannot compile: %s" % (fullpath, err))
91
def walk_python_files(paths, is_python=looks_like_python, exclude_dirs=None):
93
Recursively yield all Python source files below the given paths.
95
paths: a list of files and/or directories to be checked.
96
is_python: a function that takes a file name and checks whether it is a
98
exclude_dirs: a list of directory base names that should be excluded in
101
if exclude_dirs is None:
105
print_debug("testing: %s" % path)
106
if os.path.isfile(path):
109
elif os.path.isdir(path):
110
print_debug(" it is a directory")
111
for dirpath, dirnames, filenames in os.walk(path):
112
for exclude in exclude_dirs:
113
if exclude in dirnames:
114
dirnames.remove(exclude)
115
for filename in filenames:
116
fullpath = os.path.join(dirpath, filename)
117
print_debug("testing: %s" % fullpath)
118
if is_python(fullpath):
121
print_debug(" unknown type")
124
if __name__ == "__main__":
125
# Two simple examples/tests
126
for fullpath in walk_python_files(['.']):
129
for fullpath in walk_python_files(['.'], is_python=can_be_compiled):