diff options
| -rw-r--r-- | .gitignore | 1 | ||||
| -rw-r--r-- | avfs.py | 110 | ||||
| -rwxr-xr-x | suspicious | 101 |
3 files changed, 114 insertions, 98 deletions
diff --git a/.gitignore b/.gitignore new file mode 100644 index 0000000..bee8a64 --- /dev/null +++ b/.gitignore @@ -0,0 +1 @@ +__pycache__ @@ -0,0 +1,110 @@ +#!/usr/bin/env python3 + +""" +Traversing directories and archives with AVFS/fuse +""" + +import os +import os.path +import subprocess + +avfs_sys_mount = os.environ['HOME'] + '/.avfs' +avfs_started = False + +def sys_name(fpath): + global avfs_started + if not avfs_started: + subprocess.check_call(['mountavfs']) + avfs_started = True + return avfs_sys_mount + os.path.abspath(fpath) + +def exists(fpath): + return os.path.exists(sys_name(fpath)) + +def isdir(fpath): + return os.path.isdir(sys_name(fpath)) + +orig_open = open +def open(fpath, *pargs, **kwargs): + return orig_open(sys_name(fpath), *pargs, **kwargs) + +# AVFS has its own automatic view selection using file extensions, but it +# includes plugins (like #patch) that will lead us into an infinite loop +# if we try to do a directory traversal. Also, there are a few +# extensions we want to add. + +avfscmds = { + ('.gz', '#ugz'), + ('.tgz', '#ugz#utar'), + ('.tar.bz2', '#ubz2#utar'), + ('.bz2', '#ubz2'), + ('.bz', '#ubz2'), + ('.tbz2', '#ubz2#utar'), + ('.tbz', '#ubz2#utar'), + ('.Z', '#uz'), + ('.tpz', '#uz#utar'), + ('.tz', '#uz#utar'), + ('.taz', '#uz#utar'), + ('.a', '#uar'), + ('.deb', '#uar'), + ('.tar', '#utar'), + ('.gem', '#utar'), # Add upstream + ('.rar', '#urar'), + ('.sfx', '#urar'), + ('.zip', '#uzip'), + ('.jar', '#uzip'), + ('.ear', '#uzip'), + ('.war', '#uzip'), + ('.nupkg', '#uzip'), # Add upstream + ('.whl', '#uzip'), # Add upstream + ('.7z', '#u7z'), + ('.zoo', '#uzoo'), + ('.lha', '#ulha'), + ('.lhz', '#ulha'), + ('.arj', '#uarj'), + ('.cpio', '#ucpio'), + ('.rpm', '#rpm'), + ('.tar.xz', '#uxze#utar'), + ('.txz', '#uxze#utar'), + ('.xz', '#uxze'), + ('.lzma', '#uxze'), +} + +def guesscmd(filename): + for ext, cmd in avfscmds: + if filename.endswith(ext): + return cmd + guesscmd(filename[:-len(ext)]) + return '' + +def find(rootdir, prunedirs): + """ + Recursively list all files under rootdir, including files in archives + supported by AVFS. + """ + + sys_rootdir = sys_name(rootdir) + + for name in os.listdir(sys_rootdir): + path = rootdir + '/' + name + sys_path = sys_rootdir + '/' + name + + if os.path.isdir(sys_path): + if name not in prunedirs: + yield from find(path, prunedirs) + else: + cmd = guesscmd(name) + filtered_path = path + cmd + sys_filtered_path = sys_path + cmd + + if cmd and os.path.exists(sys_filtered_path): + if os.path.isdir(sys_filtered_path): + yield from find(filtered_path, prunedirs) + else: + yield filtered_path + else: + yield path + +if __name__ == "__main__": + import sys + for f in find(sys.argv[1], {'.git'}): + print(f) @@ -32,102 +32,7 @@ import sys import datetime import string import subprocess - -# AVFS stuff ------------------------------------------------------------------ - -avfs_sys_mount = os.environ['HOME'] + '/.avfs' -avfs_started = False -def avfs_sys_name(fpath): - if not avfs_started: - subprocess.check_call(['mountavfs']) - return avfs_sys_mount + os.path.abspath(fpath) - -def avfs_exists(fpath): - return os.path.exists(avfs_sys_name(fpath)) - -def avfs_isdir(fpath): - return os.path.isdir(avfs_sys_name(fpath)) - -def avfs_open(fpath, *pargs, **kwargs): - return open(avfs_sys_name(fpath), *pargs, **kwargs) - -# AVFS has its own automatic view selection using file extensions, but it -# includes plugins (like #patch) that will lead us into an infinite loop -# if we try to do a directory traversal. Also, there are a few -# extensions we want to add. - -avfscmds = { - ('.gz', '#ugz'), - ('.tgz', '#ugz#utar'), - ('.tar.bz2', '#ubz2#utar'), - ('.bz2', '#ubz2'), - ('.bz', '#ubz2'), - ('.tbz2', '#ubz2#utar'), - ('.tbz', '#ubz2#utar'), - ('.Z', '#uz'), - ('.tpz', '#uz#utar'), - ('.tz', '#uz#utar'), - ('.taz', '#uz#utar'), - ('.a', '#uar'), - ('.deb', '#uar'), - ('.tar', '#utar'), - ('.gem', '#utar'), # Add upstream - ('.rar', '#urar'), - ('.sfx', '#urar'), - ('.zip', '#uzip'), - ('.jar', '#uzip'), - ('.ear', '#uzip'), - ('.war', '#uzip'), - ('.nupkg', '#uzip'), # Add upstream - ('.whl', '#uzip'), # Add upstream - ('.7z', '#u7z'), - ('.zoo', '#uzoo'), - ('.lha', '#ulha'), - ('.lhz', '#ulha'), - ('.arj', '#uarj'), - ('.cpio', '#ucpio'), - ('.rpm', '#rpm'), - ('.tar.xz', '#uxze#utar'), - ('.txz', '#uxze#utar'), - ('.xz', '#uxze'), - ('.lzma', '#uxze'), -} - -def avfs_guesscmd(filename): - for ext, cmd in avfscmds: - if filename.endswith(ext): - return cmd + avfs_guesscmd(filename[:-len(ext)]) - return '' - -def avfs_find(rootdir, prunedirs): - """ - Recursively list all files under rootdir, including files in archives - supported by AVFS. - """ - - sys_rootdir = avfs_sys_name(rootdir) - - for name in os.listdir(sys_rootdir): - path = rootdir + '/' + name - sys_path = sys_rootdir + '/' + name - - if os.path.isdir(sys_path): - if name not in prunedirs: - yield from avfs_find(path, prunedirs) - else: - cmd = avfs_guesscmd(name) - filtered_path = path + cmd - sys_filtered_path = sys_path + cmd - - if cmd and os.path.exists(sys_filtered_path): - if os.path.isdir(sys_filtered_path): - yield from avfs_find(filtered_path, prunedirs) - else: - yield filtered_path - else: - yield path - -# ----------------------------------------------------------------------------- +import avfs report = {} wordscore = {} @@ -383,7 +288,7 @@ if options.displaysummary and options.summaryfile: prunedirs = {'CVS', '.git', '.bzr', '.hg', '.svn'} if len(args) > 0: for a in args: - filelist.extend(avfs_find(a, prunedirs)) + filelist.extend(avfs.find(a, prunedirs)) start = datetime.datetime.now() for file in filelist: @@ -391,7 +296,7 @@ for file in filelist: skipped += 1 continue try: - f = avfs_open(file) + f = avfs.open(file) except: print("failed to open: " + file) continue |