Peter M. Groen / oletools

Commit aede664b7c2985a4e309a8f20209eb636c383dc2

Authored by Christian Herdtweck
1 parent f75482f1

make file existence check in xglob compatible with glob

Inline Side-by-side
Showing 1 changed file with 37 additions and 1 deletions
oletools/thirdparty/xglob/xglob.py
  View file @aede664
... ... @@ -131,7 +131,7 @@ def iter_files(files, recursive=False, zip_password=None, zip_fname='*'):
131 131 else:
132 132 iglob = glob.iglob
133 133 for filespec in files:
134   - if not os.path.exists(filespec):
  134 + if not is_glob(filespec) and not os.path.exists(filespec):
135 135 raise ValueError('given path {} does not exist!'.format(filespec))
136 136 for filename in iglob(filespec):
137 137 if zip_password is not None:
... ... @@ -155,3 +155,39 @@ def iter_files(files, recursive=False, zip_password=None, zip_fname='*'):
155 155 #data = open(filename, 'rb').read()
156 156 #yield None, filename, data
157 157  
  158 +
  159 +def is_glob(filespec):
  160 + """ determine if given file specification is a single file name or a glob
  161 +
  162 + python's glob and fnmatch can only interpret ?, *, [list], and [ra-nge],
  163 + (and combinations: hex_*_[A-Fabcdef0-9]).
  164 + The special chars *?[-] can only be escaped using []
  165 + --> file_name is not a glob
  166 + --> file?name is a glob
  167 + --> file* is a glob
  168 + --> file[-._]name is a glob
  169 + --> file[?]name is not a glob (matches literal "file?name")
  170 + --> file[*]name is not a glob (matches literal "file*name")
  171 + --> file[-]name is not a glob (matches literal "file-name")
  172 + --> file-name is not a glob
  173 +
  174 + Also, obviously incorrect globs are treated as non-globs
  175 + --> file[name is not a glob (matches literal "file[name")
  176 + --> file]-[name is treated as a glob
  177 + (it is not a valid glob but detecting errors like this requires
  178 + sophisticated regular expression matching)
  179 +
  180 + Python's glob also works with globs in directory-part of path
  181 + --> dir-part of path is analyzed just like filename-part
  182 + --> thirdparty/*/xglob.py is a (valid) glob
  183 +
  184 + TODO: create a correct regexp to test for validity of ranges
  185 + """
  186 +
  187 + # remove escaped special chars
  188 + cleaned = filespec.replace('[*]', '').replace('[?]', '') \
  189 + .replace('[[]', '').replace('[]]', '').replace('[-]', '')
  190 +
  191 + # check if special chars remain
  192 + return '*' in cleaned or '?' in cleaned or \
  193 + ('[' in cleaned and ']' in cleaned)
... ...