tools/scripts/spdxcheck.py

   1 #!/usr/bin/env python3
   2 # SPDX-License-Identifier: GPL-2.0
   3 # Copyright Thomas Gleixner <tglx@linutronix.de>
   4
   5 from argparse import ArgumentParser
   6 from ply import lex, yacc
   7 import locale
   8 import traceback
   9 import fnmatch
  10 import sys
  11 import git
  12 import re
  13 import os
  14
  15 class ParserException(Exception):
  16     def __init__(self, tok, txt):
  17         self.tok = tok
  18         self.txt = txt
  19
  20 class SPDXException(Exception):
  21     def __init__(self, el, txt):
  22         self.el = el
  23         self.txt = txt
  24
  25 class SPDXdata(object):
  26     def __init__(self):
  27         self.license_files = 0
  28         self.exception_files = 0
  29         self.licenses = [ ]
  30         self.exceptions = { }
  31
  32 class dirinfo(object):
  33     def __init__(self):
  34         self.missing = 0
  35         self.total = 0
  36         self.files = []
  37
  38     def update(self, fname, basedir, miss):
  39         self.total += 1
  40         self.missing += miss
  41         if miss:
  42             fname = './' + fname
  43             bdir = os.path.dirname(fname)
  44             if bdir == basedir.rstrip('/'):
  45                 self.files.append(fname)
  46
  47 # Read the spdx data from the LICENSES directory
  48 def read_spdxdata(repo):
  49
  50     # The subdirectories of LICENSES in the kernel source
  51     # Note: exceptions needs to be parsed as last directory.
  52     # OpenOCD specific: Begin
  53     license_dirs = [ "preferred", "stand-alone", "exceptions" ]
  54     # OpenOCD specific: End
  55     lictree = repo.head.commit.tree['LICENSES']
  56
  57     spdx = SPDXdata()
  58
  59     for d in license_dirs:
  60         for el in lictree[d].traverse():
  61             if not os.path.isfile(el.path):
  62                 continue
  63
  64             exception = None
  65             for l in open(el.path, encoding="utf-8").readlines():
  66                 if l.startswith('Valid-License-Identifier:'):
  67                     lid = l.split(':')[1].strip().upper()
  68                     if lid in spdx.licenses:
  69                         raise SPDXException(el, 'Duplicate License Identifier: %s' %lid)
  70                     else:
  71                         spdx.licenses.append(lid)
  72
  73                 elif l.startswith('SPDX-Exception-Identifier:'):
  74                     exception = l.split(':')[1].strip().upper()
  75                     spdx.exceptions[exception] = []
  76
  77                 elif l.startswith('SPDX-Licenses:'):
  78                     for lic in l.split(':')[1].upper().strip().replace(' ', '').replace('\t', '').split(','):
  79                         if not lic in spdx.licenses:
  80                             raise SPDXException(None, 'Exception %s missing license %s' %(exception, lic))
  81                         spdx.exceptions[exception].append(lic)
  82
  83                 elif l.startswith("License-Text:"):
  84                     if exception:
  85                         if not len(spdx.exceptions[exception]):
  86                             raise SPDXException(el, 'Exception %s is missing SPDX-Licenses' %exception)
  87                         spdx.exception_files += 1
  88                     else:
  89                         spdx.license_files += 1
  90                     break
  91     return spdx
  92
  93 class id_parser(object):
  94
  95     reserved = [ 'AND', 'OR', 'WITH' ]
  96     tokens = [ 'LPAR', 'RPAR', 'ID', 'EXC' ] + reserved
  97
  98     precedence = ( ('nonassoc', 'AND', 'OR'), )
  99
 100     t_ignore = ' \t'
 101
 102     def __init__(self, spdx):
 103         self.spdx = spdx
 104         self.lasttok = None
 105         self.lastid = None
 106         self.lexer = lex.lex(module = self, reflags = re.UNICODE)
 107         # Initialize the parser. No debug file and no parser rules stored on disk
 108         # The rules are small enough to be generated on the fly
 109         self.parser = yacc.yacc(module = self, write_tables = False, debug = False)
 110         self.lines_checked = 0
 111         self.checked = 0
 112         self.excluded = 0
 113         self.spdx_valid = 0
 114         self.spdx_errors = 0
 115         self.spdx_dirs = {}
 116         self.dirdepth = -1
 117         self.basedir = '.'
 118         self.curline = 0
 119         self.deepest = 0
 120
 121     def set_dirinfo(self, basedir, dirdepth):
 122         if dirdepth >= 0:
 123             self.basedir = basedir
 124             bdir = basedir.lstrip('./').rstrip('/')
 125             if bdir != '':
 126                 parts = bdir.split('/')
 127             else:
 128                 parts = []
 129             self.dirdepth = dirdepth + len(parts)
 130
 131     # Validate License and Exception IDs
 132     def validate(self, tok):
 133         id = tok.value.upper()
 134         if tok.type == 'ID':
 135             if not id in self.spdx.licenses:
 136                 raise ParserException(tok, 'Invalid License ID')
 137             self.lastid = id
 138         elif tok.type == 'EXC':
 139             if id not in self.spdx.exceptions:
 140                 raise ParserException(tok, 'Invalid Exception ID')
 141             if self.lastid not in self.spdx.exceptions[id]:
 142                 raise ParserException(tok, 'Exception not valid for license %s' %self.lastid)
 143             self.lastid = None
 144         elif tok.type != 'WITH':
 145             self.lastid = None
 146
 147     # Lexer functions
 148     def t_RPAR(self, tok):
 149         r'\)'
 150         self.lasttok = tok.type
 151         return tok
 152
 153     def t_LPAR(self, tok):
 154         r'\('
 155         self.lasttok = tok.type
 156         return tok
 157
 158     def t_ID(self, tok):
 159         r'[A-Za-z.0-9\-+]+'
 160
 161         if self.lasttok == 'EXC':
 162             print(tok)
 163             raise ParserException(tok, 'Missing parentheses')
 164
 165         tok.value = tok.value.strip()
 166         val = tok.value.upper()
 167
 168         if val in self.reserved:
 169             tok.type = val
 170         elif self.lasttok == 'WITH':
 171             tok.type = 'EXC'
 172
 173         self.lasttok = tok.type
 174         self.validate(tok)
 175         return tok
 176
 177     def t_error(self, tok):
 178         raise ParserException(tok, 'Invalid token')
 179
 180     def p_expr(self, p):
 181         '''expr : ID
 182                 | ID WITH EXC
 183                 | expr AND expr
 184                 | expr OR expr
 185                 | LPAR expr RPAR'''
 186         pass
 187
 188     def p_error(self, p):
 189         if not p:
 190             raise ParserException(None, 'Unfinished license expression')
 191         else:
 192             raise ParserException(p, 'Syntax error')
 193
 194     def parse(self, expr):
 195         self.lasttok = None
 196         self.lastid = None
 197         self.parser.parse(expr, lexer = self.lexer)
 198
 199     def parse_lines(self, fd, maxlines, fname):
 200         self.checked += 1
 201         self.curline = 0
 202         fail = 1
 203         try:
 204             for line in fd:
 205                 line = line.decode(locale.getpreferredencoding(False), errors='ignore')
 206                 self.curline += 1
 207                 if self.curline > maxlines:
 208                     break
 209                 self.lines_checked += 1
 210                 if line.find("SPDX-License-Identifier:") < 0:
 211                     continue
 212                 expr = line.split(':')[1].strip()
 213                 # Remove trailing comment closure
 214                 if line.strip().endswith('*/'):
 215                     expr = expr.rstrip('*/').strip()
 216                 # Remove trailing xml comment closure
 217                 if line.strip().endswith('-->'):
 218                     expr = expr.rstrip('-->').strip()
 219                 # Special case for SH magic boot code files
 220                 if line.startswith('LIST \"'):
 221                     expr = expr.rstrip('\"').strip()
 222                 self.parse(expr)
 223                 self.spdx_valid += 1
 224                 #
 225                 # Should we check for more SPDX ids in the same file and
 226                 # complain if there are any?
 227                 #
 228                 fail = 0
 229                 break
 230
 231         except ParserException as pe:
 232             if pe.tok:
 233                 col = line.find(expr) + pe.tok.lexpos
 234                 tok = pe.tok.value
 235                 sys.stdout.write('%s: %d:%d %s: %s\n' %(fname, self.curline, col, pe.txt, tok))
 236             else:
 237                 sys.stdout.write('%s: %d:0 %s\n' %(fname, self.curline, pe.txt))
 238             self.spdx_errors += 1
 239
 240         if fname == '-':
 241             return
 242
 243         base = os.path.dirname(fname)
 244         if self.dirdepth > 0:
 245             parts = base.split('/')
 246             i = 0
 247             base = '.'
 248             while i < self.dirdepth and i < len(parts) and len(parts[i]):
 249                 base += '/' + parts[i]
 250                 i += 1
 251         elif self.dirdepth == 0:
 252             base = self.basedir
 253         else:
 254             base = './' + base.rstrip('/')
 255         base += '/'
 256
 257         di = self.spdx_dirs.get(base, dirinfo())
 258         di.update(fname, base, fail)
 259         self.spdx_dirs[base] = di
 260
 261 class pattern(object):
 262     def __init__(self, line):
 263         self.pattern = line
 264         self.match = self.match_file
 265         if line == '.*':
 266             self.match = self.match_dot
 267         elif line.endswith('/'):
 268             self.pattern = line[:-1]
 269             self.match = self.match_dir
 270         elif line.startswith('/'):
 271             self.pattern = line[1:]
 272             self.match = self.match_fn
 273
 274     def match_dot(self, fpath):
 275         return os.path.basename(fpath).startswith('.')
 276
 277     def match_file(self, fpath):
 278         return os.path.basename(fpath) == self.pattern
 279
 280     def match_fn(self, fpath):
 281         return fnmatch.fnmatchcase(fpath, self.pattern)
 282
 283     def match_dir(self, fpath):
 284         if self.match_fn(os.path.dirname(fpath)):
 285             return True
 286         return fpath.startswith(self.pattern)
 287
 288 def exclude_file(fpath):
 289     for rule in exclude_rules:
 290         if rule.match(fpath):
 291             return True
 292     return False
 293
 294 def scan_git_tree(tree, basedir, dirdepth):
 295     parser.set_dirinfo(basedir, dirdepth)
 296     for el in tree.traverse():
 297         if not os.path.isfile(el.path):
 298             continue
 299         if exclude_file(el.path):
 300             parser.excluded += 1
 301             continue
 302         with open(el.path, 'rb') as fd:
 303             parser.parse_lines(fd, args.maxlines, el.path)
 304
 305 def scan_git_subtree(tree, path, dirdepth):
 306     for p in path.strip('/').split('/'):
 307         tree = tree[p]
 308     scan_git_tree(tree, path.strip('/'), dirdepth)
 309
 310 def read_exclude_file(fname):
 311     rules = []
 312     if not fname:
 313         return rules
 314     with open(fname) as fd:
 315         for line in fd:
 316             line = line.strip()
 317             if line.startswith('#'):
 318                 continue
 319             if not len(line):
 320                 continue
 321             rules.append(pattern(line))
 322     return rules
 323
 324 if __name__ == '__main__':
 325
 326     ap = ArgumentParser(description='SPDX expression checker')
 327     ap.add_argument('path', nargs='*', help='Check path or file. If not given full git tree scan. For stdin use "-"')
 328     ap.add_argument('-d', '--dirs', action='store_true',
 329                     help='Show [sub]directory statistics.')
 330     ap.add_argument('-D', '--depth', type=int, default=-1,
 331                     help='Directory depth for -d statistics. Default: unlimited')
 332     ap.add_argument('-e', '--exclude',
 333                     help='File containing file patterns to exclude. Default: scripts/spdxexclude')
 334     ap.add_argument('-f', '--files', action='store_true',
 335                     help='Show files without SPDX.')
 336     ap.add_argument('-m', '--maxlines', type=int, default=15,
 337                     help='Maximum number of lines to scan in a file. Default 15')
 338     ap.add_argument('-v', '--verbose', action='store_true', help='Verbose statistics output')
 339     args = ap.parse_args()
 340
 341     # Sanity check path arguments
 342     if '-' in args.path and len(args.path) > 1:
 343         sys.stderr.write('stdin input "-" must be the only path argument\n')
 344         sys.exit(1)
 345
 346     try:
 347         # Use git to get the valid license expressions
 348         repo = git.Repo(os.getcwd())
 349         assert not repo.bare
 350
 351         # Initialize SPDX data
 352         spdx = read_spdxdata(repo)
 353
 354         # Initialize the parser
 355         parser = id_parser(spdx)
 356
 357     except SPDXException as se:
 358         if se.el:
 359             sys.stderr.write('%s: %s\n' %(se.el.path, se.txt))
 360         else:
 361             sys.stderr.write('%s\n' %se.txt)
 362         sys.exit(1)
 363
 364     except Exception as ex:
 365         sys.stderr.write('FAIL: %s\n' %ex)
 366         sys.stderr.write('%s\n' %traceback.format_exc())
 367         sys.exit(1)
 368
 369     try:
 370         fname = args.exclude
 371         if not fname:
 372             fname = os.path.join(os.path.dirname(__file__), 'spdxexclude')
 373         exclude_rules = read_exclude_file(fname)
 374     except Exception as ex:
 375         sys.stderr.write('FAIL: Reading exclude file %s: %s\n' %(fname, ex))
 376         sys.exit(1)
 377
 378     try:
 379         if len(args.path) and args.path[0] == '-':
 380             stdin = os.fdopen(sys.stdin.fileno(), 'rb')
 381             parser.parse_lines(stdin, args.maxlines, '-')
 382         else:
 383             if args.path:
 384                 for p in args.path:
 385                     if os.path.isfile(p):
 386                         parser.parse_lines(open(p, 'rb'), args.maxlines, p)
 387                     elif os.path.isdir(p):
 388                         scan_git_subtree(repo.head.reference.commit.tree, p,
 389                                          args.depth)
 390                     else:
 391                         sys.stderr.write('path %s does not exist\n' %p)
 392                         sys.exit(1)
 393             else:
 394                 # Full git tree scan
 395                 scan_git_tree(repo.head.commit.tree, '.', args.depth)
 396
 397             ndirs = len(parser.spdx_dirs)
 398             dirsok = 0
 399             if ndirs:
 400                 for di in parser.spdx_dirs.values():
 401                     if not di.missing:
 402                         dirsok += 1
 403
 404             if args.verbose:
 405                 sys.stderr.write('\n')
 406                 sys.stderr.write('License files:     %12d\n' %spdx.license_files)
 407                 sys.stderr.write('Exception files:   %12d\n' %spdx.exception_files)
 408                 sys.stderr.write('License IDs        %12d\n' %len(spdx.licenses))
 409                 sys.stderr.write('Exception IDs      %12d\n' %len(spdx.exceptions))
 410                 sys.stderr.write('\n')
 411                 sys.stderr.write('Files excluded:    %12d\n' %parser.excluded)
 412                 sys.stderr.write('Files checked:     %12d\n' %parser.checked)
 413                 sys.stderr.write('Lines checked:     %12d\n' %parser.lines_checked)
 414                 if parser.checked:
 415                     pc = int(100 * parser.spdx_valid / parser.checked)
 416                     sys.stderr.write('Files with SPDX:   %12d %3d%%\n' %(parser.spdx_valid, pc))
 417                 sys.stderr.write('Files with errors: %12d\n' %parser.spdx_errors)
 418                 if ndirs:
 419                     sys.stderr.write('\n')
 420                     sys.stderr.write('Directories accounted: %8d\n' %ndirs)
 421                     pc = int(100 * dirsok / ndirs)
 422                     sys.stderr.write('Directories complete:  %8d %3d%%\n' %(dirsok, pc))
 423
 424             if ndirs and ndirs != dirsok and args.dirs:
 425                 if args.verbose:
 426                     sys.stderr.write('\n')
 427                 sys.stderr.write('Incomplete directories: SPDX in Files\n')
 428                 for f in sorted(parser.spdx_dirs.keys()):
 429                     di = parser.spdx_dirs[f]
 430                     if di.missing:
 431                         valid = di.total - di.missing
 432                         pc = int(100 * valid / di.total)
 433                         sys.stderr.write('    %-80s: %5d of %5d  %3d%%\n' %(f, valid, di.total, pc))
 434
 435             if ndirs and ndirs != dirsok and args.files:
 436                 if args.verbose or args.dirs:
 437                     sys.stderr.write('\n')
 438                 sys.stderr.write('Files without SPDX:\n')
 439                 for f in sorted(parser.spdx_dirs.keys()):
 440                     di = parser.spdx_dirs[f]
 441                     for f in sorted(di.files):
 442                         sys.stderr.write('    %s\n' %f)
 443
 444             sys.exit(0)
 445
 446     except Exception as ex:
 447         sys.stderr.write('FAIL: %s\n' %ex)
 448         sys.stderr.write('%s\n' %traceback.format_exc())
 449         sys.exit(1)