10cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth#!/usr/bin/env python
20cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth
30cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth"""Script to sort the top-most block of #include lines.
40cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth
50cd94c71db19c0a53d1d97f353116e271884a336Chandler CarruthAssumes the LLVM coding conventions.
60cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth
70cd94c71db19c0a53d1d97f353116e271884a336Chandler CarruthCurrently, this script only bothers sorting the llvm/... headers. Patches
80cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruthwelcome for more functionality, and sorting other header groups.
90cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth"""
100cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth
110cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruthimport argparse
120cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruthimport os
130cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth
140cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruthdef sort_includes(f):
156cb571968901701212570cbb2ab34cf4af6ba51eChandler Carruth  """Sort the #include lines of a specific file."""
16fd025797ea5197b838f87ce57f2df5bccf27ad20Chandler Carruth
17fd025797ea5197b838f87ce57f2df5bccf27ad20Chandler Carruth  # Skip files which are under INPUTS trees or test trees.
18fd025797ea5197b838f87ce57f2df5bccf27ad20Chandler Carruth  if 'INPUTS/' in f.name or 'test/' in f.name:
19fd025797ea5197b838f87ce57f2df5bccf27ad20Chandler Carruth    return
20fd025797ea5197b838f87ce57f2df5bccf27ad20Chandler Carruth
2196ad0e8b1635eb923ef5e9c79fef8b902960dfbbChandler Carruth  ext = os.path.splitext(f.name)[1]
2296ad0e8b1635eb923ef5e9c79fef8b902960dfbbChandler Carruth  if ext not in ['.cpp', '.c', '.h', '.inc', '.def']:
2396ad0e8b1635eb923ef5e9c79fef8b902960dfbbChandler Carruth    return
2496ad0e8b1635eb923ef5e9c79fef8b902960dfbbChandler Carruth
250cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  lines = f.readlines()
2696ad0e8b1635eb923ef5e9c79fef8b902960dfbbChandler Carruth  look_for_api_header = ext in ['.cpp', '.c']
276cb571968901701212570cbb2ab34cf4af6ba51eChandler Carruth  found_headers = False
280cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  headers_begin = 0
290cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  headers_end = 0
300cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  api_headers = []
310cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  local_headers = []
320cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  project_headers = []
330cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  system_headers = []
340cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  for (i, l) in enumerate(lines):
350cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth    if l.strip() == '':
360cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth      continue
370cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth    if l.startswith('#include'):
386cb571968901701212570cbb2ab34cf4af6ba51eChandler Carruth      if not found_headers:
390cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth        headers_begin = i
406cb571968901701212570cbb2ab34cf4af6ba51eChandler Carruth        found_headers = True
410cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth      headers_end = i
420cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth      header = l[len('#include'):].lstrip()
430cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth      if look_for_api_header and header.startswith('"'):
440cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth        api_headers.append(header)
450cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth        look_for_api_header = False
460cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth        continue
471ff3d66222634fabad236423e454c0b9bd631084Chandler Carruth      if header.startswith('<') or header.startswith('"gtest/'):
480cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth        system_headers.append(header)
490cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth        continue
506cb571968901701212570cbb2ab34cf4af6ba51eChandler Carruth      if (header.startswith('"llvm/') or header.startswith('"llvm-c/') or
516cb571968901701212570cbb2ab34cf4af6ba51eChandler Carruth          header.startswith('"clang/') or header.startswith('"clang-c/')):
520cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth        project_headers.append(header)
530cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth        continue
540cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth      local_headers.append(header)
550cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth      continue
560cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth
570cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth    # Only allow comments and #defines prior to any includes. If either are
580cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth    # mixed with includes, the order might be sensitive.
596cb571968901701212570cbb2ab34cf4af6ba51eChandler Carruth    if found_headers:
600cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth      break
61998aae738d2a014cb46d6e29a585fd781f95b677Chandler Carruth    if l.startswith('//') or l.startswith('#define') or l.startswith('#ifndef'):
620cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth      continue
630cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth    break
646cb571968901701212570cbb2ab34cf4af6ba51eChandler Carruth  if not found_headers:
650cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth    return
660cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth
67c851438e3f2c587c5b8fb7ec4bf179798532b9f3Benjamin Kramer  local_headers = sorted(set(local_headers))
68c851438e3f2c587c5b8fb7ec4bf179798532b9f3Benjamin Kramer  project_headers = sorted(set(project_headers))
69c851438e3f2c587c5b8fb7ec4bf179798532b9f3Benjamin Kramer  system_headers = sorted(set(system_headers))
700cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  headers = api_headers + local_headers + project_headers + system_headers
710cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  header_lines = ['#include ' + h for h in headers]
720cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  lines = lines[:headers_begin] + header_lines + lines[headers_end + 1:]
730cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth
740cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  f.seek(0)
750cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  f.truncate()
760cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  f.writelines(lines)
770cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth
780cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruthdef main():
790cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  parser = argparse.ArgumentParser(description=__doc__)
800cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  parser.add_argument('files', nargs='+', type=argparse.FileType('r+'),
810cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth                      help='the source files to sort includes within')
820cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  args = parser.parse_args()
830cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  for f in args.files:
840cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth    sort_includes(f)
850cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth
860cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruthif __name__ == '__main__':
870cd94c71db19c0a53d1d97f353116e271884a336Chandler Carruth  main()
88