blobtracker.py

   1 import sys
   2
   3 import db
   4 from git import git
   5
   6 _blob_cache = {}
   7
   8 def get_blob(session, sha1):
   9     if sha1 in _blob_cache:
  10         return _blob_cache[sha1]
  11     b = session.query(db.Blob).filter(db.Blob.sha1==sha1).first()
  12     _blob_cache[sha1] = b
  13     return b
  14
  15 def scan_commit(session, commit):
  16     for line in git('ls-tree', '-r', commit.sha1, ret_pipe=True):
  17         assert line.endswith('\n')
  18         rest, name = line[:-1].split('\t', 1)
  19         mode, type, sha1 = rest.split(' ')
  20         blob = get_blob(session, sha1)
  21         if not blob:
  22             blob = db.Blob(sha1)
  23             session.add(blob)
  24             _blob_cache[sha1] = blob
  25         blob.update_contained_in(commit)
  26
  27 def scan_history(excludes=[]):
  28     session = db.Session()
  29     refdata = git('for-each-ref', '--format=%(objectname)')[0]
  30     refs = refdata.split()
  31     boundaries = session.query(db.Boundary).all()
  32     args = refs + ['--not'] + [b.sha1 for b in boundaries] + excludes
  33     patch_ids = {}
  34     print 'reading existing commits ...'
  35     commit_cache = set()
  36     for sha1 in session.query(db.Commit.sha1):
  37         commit_cache.add(sha1)
  38     print 'log -p | patch-id ...'
  39     pipe = git('log', '-p', '--no-merges', *args, ret_pipe=True)
  40     for line in git('patch-id', input_pipe=pipe, ret_pipe=True):
  41         patch_id, commit_sha1 = line.split()
  42         patch_ids[commit_sha1] = patch_id
  43     print 'reading trees ...'
  44     count = 0
  45     for line in git('rev-list', '--no-merges', *args, ret_pipe=True):
  46         sys.stdout.write('\r%6d' % count)
  47         sys.stdout.flush()
  48         sha1 = line.strip()
  49         if sha1 not in commit_cache:
  50             commit_cache.add(sha1)
  51             output = git('log', '-1', '--pretty=format:%ct %at', sha1)[0]
  52             adate, cdate = [int(s) for s in output.split()]
  53             commit = db.Commit(sha1, cdate, adate, patch_ids.get(sha1, None))
  54             session.add(commit)
  55             scan_commit(session, commit)
  56         count = count + 1
  57     print '\nstoring boundaries ...'
  58     for b in session.query(db.Boundary).all():
  59         session.delete(b)
  60     for r in set(refs):
  61         session.add(db.Boundary(r))
  62     session.commit()
  63
  64 if __name__ == '__main__':
  65     scan_history(sys.argv[1:])