blobtracker.py

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