def link_project(self, mproject):
"""
Loads in the commits from a project and then attempts to place new entires in the
user_corporation_project table (aka ProjectInvolvement) table.
@param mproject: a dbobjects.MasterProject element to load the data for
"""
# FIXME: this currently has no support for hierarchical master projects
projects = Project.select(Project.q.masterProjectID == mproject.id) # pylint: disable-msg=E1101
# we'll need this for building SQL queries
projarr = [x.id for x in projects]
# get the first commit on the project so we know what month to start pumping data at
firstcommit = CVSCommit.select(AND(CVSCommit.q.startDate > datetime.datetime(1997, 11, 01), # pylint: disable-msg=E1101
IN(CVSCommit.q.projectID, projarr))).min(CVSCommit.q.startDate) # pylint: disable-msg=E1101
lastcommit = CVSCommit.select(AND(CVSCommit.q.startDate > datetime.datetime(1997, 11, 01), # pylint: disable-msg=E1101
IN(CVSCommit.q.projectID, projarr))).max(CVSCommit.q.startDate) # pylint: disable-msg=E1101
if firstcommit == None:
# do something here to indicate that it couldn't process this time period
self.log.info("Project %s (id=%s) does not appear to have any commits in the window", mproject.name, mproject.id)
return
self.log.info("%s - %s %s", mproject.name, firstcommit, lastcommit)
# iterate on a monthly basis across the data so we have monthly snapshots, just as we do for Eclipse
currentdate = datetime.datetime(firstcommit.year, firstcommit.month, 1)
nextdate = add_month(currentdate)
while currentdate < lastcommit:
self.link_project_month(mproject, projarr, currentdate, nextdate)
currentdate = nextdate
nextdate = add_month(currentdate)
def link_project_month(self, mproject, projarr, currentdate, nextdate):
"""
Does the "heavy" lifting for linking individuals to projects over a time period
@param mproject: master project to link to
@param projarr: the array of child projects
@param currendate: the date to start looking
@param nextdate: the date to stop looking
"""
# get the developers who committed during that period
people = Person.select(AND(Person.q.id == PersonUser.q.personID, # pylint: disable-msg=E1101
PersonUser.q.userID == CVSCommit.q.userID, # pylint: disable-msg=E1101
CVSCommit.q.startDate >= currentdate, # pylint: disable-msg=E1101
CVSCommit.q.startDate < nextdate, # pylint: disable-msg=E1101
IN(CVSCommit.q.projectID, projarr)), distinct=True) # pylint: disable-msg=E1101
count = 0
numcorp = 0
numunknown = 0
# iterate over each of the people
for person in people:
user = User.select(AND(User.q.id == PersonUser.q.userID, # pylint: disable-msg=E1101
PersonUser.q.personID == person.id))[0] # pylint: disable-msg=E1101
# first get their company. cache it if possible
if self.corpmap.has_key(person.id):
corporation = self.corpmap[person.id]
else:
corporations = person.corporations
if len(corporations) == 0:
corporation = self.unknowncorp
else:
corporation = corporations[0]
self.corpmap[person.id] = corporation
# increment some tracking counters
if corporation == self.unknowncorp:
numunknown = numunknown + 1
else:
numcorp = numcorp + 1
# now, get their commits, we need the number of commits and the number of lines added/removed
commits = CVSCommit.select(AND(CVSCommit.q.userID == PersonUser.q.userID, # pylint: disable-msg=E1101
PersonUser.q.personID == person.id, # pylint: disable-msg=E1101
CVSCommit.q.startDate >= currentdate, # pylint: disable-msg=E1101
CVSCommit.q.startDate < nextdate, # pylint: disable-msg=E1101
IN(CVSCommit.q.projectID, projarr)), distinct=True) # pylint: disable-msg=E1101
commitids = [x.id for x in commits]
numcommits = len(commitids)
# in some cases, we don't have complete information, we just set those to empty here
linesadded = FileCommit.select(IN(FileCommit.q.cvsCommitID, commitids)).sum(FileCommit.q.linesAdded) or 0 # pylint: disable-msg=E1101
linesremoved = FileCommit.select(IN(FileCommit.q.cvsCommitID, commitids)).sum(FileCommit.q.linesRemoved) or 0 # pylint: disable-msg=E1101
linesdelta = linesadded - linesremoved
# commit the object
newpi = ProjectInvolvement(user=user, corporation=corporation,
project=mproject, year=currentdate.year,
month=currentdate.month, date=currentdate,
numCommits=numcommits, linesAdded=linesadded,
linesRemoved=linesremoved, linesDelta=linesdelta)
count = count + 1
self.log.info("%s-%s Added %d links (%d corp, %d unknown)...", currentdate, nextdate, count, numcorp, numunknown)
return count
