def readReviewComments(self):
# File with review, commit, file and developer data from comments in format:
# relationType/\\fileName/\\ownerName/\\commenterName/\\reviewNumber/\\commitHash
revFileComm = open("Data\\ReviewFilesFromComments.txt", "r")
while (True):
crtLine = revFileComm.readline()
if not crtLine:
break
lst = crtLine[:-1].split("/\\")
assert len(lst) == 6
reviewNumber = lst[4]
self.addReview(reviewNumber)
ownerName = Ownership.purifyName(lst[2])
commenterName = Ownership.purifyName(lst[3])
node1 = self.humanDict[ownerName].index
node2 = self.humanDict[commenterName].index
L = Settings.getLayer2('reviewer', 'reviewOwner')
if ownerName != commenterName:
# Add edge from commenter to owner.
self.addEdge(node2, L, node1, L, 8)
if not ('.' in lst[1]):
# All files should contain at least one '.' from the file type.
continue
fileName = (lst[1].rsplit('.', 1)[0]).replace('/', '.')[:-1]
if fileName in self.fileDict:
self.reviewDict[reviewNumber].addFile(self.fileDict[fileName])
#Add edge between reviewOwner and file on Review layer.
self.addEdge(node1, 3, self.fileDict[fileName], 3, 11)
def readIssues(self):
# Each line of file contains issue data in format:
# bugID/\version/\creation_ts/\delta_ts/\status/\resolution.
issueFile = open("Data\\BugDetails.txt")
while True:
crtLine = issueFile.readline()
if not crtLine:
break
lst = crtLine.split('/\\')
assert (len(lst) == 6)
dateTime = lst[2].split(' ', 1)
dateTime[1] = dateTime[1].replace(' ', '')
assert len(dateTime) == 2
lst[2] = Ownership.getTime(dateTime[0], dateTime[1])
dateTime = lst[3].split(' ', 1)
dateTime[1] = dateTime[1].replace(' ', '')
lst[3] = Ownership.getTime(dateTime[0], dateTime[1])
assert len(dateTime) == 2
lst[5] = lst[5][:-1] # ignore '\n'
issue = Issue.Issue(lst[0], lst[1], lst[2], lst[3], lst[4], lst[5],
self.projectList[self.projectID])
if issue.getType() == self.issueType and (lst[5] == 'FIXED' or
lst[5] == 'WORKSFORME'):
if not (issue.name in self.issueDict):
self.addIssue(issue)
issue.setIndex(self.nrIssues)
self.i_R[self.nrNodes] = {}
# only fixed issues must be added =>
# if ("CLOSED" in status_i) or ("RESOLVED" in status_i) or ("VERIFIED" in status_i) or ("FIXED" in status_i):
issueFile.close()
def readReviews(self):
reviewEdges = {}
reviewFile = open("Data\\ReviewEdges2020.txt", "r")
while True:
crtLine = reviewFile.readline()
if not crtLine:
break
lst = crtLine.split('/\\')
assert len(lst) == 3
if lst[0] == 'CommentEdge' or lst[0] == 'PCommentEdge':
# The format of current line is "edgeType/\ownerName/\commenterName".
name1 = Ownership.purifyName(lst[1])
name2 = Ownership.purifyName(lst[2][:-1])
L = Settings.getLayer2('reviewer', 'reviewOwner')
self.addEdge(self.humanDict[name2].index, L,
self.humanDict[name1].index, L, 8)
elif lst[0] == 'Review2Commit':
# The format of current line is "edgeType/\reviewNumber/\commitHash".
reviewNumber = lst[1]
self.addReview(reviewNumber)
commitHash = lst[2].replace('\n', '')
# If the commit belongs to multiple reviews, then data is invalid.
assert (
not ((commitHash in self.reviewIdForCommit) and
(self.reviewIdForCommit[commitHash] != reviewNumber)))
self.reviewIdForCommit[commitHash] = reviewNumber
else:
# The format of current line is "edgeType/\commitHash/\developerName", where
# edgeType in {AuthorEdge, UploaderEdge, ApprovalEdge, OwnerEdge}
commitHash = lst[1]
developerName = Ownership.purifyName(lst[2][:-1])
# reviewEdges[commitHash] = edges that relate to commitHash s.t edges between the
# review corresponding to commitHash can be linked with the humans in one pass.
if not (commitHash in reviewEdges):
reviewEdges[commitHash] = []
reviewEdges[commitHash].append((lst[0], developerName))
if commitHash in self.commitDict:
layer = 'patchUploader'
col = 5
if lst[0] == 'OwnerEdge':
layer = 'reviewOwner'
col = 6
elif lst[0] == 'AuthorEdge':
layer = 'author'
col = 7
elif lst[0] == 'ApprovalEdge':
layer = 'approver'
col = 8
else:
assert lst[0] == 'UploaderEdge'
L = Settings.getLayer2(layer, 'file')
# link humans to the files of the commit
if developerName in self.humanDict:
for fileNode in self.commitDict[
commitHash].modifiedFiles:
self.addEdge(self.humanDict[developerName].index,
L, fileNode, L, 10)
self.processReviewEdges(reviewEdges, 3)
reviewFile.close()
def readCommits(self, commitDataFile):
fileListForCrtCommit = []
crtCommitHash = 0
committerName = ''
authorName = ''
while (True):
crtLine = commitDataFile.readline()
if not crtLine:
break
if (crtLine == '' or crtLine == '\n'):
continue
lst = crtLine.split('/\\')
if not (('.' in lst[0]) or ('/' in lst[0]) or ('\\' in lst[0])):
# committer/\\author/\\commitHash
assert len(lst) == 3
committerName = Ownership.purifyName(lst[0])
authorName = Ownership.purifyName(lst[1])
self.addHuman(committerName, 0)
self.addHuman(authorName, 5)
commitHash = lst[2][:-1]
crtCommitHash = commitHash
self.addCommit(commitHash)
if authorName != committerName:
# add edge committer->author as cross layer edge
self.addEdge(self.humanDict[committerName].index, 1,
self.humanDict[authorName].index, 3, 2)
fileListForCrtCommit = []
else:
crtFile = crtLine.rsplit('.', 1)[0].replace("/", '.')
if not (crtFile in self.fileDict):
self.addFile(crtFile)
self.commitDict[crtCommitHash].addFile(self.fileDict[crtFile])
if len(fileListForCrtCommit) > 0:
L = Settings.getLayer2('file', 'fileC')
self.addEdge(self.fileDict[crtFile], L,
self.fileDict[fileListForCrtCommit[-1]], L, 3)
# Add edge from author of file to file.
self.addEdge(self.humanDict[authorName].index, 1,
self.fileDict[crtFile], 1, 4)
if authorName != committerName:
# Add edge from committer of file to file.
self.addEdge(self.humanDict[committerName].index, 1,
self.fileDict[crtFile], 1, 4)
fileListForCrtCommit.append(crtFile)
commitDataFile.close()
def readNameUsername(self):
# name/\username from Bugzilla
f = open("Data\\emailName2020B.txt", "rb")
while (True):
crtLine = f.readline().decode('utf-8')
if not crtLine:
break
lst = crtLine.split('/\\')
assert len(lst) == 2
name = Ownership.purifyName(lst[0])
if name in self.humanDict:
self.nameOfUsername[lst[-1][:-1]] = name
self.humanDict[name].setUserName(lst[-1][:-1])
f.close()
def readHumanRoleFile(self, humanRoleFile, fileId):
while (True):
# names in Bugzilla are read from binary files
if fileId >= 6 and fileId <= 8:
crtLine = humanRoleFile.readline().decode('utf-8')
else:
crtLine = humanRoleFile.readline()
if not crtLine:
break
# name/\email/\index
lst = crtLine.split('/\\')
assert len(lst) == 3 - (fileId >= 6 and fileId <= 8)
name = Ownership.purifyName(lst[0])
self.addHuman(name, fileId)
humanRoleFile.close()
def readIssueEdges(self):
# Each line of file contains data for an issue in format HumanRole/\name/username/\bugID.
issueEdgesFile = open("Data\\IssueEdges2020B.txt", "rb")
while True:
crtLine = issueEdgesFile.readline().decode('utf-8')
if not crtLine:
break
if crtLine == '':
continue
lst = crtLine.split('/\\')
assert (len(lst) == 3)
if (lst[0][0] != 'C'):
# Ignore "CC-Assignees".
name = Ownership.purifyName(lst[1])
bugId = lst[-1][:-1]
if not (bugId in self.issueDict):
# issue was not fixed
continue
# All humans must have been introduced previously.
assert (name in self.humanDict)
layer = 'issueReporter'
col = 5
if lst[0][0] == 'A':
layer = 'issueAssignee'
col = 6
else:
assert lst[0][0] == 'R'
self.humanDict[name].isReporter = True
L = Settings.getLayer2(layer, 'issue')
self.addEdge(self.humanDict[name].index, L,
self.issueDict[bugId], L, col)
else:
uname = lst[1]
if uname in self.nameOfUsername:
name = self.nameOfUsername[uname]
# ToDo add CCassignee edges
issueEdgesFile.close()
return self.nrIssues
def readOwnershipFile(self):
# For each modified file in the repository, the format is as follows:
# fileName/\nrCommits
# author_name/\self.author_date/\author_timezone/\added/\removed/\complexity
# Note: author_date = '%Y-%m-%d %H:%M:%S'
ownershipFile = open("Data\\ownership.txt")
Itype = Settings.getIssueType()
Ntype = Settings.getNetworkType()
valuesC = []
valuesL = []
while (True):
# Read the line with file details, then read nrCommits lines with commit details.
crtLine = ownershipFile.readline()
if not crtLine:
break
lst = crtLine.split('/\\')
# Format the file names to contain '.' instead of '/'
compName = lst[0].replace('/', '.')
# Remove the file type (i.e "dir.file.java" -> "dir.file").
compName = compName.rsplit('.', 1)[0]
if not (compName in self.fileDict):
# Ignore modified files that are not in the Contribution Network.
for i in range(int(lst[1])):
ownershipFile.readline()
continue
obj = Ownership.Ownership(compName)
for i in range(int(lst[1])):
nxtLine = ownershipFile.readline().split('/\\')
lineLen = len(nxtLine)
if lineLen == 0:
continue
obj.addModif(Ownership.getModifFromLine(nxtLine))
allCommitters = obj.authorDex[0]
L = Settings.getLayer2('committer', 'committer')
A = Settings.getLayer2('committer', 'author')
nrChangedLines = 0
if self.fileDict[obj.name] in self.posInFiles:
nrChangedLines = obj.sumAdd[0] + obj.sumRem[0]
for c1 in allCommitters:
cp = (100 * obj.authorDex[0][c1].nrCommits / obj.nrCommits[0])
# Add only minor edges. Change to cp > 50 for only major edges.
if (cp > 50):
# Add major edge
self.addEdge(self.humanDict[c1].index, 1,
self.fileDict[obj.name], 1, 14)
else:
# Add minor edge
self.addEdge(self.humanDict[c1].index, 1,
self.fileDict[obj.name], 1, -14)
valuesC.append(cp)
if nrChangedLines != 0:
c1ChangedLines = obj.authorDex[0][
c1].sumAdd + obj.authorDex[0][c1].sumRem
assert (c1ChangedLines <= nrChangedLines)
if c1ChangedLines == nrChangedLines:
lp = 100
else:
lp = (100 * (obj.authorDex[0][c1].sumAdd +
obj.authorDex[0][c1].sumRem) /
nrChangedLines)
valuesL.append(lp)
for c2 in allCommitters:
if c1 != c2:
if self.humanDict[c1].isRole[0] and self.humanDict[
c2].isRole[0]:
# Add committer to committer edge.
self.addEdge(self.humanDict[c1].index, L,
self.humanDict[c2].index, L, 13)
elif self.humanDict[c2].isRole[0] or self.humanDict[
c1].isRole[0]:
# Add committer to author edge.
self.addEdge(self.humanDict[c1].index, L,
self.humanDict[c2].index, A, 2)
# Tuple(name, nrCommits) for obj file where name is the name of the committer with
# the highest number of the commits to obj.
ownershipTuple = obj.nrCommitsOwner(0)
self.ownershipDict[self.fileDict[obj.name]] = (
ownershipTuple[0], obj.nrCommitsPercentage(0))
ownershipFile.close()