def buildCommitPart():
repos = linkOperator.selectRepoOver(5000)
logCorpus = open('commitLog.dat', "w")
codeCorpus = open('commitCode.dat', "w")
try:
print 'start'
for highRepo in repos:
path = getPath(highRepo[1])
try:
gitRe = gitResolver.GitResolver(path)
commits = gitRe.getCommits()
print path, ":", len(commits)
for commit in commits:
words = preprocessor.preprocessToWord(
commit.message.decode('utf-8'))
if len(words):
# 不是空列表
for word in words:
logCorpus.write(word.encode('utf-8'))
logCorpus.write(" ")
logCorpus.write("\n")
diffs = gitRe.getOneDiff(commit)
for diff in diffs:
diffCode = preprocessor.processDiffCode(diff.diff)
if len(diffCode):
for code in diffCode:
codeCorpus.write(code)
codeCorpus.write(" ")
codeCorpus.write("\n")
except BaseException, e:
print "***", path, ":", e
print traceback.format_exc()
print 'end'
def buildFromGit():
repos = mysqlOperator.selectAllHighRepository()
corpus = open('corpusLabel.dat', "w")
try:
for highRepo in repos:
path = getPath(highRepo[1])
try:
gitRe = gitResolver.GitResolver(path)
commits = gitRe.getCommits()
print path, ":", len(commits)
for commit in commits:
corpus.write(str(highRepo[0]).encode("utf-8"))
corpus.write("\n")
corpus.write(commit.hexsha.encode("utf-8"))
corpus.write("\n")
sens = preprocessor.preprocess(
commit.message.decode('utf-8'))
for sentence in sens:
if len(sentence): #不是空列表
for word in sentence:
corpus.write(word.encode('utf-8'))
corpus.write(" ")
corpus.write("\n")
corpus.write("\n")
except BaseException, e:
print "***", path, ":", e
except IOError, e: #检查open()是否失败，通常是IOError类型的错误
print "***", e
def buildByList(repoList, corpusName):
corpus = open('corpus/nocode%s.dat' % corpusName, "w")
commitCorpus = open('corpus/commit%s.dat' % corpusName, "w")
issueCorpus = open('corpus/issue%s.dat' % corpusName, "w")
try:
print 'start'
for i in range(len(repoList)):
repoId = repoList[i]['id']
repoPath = repoList[i]['path']
try:
# commit part
gitRe = gitResolver.GitResolver(repoPath)
commits = gitRe.getCommits()
print repoPath, ":", len(commits)
for commit in commits:
seqs = preprocessor.preprocessNoCamel(
commit.message.decode('utf-8'))
if len(seqs):
# 不是空列表
for seq in seqs:
for word in seq:
corpus.write(word.encode('utf-8'))
corpus.write(" ")
commitCorpus.write(word.encode('utf-8'))
commitCorpus.write(" ")
corpus.write("\n")
commitCorpus.write("\n")
# issue part
issues = mysqlOperator.selectAllIssueInOneRepo(repoId)
print repoId, ":", len(issues)
for issue in issues:
titleSeqs = preprocessor.preprocessNoCamel(
issue[4].decode('utf-8'))
if len(titleSeqs):
# 不是空列表
for titleSeq in titleSeqs:
for word in titleSeq:
corpus.write(word.encode('utf-8'))
corpus.write(" ")
issueCorpus.write(word.encode('utf-8'))
issueCorpus.write(" ")
corpus.write("\n")
issueCorpus.write("\n")
if issue[5]:
body = preprocessor.processHTMLNoCamel(
issue[5].decode('utf-8'))
if len(body):
# 不是空列表
for bodySeq in body:
for word in bodySeq:
corpus.write(word.encode('utf-8'))
corpus.write(" ")
issueCorpus.write(word.encode('utf-8'))
issueCorpus.write(" ")
corpus.write("\n")
issueCorpus.write("\n")
except BaseException, e:
print "***", repoId, ":", e
print traceback.format_exc()
print 'end'
def buildLinks(repoId):
print 'start'
try:
repoPath = nocodeRepoInfo.REPO_MAP[
nocodeRepoInfo.USE_REPO_INDEX]['path']
gitRepo = gitResolver.GitResolver(repoPath)
issues = mysqlOperator.selectAllIssueInOneRepo(repoId)
commits = gitRepo.getCommits()
# repoName = re.sub(r'https://github.com/', '', repo[1], 0, re.I)
print '==============', repoPath, 'Start'
for commit in commits:
commitSha = str(commit.hexsha.encode("utf-8"))
print commitSha
commitIssues = mysqlOperator.selectExistIssueOnCommit(
(repoId, commitSha))
trueLinks = []
for ci in commitIssues:
if ci[0] in trueLinks:
pass
else:
trueLinks.append(ci[0])
linkOperator.insertLink(
('true_link_%d' % repoId, repoId, commitSha, ci[0]))
for issue in issues:
if isUnlabeled(issue, gitRepo.getDateTime(commit)):
if len(commitIssues) > 0:
if issue[1] in trueLinks:
pass
else:
linkOperator.insertLink(
('false_link_%d' % repoId, repoId, commitSha,
issue[1]))
else:
pass
print '==============', repoPath, 'End'
except Exception, e:
print 'Error:', repoPath
print traceback.format_exc()
def buildIssueAndCommitSeq(repoId, repoPath, corpusName):
corpus = open('corpus/code%s.dat' % corpusName, "w")
try:
print 'start'
try:
# commit part
gitRe = gitResolver.GitResolver(repoPath)
commits = gitRe.getCommits()
print repoPath, ":", len(commits)
for commit in commits:
diffs = gitRe.getOneDiff(commit)
for diff in diffs:
diffCode = preprocessor.processDiffCode(diff.diff)
if len(diffCode):
for word in diffCode:
corpus.write(word.encode('utf-8'))
corpus.write(" ")
corpus.write("\n")
# issue part
issues = mysqlOperator.selectAllIssueInOneRepo(repoId)
print repoId, ":", len(issues)
for issue in issues:
if issue[5]:
bodycode = preprocessor.getIssueCode(
issue[5].decode('utf-8'))
if len(bodycode):
# 不是空列表
for word in bodycode:
corpus.write(word.encode('utf-8'))
corpus.write(" ")
corpus.write("\n")
except BaseException, e:
print "***", repoId, ":", e
print traceback.format_exc()
print 'end'
def buildTrainSet(trueTable, falseTable, repoId, repoPath, trueGap, falseGap,
trueCount, falseCount):
trueStart = 1
falseStart = 1
textCorpus = open('frcorpus/text%d.dat' % repoId, "w")
codeCorpus = open('frcorpus/code%d.dat' % repoId, "w")
trueLinkList = linkOperator.selectInScope(
(trueTable, trueStart, trueStart + trueCount))
falseLinkList = getRandomFalse(falseTable, falseStart,
falseStart + falseGap, falseCount)
index = 0
repo = gitResolver.GitResolver(repoPath)
try:
while len(trueLinkList) > 0 and len(falseLinkList) > 0:
print 'true: ', trueStart, ' to ', trueStart + trueCount
print 'false: ', falseStart, ' to ', falseStart + falseCount
linkList = []
for trueLink in trueLinkList:
commit = repo.getOneCommit(trueLink[1])
issue = mysqlOperator.selectOneIssue(trueLink[2])
if issue is None:
continue
comments = mysqlOperator.selectCommentInOneIssue(trueLink[2])
try:
files = repo.getFiles(trueLink[1])
except:
print 'File Fail 1:', trueLink[1]
continue
res = {}
res['type'] = 1
res['issueText'] = []
# issue body
if issue[5]:
res['issueCode'] = frpreprocesser.extractCode(
issue[5].decode('utf-8'))
res['issueText'].append(
frpreprocesser.extractText(
issue[5].decode('utf-8'))) # body
else:
res['issueCode'] = []
res['issueText'].append(
frpreprocesser.extractText(
issue[4].decode('utf-8'))) # title
for comment in comments:
res['issueText'].append(
frpreprocesser.extractText(comment[4].decode('utf-8')))
res['commitText'] = []
res['commitCode'] = []
res['commitText'].append(
frpreprocesser.extractText(commit.message.decode('utf-8')))
for changeFile in files:
if not changeFile['path'].endswith('.java'):
try:
res['commitText'].append(
frpreprocesser.extractText(
changeFile['text'].decode('utf-8')))
except:
print trueLink[1], ':', changeFile['path']
else:
codes = frpreprocesser.extractCode(
changeFile['text'].decode('utf-8'))
for code in codes:
if code in res['issueCode']:
res['commitCode'].extend(codes)
break
linkList.append(res)
writeToCorpus(textCorpus, codeCorpus, res['commitText'],
res['commitCode'])
writeToCorpus(textCorpus, codeCorpus, res['issueText'],
res['issueCode'])
for falseLink in falseLinkList:
commit = repo.getOneCommit(falseLink[1])
issue = mysqlOperator.selectOneIssue(falseLink[2])
if issue is None:
continue
comments = mysqlOperator.selectCommentInOneIssue(falseLink[2])
try:
files = repo.getFiles(falseLink[1])
except:
print 'File Fail 0:', falseLink[1]
continue
res = {}
res['type'] = 0
res['issueText'] = []
# issue body
if issue[5]:
res['issueCode'] = frpreprocesser.extractCode(
issue[5].decode('utf-8'))
res['issueText'].append(
frpreprocesser.extractText(
issue[5].decode('utf-8'))) # body
else:
res['issueCode'] = []
res['issueText'].append(
frpreprocesser.extractText(
issue[4].decode('utf-8'))) # title
for comment in comments:
res['issueText'].append(
frpreprocesser.extractText(comment[4].decode('utf-8')))
res['commitText'] = []
res['commitCode'] = []
res['commitText'].append(
frpreprocesser.extractText(commit.message.decode('utf-8')))
for changeFile in files:
if not changeFile['path'].endswith('.java'):
try:
res['commitText'].append(
frpreprocesser.extractText(
changeFile['text'].decode('utf-8')))
except:
print trueLink[1], ':', changeFile['path']
else:
codes = frpreprocesser.extractCode(
changeFile['text'].decode('utf-8'))
for code in codes:
if code in res['issueCode']:
res['commitCode'].extend(codes)
break
linkList.append(res)
writeToCorpus(textCorpus, codeCorpus, res['commitText'],
res['commitCode'])
writeToCorpus(textCorpus, codeCorpus, res['issueText'],
res['issueCode'])
index += 1
res = json.dumps(linkList, encoding="utf-8", indent=4)
trainSet = open(
'./frtrain%d/traincase%d-%d.dat' % (repoId, repoId, index),
"w")
trainSet.write(res)
trainSet.close()
print './frtrain%d/traincase%d-%d.dat' % (repoId, repoId,
index), 'Over'
trueStart += trueGap
falseStart += falseGap
trueLinkList = linkOperator.selectInScope(
(trueTable, trueStart, trueStart + trueCount))
falseLinkList = getRandomFalse(falseTable, falseStart,
falseStart + falseGap, falseCount)
except IOError, e:
print "***", e
print traceback.format_exc()
def buildTrainSet(trueTable, falseTable, repoId, repoPath, trueGap, falseGap, trueCount, falseCount):
trueStart = 1
falseStart = 1
trueLinkList = linkOperator.selectInScope((trueTable, trueStart, trueStart + trueCount))
falseLinkList = getRandomFalse(falseTable, falseStart, falseStart + falseGap, falseCount)
index = 0
repo = gitResolver.GitResolver(repoPath)
while len(trueLinkList) > 0 and len(falseLinkList) > 0:
print 'true: ', trueStart, ' to ', trueStart + trueCount
print 'false: ', falseStart, ' to ', falseStart + falseCount
linkList = []
for trueLink in trueLinkList:
commit = repo.getOneCommit(trueLink[1])
issue = mysqlOperator.selectOneIssue(trueLink[2])
if issue is None:
continue
res = {}
res['type'] = 1
res['commit'] = commit.message.decode('utf-8')
res['issuetitle'] = issue[4].decode('utf-8')
# issue body
if issue[5]:
res['issue'] = issue[5].decode('utf-8')
issueCodes = []
bodycode = preprocessor.getIssueCode(res['issue'])
if len(bodycode):
issueCodes.append(bodycode)
res['issuecode'] = issueCodes
else:
res['issue'] = ''
res['issuecode'] = []
diffs = repo.getOneDiff(commit)
diffCodes = []
for diff in diffs:
diffCode = preprocessor.processDiffCode(diff.diff)
if len(diffCode):
diffCodes.append(diffCode)
res['commitcode'] = diffCodes
linkList.append(res)
for falseLink in falseLinkList:
commit = repo.getOneCommit(falseLink[1])
issue = mysqlOperator.selectOneIssue(falseLink[2])
if issue is None:
continue
res = {}
res['type'] = 0
res['commit'] = commit.message.decode('utf-8')
res['issuetitle'] = issue[4].decode('utf-8')
# issue body
if issue[5]:
res['issue'] = issue[5].decode('utf-8')
issueCodes = []
bodycode = preprocessor.getIssueCode(res['issue'])
if len(bodycode):
issueCodes.append(bodycode)
res['issuecode'] = issueCodes
else:
res['issue'] = ''
res['issuecode'] = []
diffs = repo.getOneDiff(commit)
diffCodes = []
for diff in diffs:
diffCode = preprocessor.processDiffCode(diff.diff)
if len(diffCode):
diffCodes.append(diffCode)
res['commitcode'] = diffCodes
linkList.append(res)
index += 1
res = json.dumps(linkList, encoding="utf-8", indent=4)
trainSet = open('./codetrain%d/codetrain%d-%d.dat' % (repoId, repoId, index), "w")
trainSet.write(res)
trainSet.close()
print './codetrain%d/codetrain%d-%d.dat' % (repoId, repoId, index), 'Over'
trueStart += trueGap
falseStart += falseGap
trueLinkList = linkOperator.selectInScope((trueTable, trueStart, trueStart + trueCount))
falseLinkList = getRandomFalse(falseTable, falseStart, falseStart + falseGap, falseCount)
mysqlOperator.close()
linkOperator.close()
# -*- coding: UTF-8 -*-
from gitresolver import gitResolver
import datetime
# ioHandler.buildCorpus("model.dat", "corpus.dat")
path = 'D:/github/checkstyle'
repo = gitResolver.GitResolver(path)
# print repo.getFiles('76d6365018ec7688c8a8475b2f9aa496fbcfe88c')
print repo.getFiles('eceaa8b65a982db58d31ac901cdd751c435b1362')
def buildTrainSet(trueTable, falseTable, repoId, repoPath, trueGap, falseGap, trueCount, falseCount):
trueStart = 1
falseStart = 1
textCorpus = open('frcorpus/frtext%d.dat' % repoId, "w")
codeCorpus = open('frcorpus/frcode%d.dat' % repoId, "w")
trueLinkList = linkOperator.selectInScope((trueTable, trueStart, trueStart + trueCount))
falseLinkList = getRandomFalse(falseTable, falseStart, falseStart + falseGap, falseCount)
index = 0
repo = gitResolver.GitResolver(repoPath)
try:
while len(trueLinkList) > 0 and len(falseLinkList) > 0:
print 'true: ', trueStart, ' to ', trueStart + trueCount
print 'false: ', falseStart, ' to ', falseStart + falseCount
my_linkList = []
fr_linkList = []
for trueLink in trueLinkList:
commit = repo.getOneCommit(trueLink[1])
issue = mysqlOperator.selectOneIssue(trueLink[2])
if issue is None:
continue
my_res = {}
my_res['type'] = 1
my_res['commit'] = commit.message.decode('utf-8')
my_res['issuetitle'] = issue[4].decode('utf-8')
# issue body
if issue[5]:
my_res['issue'] = issue[5].decode('utf-8')
issueCodes = []
bodycode = preprocessor.getIssueCode(my_res['issue'])
if len(bodycode):
issueCodes.append(bodycode)
my_res['issuecode'] = issueCodes
else:
my_res['issue'] = ''
my_res['issuecode'] = []
diffs = repo.getOneDiff(commit)
diffCodes = []
for diff in diffs:
diffCode = preprocessor.processDiffCode(diff.diff)
if len(diffCode):
diffCodes.append(diffCode)
my_res['commitcode'] = diffCodes
my_linkList.append(my_res)
fr_res = {}
fr_res['type'] = 1
fr_res['issueText'] = []
# issue body
if issue[5]:
fr_res['issueCode'] = frpreprocesser.extractCode(issue[5].decode('utf-8'))
fr_res['issueText'].append(frpreprocesser.extractText(issue[5].decode('utf-8'))) # body
else:
fr_res['issueCode'] = []
fr_res['issueText'].append(frpreprocesser.extractText(issue[4].decode('utf-8'))) # title
fr_res['commitText'] = []
fr_res['commitCode'] = []
fr_res['commitText'].append(frpreprocesser.extractText(commit.message.decode('utf-8')))
comments = mysqlOperator.selectCommentInOneIssue(trueLink[2])
for comment in comments:
fr_res['issueText'].append(frpreprocesser.extractText(comment[4].decode('utf-8')))
try:
files = repo.getFiles(trueLink[1])
for changeFile in files:
if not changeFile['path'].endswith('.java'):
try:
fr_res['commitText'].append(
frpreprocesser.extractText(changeFile['text'].decode('utf-8')))
except:
print trueLink[1], ':', changeFile['path']
else:
codes = frpreprocesser.extractCode(changeFile['text'].decode('utf-8'))
for code in codes:
if code in fr_res['issueCode']:
fr_res['commitCode'].extend(codes)
break
except:
print 'File Fail 1:', trueLink[1]
fr_linkList.append(fr_res)
writeToCorpus(textCorpus, codeCorpus, fr_res['commitText'], fr_res['commitCode'])
writeToCorpus(textCorpus, codeCorpus, fr_res['issueText'], fr_res['issueCode'])
for falseLink in falseLinkList:
commit = repo.getOneCommit(falseLink[1])
issue = mysqlOperator.selectOneIssue(falseLink[2])
if issue is None:
continue
my_res = {}
my_res['type'] = 0
my_res['commit'] = commit.message.decode('utf-8')
my_res['issuetitle'] = issue[4].decode('utf-8')
# issue body
if issue[5]:
my_res['issue'] = issue[5].decode('utf-8')
issueCodes = []
bodycode = preprocessor.getIssueCode(my_res['issue'])
if len(bodycode):
issueCodes.append(bodycode)
my_res['issuecode'] = issueCodes
else:
my_res['issue'] = ''
my_res['issuecode'] = []
diffs = repo.getOneDiff(commit)
diffCodes = []
for diff in diffs:
diffCode = preprocessor.processDiffCode(diff.diff)
if len(diffCode):
diffCodes.append(diffCode)
my_res['commitcode'] = diffCodes
my_linkList.append(my_res)
fr_res = {}
fr_res['type'] = 0
fr_res['issueText'] = []
# issue body
if issue[5]:
fr_res['issueCode'] = frpreprocesser.extractCode(issue[5].decode('utf-8'))
fr_res['issueText'].append(frpreprocesser.extractText(issue[5].decode('utf-8'))) # body
else:
fr_res['issueCode'] = []
fr_res['issueText'].append(frpreprocesser.extractText(issue[4].decode('utf-8'))) # title
fr_res['commitText'] = []
fr_res['commitCode'] = []
fr_res['commitText'].append(frpreprocesser.extractText(commit.message.decode('utf-8')))
comments = mysqlOperator.selectCommentInOneIssue(falseLink[2])
for comment in comments:
fr_res['issueText'].append(frpreprocesser.extractText(comment[4].decode('utf-8')))
try:
files = repo.getFiles(falseLink[1])
for changeFile in files:
if not changeFile['path'].endswith('.java'):
try:
fr_res['commitText'].append(
frpreprocesser.extractText(changeFile['text'].decode('utf-8')))
except:
print trueLink[1], ':', changeFile['path']
else:
codes = frpreprocesser.extractCode(changeFile['text'].decode('utf-8'))
for code in codes:
if code in fr_res['issueCode']:
fr_res['commitCode'].extend(codes)
break
except:
print 'File Fail 0:', falseLink[1]
fr_linkList.append(fr_res)
writeToCorpus(textCorpus, codeCorpus, fr_res['commitText'], fr_res['commitCode'])
writeToCorpus(textCorpus, codeCorpus, fr_res['issueText'], fr_res['issueCode'])
index += 1
res = json.dumps(my_linkList, encoding="utf-8", indent=4)
trainSet = open('%s/codetrain%d-%d.dat' % (my_folder, repoId, index), "w")
trainSet.write(res)
trainSet.close()
print '%s/codetrain%d-%d.dat' % (my_folder, repoId, index), 'Over'
fres = json.dumps(fr_linkList, encoding="utf-8", indent=4)
ftrainSet = open('%s/traincase%d-%d.dat' % (fr_folder, repoId, index), "w")
ftrainSet.write(fres)
ftrainSet.close()
print '%s/traincase%d-%d.dat' % (fr_folder, repoId, index), 'Over'
trueStart += trueGap
falseStart += falseGap
trueLinkList = linkOperator.selectInScope((trueTable, trueStart, trueStart + trueCount))
falseLinkList = getRandomFalse(falseTable, falseStart, falseStart + falseGap, falseCount)
except IOError, e:
print "***", e
print traceback.format_exc()
def buildIssueAndCommit():
repos = linkOperator.selectOneRepo(50904245)
# repos = linkOperator.selectRepoOver(5000)
textCorpus = open('text50904245.dat', "w")
codeCorpus = open('code50904245.dat', "w")
try:
print 'start'
for highRepo in repos:
try:
# commit part
path = getPath(highRepo[1])
gitRe = gitResolver.GitResolver(path)
commits = gitRe.getCommits()
print path, ":", len(commits)
for commit in commits:
words = preprocessor.preprocessToWord(
commit.message.decode('utf-8'))
if len(words):
# 不是空列表
for word in words:
textCorpus.write(word.encode('utf-8'))
textCorpus.write(" ")
textCorpus.write("\n")
diffs = gitRe.getOneDiff(commit)
for diff in diffs:
diffCode = preprocessor.processDiffCode(diff.diff)
preDiffCode = preprocessor.processPreDiffCode(
diff.diff)
if len(diffCode):
for code in diffCode:
codeCorpus.write(code)
codeCorpus.write(" ")
codeCorpus.write("\n")
if len(preDiffCode):
for code in preDiffCode:
codeCorpus.write(code)
codeCorpus.write(" ")
codeCorpus.write("\n")
# issue part
issues = mysqlOperator.selectAllIssueInOneRepo(highRepo[0])
print highRepo[0], ":", len(issues)
for issue in issues:
titleWords = preprocessor.preprocessToWord(
issue[4].decode('utf-8'))
if len(titleWords):
# 不是空列表
for word in titleWords:
textCorpus.write(word.encode('utf-8'))
textCorpus.write(" ")
textCorpus.write("\n")
if issue[5]:
body = preprocessor.processHTML(
issue[5].decode('utf-8'))
bodyWords = body[1]
codeWords = body[0]
if len(bodyWords):
# 不是空列表
for word in bodyWords:
textCorpus.write(word.encode('utf-8'))
textCorpus.write(" ")
textCorpus.write("\n")
if len(codeWords):
# 不是空列表
for word in codeWords:
codeCorpus.write(word.encode('utf-8'))
codeCorpus.write(" ")
codeCorpus.write("\n")
comments = mysqlOperator.selectCommentInOneIssue(issue[1])
for comment in comments:
temp = preprocessor.processHTML(
comment[4].decode('utf-8'))
cBodyWords = temp[1]
cCodeWords = temp[0]
if len(cBodyWords):
# 不是空列表
for word in cBodyWords:
textCorpus.write(word.encode('utf-8'))
textCorpus.write(" ")
textCorpus.write("\n")
if len(cCodeWords):
# 不是空列表
for word in cCodeWords:
codeCorpus.write(word.encode('utf-8'))
codeCorpus.write(" ")
codeCorpus.write("\n")
except BaseException, e:
print "***", highRepo[0], ":", e
print traceback.format_exc()
print 'end'
comments = mysqlOperatorCopy.selectAllCommentInOneRepoDate(cou)
issueSet = []
for issue in issues:
if issue[1] not in issueSet:
issueSet.append(issue[1])
for comment in comments:
if comment[1] not in issueSet:
issueSet.append(comment[1])
return issueSet
projects = linkOperator.selectRepoOver(5000)
print 'start'
for repo in projects:
try:
gitRepo = gitResolver.GitResolver(getPath(repo[1]))
commits = gitRepo.getCommits()
print '==============', getPath(repo[1]), 'Start'
for commit in commits:
commitSha = str(commit.hexsha.encode("utf-8"))
print commitSha
commitIssues = mysqlOperatorCopy.selectExistIssueOnCommit(
(repo[0], commitSha))
trueLinks = []
for ci in commitIssues:
trueLinks.append(ci[0])
issueByDate = getIssueInDate(
(repo[0], str(gitRepo.getDateTime(commit)),
str(gitRepo.getDateTime(commit))))
for i in issueByDate:
if len(commitIssues) > 0:
"""
dot_val = 0.0
a_norm = 0.0
b_norm = 0.0
for a, b in zip(a_vect, b_vect):
dot_val += a * b
a_norm += a ** 2
b_norm += b ** 2
if a_norm == 0.0 or b_norm == 0.0:
return -1
else:
return dot_val / ((a_norm * b_norm) ** 0.5)
repoMap = {}
repoMap[12983151L] = gitResolver.GitResolver('/home/fdse/data/prior_repository/openhab/openhab1-addons')
# repos = mysqlOperator.selectAllHighRepository()
# for repo in repos:
# print type(repo[0])
# try:
# repoMap[repo[0]] = gitResolver.GitResolver(getPath(repo[1]))
# except:
# repoMap[repo[0]] = None
# TRUE_LINK_TOTAL = linkOperator.count('true_link')
# FALSE_LINK_TOTAL = linkOperator.count('false_link')
TRUE_GAP = 559
FALSE_GAP = 15000
TRUE_COUNT = 559
FALSE_COUNT = 600