def parseBuildOutput(dArgs, bDebug=False):
dFiles = {
'returnCode': 'retcode.log',
'buildTime': 'runtime.log',
#'dmesg' : 'dmesg.log',
#'stdout' : 'stdout.log',
#'stderr' : 'stderr.log',
'numObjects': 'numObjects.log'
}
dBuffer = {}
for sFileType, sFileName in dFiles.iteritems():
sFileName = os.path.join(dArgs['dirs']['output'], sFileName)
if os.path.isfile(sFileName):
with open(sFileName, 'r') as fBuilderFile:
# get file input and trim unnecessary whitespace before/after
dBuffer[sFileType] = (fBuilderFile.read()).strip()
else:
dBuffer[sFileType] = ''
if bDebug:
debug('func: parseBuildOutput() dBuffer:', json.dumps(dBuffer,
indent=4))
return dBuffer
def findProjects(sCorpusPath, dConfig):
qRedis = RedisQueue(dConfig['redis-queue-project-paths'], namespace='queue', host=dConfig['redis-loc'], port=dConfig['redis-port'])
iCount = 0
for sRoot, lDirs, lFiles in os.walk(sCorpusPath):
iLevel = sRoot.count(os.sep)
if iLevel >= 11:
del lDirs[:]
if iLevel == 11:
if dConfig['debug']: debug('func: findProjects()', 'projects-root:', sRoot, iLevel)
debug('func: findProjects()', 'projects-root:', sRoot, iLevel)
qRedis.put(sRoot)
iCount += 1
if dConfig['debug'] and iCount >= 10: break
printMsg('func: findProjects()', str(iCount), 'projects loaded into queue for processing')
def loadExistingBuilds(dConfig, sOS):
dMp = MuseProjectDB(db=dConfig['mysql-db'],
port=dConfig['mysql-port'],
user=dConfig['mysql-user'],
passwd=dConfig['mysql-passwd'],
loc=dConfig['mysql-loc'])
dMp.open()
sExistingBuilds = RedisSet(name=dConfig['redis-already-built'],
namespace='set',
host=dConfig['redis-loc'],
port=dConfig['redis-port'])
sExistingBuilds.flush()
lProjectRows = dMp.select(sSelectClause='projectName',
sTable='builtWith_' + sOS,
bDebug=dConfig['debug'])
dMp.close()
# for tProjectRow in lProjectRows:
# (sProjectName, ) = tProjectRow
# sExistingBuilds.put(sProjectName)
debug('func: loadRebuildSet()',
sOS + ' has ' + str(len(sExistingBuilds)) + ' built projects')
def main(argv):
iForks = 10
iStart = time.time()
### setup consumers
lConsumerArgs = []
# create a locking semaphore for mutex
lock = multiprocessing.Lock()
for iCtr in range(0, iForks):
lConsumerArgs.append(("lock testing procId", iCtr))
# create pool of workers -- number of workers equals the number of search strings to be processed
oConsumerPool = multiprocessing.Pool(processes=iForks,
initializer=initialize_lock,
initargs=(lock, ))
### do work -- use pool of workers to search for each search string in muse-corpus-source es index
oConsumerPool.map(test, lConsumerArgs)
oConsumerPool.close()
oConsumerPool.join()
# processBuildTargets( (dSearchStrings[ dConfig['queueBuildType'] ], 0, dArgs, dConfig) )
debug('func: main()', "all processes completed")
iEnd = time.time()
printMsg('func: main()', 'execution time:', (iEnd - iStart), 'seconds')
def findProjects(qRedis, sCorpusPath, dConfig):
lProjectPaths = []
iCount = 0
for sRoot, lDirs, lFiles in os.walk(sCorpusPath):
iLevel = sRoot.count(os.sep)
if iLevel >= 11:
del lDirs[:]
if iLevel == 11 and "github" not in sRoot:
if dConfig['debug']:
debug('func: findProjects()', 'projects-root:', sRoot, iLevel)
if dConfig['redis']:
qRedis.put(sRoot)
else:
lProjectPaths.append(sRoot)
iCount += 1
if dConfig['debug'] and iCount >= 10: break
printMsg('func: findProjects()', str(iCount),
'projects loaded into queue for processing')
return lProjectPaths
def copySource(dArgs, bDebug=False):
sCmd = 'rsync -a ' + dArgs['projectPath'] + '/ ' + dArgs['dirs'][
'source'] + '/'
if bDebug: debug('func: copySource() copy source for container:', sCmd)
os.system(sCmd)
def removeContainer(dArgs, bDebug=False):
sCmd = 'docker rm ' + dArgs['containerName']
if bDebug:
debug('func: removeContainer() removing container post build:', sCmd)
os.system(sCmd)
def recordProjectName(dArgs, bDebug=False):
# tar up container directory
sCmd = 'echo \"' + dArgs['projectName'] + '\" > ' + os.path.join(
dArgs['dirs']['output'], 'projectName.log')
if bDebug: debug('func: recordProjectName() recoding project name:', sCmd)
os.system(sCmd)
def copySource(dArgs, bDebug=False):
#sCmd = 'rsync -a ' + dArgs['projectPath'] + ' ' + dArgs['dirs']['source'] + '/'
sCmd = 'tar xzf ' + dArgs[
'projectPath'] + ' --exclude=\'.git\' --exclude=\'.svn\' -C ' + dArgs[
'dirs']['source'] + '/'
debug('func: copySource() unpack source for container:', sCmd)
if bDebug: debug('func: copySource() unpack source for container:', sCmd)
os.system(sCmd)
def copyScripts(dArgs, bDebug):
sCmd = 'rsync -a ' + dArgs['buildScripts'][
dArgs['buildType']] + ' ' + dArgs['dirs']['scripts'] + '/' + dArgs[
'script-name'] + ' && '
sCmd += 'rsync -a ' + dArgs['buildScripts']['loader'] + ' ' + dArgs[
'dirs']['scripts'] + '/'
if bDebug: debug('func: copyScripts() copy script for container:', sCmd)
os.system(sCmd)
def test(tTup):
(sMsg, iProcId) = tTup
for iCtr in range(0, 5):
lock.acquire()
time.sleep(1)
debug('func: test():', sMsg, iProcId, 'msg #', iCtr)
lock.release()
def pollBuild(dArgs, bDebug=False):
sStatus = ''
bStatus = False
sDockerStatus = subprocess.check_output(['docker', 'ps', '-a'])
if bDebug:
debug('func: startBuild() docker ps -a output:\n', sDockerStatus)
for sLine in sDockerStatus.split('\n'):
if bDebug:
debug('func: startBuild() parsed docker ps -a output:', sLine)
if dArgs['containerName'] in sLine:
sStatus = sLine
if bDebug: debug('func: pollBuild() container building status:', sStatus)
if 'Exited (' not in sStatus:
bStatus = True
if bDebug: debug('func: pollBuild() container building:', bStatus)
return bStatus
def writeBuildSummaries(dConfig):
qRedis = RedisQueue(dConfig['redis-queue-json'],
namespace='queue',
host=dConfig['redis-loc'],
port=dConfig['redis-port'])
while 1:
# get next project summary to process
sProjectSummary = qRedis.get(block=True, timeout=30)
if sProjectSummary:
# do something with summary
dProjectSummary = json.loads(sProjectSummary)
#sBuildPath = os.path.relpath(dProjectSummary['sourcePath'], '/nfscorpus/nfscorpus')
#sBuildPath = os.path.join('/nfsbuild/nfsbuild', sBuildPath)
if "_8tof" in dProjectSummary['sourcePath']:
sBuildPath = os.path.relpath(dProjectSummary['sourcePath'],
'/data/corpus_8tof')
sBuildPath = os.path.join('/data/builder_SAN/outputCyber',
sBuildPath)
if "_0to7" in dProjectSummary['sourcePath']:
sBuildPath = os.path.relpath(dProjectSummary['sourcePath'],
'/data/corpus_0to7')
sBuildPath = os.path.join('/data/builder_SAN/outputCyber',
sBuildPath)
(sBuildPath, _) = os.path.split(sBuildPath)
# ensure build directory exists
sCmd = 'mkdir -p ' + sBuildPath
if dConfig['debug']:
debug('func: writeBuildSummaries() mkdir cmd:', sCmd)
os.system(sCmd)
sJsonPath = os.path.join(sBuildPath, 'build.json')
if dConfig['debug']:
debug('func: writeBuildSummaries() sJsonPath:', sJsonPath)
with open(sJsonPath, 'w') as fJson:
fJson.write(json.dumps(dProjectSummary, indent=4))
else:
break
def copyScripts(dArgs, bDebug):
sCmd = ''
for dTarget in dArgs['targets']:
if dTarget['buildType'] not in dArgs['source-compilers'].keys():
sCmd = 'rsync -a ' + dArgs['buildScripts'][dTarget[
'buildType']] + ' ' + dArgs['dirs']['scripts'] + '/ && '
sCmd += 'rsync -a ' + dArgs['buildScripts']['loader'] + ' ' + dArgs[
'dirs']['scripts'] + '/'
if bDebug: debug('func: copyScripts() copy script for container:', sCmd)
os.system(sCmd)
def removeContainer(dArgs, bDebug=False):
sCmd = 'docker rm ' + dArgs['containerName']
if bDebug:
debug('func: removeContainer() removing container post build:', sCmd)
# enter mutex protected region
lock.acquire()
os.system(sCmd)
# sleep for 2 seconds in protected region to serialize calls to docker daemon
time.sleep(2)
# exit mutex protected region
lock.release()
def findProjects(sCorpusPath, iForks, dConfig):
lProjectPaths = []
if dConfig['redis']:
qRedis = RedisQueue(dConfig['redis-queue-name'],
namespace='queue',
host=dConfig['redis-loc'])
# ensure redis queue is empty prior to starting consumers
qRedis.flush()
iCount = 0
for sRoot, lDirs, lFiles in os.walk(sCorpusPath):
iLevel = sRoot.count(os.sep)
if iLevel >= 11:
del lDirs[:]
if iLevel == 11:
if dConfig['debug']:
debug('func: findProjects()', 'projects-root:', sRoot, iLevel)
if dConfig['redis']:
qRedis.put(sRoot)
else:
lProjectPaths.append(sRoot)
iCount += 1
if dConfig['debug'] and iCount >= 1: break
printMsg('func: findProjects()', str(iCount),
'projects loaded into queue for processing')
return lProjectPaths
def postBuildStatusUpdates(dArgs, bjson, dConfig):
dBuildArgs = {}
dMp = MuseProjectDB(db=dConfig['mysql-db'],
port=dConfig['mysql-port'],
user=dConfig['mysql-user'],
passwd=dConfig['mysql-passwd'],
loc=dConfig['mysql-loc'])
dBuildArgs['projectName'] = bjson['projectName']
dBuildArgs['projectPath'] = bjson['sourcePath']
dBuildArgs['buildTarPath'] = bjson['builds'][0]['buildTarPath']
dBuildArgs['targets'] = bjson['builds'][0]['targets']
# dBuildArgs['builder'] = bjson['containerName']
dBuildArgs['buildTime'] = bjson['builds'][0]['buildTime']
dBuildArgs['version'] = bjson['builds'][0]['version']
dBuildArgs['os'] = bjson['builds'][0]['os']
dBuildArgs['numObjectsPreBuild'] = bjson['builds'][0]['numObjectsPreBuild']
dBuildArgs['numObjectsPostBuild'] = bjson['builds'][0][
'numObjectsPostBuild']
dBuildArgs['numObjectsGenerated'] = bjson['builds'][0][
'numObjectsGenerated']
dBuildArgs['numSources'] = bjson['builds'][0]['numSources']
dBuildArgs['returnCode'] = bjson['builds'][0]['targets'][0]['returnCode']
#debug("BuildArgs: ", dBuildArgs)
if dConfig['debug']:
debug(
'func: postBuildStatusUpdates() build args prepared for mysql ingestion'
)
# commit status to database
dMp.open()
dMp.insertIntoBuildStatusTargets(dArgs=dBuildArgs, bDebug=dConfig['debug'])
dMp.insertIntoBuildStatus(dArgs=dBuildArgs, bDebug=dConfig['debug'])
dMp.close()
if dConfig['debug']:
debug(
'func: postBuildStatusUpdates() build status ingested into mysql')
def makeDirs(dArgs, bDebug=False):
lCmds = []
# initialize -- ensure old container directories aren't there
# remove container directory
sCmd = 'rm -rf ' + os.path.join(dArgs['containerPath'],
dArgs['containerName'])
lCmds.append(sCmd)
for sDirKey, sDirName in dArgs['dirs'].iteritems():
lCmds.append('mkdir -p ' + sDirName)
if bDebug:
debug('func: makeDirs() making dirs for container:',
json.dumps(lCmds, indent=4))
for sCmd in lCmds:
os.system(sCmd)
def tarUpContainerDirs(dArgs, bDebug=False):
# tar up container directory
sCmd = 'cd ' + dArgs['containerPath'] + ' && tar -zcf ' + dArgs[
'tarName'] + ' ' + dArgs['containerName'] + ' && '
# make project-specifc build directory if it does not exist
sCmd += 'mkdir -p ' + dArgs['buildPath'] + ' && '
# move tar to build directory
sCmd += 'mv ' + os.path.join(
dArgs['containerPath'],
dArgs['tarName']) + ' ' + dArgs['buildPath'] + ' && '
# remove container directory
sCmd += 'rm -rf ' + os.path.join(dArgs['containerPath'],
dArgs['containerName'])
if bDebug:
debug('func: tarUpContainerDirs() taring up container dirs:', sCmd)
os.system(sCmd)
def changePerms(tTup):
(sProjectPath, dConfig) = tTup
sLatestDir = os.path.join(sProjectPath, 'latest')
if os.path.exists(sLatestDir):
if os.path.isdir(sLatestDir):
# project-path/latest exists as a directory
if dConfig['debug']:
debug('func: changePerms()',
'changing directory permissions on', sLatestDir)
# change directory permissions to 555
os.system('find ' + sLatestDir +
' -type d -exec chmod 555 \'{}\' \;')
if dConfig['debug']:
debug('func: changePerms()', 'changing file permissions on',
sLatestDir)
# change file permissions to 4444
os.system('find ' + sLatestDir +
' -type f -exec chmod 444 \'{}\' \;')
else:
warning(
'func changePerms() latest exists but is not a directory under path:',
sLatestDir)
else:
warning('func changePerms() latest does not exist under path:',
sProjectPath, 'at', sLatestDir)
def postBuildStatusUpdates(dArgs, dBuffer, dConfig):
dBuildArgs = {}
dMp = MuseProjectDB(db=dConfig['mysql-db'],
port=dConfig['mysql-port'],
user=dConfig['mysql-user'],
passwd=dConfig['mysql-passwd'],
loc=dConfig['mysql-loc'])
lBuildTypes = dMp.getBuildTypes()
for sBuildType in lBuildTypes:
dBuildArgs[sBuildType] = False
dBuildArgs['projectName'] = dArgs['projectName']
dBuildArgs['projectPath'] = dArgs['projectPath']
dBuildArgs['buildTarPath'] = os.path.join(dArgs['buildPath'],
dArgs['tarName'])
dBuildArgs['buildTargetPath'] = dArgs['buildTargetPath']
dBuildArgs['builder'] = dArgs['containerName']
dBuildArgs['buildTime'] = dBuffer['buildTime']
#dBuildArgs['dmesg'] = dBuffer['dmesg']
dBuildArgs['version'] = dArgs['version']
dBuildArgs['os'] = dArgs['containerOS']
dBuildArgs['numObjects'] = dBuffer['numObjects']
dBuildArgs['returnCode'] = dBuffer['returnCode']
### troubleshoot serialization error
#dBuildArgs['stdout'] = dBuffer['stdout']
#dBuildArgs['stderr'] = dBuffer['stderr']
dBuildArgs[dArgs['buildType']] = True
if dConfig['debug']:
debug(
'func: postBuildStatusUpdates() build args prepared for es and mysql ingestion'
)
# commit status to elasticsearch
oES = Elasticsearch(dConfig['es-instance-locs'])
oES.index(index=dConfig['es-file-index-name'],
doc_type=dConfig['es-file-index-type'],
body=dBuildArgs,
timeout="20m",
request_timeout=600.)
if dConfig['debug']:
debug('func: postBuildStatusUpdates() build status ingested into es')
# commit status to database
dMp.open()
dMp.insertIntoBuildStatus(dArgs=dBuildArgs, bDebug=dConfig['debug'])
dMp.close()
if dConfig['debug']:
debug(
'func: postBuildStatusUpdates() build status ingested into mysql')
def startBuild(dArgs, bDebug=False):
#time.sleep( int(dArgs['containerId']) )
sCmd = 'docker run -d -m=' + dArgs[
'containerMem'] + ' --cpuset-cpus=' + dArgs['containerId']
sCmd += ' --name ' + dArgs['containerName']
sCmd += ' --ulimit nproc=2048:4096'
'''
VOLUME ["/buildArtifacts"]
VOLUME ["/output"]
VOLUME ["/scripts"]
VOLUME ["/source"]
'''
sCmd += ' -v ' + dArgs['dirs']['buildArtifacts'] + ':/buildArtifacts'
sCmd += ' -v ' + dArgs['dirs']['output'] + ':/output'
sCmd += ' -v ' + dArgs['dirs']['scripts'] + ':/scripts'
sCmd += ' -v ' + dArgs['dirs']['source'] + ':/source'
sCmd += ' ' + dArgs['imageName']
sCmd += ' /scripts/runBuild.sh'
if bDebug: debug('func: startBuild() starting container:', sCmd)
'''
use locking semaphore for mutex
noticing weird docker container spawning issues when containers are started simulateneously by multiple processes
'''
# enter mutex protected region
lock.acquire()
os.system(sCmd)
# sleep for 2 seconds in protected region to serialize calls to docker daemon
time.sleep(2)
# exit mutex protected region
lock.release()
def pollBuild(dArgs, bDebug=False):
sStatus = ''
bStatus = False
# enter mutex protected region
lock.acquire()
sDockerStatus = subprocess.check_output(['docker', 'ps', '-a'])
# sleep for 2 seconds in protected region to serialize calls to docker daemon
time.sleep(2)
# exit mutex protected region
lock.release()
if bDebug:
debug('func: startBuild() docker ps -a output:\n', sDockerStatus)
for sLine in sDockerStatus.split('\n'):
if bDebug:
debug('func: startBuild() parsed docker ps -a output:', sLine)
if dArgs['containerName'] in sLine:
sStatus = sLine
if bDebug: debug('func: pollBuild() container building status:', sStatus)
if 'Exited (' not in sStatus:
bStatus = True
if bDebug: debug('func: pollBuild() container building:', bStatus)
return bStatus
def main(argv):
# defaults
bError = False
dConfig = {}
dConfig['debug'] = False
dConfig['forks'] = 5
dConfig['mysql-db'] = 'muse'
dConfig['mysql-user'] = '******'
dConfig['mysql-passwd'] = 'muse'
dConfig['mysql-loc'] = 'muse2-int'
dConfig['mysql-port'] = 54321
dConfig['mysql'] = True
dConfig['redis-queue-json'] = 'muse-json'
dConfig['redis-set'] = 'muse-projects'
dConfig['redis-loc'] = 'muse2-int'
# dConfig['redis-port'] = '6379'
dConfig['redis-port'] = '12345'
dConfig['redis'] = True
### command line argument handling
options, remainder = getopt.getopt(sys.argv[1:], 'f:d',
['forks=', 'debug'])
# debug('func: main()', 'options:', options)
# debug('func: main()', 'remainder:', remainder)
for opt, arg in options:
if opt in ('-f', '--forks'):
try:
dConfig['forks'] = int(arg)
except ValueError as e:
bError = True
elif opt in ('-d', '--debug'):
dConfig['debug'] = True
debug('func: main()', 'dConfig:', json.dumps(dConfig, indent=4))
if bError: usage()
else:
iStart = time.time()
# prepare redis queue for producer, flush queue before starting the producer
qRedis = RedisQueue(dConfig['redis-queue-json'],
namespace='queue',
host=dConfig['redis-loc'],
port=dConfig['redis-port'])
qRedis.flush()
'''
# multi-process approach
# call producer process that populates redis queue with project path roots
pProducer = multiprocessing.Process( target=createBuildSummaries, args=(dConfig) )
pProducer.start()
### setup json writers
lConsumerArgs = []
for iCtr in range(0, dConfig['forks']):
lConsumerArgs.append( (dConfig) )
# create pool of workers
oConsumerPool = multiprocessing.Pool(processes=dConfig['forks'])
### do work -- use pool of workers to search for each search string in muse-corpus-source es index
oConsumerPool.map(writeBuildSummaries, lConsumerArgs)
# wait for the producer to complete
pProducer.join()
# wait for the consumer pool to complete
oConsumerPool.close()
oConsumerPool.join()
'''
'''
# single process approach:
'''
createBuildSummaries(dConfig)
writeBuildSummaries(dConfig)
if dConfig['debug']: debug('func: main()', "all processes completed")
iEnd = time.time()
printMsg('func: main()', 'execution time:', (iEnd - iStart), 'seconds')
def createBuildSummaries(dConfig):
qRedis = RedisQueue(dConfig['redis-queue-json'],
namespace='queue',
host=dConfig['redis-loc'],
port=dConfig['redis-port'])
dMp = MuseProjectDB(db=dConfig['mysql-db'],
port=dConfig['mysql-port'],
user=dConfig['mysql-user'],
passwd=dConfig['mysql-passwd'],
loc=dConfig['mysql-loc'])
sLimitClause = ''
if dConfig['debug']: sLimitClause = '10'
dReturnCodeLookup = {
'buildSuccess': 'success',
'buildPartial': 'partial',
'buildFail': 'fail'
}
sSelectClause = 'projectName,projectPath,buildTarPath,buildTime,version,os,numObjectsPreBuild,numObjectsPostBuild,numObjectsGenerated,numSources,buildTargetPath,configureBuildType,configureacBuildType,configureinBuildType,cmakeBuildType,makefileBuildType,antBuildType,mavenBuildType,returnCode'
lTargetTypes = [
'configureBuildType', 'configureacBuildType', 'configureinBuildType',
'cmakeBuildType', 'makefileBuildType', 'antBuildType', 'mavenBuildType'
]
dMp.open()
iProjectCount = 0
dProjects = {
'success':
RedisSet(dConfig['redis-set'] + '-success',
namespace='set',
host=dConfig['redis-loc'],
port=dConfig['redis-port']),
'partial':
RedisSet(dConfig['redis-set'] + '-partial',
namespace='set',
host=dConfig['redis-loc'],
port=dConfig['redis-port']),
'fail':
RedisSet(dConfig['redis-set'] + '-fail',
namespace='set',
host=dConfig['redis-loc'],
port=dConfig['redis-port'])
}
for sTable, sProjectBin in dReturnCodeLookup.iteritems():
# empty redis set
dProjects[sProjectBin].flush()
lProjects = dMp.select(sSelectClause='projectName',
sTable=sTable,
sOrderByClause='projectName',
sLimitClause=sLimitClause,
bDebug=dConfig['debug'])
# populate redis set with projects of each bin type
for tProject in lProjects:
(sProjectName, ) = tProject
dProjects[sProjectBin].put(sProjectName)
dProjectSummary = {}
lTargetRows = dMp.select(sSelectClause=sSelectClause,
sTable='buildStatusWithTargets',
sOrderByClause='projectName,buildTarPath',
sLimitClause=sLimitClause,
bDebug=dConfig['debug'])
for tTargetRow in lTargetRows:
dTarget = {}
(dTarget['projectName'], dTarget['projectPath'],
dTarget['buildTarPath'], dTarget['buildTime'], dTarget['version'],
dTarget['os'], dTarget['numObjectsPreBuild'],
dTarget['numObjectsPostBuild'], dTarget['numObjectsGenerated'],
dTarget['numSources'], dTarget['buildTargetPath'],
dTarget['configureBuildType'], dTarget['configureacBuildType'],
dTarget['configureinBuildType'], dTarget['cmakeBuildType'],
dTarget['makefileBuildType'], dTarget['antBuildType'],
dTarget['mavenBuildType'], dTarget['returnCode']) = tTargetRow
if dProjectSummary:
if dProjectSummary['projectName'] == dTarget['projectName']:
try:
(dBuild for dBuild in dProjectSummary['builds']
if dBuild['buildTarPath'] == dTarget['buildTarPath']
).next()
except (StopIteration) as e:
dBuild = {
'buildTarPath': dTarget['buildTarPath'],
'buildTime': dTarget['buildTime'],
'version': dTarget['version'],
'os': dTarget['os'],
'numObjectsPreBuild': dTarget['numObjectsPreBuild'],
'numObjectsPostBuild': dTarget['numObjectsPostBuild'],
'numObjectsGenerated': dTarget['numObjectsGenerated'],
'numSources': dTarget['numSources'],
'targets': []
}
dProjectSummary['builds'].append(dBuild)
dTargetSummary = {
'buildTargetPath': dTarget['buildTargetPath'],
'returnCode': dTarget['returnCode']
}
for sTargetType in lTargetTypes:
if dTarget[sTargetType] == 1:
dTargetSummary['target-type'] = sTargetType
break
dBuild['targets'].append(dTargetSummary)
else:
if dConfig['debug']:
debug('func: createBuildSummaries() dProjectSummary:',
json.dumps(dProjectSummary, indent=4))
qRedis.put(json.dumps(dProjectSummary))
iProjectCount += 1
dProjectSummary = {}
if not dProjectSummary:
# project specific build summary info
dBuild = {
'buildTarPath': dTarget['buildTarPath'],
'buildTime': dTarget['buildTime'],
'version': dTarget['version'],
'os': dTarget['os'],
'numObjectsPreBuild': dTarget['numObjectsPreBuild'],
'numObjectsPostBuild': dTarget['numObjectsPostBuild'],
'numObjectsGenerated': dTarget['numObjectsGenerated'],
'numSources': dTarget['numSources'],
'targets': []
}
dProjectSummary = {
'projectName': dTarget['projectName'],
'sourcePath': dTarget['projectPath'],
'builds': [dBuild]
}
if dTarget['projectName'] in dProjects['success']:
dProjectSummary['buildStatus'] = 'success'
elif dTarget['projectName'] in dProjects['partial']:
dProjectSummary['buildStatus'] = 'partial'
elif dTarget['projectName'] in dProjects['fail']:
dProjectSummary['buildStatus'] = 'fail'
# target specific build summary info
dTargetSummary = {
'buildTargetPath': dTarget['buildTargetPath'],
'returnCode': dTarget['returnCode']
}
for sTargetType in lTargetTypes:
if dTarget[sTargetType] == 1:
dTargetSummary['target-type'] = sTargetType
break
dBuild['targets'].append(dTargetSummary)
if dProjectSummary:
if dConfig['debug']:
debug('func: createBuildSummaries() dProjectSummary:',
json.dumps(dProjectSummary, indent=4))
qRedis.put(json.dumps(dProjectSummary))
iProjectCount += 1
dProjectSummary = {}
dMp.close()
printMsg('func: createBuildSummaries()', str(iProjectCount),
'projects queued')
def main(argv):
# defaults
bError = False
dConfig = {}
dConfig['containerImage'] = 'musebuilder'
#dConfig['containerPath'] = '/data/builder'
dConfig['containerPath'] = '/data/builder_SAN/containers'
dConfig['debug'] = False
dConfig['elasticsearch'] = True
dConfig['es-instance-locs'] = ['muse1-int', 'muse2-int', 'muse3-int']
#dConfig['es-instance-locs'] = ['muse2-int','muse3-int']
#dConfig['es-instance-locs'] = ['muse3-int']
#dConfig['es-file-index-name'] = 'muse-corpus-source'
dConfig['es-file-index-name'] = 'muse-corpus-build'
dConfig['es-file-index-type'] = 'muse-project-build'
dConfig['forks'] = 5
dConfig['hostname'] = socket.gethostname().replace('.', '')
dConfig['mysql-db'] = 'muse'
dConfig['mysql-user'] = '******'
dConfig['mysql-passwd'] = 'muse'
dConfig['mysql-loc'] = 'muse2-int'
dConfig['mysql-port'] = 54321
dConfig['mysql'] = True
dConfig['rebuild'] = False
dConfig['redis-already-built'] = 'muse-already-built-'
dConfig['redis-already-built-nate'] = 'NEWbuiltProjects'
dConfig['redis-queue-to-build'] = 'muse-to-build'
dConfig['redis-queue-building'] = 'muse-building'
dConfig['redis-loc'] = 'muse2-int'
# dConfig['redis-port'] = '6379'
dConfig['redis-port'] = '12345'
dConfig['redis'] = True
dArgs = {}
# number of attempts with each to build targets to resolve dependencies
dArgs['buildCycles'] = 2
dArgs['containerMem'] = '2g'
dArgs['buildScripts'] = {}
dArgs['buildScripts']['root'] = '/managed/scripts'
dArgs['buildScripts']['loader'] = os.path.join(
dArgs['buildScripts']['root'], 'runBuild.sh')
dArgs['buildScripts']['cmakeBuildType'] = os.path.join(
dArgs['buildScripts']['root'], 'cmake.sh')
dArgs['buildScripts']['configureBuildType'] = os.path.join(
dArgs['buildScripts']['root'], 'configure.sh')
dArgs['buildScripts']['configureacBuildType'] = os.path.join(
dArgs['buildScripts']['root'], 'configureac.sh')
dArgs['buildScripts']['configureinBuildType'] = os.path.join(
dArgs['buildScripts']['root'], 'configurein.sh')
dArgs['buildScripts']['makefileBuildType'] = os.path.join(
dArgs['buildScripts']['root'], 'make.sh')
dArgs['containerScripts'] = {}
dArgs['containerScripts']['root'] = '/scripts'
dArgs['containerScripts']['cmakeBuildType'] = os.path.join(
dArgs['containerScripts']['root'], 'cmake.sh')
dArgs['containerScripts']['configureBuildType'] = os.path.join(
dArgs['containerScripts']['root'], 'configure.sh')
dArgs['containerScripts']['configureacBuildType'] = os.path.join(
dArgs['containerScripts']['root'], 'configureac.sh')
dArgs['containerScripts']['configureinBuildType'] = os.path.join(
dArgs['containerScripts']['root'], 'configurein.sh')
dArgs['containerScripts']['makefileBuildType'] = os.path.join(
dArgs['containerScripts']['root'], 'make.sh')
dArgs['containerDirs'] = ['buildArtifacts', 'output', 'scripts', 'source']
dArgs['containerOS'] = 'ubuntu14'
dArgs['containerPath'] = dConfig['containerPath']
dArgs['imageName'] = dConfig['containerImage'] + '-' + dArgs['containerOS']
dArgs['script-name'] = 'build.sh'
'''
dArgs['build-targets'] = {
'configure' : 'configureBuildType',
'configure.ac' : 'configureacBuildType',
'configure.in' : 'configureinBuildType',
'CMakeLists.txt' : 'cmakeBuildType',
'Makefile' : 'makefileBuildType'
#'build.xml' : 'antBuildType',
#'pom.xml' : 'mavenBuildType'
}
'''
dArgs['source-compilers'] = {'cBuildType': 'gcc', 'cppBuildType': 'g++'}
'''
dArgs['source-targets'] = {
'.c' : 'cBuildType',
'.cc' : 'cppBuildType',
'.cpp' : 'cppBuildType',
'.cxx' : 'cppBuildType',
'.c++' : 'cppBuildType'
}
'''
lSupportedOSs = ['fedora20', 'fedora21', 'ubuntu12', 'ubuntu14']
### command line argument handling
options, remainder = getopt.getopt(
sys.argv[1:], 'f:o:rdy',
['forks=', 'os=', 'rebuild', 'debug', 'debug-flags'])
# debug('func: main()', 'options:', options)
# debug('func: main()', 'remainder:', remainder)
for opt, arg in options:
if opt in ('-f', '--forks'):
try:
dConfig['forks'] = int(arg)
except ValueError as e:
bError = True
elif opt in ('-o', '--os'):
if arg in lSupportedOSs:
dArgs['containerOS'] = arg
dArgs['imageName'] = dConfig['containerImage'] + '-' + dArgs[
'containerOS']
else:
bError = True
elif opt in ('-r', '--rebuild'):
dConfig['rebuild'] = True
elif opt in ('-d', '--debug'):
dConfig['debug'] = True
elif opt in ('-y', '--debug-flags'):
dArgs['source-compilers'] = {
'cBuildType': 'gcc -g3 -O0 -DDEBUG',
'cppBuildType': 'g++ -g3 -O0 -DDEBUG'
}
debug('func: main()', 'dConfig:', json.dumps(dConfig, indent=4))
if bError: usage()
else:
'''
# pre-initialization -- if projects remained in building queue, put them back in queue-to-build
qToBuildRedis = RedisQueue(name=dConfig['redis-queue-building'], name2=dConfig['redis-queue-to-build'], namespace='queue', host=dConfig['redis-loc'], port=dConfig['redis-port'])
for iCtr in range(0, len(qToBuildRedis)):
qToBuildRedis.getnpush()
'''
dConfig['redis-already-built'] = dConfig[
'redis-already-built'] + dArgs['containerOS']
sExistingBuilds = RedisSet(name=dConfig['redis-already-built'],
namespace='set',
host=dConfig['redis-loc'],
port=dConfig['redis-port'])
sExistingBuilds.flush()
if not dConfig['rebuild']:
loadExistingBuilds(dConfig, dArgs['containerOS'])
iStart = time.time()
### setup consumers
lConsumerArgs = []
# create a locking semaphore for mutex
lock = multiprocessing.Lock()
for iCtr in range(0, dConfig['forks']):
lConsumerArgs.append((iCtr, dArgs, dConfig))
# create pool of workers -- number of workers equals the number of search strings to be processed
oConsumerPool = multiprocessing.Pool(processes=dConfig['forks'],
initializer=initialize_lock,
initargs=(lock, ))
### do work -- use pool of workers to search for each search string in muse-corpus-source es index
print(lConsumerArgs)
oConsumerPool.map(processBuildTargets, lConsumerArgs)
oConsumerPool.close()
oConsumerPool.join()
# processBuildTargets( (0, dArgs, dConfig) )
if dConfig['debug']: debug('func: main()', "all processes completed")
iEnd = time.time()
printMsg('func: main()', 'execution time:', (iEnd - iStart), 'seconds')
def findProjects(sLanguage, dConfig):
# setup elasticsearch client
oES = Elasticsearch(dConfig['es-instance-locs'])
lProjects = []
iCtr = 0
dQuery = {"query": {"match_all": {}}, "fields": [sLanguage]}
if dConfig['debug']:
debug('func: findProjects() dQuery:', json.dumps(dQuery))
# scroll time set to 10 minutes, change as needed -- required for consistent results, the scroll token expires at the end of scroll time
dResponse = oES.search(index=dConfig['es-project-index-name'],
doc_type=dConfig['es-project-index-type'],
body=json.dumps(dQuery),
search_type='scan',
scroll='20m',
timeout='20m',
lowercase_expanded_terms=False)
sScrollId = dResponse['_scroll_id']
if dConfig['debug']:
debug('func: findProjects() (after initial search) dResponse: ',
dResponse)
if dConfig['debug']:
debug('func: findProjects() search hits: ', dResponse['hits']['total'])
#while not dResponse['timed_out'] and dResponse['hits']['hits']['total'] > 0:
while 'timed_out' in dResponse and not dResponse[
'timed_out'] and 'hits' in dResponse and 'total' in dResponse[
'hits'] and dResponse['hits']['total'] > 0:
dResponse = oES.scroll(scroll_id=sScrollId, scroll='20m')
sScrollId = dResponse['_scroll_id']
if ('hits'
in dResponse['hits']) and (len(dResponse['hits']['hits']) > 0):
if dConfig['debug']:
debug('func: findProjects() scroll_id:', sScrollId,
'number of hits:', len(dResponse['hits']['hits']))
if dConfig['debug'] and iCtr > 10: break
for dHit in dResponse['hits']['hits']:
iCtr += 1
if dConfig['debug']:
debug('func: findProjects()', json.dumps(dHit, indent=4))
if iCtr > 100: break
# found matches
if 'fields' in dHit and sLanguage in dHit[
'fields'] and '_id' in dHit:
lProjects.append(dHit['_id'])
else:
break
printMsg('func: findProjects() found ', str(iCtr),
' buildTargets, spawned process exiting...')
sLanguageFileName = './' + sLanguage.split('.')[1] + '.txt'
printMsg('func: findProjects() file created: ', sLanguageFileName)
with open(sLanguageFileName, 'w') as fLanguage:
for sProject in sorted(lProjects):
fLanguage.write(sProject + '\n')
return lProjects
def main(argv):
# defaults
sCorpusPath = '/data/corpus'
dConfig = {}
dConfig['debug'] = False
dConfig['redis-queue-name'] = 'muse-project-paths-perms'
dConfig['redis-loc'] = '38.100.20.212'
dConfig['redis'] = False
dConfig['time-stamp'] = datetime.datetime.now().strftime(
'%Y-%m-%dT%H:%M:%S')
iForks = 10
bError = False
### command line argument handling
options, remainder = getopt.getopt(
sys.argv[1:], 'c:f:rd',
['corpus-dir-path=', 'forks=', 'redis', 'debug'])
# debug('func: main()', 'options:', options)
# debug('func: main()', 'remainder:', remainder)
for opt, arg in options:
if opt in ('-c', '--corpus-dir-path'):
sCorpusPath = arg
elif opt in ('-d', '--debug'):
dConfig['debug'] = True
elif opt in ('-r', '--redis'):
dConfig['redis'] = True
elif opt in ('-f', '--forks'):
try:
iForks = int(arg)
except ValueError as e:
bError = True
if not os.path.isdir(sCorpusPath):
bError = True
if bError: usage()
else:
iStart = time.time()
### setup producer
lProjectPaths = []
if dConfig['redis']:
# call producer process that populates redis queue with project path roots
pProducer = multiprocessing.Process(target=findProjects,
args=(sCorpusPath, iForks,
dConfig))
pProducer.start()
else:
lProjectPaths = findProjects(sCorpusPath, iForks, dConfig)
### setup consumers
lArgs = []
# create pool of workers
oPool = multiprocessing.Pool(processes=iForks)
if dConfig['redis']:
for i in range(0, iForks):
lArgs.append(dConfig)
### do work -- use pool of workers to descend into each project path recording/ingesting all file names
oPool.map(processProjects, lArgs)
pProducer.join()
else:
for sPath in lProjectPaths:
lArgs.append((sPath, dConfig))
### do work -- use pool of workers to descend into each project path recording/ingesting all file names
oPool.map(findProjectFiles, lArgs)
oPool.close()
oPool.join()
if dConfig['debug']: debug('func: main()', "all processes completed")
iEnd = time.time()
printMsg('func: main()', 'execution time:', (iEnd - iStart), 'seconds')
def main(argv):
# defaults
sCorpusPath = '/data/builder_SAN2/RAT'
# sCorpusPath = '/data/corpus_0to7'
# sCorpusPath = '/data/corpus_8tof'
dConfig = {}
dConfig['es-bulk-chunk-size'] = 500
dConfig['debug'] = False
# binding to muse2 doesn't work right now
dConfig['es-instance-locs'] = ['muse1-int', 'muse2-int', 'muse3-int']
#dConfig['es-instance-locs'] = ['muse2-int','muse3-int']
#dConfig['es-instance-locs'] = ['muse3-int']
dConfig['es-index-name'] = 'rat-corpus-source'
dConfig['es-index-type'] = 'files'
dConfig['redis-queue-name'] = 'rat-project-paths'
dConfig['redis-loc'] = 'muse2-int'
dConfig['redis-port'] = '12345'
dConfig['redis'] = False
dConfig['time-stamp'] = datetime.datetime.now().strftime(
'%Y-%m-%d %H:%M:%S')
iForks = 5
bError = False
### command line argument handling
options, remainder = getopt.getopt(
sys.argv[1:], 'c:f:rd',
['corpus-dir-path=', 'forks=', 'redis', 'debug'])
# debug('func: main()', 'options:', options)
# debug('func: main()', 'remainder:', remainder)
for opt, arg in options:
if opt in ('-c', '--corpus-dir-path'):
sCorpusPath = arg
elif opt in ('-d', '--debug'):
dConfig['debug'] = True
elif opt in ('-r', '--redis'):
dConfig['redis'] = True
elif opt in ('-f', '--forks'):
try:
iForks = int(arg)
except ValueError as e:
bError = True
if not os.path.isdir(sCorpusPath):
bError = True
if bError: usage()
else:
iStart = time.time()
#oES = createESIndex(dConfig)
oES = Elasticsearch(dConfig['es-instance-locs'])
### setup producer
lProjectPaths = []
if dConfig['redis']:
qRedis = RedisQueue(dConfig['redis-queue-name'],
namespace='queue',
host=dConfig['redis-loc'],
port=dConfig['redis-port'])
# ensure redis queue is empty prior to starting consumers
# qRedis.flush()
# call producer process that populates redis queue with project path roots
pProducer = multiprocessing.Process(target=findProjects,
args=(qRedis, sCorpusPath,
dConfig))
pProducer.start()
else:
lProjectPaths = findProjects(None, sCorpusPath, dConfig)
### setup consumers
lArgs = []
iForks = 1
if dConfig['redis']:
# create pool of workers
oPool = multiprocessing.Pool(processes=iForks)
for i in range(0, iForks):
lArgs.append(dConfig)
### do work -- use pool of workers to descend into each project path recording/ingesting all file names
oPool.map(processProjects, lArgs)
pProducer.join()
oPool.close()
oPool.join()
else:
for sPath in lProjectPaths:
findProjectFiles((sPath, oES, dConfig))
if dConfig['debug']: debug('func: main()', "all processes completed")
# es index was created with replication turned off for speed, turn on replicating shards
turnReplicationOn(oES, dConfig)
# refresh to make the documents available for search
oES.indices.refresh(index=dConfig['es-index-name'])
# and now we can count the documents
printMsg('func: main()', 'number of documents in',
dConfig['es-index-name'], 'index: ',
oES.count(index=dConfig['es-index-name'])['count'])
iEnd = time.time()
printMsg('func: main()', 'execution time:', (iEnd - iStart), 'seconds')
def findProjectFiles(tTup):
(sProjectPath, oES, dConfig) = tTup
sProjectName = os.path.basename(sProjectPath)
oES = Elasticsearch(dConfig['es-instance-locs'])
lIgnoreDirs = ['.git', '.svn']
lProjectFiles = []
if dConfig['debug']:
debug('func: findProjectFiles()', 'project-path:', sProjectPath,
'project-name:', sProjectName)
for sRoot, lDirs, lFiles in os.walk(sProjectPath):
if len(lProjectFiles) > dConfig['es-bulk-chunk-size']:
# ingest chunk into elasticsearch
(iSuccess, lResponse) = helpers.bulk(client=oES,
actions=lProjectFiles,
timeout="20m",
request_timeout=120.)
if iSuccess < dConfig['es-bulk-chunk-size']:
warning('func: findProjectFiles() iSuccess:', iSuccess,
' expected:', dConfig['es-bulk-chunk-size'])
warning('func: findProjectFiles()', type(lResponse),
'returned by bulk api')
warning('func: findProjectFiles()',
json.dumps(lResponse, indent=4),
'returned by bulk api')
#del lProjectFiles[0 : len(lProjectFiles)]
lProjectFiles = []
if dConfig['debug']:
debug('func: findProjectFiles()', str(len(lProjectFiles)),
'files loaded into elasticsearch')
for sFile in lFiles:
# make sure dProject is emptied each loop iteration
dProject = {
'_index': dConfig['es-index-name'],
'_type': dConfig['es-index-type'],
'_source': {
'project-path': sProjectPath,
'project-name': sProjectName,
'crawl-time': dConfig['time-stamp']
}
}
sFilePath = os.path.join(sRoot, sFile)
sRelPath = os.path.relpath(sFilePath, sProjectPath)
sDecodedFile = ''
sDecodedRelPath = ''
sEncodedWith = ''
# Look for the tar file with the src code
if "_code.tgz" in sFilePath:
global counter
counter = counter + 1
print(str(counter) + ': working on: ' + sFilePath)
t = tarfile.open(sFilePath, 'r:*')
# Iterate over the files in the tar file gz
for tarinfo in t:
if tarinfo.isfile():
filename = tarinfo.name
if (".svn" not in filename and ".git" not in filename):
# make sure dProject is emptied each loop iteration
dProject = {
'_index': dConfig['es-index-name'],
'_type': dConfig['es-index-type'],
'_source': {
'project-path': sProjectPath,
'project-name': sProjectName,
'crawl-time': dConfig['time-stamp']
}
}
# append file in tar to tar path
sFile = os.path.join(sFilePath, filename)
sRelPath = os.path.relpath(sFile, sProjectPath)
sDecodedFile = ''
sDecodedRelPath = ''
sEncodedWith = ''
try:
sDecodedFile = sFile.decode('utf-8')
sDecodedRelPath = sRelPath.decode('utf-8')
sEncodedWith = 'utf-8'
except (ValueError, UnicodeDecodeError) as e:
try:
sDecodedFile = sFile.decode('latin-1')
sDecodedRelPath = sRelPath.decode(
'latin-1')
sEncodedWith = 'latin-1'
except (ValueError, UnicodeDecodeError) as e:
try:
sDecodedFile = sFile.decode('utf-16')
sDecodedRelPath = sRelPath.decode(
'utf-16')
sEncodedWith = 'utf-16'
except (ValueError,
UnicodeDecodeError) as e:
warning(
'func findProjectFiles():',
'sProjectPath:',
dProject['_source']
['project-path'], 'sProjectName:',
dProject['_source']
['project-name'], 'sFile:', sFile,
'sRelPath:', sRelPath,
'utf-8, latin-1, and utf-16 decoding failed',
'exception:', e)
print("decode failed")
sDecodedFile = ''
sDecodedRelPath = ''
sEncodedWith = ''
if sDecodedFile and sDecodedRelPath:
dProject['_source']['file'] = sDecodedFile
(_, sFileExt) = os.path.splitext(sDecodedFile)
if sFileExt:
dProject['_source']['ext'] = sFileExt[
1:].lower()
dProject['_source']['path'] = sDecodedRelPath
if dConfig['debug']:
debug('func: findProjectFiles() dProject:',
dProject, 'encoded with',
sEncodedWith)
lProjectFiles.append(dProject)
lDirs[:] = [sDir for sDir in lDirs if sDir not in lIgnoreDirs]
# ingest any stragglers remaining into elasticsearch
(iSuccess, lResponse) = helpers.bulk(client=oES,
actions=lProjectFiles,
timeout="20m",
request_timeout=120.)
if iSuccess < len(lProjectFiles):
warning('func: findProjectFiles() iSuccess:', iSuccess, ' expected:',
len(lProjectFiles))
warning('func: findProjectFiles()', type(lResponse),
'returned by bulk api')
warning('func: findProjectFiles()', json.dumps(lResponse, indent=4),
'returned by bulk api')
# del lProjectFiles[0 : len(lProjectFiles)]
lProjectFiles = []