def testConvertPackageNameToClassName(self):
data = {'foo': 'Foo',
'foobar': 'Foobar',
'foo-bar': 'FooBar'}
for input, expectedOutput in data.iteritems():
self.assertEquals(util.convertPackageNameToClassName(input),
expectedOutput)
def testConvertPackageNameToClassName(self):
data = {'foo': 'Foo',
'foobar': 'Foobar',
'foo-bar': 'FooBar'}
for input, expectedOutput in data.iteritems():
self.assertEquals(util.convertPackageNameToClassName(input),
expectedOutput)
def createProductVersion(self, fqdn, version, namespace, description,
platformLabel):
product = self.getProduct(fqdn)
if not namespace:
namespace = product.namespace
projectId = product.productId
# Check the namespace
projects._validateNamespace(namespace)
# make sure it is a valid product version
projects._validateProductVersion(version)
# initial product definition
prodDef = helperfuncs.sanitizeProductDefinition(product.name,
description, product.repositoryHostname,
product.shortname, version,
'', namespace)
label = prodDef.getDefaultLabel()
# validate the label, which will be added later. This is done
# here so the project is not created before this error occurs
if projects.validLabel.match(label) == None:
raise mint_error.InvalidLabel(label)
if platformLabel:
cclient = self.reposMgr.getUserClient()
prodDef.rebase(cclient, platformLabel)
self.setProductVersionDefinition(fqdn, version, prodDef)
try:
versionId = self.db.db.productVersions.new(projectId=projectId,
namespace=namespace, name=version, description=description,
label=prodDef.getProductDefinitionLabel(),
timeCreated=time.time())
except mint_error.DuplicateItem:
raise mint_error.DuplicateProductVersion
if product.prodtype == 'Appliance' or product.prodtype == 'PlatformFoundation':
groupName = helperfuncs.getDefaultImageGroupName(product.hostname)
className = util.convertPackageNameToClassName(groupName)
# convert from unicode
recipeStr = str(templates.write(groupTemplate,
cfg = self.cfg,
groupApplianceLabel=platformLabel,
groupName=groupName,
recipeClassName=className,
version=version) + '\n')
self.reposMgr.createSourceTrove(fqdn, groupName,
label, version,
{'%s.recipe' % groupName: recipeStr},
'Initial appliance image group template')
return versionId
def derive(repos,
cfg,
targetLabel,
troveSpec,
checkoutDir=None,
extract=False,
info=False,
callback=None):
"""
Performs all the commands necessary to create a derived recipe.
First it shadows the package, then it creates a checkout of the shadow
and converts the checkout to a derived recipe package.
Finally if extract = True, it installs an version of the binary
package into a root.
@param repos: trovesource to search for and derive packages from
@param cfg: configuration to use when deriving the package
@type cfg: ConaryConfiguration object
@param targetLabel: label to derive from
@type targetLabel: versions.Label
@param checkoutDir: directory to create the checkout in. If None,
defaults to currentDir + packageName.
@param extract: If True, creates a subdirectory of the checkout named
_ROOT_ with the contents of the binary of the derived
package.
@param info: If true, only display the information about the shadow
that would be performed if the derive command were
completed.
@param callback:
"""
origDir = os.getcwd()
try:
if callback is None:
callback = DeriveCallback()
if isinstance(troveSpec, tuple):
troveName, versionSpec, flavor = troveSpec
versionSpec = str(versionSpec)
troveSpec = cmdline.toTroveSpec(troveName, versionSpec, flavor)
else:
troveName, versionSpec, flavor = cmdline.parseTroveSpec(troveSpec)
if isinstance(targetLabel, str):
targetLabel = Label(targetLabel)
troveName, versionSpec, flavor = cmdline.parseTroveSpec(troveSpec)
result = repos.findTrove(cfg.buildLabel,
(troveName, versionSpec, flavor), cfg.flavor)
# findTrove shouldn't return multiple items for one package anymore
# when a flavor is specified.
troveToDerive, = result
# displaying output along the screen allows there to be a record
# of what operations were performed. Since this command is
# an aggregate of several commands I think that is appropriate,
# rather than simply using a progress callback.
log.info('Shadowing %s=%s[%s] onto %s' %
(troveToDerive[0], troveToDerive[1], troveToDerive[2],
targetLabel))
if info:
cfg.interactive = False
error = branch.branch(repos,
cfg,
str(targetLabel), ['%s=%s[%s]' % troveToDerive],
makeShadow=True,
sourceOnly=True,
binaryOnly=False,
allowEmptyShadow=True,
info=info)
if info or error:
return
shadowedVersion = troveToDerive[1].createShadow(targetLabel)
shadowedVersion = shadowedVersion.getSourceVersion(False)
troveName = troveName.split(':')[0]
checkoutDir = checkoutDir or troveName
checkin.checkout(repos,
cfg,
checkoutDir, ["%s=%s" % (troveName, shadowedVersion)],
callback=callback)
os.chdir(checkoutDir)
nvfs = repos.getTrovesBySource(troveToDerive[0] + ':source',
troveToDerive[1].getSourceVersion())
trvs = repos.getTroves(nvfs)
hasCapsule = [x for x in trvs if x.troveInfo.capsule.type()]
if hasCapsule:
derivedRecipeType = 'DerivedCapsuleRecipe'
removeText = ''
else:
derivedRecipeType = 'DerivedPackageRecipe'
removeText = \
"""
# This appliance uses PHP as a command interpreter but does
# not include a web server, so remove the file that creates
# a dependency on the web server
r.Remove('/etc/httpd/conf.d/php.conf')
"""
log.info('Rewriting recipe file')
recipeName = troveName + '.recipe'
className = util.convertPackageNameToClassName(troveName)
derivedRecipe = """
class %(className)sRecipe(%(recipeBaseClass)s):
name = '%(name)s'
version = '%(version)s'
def setup(r):
'''
In this recipe, you can make modifications to the package.
Examples:
# This appliance has high-memory-use PHP scripts
r.Replace('memory_limit = 8M', 'memory_limit = 32M', '/etc/php.ini')
%(removeText)s
# This appliance requires that a few binaries be replaced
# with binaries built from a custom archive that includes
# a Makefile that honors the DESTDIR variable for its
# install target.
r.addArchive('foo.tar.gz')
r.Make()
r.MakeInstall()
# This appliance requires an extra configuration file
r.Create('/etc/myconfigfile', contents='some data')
'''
""" % dict(className=className,
name=troveName,
version=shadowedVersion.trailingRevision().getVersion(),
recipeBaseClass=derivedRecipeType,
removeText=removeText)
open(recipeName, 'w').write(derivedRecipe)
log.info('Removing extra files from checkout')
conaryState = state.ConaryStateFromFile('CONARY', repos)
sourceState = conaryState.getSourceState()
# clear the factory since we don't care about how the parent trove was
# created
sourceState.setFactory('')
addRecipe = True
for (pathId, path, fileId,
version) in list(sourceState.iterFileList()):
if path == recipeName:
addRecipe = False
continue
sourceState.removeFile(pathId)
if util.exists(path):
statInfo = os.lstat(path)
try:
if statInfo.st_mode & stat.S_IFDIR:
os.rmdir(path)
else:
os.unlink(path)
except OSError, e:
log.warning("cannot remove %s: %s" % (path, e.strerror))
conaryState.write('CONARY')
if addRecipe:
checkin.addFiles([recipeName])
if extract:
log.info('extracting files from %s=%s[%s]' % (troveToDerive))
# extract to _ROOT_
extractDir = os.path.join(os.getcwd(), '_ROOT_')
ts = [(troveToDerive[0], (None, None), (troveToDerive[1],
troveToDerive[2]), True)]
cs = repos.createChangeSet(ts, recurse=True)
ChangesetExploder(cs, extractDir)
# extract to _OLD_ROOT_
secondDir = os.path.join(os.getcwd(), '_OLD_ROOT_')
cs = repos.createChangeSet(ts, recurse=True)
ChangesetExploder(cs, secondDir)
def derive(repos, cfg, targetLabel, troveSpec, checkoutDir=None,
extract=False, info=False, callback=None):
"""
Performs all the commands necessary to create a derived recipe.
First it shadows the package, then it creates a checkout of the shadow
and converts the checkout to a derived recipe package.
Finally if extract = True, it installs an version of the binary
package into a root.
@param repos: trovesource to search for and derive packages from
@param cfg: configuration to use when deriving the package
@type cfg: ConaryConfiguration object
@param targetLabel: label to derive from
@type targetLabel: versions.Label
@param checkoutDir: directory to create the checkout in. If None,
defaults to currentDir + packageName.
@param extract: If True, creates a subdirectory of the checkout named
_ROOT_ with the contents of the binary of the derived
package.
@param info: If true, only display the information about the shadow
that would be performed if the derive command were
completed.
@param callback:
"""
origDir = os.getcwd()
try:
if callback is None:
callback = DeriveCallback()
if isinstance(troveSpec, tuple):
troveName, versionSpec, flavor = troveSpec
versionSpec = str(versionSpec)
troveSpec = cmdline.toTroveSpec(troveName, versionSpec, flavor)
else:
troveName, versionSpec, flavor = cmdline.parseTroveSpec(troveSpec)
if isinstance(targetLabel, str):
targetLabel = Label(targetLabel)
troveName, versionSpec, flavor = cmdline.parseTroveSpec(troveSpec)
result = repos.findTrove(cfg.buildLabel,
(troveName, versionSpec, flavor),
cfg.flavor)
# findTrove shouldn't return multiple items for one package anymore
# when a flavor is specified.
troveToDerive, = result
# displaying output along the screen allows there to be a record
# of what operations were performed. Since this command is
# an aggregate of several commands I think that is appropriate,
# rather than simply using a progress callback.
log.info('Shadowing %s=%s[%s] onto %s' % (troveToDerive[0],
troveToDerive[1],
troveToDerive[2],
targetLabel))
if info:
cfg.interactive = False
error = branch.branch(repos, cfg, str(targetLabel),
['%s=%s[%s]'%troveToDerive],
makeShadow=True, sourceOnly=True,
binaryOnly=False, allowEmptyShadow=True,
info=info)
if info or error:
return
shadowedVersion = troveToDerive[1].createShadow(targetLabel)
shadowedVersion = shadowedVersion.getSourceVersion(False)
troveName = troveName.split(':')[0]
checkoutDir = checkoutDir or troveName
checkin.checkout(repos, cfg, checkoutDir,
["%s=%s" % (troveName, shadowedVersion)],
callback=callback)
os.chdir(checkoutDir)
nvfs = repos.getTrovesBySource(troveToDerive[0]+':source',
troveToDerive[1].getSourceVersion())
trvs = repos.getTroves(nvfs)
hasCapsule = [ x for x in trvs if x.troveInfo.capsule.type() ]
if hasCapsule:
derivedRecipeType = 'DerivedCapsuleRecipe'
removeText = ''
else:
derivedRecipeType = 'DerivedPackageRecipe'
removeText = \
"""
# This appliance uses PHP as a command interpreter but does
# not include a web server, so remove the file that creates
# a dependency on the web server
r.Remove('/etc/httpd/conf.d/php.conf')
"""
log.info('Rewriting recipe file')
recipeName = troveName + '.recipe'
className = util.convertPackageNameToClassName(troveName)
derivedRecipe = """
class %(className)sRecipe(%(recipeBaseClass)s):
name = '%(name)s'
version = '%(version)s'
def setup(r):
'''
In this recipe, you can make modifications to the package.
Examples:
# This appliance has high-memory-use PHP scripts
r.Replace('memory_limit = 8M', 'memory_limit = 32M', '/etc/php.ini')
%(removeText)s
# This appliance requires that a few binaries be replaced
# with binaries built from a custom archive that includes
# a Makefile that honors the DESTDIR variable for its
# install target.
r.addArchive('foo.tar.gz')
r.Make()
r.MakeInstall()
# This appliance requires an extra configuration file
r.Create('/etc/myconfigfile', contents='some data')
'''
""" % dict(className=className,
name=troveName,
version=shadowedVersion.trailingRevision().getVersion(),
recipeBaseClass=derivedRecipeType,
removeText=removeText)
open(recipeName, 'w').write(derivedRecipe)
log.info('Removing extra files from checkout')
conaryState = state.ConaryStateFromFile('CONARY', repos)
sourceState = conaryState.getSourceState()
# clear the factory since we don't care about how the parent trove was
# created
sourceState.setFactory('')
addRecipe=True
for (pathId, path, fileId, version) in list(sourceState.iterFileList()):
if path == recipeName:
addRecipe = False
continue
sourceState.removeFile(pathId)
if util.exists(path):
statInfo = os.lstat(path)
try:
if statInfo.st_mode & stat.S_IFDIR:
os.rmdir(path)
else:
os.unlink(path)
except OSError, e:
log.warning("cannot remove %s: %s" % (path, e.strerror))
conaryState.write('CONARY')
if addRecipe:
checkin.addFiles([recipeName])
if extract:
log.info('extracting files from %s=%s[%s]' % (troveToDerive))
# extract to _ROOT_
extractDir = os.path.join(os.getcwd(), '_ROOT_')
ts = [ (troveToDerive[0], (None, None),
(troveToDerive[1], troveToDerive[2]), True) ]
cs = repos.createChangeSet(ts, recurse = True)
ChangesetExploder(cs, extractDir)
# extract to _OLD_ROOT_
secondDir = os.path.join(os.getcwd(), '_OLD_ROOT_')
cs = repos.createChangeSet(ts, recurse = True)
ChangesetExploder(cs, secondDir)