def handle_noargs(self, **options):
fs = cluster.get_hdfs()
create_directories(fs, [REMOTE_SAMPLE_DIR.get()])
remote_dir = REMOTE_SAMPLE_DIR.get()
# Copy examples binaries
for name in os.listdir(LOCAL_SAMPLE_DIR.get()):
local_dir = fs.join(LOCAL_SAMPLE_DIR.get(), name)
remote_data_dir = fs.join(remote_dir, name)
LOG.info(
_('Copying examples %(local_dir)s to %(remote_data_dir)s\n') %
{
'local_dir': local_dir,
'remote_data_dir': remote_data_dir
})
fs.do_as_user(fs.DEFAULT_USER, fs.copyFromLocal, local_dir,
remote_data_dir)
# Copy sample data
local_dir = paths.get_thirdparty_root("sample_data")
remote_data_dir = fs.join(remote_dir, 'data')
LOG.info(
_('Copying data %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir,
'remote_data_dir': remote_data_dir
})
fs.do_as_user(fs.DEFAULT_USER, fs.copyFromLocal, local_dir,
remote_data_dir)
# Load jobs
install_sample_user()
management.call_command('loaddata',
'initial_pig_examples.json',
verbosity=2)
Document.objects.sync()
def handle_noargs(self, **options):
fs = cluster.get_hdfs()
create_directories(fs, [REMOTE_SAMPLE_DIR.get()])
remote_dir = REMOTE_SAMPLE_DIR.get()
# Copy examples binaries
for name in os.listdir(LOCAL_SAMPLE_DIR.get()):
local_dir = fs.join(LOCAL_SAMPLE_DIR.get(), name)
remote_data_dir = fs.join(remote_dir, name)
LOG.info(_('Copying examples %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir, 'remote_data_dir': remote_data_dir})
fs.do_as_user(fs.DEFAULT_USER, fs.copyFromLocal, local_dir, remote_data_dir)
# Copy sample data
local_dir = paths.get_thirdparty_root("sample_data")
remote_data_dir = fs.join(remote_dir, 'data')
LOG.info(_('Copying data %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir, 'remote_data_dir': remote_data_dir})
fs.do_as_user(fs.DEFAULT_USER, fs.copyFromLocal, local_dir, remote_data_dir)
# Load jobs
USERNAME = '******'
try:
sample_user = User.objects.get(username=USERNAME)
except User.DoesNotExist:
sample_user = User.objects.create(username=USERNAME, password='******', is_active=False, is_superuser=False, id=1100713, pk=1100713)
management.call_command('loaddata', 'initial_pig_examples.json', verbosity=2)
Document.objects.sync()
def handle_noargs(self, **options):
fs = cluster.get_hdfs()
create_directories(fs, [REMOTE_SAMPLE_DIR.get()])
remote_dir = REMOTE_SAMPLE_DIR.get()
# Copy examples binaries
for name in os.listdir(LOCAL_SAMPLE_DIR.get()):
local_dir = fs.join(LOCAL_SAMPLE_DIR.get(), name)
remote_data_dir = fs.join(remote_dir, name)
LOG.info(_('Copying examples %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir, 'remote_data_dir': remote_data_dir})
fs.do_as_user(fs.DEFAULT_USER, fs.copyFromLocal, local_dir, remote_data_dir)
# Copy sample data
local_dir = paths.get_thirdparty_root("sample_data")
remote_data_dir = fs.join(remote_dir, 'data')
LOG.info(_('Copying data %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir, 'remote_data_dir': remote_data_dir})
fs.do_as_user(fs.DEFAULT_USER, fs.copyFromLocal, local_dir, remote_data_dir)
# Load jobs
sample_user = install_sample_user()
management.call_command('loaddata', 'initial_pig_examples.json', verbosity=2)
Document.objects.sync()
if USE_NEW_EDITOR.get():
# Get or create sample user directories
home_dir = Directory.objects.get_home_directory(sample_user)
examples_dir, created = Directory.objects.get_or_create(
parent_directory=home_dir,
owner=sample_user,
name=Document2.EXAMPLES_DIR)
try:
# Don't overwrite
doc = Document.objects.get(object_id=1100713)
doc2 = Document2.objects.get(owner=sample_user, name=doc.name, type='link-pigscript')
# If document exists but has been trashed, recover from Trash
if doc2.parent_directory != examples_dir:
doc2.parent_directory = examples_dir
doc2.save()
except Document.DoesNotExist:
LOG.warn('Sample pig script document not found.')
except Document2.DoesNotExist:
if doc.content_object:
data = doc.content_object.dict
data.update({'content_type': doc.content_type.model, 'object_id': doc.object_id})
data = json.dumps(data)
doc2 = Document2.objects.create(
owner=sample_user,
parent_directory=examples_dir,
name=doc.name,
type='link-pigscript',
description=doc.description,
data=data)
LOG.info('Successfully installed sample link to pig script: %s' % (doc2.name,))
# Share with default group
examples_dir.share(sample_user, Document2Permission.READ_PERM, groups=[get_default_user_group()])
def install_pig_script(self, sample_user):
doc2 = None
name = _('UpperText')
if Document2.objects.filter(owner=sample_user,
name=name,
type='query-pig',
is_history=False).exists():
LOG.info("Sample pig editor script already installed.")
doc2 = Document2.objects.get(owner=sample_user,
name=name,
type='query-pig',
is_history=False)
else:
statement = """REGISTER hdfs://{}/piggybank.jar;
data = LOAD '{}/data/midsummer.txt' as (text:CHARARRAY);
upper_case = FOREACH data GENERATE org.apache.pig.piggybank.evaluation.string.UPPER(text);
STORE upper_case INTO '$output';
""".format(REMOTE_SAMPLE_DIR.get(), REMOTE_SAMPLE_DIR.get())
snippet_properties = {
'hadoopProperties': [],
'parameters': [],
'resources': []
}
notebook = make_notebook(
name=name,
description=_('UpperText: Example Pig script'),
editor_type='pig',
statement=statement,
status='ready',
snippet_properties=snippet_properties,
is_saved=True)
# Remove files, functions, settings from snippet properties
data = notebook.get_data()
data['snippets'][0]['properties'].pop('files')
data['snippets'][0]['properties'].pop('functions')
data['snippets'][0]['properties'].pop('settings')
try:
with transaction.atomic():
doc2 = Document2.objects.create(
owner=sample_user,
name=data['name'],
type='query-pig',
description=data['description'],
data=json.dumps(data))
except Exception as e:
LOG.exception(
"Failed to create sample pig script document: %s" % e)
# Just to be sure we delete Doc2 object incase of exception.
# Possible when there are mixed InnoDB and MyISAM tables
if doc2 and Document2.objects.filter(id=doc2.id).exists():
doc2.delete()
return doc2
def handle(self, *args, **options):
fs = cluster.get_hdfs()
create_directories(fs, [REMOTE_SAMPLE_DIR.get()])
remote_dir = REMOTE_SAMPLE_DIR.get()
sample_user = install_sample_user()
# Copy examples binaries
for name in os.listdir(LOCAL_SAMPLE_DIR.get()):
local_dir = fs.join(LOCAL_SAMPLE_DIR.get(), name)
remote_data_dir = fs.join(remote_dir, name)
LOG.info(
_('Copying examples %(local_dir)s to %(remote_data_dir)s\n') %
{
'local_dir': local_dir,
'remote_data_dir': remote_data_dir
})
fs.do_as_user(sample_user.username, fs.copyFromLocal, local_dir,
remote_data_dir)
# Copy sample data
local_dir = paths.get_thirdparty_root("sample_data")
remote_data_dir = fs.join(remote_dir, 'data')
LOG.info(
_('Copying data %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir,
'remote_data_dir': remote_data_dir
})
fs.do_as_user(sample_user.username, fs.copyFromLocal, local_dir,
remote_data_dir)
# Initialize doc2, whether editor script or link
doc2 = None
# Install editor pig script without doc1 link
LOG.info("Using Hue 4, will install pig editor sample.")
doc2 = self.install_pig_script(sample_user)
if USE_NEW_EDITOR.get():
# Get or create sample user directories
LOG.info("Creating sample user directories.")
home_dir = Directory.objects.get_home_directory(sample_user)
examples_dir, created = Directory.objects.get_or_create(
parent_directory=home_dir,
owner=sample_user,
name=Document2.EXAMPLES_DIR)
# If document exists but has been trashed, recover from Trash
if doc2 and doc2.parent_directory != examples_dir:
doc2.parent_directory = examples_dir
doc2.save()
# Share with default group
examples_dir.share(sample_user,
Document2Permission.READ_PERM,
groups=[get_default_user_group()])
def install_pig_script(self, sample_user):
doc2 = None
name = _('UpperText')
if Document2.objects.filter(owner=sample_user, name=name, type='query-pig', is_history=False).exists():
LOG.info("Sample pig editor script already installed.")
doc2 = Document2.objects.get(owner=sample_user, name=name, type='query-pig', is_history=False)
else:
statement = """REGISTER hdfs://{}/piggybank.jar;
data = LOAD '{}/data/midsummer.txt' as (text:CHARARRAY);
upper_case = FOREACH data GENERATE org.apache.pig.piggybank.evaluation.string.UPPER(text);
STORE upper_case INTO '$output';
""".format(REMOTE_SAMPLE_DIR.get(), REMOTE_SAMPLE_DIR.get())
snippet_properties = {
'hadoopProperties': [],
'parameters': [],
'resources': []
}
notebook = make_notebook(
name=name,
description=_('UpperText: Example Pig script'),
editor_type='pig',
statement=statement,
status='ready',
snippet_properties=snippet_properties,
is_saved=True
)
# Remove files, functions, settings from snippet properties
data = notebook.get_data()
data['snippets'][0]['properties'].pop('files')
data['snippets'][0]['properties'].pop('functions')
data['snippets'][0]['properties'].pop('settings')
try:
with transaction.atomic():
doc2 = Document2.objects.create(
owner=sample_user,
name=data['name'],
type='query-pig',
description=data['description'],
data=json.dumps(data)
)
except Exception, e:
LOG.exception("Failed to create sample pig script document: %s" % e)
# Just to be sure we delete Doc2 object incase of exception.
# Possible when there are mixed InnoDB and MyISAM tables
if doc2 and Document2.objects.filter(id=doc2.id).exists():
doc2.delete()
def handle(self, *args, **options):
fs = cluster.get_hdfs()
create_directories(fs, [REMOTE_SAMPLE_DIR.get()])
remote_dir = REMOTE_SAMPLE_DIR.get()
sample_user = install_sample_user()
# Copy examples binaries
for name in os.listdir(LOCAL_SAMPLE_DIR.get()):
local_dir = fs.join(LOCAL_SAMPLE_DIR.get(), name)
remote_data_dir = fs.join(remote_dir, name)
LOG.info(_('Copying examples %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir, 'remote_data_dir': remote_data_dir})
fs.do_as_user(sample_user.username, fs.copyFromLocal, local_dir, remote_data_dir)
# Copy sample data
local_dir = paths.get_thirdparty_root("sample_data")
remote_data_dir = fs.join(remote_dir, 'data')
LOG.info(_('Copying data %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir, 'remote_data_dir': remote_data_dir})
fs.do_as_user(sample_user.username, fs.copyFromLocal, local_dir, remote_data_dir)
# Initialize doc2, whether editor script or link
doc2 = None
# Install editor pig script without doc1 link
LOG.info("Using Hue 4, will install pig editor sample.")
doc2 = self.install_pig_script(sample_user)
if USE_NEW_EDITOR.get():
# Get or create sample user directories
LOG.info("Creating sample user directories.")
home_dir = Directory.objects.get_home_directory(sample_user)
examples_dir, created = Directory.objects.get_or_create(
parent_directory=home_dir,
owner=sample_user,
name=Document2.EXAMPLES_DIR)
# If document exists but has been trashed, recover from Trash
if doc2 and doc2.parent_directory != examples_dir:
doc2.parent_directory = examples_dir
doc2.save()
# Share with default group
examples_dir.share(sample_user, Document2Permission.READ_PERM, groups=[get_default_user_group()])
def handle_noargs(self, **options):
fs = cluster.get_hdfs()
create_directories(fs, [REMOTE_SAMPLE_DIR.get()])
remote_dir = REMOTE_SAMPLE_DIR.get()
# Copy examples binaries
for name in os.listdir(LOCAL_SAMPLE_DIR.get()):
local_dir = fs.join(LOCAL_SAMPLE_DIR.get(), name)
remote_data_dir = fs.join(remote_dir, name)
LOG.info(
_('Copying examples %(local_dir)s to %(remote_data_dir)s\n') %
{
'local_dir': local_dir,
'remote_data_dir': remote_data_dir
})
fs.do_as_user(fs.DEFAULT_USER, fs.copyFromLocal, local_dir,
remote_data_dir)
# Copy sample data
local_dir = paths.get_thirdparty_root("sample_data")
remote_data_dir = fs.join(remote_dir, 'data')
LOG.info(
_('Copying data %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir,
'remote_data_dir': remote_data_dir
})
fs.do_as_user(fs.DEFAULT_USER, fs.copyFromLocal, local_dir,
remote_data_dir)
# Load jobs
USERNAME = '******'
try:
sample_user = User.objects.get(username=USERNAME)
except User.DoesNotExist:
sample_user = User.objects.create(username=USERNAME,
password='******',
is_active=False,
is_superuser=False,
id=1100713,
pk=1100713)
management.call_command('loaddata',
'initial_pig_examples.json',
verbosity=2)
def handle_noargs(self, **options):
fs = cluster.get_hdfs()
create_directories(fs, [REMOTE_SAMPLE_DIR.get()])
remote_dir = REMOTE_SAMPLE_DIR.get()
# Copy examples binaries
for name in os.listdir(LOCAL_SAMPLE_DIR.get()):
local_dir = fs.join(LOCAL_SAMPLE_DIR.get(), name)
remote_data_dir = fs.join(remote_dir, name)
LOG.info(_('Copying examples %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir, 'remote_data_dir': remote_data_dir})
fs.do_as_user(fs.DEFAULT_USER, fs.copyFromLocal, local_dir, remote_data_dir)
# Copy sample data
local_dir = paths.get_thirdparty_root("sample_data")
remote_data_dir = fs.join(remote_dir, 'data')
LOG.info(_('Copying data %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir, 'remote_data_dir': remote_data_dir})
fs.do_as_user(fs.DEFAULT_USER, fs.copyFromLocal, local_dir, remote_data_dir)
# Load jobs
install_sample_user()
management.call_command('loaddata', 'initial_pig_examples.json', verbosity=2)
Document.objects.sync()
def handle_noargs(self, **options):
fs = cluster.get_hdfs()
create_directories(fs, [REMOTE_SAMPLE_DIR.get()])
remote_dir = REMOTE_SAMPLE_DIR.get()
# Copy examples binaries
for name in os.listdir(LOCAL_SAMPLE_DIR.get()):
local_dir = fs.join(LOCAL_SAMPLE_DIR.get(), name)
remote_data_dir = fs.join(remote_dir, name)
LOG.info(
_('Copying examples %(local_dir)s to %(remote_data_dir)s\n') %
{
'local_dir': local_dir,
'remote_data_dir': remote_data_dir
})
fs.do_as_user(fs.DEFAULT_USER, fs.copyFromLocal, local_dir,
remote_data_dir)
# Copy sample data
local_dir = paths.get_thirdparty_root("sample_data")
remote_data_dir = fs.join(remote_dir, 'data')
LOG.info(
_('Copying data %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir,
'remote_data_dir': remote_data_dir
})
fs.do_as_user(fs.DEFAULT_USER, fs.copyFromLocal, local_dir,
remote_data_dir)
# Load jobs
sample_user = install_sample_user()
management.call_command('loaddata',
'initial_pig_examples.json',
verbosity=2)
Document.objects.sync()
if USE_NEW_EDITOR.get():
# Get or create sample user directories
home_dir = Directory.objects.get_home_directory(sample_user)
examples_dir, created = Directory.objects.get_or_create(
parent_directory=home_dir,
owner=sample_user,
name=Document2.EXAMPLES_DIR)
try:
# Don't overwrite
doc = Document.objects.get(object_id=1100713)
doc2 = Document2.objects.get(owner=sample_user,
name=doc.name,
type='link-pigscript')
# If document exists but has been trashed, recover from Trash
if doc2.parent_directory != examples_dir:
doc2.parent_directory = examples_dir
doc2.save()
except Document.DoesNotExist:
LOG.warn('Sample pig script document not found.')
except Document2.DoesNotExist:
if doc.content_object:
data = doc.content_object.dict
data.update({
'content_type': doc.content_type.model,
'object_id': doc.object_id
})
data = json.dumps(data)
doc2 = Document2.objects.create(
owner=sample_user,
parent_directory=examples_dir,
name=doc.name,
type='link-pigscript',
description=doc.description,
data=data)
LOG.info(
'Successfully installed sample link to pig script: %s'
% (doc2.name, ))
# Share with default group
examples_dir.share(sample_user,
Document2Permission.READ_PERM,
groups=[get_default_user_group()])
def handle(self, *args, **options):
fs = cluster.get_hdfs()
create_directories(fs, [REMOTE_SAMPLE_DIR.get()])
remote_dir = REMOTE_SAMPLE_DIR.get()
sample_user = install_sample_user()
# Copy examples binaries
for name in os.listdir(LOCAL_SAMPLE_DIR.get()):
local_dir = fs.join(LOCAL_SAMPLE_DIR.get(), name)
remote_data_dir = fs.join(remote_dir, name)
LOG.info(_('Copying examples %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir, 'remote_data_dir': remote_data_dir})
fs.do_as_user(sample_user.username, fs.copyFromLocal, local_dir, remote_data_dir)
# Copy sample data
local_dir = paths.get_thirdparty_root("sample_data")
remote_data_dir = fs.join(remote_dir, 'data')
LOG.info(_('Copying data %(local_dir)s to %(remote_data_dir)s\n') % {
'local_dir': local_dir, 'remote_data_dir': remote_data_dir})
fs.do_as_user(sample_user.username, fs.copyFromLocal, local_dir, remote_data_dir)
# Initialize doc2, whether editor script or link
doc2 = None
if IS_HUE_4.get():
# Install editor pig script without doc1 link
LOG.info("Using Hue 4, will install pig editor sample.")
doc2 = self.install_pig_script(sample_user)
else:
# Install old pig script fixture
LOG.info("Using Hue 3, will install pig script fixture.")
with transaction.atomic():
management.call_command('loaddata', 'initial_pig_examples.json', verbosity=2, commit=False)
Document.objects.sync()
if USE_NEW_EDITOR.get():
# Get or create sample user directories
LOG.info("Creating sample user directories.")
home_dir = Directory.objects.get_home_directory(sample_user)
examples_dir, created = Directory.objects.get_or_create(
parent_directory=home_dir,
owner=sample_user,
name=Document2.EXAMPLES_DIR)
if not IS_HUE_4.get():
try:
# Don't overwrite
doc = Document.objects.get(object_id=1100713)
doc2 = Document2.objects.get(owner=sample_user, name=doc.name, type='link-pigscript')
except Document.DoesNotExist:
LOG.warn('Sample pig script document not found.')
except Document2.DoesNotExist:
if doc.content_object:
data = doc.content_object.dict
data.update({'content_type': doc.content_type.model, 'object_id': doc.object_id})
data = json.dumps(data)
doc2 = Document2.objects.create(
owner=sample_user,
parent_directory=examples_dir,
name=doc.name,
type='link-pigscript',
description=doc.description,
data=data)
LOG.info('Successfully installed sample link to pig script: %s' % (doc2.name,))
# If document exists but has been trashed, recover from Trash
if doc2 and doc2.parent_directory != examples_dir:
doc2.parent_directory = examples_dir
doc2.save()
# Share with default group
examples_dir.share(sample_user, Document2Permission.READ_PERM, groups=[get_default_user_group()])
