def waitForDBChange(self, since=0):
getLogger(self).debug(
"Watching for changes")
while True:
last_seq = max(self.getSeqNumber(), since)
self.stream = ChangesStream(
self.db,
feed="continuous",
since=last_seq,
heartbeat=True)
try:
for change in self.stream:
if not self.changes_callback:
return
if not change.get('last_seq', None):
if change['seq'] > self.getSeqNumber():
self.setSeqNumber(change['seq'])
if not change['id'].startswith('_design'):
getLogger(self).debug(
"Changes from another instance")
deleted = bool(change.get('deleted', False))
revision = change.get("changes")[-1].get('rev')
obj_id = change.get('id')
if not deleted:
# update cache
doc = self.db.get(obj_id)
self.addDoc(doc)
self.changes_callback(obj_id, revision, deleted)
except Exception as e:
getLogger(self).info("Some exception happened while waiting for changes")
getLogger(self).info(" The exception was: %s" % e)
def main(argv):
"""
Main method.
This method performs the following tasks:
1. Parse command line arguments
2. Retrieve credentials and connect to Cloudant and WebHDFS
3. Connect to the Cloudant `_changes` feed for checkpointed document
consumption
4. Process each change individually.
5. Upon exception throwing, store the latest checkpoint to local file and
exit.
"""
#add options into the parser
parser = configureOptions()
(options, args) = parser.parse_args()
checkRequiredArguments(options, parser)
print options
# configurations
last_seq = options.last_seq
#get credential
perm_file = '%s/.clou' % os.environ['HOME']
creds = get_creds(perm_file)
#connect to source database
s = Server('https://%s:%s@%s' %
(creds['cloudant_user'], creds['cloudant_pwd'], options.uri))
db = s[options.dbname]
#print db.info()
#connect to target hdfs cluster
hdfs = PyWebHdfsClient(host=options.hdfs_host,
port=options.hdfs_port,
user_name=creds['hdfs_user'])
hdfs.make_dir(options.hdfs_path)
#and here we consume the cloudant `_changes` feed
counter = 0
changestream = ChangesStream(db,
include_docs=True,
heartbeat=True,
since=last_seq)
for c in changestream:
#print c
try:
if counter % 100 == 0:
checkpoint(last_seq)
seq = processChange(hdfs, c, options.hdfs_path)
if seq: # protect against the last line being blank
last_seq = seq
counter += 1
except Exception:
traceback.print_exc()
checkpoint(last_seq)
os._exit(1)
checkpoint(last_seq)
def waitForDBChange(self, db_name, since=0, timeout=15000):
""" Be warned this will return after the database has a change, if
there was one before call it will return immediatly with the changes
done"""
changes = []
last_seq = max(self.getLastChangeSeq(db_name), since)
db = self._getDb(db_name)
with ChangesStream(db,
feed="longpoll",
since=last_seq,
timeout=timeout) as stream:
for change in stream:
if change['seq'] > self.getLastChangeSeq(db_name):
self.setLastChangeSeq(db_name, change['seq'])
if not change['id'].startswith('_design'):
#fake doc type for deleted objects
doc = {
'type': 'unknown',
'_deleted': 'False',
'_rev': [0]
}
if not change.get('deleted'):
doc = self.getDocument(db_name, change['id'])
changes.append(change_factory.create(doc))
if len(changes):
getLogger(self).debug("Changes from another instance")
return changes
def poll_once():
changes_stream = ChangesStream(db=self._couch_db,
since=self._last_processed_seq,
include_docs=True,
**extra_args)
for couch_change in changes_stream:
change = change_from_couch_row(
couch_change, document_store=self._document_store)
populate_change_metadata(change, SOURCE_COUCH,
self._couch_db.dbname)
yield change
self._last_processed_seq = couch_change.get('seq', None)
def run_burst(self):
"""
Use this for testing pillows. Will run through the changes stream once.
"""
changes_stream = ChangesStream(db=self.couch_db,
since=self.since,
filter=self.couch_filter,
include_docs=self.include_docs,
**self.extra_args)
for change in changes_stream:
if change:
self.processor(change)
def iter_changes(self, since, forever):
extra_args = {'feed': 'continuous'} if forever else {}
extra_args.update(self._extra_couch_view_params)
changes_stream = ChangesStream(
db=self._couch_db,
heartbeat=True,
since=since,
filter=self._couch_filter,
include_docs=self._include_docs,
**extra_args
)
for couch_change in changes_stream:
yield change_from_couch_row(couch_change, document_store=self._document_store)
def waitForDBChange(self, since=0):
"""Listen to the stream of changes provided by CouchDbKit. Process
these changes accordingly. If there's an exception while listening
to the changes, return inmediatly."""
# XXX: the while True found here shouldn't be necessary because
# changesStream already keeps listening 'for ever'. In a few tests
# I ran, this hypothesis was confirmed, but with our current setup
# i'm afraid I may be missing something. In any case, it works
# as it is, but this definitevely needs revision.
getLogger(self).debug(
"Watching for changes")
while True:
last_seq = max(self.getSeqNumber(), since)
self.stream = ChangesStream(
self.db,
feed="continuous",
since=last_seq,
heartbeat=True)
try:
for change in self.stream:
if not self.changes_callback:
return
if not change.get('last_seq', None):
if change['seq'] > self.getSeqNumber():
self.setSeqNumber(change['seq'])
if not change['id'].startswith('_design'):
getLogger(self).debug(
"Changes from another instance")
deleted = bool(change.get('deleted', False))
revision = change.get("changes")[-1].get('rev')
obj_id = change.get('id')
if not deleted:
# update cache
doc = self.db.get(obj_id)
self.addDoc(doc)
self.changes_callback(obj_id, revision, deleted)
except ResourceNotFound as e:
getLogger(self).info("The database couldn't be found")
self.no_workspace_callback()
return False
except Exception as e:
getLogger(self).info("Some exception happened while waiting for changes")
getLogger(self).info(" The exception was: %s" % e)
return False # kill thread, it's failed... in reconnection
def waitForDBChange(self, db_name, since=0, timeout=15000):
""" Be warned this will return after the database has a change, if
there was one before call it will return immediatly with the changes
done"""
changes = []
last_seq = max(self.getLastChangeSeq(db_name), since)
db = self.__getDb(db_name)
with ChangesStream(db,
feed="longpoll",
since=last_seq,
timeout=timeout) as stream:
for change in stream:
if change['seq'] > self.getLastChangeSeq(db_name):
changes.append(change)
last_seq = reduce(lambda x, y: max(y['seq'], x), changes,
self.getLastChangeSeq(db_name))
self.setLastChangeSeq(db_name, last_seq)
return changes
def iter_changes(self, since, forever):
from corehq.apps.change_feed.data_sources import SOURCE_COUCH
extra_args = {'feed': 'continuous'} if forever else {}
extra_args.update(self._extra_couch_view_params)
self._last_processed_seq = since
changes_stream = ChangesStream(db=self._couch_db,
heartbeat=True,
since=since,
filter=self._couch_filter,
include_docs=True,
**extra_args)
for couch_change in changes_stream:
change = change_from_couch_row(couch_change,
document_store=self._document_store)
populate_change_metadata(change, SOURCE_COUCH,
self._couch_db.dbname)
yield change
self._last_processed_seq = couch_change.get('seq', None)
def stream_changes(db, since, limit):
for change in ChangesStream(db=db, since=since, limit=limit):
yield CouchChange(
id=change['id'], rev=change['changes'][0]['rev'], deleted=change.get('deleted', False),
seq=change.get('seq'))