def createDB(self):
"""
Creates new DB to load SQL dump files if required
"""
db_create = MySQLDB(host=self.host,
port=self.port,
user=self.db_user,
passwd=self.db_passw)
db_create.connect()
db_create.create_database(self.db_name)
db_create.close()
def createDB(self):
"""
Creates new DB to load SQL dump files if required
"""
db_create = MySQLDB(host=self.host, port=self.port,
user=self.db_user, passwd=self.db_passw)
db_create.connect()
db_create.create_database(self.db_name)
db_create.close()
def DB_exists(self):
db_check = MySQLDB(host=self.host,
port=self.port,
user=self.db_user,
passwd=self.db_passw)
db_check.connect()
db_exists = db_check.db_exists(self.db_name)
db_check.close()
return db_exists
def DB_exists(self):
db_check = MySQLDB(host=self.host, port=self.port, user=self.db_user,
passwd=self.db_passw)
db_check.connect()
db_exists = db_check.db_exists(self.db_name)
db_check.close()
return db_exists
def execute(
self,
page_fan,
rev_fan,
page_cache_size,
rev_cache_size,
host,
port,
db_name,
db_user,
db_passw,
db_engine,
mirror,
download_files,
base_ports,
control_ports,
dumps_dir=None,
):
"""
Run data retrieval and loading actions.
Arguments:
- page_fan = Number of workers to fan out page elements parsing
- rev_fan = Number of workers to fan out rev elements parsing
- db_user = User name to connect to local database
- db_passw = Password for database user
- mirror = Base URL of site hosting XML dumps
"""
if download_files:
# TODO: Use proper logging module to track execution progress
# Choose corresponding file downloader and etl wrapper
print "Downloading new dump files from %s, for language %s" % (mirror, self.lang)
self.down = RevHistDownloader(mirror, self.lang)
# Donwload latest set of dump files
self.paths, self.date = self.down.download(self.date)
print "Got files for lang %s, date: %s" % (self.lang, self.date)
# db_name = self.lang + '_' + self.date.strip('/')
else:
# Case of dumps folder provided explicity
if dumps_dir:
# Allow specifying relative paths, as well
dumps_path = os.path.expanduser(dumps_dir)
# Retrieve path to all available files to feed ETL lines
if not os.path.exists(dumps_path):
print "No dump files will be downloaded and local folder "
print "with dump files not found. Please, specify a "
print "valid path to local folder containing dump files."
print "Program will exit now."
sys.exit()
else:
# Attempt to find list of .7z or .xml files to be processed
self.paths = glob.glob(os.path.join(dumps_path, "*.7z"))
if not self.paths:
self.paths = glob.glob(os.path.join(dumps_path, "*.xml"))
if not self.paths:
print "Directory %s" % dumps_dir
print "does not contain any valid dump file."
print "Program will exit now."
sys.exit()
# If not provided explicitly, look for default location of
# dumps directory
else:
dumps_dir = os.path.join(self.lang + "_dumps", self.date)
# Look up dump files in default directory name
if not os.path.exists(dumps_dir):
print "Default directory %s" % dumps_dir
print " containing dump files not found."
print "Program will exit now."
sys.exit()
else:
self.paths = glob.glob(dumps_dir + "/*.7z")
print "paths: " + unicode(self.paths)
# DB SCHEMA PREPARATION
db_create = MySQLDB(host=host, port=port, user=db_user, passwd=db_passw)
db_create.connect()
db_create.create_database(db_name)
db_create.close()
db_schema = MySQLDB(host=host, port=port, user=db_user, passwd=db_passw, db=db_name)
db_schema.connect()
db_schema.create_schema(engine=db_engine)
db_schema.close()
# Complete the queue of paths to be processed and STOP flags for
# each ETL subprocess
paths_queue = mp.JoinableQueue()
for path in self.paths:
paths_queue.put(path)
for x in range(self.etl_lines):
paths_queue.put("STOP")
for x in range(self.etl_lines):
new_etl = PageRevisionETL(
name="ETL-process-%s" % x,
paths_queue=paths_queue,
lang=self.lang,
page_fan=page_fan,
rev_fan=rev_fan,
page_cache_size=page_cache_size,
rev_cache_size=rev_cache_size,
db_name=db_name,
db_user=db_user,
db_passw=db_passw,
base_port=base_ports[x] + (20 * x),
control_port=control_ports[x] + (20 * x),
)
self.etl_list.append(new_etl)
print "ETL process for page and revision history defined OK."
print "Proceeding with ETL workflows. This may take time..."
# Extract, process and load information in local DB
for etl in self.etl_list:
etl.start()
# Wait for ETL lines to finish
for etl in self.etl_list:
etl.join()
# TODO: logger; ETL step completed, proceeding with data
# analysis and visualization
print "ETL process finished for language %s and date %s" % (self.lang, self.date)
# Create primary keys for all tables
# TODO: This must also be tracked by official logging module
print "Now creating primary key indexes in database tables."
print "This may take a while..."
db_pks = MySQLDB(host="localhost", port=3306, user=db_user, passwd=db_passw, db=db_name)
db_pks.connect()
db_pks.create_pks()
db_pks.close()
def execute(self, page_fan, rev_fan, page_cache_size, rev_cache_size,
mirror, download_files, base_ports, control_ports,
dumps_dir=None, debug=False):
"""
Run data retrieval and loading actions.
Arguments:
- page_fan = Number of workers to fan out page elements parsing
- rev_fan = Number of workers to fan out rev elements parsing
- db_user = User name to connect to local database
- db_passw = Password for database user
- mirror = Base URL of site hosting XML dumps
"""
print "----------------------------------------------------------"
print ("""Executing ETL:RevHistory on lang: {0} date: {1}"""
.format(self.lang, self.date))
print ("ETL lines = {0} page_fan = {1} rev_fan = {2}"
.format(self.etl_lines, page_fan, rev_fan))
print "Download files =", download_files
print "Start time is {0}".format(time.strftime("%Y-%m-%d %H:%M:%S %Z",
time.localtime()))
print "----------------------------------------------------------"
print
if download_files:
# TODO: Use proper logging module to track execution progress
# Choose corresponding file downloader and etl wrapper
print "Downloading new dump files from %s, for language %s" % (
mirror, self.lang)
self.down = download.RevHistDownloader(mirror,
self.lang, dumps_dir)
# Donwload latest set of dump files
self.paths, self.date = self.down.download(self.date)
if not self.paths:
print "Error: dump files with pages-logging info not found."
print "Program will exit now."
sys.exit()
print "Retrieved dump files for lang %s, date: %s" % (self.lang,
self.date)
print
else:
print "Looking for revision-history dump file(s) in data dir"
# Case of dumps folder provided explicity
if dumps_dir:
# Allow specifying relative paths, as well
abs_dumps_path = os.path.expanduser(dumps_dir)
dumps_path = os.path.join(abs_dumps_path,
self.lang + '_dumps', self.date)
# Retrieve path to all available files to feed ETL lines
if not os.path.exists(dumps_path):
print "No dump files will be downloaded and local folder with dump files not found."
print "Please, specify a valid path to local folder containing dump files."
print "Program will exit now."
sys.exit()
else:
# Attempt to find list of .7z or .xml files to be processed
self.paths = glob.glob(os.path.join(dumps_path,
'*pages-meta-hsitory*.7z'))
if not self.paths:
self.paths = glob.glob(os.path.join(dumps_path,
'*pages-meta-hsitory*.xml'))
if not self.paths:
print "Directory %s does not contain any valid dump file." % dumps_path
print "Program will exit now."
sys.exit()
# If not provided explicitly, look for default location of
# dumps directory
else:
dumps_dir = os.path.join("data", self.lang + '_dumps',
self.date)
# Look up dump files in default directory name
if not os.path.exists(dumps_dir):
print "Default directory %s containing dump files not found." % dumps_dir
print "Program will exit now."
sys.exit()
else:
self.paths = glob.glob(os.path.join(dumps_dir, '*pages-meta-history*.7z'))
if not self.paths:
self.paths = glob.glob(os.path.join(dumps_dir,
'*pages-meta-hsitory*.xml'))
if not self.paths:
print "Directory %s does not contain any valid dump file." % dumps_dir
print "Program will exit now."
sys.exit()
print "Found revision-history dump file(s) to process."
print
# Print list of file paths in debug mode
if debug:
print "paths: " + unicode(self.paths)
print
# Create database
# TODO: Empty correspoding tables if DB already exists
# or let the user select behaviour with config argument
if self.DB_exists():
self.create_DB(complete=False)
else:
self.create_DB(complete=True)
# First insert namespace info in DB
dump = DumpFile(self.paths[0])
db_schema = MySQLDB(host=self.host, port=self.port, user=self.db_user,
passwd=self.db_passw, db=self.db_name)
db_schema.connect()
db_schema.insert_namespaces(nsdict=dump.get_namespaces())
db_schema.close()
# Complete the queue of paths to be processed and STOP flags for
# each ETL subprocess
paths_queue = mp.JoinableQueue()
for path in self.paths:
paths_queue.put(path)
for x in range(self.etl_lines):
paths_queue.put('STOP')
for x in range(self.etl_lines):
new_etl = RevisionHistoryETL(
name="[ETL:RevHistory-%s]" % x,
paths_queue=paths_queue, lang=self.lang,
page_fan=page_fan, rev_fan=rev_fan,
page_cache_size=page_cache_size,
rev_cache_size=rev_cache_size,
db_name=self.db_name,
db_user=self.db_user, db_passw=self.db_passw,
base_port=base_ports[x]+(20*x),
control_port=control_ports[x]+(20*x)
)
self.etl_list.append(new_etl)
print "ETL:RevHistory task defined OK."
print "Proceeding with ETL workflows. This may take time..."
print
# Extract, process and load information in local DB
for etl in self.etl_list:
etl.start()
# Wait a second for new ETL process to start all subprocesses
time.sleep(1)
# Wait for ETL lines to finish
for etl in self.etl_list:
etl.join()
# Insert user info after all ETL lines have finished
# to ensure that all metadata are stored in Redis cache
# disregarding of the execution order
data_dir = os.path.join(os.getcwd(), os.path.split(self.paths[0])[0])
db_users = MySQLDB(host=self.host, port=self.port, user=self.db_user,
passwd=self.db_passw, db=self.db_name)
db_users.connect()
users_file_to_db(con=db_users, lang=self.lang,
log_file=os.path.join(data_dir, 'logs', 'users.log'),
tmp_dir=os.path.join(data_dir, 'tmp')
)
db_users.close()
# TODO: logger; ETL step completed, proceeding with data
# analysis and visualization
print "ETL:RevHistory task finished for language %s and date %s" % (
self.lang, self.date)
print
# Create primary keys for all tables
# TODO: This must also be tracked by main logging module
print "Now creating primary key indexes in database tables."
print "This may take a while..."
print
db_pks = MySQLDB(host='localhost', port=3306, user=self.db_user,
passwd=self.db_passw, db=self.db_name)
db_pks.connect()
db_pks.create_pks_revhist()
db_pks.close()
def create_DB(self, complete=False):
if complete:
db_create = MySQLDB(host=self.host, port=self.port,
user=self.db_user, passwd=self.db_passw)
db_create.connect()
db_create.create_database(self.db_name)
db_create.close()
db_schema = MySQLDB(host=self.host, port=self.port, user=self.db_user,
passwd=self.db_passw, db=self.db_name)
db_schema.connect()
db_schema.create_schema_revhist(engine=self.db_engine)
db_schema.close()
def execute(self, log_fan, log_cache_size,
mirror, download_files, base_ports, control_ports,
dumps_dir=None, debug=False):
"""
Run data retrieval and loading actions.
Arguments:
- log_fan = Number of workers to fan out logitem elements parsing
- db_user = User name to connect to local database
- db_passw = Password for database user
- mirror = Base URL of site hosting XML dumps
"""
print "----------------------------------------------------------"
print("Executing ETL:PagesLogging on lang: {0} date: {1}"
.format(self.lang, self.date))
print "log_fan =", log_fan
print "Download files =", download_files
print "Start time is {0}".format(time.strftime("%Y-%m-%d %H:%M:%S %Z",
time.localtime()))
print "----------------------------------------------------------"
print
if download_files:
# TODO: Use proper logging module to track execution progress
# Choose corresponding file downloader and etl wrapper
print """Downloading new logging dump files from %s,
for language %s""" % (mirror, self.lang)
self.down = download.LoggingDownloader(mirror,
self.lang, dumps_dir)
# Donwload latest set of dump files
self.paths, self.date = self.down.download(self.date)
if not self.paths:
print "Error: dump files with pages-logging info not found."
print "Program will exit now."
sys.exit()
print "Got files for lang %s, date: %s" % (self.lang, self.date)
else:
print "Looking for pages-logging dump file in data dir"
# Case of dumps folder provided explicity
if dumps_dir:
# Allow specifying relative paths, as well
abs_dumps_path = os.path.expanduser(dumps_dir)
dumps_path = os.path.join(abs_dumps_path,
self.lang + '_dumps', self.date)
# Retrieve path to all available files to feed ETL lines
if not os.path.exists(dumps_path):
print "No dump files will be downloaded and local folder with dump files not found."
print "Please, specify a valid path to local folder containing dump files."
print "Program will exit now."
sys.exit()
else:
# Attempt to find list of *page-logging*.gz or
# *page-logging*.xml files to be processed
self.paths = glob.glob(os.path.join(dumps_path, '*pages-logging*.gz'))
if not self.paths:
self.paths = glob.glob(os.path.join(dumps_path,
'*pages-logging*.xml'))
if not self.paths:
print "Directory %s does not contain any valid dump file." % dumps_path
print "Program will exit now."
sys.exit()
# If not provided explicitly, look for default location of
# dumps directory
else:
dumps_dir = os.path.join("data", self.lang + '_dumps',
self.date)
# Look up dump files in default directory name
if not os.path.exists(dumps_dir):
print "Default directory %s containing dump files not found." % dumps_dir
print "Program will exit now."
sys.exit()
else:
self.paths = glob.glob(os.path.join(dumps_dir, '*pages-logging*.gz'))
if not self.paths:
self.paths = glob.glob(os.path.join(dumps_dir,
'*pages-logging*.xml'))
if not self.paths:
print "Directory %s does not contain any valid dump file." % dumps_dir
print "Program will exit now."
sys.exit()
print "Found pages-logging dump file to process."
print
if debug:
print "paths: " + unicode(self.paths)
print
# Create database if it does not exist
# empty logging table otherwise
if self.DB_exists():
self.create_DB(complete=False)
else:
self.create_DB(complete=True)
new_etl = LoggingETL(name="[ETL:PagesLogging-0]",
path=self.paths, lang=self.lang,
log_fan=log_fan,
log_cache_size=log_cache_size,
db_name=self.db_name,
db_user=self.db_user, db_passw=self.db_passw,
base_port=base_ports[0]+(30),
control_port=control_ports[0]+(30)
)
print "ETL:Logging task for administrative records defined OK."
print "Proceeding with ETL workflow. This may take time..."
print
# Extract, process and load information in local DB
new_etl.start()
# Wait for ETL line to finish
new_etl.join()
# TODO: logger; ETL step completed, proceeding with data
# analysis and visualization
print "ETL:Logging task finished for lang %s and date %s" % (
self.lang, self.date)
print
# Create primary keys for all tables
# TODO: This must also be tracked by official logging module
print "Now creating primary key indexes in database tables."
print "This may take a while..."
print
db_pks = MySQLDB(host='localhost', port=3306, user=self.db_user,
passwd=self.db_passw, db=self.db_name)
db_pks.connect()
db_pks.create_pks_logitem()
db_pks.close()
def run(self):
"""
Execute workflow to import revision history data from dump files
The data loading workflow is composed of a number of processor
elements, which can be:
- Producer (P): raw input data --> input element queue
- ConsumerProducer (CP): input element queue --> insert db queue
- Consumer (C): insert db queue --> database (MySQL/MariaDB)
In this case, the logical combination is usually N:N:1 (P, CP, C)
"""
start = time.time()
print self.name, "Starting PageRevisionETL workflow at %s" % (
time.strftime("%Y-%m-%d %H:%M:%S %Z", time.localtime()))
db_ns = MySQLDB(host='localhost',
port=3306,
user=self.db_user,
passwd=self.db_passw,
db=self.db_name)
db_ns.connect()
db_pages = MySQLDB(host='localhost',
port=3306,
user=self.db_user,
passwd=self.db_passw,
db=self.db_name)
db_pages.connect()
db_revs = MySQLDB(host='localhost',
port=3306,
user=self.db_user,
passwd=self.db_passw,
db=self.db_name)
db_revs.connect()
# DATA EXTRACTION
# Use consistent naming for all child processes
xml_reader_name = '-'.join([self.name, 'xml_reader'])
page_proc_name = '-'.join([self.name, 'process_page'])
rev_proc_name = '-'.join([self.name, 'process_revision'])
page_insert_name = '-'.join([self.name, 'insert_page'])
rev_insert_name = '-'.join([self.name, 'insert_revision'])
for path in iter(self.paths_queue.get, 'STOP'):
# Start subprocess to extract elements from revision dump file
dump_file = DumpFile(path)
xml_reader = Producer(name=xml_reader_name,
target=process_xml,
kwargs=dict(dump_file=dump_file),
consumers=self.page_fan + self.rev_fan,
push_pages_port=self.base_port,
push_revs_port=self.base_port + 1,
control_port=self.control_port)
xml_reader.start()
print xml_reader_name, "started"
print self.name, "Extracting data from XML revision history file:"
print path
# List to keep tracking of page and revision workers
workers = []
db_workers_revs = []
# Create and start page processes
for worker in range(self.page_fan):
page_worker_name = '-'.join([page_proc_name, unicode(worker)])
process_page = Processor(name=page_worker_name,
target=pages_to_file,
producers=1,
consumers=1,
pull_port=self.base_port,
push_port=self.base_port + 2,
control_port=self.control_port)
process_page.start()
workers.append(process_page)
print page_worker_name, "started"
# Create and start revision processes
for worker in range(self.rev_fan):
rev_worker_name = '-'.join([rev_proc_name, unicode(worker)])
db_wrev = MySQLDB(host='localhost',
port=3306,
user=self.db_user,
passwd=self.db_passw,
db=self.db_name)
db_wrev.connect()
process_revision = Processor(name=rev_worker_name,
target=revs_to_file,
kwargs=dict(lang=self.lang),
producers=1,
consumers=1,
pull_port=self.base_port + 1,
push_port=self.base_port + 3,
control_port=self.control_port)
process_revision.start()
workers.append(process_revision)
db_workers_revs.append(db_wrev)
print rev_worker_name, "started"
# Create directory for logging files if it does not exist
log_dir = os.path.join(os.path.split(path)[0], 'logs')
tmp_dir = os.path.join(os.getcwd(), os.path.split(path)[0], 'tmp')
file_name = os.path.split(path)[1]
if not os.path.exists(log_dir):
os.makedirs(log_dir)
if not os.path.exists(tmp_dir):
os.makedirs(tmp_dir)
log_file = os.path.join(log_dir, file_name + '.log')
page_insert_db = Consumer(name=page_insert_name,
target=pages_file_to_db,
kwargs=dict(
con=db_pages,
log_file=log_file,
tmp_dir=tmp_dir,
file_rows=self.page_cache_size,
etl_prefix=self.name),
producers=self.page_fan,
pull_port=self.base_port + 2)
rev_insert_db = Consumer(name=rev_insert_name,
target=revs_file_to_db,
kwargs=dict(con=db_revs,
log_file=log_file,
tmp_dir=tmp_dir,
file_rows=self.rev_cache_size,
etl_prefix=self.name),
producers=self.rev_fan,
pull_port=self.base_port + 3)
page_insert_db.start()
print page_insert_name, "started"
rev_insert_db.start()
print rev_insert_name, "started"
print self.name, "Waiting for all processes to finish..."
print
xml_reader.join()
for w in workers:
w.join()
page_insert_db.join()
rev_insert_db.join()
# Mark this path as done
self.paths_queue.task_done()
# Mark STOP message as processed and finish
self.paths_queue.task_done()
end = time.time()
print self.name, ": All tasks done in %.4f sec." % ((end - start) / 1.)
print
db_ns.close()
db_pages.close()
db_revs.close()
for dbcon in db_workers_revs:
dbcon.close()
def run(self):
"""
Execute workflow to import logging records of actions on pages and
users from dump file
The data loading workflow is composed of a number of processor
elements, which can be:
- Producer (P): raw input data --> input element queue
- ConsumerProducer (CP): input element queue --> insert db queue
- Consumer (C): insert db queue --> database (MySQL/MariaDB)
In this case, the usual combination is 1:N:1 (P, CP, C)
"""
start = time.time()
print "Starting LoggingETL workflow at %s" % (time.strftime(
"%Y-%m-%d %H:%M:%S %Z", time.localtime()))
# DATA EXTRACTION
xml_reader_name = '-'.join([self.name, 'xml_reader'])
logitem_proc_name = '-'.join([self.name, 'process_logitem'])
logitem_insert_name = '-'.join([self.name, 'insert_logitem'])
# Start subprocess to extract elements from logging dump file
file_path = self.path[0]
dump_file = DumpFile(file_path)
xml_reader = Producer(name=xml_reader_name,
target=process_xml,
kwargs=dict(dump_file=dump_file),
consumers=self.log_fan,
push_logs_port=self.base_port,
control_port=self.control_port)
xml_reader.start()
print xml_reader_name, "started"
print self.name, "Extracting data from XML revision history file:"
print unicode(self.path[0])
# List to keep tracking of logitem workers
workers = []
# Create and start page processes
for worker in range(self.log_fan):
worker_name = '-'.join([logitem_proc_name, unicode(worker)])
process_logitems = Processor(name=worker_name,
target=logitem_to_file,
producers=1,
consumers=1,
pull_port=self.base_port,
push_port=self.base_port + 2,
control_port=self.control_port)
process_logitems.start()
workers.append(process_logitems)
print worker_name, "started"
# Create directory for logging files if it does not exist
log_dir = os.path.join(os.path.split(file_path)[0], 'logs')
tmp_dir = os.path.join(os.getcwd(), os.path.split(file_path)[0], 'tmp')
file_name = os.path.split(file_path)[1]
if not os.path.exists(log_dir):
os.makedirs(log_dir)
if not os.path.exists(tmp_dir):
os.makedirs(tmp_dir)
log_file = os.path.join(log_dir, file_name + '.log')
db_log = MySQLDB(host='localhost',
port=3306,
user=self.db_user,
passwd=self.db_passw,
db=self.db_name)
db_log.connect()
logitem_insert_db = Consumer(name=logitem_insert_name,
target=logitem_file_to_db,
kwargs=dict(con=db_log,
log_file=log_file,
tmp_dir=tmp_dir,
file_rows=self.log_cache_size,
etl_prefix=self.name),
producers=self.log_fan,
pull_port=self.base_port + 2)
print logitem_insert_name, "started"
logitem_insert_db.start()
print "Waiting for all processes to finish..."
print
xml_reader.join()
for w in workers:
w.join()
logitem_insert_db.join()
# All operations finished
end = time.time()
print "All tasks done in %.4f sec." % ((end - start) / 1.)
print
db_log.close()
def run(self):
"""
Execute workflow to import revision history data from dump files
The data loading workflow is composed of a number of processor
elements, which can be:
- Producer (P): raw input data --> input element queue
- ConsumerProducer (CP): input element queue --> insert db queue
- Consumer (C): insert db queue --> database (MySQL/MariaDB)
In this case, the logical combination is usually N:N:1 (P, CP, C)
"""
start = time.time()
print self.name, "Starting PageRevisionETL workflow at %s" % (
time.strftime("%Y-%m-%d %H:%M:%S %Z",
time.localtime()))
db_ns = MySQLDB(host='localhost', port=3306, user=self.db_user,
passwd=self.db_passw, db=self.db_name)
db_ns.connect()
db_pages = MySQLDB(host='localhost', port=3306,
user=self.db_user, passwd=self.db_passw,
db=self.db_name)
db_pages.connect()
db_revs = MySQLDB(host='localhost', port=3306, user=self.db_user,
passwd=self.db_passw, db=self.db_name)
db_revs.connect()
# DATA EXTRACTION
# Use consistent naming for all child processes
xml_reader_name = '-'.join([self.name, 'xml_reader'])
page_proc_name = '-'.join([self.name, 'process_page'])
rev_proc_name = '-'.join([self.name, 'process_revision'])
page_insert_name = '-'.join([self.name, 'insert_page'])
rev_insert_name = '-'.join([self.name, 'insert_revision'])
for path in iter(self.paths_queue.get, 'STOP'):
# Start subprocess to extract elements from revision dump file
dump_file = DumpFile(path)
xml_reader = Producer(name=xml_reader_name,
target=process_xml,
kwargs=dict(
dump_file=dump_file),
consumers=self.page_fan + self.rev_fan,
push_pages_port=self.base_port,
push_revs_port=self.base_port+1,
control_port=self.control_port)
xml_reader.start()
print xml_reader_name, "started"
print self.name, "Extracting data from XML revision history file:"
print path
# List to keep tracking of page and revision workers
workers = []
db_workers_revs = []
# Create and start page processes
for worker in range(self.page_fan):
page_worker_name = '-'.join([page_proc_name, unicode(worker)])
process_page = Processor(name=page_worker_name,
target=pages_to_file,
producers=1, consumers=1,
pull_port=self.base_port,
push_port=self.base_port+2,
control_port=self.control_port)
process_page.start()
workers.append(process_page)
print page_worker_name, "started"
# Create and start revision processes
for worker in range(self.rev_fan):
rev_worker_name = '-'.join([rev_proc_name, unicode(worker)])
db_wrev = MySQLDB(host='localhost', port=3306,
user=self.db_user,
passwd=self.db_passw, db=self.db_name)
db_wrev.connect()
process_revision = Processor(name=rev_worker_name,
target=revs_to_file,
kwargs=dict(
lang=self.lang),
producers=1, consumers=1,
pull_port=self.base_port+1,
push_port=self.base_port+3,
control_port=self.control_port)
process_revision.start()
workers.append(process_revision)
db_workers_revs.append(db_wrev)
print rev_worker_name, "started"
# Create directory for logging files if it does not exist
log_dir = os.path.join(os.path.split(path)[0], 'logs')
tmp_dir = os.path.join(os.getcwd(), os.path.split(path)[0], 'tmp')
file_name = os.path.split(path)[1]
if not os.path.exists(log_dir):
os.makedirs(log_dir)
if not os.path.exists(tmp_dir):
os.makedirs(tmp_dir)
log_file = os.path.join(log_dir, file_name + '.log')
page_insert_db = Consumer(name=page_insert_name,
target=pages_file_to_db,
kwargs=dict(con=db_pages,
log_file=log_file,
tmp_dir=tmp_dir,
file_rows=self.page_cache_size,
etl_prefix=self.name),
producers=self.page_fan,
pull_port=self.base_port+2)
rev_insert_db = Consumer(name=rev_insert_name,
target=revs_file_to_db,
kwargs=dict(con=db_revs,
log_file=log_file,
tmp_dir=tmp_dir,
file_rows=self.rev_cache_size,
etl_prefix=self.name),
producers=self.rev_fan,
pull_port=self.base_port+3)
page_insert_db.start()
print page_insert_name, "started"
rev_insert_db.start()
print rev_insert_name, "started"
print self.name, "Waiting for all processes to finish..."
print
xml_reader.join()
for w in workers:
w.join()
page_insert_db.join()
rev_insert_db.join()
# Mark this path as done
self.paths_queue.task_done()
# Mark STOP message as processed and finish
self.paths_queue.task_done()
end = time.time()
print self.name, ": All tasks done in %.4f sec." % ((end-start)/1.)
print
db_ns.close()
db_pages.close()
db_revs.close()
for dbcon in db_workers_revs:
dbcon.close()
def run(self):
"""
Execute workflow to import logging records of actions on pages and
users from dump file
The data loading workflow is composed of a number of processor
elements, which can be:
- Producer (P): raw input data --> input element queue
- ConsumerProducer (CP): input element queue --> insert db queue
- Consumer (C): insert db queue --> database (MySQL/MariaDB)
In this case, the usual combination is 1:N:1 (P, CP, C)
"""
start = time.time()
print "Starting LoggingETL workflow at %s" % (
time.strftime("%Y-%m-%d %H:%M:%S %Z",
time.localtime()))
# DATA EXTRACTION
xml_reader_name = '-'.join([self.name, 'xml_reader'])
logitem_proc_name = '-'.join([self.name, 'process_logitem'])
logitem_insert_name = '-'.join([self.name, 'insert_logitem'])
# Start subprocess to extract elements from logging dump file
file_path = self.path[0]
dump_file = DumpFile(file_path)
xml_reader = Producer(name=xml_reader_name,
target=process_xml,
kwargs=dict(
dump_file=dump_file),
consumers=self.log_fan,
push_logs_port=self.base_port,
control_port=self.control_port)
xml_reader.start()
print xml_reader_name, "started"
print self.name, "Extracting data from XML revision history file:"
print unicode(self.path[0])
# List to keep tracking of logitem workers
workers = []
# Create and start page processes
for worker in range(self.log_fan):
worker_name = '-'.join([logitem_proc_name, unicode(worker)])
process_logitems = Processor(name=worker_name,
target=logitem_to_file,
producers=1, consumers=1,
pull_port=self.base_port,
push_port=self.base_port+2,
control_port=self.control_port)
process_logitems.start()
workers.append(process_logitems)
print worker_name, "started"
# Create directory for logging files if it does not exist
log_dir = os.path.join(os.path.split(file_path)[0], 'logs')
tmp_dir = os.path.join(os.getcwd(), os.path.split(file_path)[0], 'tmp')
file_name = os.path.split(file_path)[1]
if not os.path.exists(log_dir):
os.makedirs(log_dir)
if not os.path.exists(tmp_dir):
os.makedirs(tmp_dir)
log_file = os.path.join(log_dir, file_name + '.log')
db_log = MySQLDB(host='localhost', port=3306, user=self.db_user,
passwd=self.db_passw, db=self.db_name)
db_log.connect()
logitem_insert_db = Consumer(name=logitem_insert_name,
target=logitem_file_to_db,
kwargs=dict(con=db_log,
log_file=log_file,
tmp_dir=tmp_dir,
file_rows=self.log_cache_size,
etl_prefix=self.name),
producers=self.log_fan,
pull_port=self.base_port+2)
print logitem_insert_name, "started"
logitem_insert_db.start()
print "Waiting for all processes to finish..."
print
xml_reader.join()
for w in workers:
w.join()
logitem_insert_db.join()
# All operations finished
end = time.time()
print "All tasks done in %.4f sec." % ((end-start)/1.)
print
db_log.close()
def execute(self,
log_fan,
log_cache_size,
mirror,
download_files,
base_ports,
control_ports,
dumps_dir=None,
debug=False):
"""
Run data retrieval and loading actions.
Arguments:
- log_fan = Number of workers to fan out logitem elements parsing
- db_user = User name to connect to local database
- db_passw = Password for database user
- mirror = Base URL of site hosting XML dumps
"""
print("----------------------------------------------------------")
print("Executing ETL:PagesLogging on lang: {0} date: {1}".format(
self.lang, self.date))
print("log_fan =", log_fan)
print("Download files =", download_files)
print("Start time is {0}".format(
time.strftime("%Y-%m-%d %H:%M:%S %Z", time.localtime())))
print("----------------------------------------------------------")
print()
if download_files:
# TODO: Use proper logging module to track execution progress
# Choose corresponding file downloader and etl wrapper
print("""Downloading new logging dump files from %s,
for language %s""" % (mirror, self.lang))
self.down = LoggingDownloader(mirror, self.lang, dumps_dir)
# Donwload latest set of dump files
self.paths, self.date = self.down.download(self.date)
if not self.paths:
print("Error: dump files with pages-logging info not found.")
print("Program will exit now.")
sys.exit()
print("Got files for lang %s, date: %s" % (self.lang, self.date))
else:
print("Looking for pages-logging dump file in data dir")
# Case of dumps folder provided explicity
if dumps_dir:
# Allow specifying relative paths, as well
abs_dumps_path = os.path.expanduser(dumps_dir)
dumps_path = os.path.join(abs_dumps_path, self.lang + '_dumps',
self.date)
# Retrieve path to all available files to feed ETL lines
if not os.path.exists(dumps_path):
print(
"No dump files will be downloaded and local folder with dump files not found."
)
print(
"Please, specify a valid path to local folder containing dump files."
)
print("Program will exit now.")
sys.exit()
else:
# Attempt to find list of *page-logging*.gz or
# *page-logging*.xml files to be processed
self.paths = glob.glob(
os.path.join(dumps_path, '*pages-logging*.gz'))
if not self.paths:
self.paths = glob.glob(
os.path.join(dumps_path, '*pages-logging*.xml'))
if not self.paths:
print(
"Directory %s does not contain any valid dump file."
% dumps_path)
print("Program will exit now.")
sys.exit()
# If not provided explicitly, look for default location of
# dumps directory
else:
dumps_dir = os.path.join("data", self.lang + '_dumps',
self.date)
# Look up dump files in default directory name
if not os.path.exists(dumps_dir):
print(
"Default directory %s containing dump files not found."
% dumps_dir)
print("Program will exit now.")
sys.exit()
else:
self.paths = glob.glob(
os.path.join(dumps_dir, '*pages-logging*.gz'))
if not self.paths:
self.paths = glob.glob(
os.path.join(dumps_dir, '*pages-logging*.xml'))
if not self.paths:
print(
"Directory %s does not contain any valid dump file."
% dumps_dir)
print("Program will exit now.")
sys.exit()
print("Found pages-logging dump file to process.")
print()
if debug:
print("paths: ", str(self.paths))
print()
# Create database if it does not exist
# empty logging table otherwise
if self.DB_exists():
self.create_DB(complete=False)
else:
self.create_DB(complete=True)
new_etl = LoggingETL(name="[ETL:PagesLogging-0]",
path=self.paths,
lang=self.lang,
log_fan=log_fan,
log_cache_size=log_cache_size,
db_name=self.db_name,
db_user=self.db_user,
db_passw=self.db_passw,
base_port=base_ports[0] + (30),
control_port=control_ports[0] + (30))
print("ETL:Logging task for administrative records defined OK.")
print("Proceeding with ETL workflow. This may take time...")
print()
# Extract, process and load information in local DB
new_etl.start()
# Wait for ETL line to finish
new_etl.join()
# TODO: logger; ETL step completed, proceeding with data
# analysis and visualization
print("ETL:Logging task finished for lang %s and date %s" %
(self.lang, self.date))
print()
# Create primary keys for all tables
# TODO: This must also be tracked by official logging module
print("Now creating primary key indexes in database tables.")
print("This may take a while...")
print()
db_pks = MySQLDB(host='localhost',
port=3306,
user=self.db_user,
passwd=self.db_passw,
db=self.db_name)
db_pks.connect()
db_pks.create_pks_logitem()
db_pks.close()
def create_DB(self, complete=False):
if complete:
db_create = MySQLDB(host=self.host,
port=self.port,
user=self.db_user,
passwd=self.db_passw)
db_create.connect()
db_create.create_database(self.db_name)
db_create.close()
db_schema = MySQLDB(host=self.host,
port=self.port,
user=self.db_user,
passwd=self.db_passw,
db=self.db_name)
db_schema.connect()
db_schema.create_schema_logitem(engine=self.db_engine)
db_schema.close()
def execute(self,
page_fan,
rev_fan,
page_cache_size,
rev_cache_size,
mirror,
download_files,
base_ports,
control_ports,
dumps_dir=None,
debug=False):
"""
Run data retrieval and loading actions.
Arguments:
- page_fan = Number of workers to fan out page elements parsing
- rev_fan = Number of workers to fan out rev elements parsing
- db_user = User name to connect to local database
- db_passw = Password for database user
- mirror = Base URL of site hosting XML dumps
"""
print("----------------------------------------------------------")
print(("""Executing ETL:RevHistory on lang: {0} date: {1}""".format(
self.lang, self.date)))
print(("ETL lines = {0} page_fan = {1} rev_fan = {2}".format(
self.etl_lines, page_fan, rev_fan)))
print("Download files =", download_files)
print("Start time is {0}".format(
time.strftime("%Y-%m-%d %H:%M:%S %Z", time.localtime())))
print("----------------------------------------------------------")
print()
if download_files:
# TODO: Use proper logging module to track execution progress
# Choose corresponding file downloader and etl wrapper
print("Downloading new dump files from %s, for language %s" %
(mirror, self.lang))
self.down = RevHistDownloader(mirror, self.lang, dumps_dir)
# Donwload latest set of dump files
self.paths, self.date = self.down.download(self.date)
if not self.paths:
print("Error: dump files with pages-logging info not found.")
print("Program will exit now.")
sys.exit()
print("Retrieved dump files for lang %s, date: %s" %
(self.lang, self.date))
print()
else:
print("Looking for revision-history dump file(s) in data dir")
# Case of dumps folder provided explicity
if dumps_dir:
# Allow specifying relative paths, as well
abs_dumps_path = os.path.expanduser(dumps_dir)
dumps_path = os.path.join(abs_dumps_path, self.lang + '_dumps',
self.date)
# Retrieve path to all available files to feed ETL lines
if not os.path.exists(dumps_path):
print(
"No dump files will be downloaded and local folder with dump files not found."
)
print(
"Please, specify a valid path to local folder containing dump files."
)
print("Program will exit now.")
sys.exit()
else:
# Attempt to find list of .7z or .xml files to be processed
self.paths = glob.glob(
os.path.join(dumps_path, '*pages-meta-history*.7z'))
if not self.paths:
self.paths = glob.glob(
os.path.join(dumps_path,
'*pages-meta-history*.xml'))
if not self.paths:
print(
"Directory %s does not contain any valid dump file."
% dumps_path)
print("Program will exit now.")
sys.exit()
# If not provided explicitly, look for default location of
# dumps directory
else:
dumps_dir = os.path.join("data", self.lang + '_dumps',
self.date)
# Look up dump files in default directory name
if not os.path.exists(dumps_dir):
print(
"Default directory %s containing dump files not found."
% dumps_dir)
print("Program will exit now.")
sys.exit()
else:
self.paths = glob.glob(
os.path.join(dumps_dir, '*pages-meta-history*.7z'))
if not self.paths:
self.paths = glob.glob(
os.path.join(dumps_dir,
'*pages-meta-history*.xml'))
if not self.paths:
print(
"Directory %s does not contain any valid dump file."
% dumps_dir)
print("Program will exit now.")
sys.exit()
print("Found revision-history dump file(s) to process.")
print()
# Print list of file paths in debug mode
if debug:
print("paths: ", str(self.paths))
print()
# Create database
# TODO: Empty correspoding tables if DB already exists
# or let the user select behaviour with config argument
if self.DB_exists():
self.create_DB(complete=False)
else:
self.create_DB(complete=True)
# First insert namespace info in DB
dump = DumpFile(self.paths[0])
db_schema = MySQLDB(host=self.host,
port=self.port,
user=self.db_user,
passwd=self.db_passw,
db=self.db_name)
db_schema.connect()
db_schema.insert_namespaces(nsdict=dump.get_namespaces())
db_schema.close()
# Complete the queue of paths to be processed and STOP flags for
# each ETL subprocess
paths_queue = mp.JoinableQueue()
for path in self.paths:
paths_queue.put(path)
for x in range(self.etl_lines):
paths_queue.put('STOP')
for x in range(self.etl_lines):
new_etl = RevisionHistoryETL(name="[ETL:RevHistory-%s]" % x,
paths_queue=paths_queue,
lang=self.lang,
page_fan=page_fan,
rev_fan=rev_fan,
page_cache_size=page_cache_size,
rev_cache_size=rev_cache_size,
db_name=self.db_name,
db_user=self.db_user,
db_passw=self.db_passw,
base_port=base_ports[x] + (20 * x),
control_port=control_ports[x] +
(20 * x))
self.etl_list.append(new_etl)
print("ETL:RevHistory task defined OK.")
print("Proceeding with ETL workflows. This may take time...")
print()
# Extract, process and load information in local DB
for etl in self.etl_list:
etl.start()
# Wait a second for new ETL process to start all subprocesses
time.sleep(1)
# Wait for ETL lines to finish
for etl in self.etl_list:
etl.join()
# Insert user info after all ETL lines have finished
# to ensure that all metadata are stored in Redis cache
# disregarding of the execution order
data_dir = os.path.join(os.getcwd(), os.path.split(self.paths[0])[0])
db_users = MySQLDB(host=self.host,
port=self.port,
user=self.db_user,
passwd=self.db_passw,
db=self.db_name)
db_users.connect()
users_file_to_db(con=db_users,
lang=self.lang,
log_file=os.path.join(data_dir, 'logs', 'users.log'),
tmp_dir=os.path.join(data_dir, 'tmp'))
db_users.close()
# TODO: logger; ETL step completed, proceeding with data
# analysis and visualization
print("ETL:RevHistory task finished for language %s and date %s" %
(self.lang, self.date))
print()
# Create primary keys for all tables
# TODO: This must also be tracked by main logging module
print("Now creating primary key indexes in database tables.")
print("This may take a while...")
print()
db_pks = MySQLDB(host='localhost',
port=3306,
user=self.db_user,
passwd=self.db_passw,
db=self.db_name)
db_pks.connect()
db_pks.create_pks_revhist()
db_pks.close()
