def timestamp_verifier(self):
"""
Return the time at which the data was last updated. If the
value returned by the function is newer than the cache, the
cache will be invalidated.
@return: string-formatted time '%Y-%m-%d %H:%M:%S'
"""
return max(get_table_update_time('collectionname'),
get_table_update_time('collection_collection'))
def timestamp_verifier(self):
"""
Return the time at which the data was last updated. If the
value returned by the function is newer than the cache, the
cache will be invalidated.
@return: string-formatted time '%Y-%m-%d %H:%M:%S'
"""
return max(get_table_update_time('collectionname'),
get_table_update_time('collection_collection'))
def _check_table_update_time(self, tablename):
"""Helper function to check update time of TABLENAME."""
# detect MySQL version number:
res = dbquery.run_sql("SELECT VERSION()")
mysql_server_version = res[0][0]
if mysql_server_version.startswith("5."):
# MySQL-5 provides INFORMATION_SCHEMA:
query = """SELECT UPDATE_TIME FROM INFORMATION_SCHEMA.TABLES
WHERE table_name='%s' AND table_schema='%s'""" % (
tablename,
dbquery.CFG_DATABASE_NAME,
)
tablename_update_time = str(dbquery.run_sql(query)[0][0])
elif mysql_server_version.startswith("4.1"):
# MySQL-4.1 has it on 12th position:
query = """SHOW TABLE STATUS LIKE '%s'""" % tablename
tablename_update_time = str(dbquery.run_sql(query)[0][12])
elif mysql_server_version.startswith("4.0"):
# MySQL-4.0 has it on 11th position:
query = """SHOW TABLE STATUS LIKE '%s'""" % tablename
tablename_update_time = str(dbquery.run_sql(query)[0][11])
else:
tablename_update_time = "MYSQL SERVER VERSION NOT DETECTED"
# compare it with the one detected by the function:
self.assertEqual(tablename_update_time, dbquery.get_table_update_time(tablename))
def timestamp_verifier(self):
"""
Return the time at which the data was last updated. If the
value returned by the function is newer than the cache, the
cache will be invalidated.
@return: string-formatted time '%Y-%m-%d %H:%M:%S'
"""
# This is an approximation...
return get_table_update_time('knwKBRVAL')
def timestamp_verifier(self):
"""
Return the time at which the data was last updated. If the
value returned by the function is newer than the cache, the
cache will be invalidated.
@return: string-formatted time '%Y-%m-%d %H:%M:%S'
"""
# This is an approximation...
return get_table_update_time('knwKBRVAL')
def get_database_last_updated_timestamp():
"""Return last updated timestamp for collection-related and
record-related database tables.
"""
database_tables_timestamps = []
database_tables_timestamps.append(get_table_update_time('bibrec'))
database_tables_timestamps.append(get_table_update_time('bibfmt'))
database_tables_timestamps.append(get_table_update_time('idxWORD%'))
database_tables_timestamps.append(get_table_update_time('collection%'))
database_tables_timestamps.append(get_table_update_time('portalbox'))
database_tables_timestamps.append(get_table_update_time('field%'))
database_tables_timestamps.append(get_table_update_time('format%'))
database_tables_timestamps.append(get_table_update_time('rnkMETHODNAME'))
return max(database_tables_timestamps)
def get_database_last_updated_timestamp():
"""Return last updated timestamp for collection-related and
record-related database tables.
"""
database_tables_timestamps = []
database_tables_timestamps.append(get_table_update_time('bibrec'))
database_tables_timestamps.append(get_table_update_time('bibfmt'))
database_tables_timestamps.append(get_table_update_time('idxWORD%'))
database_tables_timestamps.append(get_table_update_time('collection%'))
database_tables_timestamps.append(get_table_update_time('portalbox'))
database_tables_timestamps.append(get_table_update_time('field%'))
database_tables_timestamps.append(get_table_update_time('format%'))
database_tables_timestamps.append(get_table_update_time('rnkMETHODNAME'))
return max(database_tables_timestamps)
def get_database_last_updated_timestamp():
"""Return last updated timestamp for collection-related and
record-related database tables.
"""
database_tables_timestamps = []
database_tables_timestamps.append(get_table_update_time('bibrec'))
database_tables_timestamps.append(get_table_update_time('bibfmt'))
try:
database_tables_timestamps.append(get_table_update_time('idxWORD%'))
except ValueError:
# There are no indexes in the database. That's OK.
pass
database_tables_timestamps.append(get_table_update_time('collection%'))
database_tables_timestamps.append(get_table_update_time('portalbox'))
database_tables_timestamps.append(get_table_update_time('field%'))
database_tables_timestamps.append(get_table_update_time('format%'))
database_tables_timestamps.append(get_table_update_time('rnkMETHODNAME'))
database_tables_timestamps.append(get_table_update_time('accROLE_accACTION_accARGUMENT'))
return max(database_tables_timestamps)
def get_database_last_updated_timestamp():
"""Return last updated timestamp for collection-related and
record-related database tables.
"""
database_tables_timestamps = []
database_tables_timestamps.append(get_table_update_time("bibrec"))
database_tables_timestamps.append(get_table_update_time("bibfmt"))
try:
database_tables_timestamps.append(get_table_update_time("idxWORD%"))
except ValueError:
# There are no indexes in the database. That's OK.
pass
database_tables_timestamps.append(get_table_update_time("collection%"))
database_tables_timestamps.append(get_table_update_time("portalbox"))
database_tables_timestamps.append(get_table_update_time("field%"))
database_tables_timestamps.append(get_table_update_time("format%"))
database_tables_timestamps.append(get_table_update_time("rnkMETHODNAME"))
database_tables_timestamps.append(get_table_update_time("accROLE_accACTION_accARGUMENT", run_on_slave=True))
return max(database_tables_timestamps)
def calculate_index_term_count(config):
"""Calculate the weight of a record set based on number of enries of a
tag from the record in another index...useful for authority files"""
records = []
if config.has_section("index_term_count"):
index = config.get("index_term_count", "index_table_name")
tag = config.get("index_term_count", "index_term_value_from_tag")
# check against possible SQL injection:
dummy = get_table_update_time(index)
tag = wash_table_column_name(tag)
else:
raise Exception("Config file " + config +
" does not have index_term_count section")
return ()
task_sleep_now_if_required(can_stop_too=True)
write_message("......Processing all records")
query = "SELECT id_bibrec, value FROM bib%sx, bibrec_bib%sx WHERE tag=%%s AND id_bibxxx=id" % \
(tag[0:2], tag[0:2]) # we checked that tag is safe
records = list(run_sql(query, (tag, )))
write_message("Number of records found with the necessary tags: %s" %
len(records))
rnkset = {}
for key, value in records:
hits = 0
if len(value):
query = "SELECT hitlist from %s where term = %%s" % index # we checked that index is a table
row = run_sql(query, (value, ))
if row and row[0] and row[0][0]:
#has to be prepared for corrupted data!
try:
hits = len(intbitset(row[0][0]))
except:
hits = 0
rnkset[key] = hits
write_message("Number of records available in rank method: %s" %
len(rnkset))
return rnkset
def calculate_index_term_count(config):
"""Calculate the weight of a record set based on number of enries of a
tag from the record in another index...useful for authority files"""
records = []
if config.has_section("index_term_count"):
index = config.get("index_term_count","index_table_name")
tag = config.get("index_term_count","index_term_value_from_tag")
# check against possible SQL injection:
dummy = get_table_update_time(index)
tag = wash_table_column_name(tag)
else:
raise Exception("Config file " + config + " does not have index_term_count section")
return()
task_sleep_now_if_required(can_stop_too=True)
write_message("......Processing all records")
query = "SELECT id_bibrec, value FROM bib%sx, bibrec_bib%sx WHERE tag=%%s AND id_bibxxx=id" % \
(tag[0:2], tag[0:2]) # we checked that tag is safe
records = list(run_sql(query, (tag,)))
write_message("Number of records found with the necessary tags: %s" % len(records))
rnkset = {}
for key, value in records:
hits = 0
if len(value):
query = "SELECT hitlist from %s where term = %%s" % index # we checked that index is a table
row = run_sql(query, (value,))
if row and row[0] and row[0][0]:
#has to be prepared for corrupted data!
try:
hits = len(intbitset(row[0][0]))
except:
hits = 0
rnkset[key] = hits
write_message("Number of records available in rank method: %s" % len(rnkset))
return rnkset
def _check_table_update_time(self, tablename):
"""Helper function to check update time of TABLENAME."""
# detect MySQL version number:
res = dbquery.run_sql("SELECT VERSION()")
mysql_server_version = res[0][0]
if mysql_server_version.startswith("5."):
# MySQL-5 provides INFORMATION_SCHEMA:
query = """SELECT UPDATE_TIME FROM INFORMATION_SCHEMA.TABLES
WHERE table_name='%s'""" % tablename
tablename_update_time = str(dbquery.run_sql(query)[0][0])
elif mysql_server_version.startswith("4.1"):
# MySQL-4.1 has it on 12th position:
query = """SHOW TABLE STATUS LIKE '%s'""" % tablename
tablename_update_time = str(dbquery.run_sql(query)[0][12])
elif mysql_server_version.startswith("4.0"):
# MySQL-4.0 has it on 11th position:
query = """SHOW TABLE STATUS LIKE '%s'""" % tablename
tablename_update_time = str(dbquery.run_sql(query)[0][11])
else:
tablename_update_time = "MYSQL SERVER VERSION NOT DETECTED"
# compare it with the one detected by the function:
self.assertEqual(tablename_update_time,
dbquery.get_table_update_time(tablename))
def timestamp_verifier():
"""The standard timestamp verifier is looking at affected
tables time stamp."""
return max([get_table_update_time(table)
for table in self.affected_tables])
def timestamp_verifier():
"""The standard timestamp verifier is looking at affected
tables time stamp."""
return max([
get_table_update_time(table) for table in self.affected_tables
])
def timestamp_verifier():
return get_table_update_time("rnkCITATIONDATA")
