Cassandra Scripts

.gitignore

cassandra_config.py
*.pyc
project/

__init__.py

README.md

Cassandra CQL3 Scripts

Executables

To get a list of the keyspaces and column types in the database, use:

python list_keyspaces.py

Scripting

To print out the return of a select statement, use cql_print_select:

from cassandra_driver import query
from cassandra_cql3_formatting import cql_print_select

call_result = query('SELECT * FROM projects')
cql_print_select(call_result)

This will return:

| key                                  || name         || provider_id      |
-----------------------------------------------------------------------------
| df95812e-2d0f-11e3-97a3-080027880ca6 || New Story    || archive_storage  |
| b1325276-2d14-11e3-9fed-080027880ca6 || Fixing Bugs  || new_wave_systems |

cassandra_config.template.py

#
# Fill in this information and remove the 'template' from the filename.
#

keyspace = 'keyspace'
hostnames = ['127.0.0.1']
port = 9042

cassandra_cql3_formatting.py

#
# CQL3 Output Formatting
#
import uuid
import sys

def _measure_width(results, column):
    full_width = len(column)
    for item in results:
        # if type(item.__getattribute__(column)) is uuid.UUID:
        #     return 45
        # elif is int:
        #     current_width = len(str(item.__getattribute__(column)))
        # else:
        current_width = len(str(item.__getattribute__(column)))
        if current_width > full_width:
            full_width = current_width
    return full_width

def _line_break(column_width, columns):
    total_length = 0
    for column in columns:
        total_length += 4 + column_width[column]
    return ''.ljust(total_length, '-')

def cql_print_select(results):
    columns = results[0]._fields
    column_width = {}
    for column in columns:
        column_width[column] = _measure_width(results, column)
        sys.stdout.write("| {0} |".format(column.ljust(column_width[column], ' ')))

    sys.stdout.write('\n{0}\n'.format(_line_break(column_width, columns)))
    for item in results:
        for column in columns:
            sys.stdout.write("| {0} |".format(str(item.__getattribute__(column)).ljust(column_width[column], ' ')))
        sys.stdout.write('\n')

cassandra_driver.py

LOGGING_ON = False

if LOGGING_ON:
    import logging

    log = logging.getLogger()
    log.setLevel('DEBUG')
    handler = logging.StreamHandler()
    handler.setFormatter(logging.Formatter("%(asctime)s [%(levelname)s] %(name)s: %(message)s"))
    log.addHandler(handler)

from cassandra.cluster import Cluster
import cassandra_config as config

class Connection(object):

    def __init__(self):
        super(Connection, self).__init__()
        "".encode('utf8')
        self.cluster = Cluster(config.hostnames, port=config.port)
        self.connect()

    def connect(self):
        self.session = self.cluster.connect(keyspace=config.keyspace)

connection = Connection()
def query(query_string, parameters=None):
    return connection.session.execute(query_string, parameters=parameters)

list_keyspaces.py

SHOW_ALL = False

from cassandra_driver import query

rows = query("SELECT * FROM system.schema_columns;")

keyspaces = {}

for row in rows:
    if SHOW_ALL or (not row.keyspace_name == 'system' and not row.keyspace_name == 'system_traces'):
        if not row.keyspace_name in keyspaces:
            keyspaces[row.keyspace_name] = {}
        if not row.columnfamily_name in keyspaces[row.keyspace_name]:
            keyspaces[row.keyspace_name][row.columnfamily_name] = []
        if row.type == 'partition_key':
            keyspaces[row.keyspace_name][row.columnfamily_name].insert(0, {
                'name': '{0}'.format(row.column_name),
                'type': row.validator.replace('org.apache.cassandra.db.marshal.', ''),
                'key' : True
            })
        else:
            keyspaces[row.keyspace_name][row.columnfamily_name].append({
                'name': row.column_name,
                'type': row.validator.replace('org.apache.cassandra.db.marshal.', ''),
                'key' : False
            })

for keys in keyspaces.keys():
    print '{0}:'.format(keys)
    for columnfamily in keyspaces[keys].keys():
        print '    {0}'.format(columnfamily)
        for item in keyspaces[keys][columnfamily]:
            key = ' '
            if item['key']:
                key = '*'
            print '       {0} {1}{2}'.format(key, item['name'].ljust(28, ' '), item['type'])

requirements.txt

-e git+https://github.com/datastax/python-driver.git#egg=cassandra-driver
#cassandra-driver==1.0.0-beta4