tools/upgrade-db.py

   1 #!/usr/bin/python
   2 #
   3 # Tool for upgrading/converting a db
   4 # Requirements:
   5 # 1) Databse Schema - schema for the new database you what to upgrade to
   6 # 2) Config File - the config file that describes how to convert the db
   7 #
   8 # Notes:
   9 # 1) Will attempt to convert the db defined in  /etc/planetlab/plc_config
  10 # 2) Does not automatically drop archived database. They must be removed
  11 #    manually
  12
  13 import sys
  14 import os
  15 import getopt
  16 import pgdb
  17
  18 config = {}
  19 config_file = "/etc/planetlab/plc_config"
  20 execfile(config_file, config)
  21 upgrade_config_file = "plcdb.3-4.conf"
  22 schema_file = "planetlab4.sql"
  23 temp_dir = "/tmp"
  24
  25
  26 def usage():
  27         print "Usage: %s [OPTION] UPGRADE_CONFIG_FILE " % sys.argv[0]
  28         print "Options:"
  29         print "     -s, --schema=FILE       Upgraded Database Schema"
  30         print "     -t, --temp-dir=DIR      Temp Directory"
  31         print "     --help                  This message"
  32         sys.exit(1)
  33
  34 try:
  35         (opts, argv) = getopt.getopt(sys.argv[1:],
  36                                      "s:d:",
  37                                      ["schema=",
  38                                       "temp-dir=",
  39                                       "help"])
  40 except getopt.GetoptError, err:
  41         print "Error: ", err.msg
  42         usage()
  43
  44 for (opt, optval) in opts:
  45         if opt == "-s" or opt == "--schema":
  46                 schema_file = optval
  47         elif opt == "-d" or opt == "--temp-dir":
  48                 temp_dir = optval
  49         elif opt == "--help":
  50                 usage()
  51 try:
  52         upgrade_config_file = argv[0]
  53 except IndexError:
  54         print "Error: too few arguments"
  55         usage()
  56
  57 schema = {}
  58 inserts = []
  59 schema_items_ordered = []
  60 sequences = {}
  61 temp_tables = {}
  62
  63
  64 # load conf file for this upgrade
  65 try:
  66         upgrade_config = {}
  67         execfile(upgrade_config_file, upgrade_config)
  68         upgrade_config.pop('__builtins__')
  69         db_version_previous = upgrade_config['DB_VERSION_PREVIOUS']
  70         db_version_new = upgrade_config['DB_VERSION_NEW']
  71
  72 except IOError, fault:
  73         print "Error: upgrade config file (%s) not found. Exiting" % \
  74                 (fault)
  75         sys.exit(1)
  76 except KeyError, fault:
  77         print "Error: %s not set in upgrade confing (%s). Exiting" % \
  78                 (fault, upgrade_config_file)
  79         sys.exit(1)
  80
  81
  82
  83
  84 def connect():
  85         db = pgdb.connect(user = config['PLC_DB_USER'],
  86                   database = config['PLC_DB_NAME'])
  87         return db
  88
  89 def archive_db(database, archived_database):
  90
  91         archive_db = " dropdb -U postgres %s > /dev/null 2>&1;" \
  92                      " psql template1 postgres -qc " \
  93                      " 'ALTER DATABASE %s RENAME TO %s;';" % \
  94                      (archived_database, database, archived_database)
  95         exit_status = os.system(archive_db)
  96         if exit_status:
  97                 print "Error: unable to archive database. Upgrade failed"
  98                 sys.exit(1)
  99         #print "Status: %s has been archived. now named %s" % (database, archived_database)
 100
 101
 102 def encode_utf8(inputfile_name, outputfile_name):
 103         # rewrite a iso-8859-1 encoded file in utf8
 104         try:
 105                 inputfile = open(inputfile_name, 'r')
 106                 outputfile = open(outputfile_name, 'w')
 107                 for line in inputfile:
 108                         if line.upper().find('SET CLIENT_ENCODING') > -1:
 109                                 continue
 110                         outputfile.write(unicode(line, 'iso-8859-1').encode('utf8'))
 111                 inputfile.close()
 112                 outputfile.close()
 113         except:
 114                 print 'error encoding file'
 115                 raise
 116
 117 def create_item_from_schema(item_name):
 118
 119         try:
 120                 (type, body_list) = schema[item_name]
 121                 exit_status = os.system('psql %s %s -qc "%s" > /dev/null 2>&1' % \
 122                             (config['PLC_DB_NAME'], config['PLC_DB_USER'],"".join(body_list) ) )
 123                 if exit_status:
 124                         raise Exception
 125         except Exception, fault:
 126                 print 'Error: create %s failed. Check schema.' % item_name
 127                 sys.exit(1)
 128                 raise fault
 129
 130         except KeyError:
 131                 print "Error: cannot create %s. definition not found in %s" % \
 132                         (key, schema_file)
 133                 return False
 134
 135 def fix_row(row, table_name, table_fields):
 136
 137         if table_name in ['nodenetworks']:
 138                 # convert str bwlimit to bps int
 139                 bwlimit_index = table_fields.index('bwlimit')
 140                 if isinstance(row[bwlimit_index], int):
 141                         pass
 142                 elif row[bwlimit_index].find('mbit') > -1:
 143                         row[bwlimit_index] = int(row[bwlimit_index].split('mbit')[0]) \
 144                                             * 1000000
 145                 elif row[bwlimit_index].find('kbit') > -1:
 146                         row[bwlimit_index] = int(row[bwlimit_index].split('kbit')[0]) \
 147                                              * 1000
 148         elif table_name in ['slice_attribute']:
 149                 # modify some invalid foreign keys
 150                 attribute_type_index = table_fields.index('attribute_type_id')
 151                 if row[attribute_type_index] == 10004:
 152                         row[attribute_type_index] = 10016
 153                 elif row[attribute_type_index] == 10006:
 154                         row[attribute_type_index] = 10017
 155         elif table_name in ['slice_attribute_types']:
 156                 type_id_index = table_fields.index('attribute_type_id')
 157                 if row[type_id_index] in [10004, 10006]:
 158                         return None
 159         return row
 160
 161 def fix_table(table, table_name, table_fields):
 162         if table_name in ['slice_attribute_types']:
 163                 # remove duplicate/redundant primary keys
 164                 type_id_index = table_fields.index('attribute_type_id')
 165                 for row in table:
 166                         if row[type_id_index] in [10004, 10006]:
 167                                 table.remove(row)
 168         return table
 169
 170 def remove_temp_tables():
 171         # remove temp_tables
 172         try:
 173                 for temp_table in temp_tables:
 174                         os.remove(temp_tables[temp_table])
 175         except:
 176                 raise
 177
 178 def generate_temp_table(table_name, db):
 179         cursor = db.cursor()
 180         try:
 181                 # get upgrade directions
 182                 table_def = upgrade_config[table_name].replace('(', '').replace(')', '').split(',')
 183                 table_fields, old_fields, joins, wheres = [], [], set(), set()
 184                 for field in table_def:
 185                         field_parts = field.strip().split(':')
 186                         table_fields.append(field_parts[0])
 187                         old_fields.append(field_parts[1])
 188                         if field_parts[2:]:
 189                                 joins.update(set(filter(lambda x: not x.find('=') > -1, field_parts[2:])))
 190                                 wheres.update(set(filter(lambda x: x.find('=') > -1, field_parts[2:])))
 191
 192                 # get indices of fields that cannot be null
 193                 (type, body_list) = schema[table_name]
 194                 not_null_indices = []
 195                 for field in table_fields:
 196                         for body_line in body_list:
 197                                 if body_line.find(field) > -1 and \
 198                                    body_line.upper().find("NOT NULL") > -1:
 199                                         not_null_indices.append(table_fields.index(field))
 200
 201                 # get index of primary key
 202                 primary_key_indices = []
 203                 for body_line in body_list:
 204                         if body_line.find("PRIMARY KEY") > -1:
 205                                 primary_key = body_line
 206                                 for field in table_fields:
 207                                         if primary_key.find(field) > -1:
 208                                                 primary_key_indices.append(table_fields.index(field))
 209                                 break
 210
 211                 # get old data
 212                 get_old_data = "SELECT DISTINCT %s FROM %s" % \
 213                       (", ".join(old_fields), old_fields[0].split(".")[0])
 214                 for join in joins:
 215                         get_old_data = get_old_data + " INNER JOIN %s USING (%s) " % \
 216                                        (join.split('.')[0], join.split('.')[1])
 217                 if wheres:
 218                         get_old_data = get_old_data + " WHERE "
 219                 for where in wheres:
 220                         get_old_data = get_old_data + " %s" % where
 221                 cursor.execute(get_old_data)
 222                 rows = cursor.fetchall()
 223
 224                 # write data to a temp file
 225                 temp_file_name = '%s/%s.tmp' % (temp_dir, table_name)
 226                 temp_file = open(temp_file_name, 'w')
 227                 for row in rows:
 228                         # attempt to make any necessary fixes to data
 229                         row = fix_row(row, table_name, table_fields)
 230                         # do not attempt to write null rows
 231                         if row == None:
 232                                 continue
 233                         # do not attempt to write rows with null primary keys
 234                         if filter(lambda x: row[x] == None, primary_key_indices):
 235                                 continue
 236                         for i in range(len(row)):
 237                                 # convert nulls into something pg can understand
 238                                 if row[i] == None:
 239                                         if i in not_null_indices:
 240                                                 # XX doesnt work if column is int type
 241                                                 row[i] = ""
 242                                         else:
 243                                                 row[i] = "\N"
 244                                 if isinstance(row[i], int) or isinstance(row[i], float):
 245                                         row[i] = str(row[i])
 246                                 # escape whatever can mess up the data format
 247                                 if isinstance(row[i], str):
 248                                         row[i] = row[i].replace('\t', '\\t')
 249                                         row[i] = row[i].replace('\n', '\\n')
 250                                         row[i] = row[i].replace('\r', '\\r')
 251                         data_row = "\t".join(row)
 252                         temp_file.write(data_row + "\n")
 253                 temp_file.write("\.\n")
 254                 temp_file.close()
 255                 temp_tables[table_name] = temp_file_name
 256
 257         except KeyError:
 258                 #print "WARNING: cannot upgrade %s. upgrade def not found. skipping" % \
 259                 #       (table_name)
 260                 return False
 261         except IndexError, fault:
 262                 print "Error: error found in upgrade config file. " \
 263                       "check %s configuration. Aborting " % \
 264                       (table_name)
 265                 sys.exit(1)
 266         except:
 267                 print "Error: configuration for %s doesnt match db schema. " \
 268                       " Aborting" % (table_name)
 269                 try:
 270                         db.rollback()
 271                 except:
 272                         pass
 273                 raise
 274
 275
 276 # Connect to current db
 277 db = connect()
 278 cursor = db.cursor()
 279
 280 # determin current db version
 281 try:
 282         cursor.execute("SELECT relname from pg_class where relname = 'plc_db_version'")
 283         rows = cursor.fetchall()
 284         if not rows:
 285                 print "Warning: current db has no version. Unable to validate config file."
 286         else:
 287                 cursor.execute("SELECT version FROM plc_db_version")
 288                 rows = cursor.fetchall()
 289                 if not rows or not rows[0]:
 290                         print "Warning: current db has no version. Unable to validate config file."
 291                 elif rows[0][0] == db_version_new:
 292                         print "Status: Versions are the same. No upgrade necessary."
 293                         sys.exit()
 294                 elif not rows[0][0] == db_version_previous:
 295                         print "Stauts: DB_VERSION_PREVIOUS in config file (%s) does not" \
 296                               " match current db version %d" % (upgrade_config_file, rows[0][0])
 297                         sys.exit()
 298                 else:
 299                         print "STATUS: attempting upgrade from %d to %d" % \
 300                                 (db_version_previous, db_version_new)
 301
 302         # check db encoding
 303         sql = " SELECT pg_catalog.pg_encoding_to_char(d.encoding)" \
 304               " FROM pg_catalog.pg_database d " \
 305               " WHERE d.datname = '%s' " % config['PLC_DB_NAME']
 306         cursor.execute(sql)
 307         rows = cursor.fetchall()
 308         if rows[0][0] not in ['UTF8', 'UNICODE']:
 309                 print "WARNING: db encoding is not utf8. Attempting to encode"
 310                 db.close()
 311                 # generate db dump
 312                 dump_file = '%s/dump.sql' % (temp_dir)
 313                 dump_file_encoded = dump_file + ".utf8"
 314                 dump_cmd = 'pg_dump -i %s -U postgres -f %s > /dev/null 2>&1' % \
 315                            (config['PLC_DB_NAME'], dump_file)
 316                 if os.system(dump_cmd):
 317                         print "ERROR: during db dump. Exiting."
 318                         sys.exit(1)
 319                 # encode dump to utf8
 320                 print "Status: encoding database dump"
 321                 encode_utf8(dump_file, dump_file_encoded)
 322                 # archive original db
 323                 archive_db(config['PLC_DB_NAME'], config['PLC_DB_NAME']+'_sqlascii_archived')
 324                 # create a utf8 database and upload encoded data
 325                 recreate_cmd = 'createdb -U postgres -E UTF8 %s > /dev/null; ' \
 326                                'psql -a -U  %s %s < %s > /dev/null 2>&1;'   % \
 327                           (config['PLC_DB_NAME'], config['PLC_DB_USER'], \
 328                            config['PLC_DB_NAME'], dump_file_encoded)
 329                 print "Status: recreating database as utf8"
 330                 if os.system(recreate_cmd):
 331                         print "Error: database encoding failed. Aborting"
 332                         sys.exit(1)
 333
 334                 os.remove(dump_file_encoded)
 335                 os.remove(dump_file)
 336 except:
 337         raise
 338
 339
 340 db = connect()
 341 cursor = db.cursor()
 342
 343 # parse the schema user wishes to upgrade to
 344 try:
 345         file = open(schema_file, 'r')
 346         index = 0
 347         lines = file.readlines()
 348         while index < len(lines):
 349                 line = lines[index]
 350                 # find all created objects
 351                 if line.startswith("CREATE"):
 352                         line_parts = line.split(" ")
 353                         item_type = line_parts[1]
 354                         item_name = line_parts[2]
 355                         schema_items_ordered.append(item_name)
 356                         if item_type in ['INDEX']:
 357                                 schema[item_name] = (item_type, line)
 358
 359                         # functions, tables, views span over multiple lines
 360                         # handle differently than indexes
 361                         elif item_type in ['AGGREGATE', 'TABLE', 'VIEW']:
 362                                 fields = [line]
 363                                 while index < len(lines):
 364                                         index = index + 1
 365                                         nextline =lines[index]
 366                                         # look for any sequences
 367                                         if item_type in ['TABLE'] and nextline.find('serial') > -1:
 368                                                 sequences[item_name] = nextline.strip().split()[0]
 369                                         fields.append(nextline)
 370                                         if nextline.find(";") >= 0:
 371                                                 break
 372                                 schema[item_name] = (item_type, fields)
 373                         else:
 374                                 print "Error: unknown type %s" % item_type
 375                 elif line.startswith("INSERT"):
 376                         inserts.append(line)
 377                 index = index + 1
 378
 379 except:
 380         raise
 381
 382 print "Status: generating temp tables"
 383 # generate all temp tables
 384 for key in schema_items_ordered:
 385         (type, body_list) = schema[key]
 386         if type == 'TABLE':
 387                 generate_temp_table(key, db)
 388
 389 # disconenct from current database and archive it
 390 cursor.close()
 391 db.close()
 392
 393 print "Status: archiving database"
 394 archive_db(config['PLC_DB_NAME'], config['PLC_DB_NAME']+'_archived')
 395 os.system('createdb -U postgres -E UTF8 %s > /dev/null; ' % config['PLC_DB_NAME'])
 396
 397 print "Status: upgrading database"
 398 # attempt to create and load all items from schema into temp db
 399 try:
 400         for key in schema_items_ordered:
 401                 (type, body_list) = schema[key]
 402                 create_item_from_schema(key)
 403                 if type == 'TABLE':
 404                         if upgrade_config.has_key(key):
 405                                 # attempt to populate with temp table data
 406                                 table_def = upgrade_config[key].replace('(', '').replace(')', '').split(',')
 407                                 table_fields = [field.strip().split(':')[0] for field in table_def]
 408                                 insert_cmd = "psql %s %s -c " \
 409                                              " 'COPY %s (%s) FROM stdin;' < %s " % \
 410                                              (config['PLC_DB_NAME'], config['PLC_DB_USER'], key,
 411                                               ", ".join(table_fields), temp_tables[key] )
 412                                 exit_status = os.system(insert_cmd)
 413                                 if exit_status:
 414                                         print "Error: upgrade %s failed" % key
 415                                         sys.exit(1)
 416                                 # update the primary key sequence
 417                                 if sequences.has_key(key):
 418                                         sequence = key +"_"+ sequences[key] +"_seq"
 419                                         update_seq = "psql %s %s -c " \
 420                                              " \"select setval('%s', max(%s)) FROM %s;\" > /dev/null" % \
 421                                              (config['PLC_DB_NAME'], config['PLC_DB_USER'], sequence,
 422                                               sequences[key], key)
 423                                         exit_status = os.system(update_seq)
 424                                         if exit_status:
 425                                                 print "Error: sequence %s update failed" % sequence
 426                                                 sys.exit(1)
 427                         else:
 428                                 # check if there are any insert stmts in schema for this table
 429                                 print "Warning: %s has no temp data file. Unable to populate with old data" % key
 430                                 for insert_stmt in inserts:
 431                                         if insert_stmt.find(key) > -1:
 432                                                 insert_cmd = 'psql %s postgres -qc "%s;" > /dev/null 2>&1' % \
 433                                                 (config['PLC_DB_NAME'], insert_stmt)
 434                                                 os.system(insert_cmd)
 435 except:
 436         print "Error: failed to populate db. Unarchiving original database and aborting"
 437         undo_command = "dropdb -U postgres %s > /dev/null; psql template1 postgres -qc" \
 438                        " 'ALTER DATABASE %s RENAME TO %s;';  > /dev/null" % \
 439                        (config['PLC_DB_NAME'], config['PLC_DB_NAME']+'_archived', config['PLC_DB_NAME'])
 440         os.system(undo_command)
 441         remove_temp_tables()
 442         raise
 443
 444 remove_temp_tables()
 445
 446 print "upgrade complete"