nsz Git - libfirm/blob - scripts/statev_sql.py

   1 #! /usr/bin/env python
   2
   3 import sys
   4 import os
   5 import re
   6 import time
   7 import stat
   8 import fileinput
   9 import tempfile
  10 import optparse
  11
  12 class DummyFilter:
  13         def match(self, dummy):
  14                 return True
  15
  16 class EmitBase:
  17         def create_table(self, cols, name, type, unique):
  18                 create = 'create table if not exists %s (id int %s' % (name, unique)
  19
  20                 sorted = [None] * len(cols)
  21                 for x in cols.iterkeys():
  22                         sorted[cols[x]] = x
  23                 for x in sorted:
  24                         create += (", '%s' %s" % (x, type))
  25                 create += ');'
  26                 return create
  27
  28 class EmitMysqlInfile(EmitBase):
  29         tmpfile_mode = stat.S_IREAD | stat.S_IROTH | stat.S_IWUSR
  30
  31         def ex(self, args, tab, fname):
  32                 res = os.fork()
  33                 if res == 0:
  34                         stmt = """load data infile '%s' into table %s fields terminated by ';'""" % (fname, tab)
  35                         conn = MySQLdb.connect(**args)
  36                         c = conn.cursor()
  37                         c.execute(stmt)
  38                         conn.commit()
  39                         sys.exit(0)
  40                 return res
  41
  42         def __init__(self, options, tables, ctxcols, evcols):
  43                 import MySQLdb
  44
  45                 args = dict()
  46                 if options.password:
  47                         args['passwd'] = options.password
  48                 if not options.host:
  49                         options.host = 'localhost'
  50                 args['user'] = options.user
  51                 args['host'] = options.host
  52                 args['db']   = options.database
  53
  54                 self.conn     = MySQLdb.connect(**args)
  55                 self.ctxcols  = ctxcols
  56                 self.evcols   = evcols
  57                 self.options  = options
  58                 self.ctxtab   = tables['ctx']
  59                 self.evtab    = tables['ev']
  60
  61                 params = (tempfile.gettempdir(), os.sep, os.getpid())
  62                 self.evfifo  = '%s%sstatev_ev_%d' % params
  63                 self.ctxfifo = '%s%sstatev_ctx_%d' % params
  64
  65                 os.mkfifo(self.evfifo)
  66                 os.mkfifo(self.ctxfifo)
  67
  68                 os.chmod(self.evfifo,  self.tmpfile_mode)
  69                 os.chmod(self.ctxfifo, self.tmpfile_mode)
  70
  71                 c = self.conn.cursor()
  72                 c.execute('drop table if exists ' + self.evtab)
  73                 c.execute('drop table if exists ' + self.ctxtab)
  74                 table_ctx = self.create_table(self.ctxcols, self.ctxtab, 'char(80)', 'unique')
  75                 c.execute(table_ctx)
  76                 table_ev = self.create_table(self.evcols, self.evtab, 'double default null', '')
  77                 c.execute(table_ev)
  78                 self.conn.commit()
  79
  80                 if options.verbose:
  81                         print 'go for gold'
  82
  83                 self.pidev  = self.ex(args, self.evtab, self.evfifo)
  84                 self.pidctx = self.ex(args, self.ctxtab, self.ctxfifo)
  85
  86                 if options.verbose:
  87                         print "forked two mysql leechers: %d, %d" % (self.pidev, self.pidctx)
  88
  89                 self.evfile   = open(self.evfifo, 'w+t')
  90                 self.ctxfile  = open(self.ctxfifo, 'w+t')
  91
  92                 if options.verbose:
  93                         print 'fifo:  %s, %o' % (self.evfile.name, os.stat(self.evfile.name).st_mode)
  94                         print 'fifo:  %s, %o' % (self.ctxfile.name, os.stat(self.ctxfile.name).st_mode)
  95
  96         def ev(self, curr_id, evitems):
  97                 field = ['\N'] * len(self.evcols)
  98                 for key, val in evitems.iteritems():
  99                         index = self.evcols[key]
 100                         field[index] = val
 101                 print >> self.evfile, ('%d;' % curr_id) + ';'.join(field)
 102
 103         def ctx(self, curr_id, ctxitems):
 104                 field = ['\N'] * len(self.ctxcols)
 105                 for key, val in ctxitems.iteritems():
 106                         index = self.ctxcols[key]
 107                         field[index] = val
 108                 print >> self.ctxfile, ('%d;' % curr_id) + ';'.join(field)
 109
 110         def commit(self):
 111                 self.evfile.close()
 112                 self.ctxfile.close()
 113
 114                 os.waitpid(self.pidev, 0)
 115                 os.waitpid(self.pidctx, 0)
 116
 117                 os.unlink(self.evfile.name)
 118                 os.unlink(self.ctxfile.name)
 119
 120
 121 class EmitSqlite3(EmitBase):
 122         def __init__(self, options, tables, ctxcols, evcols):
 123                 import sqlite3
 124
 125                 if options.database == None:
 126                         print "Have to specify database (file-)name for sqlite"
 127                         sys.exit(1)
 128
 129                 if os.path.isfile(options.database):
 130                         os.unlink(options.database)
 131
 132                 self.ctxtab = tables['ctx']
 133                 self.evtab  = tables['ev']
 134                 self.conn = sqlite3.connect(options.database)
 135                 table_ctx = self.create_table(ctxcols, self.ctxtab, 'text', 'unique')
 136                 self.conn.execute(table_ctx)
 137                 table_ev = self.create_table(evcols, self.evtab, 'double', '')
 138                 self.conn.execute(table_ev)
 139
 140                 n = max(len(ctxcols), len(evcols)) + 1
 141                 q = ['?']
 142                 self.quests = []
 143                 for i in xrange(0, n):
 144                         self.quests.append(','.join(q))
 145                         q.append('?')
 146
 147         def ev(self, curr_id, evitems):
 148                 keys = ""
 149                 first = True
 150                 for key in evitems.keys():
 151                         if first:
 152                                 first = False
 153                         else:
 154                                 keys += ", "
 155                         keys += "'%s'" % (key)
 156
 157                 stmt = "insert into '%s' (id, %s) values (%s)" % (self.evtab, keys, self.quests[len(evitems)])
 158                 self.conn.execute(stmt, (curr_id,) + tuple(evitems.values()))
 159
 160         def ctx(self, curr_id, ctxitems):
 161                 keys = ""
 162                 first = True
 163                 for key in ctxitems.keys():
 164                         if first:
 165                                 first = False
 166                         else:
 167                                 keys += ", "
 168                         keys += "'%s'" % (key)
 169
 170                 stmt = "insert into '%s' (id, %s) values (%s)" % (self.ctxtab, keys, self.quests[len(ctxitems)])
 171                 self.conn.execute(stmt, (curr_id,) + tuple(ctxitems.values()))
 172
 173         def commit(self):
 174                 self.conn.commit()
 175
 176 class Conv:
 177         engines = { 'sqlite3': EmitSqlite3, 'mysql': EmitMysqlInfile }
 178         def find_heads(self):
 179                 n_ev    = 0
 180                 ctxind  = 0
 181                 evind   = 0
 182                 ctxcols = dict()
 183                 evcols  = dict()
 184
 185                 self.valid_keys = set()
 186
 187                 inp = self.input()
 188
 189                 for line in inp:
 190                         if line[0] == 'P':
 191                                 ind = line.index(';', 2)
 192                                 key = line[2:ind]
 193                                 if not ctxcols.has_key(key):
 194                                         ctxcols[key] = ctxind
 195                                         ctxind += 1
 196
 197                         elif line[0] == 'E':
 198                                 ind = line.index(';', 2)
 199                                 key = line[2:ind]
 200                                 if self.filter.match(key):
 201                                         self.n_events += 1
 202                                         if not evcols.has_key(key):
 203                                                 self.valid_keys.add(key)
 204                                                 evcols[key] = evind
 205                                                 evind += 1
 206
 207                 return (ctxcols, evcols)
 208
 209         def input(self):
 210                 return fileinput.FileInput(files=self.files, openhook=fileinput.hook_compressed)
 211
 212         def fill_tables(self):
 213                 lineno     = 0
 214                 ids        = 0
 215                 curr_id    = 0
 216                 last_push_curr_id = 0
 217                 keystack   = []
 218                 idstack    = []
 219                 curr_event = 0
 220                 last_prec  = -1
 221                 evcols     = dict()
 222                 ctxcols    = dict()
 223
 224                 for line in self.input():
 225                         lineno += 1
 226                         items = line.strip().split(';')
 227                         op    = items[0]
 228
 229                         if op == 'P':
 230                                 # flush the current events
 231                                 if len(evcols):
 232                                         self.emit.ev(last_push_curr_id, evcols)
 233                                         evcols.clear()
 234
 235                                 # push the key
 236                                 key   = items[1]
 237                                 val   = items[2]
 238                                 keystack.append(key)
 239                                 curr_id = ids
 240                                 last_push_curr_id = curr_id
 241                                 ids += 1
 242                                 idstack.append(curr_id)
 243                                 ctxcols[key] = val
 244
 245                                 self.emit.ctx(curr_id, ctxcols)
 246
 247                         elif op == 'O':
 248                                 popkey = items[1]
 249                                 key = keystack.pop()
 250
 251                                 if popkey != key:
 252                                         print "unmatched pop in line %d, push key %s, pop key: %s" % (lineno, key, popkey)
 253
 254                                 idstack.pop()
 255                                 if len(idstack) > 0:
 256                                         if len(evcols) > 0:
 257                                                 self.emit.ev(curr_id, evcols)
 258                                                 evcols.clear()
 259                                         del ctxcols[key]
 260                                         curr_id = idstack[-1]
 261                                 else:
 262                                         curr_id = -1
 263
 264                         elif op == 'E':
 265                                 key = items[1]
 266                                 if key in self.valid_keys:
 267                                         curr_event += 1
 268                                         evcols[key] = items[2]
 269
 270                                         if self.verbose:
 271                                                 prec = curr_event * 10 / self.n_events
 272                                                 if prec > last_prec:
 273                                                         last_prec = prec
 274                                                         print '%10d / %10d' % (curr_event, self.n_events)
 275
 276         def __init__(self):
 277                 parser = optparse.OptionParser('usage: %prog [options]  <event file...>')
 278                 parser.add_option("-c", "--clean",    dest="clean",    help="delete tables in advance", action="store_true", default=False)
 279                 parser.add_option("-v", "--verbose",  dest="verbose",  help="verbose messages",         action="store_true", default=False)
 280                 parser.add_option("-f", "--filter",   dest="filter",   help="regexp to filter event keys", metavar="REGEXP")
 281                 parser.add_option("-u", "--user",     dest="user",     help="user",               metavar="USER")
 282                 parser.add_option("-H", "--host",     dest="host",     help="host",               metavar="HOST")
 283                 parser.add_option("-p", "--password", dest="password", help="password",           metavar="PASSWORD")
 284                 parser.add_option("-d", "--db",       dest="database", help="database",           metavar="DB")
 285                 parser.add_option("-e", "--engine",   dest="engine",   help="engine",             metavar="ENG", default='sqlite3')
 286                 parser.add_option("-P", "--prefix",   dest="prefix",   help="table prefix",       metavar="PREFIX", default='')
 287                 (options, args) = parser.parse_args()
 288
 289                 self.n_events = 0
 290                 self.stmts    = dict()
 291                 self.verbose  = options.verbose
 292
 293                 tables = dict()
 294                 tables['ctx'] = options.prefix + 'ctx'
 295                 tables['ev']  = options.prefix + 'ev'
 296
 297                 if len(args) < 1:
 298                         parser.print_help()
 299                         sys.exit(1)
 300
 301                 self.files  = []
 302                 files       = args
 303
 304                 for file in files:
 305                         if not os.path.isfile(file):
 306                                 print "cannot find input file %s" % (file, )
 307                         else:
 308                                 self.files.append(file)
 309
 310                 if len(self.files) < 1:
 311                         print "no input file to process"
 312                         sys.exit(3)
 313
 314                 if options.filter:
 315                         self.filter = re.compile(options.filter)
 316                 else:
 317                         self.filter = DummyFilter()
 318
 319                 if options.engine in self.engines:
 320                         engine = self.engines[options.engine]
 321                 else:
 322                         print 'engine %s not found' % options.engine
 323                         print 'we offer: %s' % self.engines.keys()
 324                         sys.exit(0)
 325
 326                 if options.verbose:
 327                         print "determining schema..."
 328
 329                 (ctxcols, evcols) = self.find_heads()
 330                 if options.verbose:
 331                         print "context schema:"
 332                         print ctxcols
 333                         print "event schema:"
 334                         print evcols
 335                         print "tables:"
 336                         print tables
 337
 338                 self.emit = engine(options, tables, ctxcols, evcols)
 339
 340                 if options.verbose:
 341                         print "filling tables..."
 342                 self.fill_tables()
 343                 if options.verbose:
 344                         print "comitting..."
 345                 self.emit.commit()
 346
 347 if __name__ == "__main__":
 348         Conv()