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