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