def _open_file(self, countLines=True):
filename = os.path.join(self._distro.apple_data_dir, self._filename)
zipped = filename + ".gz"
if not os.path.exists(filename) and not os.path.exists(zipped):
if Globals.options.mount:
raise Fail("ERROR: mount failed!")
else:
while (not os.path.exists(filename)) and (not os.path.exists(zipped)):
utils.log("waiting for mount to complete for file '%s'" % filename)
time.sleep(4)
if os.path.exists(zipped):
filename = zipped
#utils.log("Opening Apple EPF file '%s'" % filename)
if not os.path.exists(filename):
utils.log("Apple EPF file '%s' does not exist!" % filename)
raise Fail("Apple EPF file '%s' does not exist!" % filename)
if filename.endswith(".gz"):
f = gzip.open(filename, 'rb')
else:
f = open(filename, 'r+b')
if countLines:
numLines = max(0, utils.getNumLines(f) - 8)
else:
numLines = 0
return f, numLines, filename
def parseCommandLine():
usage = "Usage: %prog [options] [sources]"
version = "%prog " + __version__
parser = OptionParser(usage=usage, version=version)
parser.add_option("-d", "--db", default=None, type="string",
action="store", dest="db",
help="db to connect to for output")
parser.add_option("-n", "--noop", default=False, action="store_true",
help="run in noop mode without modifying anything")
parser.add_option("-r", "--ratio", default=None, type="string",
action="store", dest="ratio",
help="where this crawler fits in to a distributed stack")
parser.add_option("-o", "--offset", default=0,
type="int", dest="offset",
help="start index of entities to import")
parser.add_option("-l", "--limit", default=None, type="int",
help="limits the number of entities to import")
(options, args) = parser.parse_args()
Globals.options = options
if options.db:
utils.init_db_config(options.db)
infile = file('autocomplete.txt', 'r')
options.count = utils.getNumLines(infile)
infile.close()
if options.ratio:
num, den = options.ratio.split('/')
num, den = int(num), int(den)
num, den = float(num), float(den)
options.offset = int(math.floor((options.count * (num - 1)) / den))
options.limit = int(math.ceil(options.count / den) + 1)
utils.log("ratio %s) offset=%d, limit=%d" % (options.ratio, options.offset, options.limit))
else:
if options.limit is None:
options.limit = options.count
options.verbose = False
return options
def _run(self):
utils.log("[%s] initializing" % self)
f, numLines, filename = self._open_file(countLines=False)
table_format = epf.parse_table_format(f, filename)
self.table_format = table_format
stale = False
self._buffer = []
self._buffer_threshold = 1024
# determine whether or not the db table already exists and attempt to
# determine if it's up-to-date s.t. we won't recalculate it if it'd
# be unnecessary.
try:
row0 = self.execute('SELECT * FROM %s LIMIT 1' % (self.table, ), error_okay=True).fetchone()
if row0 is None:
stale = True
elif len(row0) != len(dict(table_format.cols)):
stale = True
except Exception:
self.conn.rollback()
#utils.printException()
stale = True
pass
#f.close(); self._output.put(StopIteration); return
if not stale:
# table is usable as-is
utils.log("[%s] %s.%s doesn't need to be recomputed" % (self, self.dbpath, self.table))
else:
utils.log("[%s] opening '%s'" % (self, self._filename))
numLines = max(0, utils.getNumLines(f) - 8)
table_format = epf.parse_table_format(f, filename)
self.table_format = table_format
utils.log("[%s] parsing ~%d rows from '%s'" % (self, numLines, self._filename))
# initialize table
cols = []
# currently disabling primary keys for most tables
found_primary = False #(len(table_format.primary_keys) != 1)
for col in table_format.cols:
cols.append('')
for col in table_format.cols:
primary = ""
if not found_primary and col == self.primary and not self._sqlite:
#if not found_primary and col in table_format.primary_keys:
# TODO: handle the common case of multiple primary keys, which sqlite3 does not support
# TODO: defining the primary key here as opposed to after insertion is much slower!
primary = " PRIMARY KEY"
found_primary = True
col2 = table_format.cols[col]
col_type = col2['type']
if not self._sqlite:
# perform mapping between some MySQL types that Apple uses and
# their postgres equivalents
if col_type == 'DATETIME':
col_type = 'VARCHAR(100)'
elif col_type == 'LONGTEXT':
col_type = 'VARCHAR(4000)'
text = "%s %s%s" % (col, col_type, primary)
index = col2['index']
cols[index] = text
args = string.joinfields(cols, ', ')
self.execute("DROP TABLE %s" % (self.table, ), error_okay=True)
self.execute("CREATE TABLE %s (%s)" % (self.table, args), verbose=True)
if self._sqlite:
placeholder = '?'
else:
placeholder = '%s'
values_str = '(%s)' % string.joinfields((placeholder for col in table_format.cols), ', ')
self._cmd = 'INSERT INTO %s VALUES %s' % (self.table, values_str)
count = 0
for row in epf.parse_rows(f, table_format):
self._parseRowOld(row, table_format)
count += 1
if numLines > 100 and (count % (numLines / 100)) == 0:
num_rows = self.execute('SELECT COUNT(*) FROM %s' % (self.table, )).fetchone()[0]
utils.log("[%s] done parsing %s -- %d rows" % \
(self, utils.getStatusStr(count, numLines), num_rows))
self._try_flush_buffer(force=True)
if self.index:
self.execute("CREATE INDEX %s on %s (%s)" % (self.index, self.table, self.index), verbose=True)
utils.log("[%s] finished parsing %d rows" % (self, count))
f.close()
self._output.put(StopIteration)
