2010-08-18 14:30:09 -07:00
# ***** BEGIN LICENSE BLOCK *****
# Version: MPL 1.1/GPL 2.0/LGPL 2.1
#
# The contents of this file are subject to the Mozilla Public License Version
# 1.1 (the "License"); you may not use this file except in compliance with
# the License. You may obtain a copy of the License at
# http://www.mozilla.org/MPL/
#
# Software distributed under the License is distributed on an "AS IS" basis,
# WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
# for the specific language governing rights and limitations under the
# License.
#
# The Original Code is mozilla.org code
#
# The Initial Developer of the Original Code is
# Mozilla Foundation.
# Portions created by the Initial Developer are Copyright (C) 2010
# the Initial Developer. All Rights Reserved.
#
# Contributor(s):
# Taras Glek <tglek@mozilla.com>
#
# Alternatively, the contents of this file may be used under the terms of
# either the GNU General Public License Version 2 or later (the "GPL"), or
# the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
# in which case the provisions of the GPL or the LGPL are applicable instead
# of those above. If you wish to allow use of your version of this file only
# under the terms of either the GPL or the LGPL, and not to allow others to
# use your version of this file under the terms of the MPL, indicate your
# decision by deleting the provisions above and replace them with the notice
# and other provisions required by the GPL or the LGPL. If you do not delete
# the provisions above, a recipient may use your version of this file under
# the terms of any one of the MPL, the GPL or the LGPL.
#
# ***** END LICENSE BLOCK *****
import sys , os , subprocess , struct
local_file_header = [
( " signature " , " uint32 " ) ,
( " min_version " , " uint16 " ) ,
( " general_flag " , " uint16 " ) ,
( " compression " , " uint16 " ) ,
( " lastmod_time " , " uint16 " ) ,
( " lastmod_date " , " uint16 " ) ,
( " crc32 " , " uint32 " ) ,
( " compressed_size " , " uint32 " ) ,
( " uncompressed_size " , " uint32 " ) ,
( " filename_size " , " uint16 " ) ,
( " extra_field_size " , " uint16 " ) ,
( " filename " , " filename_size " ) ,
( " extra_field " , " extra_field_size " ) ,
( " data " , " compressed_size " )
]
cdir_entry = [
( " signature " , " uint32 " ) ,
( " creator_version " , " uint16 " ) ,
( " min_version " , " uint16 " ) ,
( " general_flag " , " uint16 " ) ,
( " compression " , " uint16 " ) ,
( " lastmod_time " , " uint16 " ) ,
( " lastmod_date " , " uint16 " ) ,
( " crc32 " , " uint32 " ) ,
( " compressed_size " , " uint32 " ) ,
( " uncompressed_size " , " uint32 " ) ,
( " filename_size " , " uint16 " ) ,
( " extrafield_size " , " uint16 " ) ,
( " filecomment_size " , " uint16 " ) ,
( " disknum " , " uint16 " ) ,
( " internal_attr " , " uint16 " ) ,
( " external_attr " , " uint32 " ) ,
( " offset " , " uint32 " ) ,
( " filename " , " filename_size " ) ,
( " extrafield " , " extrafield_size " ) ,
( " filecomment " , " filecomment_size " ) ,
]
cdir_end = [
( " signature " , " uint32 " ) ,
( " disk_num " , " uint16 " ) ,
( " cdir_disk " , " uint16 " ) ,
( " disk_entries " , " uint16 " ) ,
( " cdir_entries " , " uint16 " ) ,
( " cdir_size " , " uint32 " ) ,
( " cdir_offset " , " uint32 " ) ,
( " comment_size " , " uint16 " ) ,
]
type_mapping = { " uint32 " : " I " , " uint16 " : " H " }
def format_struct ( format ) :
string_fields = { }
fmt = " < "
for ( name , value ) in iter ( format ) :
try :
fmt + = type_mapping [ value ] [ 0 ]
except KeyError :
string_fields [ name ] = value
return ( fmt , string_fields )
def size_of ( format ) :
return struct . calcsize ( format_struct ( format ) [ 0 ] )
class MyStruct :
def __init__ ( self , format , string_fields ) :
self . __dict__ [ " struct_members " ] = { }
self . __dict__ [ " format " ] = format
self . __dict__ [ " string_fields " ] = string_fields
def addMember ( self , name , value ) :
self . __dict__ [ " struct_members " ] [ name ] = value
def __getattr__ ( self , item ) :
try :
return self . __dict__ [ " struct_members " ] [ item ]
except :
pass
print ( " no %s " % item )
print ( self . __dict__ [ " struct_members " ] )
raise AttributeError
def __setattr__ ( self , item , value ) :
if item in self . __dict__ [ " struct_members " ] :
self . __dict__ [ " struct_members " ] [ item ] = value
else :
raise AttributeError
def pack ( self ) :
extra_data = " "
values = [ ]
string_fields = self . __dict__ [ " string_fields " ]
struct_members = self . __dict__ [ " struct_members " ]
format = self . __dict__ [ " format " ]
for ( name , _ ) in format :
if name in string_fields :
extra_data = extra_data + struct_members [ name ]
else :
values . append ( struct_members [ name ] ) ;
return struct . pack ( format_struct ( format ) [ 0 ] , * values ) + extra_data
ENDSIG = 0x06054b50
def assert_true ( cond , msg ) :
if not cond :
raise Exception ( msg )
exit ( 1 )
class BinaryBlob :
def __init__ ( self , f ) :
self . data = open ( f , " rb " ) . read ( )
self . offset = 0
self . length = len ( self . data )
def readAt ( self , pos , length ) :
self . offset = pos + length
return self . data [ pos : self . offset ]
def read_struct ( self , format , offset = None ) :
if offset == None :
offset = self . offset
( fstr , string_fields ) = format_struct ( format )
size = struct . calcsize ( fstr )
data = self . readAt ( offset , size )
ret = struct . unpack ( fstr , data )
retstruct = MyStruct ( format , string_fields )
i = 0
for ( name , _ ) in iter ( format ) :
member_desc = None
if not name in string_fields :
member_data = ret [ i ]
i = i + 1
else :
# zip has data fields which are described by other struct fields, this does
# additional reads to fill em in
member_desc = string_fields [ name ]
member_data = self . readAt ( self . offset , retstruct . __getattr__ ( member_desc ) )
retstruct . addMember ( name , member_data )
# sanity check serialization code
data = self . readAt ( offset , self . offset - offset )
out_data = retstruct . pack ( )
assert_true ( out_data == data , " Serialization fail %d != %d " % ( len ( out_data ) , len ( data ) ) )
return retstruct
2010-09-10 09:30:07 -07:00
def optimizejar ( jar , outjar , inlog = None ) :
if inlog is not None :
inlog = open ( inlog ) . read ( ) . rstrip ( )
# in the case of an empty log still move the index forward
if len ( inlog ) == 0 :
inlog = [ ]
else :
inlog = inlog . split ( " \n " )
outlog = [ ]
2010-08-18 14:30:09 -07:00
jarblob = BinaryBlob ( jar )
dirend = jarblob . read_struct ( cdir_end , jarblob . length - size_of ( cdir_end ) )
assert_true ( dirend . signature == ENDSIG , " no signature in the end " ) ;
cdir_offset = dirend . cdir_offset
2010-09-10 09:30:07 -07:00
readahead = 0
if inlog is None and cdir_offset == 4 :
readahead = struct . unpack ( " <I " , jarblob . readAt ( 0 , 4 ) ) [ 0 ]
print ( " %s : startup data ends at byte %d " % ( outjar , readahead ) ) ;
2010-08-18 14:30:09 -07:00
jarblob . offset = cdir_offset
central_directory = [ ]
for i in range ( 0 , dirend . cdir_entries ) :
entry = jarblob . read_struct ( cdir_entry )
central_directory . append ( entry )
reordered_count = 0
2010-09-10 09:30:07 -07:00
if inlog is not None :
dup_guard = set ( )
for ordered_name in inlog :
if ordered_name in dup_guard :
continue
else :
dup_guard . add ( ordered_name )
found = False
for i in range ( reordered_count , len ( central_directory ) ) :
if central_directory [ i ] . filename == ordered_name :
# swap the cdir entries
tmp = central_directory [ i ]
central_directory [ i ] = central_directory [ reordered_count ]
central_directory [ reordered_count ] = tmp
reordered_count = reordered_count + 1
found = True
break
if not found :
print ( " Can ' t find ' %s ' in %s " % ( ordered_name , jar ) )
2010-08-18 14:30:09 -07:00
outfd = open ( outjar , " wb " )
2010-09-10 09:30:07 -07:00
out_offset = 0
if inlog is not None :
# have to put central directory at offset 4 cos 0 confuses some tools.
# This also lets us specify how many entries should be preread
dirend . cdir_offset = 4
# make room for central dir + end of dir + 4 extra bytes at front
out_offset = dirend . cdir_offset + dirend . cdir_size + size_of ( cdir_end )
outfd . seek ( out_offset )
2010-08-18 14:30:09 -07:00
cdir_data = " "
written_count = 0
2010-09-10 09:30:07 -07:00
# store number of bytes suggested for readahead
2010-08-18 14:30:09 -07:00
for entry in central_directory :
# read in the header twice..first for comparison, second time for convenience when writing out
jarfile = jarblob . read_struct ( local_file_header , entry . offset )
assert_true ( jarfile . filename == entry . filename , " Directory/Localheader mismatch " )
data = jarfile . pack ( )
outfd . write ( data )
2010-09-10 09:30:07 -07:00
old_entry_offset = entry . offset
2010-08-18 14:30:09 -07:00
entry . offset = out_offset
out_offset = out_offset + len ( data )
entry_data = entry . pack ( )
cdir_data + = entry_data
expected_len = entry . filename_size + entry . extrafield_size + entry . filecomment_size
assert_true ( len ( entry_data ) != expected_len ,
" %s entry size - expected: %d got: %d " % ( entry . filename , len ( entry_data ) , expected_len ) )
written_count + = 1
2010-09-10 09:30:07 -07:00
if inlog is not None :
if written_count == reordered_count :
readahead = out_offset
print ( " %s : startup data ends at byte %d " % ( outjar , readahead ) ) ;
elif written_count < reordered_count :
pass
#print("%s @ %d" % (entry.filename, out_offset))
elif readahead > = old_entry_offset + len ( data ) :
outlog . append ( entry . filename )
reordered_count + = 1
if inlog is None :
dirend . cdir_offset = out_offset
2010-08-18 14:30:09 -07:00
2010-09-10 09:30:07 -07:00
dirend_data = dirend . pack ( )
2010-08-18 14:30:09 -07:00
assert_true ( size_of ( cdir_end ) == len ( dirend_data ) , " Failed to serialize directory end correctly. Serialized size; %d , expected: %d " % ( len ( dirend_data ) , size_of ( cdir_end ) ) ) ;
2010-09-10 09:30:07 -07:00
2010-08-18 14:30:09 -07:00
outfd . seek ( dirend . cdir_offset )
assert_true ( len ( cdir_data ) == dirend . cdir_size , " Failed to serialize central directory correctly. Serialized size; %d , expected: %d expected-size: %d " % ( len ( cdir_data ) , dirend . cdir_size , dirend . cdir_size - len ( cdir_data ) ) ) ;
outfd . write ( cdir_data )
outfd . write ( dirend_data )
2010-09-10 09:30:07 -07:00
# for ordered jars the central directory is written in the begining of the file, so a second central-directory
# entry has to be written in the end of the file
if inlog is not None :
outfd . seek ( 0 )
outfd . write ( struct . pack ( " <I " , readahead ) ) ;
outfd . seek ( out_offset )
outfd . write ( dirend_data )
print " %s %d / %d in %s " % ( ( " Ordered " if inlog is not None else " Deoptimized " ) ,
reordered_count , len ( central_directory ) , outjar )
2010-08-18 14:30:09 -07:00
outfd . close ( )
2010-09-10 09:30:07 -07:00
return outlog
if len ( sys . argv ) != 5 :
print " Usage: --optimize|--deoptimize %s JAR_LOG_DIR IN_JAR_DIR OUT_JAR_DIR " % sys . argv [ 0 ]
2010-08-18 14:30:09 -07:00
exit ( 1 )
2010-09-10 09:30:07 -07:00
def optimize ( JAR_LOG_DIR , IN_JAR_DIR , OUT_JAR_DIR ) :
if not os . path . exists ( JAR_LOG_DIR ) :
print ( " No jar logs found in %s . No jars to optimize. " % JAR_LOG_DIR )
exit ( 0 )
2010-08-18 14:30:09 -07:00
2010-09-10 09:30:07 -07:00
ls = os . listdir ( JAR_LOG_DIR )
for logfile in ls :
if not logfile . endswith ( " .jar.log " ) :
continue
injarfile = os . path . join ( IN_JAR_DIR , logfile [ : - 4 ] )
outjarfile = os . path . join ( OUT_JAR_DIR , logfile [ : - 4 ] )
if not os . path . exists ( injarfile ) :
print " Warning: Skipping %s , %s doesn ' t exist " % ( logfile , injarfile )
continue
logfile = os . path . join ( JAR_LOG_DIR , logfile )
optimizejar ( injarfile , outjarfile , logfile )
def deoptimize ( JAR_LOG_DIR , IN_JAR_DIR , OUT_JAR_DIR ) :
if not os . path . exists ( JAR_LOG_DIR ) :
os . makedirs ( JAR_LOG_DIR )
ls = os . listdir ( IN_JAR_DIR )
for jarfile in ls :
if not jarfile . endswith ( " .jar " ) :
continue
injarfile = os . path . join ( IN_JAR_DIR , jarfile )
outjarfile = os . path . join ( OUT_JAR_DIR , jarfile )
logfile = os . path . join ( JAR_LOG_DIR , jarfile + " .log " )
log = optimizejar ( injarfile , outjarfile , None )
open ( logfile , " wb " ) . write ( " \n " . join ( log ) )
def main ( ) :
MODE = sys . argv [ 1 ]
JAR_LOG_DIR = sys . argv [ 2 ]
IN_JAR_DIR = sys . argv [ 3 ]
OUT_JAR_DIR = sys . argv [ 4 ]
if MODE == " --optimize " :
optimize ( JAR_LOG_DIR , IN_JAR_DIR , OUT_JAR_DIR )
elif MODE == " --deoptimize " :
deoptimize ( JAR_LOG_DIR , IN_JAR_DIR , OUT_JAR_DIR )
else :
print ( " Unknown mode %s " % MODE )
exit ( 1 )
2010-08-18 14:30:09 -07:00
2010-09-10 09:30:07 -07:00
if __name__ == ' __main__ ' :
main ( )