Files
pyinstaller/archive.py
T
giovannibajo 52a51e0bb1 Python 1.5.2 compatibility: use string module
git-svn-id: http://svn.pyinstaller.org/trunk@6 8dd32b29-ccff-0310-8a9a-9233e24343b1
2005-09-03 17:30:04 +00:00

392 lines
14 KiB
Python

#
# Gordon McMillan (as inspired and influenced by Greg Stein)
#
# subclasses may not need marshal or struct, but since they're
# builtin, importing is safe.
#
# While an Archive is really an abstraction for any "filesystem
# within a file", it is tuned for use with imputil.FuncImporter.
# This assumes it contains python code objects, indexed by the
# the internal name (ie, no '.py').
# See carchive.py for a more general archive (contains anything)
# that can be understood by a C program.
_verbose = 0
_listdir = None
_environ = None
import marshal
import struct
import imp
import sys
import string
_c_suffixes = filter(lambda x: x[2] == imp.C_EXTENSION, imp.get_suffixes())
for nm in ('nt', 'posix', 'dos', 'os2', 'mac'):
if nm in sys.builtin_module_names:
mod = __import__(nm)
_listdir = mod.listdir
_environ = mod.environ
break
if hasattr(sys, 'version_info'):
versuffix = '%d%d'%(sys.version_info[0],sys.version_info[1])
else:
vers = sys.version
dot1 = dot2 = 0
for i in range(len(vers)):
if vers[i] == '.':
if dot1:
dot2 = i
break
else:
dot1 = i
else:
dot2 = len(vers)
versuffix = '%s%s' % (vers[:dot1], vers[dot1+1:dot2])
if "-vi" in sys.argv[1:]:
_verbose = 1
class Archive:
""" A base class for a repository of python code objects.
The extract method is used by imputil.ArchiveImporter
to get code objects by name (fully qualified name), so
an enduser "import a.b" would become
extract('a.__init__')
extract('a.b')
"""
MAGIC = 'PYL\0'
HDRLEN = 12 # default is MAGIC followed by python's magic, int pos of toc
TOCPOS = 8
TRLLEN = 0 # default - no trailer
TOCTMPLT = {} #
os = None
_bincache = None
def __init__(self, path=None, start=0):
"Initialize an Archive. If path is omitted, it will be an empty Archive."
self.toc = None
self.path = path
self.start = start
import imp
self.pymagic = imp.get_magic()
if path is not None:
self.lib = open(self.path, 'rb')
self.checkmagic()
self.loadtoc()
####### Sub-methods of __init__ - override as needed #############
def checkmagic(self):
""" Overridable.
Check to see if the file object self.lib actually has a file
we understand.
"""
self.lib.seek(self.start) #default - magic is at start of file
if self.lib.read(len(self.MAGIC)) != self.MAGIC:
raise RuntimeError, "%s is not a valid %s archive file" \
% (self.path, self.__class__.__name__)
if self.lib.read(len(self.pymagic)) != self.pymagic:
raise RuntimeError, "%s has version mismatch to dll" % (self.path)
self.lib.read(4)
def loadtoc(self):
""" Overridable.
Default: After magic comes an int (4 byte native) giving the
position of the TOC within self.lib.
Default: The TOC is a marshal-able string.
"""
self.lib.seek(self.start + self.TOCPOS)
(offset,) = struct.unpack('=i', self.lib.read(4))
self.lib.seek(self.start + offset)
self.toc = marshal.load(self.lib)
######## This is what is called by FuncImporter #######
## Since an Archive is flat, we ignore parent and modname.
#XXX obsolete - imputil only code
## def get_code(self, parent, modname, fqname):
#### if _verbose:
#### print "I: get_code(%s, %s, %s, %s)" % (self, parent, modname, fqname)
## iname = fqname
## if parent:
## iname = '%s.%s' % (parent.__dict__.get('__iname__', parent.__name__), modname)
#### if _verbose:
#### print "I: get_code: iname is %s" % iname
## rslt = self.extract(iname) # None if not found, (ispkg, code) otherwise
#### if _verbose:
#### print 'I: get_code: rslt', rslt
## if rslt is None:
#### if _verbose:
#### print 'I: get_code: importer', getattr(parent, "__importer__", None),'self',self
## # check the cache if there is no parent or self is the parents importer
## if parent is None or getattr(parent, "__importer__", None) is self:
#### if _verbose:
#### print 'I: get_code: cached 1',iname
## file, desc = Archive._bincache.get(iname, (None, None))
#### if _verbose:
#### print 'I: get_code: file',file,'desc',desc
## if file:
## try:
## fp = open(file, desc[1])
## except IOError:
## pass
## else:
## module = imp.load_module(fqname, fp, file, desc)
## if _verbose:
## print "I: import %s found %s" % (fqname, file)
## return 0, module, {'__file__':file}
## if _verbose:
## print "I: import %s failed" % fqname
##
## return None
##
## ispkg, code = rslt
## values = {'__file__' : code.co_filename, '__iname__' : iname}
## if ispkg:
## values['__path__'] = [fqname]
## if _verbose:
## print "I: import %s found %s" % (fqname, iname)
## return ispkg, code, values
####### Core method - Override as needed #########
def extract(self, name):
""" Get the object corresponding to name, or None.
For use with imputil ArchiveImporter, object is a python code object.
'name' is the name as specified in an 'import name'.
'import a.b' will become:
extract('a') (return None because 'a' is not a code object)
extract('a.__init__') (return a code object)
extract('a.b') (return a code object)
Default implementation:
self.toc is a dict
self.toc[name] is pos
self.lib has the code object marshal-ed at pos
"""
ispkg, pos = self.toc.get(name, (0,None))
if pos is None:
return None
self.lib.seek(self.start + pos)
return ispkg, marshal.load(self.lib)
########################################################################
# Informational methods
def contents(self):
"""Return a list of the contents
Default implementation assumes self.toc is a dict like object.
Not required by ArchiveImporter.
"""
return self.toc.keys()
########################################################################
# Building
####### Top level method - shouldn't need overriding #######
def build(self, path, lTOC):
"""Create an archive file of name 'path'.
lTOC is a 'logical TOC' - a list of (name, path, ...)
where name is the internal name, eg 'a'
and path is a file to get the object from, eg './a.pyc'.
"""
self.path = path
self.lib = open(path, 'wb')
#reserve space for the header
if self.HDRLEN:
self.lib.write('\0'*self.HDRLEN)
#create an empty toc
if type(self.TOCTMPLT) == type({}):
self.toc = {}
else: # assume callable
self.toc = self.TOCTMPLT()
for tocentry in lTOC:
self.add(tocentry) # the guts of the archive
tocpos = self.lib.tell()
self.save_toc(tocpos)
if self.TRLLEN:
self.save_trailer(tocpos)
if self.HDRLEN:
self.update_headers(tocpos)
self.lib.close()
####### manages keeping the internal TOC and the guts in sync #######
def add(self, entry):
"""Override this to influence the mechanics of the Archive.
Assumes entry is a seq beginning with (nm, pth, ...) where
nm is the key by which we'll be asked for the object.
pth is the name of where we find the object. Overrides of
get_obj_from can make use of further elements in entry.
"""
if self.os is None:
import os
self.os = os
nm = entry[0]
pth = entry[1]
pynm, ext = self.os.path.splitext(self.os.path.basename(pth))
ispkg = pynm == '__init__'
assert ext in ('.pyc', '.pyo')
self.toc[nm] = (ispkg, self.lib.tell())
f = open(entry[1], 'rb')
f.seek(8) #skip magic and timestamp
self.lib.write(f.read())
def save_toc(self, tocpos):
"""Default - toc is a dict
Gets marshaled to self.lib
"""
marshal.dump(self.toc, self.lib)
def save_trailer(self, tocpos):
"""Default - not used"""
pass
def update_headers(self, tocpos):
"""Default - MAGIC + Python's magic + tocpos"""
self.lib.seek(self.start)
self.lib.write(self.MAGIC)
self.lib.write(self.pymagic)
self.lib.write(struct.pack('=i', tocpos))
class DummyZlib:
def decompress(self, data):
return data
def compress(self, data, lvl):
return data
import iu
##############################################################
#
# ZlibArchive - an archive with compressed entries
#
class ZlibArchive(Archive):
MAGIC = 'PYZ\0'
TOCPOS = 8
HDRLEN = 16
TRLLEN = 0
TOCTMPLT = {}
LEVEL = 9
def __init__(self, path=None, offset=None, level=9):
if path is None:
offset = 0
elif offset is None:
for i in range(len(path)-1, -1, -1):
if path[i] == '?':
offset = int(path[i+1:])
path = path[:i]
break
else:
offset = 0
self.LEVEL = level
Archive.__init__(self, path, offset)
# dynamic import so not imported if not needed
global zlib
if self.LEVEL:
try:
import zlib
except ImportError:
zlib = DummyZlib()
else:
zlib = DummyZlib()
def extract(self, name):
(ispkg, pos, lngth) = self.toc.get(name, (0, None, 0))
if pos is None:
return None
self.lib.seek(self.start + pos)
try:
co = marshal.loads(zlib.decompress(self.lib.read(lngth)))
except EOFError:
raise ImportError, "PYZ entry '%s' failed to unmarshal" % name
return ispkg, co
def add(self, entry):
if self.os is None:
import os
self.os = os
nm = entry[0]
pth = entry[1]
base, ext = self.os.path.splitext(self.os.path.basename(pth))
ispkg = base == '__init__'
try:
txt = open(pth[:-1], 'r').read()+'\n'
except (IOError, OSError):
try:
f = open(pth, 'rb')
f.seek(8) #skip magic and timestamp
bytecode = f.read()
marshal.loads(bytecode).co_filename # to make sure it's valid
obj = zlib.compress(bytecode, self.LEVEL)
except (IOError, ValueError, EOFError, AttributeError):
raise ValueError("bad bytecode in %s and no source" % pth)
else:
txt = string.join(string.split(txt, '\r\n'), '\n')
try:
co = compile(txt, "%s/%s" % (self.path, nm), 'exec')
except SyntaxError, e:
print "Syntax error in", pth[:-1]
print e.args
raise
obj = zlib.compress(marshal.dumps(co), self.LEVEL)
self.toc[nm] = (ispkg, self.lib.tell(), len(obj))
self.lib.write(obj)
def update_headers(self, tocpos):
"""add level"""
Archive.update_headers(self, tocpos)
self.lib.write(struct.pack('!i', self.LEVEL))
def checkmagic(self):
Archive.checkmagic(self)
self.LEVEL = struct.unpack('!i', self.lib.read(4))[0]
class PYZOwner(iu.Owner):
def __init__(self, path):
self.pyz = ZlibArchive(path)
iu.Owner.__init__(self, path)
def getmod(self, nm, newmod=imp.new_module):
rslt = self.pyz.extract(nm)
if rslt is None:
return None
ispkg, co = rslt
mod = newmod(nm)
try:
mod.__file__ = co.co_filename
except AttributeError:
raise ImportError, "PYZ entry '%s' (%s) is not a valid code object" % (nm, repr(co))
if ispkg:
if _environ.has_key('_MEIPASS2'):
localpath = _environ['_MEIPASS2'][:-1]
else:
localpath = iu._os_path_dirname(self.path)
mod.__path__ = [self.path, localpath, iu._os_path_dirname(mod.__file__)]
#print "PYZOwner setting %s's __path__: %s" % (nm, mod.__path__)
importer = iu.PathImportDirector(mod.__path__,
{self.path:PkgInPYZImporter(nm, self),
localpath:ExtInPkgImporter(localpath, nm)},
[iu.DirOwner])
mod.__importsub__ = importer.getmod
mod.__co__ = co
return mod
class PkgInPYZImporter:
def __init__(self, name, owner):
self.name = name
self.owner = owner
def getmod(self, nm):
#print "PkgInPYZImporter.getmod %s -> %s" % (nm, self.name+'.'+nm)
return self.owner.getmod(self.name+'.'+nm)
class ExtInPkgImporter(iu.DirOwner):
def __init__(self, path, prefix):
iu.DirOwner.__init__(self, path)
self.prefix = prefix
def getmod(self, nm):
return iu.DirOwner.getmod(self, self.prefix+'.'+nm)
#XXX this should also get moved out
##iu._globalownertypes.insert(0, PYZOwner)
##iu.ImportManager().install()