def extract(location, target_dir, arch_type='*'):
"""
Extract all files from a 7zip-supported archive file at location in the
target_dir directory. Return a list of warning messages.
Raise exception on errors.
`arch_type` is the type of 7zip archive passed to the -t 7zip option. Can be
None.
"""
assert location
assert target_dir
abs_location = os.path.abspath(os.path.expanduser(location))
abs_target_dir = os.path.abspath(os.path.expanduser(target_dir))
if is_rar(location):
raise ExtractErrorFailedToExtract('RAR extraction disactivated')
# note: there are some issues with the extraction of debian .deb ar files
# see sevenzip bug http://sourceforge.net/p/sevenzip/bugs/1472/
# 7z arguments
extract = 'x'
yes_to_all = '-y'
# NB: we use t* to ensure that all archive types are honored
if not arch_type:
arch_type = ''
else:
arch_type = '-t' + arch_type
# pass an empty password so that extraction with passwords WILL fail
password = '******'
# renaming may not behave the same way on all OSes in particular Mac and Windows
auto_rename_dupe_names = '-aou'
# These things do not work well with p7zip for now:
# - ensure that we treat the FS as case insensitive even if it is
# this ensure we have consistent names across OSes
# case_insensitive = '-ssc-'
# - force any console output to be UTF-8 encoded
# TODO: add this may be for a UTF output on Windows only
# output_as_utf = '-sccUTF-8'
# working_tmp_dir = '-w<path>'
# NB: we force running in the GMT timezone, because 7z is unable to set
# the TZ correctly when the archive does not contain TZ info. This does
# not work on Windows, because 7z is not using the TZ env var there.
timezone = dict(os.environ)
timezone.update({u'TZ': u'GMT'})
timezone = command.get_env(timezone)
# Note: 7z does extract in the current directory so we cwd to the target dir first
args = [
extract, yes_to_all, auto_rename_dupe_names, arch_type, abs_location,
password
]
lib_dir = get_location(EXTRACTCODE_7ZIP_LIBDIR)
cmd_loc = get_location(EXTRACTCODE_7ZIP_EXE)
ex_args = dict(
cmd_loc=cmd_loc,
args=args,
lib_dir=lib_dir,
cwd=abs_target_dir,
env=timezone,
)
if TRACE:
logger.debug('extract: args:')
pprint(ex_args)
rc, stdout, stderr = command.execute2(**ex_args)
if rc != 0:
if TRACE:
logger.debug(
'extract: failure: {rc}\nstderr: {stderr}\nstdout: {stdout}\n'.
format(**locals()))
error = get_7z_errors(stdout) or UNKNOWN_ERROR
raise ExtractErrorFailedToExtract(error)
extractcode.remove_backslashes_and_dotdots(abs_target_dir)
return get_7z_warnings(stdout)
def list_entries(location, arch_type='*'):
"""
Tield Entry tuples for each entry found in a 7zip-supported archive file at
`location`. Use the provided 7zip `arch_type` CLI archive type code (e.g. with
the "-t* 7z" cli type option) (can be None).
"""
assert location
abs_location = os.path.abspath(os.path.expanduser(location))
if is_rar(location):
return []
# 7z arguments
listing = 'l'
# NB: we use t* to ensure that all archive types are honored
if not arch_type:
arch_type = ''
else:
arch_type = '-t' + arch_type
# pass an empty password so that extraction with passwords WILL fail
password = '******'
tech_info = '-slt'
output_as_utf = ''
if on_windows:
output_as_utf = '-sccUTF-8'
# NB: we force running in the GMT timezone, because 7z is unable to set
# the TZ correctly when the archive does not contain TZ info. This does
# not work on Windows, because 7z is not using the TZ env var there.
timezone = dict(os.environ)
timezone.update({u'TZ': u'GMT'})
timezone = command.get_env(timezone)
args = [
listing,
tech_info,
arch_type,
output_as_utf,
abs_location,
password,
]
lib_dir = get_location(EXTRACTCODE_7ZIP_LIBDIR)
cmd_loc = get_location(EXTRACTCODE_7ZIP_EXE)
rc, stdout, stderr = command.execute2(cmd_loc=cmd_loc,
args=args,
lib_dir=lib_dir,
env=timezone,
to_files=True)
if TRACE:
logger.debug(
'list_entries: rc: {rc}\nstderr: file://{stderr}\nstdout: file://{stdout}\n'
.format(**locals()))
if rc != 0:
# FIXME: this test is useless
_error = get_7z_errors(stdout) or UNKNOWN_ERROR
# the listing was produced as UTF on windows to avoid damaging binary
# paths in console outputs
utf = bool(output_as_utf)
return parse_7z_listing(stdout, utf)
def build_7z_extract_command(location,
target_dir,
single_entry=None,
arch_type='*'):
"""
Return a mapping of 7z command line aguments to extract the archive at
`location` to `target_dir`.
If `single_entry` contains an Entry, provide the command to extract only
that single entry "path" in the current directory without any leading path.
"""
# 7z arguments
if single_entry:
# do not use full path
extract = 'e'
else:
extract = 'x'
yes_to_all = '-y'
# NB: we use t* to ensure that all archive types are honored
if not arch_type:
arch_type = ''
else:
arch_type = '-t' + arch_type
# pass an empty password so that extraction with passwords WILL fail
password = '******'
# renaming may not behave the same way on all OSes in particular Mac and Windows
auto_rename_dupe_names = '-aou'
# Ensure that we treat the FS as case insensitive if that's what it is
# -ssc Set case-sensitive mode. It's default for Posix/Linux systems.
# -ssc- Set case-insensitive mode. It's default for Windows systems.
# historically, this was not needed on macOS, but now APFS is case
# insentitive as a default
if on_windows or on_macos_14_or_higher or not is_case_sensitive_fs:
case_sensitive = '-ssc-'
else:
case_sensitive = '-ssc'
# These does not work well with p7zip for now:
# - force any console output to be UTF-8 encoded
# TODO: add this may be for a UTF output on Windows only
# output_as_utf = '-sccUTF-8'
# working_tmp_dir = '-w<path>'
# NB: we force running in the GMT timezone, because 7z is unable to set
# the TZ correctly when the archive does not contain TZ info. This does
# not work on Windows, because 7z is not using the TZ env var there.
timezone = dict(os.environ)
timezone.update({u'TZ': u'GMT'})
timezone = command.get_env(timezone)
# Note: 7z does extract in the current directory so we cwd to the target dir first
args = [
extract,
yes_to_all,
case_sensitive,
auto_rename_dupe_names,
arch_type,
password,
'--',
location,
]
if single_entry:
args += [shlex_quote(single_entry.path)]
lib_dir, cmd_loc = get_bin_locations()
ex_args = dict(
cmd_loc=cmd_loc,
args=args,
lib_dir=lib_dir,
cwd=target_dir,
env=timezone,
)
if TRACE:
logger.debug('extract: args:')
pprint.pprint(ex_args)
return ex_args