The metaclass uese the 'nested' attribute, while the existing code
accidentially used the 'children' attribute. The latter is used
by instances for actual child classes, while the Class/nested
attribute is for the list of classes whose instancse could be potential
children.
Change-Id: I968bd84d074dcdcec37d99be5d3d4edac9c35a0c
The length value "of course" is a hex value, don't use %02u but %02x
This fixes any eUICC command with a Lc > 10 bytes.
Change-Id: I1e1efbfb9916fc43699602cc889cf4b3d42736f2
The encoding was missing a "CHOICE" container and missed the
fact that the refreshFlag presence is mandatory for enable+disable.
Change-Id: I12e2b16b2c1b4b01dfad0d1fb485399827f25ddc
If a TLV was elementary (no nested IEs), and it had only a single
integer content whose value is 0, we erroneously encoded that as
zero-length TLV (len=0, no value part):
>>> rf = pySim.euicc.RefreshFlag(decoded=0);
>>> rf.to_bytes()
b''
>>> rf.to_tlv()
b'\x81\x00'
After this change it is correct:
>>> rf = pySim.euicc.RefreshFlag(decoded=0);
>>> rf.to_bytes()
b'\x00'
>>> rf.to_tlv()
b'\x81\x01\x00'
Change-Id: I5f4c0555cff7df9ccfc4a56da12766d1bf89122f
If we have a list of dicts, and we flatten that into a dict: Only
do that if there are no dicts with duplocate key values in the list,
as otherwise we will loose information during the transformation.
Change-Id: I7f6d03bf323a153f3172853a3ef171cbec8aece7
Closes: OS#6288
Ever since commit 30de9fd8ab in July
we are (properly) using snake_case names in the from_dict (to become
bijective with to_dict). This code was not updated by accident,
creating an exception when using the `aram_get_config`
Change-Id: If216b56b38ab17d13896074aa726278b9ba16923
Related: OS#6119
So far we implemented only one round of "Send the APDU, get SW=61xx,
call GET RESPONSE". This permitted us to receive only data up to 256
bytes.
Let's extend that to doing multiple rounds, concatenating the result.
This allows us to obtain arbitrary-length data from the card.
See Annex C.1 of ETSI TS 102 221 for examples showing multiple 61xx
iterations.
Change-Id: Ib17da655aa0b0eb203c29dc92690c81bd1300778
Closes: OS#6287
The point of this is to move generic code out of pySim-shell.py,
paving the way for more/other executables using the full power of
our class model without having to reinvent the wheel.
Change-Id: Icf557ed3064ef613ed693ce28bd3514a97a938bd
It's better for the human reader (and more obvious that it's a boolean
value) if we decode single Bits as True/False instead of 1/0.
Change-Id: Ib025f9c4551af7cf57090a0678ab0f66a6684fa4
This increases test coverage and also shows where we so far only
have decoders but no encoders yet
Change-Id: I7932bab7c81a2314c1b9477f50b82a46f24d074e
Don't even send any non-decimal PIN values to the card, but reject
them when parsing the command arguments.
Change-Id: Icec1698851471af7f76f20201dcdcfcd48ddf365
before this patch:
pySIM-shell (00:MF)> echo foo bar baz
usage: echo [-h] string
echo: error: unrecognized arguments: bar baz
after this patch:
pySIM-shell (00:MF)> echo foo bar baz
foo bar baz
Change-Id: I1369bc3aa975865e3a8a574c132e469813a9f6b9
Let's make sure we don't even bother to ask the card to verify
anything as ADM1 pin which is not either a sequence of decimal digits
or an even number of hex digits (even number of bytes).
Change-Id: I4a193a3cf63462fad73d145ab1481070ddf767ca
Let's add a proper argparser instance for the 'verify_adm' command,
avoiding situations where the user types 'verif_adm --help' and then
--help is interpreted as the PIN value, removing one more attempt from
the failed ADM1 counter.
Let's use that opportunity to improve the documentation of the command.
Change-Id: I3321fae66a11efd00c53b66c7890fce84796e658
pySim-shell output has changed over time, so some examples were
showing outdated content. Let's update those.
Change-Id: I4058719c32b61689522e90eba37253e8accb8ba5
this is far from being complete, just some basic first commands
to get the certificates and eIM configuration.
Change-Id: Ie05108e635ed9c6de10f0ba431cb1b13893f6be8
This just adds basic support for the ISD-R application and its
associated STORE DATA command which is used for the ES10x interfaces
between off-card entities and the on-card ISD-R.
Change-Id: Ieab37b083e25d3f36c20f6e9ed3e4bdfdd14a42a
Closes: OS#5637
We used to support only single-byte tags in bertlv_encode_tag,
let's fix that. The easy option is to simply call bertlv_parse_tag,
as that already supported multi-byte tags.
Change-Id: If0bd9137883c4c8b01c4dfcbb53cabeee5c1ce2b
Now that pySim-shell can switch between logical channels, let's state
the currently used logical channel in the prompt.
Change-Id: I45781a6fba205eeb4ac7f58d5cb642b7131bdd88
Related: OS#6230
We've already had the 'open_channel' and 'close_channel' commands,
which were sent to (and acknowledged by) the card. However,
those commands didn't affect the pySim-shell state, i.e. all
communication would still happen through the default channel '0'.
With this patch we introduce a 'switch_channel' command, using which
the user can determine which of the (previously opened) logical channels
shall be used by pySim-shell.
Change-Id: Ia76eb45c4925882ae6866e50b64d9610bd4d546d
Closes: OS#6230
Now that pySim-shell is aware of logical channels and issues almost
all of its APDUs on the currently selected channel, we must also make
sure that ADF selection by AID (implemented by the CardBase class)
issues the SELECT on the respective logical channel.
Before this patch, SELECT ADF by AID would always be issued on the
primary logical channel (0), irrespective of the currently active
RuntimeLchan.
Change-Id: Idf05c297e6a2e24ca539408b8912e348c0782bb4
Related: OS#6230
This new approach will "fork" separate SimCardCommands instances
for each RuntimeLchan. Higher-layer code should now always use the
RuntimeLchan.scc rather than the RuntimeState.card._scc in order to
make sure commands use the correct logical channel.
Change-Id: I13e2e871f2afc2460d9fd1cd566de42267c7d389
Related: OS#6230
Historically we always only had one instance of SimCardCommands, but
with this patch we can now have multiple instances, one for each lchan.
The SimCardCommands class is aware of the logical channel it runs on
and will patch the CLA byte accordingly.
Change-Id: Ibe5650dedc0f7681acf82018a86f83377ba81d30
Related: OS#6230
This fixes the below error during build of the documentation:
pysim/docs/shell.rst:349: ERROR: "<class 'pySim-shell.PySimCommands'>" has no attribute "apdu_cmd_parser"
Change-Id: If89b66a45ea18b5a3fc56bf77b05e679463da5a8
We already have the open channel and close_channel commands in
pySim-shell. They are sent to the card and acknowledged, respectively.
We also already do have code that can track multiple different logical
channels (the rs.lchan array). However, this is currently only used by
pySim-trace, and not by pySim-shell. Let's change that.
Change-Id: Idacee2dc57e8afe85c79bc85b259064e7f5b83a2
Related: OS#6230
While our base classes (TransparentEF / LinFixedEF) always have the
dsecription as 4th argument after "fid, sfid, name", most of the derived
file-specific classes do not share that same argument order.
As seen in the bug fixed by previous Change-Id I7f32c9fd01094620b68b0e54536ecc6cdbe67903
this can have serious consequences. Let's avoid using unnamed
(positional) arguments for the description text altogether.
Change-Id: Icfb3fd1bae038c54fa14a91aa9f75219d839968c
We were using positional arguments when instantiating instances
of classes like EF_5GS3GPPLOCI with non-default names/fids/...
However, we got the argument order wrong and were passing the
description string in the position of the file size, which causes
exceptions like the following from pySim-trace:
Traceback (most recent call last):
File "/home/laforge/projects/git/pysim/./pySim-trace.py", line 198, in <module>
tracer.main()
File "/home/laforge/projects/git/pysim/./pySim-trace.py", line 125, in main
inst.process(self.rs)
File "/home/laforge/projects/git/pysim/pySim/apdu/__init__.py", line 259, in process
self.processed = method(self.lchan)
File "/home/laforge/projects/git/pysim/pySim/apdu/ts_102_221.py", line 152, in process_on_lchan
if self.cmd_dict['offset'] != 0 or self.lr < self.file.size[0]:
TypeError: '<' not supported between instances of 'int' and 'str'
Let's use named initializers for any arguments after the usual "fid, sfid, name"
initial arguments.
Change-Id: I7f32c9fd01094620b68b0e54536ecc6cdbe67903
This adds a new operation mode for pySim-shell, where a single command
can be passed to pySim-shell, which then is executed before pySim-shell
terminates.
Example: ./pySim-shell.py -p0 export --json
Change-Id: I0ed379b23a4b1126006fd8f9e7ba2ba07fb01ada
Closes: OS#6088
Rather than having to know and explicitly list every CardApplication,
let's iterate over the __subclasses__ of the CardApplication base class.
Change-Id: Ia6918e49d73d80acfaf09506e604d4929d37f1b6
This profile has always been a hack/work-around for the situation that
a classic GSM SIM is not a UICC, and we didn't yet have the concept of
CardProfileAddons yet, so there was no way to probe and add something
to an UICC which was not an application with its own AID/ADF.
Since now we have CardProfileAddons (including one for GSM SIM),
and pySim-trace (the other user of CardProfileUICCSIM) has also switched
over to using CardProfileUICC + addons, we can remove this work-around.
Change-Id: I45cec68d72f2003123da4c3f86ed6a5a90988bd8
We have a strict "one CardProfile per card" rule. For a modern UICC
without legacy SIM support, that works great, as all applications
have AID and ADF and can hence be enumerated/detected that way.
However, in reality there are mostly UICC that have legacy SIM, GSM-R
or even CDMA support, all of which are not proper UICC applications
for historical reasons.
So instead of having hard-coded hacks in various places, let's introduce
the new concept of a CardProfileAddon. Every profile can have any
number of those. When building up the RuntimeState, we iterate over the
CardProfile addons, and probe which of those are actually on the card.
For those discovered, we add their files to the filesystem hierarchy.
Change-Id: I5866590b6d48f85eb889c9b1b8ab27936d2378b9
As pySim.cdma_ruim was not imported by test_files.py, the unit tests
were apparently never executed and hence didn't pass. Let's fix both
of those problems.
Change-Id: Icdf4621eb68d05a4948ae9efeb81a007d48e1bb7
Hence move this from the derived classes into the respective base
classes SimCardBase and UiccCardBase
Change-Id: Iad197c2b560c5ea05c54a122144361de5742aafd
Those old flat dicts indicating FID to string-name mapping have long
been obsoleted by the pySim.filsystem based classes.
Change-Id: I20ceea3fdb02ee70d8c8889c078b2e5a0f17c83b
There are some functions / classes which are only needed by the legacy
tools pySim-{read,prog}, bypassing our modern per-file transcoder
classes. Let's move this code to the pySim/legacy sub-directory,
rendering pySim.legacy.* module names.
The long-term goal is to get rid of those and have all code use the
modern pySim/filesystem classes for reading/decoding/encoding/writing
any kind of data on cards.
Change-Id: Ia8cf831929730c48f90679a83d69049475cc5077
This introduces an internal split between
* the code that is shared between pySim-shell and legacy tools, which is
now in the new class hierarchy {Card,SimCard,UiccCard}Base
* the code that is only used by legacy tools,
which is using the old class names inherited from the *Base above
All users still go through the legacy {Sim,Usim,Isim}Card classes, they
will be adjusted in subsequent patches.
Change-Id: Id36140675def5fc44eedce81fc7b09e0adc527e1
This was currently not handled in build_select_path_to(), resulting in
weird exceptions like 'Cannot determine path from MF(3f00) to MF(3f00)'
Change-Id: I41b9f047ee5dc6b91b487f370f011af994aaca04
The get_data shell command didn't have any interactive help / syntax,
and no meaningful error message in case an unknown data object name
was specified by the user. Let's fix that.
Change-Id: I09faaf5d45118635cf832c8c513033aede1427e5
This is all quite complicated. In general, the TLV_IE.to_dict() method
obviously is expected to return a dict (with key equal to the snake-case
name of the class, value to the decode IE value). This single-entry
dict can then be passed back to the from_dict() method to build the
binary representation.
However, with a TLV_IE_Collection, any TLV_IE can occur any number of
times, so we need an array to represent it (dict would need unique key,
which doesn't exist in multiple instances of same TLV IE). Hence, the
TLV_IE_Collection.to_dict() method actually returns a list of dicts,
rather than a dict itself. Each dict in the list represents one TLV_IE.
When encoding such a TLV_IE_Collection back from the list-of-dicts, we
so far didn't handle this special case and tried to de-serialize with
a class-name-keyed dict, which doesn't work.
This patch fixes a regression in the aram_store_ref_ar_do pySim-shell
command which got introduced in Change-Id I3dd5204510e5c32ef1c4a999258d87cb3f1df8c8
While we're fixing it, add some additional comments to why things are
how they are.
Change-Id: Ibdd30cf1652c864f167b1b655b49a87941e15fd5
An invalid variable used in a raise ValueError() would cause a further
exception, depriving the user of a meaningful error message.
Change-Id: I6eb31b91bd69c311f07ff259a424edc58b57529a
The TLV_IE_Collection, just like the individual TLV classes, do
use their snake-style names when converting from binary to dict
using the to_dict() method. It is inconsistent (and a bug) to
expect the CamelCase names during encoding (from_dict). After all,
we want the output of to_dict() to be used as input to from_dict().
Change-Id: Iabd1ad98c3878659d123eef919c22ca824886f8a
So far, if no known programmable card (like sysmoISIM) has been found,
we were using the SimCard base class. However, once we detect an UICC,
we should have switched to the UsimCard class, as otherwise the various
methods called by USIM/ISIM specific commands don't exist and we get
weird 'SimCard' object has no attribute 'update_ust' execptions.
The entire auto-detection and the legacy SimCard / UsimCard classes
are showing the legacy of the code base and should probably be
re-architected. However, let's fix the apparent bug for now.
Change-Id: I5a863198084250458693f060ca10b268a58550a1
Closes: OS#6055
Now that we have fixed OS#6073 in the previous commit, we can enable
the so-far disabled encoder tests for EF.{DOMAIN,IMPU,IMPI} and
remove associated FIXMEs.
Change-Id: I79bfc5b77122907d6cc2f75605f9331b5e650286
The existing IE.from_dict() method *supposedly* accepts a dict as
input value, but it actually expects the raw decoded value, unless it is
a nested IE. This is inconsistent in various ways, and results in a bug
visible at a higher layer, such as files like EF.{DOMAIN,IMPI,IMPU},
which are transparent files containing a single BER-TLV IE.
Decoding such files worked, but re-encoding them did not, due to the
fact that we'd pass a dict to the from_dict method, which then gets
assigned to self.decoded and further passed along to any later actual
encoder function like to_bytes or to_tlv. In that instance, the dict
might be handed to a self._construct which has no idea how to process
the dict, as it expects the raw decoded value.
Change-Id: I3dd5204510e5c32ef1c4a999258d87cb3f1df8c8
Closes: OS#6073
Related: OS#6072
smpp.pdu.pdu_types.DataCodingScheme.GSM_MESSAGE_CLASS very much exists,
and I can prove that manually in the python shell. So let's assume this
is a pylint bug and work around it
pySim/sms.py:72:21: E1101: Instance of 'DataCodingScheme' has no 'GSM_MESSAGE_CLASS' member (no-member)
Change-Id: Iab34bae06940fecf681af9f45b8657e9be8cbc7b
In cmd2, the upstream authors decided to rename a method in 2.0.0
without providing a backwards compatibility wrapper. Let's add that
locally.
Change-Id: Iaa17b93db13ba330551799cce5f0388c78217224
Closes: OS#6071
Rather than writing one test class with associated method for each
OTA algorithm / test, let's do this in a data-driven way, where new
test cases just have to provide test data, while the code iterates over
it.
Change-Id: I8789a21fa5a4793bdabd468adc9fee3b6e633c25
This is important to produce the right command syntax when generating
command line reference in the user manual. However, we shouldn't add
this kludge to the individual programs, but only to the documentation
using the :prog: syntax.
Change-Id: I2ec7ab00c63d5d386f187e54755c71ffc2dce429
The routing indicator is BCD-encoded but has an arbitrary length of
1, 2, 3 or 4 digits.
In order to support the odd lengths of 1 or 3, we must not pad on the
byte level, but on the nibble level. This requires a slight extension of
the Rpad() Adapter.
Change-Id: I6c26dccdd570de7b7a4cd48338068e230340ec7c
Fixes: OS#6054
The K value in case of TUAK can be 16 or 32 bytes long. We used to
permit/parse/display 32 bytes even if only 16 bytes was configured.
Let's enforce the correct length of "K".
Fixes: OS#6053
Change-Id: Ia0f9a2138f16dce72f3118001e95baa1c80f23ce
We've had a "suspend_uicc" command since commit
ec95053249 in 2021, but didn't yet
have the corresponding "resume" pair.
Note that you cannot really execute this in a reasonable way from
within pySim, as it is required to power-cycle the card
between SUSPEND and RESUME, see TS 102 221 Section 11.1.22.3.2
Change-Id: I3322fde74f680e77954e1d3e18a32ef5662759f2
The SUSPEND UICC command is a TS 102 221 (UICC) command, so move
it to the UICC Card Profile.
Also, make sure that any shell command sets specified in the
CardProfile are actually installed during equip().
Change-Id: I574348951f06b749aeff986589186110580328bc
prior to this patch, the suspend_uicc command would always cause a
python exception as a list of integers was returned by decode_duration rather than a single integer (that can be used with %u format string).
Change-Id: I981e9d46607193176b28cb574564e6da546501ba
This avoids error messages about re-registering the same TS 102 222
commands during executing the 'equip' command.
Change-Id: I3567247fe84e928e3ef404c07eff8250ef04dfe9
So for some weird historical reasons, the same python module is
available as pycryptodome (Crypto.* namespace) and pycryptodomex
(Cryptodome.* namespace). See the following information on the project
homepage: https://www.pycryptodome.org/src/installation
To make things extra-weird, Debian choose to package pycryptodomex as
python3-pycryptodome
(https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=886291).
So in order to support both Debian-packaged and differently-installed
packages, let's switch to pycryotodomex on all platforms/installers.
Change-Id: I04daed01f51f9702595ef9f9e0d7fcdf1e4adb62
When using AES CMAC for authentication of OTA messages, we must not pad
the user data before calling the CMAC function. This is unlike the DES
MAC, where padding to the DES block size is mandatory.
This bug was discovered when trying to talk OTA with AES to a
sysmoISIM-SJA5. This patch makes the OTA AES interoperate with the
card. Also, with this patch the cryptographic results of pySim/ota.py
are identical to those of the java code
org.opentelecoms.gsm0348.impl.crypto.CipheringManager
Change-Id: I4b40b5857f95ccb21c35795abe7a1995e368bac3
Unfortunately, TUAK requires a number of additional (and
differently-sized) parameters, so the format of EF.USIM_AUTH_KEY
differs significantly depending on TUAK or non-TUAK case.
Change-Id: I0dcfe05777510fb34973dc2259b137133d8e199d
This implicitly adds support for JSON->binary encoding, not just
decoding (previous code predating construct support).
Change-Id: I0994d9f66a504dd3c60b43ed5cf6645515dcbc6a
According to TS 23.003 Section 28.15 and 28.16 both GLI and GCI
are NAI as defined in IETF RFC 7542, which in turn specifies they
are encoded in UTF-8.
Change-Id: I0a82bd0d0a2badd7bc4a1f8de2c3e3c144ee5b12
This file is rather important for 5G SA operation, so we should have
a proper encoder/decoder in place.
Change-Id: I1b37fdfc2807976880b2cafb61951f08eebeb344
The cards are 99% software-compatible to the SJA2, so let's just
derive the SJA5 class from the SJA2
Change-Id: I706631baaf447c49904277886bc9a3f6ba3f5532
DF.SAIP (SIMalliance Interoperable Profile) is not part of 31.102,
but something from the eSIM/eUICC universe of TCA (formerly known as
SIMalliance). However, as 3GPP does not specify how/where the card
stores the information required for SUCI calculation, the
TCA/SIMalliance standard is the only standard there is. Some CardOS
start to use this standard even for non-eSIM/eUICC use cases.
Change-Id: Iffb65af335dfdbd7791fca9a0a6ad4b79814a57c
The FID in ADF.USIM is different from the FID in DF.GSM. So while
we can re-use the ts_51_011 EF_NIA class definition, we must pass in
a different fid to the constructor.
Change-Id: Ib414d5b476666e276824266e33b341175a2ee05a
Harald Welte