Jan A. Botha

25 papers A* 4A 1B 1Journal 13Unranked 6
YearRankTypeTitle / Venue / Authors
2023 J jnl
Trans. Assoc. Comput. Linguistics
Parker Riley, Timothy Dozat, Jan A. Botha, Xavier Garcia, Dan Garrette, Jason Riesa, Orhan Firat, Noah Constant
2023 J jnl
CoRR
Rohan Anil, Andrew M. Dai, Orhan Firat, Melvin Johnson, Dmitry Lepikhin, Alexandre Passos, Siamak Shakeri, Emanuel Taropa, Paige Bailey, Zhifeng Chen, Eric Chu, Jonathan H. Clark, Laurent El Shafey, Yanping Huang, Kathy Meier-Hellstern, Gaurav Mishra, Erica Moreira, Mark Omernick, Kevin Robinson, Sebastian Ruder, Yi Tay, Kefan Xiao, Yuanzhong Xu, Yujing Zhang, Gustavo Hernández Ábrego, Junwhan Ahn, Jacob Austin, Paul Barham, Jan A. Botha, James Bradbury, Siddhartha Brahma, Kevin Brooks, Michele Catasta, Yong Cheng, Colin Cherry, Christopher A. Choquette-Choo, Aakanksha Chowdhery, Clément Crepy, Shachi Dave, Mostafa Dehghani, Sunipa Dev, Jacob Devlin, Mark Díaz, Nan Du, Ethan Dyer, Vladimir Feinberg, Fangxiaoyu Feng, Vlad Fienber, Markus Freitag, Xavier Garcia, Sebastian Gehrmann, Lucas Gonzalez, et al.
2023 conf
EMNLP (Findings)
Sebastian Gehrmann, Sebastian Ruder, Vitaly Nikolaev, Jan A. Botha, Michael Chavinda, Ankur P. Parikh, Clara Rivera
2022 J jnl
CoRR
Parker Riley, Timothy Dozat, Jan A. Botha, Xavier Garcia, Dan Garrette, Jason Riesa, Orhan Firat, Noah Constant
2022 J jnl
CoRR
Sebastian Gehrmann, Sebastian Ruder, Vitaly Nikolaev, Jan A. Botha, Michael Chavinda, Ankur P. Parikh, Clara Rivera
2021 conf
ACL/IJCNLP (2)
Nicholas FitzGerald, Daniel M. Bikel, Jan A. Botha, Daniel Gillick, Tom Kwiatkowski, Andrew McCallum
2021 J jnl
CoRR
Nicholas FitzGerald, Jan A. Botha, Daniel Gillick, Daniel M. Bikel, Tom Kwiatkowski, Andrew McCallum
2021 conf
EMNLP (1)
Sebastian Ruder, Noah Constant, Jan A. Botha, Aditya Siddhant, Orhan Firat, Jinlan Fu, Pengfei Liu, Junjie Hu, Dan Garrette, Graham Neubig, Melvin Johnson
2021 J jnl
CoRR
Sebastian Ruder, Noah Constant, Jan A. Botha, Aditya Siddhant, Orhan Firat, Jinlan Fu, Pengfei Liu, Junjie Hu, Graham Neubig, Melvin Johnson
2020 conf
EMNLP (1)
Julian Michael, Jan A. Botha, Ian Tenney
2020 J jnl
CoRR
Julian Michael, Jan A. Botha, Ian Tenney
2020 conf
EMNLP (1)
Jan A. Botha, Zifei Shan, Daniel Gillick
2020 J jnl
CoRR
Jan A. Botha, Zifei Shan, Daniel Gillick
2018 A* conf
EMNLP
Jan A. Botha, Manaal Faruqui, John Alex, Jason Baldridge, Dipanjan Das
2018 J jnl
CoRR
Jan A. Botha, Manaal Faruqui, John Alex, Jason Baldridge, Dipanjan Das
2017 A* conf
EMNLP
Jan A. Botha, Emily Pitler, Ji Ma, Anton Bakalov, Alex Salcianu, David Weiss, Ryan T. McDonald, Slav Petrov
2017 J jnl
CoRR
Jan A. Botha, Emily Pitler, Ji Ma, Anton Bakalov, Alex Salcianu, David Weiss, Ryan T. McDonald, Slav Petrov
2016 conf
ACL (1)
Jan Buys, Jan A. Botha
2016 J jnl
CoRR
Jan Buys, Jan A. Botha
2015 J jnl
CoRR
Jan A. Botha
2014 A* conf
ICML
Jan A. Botha, Phil Blunsom
2014 J jnl
CoRR
Jan A. Botha, Phil Blunsom
2013 A* conf
EMNLP
Jan A. Botha, Phil Blunsom
2012 B conf
COLING
Jan A. Botha, Chris Dyer, Phil Blunsom
2012 A conf
EACL
Jan A. Botha
redb/extractors/pe_extractors/pe_signature_old.py
← Index redb/extractors/pe_extractors/pe_signature_old.py python
import inspect
import pefile

from hashlib import md5, sha1, sha256

from asn1crypto import cms, pem, x509, core

from redb.extractors.enum import Tag
from redb.extractors.pe_extractor import PEExtractor
from redb.models.dataclasses import PECertificate, PESigner


class PESignatureExtractor(PEExtractor):
    """
    RFC 3161 - Internet X.509 Public Key Infrastructure Time-Stamp Protocol (TSP)
    Section 2.4.2 defines the TSTInfo structure.
    https://tools.ietf.org/html/rfc3161#section-2.4.2

    Microsoft Authenticode Time Stamping Specification
    https://download.microsoft.com/download/9/c/5/9c5b2167-8017-4bae-9fde-d599bac8184a/Authenticode_PE.docx
    """

    def __init__(
        self,
        filepath,
        log,
        index_prefix=None,
        elastic_index=None,
        known_benign=False,
        known_malicious=False,
        pe=None,
    ):
        super().__init__(
            filepath,
            log,
            index_prefix,
            elastic_index,
            known_benign,
            known_malicious,
            pe,
        )
        self.elastic_index = self.index_prefix + "-pe_signature"
        self.log.debug(inspect.currentframe().f_code.co_name)

    def tag(self):
        return Tag.PE_SIGNATURE.value

    @staticmethod
    def _extract_certificate_info(cert_data):
        # self.log.debug(inspect.currentframe().f_code.co_name)

        certificate = x509.Certificate.load(cert_data)
        certificate_serial_number = format(certificate.serial_number, "x").upper()
        grouped_serial_number = ":".join(
            [
                certificate_serial_number[i : i + 2]
                for i in range(0, len(certificate_serial_number), 2)
            ]
        )
        return PECertificate(
            _id=format(certificate.serial_number, "x").upper(),
            certificate_serial_number=grouped_serial_number,
            certificate_issuer=certificate.issuer.human_friendly,
            certificate_subject=certificate.subject.human_friendly,
            certificate_valid_from=certificate["tbs_certificate"]["validity"][
                "not_before"
            ].native.strftime("%Y-%m-%d %H:%M:%S"),
            certificate_valid_to=certificate["tbs_certificate"]["validity"][
                "not_after"
            ].native.strftime("%Y-%m-%d %H:%M:%S"),
            certificate_thumbprint=sha1(cert_data).hexdigest().upper(),
            certificate_md5=md5(cert_data).hexdigest().upper(),
            certificate_sha256=sha256(cert_data).hexdigest().upper(),
        )

    @staticmethod
    def _parse_microsoft_time_stamp(ts_token):
        class TSTInfo(core.Sequence):
            _fields = [
                ("version", core.Integer),
                ("policy_id", core.ObjectIdentifier),
                ("message_imprint", core.Sequence),
                ("serial_number", core.Integer),
                ("gen_time", core.GeneralizedTime),
                ("accuracy", core.Sequence, {"optional": True}),
                ("ordering", core.Boolean, {"optional": True}),
                ("nonce", core.Integer, {"optional": True}),
                ("tsa", core.Sequence, {"optional": True}),
                ("extensions", core.Sequence, {"optional": True}),
            ]

        # The content should be SignedData
        signed_data = ts_token["content"]

        # The actual timestamp should be in the encapContentInfo
        encap_content_info = signed_data["encap_content_info"]
        tst_info = TSTInfo.load(encap_content_info["content"])

        # Extract the genTime field from TSTInfo, type(genTime): <class 'datetime.datetime'>
        gen_time = tst_info["gen_time"].native

        return gen_time

    def _extract_signer_info(self, signer):
        self.log.debug(inspect.currentframe().f_code.co_name)
        signing_time = None

        for attr_set in [signer["signed_attrs"], signer["unsigned_attrs"]]:
            if attr_set:
                for attribute in attr_set:
                    attr_type = attribute["type"].dotted
                    if attr_type == "1.2.840.113549.1.9.5":  # signingTime
                        signing_time = attribute["values"][0].native
                        break
                    elif (
                        attr_type == "1.3.6.1.4.1.311.3.2.1"
                        or attr_type == "1.3.6.1.4.1.311.3.3.1"
                    ):  # microsoft_time_stamp_token
                        try:
                            ts_token = attribute["values"][0].native
                            signing_time = self._parse_microsoft_time_stamp(ts_token)
                            self.log.debug(
                                f"Signing Time from Microsoft Time Stamp Token: {signing_time}"
                            )
                        except Exception as e:
                            self.log.error(
                                f"Error parsing Microsoft Time Stamp Token: {e}"
                            )
                            self.log.error("Token structure:", attribute["values"][0])
                    elif attr_type == "1.2.840.113549.1.9.6":  # counterSignature
                        counter_sig_values = attribute["values"]
                        for counter_sig_value in counter_sig_values:
                            try:
                                counter_sig_info = cms.SignerInfo.load(
                                    counter_sig_value.native
                                )
                                cs_signed_attrs = counter_sig_info["signed_attrs"]
                                for cs_attr in cs_signed_attrs:
                                    cs_attr_type = cs_attr["type"].dotted
                                    if (
                                        cs_attr_type == "1.2.840.113549.1.9.5"
                                    ):  # signingTime
                                        signing_time = cs_attr["values"][0].native
                                        self.log.debug(
                                            "Signing Time (from countersignature):",
                                            signing_time,
                                        )
                                        break
                                if signing_time:
                                    break
                            except Exception as e:
                                self.log.error(
                                    f"Error processing countersignature: {e}"
                                )
                    if signing_time:
                        break
            if signing_time:
                break
        signer_serial_number = format(
            signer["sid"].native["serial_number"], "x"
        ).upper()
        grouped_serial_number = " ".join(
            [
                signer_serial_number[i : i + 2]
                for i in range(0, len(signer_serial_number), 2)
            ]
        )
        return PESigner(
            # this is to avoid duplicate _id with the Signer certificate used to sign as they are in the same ES index
            _id=format(signer["sid"].native["serial_number"], "x").upper()
            + signer["digest_algorithm"]["algorithm"].native,
            signer_serial_number=grouped_serial_number,
            signer_digest_algorithm=signer["digest_algorithm"]["algorithm"].native,
            signer_signature_algorithm=signer["signature_algorithm"][
                "algorithm"
            ].native,
            signing_time=(
                signing_time.strftime("%Y-%m-%d %H:%M:%S") if signing_time else None
            ),
            # "countersigners": extract_countersigners(signer), # TODO: Implement this
        )

    def _extract_signature_info(self):
        self.log.debug(inspect.currentframe().f_code.co_name)
        address = self.pe.OPTIONAL_HEADER.DATA_DIRECTORY[
            pefile.DIRECTORY_ENTRY["IMAGE_DIRECTORY_ENTRY_SECURITY"]
        ].VirtualAddress
        size = self.pe.OPTIONAL_HEADER.DATA_DIRECTORY[
            pefile.DIRECTORY_ENTRY["IMAGE_DIRECTORY_ENTRY_SECURITY"]
        ].Size

        addr_8 = address + 8
        signature_data = bytes(
            self.pe.write()[addr_8 : addr_8 + size]  # noqa E203
        )  # Ensure this is a bytes object
        if pem.detect(signature_data):
            signature_data = pem.unarmor(signature_data)

        content_info = cms.ContentInfo.load(signature_data)
        signed_data = content_info["content"]

        certificates = signed_data["certificates"]
        signers = signed_data["signer_infos"]

        certificates_list = []
        for cert in certificates:
            cert_info = self._extract_certificate_info(cert.chosen.dump())
            certificates_list.append(cert_info)

        signer_info_list = []
        for signer in signers:
            signer_info = self._extract_signer_info(signer)
            signer_info_list.append(signer_info)

        return certificates_list, signer_info_list

    def extract(self):
        try:
            self.log.debug(inspect.currentframe().f_code.co_name)
            if self._is_signed():
                certificates_list, signers_list = self._extract_signature_info()
                # self.export_to_elastic([PECodeSigning(certificates_list, signers_list)])
                self.export_to_elastic(certificates_list, Tag.PE_CERTIFICATE.value)
                self.export_to_elastic(signers_list, Tag.PE_SIGNER.value)
            return True
        except Exception as e:
            self.log.error(f"Error extracting PE Signature: {e}")
            return None