Xiang Li

19 papers A* 3Journal 4Unranked 12
YearRankTypeTitle / Venue / Authors
2024 conf
ACL (1)
Qi Cheng, Michael Boratko, Pranay Kumar Yelugam, Tim O'Gorman, Nalini Singh, Andrew McCallum, Xiang Li
2024 A* conf
EMNLP
Huihan Li, Yuting Ning, Zeyi Liao, Siyuan Wang, Xiang Li, Ximing Lu, Wenting Zhao, Faeze Brahman, Yejin Choi, Xiang Ren
2024 conf
EMNLP (Findings)
Mohamed Elaraby, Diane J. Litman, Xiang Li, Ahmed Magooda
2024 conf
NAACL-HLT (Findings)
Tenghao Huang, Dongwon Jung, Vaibhav Kumar, Mohammad Kachuee, Xiang Li, Puyang Xu, Muhao Chen
2024 conf
NAACL-HLT
Wenting Zhao, Justin T. Chiu, Jena D. Hwang, Faeze Brahman, Jack Hessel, Sanjiban Choudhury, Yejin Choi, Xiang Li, Alane Suhr
2023 A* conf
EMNLP
Anshita Gupta, Debanjan Mondal, Akshay Krishna Sheshadri, Wenlong Zhao, Xiang Li, Sarah Wiegreffe, Niket Tandon
2022 conf
ACL (1)
Shib Sankar Dasgupta, Michael Boratko, Siddhartha Mishra, Shriya Atmakuri, Dhruvesh Patel, Xiang Li, Andrew McCallum
2020 A* conf
NeurIPS
Shib Sankar Dasgupta, Michael Boratko, Dongxu Zhang, Luke Vilnis, Xiang Li, Andrew McCallum
2020 J jnl
CoRR
Anshuman Mishra, Dhruvesh Patel, Aparna Vijayakumar, Xiang Li, Pavan Kapanipathi, Kartik Talamadupula
2020 conf
EMNLP (1)
Michael Boratko, Xiang Li, Tim O'Gorman, Rajarshi Das, Dan Le, Andrew McCallum
2020 conf
*SEM@COLING
Anshuman Mishra, Dhruvesh Patel, Aparna Vijayakumar, Xiang Li, Pavan Kapanipathi, Kartik Talamadupula
2020 J jnl
CoRR
Anshuman Mishra, Dhruvesh Patel, Aparna Vijayakumar, Xiang Li, Pavan Kapanipathi, Kartik Talamadupula
2020 conf
AKBC
Dhruvesh Patel, Shib Sankar Dasgupta, Michael Boratko, Xiang Li, Luke Vilnis, Andrew McCallum
2019 conf
ICLR
Xiang Li, Luke Vilnis, Dongxu Zhang, Michael Boratko, Andrew McCallum
2018 conf
ACL (1)
Luke Vilnis, Xiang Li, Shikhar Murty, Andrew McCallum
2018 J jnl
CoRR
Luke Vilnis, Xiang Li, Shikhar Murty, Andrew McCallum
2017 J jnl
CoRR
Xiang Li, Luke Vilnis, Andrew McCallum
2016 conf
ACL (1)
Xiang Li, Aynaz Taheri, Lifu Tu, Kevin Gimpel
2016 conf
eScience
Xiang Li, Xiaoyang Xu, Tanu Malik
redb/extractors/macho_extractor.py
← Index redb/extractors/macho_extractor.py python
import logging
from abc import ABCMeta, abstractmethod
import inspect
import sys
import os

import machofile

from redb.extractors.extractor import Extractor

logger = logging.getLogger(__name__)


@abstractmethod
class MachOExtractor(Extractor, metaclass=ABCMeta):

    def __init__(
        self,
        filepath,
        log,
        exporters=None,
        index_prefix=None,
        elastic_index=None,
        known_benign=False,
        known_malicious=False,
        macho=None,
    ):
        # Read binary and parse machofile BEFORE calling super().__init__
        # This avoids reading the file twice
        with open(filepath, "rb") as f:
            binary_data = f.read()

        # Parse machofile with binary data
        self.macho = macho if macho else self._generate_machofile_object(binary_data)

        # Extract hashes from machofile to pass to parent
        precomputed_hashes = None
        if self.macho:
            try:
                general_info = self.macho.get_general_info()
                if general_info:
                    # For FAT binaries, get_general_info() returns dict with 'fat' key
                    # For single-arch, it returns the info directly
                    if 'fat' in general_info:
                        fat_info = general_info['fat']
                        precomputed_hashes = {
                            'MD5': fat_info.get('MD5'),
                            'SHA1': fat_info.get('SHA1'),
                            'SHA256': fat_info.get('SHA256'),
                        }
                    else:
                        precomputed_hashes = {
                            'MD5': general_info.get('MD5'),
                            'SHA1': general_info.get('SHA1'),
                            'SHA256': general_info.get('SHA256'),
                        }
            except Exception as e:
                logger.debug(f"Could not get hashes from machofile: {e}")

        super().__init__(
            filepath,
            log,
            exporters,
            index_prefix,
            elastic_index,
            known_benign,
            known_malicious,
            precomputed_hashes=precomputed_hashes,
        )

        # Store binary data so base class doesn't re-read
        self._binary_data = binary_data

    @property
    def binary(self):
        """Override to use already-read binary data."""
        return self._binary_data

    def _generate_machofile_object(self, binary_data):
        """Generate and parse a machofile object from binary data."""
        macho = None
        try:
            macho = machofile.UniversalMachO(data=binary_data)
            if not macho:
                raise Exception("Empty file?")

            # Parse the MachO object once during initialization
            macho.parse()

        except Exception as e:
            logger.error(f"Format error parsing MachO: {e}")
        return macho

    # def _is_macho_file(self):
    #     """Check if the file is a valid Mach-O binary."""
    #     try:
    #         if not self.macho:
    #             return False
            
    #         # For Universal/FAT binaries, check if any architecture is valid
    #         if hasattr(self.macho, 'is_fat') and self.macho.is_fat:
    #             return len(self.macho.architectures) > 0
    #         else:
    #             # Single architecture binary
    #             return hasattr(self.macho, 'macho') and self.macho.macho is not None
    #     except Exception as e:
    #         self.log.error(f"Error checking Mach-O file: {e}")
    #         return False

    def _is_signed(self):
        """Check if the Mach-O binary is code signed using new API."""
        try:
            if not self.macho:
                return False

            # Get architectures using new API
            architectures = self.macho.get_architectures()

            # For each architecture, check if signed
            for arch in architectures:
                try:
                    signature_info = self.macho.get_code_signature_info(arch=arch)
                    if signature_info and signature_info.get('signed', False):
                        return True
                except Exception:
                    continue

            return False
        except Exception as e:
            self.log.error(f"Error checking Mach-O signature: {e}")
            return False

    def _get_architectures(self):
        """Get list of architectures in the Mach-O binary using new API."""
        try:
            if not self.macho:
                return []

            # Use new API method
            architectures = self.macho.get_architectures()
            return architectures if architectures else []
        except Exception as e:
            self.log.error(f"Error getting architectures: {e}")
            return []

    # def _get_macho_for_arch(self, arch_name=None):
    #     """Get MachO instance for specific architecture or default."""
    #     try:
    #         if not self.macho:
    #             return None
            
    #         if hasattr(self.macho, 'is_fat') and self.macho.is_fat:
    #             if arch_name:
    #                 return self.macho.architectures.get(arch_name)
    #             else:
    #                 # Return first available architecture
    #                 return next(iter(self.macho.architectures.values())) if self.macho.architectures else None
    #         else:
    #             # Single architecture binary
    #             return self.macho.macho if hasattr(self.macho, 'macho') else None
    #     except Exception as e:
    #         self.log.error(f"Error getting MachO for architecture: {e}")
    #         return None

    # def _get_formatted_header_values(self, header):
    #     """Get both raw and human-readable header values."""
    #     try:
    #         macho_instance = self._get_macho_for_arch()
    #         if not macho_instance:
    #             return None
            
    #         # Parse the MachO if not already parsed
    #         if not hasattr(macho_instance, 'header') or not macho_instance.header:
    #             macho_instance.parse()
            
    #         # Get human-readable values using machofile's formatting methods
    #         magic_str = macho_instance.format_magic_value(header.get('magic', 0))
            
    #         # Simple CPU type mapping since CPU_TYPE_MAP is not exposed
    #         cputype = header.get('cputype', 0)
    #         if cputype == 0x7:
    #             cputype_str = "x86"
    #         elif cputype == 0x1000007:
    #             cputype_str = "x86_64"
    #         elif cputype == 0xC:
    #             cputype_str = "ARM"
    #         elif cputype == 0x100000C:
    #             cputype_str = "ARM 64-bit"
    #         else:
    #             cputype_str = str(cputype)
            
    #         cpusubtype_str = macho_instance.decode_cpusubtype(header.get('cputype', 0), header.get('cpusubtype', 0))
    #         filetype_str = macho_instance.format_file_type(header.get('filetype', 0))
    #         flags_str = macho_instance.decode_flags(header.get('flags', 0))
            
    #         return {
    #             'raw': {
    #                 'magic': header.get('magic', 0),
    #                 'cputype': header.get('cputype', 0),
    #                 'cpusubtype': header.get('cpusubtype', 0),
    #                 'filetype': header.get('filetype', 0),
    #                 'flags': header.get('flags', 0),
    #             },
    #             'formatted': {
    #                 'magic_str': magic_str,
    #                 'cputype_str': cputype_str,
    #                 'cpusubtype_str': cpusubtype_str,
    #                 'filetype_str': filetype_str,
    #                 'flags_str': flags_str,
    #             }
    #         }
    #     except Exception as e:
    #         self.log.error(f"Error formatting header values: {e}")
    #         return None