Ian Kelley

17 papers A 2Journal 6Unranked 8
YearRankTypeTitle / Venue / Authors
2014 conf
ScienceCloud@HPDC
Ian Kelley
2014 conf
DICT@HPDC
Ian Kelley, Joshua Evan Blumenstock
2012
Ian Kelley
2011 conf
IPDPS Workshops
AbdelHamid Elwaer, Andrew Harrison, Ian Kelley, Ian J. Taylor
2009 J jnl
Future Gener. Comput. Syst.
Carlo Mastroianni, Pasquale Cozza, Domenico Talia, Ian Kelley, Ian J. Taylor
2008 conf
CoreGRID Integration Workshop
Stavros Isaiadis, Vladimir Getov, Ian Kelley, Ian J. Taylor
2008 conf
ChinaGrid
Gilles Fedak, Haiwu He, Oleg Lodygensky, Zoltán Balaton, Zoltán Farkas, Gabor Gombás, Péter Kacsuk, Róbert Lovas, Csaba Attila Marosi, Ian Kelley, Ian J. Taylor, Gábor Terstyánszky, Tamás Kiss, Miguel Cárdenas-Montes, Ad Emmen, Filipe Araújo
2008 conf
CoreGRID Integration Workshop
Zoltán Balaton, Zoltán Farkas, Gabor Gombás, Péter Kacsuk, Róbert Lovas, Csaba Attila Marosi, Gábor Terstyánszky, Tamás Kiss, Oleg Lodygensky, Gilles Fedak, Ad Emmen, Ian Kelley, Ian J. Taylor, Miguel Cárdenas-Montes, Filipe Araújo
2008 J jnl
Parallel Process. Lett.
Zoltán Balaton, Zoltán Farkas, Gabor Gombás, Péter Kacsuk, Róbert Lovas, Csaba Attila Marosi, Ad Emmen, Gábor Terstyánszky, Tamás Kiss, Ian Kelley, Ian J. Taylor, Oleg Lodygensky, Miguel Cárdenas-Montes, Gilles Fedak, Filipe Araújo
2008 J jnl
Parallel Process. Lett.
Fernando Costa, Luís Moura Silva, Gilles Fedak, Ian Kelley
2008 A conf
IPDPS
Fernando Costa, Luís Moura Silva, Ian Kelley, Gilles Fedak
2007 J jnl
Concurr. Comput. Pract. Exp.
Chongjie Zhang, Chirag Dekate, Gabrielle Allen, Ian Kelley, Jon MacLaren
2007 J jnl
Concurr. Comput. Pract. Exp.
Chongjie Zhang, Ian Kelley, Gabrielle Allen
2007 conf
CoreGRID Workshop - Making Grids Work
Fernando Costa, Luís Moura Silva, Ian Kelley, Ian J. Taylor
2006 conf
HealthGrid
Andrew Harrison, Ian Kelley, Emil Mieilica, Adina Riposan, Ian J. Taylor
2005 J jnl
Future Gener. Comput. Syst.
Ruxandra Bondarescu, Gabrielle Allen, Greg Daues, Ian Kelley, Michael Russell, Edward Seidel, John Shalf, Malcolm Tobias
2002 A conf
HPDC
Gabrielle Allen, Kelly Davis, Thomas Dramlitsch, Tom Goodale, Ian Kelley, Gerd Lanfermann, Jason Novotny, Thomas Radke, Kashif Rasul, Michael Russell, Edward Seidel, Oliver Wehrens
redb/extractors/pe_extractors/pe_features.py
← Index redb/extractors/pe_extractors/pe_features.py python
import base64
import inspect
import json
from pprint import pprint
import re

import pefile

from redb.ext.rich_header import get_rich_idVersion
from redb.extractors.enum import Tag
from redb.extractors.pe_extractor import PEExtractor
from redb.extractors.pe_extractors.pe_dotnet import PEDotNetExtractor
from redb.models.dataclasses import PE
from datetime import datetime, timezone
from typing import Any


class PEFeaturesExtractor(PEExtractor):

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

    def _extract_type(self):
        self.log.debug(inspect.currentframe().f_code.co_name)
        if self.pe.is_dll():
            return "DLL"
        elif self.pe.is_driver():
            return "DRIVER"
        elif self.pe.is_exe():
            return "EXE"

    def _extract_architecture(self):
        self.log.debug(inspect.currentframe().f_code.co_name)
        mt = {"0x14c": "x86", "0x0200": "Itanium", "0x8664": "x64"}
        machine_value = self.pe.FILE_HEADER.Machine
        if isinstance(machine_value, int):
            return mt.get(str(hex(machine_value)), "")
        return str(machine_value) + " => Not x86/64 or Itanium"

    def _extract_rich_header(self):
        """Extract the Rich header from the PE file

        to decode it back from base64, use:
        rh_tmp = json.loads(rh_b64_string)
        rh_decoded = {}
        for k in rh_tmp:
            if isinstance(rh_tmp[k], str):
                rh_decoded[k] = base64.b64decode(rh_tmp[k])
            else:
                rh_decoded[k] = rh_tmp[k]
        """
        self.log.debug(inspect.currentframe().f_code.co_name)
        rh = self.pe.parse_rich_header()
        rich_header_b64 = {}
        rich_header_infos = []
        if rh:
            for key in rh:
                if isinstance(rh[key], bytes):
                    rich_header_b64[key] = base64.b64encode(rh[key]).decode("utf-8")
                else:
                    rich_header_b64[key] = rh[key]
            rich_header_infos = self._get_rich_header_infos(rh)
        return json.dumps(rich_header_b64) if rich_header_b64 else None, "\n".join(
            item for item in rich_header_infos
        ) if rich_header_infos else None

    def _get_rich_header_infos(self, rh):
        """Parse Rich header information from the PE Rich Header dump"""
        self.log.debug(inspect.currentframe().f_code.co_name)

        rich_header_infos = []
        try:
            # Get list of @Comp.IDs and counts from Rich header
            # Elements in rich_fields at even indices are @Comp.IDs
            # Elements in rich_fields at odd indices are counts
            # example:                  'values': [8681481,
            #                             1,
            #                             9795593,
            #                             1]
            rich_fields = rh.get("values", [])
            if len(rich_fields) % 2 != 0:
                self.log.info(f"rich header extraction stopped for {self.hash.sha256}")
                return None

            comp_id = None
            for i in rich_fields:
                if rich_fields.index(i) % 2 == 0:
                    # even -> save value
                    comp_id = get_rich_idVersion(i)
                else:
                    # odd -> add to list
                    if comp_id:
                        rich_header_infos.append(f"{comp_id} count={i}")
                        comp_id = None
        except Exception as e:
            self.log.error(
                f"Extract rich header error {self.hash.sha256} Exception: {e}"
            )
        return rich_header_infos

    def _extract_version_info(self):
        """Extract the VS_VERSIONINFO field in a PE file

        Returns:
        vsinfo: a list of "key:value" strings from VS_INFORMATION content.
               None if no VS_INFORMATION content is present.
        """
        self.log.debug(inspect.currentframe().f_code.co_name)
        version_info = []
        try:
            if hasattr(self.pe, "VS_VERSIONINFO") and hasattr(self.pe, "FileInfo"):
                for finfo in self.pe.FileInfo:
                    for entry in finfo:
                        if hasattr(entry, "StringTable"):
                            for st_entry in entry.StringTable:
                                for key, str_entry in list(st_entry.entries.items()):
                                    version_info.append(f"{key.decode()}:{str_entry.decode()}")
        except Exception as e:
            self.log.error(
                f"Extract VersionInfo error {self.hash.sha256}  Exception: {e}"
            )
        return version_info if version_info else None

    def _extract_exports(self):
        """
        Returns:
        export_library_name: a str representing the name of the export library.
                            An empty string if no export library is present.
        exp_symbols_list: a list of function names exported as per DIRECTORY_ENTRY_EXPORT.symbols
                            An empty list if no export library is present.
        export_timestamp: a timestamp representing the time of the export.
        """
        self.log.debug(inspect.currentframe().f_code.co_name)
        export_symbols_list = []
        export_library_name = ""
        export_library_name_raw = ""
        export_timestamp = None
        try:
            if hasattr(self.pe, "DIRECTORY_ENTRY_EXPORT"):
                export_timestamp = self.pe.DIRECTORY_ENTRY_EXPORT.struct.TimeDateStamp
                export_directory = self.pe.DIRECTORY_ENTRY_EXPORT
                export_library_name_rva = export_directory.struct.Name
                try:
                    export_library_name = self.remove_non_utf8(
                        self.pe.get_string_at_rva(export_library_name_rva)
                    ).decode()
                    export_library_name_raw = self.pe.get_string_at_rva(
                        export_library_name_rva
                    ).__str__()
                except Exception as e:
                    self.log.error(
                        f"Error while getting export library name {self.pe.get_string_at_rva(export_library_name_rva)} for file: {self.hash.sha256} {e}"
                    )
                    export_library_name = "REDB_ERROR"
                if hasattr(self.pe.DIRECTORY_ENTRY_EXPORT, "symbols"):
                    for exp in self.pe.DIRECTORY_ENTRY_EXPORT.symbols:
                        export_symbols_list.append(
                            exp.name.decode() if exp.name else None
                        )
        except Exception as e:
            self.log.error(f"Extract exports error {self.hash.sha256} Exception: {e}")

        return (
            export_library_name if export_library_name else None,
            export_library_name_raw if export_library_name_raw else None,
            export_symbols_list if export_symbols_list else None,
            export_timestamp if export_timestamp else None,
        )

    def _extract_dbg_info(self):
        """Extract debug information from PE if present

        At the moment this function parses only the DEBUG_ENTRY Structure where Type
        field == 1, i.e. IMAGE_DEBUG_TYPE_CODEVIEW, where the pdb is eventually present.

        Returns:
        debug_entry: the dump of pe.DIRECTORY_ENTRY_DEBUG.struct only for the type 1
                    "IMAGE_DEBUG_TYPE_CODEVIEW". "None" if not present.
        debug_time: a int representing the epoch timestamp as present in the
                    DIRECTORY_ENTRY_DEBUG. "None" if not present.
        pdb_info: a str with the pdb path. "None" if not present.
        """
        self.log.debug(inspect.currentframe().f_code.co_name)
        dbg_timestamp = None
        dbg_timestamp_utc = ""
        dbg_pdb_info = ""
        dbg_pdb_info_raw = ""
        dbg_struct = ""
        try:
            if hasattr(self.pe, "DIRECTORY_ENTRY_DEBUG"):
                for debug_entry in self.pe.DIRECTORY_ENTRY_DEBUG:
                    if (
                        debug_entry.struct.Type
                        == pefile.DEBUG_TYPE["IMAGE_DEBUG_TYPE_CODEVIEW"]
                    ):
                        dbg_timestamp = debug_entry.struct.TimeDateStamp
                        dbg_timestamp_utc = datetime.fromtimestamp(
                            dbg_timestamp, timezone.utc
                        ).strftime("%Y-%m-%d %H:%M:%S")
                        # dbg_struct = base64.b64encode(debug_entry.entry).decode()
                        dbg_struct = debug_entry.entry.__str__()
                        if hasattr(debug_entry.entry, "PdbFileName"):
                            dbg_pdb_info = self.remove_non_utf8(
                                debug_entry.entry.PdbFileName.rstrip(b"\x00")
                            ).decode()
                            dbg_pdb_info_raw = debug_entry.entry.PdbFileName.rstrip(
                                b"\x00"
                            ).__str__()
                            # if dbg_pdb_info:
                            #     dbg_pdb_info = dbg_pdb_info.rstrip(b"\x00")
                            #     dbg_pdb_info = dbg_pdb_info.decode()

        except Exception as e:
            self.log.error(f"Extract DBG info error {self.hash.sha256} Exception: {e}")
        return (
            dbg_timestamp if dbg_timestamp else None,
            dbg_timestamp_utc if dbg_timestamp_utc else None,
            dbg_pdb_info if dbg_pdb_info else None,
            dbg_pdb_info_raw if dbg_pdb_info_raw else None,
            dbg_struct if dbg_struct else None,
        )

    def _extract_tls_info(self):
        """Check for the presence of Thread Local Storage and related extract
        callback addresses.

        Taken from the original version of PEScanner, as the python3 porting
        "ext_pescanner" does not have it.

        Returns:
        pe.DIRECTORY_ENTRY_TLS.struct: a pefile.Structure type containing the dump
                    of the TLS structure and content. "None" is no TLS is present.
        callbacks: a list containing the TLS callbacks addresse. Empty list if no
                    address is found, "None" is no TLS is present.
        """
        self.log.debug(inspect.currentframe().f_code.co_name)
        callbacks = []
        tls_dir_struct = ""
        try:
            if (
                hasattr(self.pe, "DIRECTORY_ENTRY_TLS")
                and self.pe.DIRECTORY_ENTRY_TLS
                and self.pe.DIRECTORY_ENTRY_TLS.struct
                and self.pe.DIRECTORY_ENTRY_TLS.struct.AddressOfCallBacks
            ):
                # tls_dir_struct = base64.b64encode(self.pe.DIRECTORY_ENTRY_TLS.struct).decode()
                tls_dir_struct = self.pe.DIRECTORY_ENTRY_TLS.struct.__str__()
                callback_array_rva = (
                    self.pe.DIRECTORY_ENTRY_TLS.struct.AddressOfCallBacks
                    - self.pe.OPTIONAL_HEADER.ImageBase
                )
                # Originally it was while True
                # todo while can't be used cause risky
                # what maximum range makes sense to add here?
                for idx in range(10000):
                    func = self.pe.get_dword_from_data(
                        self.pe.get_data(callback_array_rva + 4 * idx, 4), 0
                    )
                    if func == 0:
                        break
                    callbacks.append(func)
        except Exception as e:
            self.log.error(f"Extract TLS error {self.hash.sha256} Exception: {e}")
        return callbacks if callbacks else None, tls_dir_struct if tls_dir_struct else None

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

    def extract(self):
        try:
            self.log.debug(inspect.currentframe().f_code.co_name)
            (
                export_library_name,
                export_library_name_raw,
                export_symbols_list,
                export_timestamp,
            ) = self._extract_exports()
            (
                dbg_timestamp,
                dbg_timestamp_utc,
                dbg_pdb_info,
                dbg_pdb_info_raw,
                dbg_struct,
            ) = self._extract_dbg_info()
            tls_callbacks, tls_struct = self._extract_tls_info()
            rich_header_dump, rich_header_parsed = self._extract_rich_header()

            is_dotnet = self._check_dotnet()

            number_of_resources = 0
            number_of_imports = 0
            if hasattr(self.pe, "DIRECTORY_ENTRY_RESOURCE"):
                number_of_resources = len(self.pe.DIRECTORY_ENTRY_RESOURCE.entries)
            if hasattr(self.pe, "DIRECTORY_ENTRY_IMPORT"):
                number_of_imports = len(self.pe.DIRECTORY_ENTRY_IMPORT)

            self.pe_features = PE(
                type=self._extract_type(),
                magic=hex(self.pe.OPTIONAL_HEADER.Magic),
                entry_point=hex(self.pe.OPTIONAL_HEADER.AddressOfEntryPoint),
                major_linker_version=self.pe.OPTIONAL_HEADER.MajorLinkerVersion,
                minor_linker_version=self.pe.OPTIONAL_HEADER.MinorLinkerVersion,
                target_machine=self.pe.FILE_HEADER.Machine,
                architecture=self._extract_architecture(),
                compilation_time=self.pe.FILE_HEADER.TimeDateStamp,
                compilation_time_utc=datetime.fromtimestamp(
                    self.pe.FILE_HEADER.TimeDateStamp, timezone.utc
                ).strftime("%Y-%m-%d %H:%M:%S"),
                rich_header_dump=rich_header_dump,
                rich_header_parsed=rich_header_parsed,
                dos_header=self.pe.DOS_HEADER.__str__(),
                nt_header=self.pe.NT_HEADERS.__str__(),
                optional_header=self.pe.OPTIONAL_HEADER.__str__(),
                file_header=self.pe.FILE_HEADER.__str__(),
                version_info=self._extract_version_info(),
                export_library_name=export_library_name,
                export_library_name_raw=export_library_name_raw,
                export_symbols_list=export_symbols_list,
                export_timestamp=export_timestamp,
                is_dotnet=is_dotnet,
                dbg_timestamp=dbg_timestamp,
                dbg_timestamp_utc=dbg_timestamp_utc,
                dbg_pdb_info=dbg_pdb_info,
                dbg_pdb_info_raw=dbg_pdb_info_raw,
                dbg_struct=dbg_struct,
                tls_callbacks=tls_callbacks,
                tls_struct=tls_struct,
                is_signed=self._is_signed(),
                has_overlay=self._has_overlay(),
                number_of_sections=len(self.pe.sections),
                number_of_imports=number_of_imports,
                number_of_exports=(
                    len(export_symbols_list) if export_symbols_list else 0
                ),
                number_of_resources=number_of_resources,
            )
            return self.pe_features
        except Exception as e:
            self.log.error(f"Error extracting PE features {self.hash.sha256}: {e}")
            return None

    def prepare_export_data(self, exporter_type: str) -> Any:
        self.log.debug(inspect.currentframe().f_code.co_name)
        if exporter_type == "ElasticsearchExporter":
            return self.pe_features
        elif exporter_type == "ClickHouseExporter":
            try:
                pe_dump = self.pe.dump_dict()   
                # Convert PE headers directly to JSON
                dos_header_raw = json.dumps(pe_dump['DOS_HEADER'])
                nt_header_raw = json.dumps(pe_dump['NT_HEADERS'])
                optional_header_raw = json.dumps(pe_dump['OPTIONAL_HEADER'])
                file_header_raw = json.dumps(pe_dump['FILE_HEADER'])

                # Convert magic and entry_point from hex strings to integers
                magic_raw = int(self.pe_features.magic, 16) if isinstance(self.pe_features.magic, str) else self.pe_features.magic
                entry_point = int(self.pe_features.entry_point, 16) if isinstance(self.pe_features.entry_point, str) else self.pe_features.entry_point

                # Convert target_machine to string if it's an integer
                target_machine = str(self.pe_features.target_machine) if isinstance(self.pe_features.target_machine, int) else self.pe_features.target_machine

                # debug struct
                dbg_struct_raw = {}
                if self.pe_features.dbg_struct:
                    if hasattr(self.pe, "DIRECTORY_ENTRY_DEBUG"):
                        for debug_entry in self.pe.DIRECTORY_ENTRY_DEBUG:
                            if (
                                debug_entry.struct.Type
                                == pefile.DEBUG_TYPE["IMAGE_DEBUG_TYPE_CODEVIEW"]
                            ):
                                dbg_struct_raw = json.dumps(debug_entry.struct.__dict__)

                # tls struct
                tls_struct_raw = {}
                if self.pe_features.tls_struct:
                    if (
                        hasattr(self.pe, "DIRECTORY_ENTRY_TLS")
                        and self.pe.DIRECTORY_ENTRY_TLS
                        and self.pe.DIRECTORY_ENTRY_TLS.struct
                        and self.pe.DIRECTORY_ENTRY_TLS.struct.AddressOfCallBacks
                    ):
                        tls_struct_raw = json.dumps(self.pe.DIRECTORY_ENTRY_TLS.struct.__dict__)

                # Ensure arrays are properly initialized
                rich_header_parsed = []
                if self.pe_features.rich_header_parsed:
                    if isinstance(self.pe_features.rich_header_parsed, str):
                        rich_header_parsed = [x for x in self.pe_features.rich_header_parsed.split("\n") if x]
                    elif isinstance(self.pe_features.rich_header_parsed, list):
                        rich_header_parsed = self.pe_features.rich_header_parsed

                version_info = self.pe_features.version_info if self.pe_features.version_info else []
                version_info_raw = {}
                if version_info:
                    for finfo in self.pe.FileInfo:
                        for entry in finfo:
                            if hasattr(entry, "StringTable"):
                                for st_entry in entry.StringTable:
                                    for key, str_entry in list(st_entry.entries.items()):
                                        version_info_raw[key.decode()] = str_entry.decode()
                version_info_raw = json.dumps(version_info_raw) if version_info_raw else "{}"

                export_symbols_list = self.pe_features.export_symbols_list if self.pe_features.export_symbols_list else []
                tls_callbacks = self.pe_features.tls_callbacks if self.pe_features.tls_callbacks else []

                # Prepare data array
                data = [[
                    self.sha256,
                    self.md5,
                    self.sha1,
                    self.pe_features.dos_header,
                    dos_header_raw,
                    self.pe_features.nt_header,
                    nt_header_raw,
                    self.pe_features.optional_header,
                    optional_header_raw,
                    self.pe_features.file_header,
                    file_header_raw,
                    magic_raw,
                    entry_point,
                    self.pe_features.major_linker_version,
                    self.pe_features.minor_linker_version,
                    target_machine,
                    self.pe_features.architecture,
                    self.pe_features.compilation_time,
                    1 if self.pe_features.is_dotnet else 0,
                    1 if self.pe_features.is_signed else 0,
                    1 if self.pe_features.has_overlay else 0,
                    self.pe_features.number_of_sections,
                    self.pe_features.number_of_imports,
                    self.pe_features.number_of_exports,
                    self.pe_features.number_of_resources,
                    self.pe_features.type,
                    self.pe_features.dbg_struct,
                    dbg_struct_raw,
                    self.pe_features.dbg_timestamp,
                    self.pe_features.dbg_pdb_info,
                    self.pe_features.dbg_pdb_info_raw,
                    self.pe_features.tls_struct,
                    tls_struct_raw,
                    self.pe_features.export_timestamp,
                    self.pe_features.export_library_name,
                    self.pe_features.export_library_name_raw,
                    self.pe_features.rich_header_dump if self.pe_features.rich_header_dump else "{}",
                    rich_header_parsed,
                    version_info,
                    version_info_raw,
                    export_symbols_list,
                    tls_callbacks,
                    datetime.now(timezone.utc)
                ]]

                column_names = [
                    'sha256', 'md5', 'sha1',
                    'dos_header', 'dos_header_raw', 'nt_header', 'nt_header_raw', 'optional_header', 'optional_header_raw', 'file_header', 'file_header_raw',
                    'magic_raw', 'entry_point',
                    'major_linker_version', 'minor_linker_version',
                    'target_machine', 'architecture',
                    'compilation_time',
                    'is_dotnet', 'is_signed', 'has_overlay',
                    'number_of_sections', 'number_of_imports', 'number_of_exports', 'number_of_resources',
                    'type',
                    'dbg_struct', 'dbg_struct_raw', 'dbg_timestamp', 'dbg_pdb_info', 'dbg_pdb_info_raw',
                    'tls_struct', 'tls_struct_raw', 'export_timestamp', 'export_library_name', 'export_library_name_raw',
                    'rich_header_dump', 'rich_header_parsed', 'version_info', 'version_info_raw',
                    'export_symbols_list', 'tls_callbacks',
                    'analysis_date'
                ]

                column_type_names = [
                    'FixedString(64)', 'FixedString(32)', 'FixedString(40)',
                    'String', 'JSON', 'String', 'JSON', 'String', 'JSON', 'String', 'JSON',
                    'UInt16', 'UInt32',
                    'UInt8', 'UInt8',
                    'LowCardinality(String)', 'Enum8(\'x86\' = 1, \'Itanium\' = 2, \'x64\' = 3)',
                    'UInt64',
                    'UInt8', 'UInt8', 'UInt8',
                    'UInt16', 'UInt16', 'UInt16', 'UInt16',
                    'Enum8(\'DLL\' = 1, \'EXE\' = 2, \'DRIVER\' = 3)',
                    'Nullable(String)', 'JSON', 'Nullable(UInt64)', 'Nullable(String)', 'Nullable(String)',
                    'Nullable(String)', 'JSON', 'Nullable(UInt64)', 'Nullable(String)', 'Nullable(String)',
                    'JSON', 'Array(Nullable(String))', 'Array(Nullable(String))', 'JSON',
                    'Array(Nullable(String))', 'Array(Nullable(UInt64))',
                    'DateTime64(3, \'UTC\')'
                ]

                if not data:
                    return None

                return (data, column_names, column_type_names)
            except Exception as e:
                self.log.error(f"Error preparing export data: {e}")
                raise

    def get_clickhouse_table(self) -> str:
        return "redb_pe_features"