Manish Nagireddy

27 papers A* 4Misc 1Journal 17Unranked 5
YearRankTypeTitle / Venue / Authors
2025 J jnl
CoRR
Yue Huang, Hang Hua, Yujun Zhou, Pengcheng Jing, Manish Nagireddy, Inkit Padhi, Greta Dolcetti, Zhangchen Xu, Subhajit Chaudhury, Ambrish Rawat, Liubov Nedoshivina, Pin-Yu Chen, Prasanna Sattigeri, Xiangliang Zhang
2025 Misc conf
ICASSP
Pierre L. Dognin, Jesus Rios, Ronny Luss, Prasanna Sattigeri, Miao Liu, Inkit Padhi, Matthew Riemer, Manish Nagireddy, Kush R. Varshney, Djallel Bouneffouf
2025 conf
NAACL (Industry Track)
Inkit Padhi, Manish Nagireddy, Giandomenico Cornacchia, Subhajit Chaudhury, Tejaswini Pedapati, Pierre L. Dognin, Keerthiram Murugesan, Erik Miehling, Martín Santillán Cooper, Kieran Fraser, Giulio Zizzo, Muhammad Zaid Hameed, Mark Purcell, Michael Desmond, Qian Pan, Inge Vejsbjerg, Elizabeth M. Daly, Michael Hind, Werner Geyer, Ambrish Rawat, Kush R. Varshney, Prasanna Sattigeri
2025 conf
ACL (1)
Lucas Monteiro Paes, Dennis Wei, Hyo Jin Do, Hendrik Strobelt, Ronny Luss, Amit Dhurandhar, Manish Nagireddy, Karthikeyan Natesan Ramamurthy, Prasanna Sattigeri, Werner Geyer, Soumya Ghosh
2025 A* conf
ICLR
Bruce W. Lee, Inkit Padhi, Karthikeyan Natesan Ramamurthy, Erik Miehling, Pierre L. Dognin, Manish Nagireddy, Amit Dhurandhar
2025 J jnl
Trans. Mach. Learn. Res.
Hussein Mozannar, Valerie Chen, Mohammed Alsobay, Subhro Das, Sebastian Zhao, Dennis Wei, Manish Nagireddy, Prasanna Sattigeri, Ameet Talwalkar, David A. Sontag
2024 J jnl
CoRR
Swapnaja Achintalwar, Ioana Baldini, Djallel Bouneffouf, Joan Byamugisha, Maria Chang, Pierre L. Dognin, Eitan Farchi, Ndivhuwo Makondo, Aleksandra Mojsilovic, Manish Nagireddy, Karthikeyan Natesan Ramamurthy, Inkit Padhi, Orna Raz, Jesus Rios, Prasanna Sattigeri, Moninder Singh, Siphiwe Thwala, Rosario A. Uceda-Sosa, Kush R. Varshney
2024 J jnl
IEEE Internet Comput.
Swapnaja Achintalwar, Ioana Baldini, Djallel Bouneffouf, Joan Byamugisha, Maria Chang, Pierre L. Dognin, Eitan Farchi, Ndivhuwo Makondo, Aleksandra Mojsilovic, Manish Nagireddy, Karthikeyan Natesan Ramamurthy, Inkit Padhi, Orna Raz, Jesus Rios, Prasanna Sattigeri, Moninder Singh, Siphiwe Thwala, Rosario A. Uceda-Sosa, Kush R. Varshney
2024 A* conf
IJCAI
Inkit Padhi, Pierre L. Dognin, Jesus Rios, Ronny Luss, Swapnaja Achintalwar, Matthew Riemer, Miao Liu, Prasanna Sattigeri, Manish Nagireddy, Kush R. Varshney, Djallel Bouneffouf
2024 J jnl
CoRR
Pierre L. Dognin, Jesus Rios, Ronny Luss, Inkit Padhi, Matthew D. Riemer, Miao Liu, Prasanna Sattigeri, Manish Nagireddy, Kush R. Varshney, Djallel Bouneffouf
2024 A* conf
KDD
Manish Nagireddy, Bernat Guillen Pegueroles, Ioana Baldini
2024 J jnl
CoRR
Swapnaja Achintalwar, Adriana Alvarado Garcia, Ateret Anaby-Tavor, Ioana Baldini, Sara E. Berger, Bishwaranjan Bhattacharjee, Djallel Bouneffouf, Subhajit Chaudhury, Pin-Yu Chen, Lamogha Chiazor, Elizabeth M. Daly, Rogério Abreu de Paula, Pierre L. Dognin, Eitan Farchi, Soumya Ghosh, Michael Hind, Raya Horesh, George Kour, Ja Young Lee, Erik Miehling, Keerthiram Murugesan, Manish Nagireddy, Inkit Padhi, David Piorkowski, Ambrish Rawat, Orna Raz, Prasanna Sattigeri, Hendrik Strobelt, Sarathkrishna Swaminathan, Christoph Tillmann, Aashka Trivedi, Kush R. Varshney, Dennis Wei, Shalisha Witherspoon, Marcel Zalmanovici
2024 J jnl
CoRR
Inkit Padhi, Manish Nagireddy, Giandomenico Cornacchia, Subhajit Chaudhury, Tejaswini Pedapati, Pierre L. Dognin, Keerthiram Murugesan, Erik Miehling, Martin Santillan Cooper, Kieran Fraser, Giulio Zizzo, Muhammad Zaid Hameed, Mark Purcell, Michael Desmond, Qian Pan, Zahra Ashktorab, Inge Vejsbjerg, Elizabeth M. Daly, Michael Hind, Werner Geyer, Ambrish Rawat, Kush R. Varshney, Prasanna Sattigeri
2024 conf
EMNLP (Findings)
Erik Miehling, Manish Nagireddy, Prasanna Sattigeri, Elizabeth Daly, David Piorkowski, John T. Richards
2024 J jnl
CoRR
Erik Miehling, Manish Nagireddy, Prasanna Sattigeri, Elizabeth M. Daly, David Piorkowski, John T. Richards
2024 J jnl
CoRR
Lucas Monteiro Paes, Dennis Wei, Hyo Jin Do, Hendrik Strobelt, Ronny Luss, Amit Dhurandhar, Manish Nagireddy, Karthikeyan Natesan Ramamurthy, Prasanna Sattigeri, Werner Geyer, Soumya Ghosh
2024 J jnl
CoRR
Bruce W. Lee, Inkit Padhi, Karthikeyan Natesan Ramamurthy, Erik Miehling, Pierre L. Dognin, Manish Nagireddy, Amit Dhurandhar
2024 A* conf
AAAI
Manish Nagireddy, Lamogha Chiazor, Moninder Singh, Ioana Baldini
2024 J jnl
CoRR
Hussein Mozannar, Valerie Chen, Mohammed Alsobay, Subhro Das, Sebastian Zhao, Dennis Wei, Manish Nagireddy, Prasanna Sattigeri, Ameet Talwalkar, David A. Sontag
2024 conf
EMNLP (Industry Track)
Inkit Padhi, Karthikeyan Natesan Ramamurthy, Prasanna Sattigeri, Manish Nagireddy, Pierre L. Dognin, Kush R. Varshney
2024 J jnl
CoRR
Inkit Padhi, Karthikeyan Natesan Ramamurthy, Prasanna Sattigeri, Manish Nagireddy, Pierre L. Dognin, Kush R. Varshney
2024 J jnl
CoRR
Manish Nagireddy, Inkit Padhi, Soumya Ghosh, Prasanna Sattigeri
2023 J jnl
CoRR
Manish Nagireddy, Moninder Singh, Samuel C. Hoffman, Evaline Ju, Karthikeyan Natesan Ramamurthy, Kush R. Varshney
2023 J jnl
CoRR
Manish Nagireddy, Lamogha Chiazor, Moninder Singh, Ioana Baldini
2022 J jnl
CoRR
Nil-Jana Akpinar, Manish Nagireddy, Logan Stapleton, Hao Fei Cheng, Haiyi Zhu, Zhiwei Steven Wu, Hoda Heidari
2022 conf
FAccT
Wesley Hanwen Deng, Manish Nagireddy, Michelle Seng Ah Lee, Jatinder Singh, Zhiwei Steven Wu, Kenneth Holstein, Haiyi Zhu
2022 J jnl
CoRR
Wesley Hanwen Deng, Manish Nagireddy, Michelle Seng Ah Lee, Jatinder Singh, Zhiwei Steven Wu, Kenneth Holstein, Haiyi Zhu
tests/unit/test_apk_cfg_features.py
← Index tests/unit/test_apk_cfg_features.py python
"""
Unit tests for APK CFG feature parity with the Binary Ninja pipeline.

Tests advanced CFG features computed from smali (topology hash, MD-index,
WL-MinHash, prime product, packed adjacency, block features export) and
the slimmed similarity table.

All tests use mocked smali input — no JADX/apktool/Java required.
"""
import pytest

from redb.extractors.decompiler.apk.smali_cfg import (
    SmaliCFGMetrics,
    compute_cfg_metrics,
)
from redb.extractors.decompiler.apk.method_extractor import (
    SMALI_OP_PRIMES,
    compute_prime_product_smali,
    count_call_instructions,
)

pytestmark = [pytest.mark.unit, pytest.mark.apk, pytest.mark.decompile]


# ===================================================================
# Sample smali bodies for testing
# ===================================================================

# Simple linear method: 3 instructions, 1 block
SMALI_LINEAR = """\
    const/4 v0, 0x0
    invoke-virtual {p0, v0}, Lcom/Foo;->bar(I)V
    return-void
"""

# Diamond CFG: if-else with two paths merging
SMALI_DIAMOND = """\
    const/4 v0, 0x1
    if-eqz v0, :cond_0
    invoke-virtual {p0}, Lcom/Foo;->pathA()V
    goto :goto_0
    :cond_0
    invoke-virtual {p0}, Lcom/Foo;->pathB()V
    :goto_0
    return-void
"""

# Loop: a back edge from goto to a label before it
SMALI_LOOP = """\
    const/4 v0, 0x0
    :loop_start
    add-int/lit8 v0, v0, 0x1
    if-lt v0, v1, :loop_start
    return-void
"""

# Multi-block with several invoke calls
SMALI_MULTI_CALL = """\
    invoke-virtual {p0}, Lcom/Foo;->a()V
    invoke-static {v0}, Lcom/Bar;->b(I)V
    invoke-direct {p0, v1}, Lcom/Baz;-><init>(I)V
    const/4 v0, 0x0
    return-void
"""


# ===================================================================
# TestSmaliCFGMetricsAdvanced — new fields in SmaliCFGMetrics
# ===================================================================

class TestSmaliCFGMetricsAdvanced:
    """Test that advanced CFG fields are populated by compute_cfg_metrics."""

    def test_topology_hash_is_bytes(self):
        m = compute_cfg_metrics(SMALI_DIAMOND)
        assert isinstance(m.cfg_topology_hash, bytes)
        assert len(m.cfg_topology_hash) == 16

    def test_topology_hash_nonzero_for_multiblock(self):
        m = compute_cfg_metrics(SMALI_DIAMOND)
        assert m.cfg_topology_hash != b'\x00' * 16

    def test_topology_hash_zero_for_empty(self):
        m = compute_cfg_metrics("")
        assert m.cfg_topology_hash == b'\x00' * 16

    def test_identical_cfg_same_topology_hash(self):
        """Two smali methods with the same CFG structure produce the same hash."""
        smali_a = """\
    const/4 v0, 0x1
    if-eqz v0, :cond_0
    const/4 v1, 0x2
    goto :goto_0
    :cond_0
    const/4 v1, 0x3
    :goto_0
    return-void
"""
        smali_b = """\
    const/4 v5, 0x9
    if-nez v5, :cond_0
    add-int v6, v5, v5
    goto :goto_0
    :cond_0
    sub-int v6, v5, v5
    :goto_0
    return-void
"""
        m_a = compute_cfg_metrics(smali_a)
        m_b = compute_cfg_metrics(smali_b)
        assert m_a.cfg_topology_hash == m_b.cfg_topology_hash

    def test_different_cfg_different_topology_hash(self):
        m_linear = compute_cfg_metrics(SMALI_LINEAR)
        m_diamond = compute_cfg_metrics(SMALI_DIAMOND)
        assert m_linear.cfg_topology_hash != m_diamond.cfg_topology_hash

    def test_md_index_topdown_nonzero(self):
        m = compute_cfg_metrics(SMALI_DIAMOND)
        assert m.md_index_topdown != 0

    def test_md_index_bottomup_nonzero(self):
        m = compute_cfg_metrics(SMALI_DIAMOND)
        assert m.md_index_bottomup != 0

    def test_md_index_zero_for_empty(self):
        m = compute_cfg_metrics("")
        assert m.md_index_topdown == 0
        assert m.md_index_bottomup == 0

    def test_wl_minhash_length(self):
        m = compute_cfg_metrics(SMALI_DIAMOND)
        assert len(m.wl_minhash) == 128

    def test_wl_minhash_sentinel_for_empty(self):
        m = compute_cfg_metrics("")
        assert m.wl_minhash == [255] * 128

    def test_wl_minhash_values_are_uint8(self):
        m = compute_cfg_metrics(SMALI_DIAMOND)
        for val in m.wl_minhash:
            assert 0 <= val <= 255

    def test_cfg_adjacency_nonempty_for_multiblock(self):
        m = compute_cfg_metrics(SMALI_DIAMOND)
        assert len(m.cfg_adjacency) > 0

    def test_cfg_adjacency_empty_for_empty(self):
        m = compute_cfg_metrics("")
        assert m.cfg_adjacency == []

    def test_cfg_adjacency_packed_format(self):
        """Each entry should be (src << 16) | tgt."""
        m = compute_cfg_metrics(SMALI_DIAMOND)
        for edge in m.cfg_adjacency:
            assert isinstance(edge, int)
            src = edge >> 16
            tgt = edge & 0xFFFF
            assert src < m.block_count
            assert tgt < m.block_count

    def test_block_features_exported(self):
        m = compute_cfg_metrics(SMALI_DIAMOND)
        assert len(m.block_features) == m.block_count
        for bf in m.block_features:
            assert len(bf) == 8  # 7 categories + successor_count

    def test_cfg_feature_tlsh_none_for_small(self):
        """Small methods produce None (< 50 bytes of feature data)."""
        m = compute_cfg_metrics(SMALI_LINEAR)
        # Single-block linear method — too small for TLSH
        assert m.cfg_feature_tlsh is None

    def test_loop_count_with_back_edge(self):
        m = compute_cfg_metrics(SMALI_LOOP)
        assert m.loop_count >= 1


# ===================================================================
# TestPrimeProductSmali
# ===================================================================

class TestPrimeProductSmali:

    def test_empty_returns_zero(self):
        assert compute_prime_product_smali("") == 0

    def test_none_returns_zero(self):
        assert compute_prime_product_smali(None) == 0

    def test_known_computation(self):
        """Single invoke instruction → CALL category → prime 17."""
        smali = "    invoke-virtual {p0}, Lcom/Foo;->bar()V"
        assert compute_prime_product_smali(smali) == 17

    def test_two_instructions(self):
        """const (CONST→2) * invoke (CALL→17) = 34."""
        smali = """\
    const/4 v0, 0x0
    invoke-virtual {p0, v0}, Lcom/Foo;->bar(I)V
"""
        assert compute_prime_product_smali(smali) == 2 * 17

    def test_skips_directives_and_labels(self):
        """Directives, labels, and comments should be skipped."""
        smali = """\
.registers 2
.line 10
    :label
    # comment
    const/4 v0, 0x0
"""
        # Only const/4 → CONST → prime 2
        assert compute_prime_product_smali(smali) == 2

    def test_position_independent(self):
        """Reordering instructions produces the same product (multiplication is commutative)."""
        smali_a = """\
    add-int v0, v1, v2
    invoke-virtual {p0}, Lcom/Foo;->bar()V
"""
        smali_b = """\
    invoke-virtual {p0}, Lcom/Foo;->bar()V
    add-int v0, v1, v2
"""
        assert compute_prime_product_smali(smali_a) == compute_prime_product_smali(smali_b)

    def test_all_categories_have_primes(self):
        """Every category in SMALI_OP_PRIMES is a positive integer."""
        for cat, prime in SMALI_OP_PRIMES.items():
            assert isinstance(prime, int)
            assert prime >= 1

    def test_mod_2_64(self):
        """Result should be mod 2^64 (doesn't overflow)."""
        # A very long method — product stays within uint64
        smali = "\n".join(["    mul-int v0, v1, v2"] * 1000)
        result = compute_prime_product_smali(smali)
        assert 0 <= result < 2**64


# ===================================================================
# TestCallCount
# ===================================================================

class TestCallCount:

    def test_empty(self):
        assert count_call_instructions("") == 0

    def test_none(self):
        assert count_call_instructions(None) == 0

    def test_single_invoke(self):
        smali = "    invoke-virtual {p0}, Lcom/Foo;->bar()V"
        assert count_call_instructions(smali) == 1

    def test_multiple_invoke_types(self):
        assert count_call_instructions(SMALI_MULTI_CALL) == 3

    def test_non_invoke_not_counted(self):
        smali = """\
    const/4 v0, 0x0
    add-int v1, v0, v0
    return-void
"""
        assert count_call_instructions(smali) == 0


# ===================================================================
# TestSimilarityTableSlimmed
# ===================================================================

class TestSimilarityTableSlimmed:
    """Verify the similarity table no longer contains moved fields."""

    def test_export_similarity_no_block_count(self):
        """block_count should not be in similarity export columns."""
        from redb.extractors.decompiler.DecompileAPK import DecompileAPK
        from unittest.mock import MagicMock

        extractor = DecompileAPK.__new__(DecompileAPK)
        extractor.log = MagicMock()
        extractor.sha256 = "a" * 64
        extractor.sha1 = "b" * 40
        extractor.md5 = "c" * 32
        extractor.analysis_results = {
            "sha256": "a" * 64,
            "sha1": "b" * 40,
            "md5": "c" * 32,
            "similarity_metrics": [{
                "smali_method_hash": "d" * 64,
                "cyclomatic_complexity": 3,
                "ssdeep_smali": None,
                "tlsh_smali": None,
                "minhash": [1, 2, 3],
            }],
        }

        export = extractor.prepare_export_data("ClickHouseExporter")
        sim_cols = export["method_similarity_metrics"]["column_names"]
        assert "block_count" not in sim_cols
        assert "edge_count" not in sim_cols
        assert "loop_count" not in sim_cols
        assert "max_depth" not in sim_cols
        assert "max_fan_out" not in sim_cols
        # These should still be there
        assert "smali_method_hash" in sim_cols
        assert "ssdeep_smali" in sim_cols
        assert "minhash" in sim_cols
        # cyclomatic_complexity moved to cfg table
        assert "cyclomatic_complexity" not in sim_cols


# ===================================================================
# TestCFGTableExport
# ===================================================================

class TestCFGTableExport:
    """Verify the new CFG table is exported correctly."""

    def test_export_cfg_table_present(self):
        from redb.extractors.decompiler.DecompileAPK import DecompileAPK
        from unittest.mock import MagicMock

        extractor = DecompileAPK.__new__(DecompileAPK)
        extractor.log = MagicMock()
        extractor.analysis_results = {
            "sha256": "a" * 64,
            "sha1": "b" * 40,
            "md5": "c" * 32,
            "cfg": [{
                "smali_method_hash": "d" * 64,
                "cfg_topology_hash": b'\x01' * 16,
                "block_count": 4,
                "edge_count": 5,
                "instructions_count": 20,
                "call_count": 3,
                "cyclomatic_complexity": 3,
                "loop_count": 1,
                "max_depth": 2,
                "max_fan_out": 2,
                "md_index_topdown": 12345,
                "md_index_bottomup": 67890,
                "prime_product_smali": 999,
                "cfg_feature_tlsh": None,
                "wl_minhash": [0] * 128,
                "bb_features": [[10, 1, 0, 2, 1, 0, 0, 2]] * 4,
                "cfg_adjacency": [0x00010000, 0x00010002],
            }],
        }

        export = extractor.prepare_export_data("ClickHouseExporter")
        assert "cfg_methods" in export
        cfg_export = export["cfg_methods"]
        assert cfg_export["table"] == "code_apk_cfg_methods"
        assert len(cfg_export["data"]) == 1
        assert "cfg_topology_hash" in cfg_export["column_names"]
        assert "prime_product_smali" in cfg_export["column_names"]
        assert "wl_minhash" in cfg_export["column_names"]
        assert "bb_features" in cfg_export["column_names"]
        assert "cfg_adjacency" in cfg_export["column_names"]
        # Verify column count matches data width
        assert len(cfg_export["column_names"]) == len(cfg_export["data"][0])
        assert len(cfg_export["column_type_names"]) == len(cfg_export["data"][0])