From 5c6a298ab8bf1ea2e56c2dd339822359ca1c95fd Mon Sep 17 00:00:00 2001
From: Josh Feather <142008135+josh-feather@users.noreply.github.com>
Date: Wed, 14 Feb 2024 11:23:45 +0000
Subject: [PATCH 1/6] Fix order of precedence in Config enumeration

---
 lib/cuckoo/common/config.py | 13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)

diff --git a/lib/cuckoo/common/config.py b/lib/cuckoo/common/config.py
index 0984ead4e94..af3b840b90b 100644
--- a/lib/cuckoo/common/config.py
+++ b/lib/cuckoo/common/config.py
@@ -107,12 +107,13 @@ def __init__(self, fname_base: str = "cuckoo"):
         self._read_files(files)
 
     def _get_files_to_read(self, fname_base):
-        files = [
-            os.path.join(CUCKOO_ROOT, "conf", f"{fname_base}.conf.default"),
-            os.path.join(CUCKOO_ROOT, "conf", f"{fname_base}.conf"),
-            os.path.join(CUSTOM_CONF_DIR, f"{fname_base}.conf"),
-        ]
-        files.extend(sorted(glob.glob(os.path.join(CUCKOO_ROOT, "conf", f"{fname_base}.conf.d", "*.conf"))))
+        # Allows test workflows to ignore custom root configs
+        include_root_configs = "CAPE_DISABLE_ROOT_CONFIGS" not in os.environ
+        files = [os.path.join(CUCKOO_ROOT, "conf", f"{fname_base}.conf.default")]
+        if include_root_configs:
+            files.append(os.path.join(CUCKOO_ROOT, "conf", f"{fname_base}.conf"))
+            files.extend(sorted(glob.glob(os.path.join(CUCKOO_ROOT, "conf", f"{fname_base}.conf.d", "*.conf"))))
+        files.append(os.path.join(CUSTOM_CONF_DIR, f"{fname_base}.conf"))
         files.extend(sorted(glob.glob(os.path.join(CUSTOM_CONF_DIR, f"{fname_base}.conf.d", "*.conf"))))
         return files
 

From ccdc8fa15d9f34dc709861c4f154478621e72b32 Mon Sep 17 00:00:00 2001
From: Josh Feather <142008135+josh-feather@users.noreply.github.com>
Date: Wed, 14 Feb 2024 11:26:55 +0000
Subject: [PATCH 2/6] Flush Error table between database tests

---
 tests/test_database.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/tests/test_database.py b/tests/test_database.py
index a14d0c47f65..d44b67eb7b9 100644
--- a/tests/test_database.py
+++ b/tests/test_database.py
@@ -14,7 +14,7 @@
 from lib.cuckoo.common.exceptions import CuckooOperationalError
 from lib.cuckoo.common.path_utils import path_mkdir
 from lib.cuckoo.common.utils import store_temp_file
-from lib.cuckoo.core.database import Database, Machine, Tag, Task, machines_tags, tasks_tags
+from lib.cuckoo.core.database import Database, Error, Machine, Tag, Task, machines_tags, tasks_tags
 
 
 @pytest.fixture(autouse=True)
@@ -57,11 +57,13 @@ def setup_method(self, method):
             stmt3 = delete(machines_tags)
             stmt4 = delete(tasks_tags)
             stmt5 = delete(Tag)
+            stmt6 = delete(Error)
             self.session.execute(stmt)
             self.session.execute(stmt2)
             self.session.execute(stmt3)
             self.session.execute(stmt4)
             self.session.execute(stmt5)
+            self.session.execute(stmt6)
             self.session.commit()
 
     def teardown_method(self):

From 2425f053a3e2f372174ac7613ec6c06feeeb666b Mon Sep 17 00:00:00 2001
From: Josh Feather <142008135+josh-feather@users.noreply.github.com>
Date: Wed, 14 Feb 2024 11:29:31 +0000
Subject: [PATCH 3/6] Make tests only use the '*.conf.default' files unless
 necessary

---
 tests/conftest.py    |  14 ++++++
 tests/test_config.py | 103 +++++++++++++------------------------------
 2 files changed, 44 insertions(+), 73 deletions(-)

diff --git a/tests/conftest.py b/tests/conftest.py
index ff0c06f3d42..bcccc2f9be9 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -1,8 +1,22 @@
+import pathlib
 import pytest
+from lib.cuckoo.common.config import ConfigMeta
 
+import lib.cuckoo.common.config
 import lib.cuckoo.core.database
 
 
 @pytest.fixture
 def tmp_cuckoo_root(monkeypatch, tmp_path):
     monkeypatch.setattr(lib.cuckoo.core.database, "CUCKOO_ROOT", str(tmp_path))
+    yield tmp_path
+
+
+@pytest.fixture(autouse=True)
+def custom_conf_path(request, monkeypatch, tmp_cuckoo_root):
+    ConfigMeta.reset()
+    monkeypatch.setenv("CAPE_DISABLE_ROOT_CONFIGS", "1")
+    path: pathlib.Path = tmp_cuckoo_root / "custom" / "conf"
+    path.mkdir(mode=0o755, parents=True)
+    monkeypatch.setattr(lib.cuckoo.common.config, "CUSTOM_CONF_DIR", str(path))
+    yield path
\ No newline at end of file
diff --git a/tests/test_config.py b/tests/test_config.py
index 19fd06b1081..d4a9a48aafc 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -7,8 +7,7 @@
 
 import pytest
 
-import lib.cuckoo.common.config
-from lib.cuckoo.common.config import AnalysisConfig, Config, ConfigMeta
+from lib.cuckoo.common.config import AnalysisConfig, Config
 from lib.cuckoo.common.exceptions import CuckooOperationalError
 from lib.cuckoo.common.path_utils import path_write_file
 
@@ -31,11 +30,6 @@ def analysis_config(tmp_path):
     yield AnalysisConfig(path)
 
 
-@pytest.fixture(autouse=True)
-def reset():
-    ConfigMeta.reset()
-
-
 class TestAnalysisConfig:
     def test_get_option_exist(self, analysis_config):
         """Fetch an option of each type from default config file."""
@@ -67,60 +61,34 @@ def write_config(path, content):
     path_write_file(path, textwrap.dedent(content), mode="text")
 
 
-@pytest.fixture
-def custom_conf_dir(monkeypatch, tmp_path):
-    custom_dir = tmp_path / "custom"
-    custom_dir.mkdir()
-    monkeypatch.setattr(lib.cuckoo.common.config, "CUSTOM_CONF_DIR", str(custom_dir))
-    yield custom_dir
-
-
-@pytest.fixture
-def default_config(monkeypatch, tmp_path):
-    default_dir = tmp_path / "default"
-    default_dir.mkdir()
-    default_conf = default_dir / "conf" / "cuckoo.conf"
-    default_conf.parent.mkdir()
-    write_config(
-        default_conf,
-        """
-        [cuckoo]
-        debug = false
-        analysis_timeout = 120
-        """,
-    )
-    monkeypatch.setattr(lib.cuckoo.common.config, "CUCKOO_ROOT", str(default_dir))
-    yield default_conf
-
-
 class TestConfig:
-    def test_option_override(self, custom_conf_dir, default_config):
+    def test_option_override(self, custom_conf_path):
         """Fetch an option of each type from default config file."""
-        custom_conf = custom_conf_dir / "cuckoo.conf"
+        custom_conf = custom_conf_path / "cuckoo.conf"
         write_config(
             custom_conf,
             """
             [cuckoo]
-            debug = true
+            machinery_screenshots = true
             """,
         )
         config = Config("cuckoo")
 
         # This was overridden in the custom config.
-        assert config.get("cuckoo")["debug"] is True
+        assert config.get("cuckoo")["machinery_screenshots"] is True
         # This was inherited from the default config.
-        assert config.get("cuckoo")["analysis_timeout"] == 120
+        assert config.get("cuckoo")["max_analysis_count"] == 0
 
-    def test_nonexistent_custom_file(self, custom_conf_dir, default_config):
+    def test_nonexistent_custom_file(self):
         """Verify that there are no problems when the file to be processed does not
         exist in the custom config dir.
         """
         config = Config("cuckoo")
-        assert config.get("cuckoo")["debug"] is False
-        assert config.get("cuckoo")["analysis_timeout"] == 120
+        assert config.get("cuckoo")["machinery_screenshots"] is False
+        assert config.get("cuckoo")["max_analysis_count"] == 0
 
-    def test_subdirs(self, custom_conf_dir):
-        api_conf = custom_conf_dir / "api.conf"
+    def test_subdirs(self, custom_conf_path):
+        api_conf = custom_conf_path / "api.conf"
         write_config(
             api_conf,
             """
@@ -128,7 +96,7 @@ def test_subdirs(self, custom_conf_dir):
             ratelimit = no
             """,
         )
-        api_conf_d_dir = custom_conf_dir / "api.conf.d"
+        api_conf_d_dir = custom_conf_path / "api.conf.d"
         api_conf_d_dir.mkdir()
         host_specific_conf = api_conf_d_dir / "01_host_specific.conf"
         url = "https://somehost.example.com"
@@ -149,7 +117,7 @@ def test_subdirs(self, custom_conf_dir):
         # api.conf.d/01_host_specific.conf.
         assert config.get("api")["url"] == url
 
-    def test_singleton_configs(self, default_config):
+    def test_singleton_configs(self):
         """Verify that Config objects that were passed the same "file_name" argument
         are reused.
         """
@@ -157,54 +125,43 @@ def test_singleton_configs(self, default_config):
         config2 = Config("cuckoo")
         assert config1 is config2
 
-    def test_environment_interpolation(self, default_config, custom_conf_dir, monkeypatch):
+    def test_environment_interpolation(self, custom_conf_path, monkeypatch):
         """Verify that environment variables are able to be referenced in config
         files.
         """
-        default_dir = default_config.parent
-        aux_conf = default_dir / "auxiliary.conf"
-        write_config(
-            aux_conf,
-            """
-            [virustotaldl]
-            enabled = no
-            #dlintelkey = SomeKeyWithDLAccess
-            dlpath = /tmp/
-            """,
-        )
-        custom_conf = custom_conf_dir / "auxiliary.conf"
+        custom_conf = custom_conf_path / "processing.conf"
         write_config(
             custom_conf,
             """
-            [virustotaldl]
-            enabled = yes
-            dlintelkey = %(ENV:DLINTELKEY)s
+            [virustotal]
+            on_demand = yes
+            key = %(ENV:DLINTELKEY)s
             """,
         )
 
         custom_secret = "MyReallySecretKeyWithAPercent(%)InIt"
         monkeypatch.setenv("DLINTELKEY", custom_secret)
-        config = Config("auxiliary")
-        section = config.get("virustotaldl")
+        config = Config("processing")
+        section = config.get("virustotal")
         # Inherited from default config
-        assert section.dlpath == "/tmp/"
-        # Overridden from custom config
         assert section.enabled is True
+        # Overridden from custom config
+        assert section.on_demand is True
         # Overridden from custom config and uses environment variable
-        assert section.dlintelkey == custom_secret
+        assert section.key == custom_secret
 
-    def test_missing_environment_interpolation(self, default_config, custom_conf_dir, monkeypatch):
+    def test_missing_environment_interpolation(self, custom_conf_path, monkeypatch):
         """Verify that an exception is raised if an ENV variable is to be used in a
         config file, but that variable is not present in the environment.
         """
-        default_dir = default_config.parent
-        aux_conf = default_dir / "auxiliary.conf"
+        custom_conf = custom_conf_path / "processing.conf"
         write_config(
-            aux_conf,
+            custom_conf,
             """
-            [foo]
-            bar = %(ENV:IDONTEXIST)s
+            [virustotal]
+            key = %(ENV:IDONTEXIST)s
             """,
         )
+
         with pytest.raises(configparser.InterpolationMissingOptionError):
-            _ = Config("auxiliary")
+            _ = Config("processing")

From 35829829a1336a4b8b1427beb842cf756d938027 Mon Sep 17 00:00:00 2001
From: Josh Feather <142008135+josh-feather@users.noreply.github.com>
Date: Wed, 14 Feb 2024 11:29:59 +0000
Subject: [PATCH 4/6] Additional test_file_extra_info skip criteria, move
 Config load into extractor_ctx

---
 .../file_extra_info_modules/__init__.py       |  2 +-
 tests/test_file_extra_info.py                 | 59 ++++++++++++++-----
 2 files changed, 45 insertions(+), 16 deletions(-)

diff --git a/lib/cuckoo/common/integrations/file_extra_info_modules/__init__.py b/lib/cuckoo/common/integrations/file_extra_info_modules/__init__.py
index d28218c39d7..4fb477ae3d9 100644
--- a/lib/cuckoo/common/integrations/file_extra_info_modules/__init__.py
+++ b/lib/cuckoo/common/integrations/file_extra_info_modules/__init__.py
@@ -11,7 +11,6 @@
 from lib.cuckoo.common.config import Config
 from lib.cuckoo.common.path_utils import path_mkdir, path_object
 
-cfg = Config()
 log = logging.getLogger(__name__)
 
 
@@ -50,6 +49,7 @@ def wrapped(*args, **kwargs):
 
 @contextlib.contextmanager
 def extractor_ctx(filepath, tool_name, prefix=None):
+    cfg = Config()
     folder = os.path.join(cfg.cuckoo.get("tmppath", "/tmp"), "cape-external")
     path_mkdir(folder, exist_ok=True)
 
diff --git a/tests/test_file_extra_info.py b/tests/test_file_extra_info.py
index 20b24dfa485..8068687380b 100644
--- a/tests/test_file_extra_info.py
+++ b/tests/test_file_extra_info.py
@@ -8,11 +8,15 @@
 
 from lib.cuckoo.common.integrations import file_extra_info
 
+self_extraction_dir = (pathlib.Path(__file__).parent / "data" / "selfextraction")
 
 @pytest.mark.skipif(
-    not (pathlib.Path(__file__).parent / "data" / "selfextraction").exists(), reason="Required data file is not present"
+    not (self_extraction_dir).exists(), reason="Required data file is not present"
 )
 class TestFileExtraInfo:
+    @pytest.mark.skipif(
+        not (self_extraction_dir / "0ea5e25b12ab314bc9a0569c3ca756f205f40b792119f8e0fc62c874628dfea0.msi").exists(), reason="Required data file is not present"
+    )   
     def test_generic_file_extractors(self):
         results = {}
         data_dictionary = {"type": "MSI Installer"}
@@ -20,7 +24,7 @@ def test_generic_file_extractors(self):
         tmpdir = tempfile.mkdtemp()
         duplicated = {"sha256": set()}
         file_extra_info.generic_file_extractors(
-            "tests/data/selfextraction/0ea5e25b12ab314bc9a0569c3ca756f205f40b792119f8e0fc62c874628dfea0.msi",
+            f"{self_extraction_dir}/0ea5e25b12ab314bc9a0569c3ca756f205f40b792119f8e0fc62c874628dfea0.msi",
             tmpdir,
             data_dictionary,
             options_dict,
@@ -31,6 +35,9 @@ def test_generic_file_extractors(self):
         assert data_dictionary["extracted_files_tool"] == "MsiExtract"
         assert len(data_dictionary["extracted_files"]) == 2
 
+    @pytest.mark.skipif(
+        not (self_extraction_dir / "5b354397f6393ed777639b7d40dec3f37215dcb5078c63993e8a9703e819e2bc.inno").exists(), reason="Required data file is not present"
+    )
     def test_generic_file_extractors_no_tests(self):
         results = {}
         data_dictionary = {"die": ["Inno Setup"], "type": ""}
@@ -38,7 +45,7 @@ def test_generic_file_extractors_no_tests(self):
         tmpdir = tempfile.mkdtemp()
         duplicated = {"sha256": set()}
         file_extra_info.generic_file_extractors(
-            "tests/data/selfextraction/5b354397f6393ed777639b7d40dec3f37215dcb5078c63993e8a9703e819e2bc.inno",
+            f"{self_extraction_dir}/5b354397f6393ed777639b7d40dec3f37215dcb5078c63993e8a9703e819e2bc.inno",
             tmpdir,
             data_dictionary,
             options_dict,
@@ -51,20 +58,23 @@ def test_generic_file_extractors_no_tests(self):
     @pytest.mark.skip(reason="Not implemented yet")
     def test_batch_extract(self):
         extracted_files = file_extra_info.batch_extract(
-            file="tests/data/selfextraction/",
+            file=f"{self_extraction_dir}/",
         )
         assert len(extracted_files["result"]["extracted_files"]) == 4
 
+    @pytest.mark.skipif(
+        not (self_extraction_dir / "c738cdd8ec0d65769e17eed1d6fe371893a2972b7a432c6532446d225e166733.vbe").exists(), reason="Required data file is not present"
+    )
     def test_vbe_extract(self):
         extracted_files = file_extra_info.vbe_extract(
-            file="tests/data/selfextraction/c738cdd8ec0d65769e17eed1d6fe371893a2972b7a432c6532446d225e166733.vbe",
+            file=f"{self_extraction_dir}/c738cdd8ec0d65769e17eed1d6fe371893a2972b7a432c6532446d225e166733.vbe",
         )
         assert len(extracted_files["result"]["extracted_files"]) == 1
 
     @pytest.mark.skip(reason="Not implemented yet")
     def test_eziriz_deobfuscate(self):
         extracted_files = file_extra_info.eziriz_deobfuscate(
-            file="tests/data/selfextraction/",
+            file=f"{self_extraction_dir}/",
             data_dictionary={"die": ["Eziriz .NET Reactor"]} ** {"test": True, "options": {}},
         )
         self.assertEqual(len(extracted_files["result"]["extracted_files"]), 4, "Failed to extract.")
@@ -72,21 +82,27 @@ def test_eziriz_deobfuscate(self):
     @pytest.mark.skip(reason="Not implemented yet")
     def test_de4dot_deobfuscate(self):
         extracted_files = file_extra_info.de4dot_deobfuscate(
-            file="tests/data/selfextraction/", filetype="Mono", **{"test": True, "options": {}}
+            file=f"{self_extraction_dir}/", filetype="Mono", **{"test": True, "options": {}}
         )
         self.assertEqual(len(extracted_files["result"]["extracted_files"]), 4, "Failed to extract.")
 
+    @pytest.mark.skipif(
+        not (self_extraction_dir / "0ea5e25b12ab314bc9a0569c3ca756f205f40b792119f8e0fc62c874628dfea0.msi").exists(), reason="Required data file is not present"
+    )
     def test_msi_extract(self):
         extracted_files = file_extra_info.msi_extract(
-            file="tests/data/selfextraction/0ea5e25b12ab314bc9a0569c3ca756f205f40b792119f8e0fc62c874628dfea0.msi",
+            file=f"{self_extraction_dir}/0ea5e25b12ab314bc9a0569c3ca756f205f40b792119f8e0fc62c874628dfea0.msi",
             filetype="MSI Installer",
             **{"tests": True, "options": {}},
         )
         assert len(extracted_files["result"]["extracted_files"]) == 2
 
+    @pytest.mark.skipif(
+        not (self_extraction_dir / "5b354397f6393ed777639b7d40dec3f37215dcb5078c63993e8a9703e819e2bc.inno").exists(), reason="Required data file is not present"
+    )
     def test_Inno_extract(self):
         extracted_files = file_extra_info.Inno_extract(
-            file="tests/data/selfextraction/5b354397f6393ed777639b7d40dec3f37215dcb5078c63993e8a9703e819e2bc.inno",
+            file=f"{self_extraction_dir}/5b354397f6393ed777639b7d40dec3f37215dcb5078c63993e8a9703e819e2bc.inno",
             data_dictionary={"die": ["Inno Setup"]},
         )
         assert len(extracted_files["result"]["extracted_files"]) == 1
@@ -95,22 +111,25 @@ def test_Inno_extract(self):
     @pytest.mark.skip(reason="Not implemented yet - need to include community repo")
     def test_kixtart_extract(self):
         extracted_files = file_extra_info.kixtart_extract(
-            file="tests/data/selfextraction/d0d415dbe02e893fb1b2d6112c0f38d8ce65ab3268c896bfc64ba06096d4d09a.kix",
+            file=f"{self_extraction_dir}/d0d415dbe02e893fb1b2d6112c0f38d8ce65ab3268c896bfc64ba06096d4d09a.kix",
         )
         assert len(extracted_files["result"]["extracted_files"]) == 4
 
     @pytest.mark.skip(reason="Not implemented yet")
     def test_UnAutoIt_extract(self):
         extracted_files = file_extra_info.kixtart_extract(
-            file="tests/data/selfextraction/5b354397f6393ed777639b7d40dec3f37215dcb5078c63993e8a9703e819e2bc.inno",
+            file=f"{self_extraction_dir}/5b354397f6393ed777639b7d40dec3f37215dcb5078c63993e8a9703e819e2bc.inno",
             data_dictionary={"yara": [{"name": "AutoIT_Compiled"}]},
             **{"test": True, "options": {}},
         )
         assert len(extracted_files["result"]["extracted_files"]) == 4
 
+    @pytest.mark.skipif(
+        not (self_extraction_dir / "1b0c4149df7892b2497c955dc393ed49e458062324a81288589b15492ce8b50b.upx").exists(), reason="Required data file is not present"
+    )
     def test_UPX_unpack(self):
         extracted_files = file_extra_info.UPX_unpack(
-            file="tests/data/selfextraction/1b0c4149df7892b2497c955dc393ed49e458062324a81288589b15492ce8b50b.upx",
+            file=f"{self_extraction_dir}/1b0c4149df7892b2497c955dc393ed49e458062324a81288589b15492ce8b50b.upx",
             filetype="UPX compressed",
             data_dictionary={},
             **{"test": True, "options": {}},
@@ -120,9 +139,12 @@ def test_UPX_unpack(self):
             "1b0c4149df7892b2497c955dc393ed49e458062324a81288589b15492ce8b50b.upx_unpacked"
         ]
 
+    @pytest.mark.skipif(
+        not (self_extraction_dir / "ab77ea6ad4b6766e0db88d4f49c2c0075ba18b3573d7c6d07ee878bd6e71c388.7z").exists(), reason="Required data file is not present"
+    )
     def test_SevenZip_unpack(self):
         extracted_files = file_extra_info.SevenZip_unpack(
-            file="tests/data/selfextraction/ab77ea6ad4b6766e0db88d4f49c2c0075ba18b3573d7c6d07ee878bd6e71c388.7z",
+            file=f"{self_extraction_dir}/ab77ea6ad4b6766e0db88d4f49c2c0075ba18b3573d7c6d07ee878bd6e71c388.7z",
             data_dictionary={"die": ["7-zip Installer data"]},
             filetype="",
             **{"test": True, "options": {}},
@@ -137,18 +159,25 @@ def test_SevenZip_unpack(self):
             "自述.txt",
         ]
 
+    @pytest.mark.skipif(
+        not (self_extraction_dir / "9e69c36d967afbb1a948a022fcfb1a6384b35b233a47e7d859145db19018d21e.sfx").exists(), reason="Required data file is not present"
+    )
     def test_RarSFX_extract(self):
         extracted_files = file_extra_info.RarSFX_extract(
-            file="tests/data/selfextraction/9e69c36d967afbb1a948a022fcfb1a6384b35b233a47e7d859145db19018d21e.sfx",
+            file=f"{self_extraction_dir}/9e69c36d967afbb1a948a022fcfb1a6384b35b233a47e7d859145db19018d21e.sfx",
             data_dictionary={"type": "RAR self-extracting archive"},
             options={},
         )
         assert len(extracted_files["result"]["extracted_files"]) == 3
         assert sorted(extracted_files["result"]["extracted_files"]) == ["Manag.exe", "mLib.cs", "x64.xr"]
 
+
+    @pytest.mark.skipif(
+        not (self_extraction_dir / "12c4d9eddce807d10e3578fcf2918366def586ec374a35957880a65dbd467efc.one").exists(), reason="Required data file is not present"
+    )   
     def test_office_one_extract(self):
         extracted_files = file_extra_info.office_one(
-            file="tests/data/selfextraction/12c4d9eddce807d10e3578fcf2918366def586ec374a35957880a65dbd467efc.one",
+            file=f"{self_extraction_dir}/12c4d9eddce807d10e3578fcf2918366def586ec374a35957880a65dbd467efc.one",
         )
         assert len(extracted_files["result"]["extracted_files"]) == 6
         assert sorted(extracted_files["result"]["extracted_files"]) == [

From 96602f7cc8d02cf5c22a4cf00452151fd35052b9 Mon Sep 17 00:00:00 2001
From: doomedraven <doommedraven@gmail.com>
Date: Thu, 15 Feb 2024 14:26:58 +0100
Subject: [PATCH 5/6] fix conf load on each file processing

---
 .../common/integrations/file_extra_info.py    | 28 +++++++++----------
 .../file_extra_info_modules/__init__.py       |  5 +---
 2 files changed, 15 insertions(+), 18 deletions(-)

diff --git a/lib/cuckoo/common/integrations/file_extra_info.py b/lib/cuckoo/common/integrations/file_extra_info.py
index 3b10b0f7e6c..105fb11ea08 100644
--- a/lib/cuckoo/common/integrations/file_extra_info.py
+++ b/lib/cuckoo/common/integrations/file_extra_info.py
@@ -59,7 +59,7 @@
 
 DuplicatesType = DefaultDict[str, Set[str]]
 
-
+cfg = Config()
 processing_conf = Config("processing")
 selfextract_conf = Config("selfextract")
 
@@ -133,7 +133,7 @@
 
 exclude_startswith = ("parti_",)
 excluded_extensions = (".parti",)
-
+tools_folder = os.path.join(cfg.cuckoo.get("tmppath", "/tmp"), "cape-external")
 
 def static_file_info(
     data_dictionary: dict,
@@ -515,7 +515,7 @@ def generic_file_extractors(
 
 
 def _generic_post_extraction_process(file: str, tool_name: str, decoded: str) -> SuccessfulExtractionReturnType:
-    with extractor_ctx(file, tool_name) as ctx:
+    with extractor_ctx(file, tool_name, folder=tools_folder) as ctx:
         basename = f"{os.path.basename(file)}_decoded"
         decoded_file_path = os.path.join(ctx["tempdir"], basename)
         _ = path_write_file(decoded_file_path, decoded, mode="text")
@@ -595,7 +595,7 @@ def eziriz_deobfuscate(file: str, *, data_dictionary: dict, **_) -> ExtractorRet
         log.error("You need to add execution permissions: chmod a+x data/NETReactorSlayer.CLI")
         return
 
-    with extractor_ctx(file, "eziriz", prefix="eziriz_") as ctx:
+    with extractor_ctx(file, "eziriz", prefix="eziriz_", folder=tools_folder) as ctx:
         tempdir = ctx["tempdir"]
         dest_path = os.path.join(tempdir, os.path.basename(file))
         _ = run_tool(
@@ -628,7 +628,7 @@ def de4dot_deobfuscate(file: str, *, filetype: str, **_) -> ExtractorReturnType:
         log.error("Missed dependency: sudo apt install de4dot")
         return
 
-    with extractor_ctx(file, "de4dot", prefix="de4dot_") as ctx:
+    with extractor_ctx(file, "de4dot", prefix="de4dot_", folder=tools_folder) as ctx:
         tempdir = ctx["tempdir"]
         dest_path = os.path.join(tempdir, os.path.basename(file))
         _ = run_tool(
@@ -657,7 +657,7 @@ def msi_extract(file: str, *, filetype: str, **kwargs) -> ExtractorReturnType:
 
     extracted_files = []
     # sudo apt install msitools or 7z
-    with extractor_ctx(file, "MsiExtract", prefix="msidump_") as ctx:
+    with extractor_ctx(file, "MsiExtract", prefix="msidump_", folder=tools_folder) as ctx:
         tempdir = ctx["tempdir"]
         output = False
         if not kwargs.get("tests"):
@@ -705,7 +705,7 @@ def Inno_extract(file: str, *, data_dictionary: dict, **_) -> ExtractorReturnTyp
         log.error("Missed dependency: sudo apt install innoextract")
         return
 
-    with extractor_ctx(file, "InnoExtract", prefix="innoextract_") as ctx:
+    with extractor_ctx(file, "InnoExtract", prefix="innoextract_", folder=tools_folder) as ctx:
         tempdir = ctx["tempdir"]
         run_tool(
             [selfextract_conf.Inno_extract.binary, file, "--output-dir", tempdir],
@@ -731,7 +731,7 @@ def kixtart_extract(file: str, **_) -> ExtractorReturnType:
     if not data.startswith(b"\x1a\xaf\x06\x00\x00\x10"):
         return
 
-    with extractor_ctx(file, "Kixtart", prefix="kixtart_") as ctx:
+    with extractor_ctx(file, "Kixtart", prefix="kixtart_", folder=tools_folder) as ctx:
         tempdir = ctx["tempdir"]
         kix = Kixtart(file, dump_dir=tempdir)
         kix.decrypt()
@@ -756,7 +756,7 @@ def UnAutoIt_extract(file: str, *, data_dictionary: dict, **_) -> ExtractorRetur
         UN_AUTOIT_NOTIF = True
         return
 
-    with extractor_ctx(file, "UnAutoIt", prefix="unautoit_") as ctx:
+    with extractor_ctx(file, "UnAutoIt", prefix="unautoit_", folder=tools_folder) as ctx:
         tempdir = ctx["tempdir"]
         output = run_tool(
             [unautoit_binary, "extract-all", "--output-dir", tempdir, file],
@@ -778,7 +778,7 @@ def UPX_unpack(file: str, *, filetype: str, data_dictionary: dict, **_) -> Extra
     ):
         return
 
-    with extractor_ctx(file, "UnUPX", prefix="unupx_") as ctx:
+    with extractor_ctx(file, "UnUPX", prefix="unupx_", folder=tools_folder) as ctx:
         basename = f"{os.path.basename(file)}_unpacked"
         dest_path = os.path.join(ctx["tempdir"], basename)
         output = run_tool(
@@ -852,7 +852,7 @@ def SevenZip_unpack(file: str, *, filetype: str, data_dictionary: dict, options:
     else:
         return
 
-    with extractor_ctx(file, tool, prefix=prefix) as ctx:
+    with extractor_ctx(file, tool, prefix=prefix, folder=tools_folder) as ctx:
         tempdir = ctx["tempdir"]
         HAVE_SFLOCK = False
         if HAVE_SFLOCK:
@@ -892,7 +892,7 @@ def RarSFX_extract(file, *, data_dictionary, options: dict, **_) -> ExtractorRet
         log.warning("Missed UnRar binary: /usr/bin/unrar. sudo apt install unrar")
         return
 
-    with extractor_ctx(file, "UnRarSFX", prefix="unrar_") as ctx:
+    with extractor_ctx(file, "UnRarSFX", prefix="unrar_", folder=tools_folder) as ctx:
         tempdir = ctx["tempdir"]
         password = options.get("password", "infected")
         output = run_tool(
@@ -914,7 +914,7 @@ def office_one(file, **_) -> ExtractorReturnType:
     ):
         return
 
-    with extractor_ctx(file, "OfficeOne", prefix="office_one") as ctx:
+    with extractor_ctx(file, "OfficeOne", prefix="office_one", folder=tools_folder) as ctx:
         tempdir = ctx["tempdir"]
         try:
             document = OneNoteExtractor(path_read_file(file))
@@ -937,7 +937,7 @@ def msix_extract(file: str, *, data_dictionary: dict, **_) -> ExtractorReturnTyp
     ):
         return
 
-    with extractor_ctx(file, "MSIX", prefix="msixdump_") as ctx:
+    with extractor_ctx(file, "MSIX", prefix="msixdump_", folder=tools_folder) as ctx:
         tempdir = ctx["tempdir"]
         _ = run_tool(
             ["unzip", file, "-d", tempdir],
diff --git a/lib/cuckoo/common/integrations/file_extra_info_modules/__init__.py b/lib/cuckoo/common/integrations/file_extra_info_modules/__init__.py
index 4fb477ae3d9..fc22fcdf8f4 100644
--- a/lib/cuckoo/common/integrations/file_extra_info_modules/__init__.py
+++ b/lib/cuckoo/common/integrations/file_extra_info_modules/__init__.py
@@ -8,7 +8,6 @@
 import timeit
 from typing import List, Optional, TypedDict
 
-from lib.cuckoo.common.config import Config
 from lib.cuckoo.common.path_utils import path_mkdir, path_object
 
 log = logging.getLogger(__name__)
@@ -48,9 +47,7 @@ def wrapped(*args, **kwargs):
 
 
 @contextlib.contextmanager
-def extractor_ctx(filepath, tool_name, prefix=None):
-    cfg = Config()
-    folder = os.path.join(cfg.cuckoo.get("tmppath", "/tmp"), "cape-external")
+def extractor_ctx(filepath, tool_name, prefix=None, folder="/tmp/cape-external"):
     path_mkdir(folder, exist_ok=True)
 
     tempdir = tempfile.mkdtemp(prefix=prefix, dir=folder)

From 3048bf21d71ec59fe62c4ba936670c439ff86802 Mon Sep 17 00:00:00 2001
From: Josh Feather <142008135+josh-feather@users.noreply.github.com>
Date: Thu, 15 Feb 2024 14:59:16 +0000
Subject: [PATCH 6/6] Set tools_folder to tmp in test_file_extra_info

---
 tests/test_file_extra_info.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/tests/test_file_extra_info.py b/tests/test_file_extra_info.py
index 8068687380b..affe09eba15 100644
--- a/tests/test_file_extra_info.py
+++ b/tests/test_file_extra_info.py
@@ -8,6 +8,11 @@
 
 from lib.cuckoo.common.integrations import file_extra_info
 
+@pytest.fixture(autouse=True)
+def set_tools_folder():
+    file_extra_info.tools_folder = "/tmp"
+    yield
+
 self_extraction_dir = (pathlib.Path(__file__).parent / "data" / "selfextraction")
 
 @pytest.mark.skipif(