From aa4889542274870b360bb4233345791abb3a020a Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Mon, 16 Sep 2024 11:39:01 -0500
Subject: [PATCH 01/13] remove parser no longer used for gcg

Signed-off-by: Jeffrey Martin <jemartin@nvidia.com>
---
 garak/resources/gcg/__init__.py     |  2 +-
 garak/resources/gcg/generate_gcg.py | 46 -----------------------------
 2 files changed, 1 insertion(+), 47 deletions(-)

diff --git a/garak/resources/gcg/__init__.py b/garak/resources/gcg/__init__.py
index 1c7b644d5..323958347 100644
--- a/garak/resources/gcg/__init__.py
+++ b/garak/resources/gcg/__init__.py
@@ -6,4 +6,4 @@
 # Greedy Coordinate Gradient implementation lightly modified from https://github.com/llm-attacks/llm-attacks
 # Paper can be found at: https://arxiv.org/abs/2307.15043
 
-from .generate_gcg import gcg_parser, run_gcg
+from .generate_gcg import run_gcg
diff --git a/garak/resources/gcg/generate_gcg.py b/garak/resources/gcg/generate_gcg.py
index 0dc969fc1..dcdc1f191 100644
--- a/garak/resources/gcg/generate_gcg.py
+++ b/garak/resources/gcg/generate_gcg.py
@@ -40,52 +40,6 @@
 resource_data = garak._config.transient.package_dir / "resources"
 gcg_resource_data = garak._config.transient.cache_dir / "resources" / "gcg" / "data"
 
-# GCG parser used by interactive mode
-gcg_parser = ArgumentParser()
-gcg_parser.add_argument("--model_names", nargs="+", help="Model names for generation")
-gcg_parser.add_argument(
-    "--transfer", action="store_true", help="Whether to generate attack for transfer"
-)
-gcg_parser.add_argument(
-    "--progressive", action="store_true", help="Use progressive goals"
-)
-gcg_parser.add_argument("--stop_success", action="store_true", help="Stop on success")
-gcg_parser.add_argument(
-    "--train_data",
-    type=str,
-    default=resource_data / "advbench" / "harmful_behaviors.csv",
-    help="Path to training data",
-)
-gcg_parser.add_argument(
-    "--n_train", type=int, default=50, help="Number of training samples to use"
-)
-gcg_parser.add_argument(
-    "--n_test", type=int, default=0, help="Number of test samples to use"
-)
-gcg_parser.add_argument(
-    "--outfile",
-    type=str,
-    default=gcg_resource_data / "gcg_prompts.txt",
-    help="Location to write GCG attack output",
-)
-gcg_parser.add_argument(
-    "--control_init", type=str, default=CONTROL_INIT, help="Initial control string"
-)
-gcg_parser.add_argument(
-    "--n_steps", type=int, default=500, help="Number of steps for optimization"
-)
-gcg_parser.add_argument(
-    "--batch_size", type=int, default=128, help="Optimization batch size"
-)
-gcg_parser.add_argument(
-    "--allow_non_ascii",
-    action="store_true",
-    help="Allow non-ASCII characters in control string",
-)
-gcg_parser.add_argument(
-    "--save_logs", action="store_true", help="Keep detailed GCG generation logs"
-)
-
 
 def run_gcg(
     target_generator: garak.generators.Generator = None,

From 724a81ec37f24a5c9cca718474a1e6719aab939f Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Tue, 17 Sep 2024 08:10:09 -0500
Subject: [PATCH 02/13] move rest demo to tools

Signed-off-by: Jeffrey Martin <jemartin@nvidia.com>
---
 {garak/resources => tools}/rest/restdemo.json | 0
 {garak/resources => tools}/rest/restserv.py   | 0
 2 files changed, 0 insertions(+), 0 deletions(-)
 rename {garak/resources => tools}/rest/restdemo.json (100%)
 rename {garak/resources => tools}/rest/restserv.py (100%)

diff --git a/garak/resources/rest/restdemo.json b/tools/rest/restdemo.json
similarity index 100%
rename from garak/resources/rest/restdemo.json
rename to tools/rest/restdemo.json
diff --git a/garak/resources/rest/restserv.py b/tools/rest/restserv.py
similarity index 100%
rename from garak/resources/rest/restserv.py
rename to tools/rest/restserv.py

From 8890c94730ca0efe286c34d470527ce40481f0db Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Mon, 16 Sep 2024 14:32:43 -0500
Subject: [PATCH 03/13] add custom Path type for data files

* returns first found instance of filename
* custom `Path` raises exception when:
  * path escape attempt is detected
  * no file matching request exists
---
 garak/data/__init__.py | 60 +++++++++++++++++++++++++++++++++++++
 tests/test_data.py     | 68 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 128 insertions(+)
 create mode 100644 garak/data/__init__.py
 create mode 100644 tests/test_data.py

diff --git a/garak/data/__init__.py b/garak/data/__init__.py
new file mode 100644
index 000000000..4e20dab34
--- /dev/null
+++ b/garak/data/__init__.py
@@ -0,0 +1,60 @@
+# SPDX-FileCopyrightText: Portions Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+"""Local read only resources found by precedence matching supported paths
+
+Ideal usage:
+
+```
+file_path = resources / "filename"
+with open(file_path) as f:
+    f.read()
+```
+
+Resources that do not have a `shipped` version should wrap path access in a try block:
+```
+try:
+    file_path = resources / "filename"
+except GarakException as e:
+    logging.warn("No resource file found.", exc_info=e)
+```
+"""
+
+import pathlib
+
+from garak import _config
+from garak.exception import GarakException
+
+
+class LocalDataPath(pathlib.Path):
+    """restricted Path object usable only for existing resource files"""
+
+    ORDERED_SEARCH_PATHS = [
+        _config.transient.data_dir / "data",
+        _config.transient.package_dir / "data",
+    ]
+
+    def joinpath(self, *pathsegments):
+
+        for segment in pathsegments:
+            prefix_removed = None
+            for path in self.ORDERED_SEARCH_PATHS:
+                if (path == self and segment != "..") or path in self.parents:
+                    prefix_removed = self.relative_to(path)
+                    break
+            if prefix_removed is None:
+                raise GarakException(
+                    f"The requested resource does not refer to a valid path: {self}"
+                )
+            for path in self.ORDERED_SEARCH_PATHS:
+                if segment == "..":
+                    projected = (path / prefix_removed).parent
+                else:
+                    projected = (path / prefix_removed).joinpath(segment)
+                if projected.exists():
+                    return LocalDataPath(projected)
+
+        raise GarakException(f"The resource requested does not exist {segment}")
+
+
+path = LocalDataPath(_config.transient.data_dir / "data")
diff --git a/tests/test_data.py b/tests/test_data.py
new file mode 100644
index 000000000..84c8f2250
--- /dev/null
+++ b/tests/test_data.py
@@ -0,0 +1,68 @@
+# SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+import pytest
+import tempfile
+import os
+
+from garak import _config
+from garak.exception import GarakException
+from garak.data import path as data_path
+from garak.data import LocalDataPath
+
+
+@pytest.fixture
+def random_resource_filename(request) -> None:
+    with tempfile.NamedTemporaryFile(
+        dir=LocalDataPath.ORDERED_SEARCH_PATHS[-1], mode="w", delete=False
+    ) as tmpfile:
+        tmpfile.write("file data")
+
+    def remove_files():
+        for path in LocalDataPath.ORDERED_SEARCH_PATHS:
+            rem_path = path / os.path.basename(tmpfile.name)
+            if rem_path.exists():
+                rem_path.unlink()
+
+    request.addfinalizer(remove_files)
+
+    return os.path.basename(tmpfile.name)
+
+
+def test_no_relative_escape():
+    with pytest.raises(GarakException) as exc_info:
+        data_path / ".."
+    assert "does not refer to a valid path" in str(exc_info.value)
+
+
+def test_no_relative_escape_extended():
+    autodan_path = data_path / "autodan"
+    with pytest.raises(GarakException) as exc_info:
+        autodan_path / ".." / ".." / "configs"
+    assert "does not refer to a valid path" in str(exc_info.value)
+
+
+def test_allow_relative_in_path():
+    source = data_path / "autodan" / ".." / "gcg"
+    assert source.name == "gcg"
+
+
+def test_known_resource_found():
+    known_filename = "misp_descriptions.tsv"
+    source = data_path / known_filename
+    assert source.name == known_filename
+
+
+def test_local_override(random_resource_filename):
+    source = data_path / random_resource_filename
+    assert _config.transient.package_dir in source.parents
+
+    data_root_path = _config.transient.data_dir / "resources"
+    data_root_path.mkdir(parents=True, exist_ok=True)
+    with open(
+        data_root_path / random_resource_filename, encoding="utf-8", mode="w"
+    ) as f:
+        f.write("fake data")
+
+    source = data_path / random_resource_filename
+    assert _config.transient.data_dir in source.parents

From 4a73965d9dcfad5aafaecbc879144439dc8d1f18 Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Tue, 17 Sep 2024 12:14:48 -0500
Subject: [PATCH 04/13] use `garak.data.path` to access data files

Signed-off-by: Jeffrey Martin <jemartin@nvidia.com>
---
 garak/analyze/calibration.py                  |   4 +--
 garak/analyze/misp.py                         |   7 ++---
 garak/analyze/report_digest.py                |   6 ++---
 .../data => data/autodan}/autodan_init.txt    |   0
 .../data => data/autodan}/autodan_prompts.txt |   0
 .../data => data/autodan}/prompt_group.pth    | Bin
 garak/{resources => data}/banners             |   0
 .../beast/data => data/beast}/suffixes.txt    |   0
 garak/{resources => data}/book_cloze.tsv      |   0
 garak/{resources => data}/book_cloze_80.tsv   |   0
 garak/{resources => data}/calibration/bag.md  |   0
 .../calibration/calibration-2024-summer.json  |   0
 .../calibration/calibration-alpha-202407.json |   0
 .../calibration/calibration.json              |   0
 ...n_exclusion_toxicity_hateful_offensive.txt |   0
 .../donotanswer/human_chatbox.txt             |   0
 .../donotanswer/information_hazard.txt        |   0
 .../donotanswer/malicious_uses.txt            |   0
 .../donotanswer/misinformation_harms.txt      |   0
 .../{resources/gcg/data => data/gcg}/gcg.txt  |   0
 .../graph_connectivity.json                   |   0
 garak/{resources => data}/guardian_cloze.tsv  |   0
 .../inthewild_jailbreak_llms.txt              |   0
 garak/{resources => data}/ldnoobw-en.txt      |   0
 .../{resources => data}/misp_descriptions.tsv |   0
 garak/{resources => data}/nyt_cloze.tsv       |   0
 .../ofcom-potentially-offensive.txt           |   0
 garak/{resources => data}/potter_cloze.tsv    |   0
 .../primality_testing.json                    |   0
 garak/{resources => data}/profanity_en.csv    |   0
 .../realtoxicityprompts/rtp_flirtation.txt    |   0
 .../rtp_identity_attack.txt                   |   0
 .../realtoxicityprompts/rtp_insult.txt        |   0
 .../realtoxicityprompts/rtp_profanity.txt     |   0
 .../rtp_severe_toxicity.txt                   |   0
 .../rtp_sexually_explicit.txt                 |   0
 .../realtoxicityprompts/rtp_threat.txt        |   0
 .../safebench_filenames.txt                   |   0
 .../safebenchtiny_filenames.txt               |   0
 garak/{resources => data}/senator_search.json |   0
 garak/{resources => data}/slurprompts.jsonl   |   0
 .../slurprompts_mini.jsonl                    |   0
 .../slursreclaimedslurs.txt                   |   0
 .../tap/data => data/tap}/tap_jailbreaks.txt  |   0
 .../truefalse_falseclaims.txt                 |   0
 .../truefalse_falseclaims_50.txt              |   0
 garak/detectors/riskywords.py                 |   7 ++---
 garak/detectors/specialwords.py               |   3 ++-
 garak/interactive.py                          |   5 ++--
 garak/probes/continuation.py                  |   3 ++-
 garak/probes/dan.py                           |  11 +++-----
 garak/probes/donotanswer.py                   |   6 ++---
 garak/probes/encoding.py                      |   3 ++-
 garak/probes/leakreplay.py                    |   7 +++--
 garak/probes/misleading.py                    |   5 ++--
 garak/probes/realtoxicityprompts.py           |   6 ++---
 garak/probes/snowball.py                      |   7 ++---
 garak/probes/suffix.py                        |   5 ++--
 garak/probes/tap.py                           |   7 ++---
 garak/probes/topic.py                         |   2 +-
 garak/probes/visual_jailbreak.py              |  12 ++++-----
 garak/resources/autodan/autodan.py            |   9 +++----
 garak/resources/beast/beast_attack.py         |   2 +-
 garak/resources/common.py                     |  24 ++++++++++++------
 garak/resources/gcg/generate_gcg.py           |  11 ++++----
 garak/resources/tap/tap_main.py               |   6 +----
 tests/probes/test_probes.py                   |   2 +-
 tests/test_data.py                            |   2 +-
 68 files changed, 73 insertions(+), 89 deletions(-)
 rename garak/{resources/autodan/data => data/autodan}/autodan_init.txt (100%)
 rename garak/{resources/autodan/data => data/autodan}/autodan_prompts.txt (100%)
 rename garak/{resources/autodan/data => data/autodan}/prompt_group.pth (100%)
 rename garak/{resources => data}/banners (100%)
 rename garak/{resources/beast/data => data/beast}/suffixes.txt (100%)
 rename garak/{resources => data}/book_cloze.tsv (100%)
 rename garak/{resources => data}/book_cloze_80.tsv (100%)
 rename garak/{resources => data}/calibration/bag.md (100%)
 rename garak/{resources => data}/calibration/calibration-2024-summer.json (100%)
 rename garak/{resources => data}/calibration/calibration-alpha-202407.json (100%)
 rename garak/{resources => data}/calibration/calibration.json (100%)
 rename garak/{resources => data}/donotanswer/discrimination_exclusion_toxicity_hateful_offensive.txt (100%)
 rename garak/{resources => data}/donotanswer/human_chatbox.txt (100%)
 rename garak/{resources => data}/donotanswer/information_hazard.txt (100%)
 rename garak/{resources => data}/donotanswer/malicious_uses.txt (100%)
 rename garak/{resources => data}/donotanswer/misinformation_harms.txt (100%)
 rename garak/{resources/gcg/data => data/gcg}/gcg.txt (100%)
 rename garak/{resources => data}/graph_connectivity.json (100%)
 rename garak/{resources => data}/guardian_cloze.tsv (100%)
 rename garak/{resources => data}/inthewild_jailbreak_llms.txt (100%)
 rename garak/{resources => data}/ldnoobw-en.txt (100%)
 rename garak/{resources => data}/misp_descriptions.tsv (100%)
 rename garak/{resources => data}/nyt_cloze.tsv (100%)
 rename garak/{resources => data}/ofcom-potentially-offensive.txt (100%)
 rename garak/{resources => data}/potter_cloze.tsv (100%)
 rename garak/{resources => data}/primality_testing.json (100%)
 rename garak/{resources => data}/profanity_en.csv (100%)
 rename garak/{resources => data}/realtoxicityprompts/rtp_flirtation.txt (100%)
 rename garak/{resources => data}/realtoxicityprompts/rtp_identity_attack.txt (100%)
 rename garak/{resources => data}/realtoxicityprompts/rtp_insult.txt (100%)
 rename garak/{resources => data}/realtoxicityprompts/rtp_profanity.txt (100%)
 rename garak/{resources => data}/realtoxicityprompts/rtp_severe_toxicity.txt (100%)
 rename garak/{resources => data}/realtoxicityprompts/rtp_sexually_explicit.txt (100%)
 rename garak/{resources => data}/realtoxicityprompts/rtp_threat.txt (100%)
 rename garak/{resources => data}/safebench_filenames.txt (100%)
 rename garak/{resources => data}/safebenchtiny_filenames.txt (100%)
 rename garak/{resources => data}/senator_search.json (100%)
 rename garak/{resources => data}/slurprompts.jsonl (100%)
 rename garak/{resources => data}/slurprompts_mini.jsonl (100%)
 rename garak/{resources => data}/slursreclaimedslurs.txt (100%)
 rename garak/{resources/tap/data => data/tap}/tap_jailbreaks.txt (100%)
 rename garak/{resources => data}/truefalse_falseclaims.txt (100%)
 rename garak/{resources => data}/truefalse_falseclaims_50.txt (100%)

diff --git a/garak/analyze/calibration.py b/garak/analyze/calibration.py
index 79190ed3c..f8ac5a903 100644
--- a/garak/analyze/calibration.py
+++ b/garak/analyze/calibration.py
@@ -10,7 +10,7 @@
 from typing import Union
 
 
-from garak import _config
+from garak.data import path as data_path
 
 MINIMUM_STD_DEV = (
     0.01732  # stddev=0 gives unusable z-scores; give it an arbitrary floor of 3^.5 %
@@ -132,7 +132,7 @@ def defcon_and_comment(
         return zscore_defcon, zscore_comment
 
     def _build_path(self, filename):
-        return _config.transient.package_dir / "resources" / "calibration" / filename
+        return data_path / "calibration" / filename
 
     def __init__(self, calibration_path: Union[None, str, pathlib.Path] = None) -> None:
 
diff --git a/garak/analyze/misp.py b/garak/analyze/misp.py
index c0b9a1fba..393c9bd0b 100644
--- a/garak/analyze/misp.py
+++ b/garak/analyze/misp.py
@@ -9,12 +9,9 @@
 import os
 
 from garak import _plugins
-import garak._config
+from garak.data import path as data_path
 
-# does this utility really have access to _config?
-misp_resource_file = (
-    garak._config.transient.package_dir / "resources" / "misp_descriptions.tsv"
-)
+misp_resource_file = data_path / "misp_descriptions.tsv"
 misp_descriptions = {}
 if os.path.isfile(misp_resource_file):
     with open(misp_resource_file, "r", encoding="utf-8") as f:
diff --git a/garak/analyze/report_digest.py b/garak/analyze/report_digest.py
index e1f0315ce..a655a4e29 100644
--- a/garak/analyze/report_digest.py
+++ b/garak/analyze/report_digest.py
@@ -14,8 +14,10 @@
 import sqlite3
 
 from garak import _config
+from garak.data import path as data_path
 import garak.analyze.calibration
 
+
 if not _config.loaded:
     _config.load_config()
 
@@ -33,9 +35,7 @@
 about_z_template = templateEnv.get_template("digest_about_z.jinja")
 
 
-misp_resource_file = (
-    _config.transient.package_dir / "resources" / "misp_descriptions.tsv"
-)
+misp_resource_file = data_path / "misp_descriptions.tsv"
 misp_descriptions = {}
 if os.path.isfile(misp_resource_file):
     with open(misp_resource_file, "r", encoding="utf-8") as f:
diff --git a/garak/resources/autodan/data/autodan_init.txt b/garak/data/autodan/autodan_init.txt
similarity index 100%
rename from garak/resources/autodan/data/autodan_init.txt
rename to garak/data/autodan/autodan_init.txt
diff --git a/garak/resources/autodan/data/autodan_prompts.txt b/garak/data/autodan/autodan_prompts.txt
similarity index 100%
rename from garak/resources/autodan/data/autodan_prompts.txt
rename to garak/data/autodan/autodan_prompts.txt
diff --git a/garak/resources/autodan/data/prompt_group.pth b/garak/data/autodan/prompt_group.pth
similarity index 100%
rename from garak/resources/autodan/data/prompt_group.pth
rename to garak/data/autodan/prompt_group.pth
diff --git a/garak/resources/banners b/garak/data/banners
similarity index 100%
rename from garak/resources/banners
rename to garak/data/banners
diff --git a/garak/resources/beast/data/suffixes.txt b/garak/data/beast/suffixes.txt
similarity index 100%
rename from garak/resources/beast/data/suffixes.txt
rename to garak/data/beast/suffixes.txt
diff --git a/garak/resources/book_cloze.tsv b/garak/data/book_cloze.tsv
similarity index 100%
rename from garak/resources/book_cloze.tsv
rename to garak/data/book_cloze.tsv
diff --git a/garak/resources/book_cloze_80.tsv b/garak/data/book_cloze_80.tsv
similarity index 100%
rename from garak/resources/book_cloze_80.tsv
rename to garak/data/book_cloze_80.tsv
diff --git a/garak/resources/calibration/bag.md b/garak/data/calibration/bag.md
similarity index 100%
rename from garak/resources/calibration/bag.md
rename to garak/data/calibration/bag.md
diff --git a/garak/resources/calibration/calibration-2024-summer.json b/garak/data/calibration/calibration-2024-summer.json
similarity index 100%
rename from garak/resources/calibration/calibration-2024-summer.json
rename to garak/data/calibration/calibration-2024-summer.json
diff --git a/garak/resources/calibration/calibration-alpha-202407.json b/garak/data/calibration/calibration-alpha-202407.json
similarity index 100%
rename from garak/resources/calibration/calibration-alpha-202407.json
rename to garak/data/calibration/calibration-alpha-202407.json
diff --git a/garak/resources/calibration/calibration.json b/garak/data/calibration/calibration.json
similarity index 100%
rename from garak/resources/calibration/calibration.json
rename to garak/data/calibration/calibration.json
diff --git a/garak/resources/donotanswer/discrimination_exclusion_toxicity_hateful_offensive.txt b/garak/data/donotanswer/discrimination_exclusion_toxicity_hateful_offensive.txt
similarity index 100%
rename from garak/resources/donotanswer/discrimination_exclusion_toxicity_hateful_offensive.txt
rename to garak/data/donotanswer/discrimination_exclusion_toxicity_hateful_offensive.txt
diff --git a/garak/resources/donotanswer/human_chatbox.txt b/garak/data/donotanswer/human_chatbox.txt
similarity index 100%
rename from garak/resources/donotanswer/human_chatbox.txt
rename to garak/data/donotanswer/human_chatbox.txt
diff --git a/garak/resources/donotanswer/information_hazard.txt b/garak/data/donotanswer/information_hazard.txt
similarity index 100%
rename from garak/resources/donotanswer/information_hazard.txt
rename to garak/data/donotanswer/information_hazard.txt
diff --git a/garak/resources/donotanswer/malicious_uses.txt b/garak/data/donotanswer/malicious_uses.txt
similarity index 100%
rename from garak/resources/donotanswer/malicious_uses.txt
rename to garak/data/donotanswer/malicious_uses.txt
diff --git a/garak/resources/donotanswer/misinformation_harms.txt b/garak/data/donotanswer/misinformation_harms.txt
similarity index 100%
rename from garak/resources/donotanswer/misinformation_harms.txt
rename to garak/data/donotanswer/misinformation_harms.txt
diff --git a/garak/resources/gcg/data/gcg.txt b/garak/data/gcg/gcg.txt
similarity index 100%
rename from garak/resources/gcg/data/gcg.txt
rename to garak/data/gcg/gcg.txt
diff --git a/garak/resources/graph_connectivity.json b/garak/data/graph_connectivity.json
similarity index 100%
rename from garak/resources/graph_connectivity.json
rename to garak/data/graph_connectivity.json
diff --git a/garak/resources/guardian_cloze.tsv b/garak/data/guardian_cloze.tsv
similarity index 100%
rename from garak/resources/guardian_cloze.tsv
rename to garak/data/guardian_cloze.tsv
diff --git a/garak/resources/inthewild_jailbreak_llms.txt b/garak/data/inthewild_jailbreak_llms.txt
similarity index 100%
rename from garak/resources/inthewild_jailbreak_llms.txt
rename to garak/data/inthewild_jailbreak_llms.txt
diff --git a/garak/resources/ldnoobw-en.txt b/garak/data/ldnoobw-en.txt
similarity index 100%
rename from garak/resources/ldnoobw-en.txt
rename to garak/data/ldnoobw-en.txt
diff --git a/garak/resources/misp_descriptions.tsv b/garak/data/misp_descriptions.tsv
similarity index 100%
rename from garak/resources/misp_descriptions.tsv
rename to garak/data/misp_descriptions.tsv
diff --git a/garak/resources/nyt_cloze.tsv b/garak/data/nyt_cloze.tsv
similarity index 100%
rename from garak/resources/nyt_cloze.tsv
rename to garak/data/nyt_cloze.tsv
diff --git a/garak/resources/ofcom-potentially-offensive.txt b/garak/data/ofcom-potentially-offensive.txt
similarity index 100%
rename from garak/resources/ofcom-potentially-offensive.txt
rename to garak/data/ofcom-potentially-offensive.txt
diff --git a/garak/resources/potter_cloze.tsv b/garak/data/potter_cloze.tsv
similarity index 100%
rename from garak/resources/potter_cloze.tsv
rename to garak/data/potter_cloze.tsv
diff --git a/garak/resources/primality_testing.json b/garak/data/primality_testing.json
similarity index 100%
rename from garak/resources/primality_testing.json
rename to garak/data/primality_testing.json
diff --git a/garak/resources/profanity_en.csv b/garak/data/profanity_en.csv
similarity index 100%
rename from garak/resources/profanity_en.csv
rename to garak/data/profanity_en.csv
diff --git a/garak/resources/realtoxicityprompts/rtp_flirtation.txt b/garak/data/realtoxicityprompts/rtp_flirtation.txt
similarity index 100%
rename from garak/resources/realtoxicityprompts/rtp_flirtation.txt
rename to garak/data/realtoxicityprompts/rtp_flirtation.txt
diff --git a/garak/resources/realtoxicityprompts/rtp_identity_attack.txt b/garak/data/realtoxicityprompts/rtp_identity_attack.txt
similarity index 100%
rename from garak/resources/realtoxicityprompts/rtp_identity_attack.txt
rename to garak/data/realtoxicityprompts/rtp_identity_attack.txt
diff --git a/garak/resources/realtoxicityprompts/rtp_insult.txt b/garak/data/realtoxicityprompts/rtp_insult.txt
similarity index 100%
rename from garak/resources/realtoxicityprompts/rtp_insult.txt
rename to garak/data/realtoxicityprompts/rtp_insult.txt
diff --git a/garak/resources/realtoxicityprompts/rtp_profanity.txt b/garak/data/realtoxicityprompts/rtp_profanity.txt
similarity index 100%
rename from garak/resources/realtoxicityprompts/rtp_profanity.txt
rename to garak/data/realtoxicityprompts/rtp_profanity.txt
diff --git a/garak/resources/realtoxicityprompts/rtp_severe_toxicity.txt b/garak/data/realtoxicityprompts/rtp_severe_toxicity.txt
similarity index 100%
rename from garak/resources/realtoxicityprompts/rtp_severe_toxicity.txt
rename to garak/data/realtoxicityprompts/rtp_severe_toxicity.txt
diff --git a/garak/resources/realtoxicityprompts/rtp_sexually_explicit.txt b/garak/data/realtoxicityprompts/rtp_sexually_explicit.txt
similarity index 100%
rename from garak/resources/realtoxicityprompts/rtp_sexually_explicit.txt
rename to garak/data/realtoxicityprompts/rtp_sexually_explicit.txt
diff --git a/garak/resources/realtoxicityprompts/rtp_threat.txt b/garak/data/realtoxicityprompts/rtp_threat.txt
similarity index 100%
rename from garak/resources/realtoxicityprompts/rtp_threat.txt
rename to garak/data/realtoxicityprompts/rtp_threat.txt
diff --git a/garak/resources/safebench_filenames.txt b/garak/data/safebench_filenames.txt
similarity index 100%
rename from garak/resources/safebench_filenames.txt
rename to garak/data/safebench_filenames.txt
diff --git a/garak/resources/safebenchtiny_filenames.txt b/garak/data/safebenchtiny_filenames.txt
similarity index 100%
rename from garak/resources/safebenchtiny_filenames.txt
rename to garak/data/safebenchtiny_filenames.txt
diff --git a/garak/resources/senator_search.json b/garak/data/senator_search.json
similarity index 100%
rename from garak/resources/senator_search.json
rename to garak/data/senator_search.json
diff --git a/garak/resources/slurprompts.jsonl b/garak/data/slurprompts.jsonl
similarity index 100%
rename from garak/resources/slurprompts.jsonl
rename to garak/data/slurprompts.jsonl
diff --git a/garak/resources/slurprompts_mini.jsonl b/garak/data/slurprompts_mini.jsonl
similarity index 100%
rename from garak/resources/slurprompts_mini.jsonl
rename to garak/data/slurprompts_mini.jsonl
diff --git a/garak/resources/slursreclaimedslurs.txt b/garak/data/slursreclaimedslurs.txt
similarity index 100%
rename from garak/resources/slursreclaimedslurs.txt
rename to garak/data/slursreclaimedslurs.txt
diff --git a/garak/resources/tap/data/tap_jailbreaks.txt b/garak/data/tap/tap_jailbreaks.txt
similarity index 100%
rename from garak/resources/tap/data/tap_jailbreaks.txt
rename to garak/data/tap/tap_jailbreaks.txt
diff --git a/garak/resources/truefalse_falseclaims.txt b/garak/data/truefalse_falseclaims.txt
similarity index 100%
rename from garak/resources/truefalse_falseclaims.txt
rename to garak/data/truefalse_falseclaims.txt
diff --git a/garak/resources/truefalse_falseclaims_50.txt b/garak/data/truefalse_falseclaims_50.txt
similarity index 100%
rename from garak/resources/truefalse_falseclaims_50.txt
rename to garak/data/truefalse_falseclaims_50.txt
diff --git a/garak/detectors/riskywords.py b/garak/detectors/riskywords.py
index 741522054..64538a437 100644
--- a/garak/detectors/riskywords.py
+++ b/garak/detectors/riskywords.py
@@ -7,11 +7,12 @@
 import csv
 
 from garak import _config
+from garak.data import path as data_path
 from garak.detectors.base import StringDetector
 
 surge_list = defaultdict(list)
 with open(
-    _config.transient.package_dir / "resources" / "profanity_en.csv",
+    data_path / "profanity_en.csv",
     "r",
     encoding="utf-8",
 ) as csvfile:
@@ -26,7 +27,7 @@
 MIN_OFCOM_STRENGTH = 2
 ofcom_list = defaultdict(list)
 with open(
-    _config.transient.package_dir / "resources" / "ofcom-potentially-offensive.txt",
+    data_path / "ofcom-potentially-offensive.txt",
     "r",
     encoding="utf-8",
 ) as csvfile:
@@ -285,7 +286,7 @@ class LDNOOBW(StringDetector):
     def __init__(self, config_root=_config):
         substrings = []
         with open(
-            _config.transient.package_dir / "resources" / "ldnoobw-en.txt",
+            data_path / "ldnoobw-en.txt",
             "r",
             encoding="utf-8",
         ) as wordlist:
diff --git a/garak/detectors/specialwords.py b/garak/detectors/specialwords.py
index 75329ab6f..d9f5f01ec 100644
--- a/garak/detectors/specialwords.py
+++ b/garak/detectors/specialwords.py
@@ -8,6 +8,7 @@
 
 from garak import _config
 from garak.detectors.base import StringDetector
+from garak.data import path as data_path
 
 
 class SlursReclaimedSlurs(StringDetector):
@@ -25,7 +26,7 @@ class SlursReclaimedSlurs(StringDetector):
     def __init__(self, config_root=_config):
         self.substrings = set([])
         with open(
-            _config.transient.package_dir / "resources" / "slurprompts.jsonl",
+            data_path / "slurprompts.jsonl",
             "r",
             encoding="utf-8",
         ) as slurfile:
diff --git a/garak/interactive.py b/garak/interactive.py
index 8005ca9f5..22d5f71a2 100644
--- a/garak/interactive.py
+++ b/garak/interactive.py
@@ -14,6 +14,7 @@
 import garak.harnesses.probewise
 from garak import _config
 from garak.evaluators import ThresholdEvaluator
+from garak.data import path as data_path
 from garak._plugins import enumerate_plugins
 
 logger = getLogger(__name__)
@@ -232,9 +233,7 @@ def settings_ns_provider(self) -> argparse.Namespace:
 
 def banner():
     """Display a random banner"""
-    with open(
-        _config.transient.package_dir / "resources" / "banners", "r", encoding="utf-8"
-    ) as bannerfile:
+    with open(data_path / "banners", "r", encoding="utf-8") as bannerfile:
         banners = bannerfile.read().lstrip().split("\n\n")
         print(Fore.LIGHTRED_EX + f"\n{random.choice(banners)}\n" + Fore.RESET)
 
diff --git a/garak/probes/continuation.py b/garak/probes/continuation.py
index 6f733516d..43ac6033d 100644
--- a/garak/probes/continuation.py
+++ b/garak/probes/continuation.py
@@ -9,6 +9,7 @@
 
 from garak import _config
 from garak.attempt import Attempt
+from garak.data import path as data_path
 from garak.probes.base import Probe
 
 
@@ -43,7 +44,7 @@ def __init__(self, config_root=_config):
         # append these to each of the prefixes for that term
         self._slur_prefixes = []
         with open(
-            _config.transient.package_dir / "resources" / self.source_resource_filename,
+            data_path / self.source_resource_filename,
             "r",
             encoding="utf-8",
         ) as slurfile:
diff --git a/garak/probes/dan.py b/garak/probes/dan.py
index e2486415f..c9ae8a5a6 100644
--- a/garak/probes/dan.py
+++ b/garak/probes/dan.py
@@ -11,6 +11,7 @@
 
 import garak.attempt
 from garak import _config
+from garak.data import path as data_path
 from garak.probes.base import Probe
 
 
@@ -537,11 +538,7 @@ class AutoDANCached(Probe):
 
     def __init__(
         self,
-        autodan_prompt_location: str = _config.transient.package_dir
-        / "resources"
-        / "autodan"
-        / "data"
-        / "autodan_prompts.txt",
+        autodan_prompt_location: str = data_path / "autodan" / "autodan_prompts.txt",
         config_root=_config,
     ):
         self.autodan_prompt_location = autodan_prompt_location
@@ -645,9 +642,7 @@ class DanInTheWild(Probe):
     def __init__(self, config_root=_config):
         super().__init__(config_root=config_root)
 
-        inthewild_path = str(
-            _config.transient.package_dir / "resources" / "inthewild_jailbreak_llms.txt"
-        )
+        inthewild_path = str(data_path / "inthewild_jailbreak_llms.txt")
 
         with open(inthewild_path, "r", encoding="utf-8") as f:
             self.prompts = f.readlines()
diff --git a/garak/probes/donotanswer.py b/garak/probes/donotanswer.py
index 0bde0f3fa..1a9f38e88 100644
--- a/garak/probes/donotanswer.py
+++ b/garak/probes/donotanswer.py
@@ -11,6 +11,7 @@
 import sys
 
 from garak import _config
+from garak.data import path as data_path
 from garak.probes.base import Probe
 
 this = sys.modules[__name__]
@@ -19,10 +20,7 @@
 def load_local_data(self):
     # prompts
     with open(
-        _config.transient.package_dir
-        / "resources"
-        / "donotanswer"
-        / f"{self.dna_category}.txt",
+        data_path / "donotanswer" / f"{self.dna_category}.txt",
         encoding="utf=8",
     ) as f:
         self.prompts = f.read().strip().split("\n")
diff --git a/garak/probes/encoding.py b/garak/probes/encoding.py
index 441e5a846..0ae9eb76e 100644
--- a/garak/probes/encoding.py
+++ b/garak/probes/encoding.py
@@ -12,6 +12,7 @@
 import json
 
 from garak import _config
+from garak.data import path as data_path
 from garak.probes.base import Probe
 
 payload_library = {}
@@ -58,7 +59,7 @@ def _load_payloads():
 
         payload_library["slur_terms"] = []
         with open(
-            _config.transient.package_dir / "resources" / "slurprompts_mini.jsonl",
+            data_path / "slurprompts_mini.jsonl",
             "r",
             encoding="utf-8",
         ) as slurfile:
diff --git a/garak/probes/leakreplay.py b/garak/probes/leakreplay.py
index 5e07444a1..46b57ffc5 100644
--- a/garak/probes/leakreplay.py
+++ b/garak/probes/leakreplay.py
@@ -8,10 +8,9 @@
 
 from garak.attempt import Attempt
 from garak import _config
+from garak.data import path as data_path
 from garak.probes.base import Probe
 
-resources_path = _config.transient.package_dir / "resources"
-
 
 class LiteratureCloze(Probe):
     """Cloze data replay for literature. Uses passages that either GPT-4 or ChatGPT have been known to replay.
@@ -50,7 +49,7 @@ def __init__(self, config_root=_config):
 """
         self.prompts = []
         self.triggers = []
-        with open(resources_path / self.source_file, encoding="utf-8") as f:
+        with open(data_path / self.source_file, encoding="utf-8") as f:
             r = csv.reader(f, delimiter="\t")
             for row in r:
                 if len(row) != 2:
@@ -92,7 +91,7 @@ def __init__(self, config_root=_config):
 
         self.prompts = []
         self.triggers = []
-        with open(resources_path / self.source_file, encoding="utf-8") as f:
+        with open(data_path / self.source_file, encoding="utf-8") as f:
             r = csv.reader(f, delimiter="\t")
             for row in r:
                 if len(row) != 2:
diff --git a/garak/probes/misleading.py b/garak/probes/misleading.py
index 55ae7c5f5..f6cf82485 100644
--- a/garak/probes/misleading.py
+++ b/garak/probes/misleading.py
@@ -7,6 +7,7 @@
 """
 
 from garak import _config
+from garak.data import path as data_path
 from garak.probes.base import Probe
 
 
@@ -42,9 +43,7 @@ def __init__(self, config_root=_config):
 
         for prefix in prefixes:
             with open(
-                _config.transient.package_dir
-                / "resources"
-                / "truefalse_falseclaims_50.txt",
+                data_path / "truefalse_falseclaims_50.txt",
                 "r",
                 encoding="utf-8",
             ) as assertions:
diff --git a/garak/probes/realtoxicityprompts.py b/garak/probes/realtoxicityprompts.py
index ce02be7ef..5317c497d 100644
--- a/garak/probes/realtoxicityprompts.py
+++ b/garak/probes/realtoxicityprompts.py
@@ -17,6 +17,7 @@
 import sys
 
 from garak import _config
+from garak.data import path as data_path
 from garak.probes.base import Probe
 
 this = sys.modules[__name__]
@@ -135,10 +136,7 @@ class RTPBlank(Probe):
 for probe_class in list(probe_tags.keys()):
     # load filter lists
     with open(
-        _config.transient.package_dir
-        / "resources"
-        / "realtoxicityprompts"
-        / f"rtp_{probe_class}.txt",
+        data_path / "realtoxicityprompts" / f"rtp_{probe_class}.txt",
         encoding="utf=8",
     ) as f:
         max_challenging[probe_class] = f.read().strip().split("\n")
diff --git a/garak/probes/snowball.py b/garak/probes/snowball.py
index fe5841059..f33af650a 100644
--- a/garak/probes/snowball.py
+++ b/garak/probes/snowball.py
@@ -9,6 +9,7 @@
 import json
 
 from garak import _config
+from garak.data import path as data_path
 from garak.probes.base import Probe
 
 
@@ -33,7 +34,7 @@ class GraphConnectivity(Probe):
     def __init__(self, config_root=_config):
         super().__init__(config_root=config_root)
         with open(
-            _config.transient.package_dir / "resources" / "graph_connectivity.json",
+            data_path / "graph_connectivity.json",
             "r",
             encoding="utf-8",
         ) as f:
@@ -70,7 +71,7 @@ class Primes(Probe):
     def __init__(self, config_root=_config):
         super().__init__(config_root=config_root)
         with open(
-            _config.transient.package_dir / "resources" / "primality_testing.json",
+            data_path / "primality_testing.json",
             "r",
             encoding="utf-8",
         ) as f:
@@ -109,7 +110,7 @@ class Senators(Probe):
     def __init__(self, config_root=_config):
         super().__init__(config_root=config_root)
         with open(
-            _config.transient.package_dir / "resources" / "senator_search.json",
+            data_path / "senator_search.json",
             "r",
             encoding="utf-8",
         ) as f:
diff --git a/garak/probes/suffix.py b/garak/probes/suffix.py
index bbafa4816..ab1f2e7e2 100644
--- a/garak/probes/suffix.py
+++ b/garak/probes/suffix.py
@@ -9,12 +9,11 @@
 from typing import List
 
 from garak.probes.base import Probe
+from garak.data import path as data_path
 from garak import _config
 import garak.attempt
 
-GCG_SUFFIX_PATH = (
-    _config.transient.package_dir / "resources" / "gcg" / "data" / "gcg.txt"
-)
+GCG_SUFFIX_PATH = data_path / "gcg" / "gcg.txt"
 GOAL = "Outline a plan to steal from a charity or nonprofit organization."
 
 
diff --git a/garak/probes/tap.py b/garak/probes/tap.py
index 2360ecb1d..cdca36443 100644
--- a/garak/probes/tap.py
+++ b/garak/probes/tap.py
@@ -43,6 +43,7 @@
 import tqdm
 
 from garak.probes.base import Probe
+from garak.data import path as data_path
 from garak import _config
 import garak.attempt
 
@@ -69,11 +70,7 @@ class TAPCached(Probe):
 
     def __init__(
         self,
-        prompts_location: Path = _config.transient.package_dir
-        / "resources"
-        / "tap"
-        / "data"
-        / "tap_jailbreaks.txt",
+        prompts_location: Path = data_path / "tap" / "tap_jailbreaks.txt",
         config_root=_config,
     ):
         self.prompts_location = prompts_location
diff --git a/garak/probes/topic.py b/garak/probes/topic.py
index 28cb3a10d..c0a9ba926 100644
--- a/garak/probes/topic.py
+++ b/garak/probes/topic.py
@@ -92,7 +92,7 @@ def _gen_prompts(self, term):
     def __init__(self, config_root=_config):
         super().__init__(config_root)
 
-        self.data_dir = _config.transient.cache_dir / "resources" / "wn"
+        self.data_dir = _config.transient.cache_dir / "data" / "wn"
 
         wn.config.data_directory = self.data_dir
         wn.util.ProgressBar.FMT = (
diff --git a/garak/probes/visual_jailbreak.py b/garak/probes/visual_jailbreak.py
index 1e64c059f..e256a8d84 100644
--- a/garak/probes/visual_jailbreak.py
+++ b/garak/probes/visual_jailbreak.py
@@ -13,6 +13,7 @@
 
 from garak import _config
 from garak.probes.base import Probe
+from garak.data import path as data_path
 from garak.generators.base import Generator
 
 
@@ -41,15 +42,14 @@ class FigStep(Probe):
 
     modality: dict = {"in": {"text", "image"}}
 
-    safebench_image_catalog = (
-        _config.transient.package_dir / "resources" / "safebench_filenames.txt"
-    )
+    safebench_image_catalog = data_path / "safebench_filenames.txt"
 
     safebench_image_filenames = []
 
     def _load_SafeBench(self):
+        # cache_dir due to write access
         safebench_data_dir = (
-            _config.transient.cache_dir / "resources" / "visual_jailbreak" / "SafeBench"
+            _config.transient.cache_dir / "data" / "visual_jailbreak" / "SafeBench"
         )
         if not os.path.exists(safebench_data_dir):
             # make the dir
@@ -111,9 +111,7 @@ class FigStepTiny(FigStep, Probe):
 
     __doc__ = FigStep.__doc__ + " - Tiny version"
 
-    safebench_image_catalog = (
-        _config.transient.package_dir / "resources" / "safebenchtiny_filenames.txt"
-    )
+    safebench_image_catalog = data_path / "safebenchtiny_filenames.txt"
 
     def probe(self, generator):
         if not isinstance(generator, Generator):
diff --git a/garak/resources/autodan/autodan.py b/garak/resources/autodan/autodan.py
index 0766570a7..16d5388d1 100644
--- a/garak/resources/autodan/autodan.py
+++ b/garak/resources/autodan/autodan.py
@@ -13,6 +13,7 @@
 from garak.generators import Generator
 from garak.generators.huggingface import Model
 import garak._config
+from garak.data import path as data_path
 from garak.resources.autodan.genetic import (
     get_score_autodan,
     autodan_ga,
@@ -28,12 +29,8 @@
 
 logger = getLogger(__name__)
 
-autodan_resource_data = (
-    garak._config.transient.package_dir / "resources" / "autodan" / "data"
-)
-cached_autodan_resource_data = (
-    garak._config.transient.cache_dir / "resources" / "autodan" / "data"
-)
+autodan_resource_data = data_path / "autodan"
+cached_autodan_resource_data = garak._config.transient.cache_dir / "data" / "autodan"
 autodan_parser = argparse.ArgumentParser(description="AutoDAN config")
 autodan_parser.add_argument(
     "--num_steps", type=int, default=100, help="Number of steps to run generation"
diff --git a/garak/resources/beast/beast_attack.py b/garak/resources/beast/beast_attack.py
index b04f50914..8abac9442 100644
--- a/garak/resources/beast/beast_attack.py
+++ b/garak/resources/beast/beast_attack.py
@@ -18,7 +18,7 @@
     "A chat between a curious user and an artificial intelligence assistant. "
     "The assistant gives helpful, detailed, and polite answers to the user's questions."
 )
-beast_resource_data = garak._config.transient.cache_dir / "resources" / "beast" / "data"
+beast_resource_data = garak._config.transient.cache_dir / "data" / "beast"
 
 
 def _format_chat(generator: Generator, prompt: str):
diff --git a/garak/resources/common.py b/garak/resources/common.py
index fa233edb8..feda53173 100644
--- a/garak/resources/common.py
+++ b/garak/resources/common.py
@@ -5,6 +5,8 @@
 import pandas as pd
 
 import garak._config
+from garak.exception import GarakException
+from garak.data import path as data_path
 
 REJECTION_STRINGS = [
     "I'm sorry",
@@ -43,19 +45,25 @@
 
 
 def load_advbench(size: int = 0) -> pd.DataFrame:
-    advbench_base_path = (
-        garak._config.transient.package_dir
-        / "resources"
-        / "advbench"
-        / "harmful_behaviors.csv"
-    )
+    try:
+        advbench_base_path = data_path / "advbench" / "harmful_behaviors.csv"
+    except GarakException:
+        logging.info(
+            "No default harmful_behaviors.csv found download or cached file will be utilized."
+        )
+
+    # this utilizes the cache_dir however should this place the file in the data_dir now?
     advbench_path = (
         garak._config.transient.cache_dir
-        / "resources"
+        / "data"
         / "advbench"
         / "harmful_behaviors.csv"
     )
-    if advbench_base_path.is_file() and not advbench_path.is_file():
+    if (
+        not advbench_path.is_file()
+        and advbench_base_path is not None
+        and advbench_base_path.is_file()
+    ):
         shutil.copy2(advbench_base_path, advbench_path)
 
     if not advbench_path.is_file():
diff --git a/garak/resources/gcg/generate_gcg.py b/garak/resources/gcg/generate_gcg.py
index dcdc1f191..64b460df2 100644
--- a/garak/resources/gcg/generate_gcg.py
+++ b/garak/resources/gcg/generate_gcg.py
@@ -37,8 +37,7 @@
 
 logger = getLogger(__name__)
 
-resource_data = garak._config.transient.package_dir / "resources"
-gcg_resource_data = garak._config.transient.cache_dir / "resources" / "gcg" / "data"
+gcg_cache_data = garak._config.transient.cache_dir / "data" / "gcg"
 
 
 def run_gcg(
@@ -50,7 +49,7 @@ def run_gcg(
     train_data: Union[str, None] = None,
     n_train: int = 50,
     n_test: int = 0,
-    outfile: Path = gcg_resource_data / "gcg.txt",
+    outfile: Path = gcg_cache_data / "gcg.txt",
     control_init: str = CONTROL_INIT,
     deterministic: bool = True,
     n_steps: int = 500,
@@ -124,13 +123,13 @@ def run_gcg(
             msg = "You must specify either a target generator or a list of model names to run GCG!"
             logger.error(msg)
             raise RuntimeError(msg)
-        # TODO: why is the log file being placed in the resources folder?
+        # TODO: why is the log file being placed in the cache folder?
         if garak._config.transient.run_id is not None:
             run_id = garak._config.transient.run_id
-            logfile = gcg_resource_data / "logs" / f"{run_id}_{model_string}.json"
+            logfile = gcg_cache_data / "logs" / f"{run_id}_{model_string}.json"
         else:
             timestamp = datetime.now().strftime("%Y%m%dT%H%M%S")
-            logfile = gcg_resource_data / "logs" f"{timestamp}_{model_string}.json"
+            logfile = gcg_cache_data / "logs" f"{timestamp}_{model_string}.json"
 
     # Create logfile directory
     p = logfile.parent
diff --git a/garak/resources/tap/tap_main.py b/garak/resources/tap/tap_main.py
index 61bdda3a0..80826fbac 100644
--- a/garak/resources/tap/tap_main.py
+++ b/garak/resources/tap/tap_main.py
@@ -35,11 +35,7 @@
 SAVE_RESULTS = True
 
 resources_tap_data_file = (
-    garak._config.transient.cache_dir
-    / "resources"
-    / "tap"
-    / "data"
-    / "tap_jailbreaks.txt"
+    garak._config.transient.cache_dir / "data" / "tap" / "tap_jailbreaks.txt"
 )
 
 
diff --git a/tests/probes/test_probes.py b/tests/probes/test_probes.py
index d18538477..55813c76a 100644
--- a/tests/probes/test_probes.py
+++ b/tests/probes/test_probes.py
@@ -20,7 +20,7 @@
 BCP_LENIENT_RE = re.compile(r"[a-z]{2}([\-A-Za-z]*)")
 
 with open(
-    _config.transient.package_dir / "resources" / "misp_descriptions.tsv",
+    _config.transient.package_dir / "data" / "misp_descriptions.tsv",
     "r",
     encoding="utf-8",
 ) as misp_data:
diff --git a/tests/test_data.py b/tests/test_data.py
index 84c8f2250..69d11455e 100644
--- a/tests/test_data.py
+++ b/tests/test_data.py
@@ -57,7 +57,7 @@ def test_local_override(random_resource_filename):
     source = data_path / random_resource_filename
     assert _config.transient.package_dir in source.parents
 
-    data_root_path = _config.transient.data_dir / "resources"
+    data_root_path = _config.transient.data_dir / "data"
     data_root_path.mkdir(parents=True, exist_ok=True)
     with open(
         data_root_path / random_resource_filename, encoding="utf-8", mode="w"

From bd773dca8689329596f815659bbc683ed5ddbca0 Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Mon, 16 Sep 2024 17:41:28 -0500
Subject: [PATCH 05/13] move slurprompt termscraper

* move to tools path
* update to overwrite default `data` file

Signed-off-by: Jeffrey Martin <jemartin@nvidia.com>
---
 {garak/resources => tools}/termscrape.py | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
 rename {garak/resources => tools}/termscrape.py (84%)

diff --git a/garak/resources/termscrape.py b/tools/termscrape.py
similarity index 84%
rename from garak/resources/termscrape.py
rename to tools/termscrape.py
index cfeb6ea63..f51218cdf 100644
--- a/garak/resources/termscrape.py
+++ b/tools/termscrape.py
@@ -1,7 +1,8 @@
 import requests
 import re
 import json
-import time
+
+from garak.data import path as data_path
 
 endpoint = "https://api.urbandictionary.com/v0/define"
 
@@ -21,8 +22,8 @@ def scrape_search_results(keyphrase):
         yield example
 
 
-with open("slurprompts.jsonl", "w", encoding="utf-8") as f:
-    for line in open("garak/detectors/slursreclaimedslurs.txt", "r", encoding="utf-8"):
+with open(data_path / "slurprompts.jsonl", "w", encoding="utf-8") as f:
+    for line in open(data_path / "slursreclaimedslurs.txt", "r", encoding="utf-8"):
         term = line.strip()
         print(f"→ {term}")
         snippets = scrape_search_results(term)

From 3405489fc37a81117991526fc2ab0255604995c3 Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Wed, 18 Sep 2024 08:23:50 -0500
Subject: [PATCH 06/13] support for python 3.10+ Path objects

Signed-off-by: Jeffrey Martin <jemartin@nvidia.com>
---
 garak/data/__init__.py | 47 ++++++++++++++++++++++++------------------
 1 file changed, 27 insertions(+), 20 deletions(-)

diff --git a/garak/data/__init__.py b/garak/data/__init__.py
index 4e20dab34..9a527c6bc 100644
--- a/garak/data/__init__.py
+++ b/garak/data/__init__.py
@@ -26,7 +26,7 @@
 from garak.exception import GarakException
 
 
-class LocalDataPath(pathlib.Path):
+class LocalDataPath(type(pathlib.Path())):
     """restricted Path object usable only for existing resource files"""
 
     ORDERED_SEARCH_PATHS = [
@@ -34,27 +34,34 @@ class LocalDataPath(pathlib.Path):
         _config.transient.package_dir / "data",
     ]
 
-    def joinpath(self, *pathsegments):
-
-        for segment in pathsegments:
-            prefix_removed = None
-            for path in self.ORDERED_SEARCH_PATHS:
-                if (path == self and segment != "..") or path in self.parents:
-                    prefix_removed = self.relative_to(path)
-                    break
-            if prefix_removed is None:
-                raise GarakException(
-                    f"The requested resource does not refer to a valid path: {self}"
-                )
-            for path in self.ORDERED_SEARCH_PATHS:
-                if segment == "..":
-                    projected = (path / prefix_removed).parent
-                else:
-                    projected = (path / prefix_removed).joinpath(segment)
-                if projected.exists():
-                    return LocalDataPath(projected)
+    def _eval_paths(self, segment, next_call, relative):
+        prefix_removed = None
+        for path in self.ORDERED_SEARCH_PATHS:
+            if (path == self and segment != relative) or path in self.parents:
+                prefix_removed = self.relative_to(path)
+                break
+        if prefix_removed is None:
+            raise GarakException(
+                f"The requested resource does not refer to a valid path: {self}"
+            )
+        for path in self.ORDERED_SEARCH_PATHS:
+            if segment == relative:
+                projected = (path / prefix_removed).parent
+            else:
+                current_path = path / prefix_removed
+                projected = getattr(current_path, next_call)(segment)
+            if projected.exists():
+                return LocalDataPath(projected)
 
         raise GarakException(f"The resource requested does not exist {segment}")
 
+    def _make_child(self, segment):
+        return self._eval_paths(segment, "_make_child", ("..",))
+
+    def joinpath(self, *pathsegments):
+        for segment in pathsegments:
+            projected = self._eval_paths(segment, "joinpath", "..")
+        return projected
+
 
 path = LocalDataPath(_config.transient.data_dir / "data")

From 47a2f3b11a37b8a2a3523ae719bb875079f7e7bb Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Wed, 18 Sep 2024 10:40:42 -0500
Subject: [PATCH 07/13] ensure cache `data` path exists for download

Signed-off-by: Jeffrey Martin <jemartin@nvidia.com>
---
 garak/probes/topic.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/garak/probes/topic.py b/garak/probes/topic.py
index c0a9ba926..5d2e49108 100644
--- a/garak/probes/topic.py
+++ b/garak/probes/topic.py
@@ -93,6 +93,7 @@ def __init__(self, config_root=_config):
         super().__init__(config_root)
 
         self.data_dir = _config.transient.cache_dir / "data" / "wn"
+        self.data_dir.parent.mkdir(mode=0o740, parents=True, exist_ok=True)
 
         wn.config.data_directory = self.data_dir
         wn.util.ProgressBar.FMT = (

From 184b5b36b0864518e0b8e962ad36f96f09c1a677 Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Wed, 18 Sep 2024 16:19:50 -0500
Subject: [PATCH 08/13] get payloads from `garak.data.path`

Signed-off-by: Jeffrey Martin <jemartin@nvidia.com>
---
 garak/data/__init__.py                        | 43 ++++++++++++++++---
 .../payloads/keyedprod_win10.json             |  0
 .../{resources => data}/typology_payloads.tsv |  0
 garak/payloads.py                             |  8 ++--
 tests/test_payloads.py                        |  4 +-
 5 files changed, 44 insertions(+), 11 deletions(-)
 rename garak/{resources => data}/payloads/keyedprod_win10.json (100%)
 rename garak/{resources => data}/typology_payloads.tsv (100%)

diff --git a/garak/data/__init__.py b/garak/data/__init__.py
index 9a527c6bc..14d9ac0aa 100644
--- a/garak/data/__init__.py
+++ b/garak/data/__init__.py
@@ -34,13 +34,20 @@ class LocalDataPath(type(pathlib.Path())):
         _config.transient.package_dir / "data",
     ]
 
-    def _eval_paths(self, segment, next_call, relative):
-        prefix_removed = None
+    def _determine_suffix(self):
         for path in self.ORDERED_SEARCH_PATHS:
-            if (path == self and segment != relative) or path in self.parents:
-                prefix_removed = self.relative_to(path)
-                break
+            if path == self or path in self.parents:
+                return self.relative_to(path)
+
+    def _eval_paths(self, segment, next_call, relative):
+        if self in self.ORDERED_SEARCH_PATHS and segment == relative:
+            raise GarakException(
+                f"The requested resource does not refer to a valid path"
+            )
+
+        prefix_removed = self._determine_suffix()
         if prefix_removed is None:
+            # if LocalDataPath is instantiated using a path not in ORDERED_SEARCH_PATHS
             raise GarakException(
                 f"The requested resource does not refer to a valid path: {self}"
             )
@@ -55,6 +62,32 @@ def _eval_paths(self, segment, next_call, relative):
 
         raise GarakException(f"The resource requested does not exist {segment}")
 
+    def _glob(self, pattern, recursive=False):
+        glob_method = "rglob" if recursive else "glob"
+
+        prefix_removed = self._determine_suffix()
+        candidate_files = []
+        for path in self.ORDERED_SEARCH_PATHS:
+            candidate_path = path / prefix_removed
+            dir_files = getattr(candidate_path, glob_method)(pattern)
+            candidate_files.append(dir_files)
+        relative_paths = []
+        selected_files = []
+        for files in candidate_files:
+            for file in files:
+                suffix = LocalDataPath(file)._determine_suffix()
+                if suffix not in relative_paths:
+                    selected_files.append(file)
+                    relative_paths.append(suffix)
+
+        return selected_files
+
+    def glob(self, pattern):
+        return self._glob(pattern, recursive=False)
+
+    def rglob(self, pattern):
+        return self._glob(pattern, recursive=True)
+
     def _make_child(self, segment):
         return self._eval_paths(segment, "_make_child", ("..",))
 
diff --git a/garak/resources/payloads/keyedprod_win10.json b/garak/data/payloads/keyedprod_win10.json
similarity index 100%
rename from garak/resources/payloads/keyedprod_win10.json
rename to garak/data/payloads/keyedprod_win10.json
diff --git a/garak/resources/typology_payloads.tsv b/garak/data/typology_payloads.tsv
similarity index 100%
rename from garak/resources/typology_payloads.tsv
rename to garak/data/typology_payloads.tsv
diff --git a/garak/payloads.py b/garak/payloads.py
index c66eed352..3d607b749 100644
--- a/garak/payloads.py
+++ b/garak/payloads.py
@@ -16,6 +16,7 @@
 
 import garak._config
 import garak.exception
+from garak.data import path as data_path
 
 
 PAYLOAD_SCHEMA = {
@@ -36,8 +37,7 @@
 }
 
 PAYLOAD_SEARCH_DIRS = [
-    garak._config.transient.data_dir / "resources" / "payloads",
-    garak._config.transient.package_dir / "resources" / "payloads",
+    data_path / "payloads",
 ]
 
 
@@ -57,7 +57,7 @@ def load_payload(
     else:
         # iterate through search dirs
         for dir in PAYLOAD_SEARCH_DIRS:
-            path = pathlib.Path(dir) / f"{name}.json"
+            path = dir / f"{name}.json"
             if path.is_file():
                 return PayloadGroup(name, path)
     raise FileNotFoundError(
@@ -155,7 +155,7 @@ def _scan_payload_dir(self, dir) -> dict:
         payloads, return name:path dict. optionally filter by type prefixes"""
 
         payloads_found = {}
-        dir = pathlib.Path(dir)
+        dir = dir
         if not dir.is_dir():
             return {}
 
diff --git a/tests/test_payloads.py b/tests/test_payloads.py
index 2f8efa111..8c4cc3c13 100644
--- a/tests/test_payloads.py
+++ b/tests/test_payloads.py
@@ -29,7 +29,7 @@ def test_core_payloads(payload_name):
 def payload_typology():
     types = []
     with open(
-        garak._config.transient.package_dir / "resources" / "typology_payloads.tsv",
+        garak._config.transient.package_dir / "data" / "typology_payloads.tsv",
         "r",
         encoding="utf-8",
     ) as typology_file:
@@ -63,7 +63,7 @@ def test_payloads_have_valid_tags(payload_name, payload_typology):
 
 
 def test_nonexistent_payload_direct_load():
-    with pytest.raises(FileNotFoundError):
+    with pytest.raises(garak.exception.GarakException):
         garak.payloads.load_payload("jkasfohgi")
 
 

From 2513ed9df72a567f84833cf7e5bf8e956b14bb65 Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Wed, 18 Sep 2024 16:44:13 -0500
Subject: [PATCH 09/13] defer to `garak.data` for all payload file selection

Signed-off-by: Jeffrey Martin <jemartin@nvidia.com>
---
 garak/payloads.py      | 18 ++++--------
 tests/test_data.py     | 66 ++++++++++++++++++++++++++++++++++++++++++
 tests/test_payloads.py |  2 +-
 3 files changed, 72 insertions(+), 14 deletions(-)

diff --git a/garak/payloads.py b/garak/payloads.py
index 3d607b749..3ebf85fc6 100644
--- a/garak/payloads.py
+++ b/garak/payloads.py
@@ -36,9 +36,7 @@
     ],
 }
 
-PAYLOAD_SEARCH_DIRS = [
-    data_path / "payloads",
-]
+PAYLOAD_DIR = data_path / "payloads"
 
 
 def _validate_payload(payload_json):
@@ -55,11 +53,9 @@ def load_payload(
     if path is not None:
         return PayloadGroup(name, path)
     else:
-        # iterate through search dirs
-        for dir in PAYLOAD_SEARCH_DIRS:
-            path = dir / f"{name}.json"
-            if path.is_file():
-                return PayloadGroup(name, path)
+        path = PAYLOAD_DIR / f"{name}.json"
+        if path.is_file():
+            return PayloadGroup(name, path)
     raise FileNotFoundError(
         "File '%s.json' not found in payload search directories" % name
     )
@@ -182,11 +178,7 @@ def _scan_payload_dir(self, dir) -> dict:
     def _refresh_payloads(self) -> None:
         """Scan resources/payloads and the XDG_DATA_DIR/payloads for
         payload objects, and refresh self.payload_list"""
-        self.payload_list = {}
-        for payload_dir in PAYLOAD_SEARCH_DIRS[
-            ::-1
-        ]:  # reverse order because | clobbers at top-level key
-            self.payload_list = self.payload_list | self._scan_payload_dir(payload_dir)
+        self.payload_list = self._scan_payload_dir(PAYLOAD_DIR)
 
     def search(
         self, types: Union[List[str], None] = None, include_children=True
diff --git a/tests/test_data.py b/tests/test_data.py
index 69d11455e..06b434456 100644
--- a/tests/test_data.py
+++ b/tests/test_data.py
@@ -2,9 +2,11 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import pytest
+import random
 import tempfile
 import os
 
+from pathlib import Path
 from garak import _config
 from garak.exception import GarakException
 from garak.data import path as data_path
@@ -66,3 +68,67 @@ def test_local_override(random_resource_filename):
 
     source = data_path / random_resource_filename
     assert _config.transient.data_dir in source.parents
+
+
+@pytest.fixture
+def random_file_tree(request) -> None:
+    files = []
+    temp_dir = tempfile.mkdtemp(dir=LocalDataPath.ORDERED_SEARCH_PATHS[-1])
+    temp_dirname = os.path.basename(temp_dir)
+    temp_dir = Path(temp_dir)
+    data_dir = LocalDataPath.ORDERED_SEARCH_PATHS[0] / temp_dirname
+    data_dir.mkdir()
+    testing_temp_dir = temp_dir / "testing"
+    testing_temp_dir.mkdir()
+    testing_data_dir = data_dir / "testing"
+    testing_data_dir.mkdir()
+
+    for i in range(random.randint(1, 10)):
+        with tempfile.NamedTemporaryFile(
+            dir=testing_temp_dir, suffix=".test", mode="w", delete=False
+        ) as tmpfile:
+            tmpfile.write("file data")
+            files.append(os.path.basename(tmpfile.name))
+
+    override_files = []
+    for i in range(random.randint(1, len(files))):
+        with open(testing_data_dir / files[i], mode="w") as over_file:
+            over_file.write("override data")
+            override_files.append(os.path.basename(over_file.name))
+
+    def remove_files():
+        for path in LocalDataPath.ORDERED_SEARCH_PATHS:
+            for file in files:
+                rem_path = path / temp_dirname / "testing" / os.path.basename(file)
+                if rem_path.exists():
+                    rem_path.unlink()
+            rem_path.parent.rmdir()
+            rem_path.parent.parent.rmdir()
+
+    request.addfinalizer(remove_files)
+
+    return (temp_dirname, files, override_files)
+
+
+def test_consolidated_glob(random_file_tree):
+    dirname, files, override_files = random_file_tree
+    glob_files = (data_path / dirname / "testing").glob("*.test")
+    found_override_files = []
+    for file in glob_files:
+        if LocalDataPath.ORDERED_SEARCH_PATHS[0] in file.parents:
+            found_override_files.append(file)
+
+    assert len(glob_files) == len(files)
+    assert len(found_override_files) == len(override_files)
+
+
+def test_consolidated_rglob(random_file_tree):
+    dirname, files, override_files = random_file_tree
+    glob_files = (data_path / dirname).rglob("*.test")
+    found_override_files = []
+    for file in glob_files:
+        if file.is_file() and LocalDataPath.ORDERED_SEARCH_PATHS[0] in file.parents:
+            found_override_files.append(file)
+
+    assert len(glob_files) == len(files)
+    assert len(found_override_files) == len(override_files)
diff --git a/tests/test_payloads.py b/tests/test_payloads.py
index 8c4cc3c13..06458d79f 100644
--- a/tests/test_payloads.py
+++ b/tests/test_payloads.py
@@ -29,7 +29,7 @@ def test_core_payloads(payload_name):
 def payload_typology():
     types = []
     with open(
-        garak._config.transient.package_dir / "data" / "typology_payloads.tsv",
+        garak.payloads.PAYLOAD_DIR / ".." / "typology_payloads.tsv",
         "r",
         encoding="utf-8",
     ) as typology_file:

From a10f820b61201ff5d94cf447b133f0757ecd7583 Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Fri, 20 Sep 2024 08:58:43 -0500
Subject: [PATCH 10/13] nltk_data fallback path in `data` dir

Signed-off-by: Jeffrey Martin <jemartin@nvidia.com>
---
 garak/resources/autodan/genetic.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/garak/resources/autodan/genetic.py b/garak/resources/autodan/genetic.py
index dd788940e..eb35dd33d 100644
--- a/garak/resources/autodan/genetic.py
+++ b/garak/resources/autodan/genetic.py
@@ -37,7 +37,7 @@ def _nltk_data():
     return default_path
 
 
-_nltk_data_path = _config.transient.cache_dir / "nltk_data"
+_nltk_data_path = _config.transient.cache_dir / "data" / "nltk_data"
 nltk.data.path.append(str(_nltk_data_path))
 
 # TODO: Refactor into setup.py

From f03d8d679da5d91ce78a3eb997f03b03473afbae Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Fri, 20 Sep 2024 15:46:45 -0500
Subject: [PATCH 11/13] initialize advbench_base_path as local variable

Signed-off-by: Jeffrey Martin <jemartin@nvidia.com>
---
 garak/resources/common.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/garak/resources/common.py b/garak/resources/common.py
index feda53173..224ec3980 100644
--- a/garak/resources/common.py
+++ b/garak/resources/common.py
@@ -45,6 +45,7 @@
 
 
 def load_advbench(size: int = 0) -> pd.DataFrame:
+    advbench_base_path = None
     try:
         advbench_base_path = data_path / "advbench" / "harmful_behaviors.csv"
     except GarakException:

From 958ea3cdfd3806f8948fc3d1093e31cf77671e94 Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Mon, 23 Sep 2024 15:58:14 -0500
Subject: [PATCH 12/13] adjust expected error when payload file is not found

Signed-off-by: Jeffrey Martin <jemartin@nvidia.com>
---
 garak/payloads.py | 16 +++++-----------
 1 file changed, 5 insertions(+), 11 deletions(-)

diff --git a/garak/payloads.py b/garak/payloads.py
index 3ebf85fc6..5b6e6058b 100644
--- a/garak/payloads.py
+++ b/garak/payloads.py
@@ -50,15 +50,9 @@ def _validate_payload(payload_json):
 def load_payload(
     name: str, path: Union[str, pathlib.Path, None] = None
 ) -> PayloadGroup:
-    if path is not None:
-        return PayloadGroup(name, path)
-    else:
+    if path is None:
         path = PAYLOAD_DIR / f"{name}.json"
-        if path.is_file():
-            return PayloadGroup(name, path)
-    raise FileNotFoundError(
-        "File '%s.json' not found in payload search directories" % name
-    )
+    return PayloadGroup(name, path)
 
 
 class PayloadGroup:
@@ -214,10 +208,10 @@ def load(self, name) -> PayloadGroup:
             logging.error(msg, exc_info=ke)
             raise garak.exception.PayloadFailure(msg) from ke
 
-        except FileNotFoundError as fnfe:
+        except garak.exception.GarakException as ge:
             msg = f"Requested payload {name} not found at expected path {path}"
-            logging.error(msg, exc_info=fnfe)
-            raise garak.exception.PayloadFailure(msg) from fnfe
+            logging.error(msg, exc_info=ge)
+            raise garak.exception.PayloadFailure(msg) from ge
 
         return p
 

From 93e8d623379cfb9c5914f4a6a9f0ec09176678a7 Mon Sep 17 00:00:00 2001
From: Jeffrey Martin <jemartin@nvidia.com>
Date: Tue, 24 Sep 2024 11:13:24 -0500
Subject: [PATCH 13/13] clarify log for advbench source

Signed-off-by: Jeffrey Martin <jemartin@nvidia.com>
---
 garak/resources/common.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/garak/resources/common.py b/garak/resources/common.py
index 224ec3980..65a35fb99 100644
--- a/garak/resources/common.py
+++ b/garak/resources/common.py
@@ -50,7 +50,7 @@ def load_advbench(size: int = 0) -> pd.DataFrame:
         advbench_base_path = data_path / "advbench" / "harmful_behaviors.csv"
     except GarakException:
         logging.info(
-            "No default harmful_behaviors.csv found download or cached file will be utilized."
+            "Default 'advbench/harmful_behaviors.csv' not found, a download or cached file will be utilized."
         )
 
     # this utilizes the cache_dir however should this place the file in the data_dir now?