update: rename data/misp_descriptions.tsv to data/tags.misp.tsv

garak/analyze/misp.py

@@ -27,13 +27,13 @@
 
 
 def misp_report(include_untagged: bool = True) -> None:
-    misp_resource_file = data_path / "misp_descriptions.tsv"
-    misp_descriptions = {}
+    misp_resource_file = data_path / "tags.misp.tsv"
+    tag_descriptions = {}
     if os.path.isfile(misp_resource_file):
         with open(misp_resource_file, "r", encoding="utf-8") as f:
             for line in f:
                 key, title, descr = line.strip().split("\t")
-                misp_descriptions[key] = (title, descr)
+                tag_descriptions[key] = (title, descr)
 
     probes_per_tag = defaultdict(list)
 
@@ -46,11 +46,11 @@ def misp_report(include_untagged: bool = True) -> None:
         if tags == [] and include_untagged:
             print(f"{plugin_name}: no tags defined")
         for tag in tags:
-            if tag not in misp_descriptions:
-                print(f"{plugin_name}: tag {tag} undefined in misp_descriptions.tsv")
+            if tag not in tag_descriptions:
+                print(f"{plugin_name}: tag {tag} undefined in garak/data/tags.misp.tsv")
             probes_per_tag[tag].append(plugin_name)
 
-    for misp_tag in misp_descriptions.keys():
+    for misp_tag in tag_descriptions.keys():
         if len(probes_per_tag[misp_tag]) == 0:
             print(f"{misp_tag}: zero probes testing this")
         else:

garak/analyze/report_digest.py

@@ -45,13 +45,13 @@
 about_z_template = templateEnv.get_template("digest_about_z.jinja")
 
 
-misp_resource_file = data_path / "misp_descriptions.tsv"
-misp_descriptions = {}
+misp_resource_file = data_path / "tags.misp.tsv"
+tag_descriptions = {}
 if os.path.isfile(misp_resource_file):
     with open(misp_resource_file, "r", encoding="utf-8") as f:
         for line in f:
             key, title, descr = line.strip().split("\t")
-            misp_descriptions[key] = (title, descr)
+            tag_descriptions[key] = (title, descr)
 
 
 def map_absolute_score(score: float) -> int:
@@ -239,8 +239,8 @@ def _get_group_info(probe_group, group_score, taxonomy, config=_config) -> dict:
         )
     elif probe_group != "other":
         probe_group_name = f"{taxonomy}:{probe_group}"
-        if probe_group_name in misp_descriptions:
-            probe_group_name, group_doc = misp_descriptions[probe_group_name]
+        if probe_group_name in tag_descriptions:
+            probe_group_name, group_doc = tag_descriptions[probe_group_name]
     else:
         probe_group_name = "Uncategorized"
 

tests/probes/test_probes.py

@@ -8,6 +8,7 @@
 
 from garak import _config, _plugins
 import garak.probes
+
 PROBES = [classname for (classname, active) in _plugins.enumerate_plugins("probes")]
 
 DETECTORS = [
@@ -20,7 +21,7 @@
 
 
 with open(
-    _config.transient.package_dir / "data" / "misp_descriptions.tsv",
+    _config.transient.package_dir / "data" / "tags.misp.tsv",
     "r",
     encoding="utf-8",
 ) as misp_data:
@@ -74,7 +75,9 @@ def test_probe_metadata(classname):
     p = _plugins.load_plugin(classname)
     assert isinstance(p.goal, str), "probe goals should be a text string"
     assert len(p.goal) > 0, "probes must state their general goal"
-    assert p.lang is not None and (p.lang == "*" or langcodes.tag_is_valid(p.lang)), "lang must be either * or a BCP47 code"
+    assert p.lang is not None and (
+        p.lang == "*" or langcodes.tag_is_valid(p.lang)
+    ), "lang must be either * or a BCP47 code"
     assert isinstance(
         p.doc_uri, str
     ), "probes should give a doc uri describing/citing the attack"

tests/test_data.py

@@ -50,7 +50,7 @@ def test_allow_relative_in_path():
 
 
 def test_known_resource_found():
-    known_filename = "misp_descriptions.tsv"
+    known_filename = "tags.misp.tsv"
     source = data_path / known_filename
     assert source.name == known_filename