Add the Hu CFGED algorithm for eval

mahaloz · Feb 8, 2024 · 1e7e4f3 · 1e7e4f3
1 parent 4310ccb
commit 1e7e4f3
Show file tree

Hide file tree

Showing 3 changed files with 21 additions and 2 deletions.
diff --git a/sailreval/__init__.py b/sailreval/__init__.py
@@ -22,6 +22,7 @@ class SAILR_METRICS:
     GED_MAX = "ged_max"
     GED_EXACT = "ged_exact"
     CFGED = "cfged"
+    HU_CFGED = "hu_cfged"
     GRAPH_SIZE = "graph_size"
     BLOCK_COUNT = "block_count"
     DEC_TIME = "dec_time"

diff --git a/sailreval/metrics/__init__.py b/sailreval/metrics/__init__.py
@@ -7,7 +7,7 @@
 from .nesting_count import count_nesting_ifs
 from .duplication import count_func_calls
 from .total_bools import count_if_bools
-from .ged_to_source import ged_upperbound_score, cfg_edit_distance, ged_max_score, block_count, ged_exact_score, graph_size
+from .ged_to_source import ged_upperbound_score, cfg_edit_distance, ged_max_score, block_count, ged_exact_score, graph_size, hu_cfged_score
 from .post_metrics import norm_cfged, zero_cfged, funcs_w_goto, goto_func_cfged
 
 from sailreval import SAILR_METRICS
@@ -36,6 +36,7 @@
     SAILR_METRICS.GED_UPPERBOUND: ged_upperbound_score,
     SAILR_METRICS.GED_MAX: ged_max_score,
     SAILR_METRICS.GED_EXACT: ged_exact_score,
+    SAILR_METRICS.HU_CFGED: hu_cfged_score,
 
     # Important Metrics
     SAILR_METRICS.GOTO_COUNT: count_total_gotos,

diff --git a/sailreval/metrics/ged_to_source.py b/sailreval/metrics/ged_to_source.py
@@ -3,7 +3,7 @@
 from typing import Dict
 
 from pyjoern.mapping import correct_decompiler_mappings, read_line_maps
-from cfgutils.similarity import ged_max, ged_upperbound, ged_exact
+from cfgutils.similarity import ged_max, ged_upperbound, ged_exact, hu_cfged
 from cfgutils.similarity import cfg_edit_distance as _cfg_edit_distance
 import networkx as nx
 
@@ -160,3 +160,20 @@ def block_count(
         return None
 
     return len(dec_cfg.nodes)
+
+
+#
+# More special case GED algorithms
+#
+
+def hu_cfged_score(
+    func_name, client, source_cfgs: Dict[str, nx.DiGraph] = None, dec_cfgs: Dict[str, nx.DiGraph] = None,
+    decompiler=None, binary_path=None, **kwargs
+):
+    if decompiler == "source":
+        return float(0)
+    source_cfg, dec_cfg = _verify_has_valid_graphs(func_name, client, source_cfgs, dec_cfgs, decompiler, binary_path)
+    if source_cfg is None or dec_cfg is None:
+        return None
+
+    return hu_cfged(dec_cfg, source_cfg)