improve blintdb performance

Signed-off-by: Aryan Rajoria <[email protected]>

Author: aryan-rajoria

blint/db.py

@@ -11,20 +11,20 @@
 
 import concurrent
 
-def get_bnames_ename(symbols_name):
-    """
-    Older algorithm with many false positives
-    """
-    bin_name = []
-    eid = get_export_id(symbols_name)
-    bid_list = get_bid_using_fid(eid)
-    if bid_list:
-        bin_name.extend(get_bname(bid) for bid in bid_list)
-    return bin_name
+# def get_bnames_ename(symbols_name):
+#     """
+#     Older algorithm with many false positives
+#     """
+#     bin_name = []
+#     eid = get_export_id(symbols_name)
+#     bid_list = get_bid_using_fid(eid)
+#     if bid_list:
+#         bin_name.extend(get_bname(bid) for bid in bid_list)
+#     return bin_name
 
 def return_binaries_detected(eid):
     """
-    New scoring algorithm
+    Current scoring algorithm
     """
     binaries_detected_dict = {}
     bid_list = get_bid_using_fid(eid)
@@ -64,6 +64,14 @@ def detect_binaries_utilized(sybmols_list) -> set:
 
     return binary_detected
 
+def get_bid_using_ename(export_name):
+    BLINTDB_LOC = os.getenv("BLINTDB_LOC")
+    with closing(sqlite3.connect(BLINTDB_LOC)) as connection:
+        with closing(connection.cursor()) as c:
+            c.execute("SELECT bid from BinariesExports where eid = (SELECT rowid from Exports where infunc=?)", (export_name,))
+            res = c.fetchall()
+        connection.commit()
+    return map(lambda x: x[0], res) if res else None
 
 def get_export_id(export_name):
     BLINTDB_LOC = os.getenv("BLINTDB_LOC")

blint/sbom.py

@@ -456,7 +456,9 @@ def process_exe_file(
         if os.environ.get("USE_BLINTDB", "") in ["1", "true"]:
             # utilize voting logic along with blitndb
             # we iterate through each symbol and try to find a match in the database
-            dynamic_symbols_list = metadata.get("dynamic_symbols", [])
+            print("Utilizing blint_db")
+            # TODO: utilize both symtab_symbols and dynamic_symbols
+            dynamic_symbols_list = metadata.get("symtab_symbols", [])
             binaries_detected = detect_binaries_utilized(dynamic_symbols_list)
             # adds the components in a similar way to dynamic entries
             for binary in binaries_detected: