refactor: scan_file() in versionscanner.py (#1226)

* refactor: scan_file() in versionscanner.py

* fix: isort and black

Author: peb-peb

cve_bin_tool/version_scanner.py

@@ -90,31 +90,17 @@ def remove_skiplist(self, skips):
     def print_checkers(self):
         self.logger.info(f'Checkers: {", ".join(self.checkers.keys())}')
 
-    def scan_file(self, filename):
-        """Scans a file to see if it contains any of the target libraries,
-        and whether any of those contain CVEs"""
-
-        self.logger.debug(f"Scanning file: {filename}")
-        self.total_scanned_files += 1
-
-        # Do not try to scan symlinks
-        if os.path.islink(filename):
-            return None
-
-        # Ensure filename is a file
-        if not os.path.isfile(filename):
-            self.logger.warning(f"Invalid file {filename} cannot be scanned")
-            return None
+    def is_executable(self, filename):
+        """check if file is an ELF binary file"""
 
-        # step 1: check if it's an ELF binary file
         if inpath("file"):
             # use system file if available (for performance reasons)
             output = subprocess.check_output(["file", filename])
             output = output.decode(sys.stdout.encoding)
 
             if "cannot open" in output:
                 self.logger.warning(f"Unopenable file {filename} cannot be scanned")
-                return None
+                return False
 
             if (
                 ("LSB " not in output)
@@ -126,11 +112,16 @@ def scan_file(self, filename):
                 and ("PKG-INFO: " not in output)
                 and ("METADATA: " not in output)
             ):
-                return None
+                return False
         # otherwise use python implementation of file
         elif not is_binary(filename):
-            return None
-        # parse binary file's strings
+            return False
+
+        return True, output
+
+    def parse_strings(self, filename):
+        """parse binary file's strings"""
+
         if inpath("strings"):
             # use "strings" on system if available (for performance)
             lines = (
@@ -140,7 +131,38 @@ def scan_file(self, filename):
             )
         else:
             # Otherwise, use python implementation
-            lines = Strings(filename).parse()
+            s = Strings(filename)
+            lines = s.parse()
+        return lines
+
+    def scan_file(self, filename):
+        """Scans a file to see if it contains any of the target libraries,
+        and whether any of those contain CVEs"""
+
+        self.logger.debug(f"Scanning file: {filename}")
+        self.total_scanned_files += 1
+
+        # Do not try to scan symlinks
+        if os.path.islink(filename):
+            return None
+
+        # Ensure filename is a file
+        if not os.path.isfile(filename):
+            self.logger.warning(f"Invalid file {filename} cannot be scanned")
+            return None
+
+        # check if it's an ELF binary file
+        try:
+            t, output = self.is_executable(filename)
+        except:
+            t = self.is_executable(filename)
+
+        if not t:
+            return None
+
+        # parse binary file's strings
+        lines = self.parse_strings(filename)
+
         #  If python package then strip the lines to avoid detecting other product strings
         if "PKG-INFO: " in output or "METADATA: " in output:
             lines = lines[1:3]