Added options to include/exclude files by name via regex

src/binwalk/core/module.py

@@ -354,7 +354,7 @@ class Module(object):
         self.target_file_list += self.extractor.pending
         self.extractor.pending = []
 
-        if self.target_file_list:
+        while self.target_file_list:
             next_target_file = self.target_file_list.pop(0)
 
             # Values in self.target_file_list are either already open files (BlockFile instances), or paths
@@ -364,8 +364,17 @@ class Module(object):
             else:
                 fp = next_target_file
 
-            self.status.clear()
-            self.status.total = fp.length
+            if not fp:
+                break
+            else:
+                if self.config.file_name_filter(fp) == False:
+                    fp.close()
+                    fp = None
+                    continue
+                else:
+                    self.status.clear()
+                    self.status.total = fp.length
+                    break
 
         if fp is not None:
             self.current_target_file_name = fp.path

src/binwalk/modules/general.py

@@ -2,6 +2,7 @@
 
 import io
 import os
+import re
 import sys
 import argparse
 import binwalk.core.idb
@@ -69,6 +70,16 @@ class General(Module):
                long='help',
                kwargs={'show_help' : True},
                description='Show help output'),
+        Option(short='a',
+               long='finclude',
+               type=str,
+               kwargs={'file_name_include_regex' : ""},
+               description='Only scan files whose names match this regex'),
+        Option(short='p',
+               long='fexclude',
+               type=str,
+               kwargs={'file_name_exclude_regex' : ""},
+               description='Do not scan files whose names match this regex'),
         Option(long=None,
                short=None,
                type=binwalk.core.common.BlockFile,
@@ -95,6 +106,8 @@ class General(Module):
         Kwarg(name='show_help', default=False),
         Kwarg(name='keep_going', default=False),
         Kwarg(name='subclass', default=io.FileIO),
+        Kwarg(name='file_name_include_regex', default=None),
+        Kwarg(name='file_name_exclude_regex', default=None),
     ]
 
     PRIMARY = False
@@ -110,6 +123,12 @@ class General(Module):
         self._open_target_files()
         self._set_verbosity()
 
+        # Build file name filter regex rules
+        if self.file_name_include_regex:
+            self.file_name_include_regex = re.compile(self.file_name_include_regex)
+        if self.file_name_exclude_regex:
+            self.file_name_exclude_regex = re.compile(self.file_name_exclude_regex)
+
         self.settings = binwalk.core.settings.Settings()
         self.display = binwalk.core.display.Display(log=self.log_file,
                                                     csv=self.csv,
@@ -135,6 +154,22 @@ class General(Module):
         if len(self.target_files) > 1 and not self.verbose:
             self.verbose = True
 
+    def file_name_filter(self, fp):
+        '''
+        Checks to see if a file should be scanned based on file name include/exclude filters.
+        Most useful for matryoshka scans where only certian files are desired.
+
+        @fp - An instances of binwalk.common.BlockFile
+
+        Returns True if the file should be scanned, False if not.
+        '''
+        if self.file_name_include_regex and not self.file_name_include_regex.search(fp.name):
+            return False
+        if self.file_name_exclude_regex and self.file_name_exclude_regex.search(fp.name):
+            return False
+
+        return True
+
     def open_file(self, fname, length=None, offset=None, swap=None, block=None, peek=None):
         '''
         Opens the specified file with all pertinent configuration settings.