Fixed cpio extraction bug; added extra extraction debugging statements

src/binwalk/modules/entropy.py

@@ -38,7 +38,7 @@ class Entropy(Module):
                    kwargs={'do_plot' : False},
                    description='Do not generate an entropy plot graph'),
             Option(short='Q',
-                   long='legend',
+                   long='nlegend',
                    kwargs={'show_legend' : False},
                    description='Omit the legend from the entropy plot graph'),
     ]

src/binwalk/modules/extractor.py

@@ -114,9 +114,15 @@ class Extractor(Module):
         else:
             size = r.size
 
+        if r.valid:
+            binwalk.core.common.debug("Extractor callback for %s:%d [%s & %s & %s]" % (r.file.name, r.offset, str(r.valid), str(r.display), str(r.extract)))
+        
         # Only extract valid results displayed to the user and marked for extraction
+        # TODO: Results excluded via -x/-y options should be marked as invalid; filtering on r.display means that
+        #       with -q specified, nothing gets extracted!
         if r.valid and r.display and r.extract:
             # Do the extraction
+            binwalk.core.common.debug("Attempting extraction...")
             (extraction_directory, dd_file) = self.extract(r.offset, r.description, r.file.name, size, r.name)
 
             # If the extraction was successful, self.extract will have returned the output directory and name of the dd'd file
@@ -541,6 +547,8 @@ class Extractor(Module):
         tmp = None
         retval = True
 
+        binwalk.core.common.debug("Running extractor '%s'" % str(cmd))
+
         try:
             if callable(cmd):
                 try:

src/binwalk/plugins/cpio.py

@@ -10,6 +10,7 @@ class CPIOPlugin(binwalk.core.plugin.Plugin):
     def pre_scan(self):
         # Be sure to re-set this at the beginning of every scan
         self.found_archive = False
+        self.found_archive_in_file = None
 
     def scan(self, result):
         if result.valid:
@@ -17,8 +18,9 @@ class CPIOPlugin(binwalk.core.plugin.Plugin):
             # Displaying each entry is useful, as it shows what files are contained in the archive,
             # but we only want to extract the archive when the first entry is found.
             if result.description.startswith('ASCII cpio archive'):
-                if not self.found_archive:
+                if not self.found_archive or self.found_archive_in_file != result.file.name:
                     # This is the first entry. Set found_archive and allow the scan to continue normally.
+                    self.found_archive_in_file = result.file.name
                     self.found_archive = True
                     result.extract = True
                 elif 'TRAILER!!!' in result.description:
@@ -29,3 +31,9 @@ class CPIOPlugin(binwalk.core.plugin.Plugin):
                     # The first entry has already been found and this is not the last entry, or the last entry 
                     # has not yet been found. Don't extract.
                     result.extract = False
+            else:
+                # If this was a valid non-CPIO archive result, reset these values; else, a previous
+                # false positive CPIO result could leave these set, causing a subsequent valid CPIO
+                # result to not be extracted.
+                self.found_archive = False
+                self.found_archive_in_file = None