From 5896a6acc3ac7bfb33915781263e2adf5ee62d5a Mon Sep 17 00:00:00 2001
From: Andrew Lamoureux <andrew@vector35.com>
Date: Sat, 11 May 2019 13:54:01 -0400
Subject: kaitai: fix fields with values 0 being hidden

---
 python/examples/kaitai/__main__.py  | 135 +++++++++--------
 python/examples/kaitai/kshelpers.py | 289 +++++++++++++++++++++++++++++++++++-
 2 files changed, 356 insertions(+), 68 deletions(-)

(limited to 'python')

diff --git a/python/examples/kaitai/__main__.py b/python/examples/kaitai/__main__.py
index 2accd2e5..33e8e549 100755
--- a/python/examples/kaitai/__main__.py
+++ b/python/examples/kaitai/__main__.py
@@ -9,7 +9,6 @@ import sys
 import types
 import importlib
 
-
 if sys.version_info[0] == 2:
 	import kaitaistruct
 	import kshelpers
@@ -37,70 +36,62 @@ LCYAN = '\033[1;36m'
 LGRAY = '\033[1;37m'
 
 def dump(obj, depth=0):
-	dump_exceptions = ['_root', '_parent', '_io', 'SEQ_FIELDS']
-
 	indent = '    '*depth
 
-	if isinstance(obj, kaitaistruct.KaitaiStruct):
-		fieldNames = []
-		for candidate in dir(obj):
-			if candidate != '_debug' and candidate.startswith('_'):
-				continue
-			if candidate in dump_exceptions:
-				continue
-			try:
-				if getattr(obj, candidate, False):
-					fieldNames.append(candidate)
-			except Exception:
-				pass
+	print(('%s'+PURPLE+'%s'+NORMAL) % (indent, repr(obj)))
 
-		for fieldName in fieldNames:
+	kshelpers.exercise(obj)
+	for fieldName in kshelpers.getFieldNamesPrint(obj):
+		subObj = None
+		try:
 			subObj = getattr(obj, fieldName)
-
-			if type(subObj) == types.MethodType:
-				pass
-			#elif type(subObj) == types.TypeType:
-			elif isinstance(subObj, type):
-				pass
-			elif fieldName == '_debug':
-				print(('%s._debug:'+RED+' %s'+NORMAL) % (indent, repr(subObj)))
-			#elif type(subObj) == types.ListType:
-			elif isinstance(subObj, list):
-				if len(subObj)>0 and isinstance(subObj[0], kaitaistruct.KaitaiStruct):
-					for i in range(len(subObj)):
-						print('%s.%s[%d]:' % (indent, fieldName, i))
-						dump(subObj[i], depth+1)
-				else:
-					print('%s.%s: %s' % (indent, fieldName, str(subObj)))
-			#elif type(subObj) == types.DictionaryType:
-			elif isinstance(subObj, dict):
-				print('%s.%s: %s' % (indent, fieldName, subObj))
-
-			#elif type(subObj) == types.StringType:
-			elif isinstance(subObj, str):
-				if len(subObj) <= 16:
-					print(('%s.%s: '+CYAN+'%s'+NORMAL) % (indent, fieldName, repr(subObj)))
-				else:
-					print(('%s.%s: '+CYAN+'%s...'+NORMAL+' 0x%X (%d) bytes total') % \
-						(indent, fieldName, repr(subObj[0:16]), len(subObj), len(subObj))
-					)
-
-			elif type(subObj) == int:
-				print(('%s.%s: '+YELLOW+'0x%X '+NORMAL+'('+YELLOW+'%d'+NORMAL+')') % (indent, fieldName, subObj, subObj))
-
-			elif str(type(subObj)).startswith('<enum '):
-				print(('%s.%s: '+'%s') % (indent, fieldName, repr(subObj)))
-
-			elif isinstance(subObj, kaitaistruct.KaitaiStruct):
-				print('%s.%s:' % (indent, fieldName))
-				dump(subObj, depth+1)
-
-			else:
-				print('%s.%s: %s' % (indent, fieldName, type(subObj)))
-	else:
-		print((PURPLE+'%s%s'+NORMAL) % (indent, repr(obj)))
-		#else:
-		#	print('%s%s: %s' % (indent, fieldName, repr(subObj)))
+		except Exception:
+			continue
+		if subObj == None:
+			continue
+
+		subObjStr = kshelpers.objToStr(subObj)
+
+		color = ''
+
+		if type(subObj) == types.MethodType:
+			pass
+		elif isinstance(subObj, type):
+			pass
+		elif fieldName == '_debug':
+			color = RED
+		elif isinstance(subObj, list):
+			pass
+		elif isinstance(subObj, dict):
+			pass
+		elif isinstance(subObj, str):
+			color = CYAN
+		elif isinstance(subObj, bytes):
+			color = CYAN
+		elif type(subObj) == int:
+			color = YELLOW
+		elif str(type(subObj)).startswith('<enum '):
+			color = GREEN
+			pass
+
+		if color:
+			print('%s.%s: %s%s%s' % (indent, fieldName, color, subObjStr, NORMAL))
+		else:
+			print('%s.%s: %s' % (indent, fieldName, subObjStr))
+
+	for fieldName in kshelpers.getFieldNamesDescend(obj):
+		subObj = getattr(obj, fieldName)
+
+		#print('recurring on: %s' % repr(subObj))
+
+		if isinstance(subObj, list):
+			for (i, tmp) in enumerate(subObj):
+				print('%s.%s[%d]:' % (indent, fieldName, i))
+				dump(subObj[i], depth+1)
+		else:
+			print('%s.%s:' % (indent, fieldName))
+			#print(dir(subObj))
+			dump(subObj, depth+1)
 
 if __name__ == '__main__':
 	if not sys.argv[1:]:
@@ -108,10 +99,17 @@ if __name__ == '__main__':
 
 	cmd = sys.argv[1]
 
-	if cmd == 'dump':
-		fpath = sys.argv[2]
-		parsed = kshelpers.parseFpath(fpath)
-		dump(parsed)
+	if cmd in ['dump0']:
+		kshelpers.setFieldExceptionLevel0()
+		dump(kshelpers.parseFpath(sys.argv[2]))
+
+	if cmd in ['dump1']:
+		kshelpers.setFieldExceptionLevel1()
+		dump(kshelpers.parseFpath(sys.argv[2]))
+
+	if cmd in ['dump', 'dump2']:
+		kshelpers.setFieldExceptionLevel2()
+		dump(kshelpers.parseFpath(sys.argv[2]))
 
 	if cmd == 'import':
 		print('importing every format...')
@@ -176,3 +174,10 @@ if __name__ == '__main__':
 		parsed = parse_io(kaitaiIo)
 		print(parsed.header.program_headers)
 
+	if cmd == 'pdb':
+		ksobj = kshelpers.parseFpath(sys.argv[2])
+		print('parsed object is in ksobj')
+		import pdb
+		pdb.set_trace()
+
+
diff --git a/python/examples/kaitai/kshelpers.py b/python/examples/kaitai/kshelpers.py
index 283c90d7..2d465346 100644
--- a/python/examples/kaitai/kshelpers.py
+++ b/python/examples/kaitai/kshelpers.py
@@ -4,10 +4,13 @@ import traceback
 
 import io
 import os
+import re
 import sys
-import struct
 import types
+import struct
+import binascii
 import importlib
+import collections
 
 from binaryninja import log
 
@@ -148,6 +151,286 @@ def parseIo(ioObj, ksModuleName=None):
 
 	return parsed
 
+#------------------------------------------------------------------------------
+# misc
+#------------------------------------------------------------------------------
+
+def objToStr(obj):
+	objType = type(obj)
+
+	# blacklist: functions, types, callables
+	#
+	if isinstance(obj, type):
+		#print('reject %s because its a type' % fieldName)
+		return '(type)'
+	elif hasattr(obj, '__call__'):
+		#print('reject %s because its a callable' % fieldName)
+		return '(callable)'
+
+	result = None
+
+	# whitelist: strings, unicodes, bytes, ints, bools, enums
+	#
+	if obj == None:
+		return 'None'
+	elif isinstance(obj, str):
+		if len(obj) > 8:
+			result = '%s...%s (0x%X==%d chars total)' % \
+				(repr(obj[0:8]), repr(obj[-1]), len(obj), len(obj))
+		else:
+			result = repr(obj)
+	elif isinstance(obj, bytes):
+		if len(obj) > 8:
+			result = binascii.hexlify(obj[0:8]).decode('utf-8') + '...' + \
+				('%02X' % obj[-1]) + ' (0x%X==%d bytes total)' % (len(obj), len(obj))
+		else:
+			result = binascii.hexlify(obj).decode('utf-8')
+	# note: bool needs to appear before int (else int determination will dominate)
+	elif isinstance(obj, bool):
+		result = '%s' % (obj)
+	elif isinstance(obj, int):
+		result = '0x%X (%d)' % (obj, obj)
+	elif str(objType).startswith('<enum '):
+		result = '%s' % (obj)
+	elif isinstance(obj, list):
+		result = repr(obj)
+	elif isinstance(obj, kaitaistruct.KaitaiStruct):
+		return re.match(r'^.*\.(\w+) object at ', repr(obj)).group(1)
+	elif isinstance(obj, kaitaistruct.KaitaiStream):
+		return re.match(r'^.*\.(\w+) object at ', repr(obj)).group(1)
+	elif isinstance(obj, collections.defaultdict):
+		# probably _debug
+		result = repr(obj)
+	else:
+		result = '(unknown type %s)' % (str(objType))
+
+	return result
+
+# access all fields that may be properties, which could compute internal results
+# (often '_m_XXX' fields)
+def exercise(ksobj):
+	for candidate in dir(ksobj):
+		#if candidate.startswith('_') and (not candidate.startswith('_m_')):
+		#	continue
+		try:
+			foo = getattr(ksobj, candidate, False)
+		except Exception:
+			pass
+
+# get the [start,end) data for a given field within a ks object
+#
+# abstracts away:
+# * the debug['arr'] stuff, you just give it 'foo' or 'foo[3]'
+# * the 'foo' vs. '_m_foo' complication, you just give it 'foo'
+#
+def getFieldRange(ksobj, fieldName:str, restrictedToRoot=False):
+	if restrictedToRoot:
+		if ksobj._io != ksobj._root._io:
+			return None
+
+	# does given kaitai object even have ._debug?
+	debug = None
+	try:
+		debug = getattr(ksobj, '_debug')
+	except Exception:
+		return None
+
+	# divide up if request field is list, like "foo[3]"
+	tmp = None
+	if fieldName.endswith(']'):
+		m = re.match(r'^(.*)\[(\d+)\]$', fieldName)
+		fieldName = m.group(1)
+		index = int(m.group(2))
+
+		tmp = None
+		if not fieldName.startswith('_m_'):
+			if '_m_'+fieldName in debug:
+				tmp = debug['_m_'+fieldName]['arr'][index]
+		if not tmp:
+			tmp = debug[fieldName]['arr'][index]
+	else:
+		tmp = None
+		if not fieldName.startswith('_m_'):
+			if '_m_'+fieldName in debug:
+				tmp = debug['_m_'+fieldName]
+		if not tmp:
+			tmp = debug[fieldName]
+
+	if not tmp:
+		return None
+
+	return (tmp['start'], tmp['end'])
+
+#------------------------------------------------------------------------------
+# kaitai object field control
+#------------------------------------------------------------------------------
+
+# certain fields in the kaitai python object we:
+# - should not DESCEND into (eg: ._parent, ._root)
+# - should not PRINT (eg: ._io)
+
+fieldDescendExceptions = ['_parent', '_root']
+fieldDescendExceptionsPatterns = []
+
+fieldPrintExceptions = []
+fieldPrintExceptionsPatterns = []
+
+def isFieldExceptionDescend(fieldName):
+	global fieldDescendExceptions, fieldDescendExceptionsPatterns
+
+	if fieldName in fieldDescendExceptions:
+		return True
+
+	for fep in fieldDescendExceptionsPatterns:
+		if re.match(fep, fieldName):
+			return True
+
+	return False
+
+def isFieldExceptionPrint(fieldName):
+	global fieldExceptions, fieldExceptionsPatterns
+
+	if fieldName in fieldPrintExceptions:
+		return True
+
+	for fep in fieldPrintExceptionsPatterns:
+		if re.match(fep, fieldName):
+			return True
+
+	return False
+
+def setFieldExceptionLevel0():
+	global fieldDescendExceptions, fieldDescendExceptionsPatterns
+	global fieldPrintExceptions, fieldPrintExceptionsPatterns
+	fieldDescendExceptions = ['_parent', '_root']
+	fieldDescendExceptionsPatterns = []
+	fieldPrintExceptions = []
+	fieldPrintExceptionsPatterns = []
+
+def setFieldExceptionLevel1():
+	global fieldDescendExceptions, fieldDescendExceptionsPatterns
+	global fieldPrintExceptions, fieldPrintExceptionsPatterns
+
+	setFieldExceptionLevel0()
+
+	fieldPrintExceptionsPatterns += [r'_raw__.*$']
+	fieldPrintExceptions += ['_is_le', '_root', '_parent', '_debug']
+	fieldPrintExceptions += ['_read', '_read_be', '_read_le']
+	fieldPrintExceptions += ['from_bytes', 'from_file', 'from_io']
+	fieldPrintExceptions += ['SEQ_FIELDS']
+
+def setFieldExceptionLevel2():
+	global fieldDescendExceptions, fieldDescendExceptionsPatterns
+	global fieldPrintExceptions, fieldPrintExceptionsPatterns
+
+	setFieldExceptionLevel1()
+
+	#fieldPrintExceptions += ['_io']
+	fieldPrintExceptionsPatterns += [r'^_m_.*$', r'^__.*$']
+	fieldDescendExceptionsPatterns += [r'^_m_.*$']
+
+#------------------------------------------------------------------------------
+# kaitai object exploring stuff
+#------------------------------------------------------------------------------
+
+# return all field names qualified for printing
+#
+def getFieldNamesPrint(ksobj):
+	result = []
+
+	for fieldName in dir(ksobj):
+		if isFieldExceptionPrint(fieldName):
+			continue
+
+		try:
+			subobj = getattr(ksobj, fieldName, False)
+
+			# do not return kaitai objects (are for descending, not printing)
+			if isinstance(subobj, kaitaistruct.KaitaiStruct):
+				continue
+			elif isinstance(subobj, list):
+				if len(subobj)<=0 or isinstance(subobj[0], kaitaistruct.KaitaiStruct):
+					continue
+
+			#print('%s is ok' % fieldName)
+			#print('%s is instance? %s' % (fieldName, isinstance(subobj, kaitaistruct.KaitaiStruct)))
+			result.append(fieldName)
+		except Exception:
+			pass
+
+	return result
+
+# return all field names required for descending
+#
+# IN:	kaitai object
+# OUT:	field names that are either:
+#		- kaitai objects
+#		- lists of kaitai objects
+#
+def getFieldNamesDescend(ksobj):
+	result = []
+
+	for fieldName in dir(ksobj):
+		if isFieldExceptionDescend(fieldName):
+			continue
+
+		try:
+			subobj = getattr(ksobj, fieldName, False)
+
+			if isinstance(subobj, kaitaistruct.KaitaiStruct):
+				result += [fieldName]
+			elif isinstance(subobj, list):
+				if len(subobj)>0 and isinstance(subobj[0], kaitaistruct.KaitaiStruct):
+					result += [fieldName]
+		except Exception:
+			pass
+
+	return result
+
+# compute all kaitai objects linked to from the given object
+#
+# IN:	kaitai object
+# OUT:	[obj0, obj1, obj2, ...]
+#
+def getLinkedKaitaiObjects(ksobj):
+	result = set()
+
+	for fieldName in getFieldNamesDescend(ksobj):
+		subobj = getattr(ksobj, fieldName, False)
+		if isinstance(subobj, list):
+			for tmp in subobj:
+				result.add(tmp)
+		else:
+			result.add(subobj)
+
+	return result
+
+# compute all kaitai objects linked to from the given object, and from its
+# descendents, and so on...
+def getLinkedKaitaiObjectsAll(ksobj, depth=0):
+	#if depth > 2:
+	#	return []
+
+	exercise(ksobj)
+
+	result = set([ksobj])
+
+	linkedObjects = getLinkedKaitaiObjects(ksobj)
+	for subobj in linkedObjects:
+		subResult = getLinkedKaitaiObjectsAll(subobj, depth+1)
+		result = result.union(subResult)
+
+	return result
+
+def getDepth(ksobj, depth=0):
+	result = depth
+
+	exercise(ksobj)
+	for subObj in getLinkedKaitaiObjects(ksobj):
+		result = max(result, getDepth(subObj, depth+1))
+
+	return result
+
 #------------------------------------------------------------------------------
 # Kaitai IO Wrapper
 #------------------------------------------------------------------------------
@@ -314,8 +597,8 @@ def buildQtree(ksobj):
 		if candidate in exceptions:
 			continue
 		try:
-			if getattr(ksobj, candidate, False):
-				fieldNames.add(candidate)
+			getattr(ksobj, candidate)
+			fieldNames.add(candidate)
 		except Exception:
 			pass
 
-- 
cgit v1.3.1