- Code cleanup.

- Removed the need for local temp files.

test/multiprotocoltest.py

@@ -4,24 +4,6 @@
 #
 # (c)2004 Cas Cremers
 #
-# Input of this script:
-#
-# 	- A number on the commandline of stuff to test
-# 	- A list of files on stdin to be used (lines starting with '#' are
-# 	ignored)
-#
-#
-# Tips and tricks:
-#
-# Use e.g.
-#	$ ulimit -v 100000
-# to counteract memory problems
-# 
-# If you know where to look, use
-#	$ ls s*.spdl t*.spdl -1 | ./multiprotocoltest.py 2
-# To verify combos of protocols starting with s and t
-#
-
 # ***********************
 # 	MODULES
 # ***********************
@@ -31,9 +13,9 @@ import sys
 import string
 import commands
 import copy
-from tempfile import NamedTemporaryFile
 from optparse import OptionParser
 
+# My own stuff
 import tuplesdo
 import scythertest
 import protocollist
@@ -43,28 +25,12 @@ import protocollist
 # 	PARAMETERS
 # ***********************
 
-# Temporary files
-TempFileList = "scyther-blap.tmp"
-TempFileTuples = "scyther-blip.tmp"
-
-# External programs
-TupleProgram = "./tuples.py"
-
-# Some default settings for Agents, untrusted e with sk(e) and k(a,e) etc.
-ScytherDefaults	= "--summary"
-IncludeProtocols = '../spdl/spdl-defaults.inc'
-
-# Some protocols are causing troubles: this is a hard-coded filter to exclude
-# the problem children. Unfair, yes. Practical, yes.
-#SkipList = [ 'gong-nonce.spdl', 'gong-nonce-b.spdl', 'splice-as-hc.spdl', 'kaochow-palm.spdl' ]
-SkipList = []
-
 ClaimToResultMap = {}		# maps protocol claims to correctness in singular tests (0,1)
 ProtocolToFileMap = {}		# maps protocol names to file names
 ProtocolToStatusMap = {}	# maps protocol names to status: 0 all false, 1 all correct, otherwise (2) mixed
 ProtocolToEffectsMap = {}	# maps protocols that help create multiple flaws, to the protocol names of the flaws they caused
 
-CommandPrefix = "not yet initialised."
+CommandPrefix = ""
 ArgumentsList = []			# argument lists that have been displayed onscreen
 
 # Ugly hack. Works.
@@ -122,6 +88,8 @@ def PrintProtStatus (file, prname):
 # Returns a dictionary of claim -> bool; where 1 means that it is
 # correct, and 0 means that it is false (i.e. there exists an attack)
 def ScytherEval (plist):
+	global options
+
 	# Flush before trying (possibly fatal) external commands
 	sys.stdout.flush()
 	sys.stderr.flush()
@@ -173,6 +141,8 @@ LastProgress = {}
 ProgressBarWidth = 50
 
 def ShowProgress (i,n,txt):
+	global options
+
 	def IntegerPart (x):
 		return int (( x * i ) / n)
 	
@@ -181,13 +151,13 @@ def ShowProgress (i,n,txt):
 	percentage = IntegerPart (100)
 	factor = IntegerPart (ProgressBarWidth)
 
-	showme = 0
+	showme = False
 	if LastProgress.has_key(n):
 		if LastProgress[n]<>(factor,txt):
-			showme = 1
+			showme = True
 	else:
-		showme = 1
-	if showme == 1:
+		showme = True
+	if showme:
 		bar = "\r["
 		i = 0
 		while i < ProgressBarWidth:
@@ -202,6 +172,8 @@ def ShowProgress (i,n,txt):
 	LastProgress[n] = (factor, txt)
 
 def ClearProgress (n,txt):
+	global options
+
 	if not options.progressbar:
 		return
 	bar = " " * (1 + ProgressBarWidth + 2 + 5 + len(txt))
@@ -279,7 +251,7 @@ def DescribeContext (filep, protocols, claim):
 			# the protocol of the claim. However, if the
 			# partner protocol is completely correct or
 			# completely false, we summarize.
-			summary = 0
+			summary = False
 			all = 0
 			if claim.split()[0] <> prname:
 				count = [0,0]
@@ -287,12 +259,12 @@ def DescribeContext (filep, protocols, claim):
 					count[v] = count[v]+1
 				if count[0] == 0 and count[1] > 0:
 					all = 1
-					summary = 1
+					summary = True
 				if count[1] == 0 and count[0] > 0:
 					all = 0
-					summary = 1
+					summary = True
 				
-			if summary == 1:
+			if summary:
 				DC_Claim (cl.split()[0] + " *ALL*", all)
 			else:
 				for cl,v in cllist:
@@ -315,7 +287,7 @@ def RequiresAllProtocols (protocols, claim):
 		# claim was always false
 		return 0
 	# check for simple cases
-	if int(TupleWidth) <= 2:
+	if TupleWidth <= 2:
 		# nothing to remove
 		return 1
 
@@ -387,6 +359,11 @@ def SignalAttack (protocols, claim):
 #
 # Furthermore, TempFileList is created.
 
+def main():
+	global options
+	global processed, newattacks, StartSkip
+	global TupleWidth, TupleCount
+
 	parser = OptionParser()
 	scythertest.default_options(parser)
 	parser.add_option("-t","--tuplewidth", dest="tuplewidth",
@@ -414,13 +391,9 @@ if StartPercentage < 0 or StartPercentage > 100:
 	# Send protocollist to temp file (is this necessary?)
 	ProtocolFileList = protocollist.select(options.protocols)
 	ProtocolCount = len(ProtocolFileList)
-outp = open(TempFileList, 'w')
-for l in ProtocolFileList:
-	outp.write(l + "\n") 
-outp.close()
 
 	# Determine arguments
-TupleWidth = str(options.tuplewidth)
+	TupleWidth = int(options.tuplewidth)
 
 	# Match
 	ScytherMethods = "--match=" + str(options.match)
@@ -433,7 +406,8 @@ ScytherMethods = "--match=" + str(options.match)
 	# The script first computes the singular results for all the protocols
 	# and stores this in an array, or something like that.
 
-print "Evaluating tuples of", TupleWidth, "for", ProtocolCount, "protocols, using the command '" + CommandPrefix + "'"
+	TupleCount = tuplesdo.tuples_count(ProtocolCount, TupleWidth)
+	print "Evaluating", TupleCount, "tuples of", TupleWidth, "for", ProtocolCount, "protocols."
 	i = 0
 	while i < ProtocolCount:
 		ShowProgress (i, ProtocolCount,ProtocolFileList[i]+safetxt)
@@ -449,28 +423,11 @@ print "Correct protocols: ", GetKeys (ProtocolToStatusMap, 1)
 	print "Partly flawed protocols: ", GetKeys (ProtocolToStatusMap, 2)
 	print "Completely flawed protocols: ", GetKeys (ProtocolToStatusMap, 0)
 
-# Computation of combined list.
-#----------------------------------------------------------------------
-#
-# We use the tuple script to generate the list of tuples we need.
-# We use a temporary file (again) to store that list.
-# This requires that 'tuples.py' is in the same directory.
-
-lstatus=os.system(TupleProgram + ' ' + TupleWidth + ' <' + TempFileList + ' >' + TempFileTuples)
-
-inp = open(TempFileTuples, 'r')
-TupleCount = 0
-for x in inp:
-	TupleCount = TupleCount + 1
-inp.close()
-print "Commencing test for", TupleCount, "protocol combinations."
-
 	# Testing of protocol tuples
 	#----------------------------------------------------------------------
 	#
 	# We take the list of tuples and test each combination.
 
-inp = open(TempFileTuples, 'r')
 	processed = 0
 	newattacks = 0
 	StartSkip = 0
@@ -483,22 +440,23 @@ if StartPercentage > 0:
 	#
 	# Check all these protocols
 	#
-for tline in inp:
+	def process(protocols):
+		global processed, newattacks, StartSkip
+
 		if (processed >= StartSkip):
 			#
 			# Get the next tuple
 			#
-		protocols = tline.split()
 			ShowProgress (processed, TupleCount, " ".join(protocols) + safetxt)
 			#
 			# Determine whether there are valid claims at all in
 			# this set of file names
 			#
-		has_valid_claims = 0
+			has_valid_claims = False
 			for prname in GetListKeys (ProtocolToFileMap, protocols):
 				if ProtocolToStatusMap[prname] != 0:
-				has_valid_claims = 1
-		if has_valid_claims == 1:
+					has_valid_claims = True
+			if has_valid_claims:
 				#
 				# Use Scyther to verify the claims
 				#
@@ -515,7 +473,8 @@ for tline in inp:
 				
 		# Next!
 		processed = processed + 1
-inp.close()
+
+	tuplesdo.tuples_do(process,ProtocolFileList,TupleWidth)
 
 	ClearProgress (TupleCount, safetxt)
 	print "Processed", processed,"tuple combinations in total."
@@ -531,3 +490,7 @@ if newattacks > 0:
 
 	sys.stdout.flush()
 	sys.stderr.flush()
+
+
+if __name__ == '__main__':
+	main()

test/tuple.py

@@ -1,21 +0,0 @@
-# Tuple module
-#
-# tuplesDo generates all unordered sets (in a list) of size n of the
-# elements of the list l. The resulting lists (of length n) are passed
-# to the function f.
-
-def tuplesDo (f,l,n):
-	def tuplesDoRecurse (l,r):
-		if r and (len(r) == n):
-			f(r)
-		else:
-			if l and (n > 0):
-				# Larger size: we have options
-				# Option 1: include first
-				tuplesDoRecurse (l[1:], r + [l[0]])
-				# Option 2: exclude first
-				tuplesDoRecurse (l[1:], r)
-
-	tuplesDoRecurse (l,[])
-
-

test/tuples.py

@@ -1,51 +0,0 @@
-#!/usr/bin/python
-#
-# Given a number of input lines on std and an argument int, this program
-# generates unordered tuples, e.g.:
-#
-# arg:	2
-# in:	a
-# 	b
-# 	c
-# 	d
-#
-# out:	a,b
-# 	a,c
-# 	a,d
-# 	b,c
-# 	b,d
-# 	c,d
-#
-# This should make it clear what happens.
-#
-import sys
-import string
-import tuplesdo
-
-# Retrieve the tuple width
-tuplesize = int(sys.argv[1])
-
-# Read stdin into list and count
-list = []
-loop = 1
-while loop:
-	line = sys.stdin.readline()
-	if line != '':
-		# not the end of the input
-		line = string.strip(line)
-		if line != '':
-			# not a blank line
-			list.append(line)
-	else:
-		# end of the input
-		loop = 0
-
-def tuplesPrint (l, n):
-	def f (resultlist):
-		print " ".join(resultlist)
-
-	tuplesdo.tuplesDo (f, l, n)
-
-# Generate tuples...
-tuplesPrint (list, tuplesize)
-# Thanks for your attention

test/tuplesdo.py

@@ -4,18 +4,41 @@
 # elements of the list l. The resulting lists (of length n) are passed
 # to the function f.
 
-def tuplesDo (f,l,n):
-	def tuplesDoRecurse (l,r):
+
+# First some generic combinatorial stuff
+
+def faculty_gen(n,k):
+	if n <= k:
+		return 1
+	else:
+		return n * faculty_gen(n-1,k)
+
+def faculty(n):
+	return faculty_gen(n,1)
+
+def binomial(n,k):
+	b1 = faculty_gen(n,k)
+	b2 = faculty(n-k)
+	return b1/b2
+
+
+# How many elements will there be?
+def tuples_count (l,n):
+	return binomial(l,n)
+
+# Generate those elements, and apply f
+def tuples_do (f,l,n):
+	def recurse (l,r):
 		if r and (len(r) == n):
 			f(r)
 		else:
 			if l and (n > 0):
 				# Larger size: we have options
 				# Option 1: include first
-				tuplesDoRecurse (l[1:], r + [l[0]])
+				recurse (l[1:], r + [l[0]])
 				# Option 2: exclude first
-				tuplesDoRecurse (l[1:], r)
+				recurse (l[1:], r)
 
-	tuplesDoRecurse (l,[])
+	recurse (l,[])