def diff_side(dumpair, addrpair, score, sources, level=3):
d1,d2 = dumpair
a1,a2 = addrpair
try:
f1 = d1.Fun(a1)
f2 = d2.Fun(a2)
except:
sc = "←(%s)→" % ('{0:^+.2g}'.format(score))
head = "<h%d>%s in %s %s %s in %s</h%d>" % (level, disasm.guess_data(d1,a1), d1.bin, sc, disasm.guess_data(d2,a2), d2.bin, level)
srch = "<small>"
for srcpair,sco in sources:
srch += diff_side(dumpair, srcpair, sco, [], level=level+1)
srch += "</small>"
return head + srch
A = fileutil.capture(f1.disasm)[1].split("\n")
B = fileutil.capture(f2.disasm)[1].split("\n")
#~ a = [l.split("\t")[0] + " " + string.join(l.split("\t")[2:], "\t")[:60] for l in a]
#~ b = [l.split("\t")[0] + " " + string.join(l.split("\t")[2:], "\t")[:60] for l in b]
a = [string.join(l.split("\t")[2:], "\t")[:60] for l in A]
b = [string.join(l.split("\t")[2:], "\t")[:60] for l in B]
h = difflib.HtmlDiff()
sc = "←[%s]→" % ('{0:^+.2g}'.format(score))
head = "<h%d>%s in %s %s %s in %s</h%d>" % (level, f1.name, f1.dump.bin, sc, f2.name, f2.dump.bin, level)
htm = head + h.make_table(a, b, f1.dump.bin, f2.dump.bin)
for i,l in enumerate(A):
try: addr = int(l.split("\t")[0][:-1],16)
except: continue
r = r'<td class="diff_header" id="from([0-9_]+)%d">%d</td>' % (i+1,i+1)
new = r'<td class="diff_header" id="from\\1%d">%x:</td>' % (i+1,addr)
htm = re.sub(r,new,htm)
for i,l in enumerate(B):
try: addr = int(l.split("\t")[0][:-1],16)
except: continue
r = r'<td class="diff_header" id="to([0-9_]+)%d">%d</td>' % (i+1,i+1)
new = r'<td class="diff_header" id="to\\1%d">%x:</td>' % (i+1,addr)
htm = re.sub(r,new,htm)
return htm
def data_match_funcpair(dumpair, addrpair, log=True):
if type(log) == file:
matchlog = log
else:
matchlog = sys.stdout
verbose = bool(log)
d1,d2 = dumpair
a1,a2 = addrpair
e1 = d1.FUNCS[a1]
e2 = d2.FUNCS[a2]
#~ print e1-a1, e2-a2
#~ print d1.addr2sigs[a1]
#~ print d2.addr2sigs[a2]
ref1 = len(d1.REF2AS[a1]) # how many times it was referenced
ref2 = len(d2.REF2AS[a2])
refm = (ref1 + ref2) / 2.0
reftomismatch = abs(ref1 - ref2) / refm if refm else 0
if verbose: print >> matchlog, "refto mismatch: %.2g (%.2g vs %d)" % (reftomismatch, ref1, ref2)
ref1 = len(disasm.find_refs(d1, func=a1))
ref2 = len(disasm.find_refs(d2, func=a2))
refm = (ref1 + ref2) / 2.0
reffrommismatch = abs(ref1 - ref2) / refm if refm else 0
if verbose: print >> matchlog, "reffrom mismatch: %.2g (%.2g vs %d)" % (reffrommismatch, ref1, ref2)
assert e1-a1 == e2-a2
smallmatch_ok = 0
smallmatch_total = 0
smallmatch = 0
bigmatch_ok = 0
bigmatch_total = 0
bigmatch = 0
datamatches = []
stringmatch_ok = 0
stringmatch_total = 0
stringmatch = 0
if verbose: print >> matchlog, funcpair((d1,d2),(a1,a2))
# refs from this function, matched line-by-line
refpairs = []
for off in range(0, e1-a1, 4):
for i in range(2): # max 2 refs at a single address; first is raw value, second is pointer
try:
c1,c2 = None,None
c1 = d1.A2REFS[a1+off][i]
c2 = d2.A2REFS[a2+off][i]
refpairs.append((c1,c2))
except:
pass
# if there is a single ref (call) to this function, consider it also
# (i.e. the function which calls this should also match)
c1 = d1.REF2AS[a1]
c2 = d2.REF2AS[a2]
if len(c1)==1 and len(c2)==1:
c1,c2 = disasm.which_func(d1,c1[0]), disasm.which_func(d2,c2[0])
refpairs.append((c1, c2))
if verbose:
print >> matchlog, "unique ref: %s called from %s <---> %s called from %s" \
% (disasm.guess_data(d1,a1),
disasm.guess_data(d1,c1),
disasm.guess_data(d2,a2),
disasm.guess_data(d2,c2))
for c1,c2 in refpairs:
if c1 is not None or c2 is not None:
#~ if verbose: print >> matchlog, "%f:%s --- %f:%s" % (a1+off,c1,a2+off,c2)
if c1 < 0x1000 and c2 < 0x1000:
smallmatch_total += 1
if c1 == c2: smallmatch_ok += 1
else:
s1 = disasm.GuessString(d1, c1)
s2 = disasm.GuessString(d2, c2)
if s1 or s2:
stringmatch_total += 1
ds = 0
if s1 is None or s2 is None:
ds = -1 # penalty if one has string and other doesn't
else:
ratio = difflib.SequenceMatcher(None, s1, s2).ratio()
if ratio > 0.7: ds = ratio
if len(s1) <= 5: ds /= 2 # penalty for small strings
stringmatch_ok += ds
if verbose:
print >> matchlog, "string pair [match=%.2g]: %s <---> %s" % (ds, repr(s1), repr(s2))
else:
bigmatch_total += 1
if c1==c2:
bigmatch_ok += 1
elif c1 is not None and c2 is not None:
cm = ((c1+c2)/2)
if abs(c1 - c2) / cm < 1:
bigmatch_ok += 0.5 # same order of magnitude
datamatches.append((c1,c2))
if smallmatch_total:
smallmatch = smallmatch_ok / smallmatch_total
if verbose: print >> matchlog, "small numbers match: %.2g (%d / %d)" % (smallmatch, smallmatch_ok, smallmatch_total)
if bigmatch_total:
bigmatch = bigmatch_ok / bigmatch_total
if verbose: print >> matchlog, "big numbers match: %.2g (%d / %d)" % (bigmatch, bigmatch_ok, bigmatch_total)
if stringmatch_total:
stringmatch = stringmatch_ok / stringmatch_total
if verbose: print >> matchlog, "STRING MATCH: %.2g (%.2g / %d)" % (stringmatch, stringmatch_ok, stringmatch_total)
score = (smallmatch-0.5) * math.sqrt(smallmatch_total) + \
(bigmatch-0.5) * math.sqrt(bigmatch_total) + \
20 * (stringmatch-0.5) * math.sqrt(stringmatch_total) - \
reftomismatch * 2 - \
reffrommismatch * 2
if verbose: print >> matchlog, "score: %.3g\n" % score
return score, datamatches