def test_jumptargets():
input = "01: char '2f'\n13: span [(30-39)]\n37: testcode [(1)(3)(39)] -> 13\n42: jmp -> 37"
instructionlist = relabel(parse(input))
assert not instructionlist[0].isjumptarget
assert instructionlist[1].isjumptarget
assert instructionlist[2].isjumptarget
assert not instructionlist[3].isjumptarget
def test_json(self):
with open("examples/jsonpattern") as f:
p = f.read()
with open("examples/jsoninput") as f:
i = f.read()
instructionlist = relabel(parse(runpattern(p)))
self.run_string(instructionlist, i)
def test_span(self):
#this instructionset should match an arbitrarily long string of
#'a's and 'b's, terminating with another character
pattern = 'lpeg.R("az")^0 * lpeg.R("09")^0'
instructionlist = relabel(parse(runpattern(pattern)))
input = "abbac" * 100 + "c" + "012312312445598" * 10
self.run_string(instructionlist, input)
def test_optimize_testset(self):
pattern = 'lpeg.P{lpeg.S"hH" * lpeg.P"allo" + 1 * lpeg.V(1)}^0'
instrs = relabel(parse(runpattern(pattern)))
input = "z" * 100 + "Halloyxxddcccxxddffhallojjjjjjgfffdssadgh" * 50
self.run_string(instrs,
input,
optimize_char=True,
optimize_testchar=True)
def test_optimize_char(self):
pattern = 'lpeg.P{lpeg.P"Hallo" + 1 * lpeg.V(1)}^0'
instrs = relabel(parse(runpattern(pattern)))
input = "z" * 100 + "Hallo" * 50
self.run_string(instrs,
input,
optimize_char=True,
optimize_testchar=True)
def test_optimize_char_1(self):
flags = self.flags
#tests if the code runs in principle
from pypeg.vm import look_for_chars, match_many_chars
pattern = 'lpeg.P"Hallo"'
bytecode = relabel(parse(runpattern(pattern)))
assert look_for_chars(
bytecode,
0,
) == 5
assert match_many_chars(bytecode, 0, 5, "Hallo", 0)
def test_url(self):
pattern = """lpeg.P{
"S";
S = lpeg.V("URL") + lpeg.P(1) * lpeg.V("S"),
URL = lpeg.Cp()*lpeg.C(lpeg.P"http" * lpeg.V("urlchar")^3),
urlchar = lpeg.R("az","AZ","09") + lpeg.S("-._~:/?#@!$&*+,;=")}^0"""
instructionlist = relabel(parse(runpattern(pattern)))
input = "das hier ist eine url:https://www3.hhu.de/stups/downloads/pdf/BoCuFiRi09_246.pdf und das hier nicht 192.168.13.37" * 100
self.run_string(instructionlist, input)
def test_relabel():
input = "01: char '2f'\n13: span [(30-39)]\n37: testcode [(1)(3)(39)] -> 13"
instructionlist = parse(input)
output = relabel(instructionlist)
assert output[0] == Instruction(label=0, name="char", character="/")
assert output[1] == Instruction(label=1,
name="span",
charlist=[CharRange("0", "9")])
assert output[2] == Instruction(label=2,
name="testcode",
charlist=[
SingleChar(chr(0x1)),
SingleChar(chr(0x3)),
SingleChar(chr(0x39))
],
goto=1)
def make_instructionlist(pattern):
bytecodestring = runpattern(pattern)
instructionlist = parse(bytecodestring)
return relabel(instructionlist)
def test_set(self):
pattern = 'lpeg.P{lpeg.P"c" + (lpeg.P"a"+lpeg.P"z") * lpeg.V(1)}'
instructionlist = relabel(parse(runpattern(pattern)))
input = "azaaazaz" * 100 + "c"
self.run_string(instructionlist, input)
def test_email(self):
pattern = '(lpeg.P{ lpeg.C(lpeg.R("az","AZ","09")^1*lpeg.P("@")*lpeg.R("az","AZ","09")^1*lpeg.P(".de")) + 1 * lpeg.V(1)})^0'
instructionlist = relabel(parse(runpattern(pattern)))
input = " und es endet mit noch ner mail: [email protected] kommt was:, asdfasdf [email protected]" * 100
self.run_string(instructionlist, input)
def test_complex(self):
pattern = '(lpeg.P"aa"+lpeg.P"zz")^0'
instructionlist = relabel(parse(runpattern(pattern)))
input = "aa" * 100 + "zz" * 50 + "aaaa"
self.run_string(instructionlist, input)