def test_token_iter_newline_lbox(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
new1 = TextNode(Terminal("'a"))
new2 = TextNode(Terminal("\r"))
new3 = TextNode(Terminal("b"))
new4 = TextNode(MagicTerminal("<SQL>"))
new5 = TextNode(Terminal("c'"))
bos.insert_after(new1)
new1.insert_after(new2)
new2.insert_after(new3)
new3.insert_after(new4)
new4.insert_after(new5)
it = self.lexer.get_token_iter(new1)
assert it.next() == (["'a", "\r", "b", lbph, "c'"], "string", 1, [
TextNode(Terminal("'a")),
TextNode(Terminal("\r")),
TextNode(Terminal("b")),
TextNode(MagicTerminal("<SQL>")),
TextNode(Terminal("c'"))
])
with pytest.raises(StopIteration):
it.next()
def test_relex_stop(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
new = TextNode(Terminal("1+2"))
old1 = TextNode(Terminal("*"))
old2 = TextNode(Terminal("3"))
old2.lookup = "INT"
bos.insert_after(new)
new.insert_after(old1)
old1.insert_after(old2)
self.relex(new)
assert ast.parent.symbol == Nonterminal("Root")
assert isinstance(ast.parent.children[0], BOS)
assert isinstance(ast.parent.children[-1], EOS)
node = bos.next_term
assert node.symbol == Terminal("1")
node = node.next_term
assert node.symbol == Terminal("+")
node = node.next_term
assert node.symbol == Terminal("2")
node = node.next_term
assert node.symbol == Terminal("*")
node = node.next_term
assert node.symbol == Terminal("3")
node = node.next_term
assert isinstance(node, EOS)
def test_stringwrapper(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text1 = TextNode(Terminal("abc"))
text2 = TextNode(Terminal("+"))
text3 = TextNode(Terminal("1"))
text4 = TextNode(Terminal("*"))
text5 = TextNode(Terminal("3456"))
bos.insert_after(text1)
text1.insert_after(text2)
text2.insert_after(text3)
text3.insert_after(text4)
text4.insert_after(text5)
wrapper = StringWrapper(text1)
assert wrapper[0] == "a"
assert wrapper[2] == "c"
assert wrapper[3] == "+"
assert wrapper[4] == "1"
assert wrapper[5] == "*"
assert wrapper[6] == "3"
assert wrapper[9] == "6"
s = "abc+1*3456"
for i in range(len(s)):
for j in range(len(s)):
assert wrapper[i:j] == s[i:j]
print(i, j, wrapper[i:j])
def test_relex3(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
new1 = TextNode(Terminal("1+2"))
new2 = TextNode(Terminal("345"))
new3 = TextNode(Terminal("6+"))
new4 = TextNode(Terminal("789")) # this should never be touched
new5 = TextNode(Terminal("+")) # this should never be touched
bos.insert_after(new1)
new1.insert_after(new2)
new2.insert_after(new3)
new3.insert_after(new4)
new4.insert_after(new5)
self.relex(new1)
assert ast.parent.symbol == Nonterminal("Root")
assert isinstance(ast.parent.children[0], BOS)
assert isinstance(ast.parent.children[-1], EOS)
node = bos.next_term; assert node.symbol == Terminal("1")
node = node.next_term; assert node.symbol == Terminal("+")
node = node.next_term; assert node.symbol == Terminal("23456")
node = node.next_term; assert node.symbol == Terminal("+")
# check that 789 hasn't been relexed
assert node.next_term is new4
assert node.next_term.symbol is new4.symbol
def test_relex_altered_string(self):
lexer = IncrementalLexer("""
"\"[^\"]*\"":str
"[0-9]+":INT
"\+":PLUS
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text1 = TextNode(Terminal("123+\"\""))
bos.insert_after(text1)
lexer.relex(text1)
assert bos.next_term.symbol == Terminal("123")
assert bos.next_term.lookup == "INT"
assert bos.next_term.lookahead == 1
assert bos.next_term.next_term.symbol == Terminal("+")
assert bos.next_term.next_term.lookup == "PLUS"
assert bos.next_term.next_term.lookahead == 0
assert bos.next_term.next_term.next_term.symbol == Terminal("\"\"")
assert bos.next_term.next_term.next_term.lookup == "str"
assert bos.next_term.next_term.next_term.lookahead == 0
string = bos.next_term.next_term.next_term
string.symbol.name = "\"abc\""
lexer.relex(string)
def test_triplequotes1(self):
lexer = IncrementalLexer("""
"\"\"\"[^\"]*\"\"\"":triplestring
"\"[^\"]*\"":string
"[a-z]+":var
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text1 = TextNode(Terminal("\"\"\""))
text2 = TextNode(Terminal("abc"))
text3 = TextNode(Terminal("\"\"\""))
bos.insert_after(text1)
text1.insert_after(text2)
text2.insert_after(text3)
lexer.relex(text1)
assert bos.next_term.symbol == Terminal("\"\"\"abc\"\"\"")
assert bos.next_term.lookup == "triplestring"
bos.next_term.symbol.name = "\"\"\"ab\"\"\"c\"\"\""
pytest.raises(LexingError, lexer.relex, bos.next_term)
bos.next_term.symbol.name = "\"\"\"ab\"\"\"c\"\""
lexer.relex(bos.next_term)
def test_multitoken_real_lbox_relex(self):
lexer = IncrementalLexer("""
"\"[^\"]*\"":str
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text1 = TextNode(Terminal("\"abc"))
lbox = TextNode(MagicTerminal("<SQL>"))
text2 = TextNode(Terminal("def\""))
bos.insert_after(text1)
text1.insert_after(lbox)
lbox.insert_after(text2)
lexer.relex(text1)
assert bos.next_term.lookup == "str"
assert bos.next_term == mk_multitextnode(
[Terminal("\"abc"),
MagicTerminal("<SQL>"),
Terminal("def\"")])
bos.next_term.children[0].symbol.name = "\"ab\rc"
lexer.relex(bos.next_term)
assert bos.next_term == mk_multitextnode([
Terminal("\"ab"),
Terminal("\r"),
Terminal("c"),
MagicTerminal("<SQL>"),
Terminal("def\"")
])
def test_multitoken_real_lbox_relex_cut_off_string(self):
lexer = IncrementalLexer("""
"\"[^\"]*\"":str
"[a-z]+":var
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text1 = TextNode(Terminal("\"abc"))
lbox = TextNode(MagicTerminal("<SQL>"))
text2 = TextNode(Terminal("def\""))
bos.insert_after(text1)
text1.insert_after(lbox)
lbox.insert_after(text2)
lexer.relex(text1)
assert bos.next_term.lookup == "str"
assert bos.next_term == mk_multitextnode(
[Terminal("\"abc"),
MagicTerminal("<SQL>"),
Terminal("def\"")])
assert bos.next_term.lookahead == 0
bos.next_term.children[2].symbol.name = "d\"ef\""
pytest.raises(LexingError, lexer.relex, bos.next_term)
bos.next_term.children[2].symbol.name = "d\"ef"
lexer.relex(bos.next_term)
assert bos.next_term == mk_multitextnode(
[Terminal("\"abc"),
MagicTerminal("<SQL>"),
Terminal("d\"")])
assert bos.next_term.next_term.symbol.name == "ef"
def test_multitoken_real_lbox_multiple(self):
lexer = IncrementalLexer("""
"\"[^\"]*\"":str
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
n1 = TextNode(Terminal("\"abc"))
n2 = TextNode(MagicTerminal("<SQL>"))
n3 = TextNode(Terminal("def"))
n4 = TextNode(MagicTerminal("<Calc>"))
n5 = TextNode(Terminal("ghi\""))
bos.insert_after(n1)
n1.insert_after(n2)
n2.insert_after(n3)
n3.insert_after(n4)
n4.insert_after(n5)
lexer.relex(n1)
assert bos.next_term.lookup == "str"
assert bos.next_term == mk_multitextnode([
Terminal("\"abc"),
MagicTerminal("<SQL>"),
Terminal("def"),
MagicTerminal("<Calc>"),
Terminal("ghi\"")
])
def test_multitoken_real_lbox_cut_off_string(self):
lexer = IncrementalLexer("""
"\"[^\"]*\"":str
"[a-z]+":var
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text1 = TextNode(Terminal("\"abc"))
lbox = TextNode(MagicTerminal("<SQL>"))
text2 = TextNode(Terminal("d\"ef\"g"))
bos.insert_after(text1)
text1.insert_after(lbox)
lbox.insert_after(text2)
pytest.raises(LexingError, lexer.relex, text1)
assert type(bos.next_term) is MultiTextNode
assert bos.next_term.children[0] is text1
assert bos.next_term.children[1] is lbox
assert bos.next_term.children[2] is text2
assert text2.symbol.name == "d\""
assert bos.next_term.next_term.symbol.name == "ef"
leftover = bos.next_term.next_term.next_term
assert leftover.symbol.name == "\"g"
leftover.symbol.name = "g"
leftover.changed = True
lexer.relex(leftover)
assert bos.next_term.lookup == "str"
assert bos.next_term == mk_multitextnode(
[Terminal("\"abc"),
MagicTerminal("<SQL>"),
Terminal("d\"")])
assert bos.next_term.next_term.symbol.name == "efg"
def test_multi(self):
ast = AST()
ast.init()
new = TextNode(Terminal('--[[test\rtest]]'))
ast.parent.children[0].insert_after(new)
it = self.lexer.get_token_iter(new)
assert next(it) == (['--[[test', '\r', 'test]]'], "mcomment", 0, [TextNode(Terminal('--[[test\rtest]]'))], 0)
def test_stringwrapper(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text1 = TextNode(Terminal("abc"))
text2 = TextNode(Terminal("+"))
text3 = TextNode(Terminal("1"))
text4 = TextNode(Terminal("*"))
text5 = TextNode(Terminal("3456"))
bos.insert_after(text1)
text1.insert_after(text2)
text2.insert_after(text3)
text3.insert_after(text4)
text4.insert_after(text5)
wrapper = StringWrapper(text1)
assert wrapper[0] == "a"
assert wrapper[2] == "c"
assert wrapper[3] == "+"
assert wrapper[4] == "1"
assert wrapper[5] == "*"
assert wrapper[6] == "3"
assert wrapper[9] == "6"
s = "abc+1*3456"
for i in range(len(s)):
for j in range(len(s)):
assert wrapper[i:j] == s[i:j]
print(i,j,wrapper[i:j])
def test_lookahead(self):
ast = AST()
ast.init()
new = TextNode(Terminal('--[[test\rtest'))
ast.parent.children[0].insert_after(new)
it = self.lexer.get_token_iter(new)
assert next(it) == ('--[[test', "scomment", 6, [TextNode(Terminal('--[[test\rtest'))], -5)
def test_simple(self):
ast = AST()
ast.init()
new = TextNode(Terminal('--[[testtest]]'))
ast.parent.children[0].insert_after(new)
it = self.lexer.get_token_iter(new)
assert next(it) == ('--[[testtest]]', "mcomment", 0, [TextNode(Terminal('--[[testtest]]'))], 0)
def test_simple(self):
ast = AST()
ast.init()
new = TextNode(Terminal("asd"))
ast.parent.children[0].insert_after(new)
it = self.lexer.get_token_iter(new)
assert next(it) == ("asd", "NAME", 1, [TextNode(Terminal("asd"))], 0)
def test_relex(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
new = TextNode(Terminal("1 + 2 * 3"))
bos.insert_after(new)
self.relex(new)
assert ast.parent.symbol == Nonterminal("Root")
assert isinstance(ast.parent.children[0], BOS)
assert isinstance(ast.parent.children[-1], EOS)
node = bos.next_term
assert node.symbol == Terminal("1")
assert node.lookahead == 1
node = node.next_term
assert node.symbol == Terminal(" ")
assert node.lookahead == 1
node = node.next_term
assert node.symbol == Terminal("+")
assert node.lookahead == 0
node = node.next_term
assert node.symbol == Terminal(" ")
node = node.next_term
assert node.symbol == Terminal("2")
node = node.next_term
assert node.symbol == Terminal(" ")
node = node.next_term
assert node.symbol == Terminal("*")
node = node.next_term
assert node.symbol == Terminal(" ")
node = node.next_term
assert node.symbol == Terminal("3")
node = node.next_term
assert isinstance(node, EOS)
def test_backwards_lexing(self):
lexer = IncrementalLexer("""
"::=":doublecolon
"=":equal
":":singlecolon
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text = TextNode(Terminal(":"))
bos.insert_after(text)
lexer.relex(text)
assert bos.next_term.symbol.name == ":"
assert bos.next_term.lookup == "singlecolon"
assert text.lookahead == 1
text2 = TextNode(Terminal(":"))
text.insert_after(text2)
lexer.relex(text2)
assert text2.lookahead == 1
assert bos.next_term.symbol.name == ":"
assert bos.next_term.next_term.symbol.name == ":"
text3 = TextNode(Terminal("="))
text2.insert_after(text3)
lexer.relex(text3)
assert bos.next_term.symbol.name == "::="
assert isinstance(bos.next_term.next_term, EOS)
def test_multitoken_relex_merge(self):
lexer = IncrementalLexer("""
"\"[a-z\r\x80]*\"":str
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text = TextNode(Terminal("\"abc\rde\rf\""))
bos.insert_after(text)
lexer.relex(text)
assert bos.next_term.lookup == "str"
assert bos.next_term == mk_multitextnode([
Terminal("\"abc"),
Terminal("\r"),
Terminal("de"),
Terminal("\r"),
Terminal("f\"")
])
bos.next_term.children.pop(3) # remove a newline
bos.next_term.update_children()
child0 = bos.next_term.children[0]
child1 = bos.next_term.children[1]
child2 = bos.next_term.children[2]
lexer.relex(bos.next_term)
assert bos.next_term == mk_multitextnode(
[Terminal("\"abc"),
Terminal("\r"),
Terminal("def\"")])
assert bos.next_term.children[0] is child0
assert bos.next_term.children[1] is child1
assert bos.next_term.children[2] is child2
def test_relex3(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
new1 = TextNode(Terminal("1+2"))
new2 = TextNode(Terminal("345"))
new3 = TextNode(Terminal("6+"))
new4 = TextNode(Terminal("789")) # this should never be touched
new4.lookup = "INT"
new5 = TextNode(Terminal("+")) # this should never be touched
new5.lookup = "plus"
bos.insert_after(new1)
new1.insert_after(new2)
new2.insert_after(new3)
new3.insert_after(new4)
new4.insert_after(new5)
self.relex(new1)
assert ast.parent.symbol == Nonterminal("Root")
assert isinstance(ast.parent.children[0], BOS)
assert isinstance(ast.parent.children[-1], EOS)
node = bos.next_term
assert node.symbol == Terminal("1")
node = node.next_term
assert node.symbol == Terminal("+")
node = node.next_term
assert node.symbol == Terminal("23456")
node = node.next_term
assert node.symbol == Terminal("+")
# check that 789 hasn't been relexed
assert node.next_term is new4
assert node.next_term.symbol is new4.symbol
def test_multi_to_multi_normal_multi(self):
lexer = IncrementalLexer("""
"\"[a-z\r\x80]*\"":str
"[a-z]+":var
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
n1 = mk_multitextnode([
Terminal("\"ab"),
Terminal("\r"),
Terminal("cd\"xy\"ef"),
Terminal("\r"),
Terminal("gaaaah\"")
])
bos.insert_after(n1)
lexer.relex(n1)
assert bos.next_term == mk_multitextnode(
[Terminal("\"ab"),
Terminal("\r"),
Terminal("cd\"")])
assert bos.next_term.next_term.symbol == Terminal("xy")
assert bos.next_term.next_term.next_term == mk_multitextnode(
[Terminal("\"ef"),
Terminal("\r"),
Terminal("gaaaah\"")])
assert bos.next_term.next_term.next_term.next_term is eos
def test_normal_and_multi_to_multi2(self):
lexer = IncrementalLexer("""
"\"[a-z\r\x80]*\"":str
"[a-z]+":var
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
n1 = TextNode(Terminal("\"abc"))
n2 = TextNode(Terminal("\r"))
n3 = mk_multitextnode(
[Terminal("def"),
Terminal("\r"), Terminal("gh\"")])
bos.insert_after(n1)
n1.insert_after(n2)
n2.insert_after(n3)
lexer.relex(n1)
assert bos.next_term == mk_multitextnode([
Terminal("\"abc"),
Terminal("\r"),
Terminal("def"),
Terminal("\r"),
Terminal("gh\"")
])
assert bos.next_term.next_term is eos
def test_multix2_to_multi(self):
lexer = IncrementalLexer("""
"\"[a-z\r\x80]*\"":str
"[a-z]+":var
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
n1 = mk_multitextnode(
[Terminal("\"ab"),
Terminal("\r"), Terminal("cd")])
n2 = mk_multitextnode(
[Terminal("ef"), Terminal("\r"),
Terminal("gh\"")])
bos.insert_after(n1)
n1.insert_after(n2)
lexer.relex(n1)
assert bos.next_term == mk_multitextnode([
Terminal("\"ab"),
Terminal("\r"),
Terminal("cdef"),
Terminal("\r"),
Terminal("gh\"")
])
assert bos.next_term.next_term is eos
def test_normal_to_normal_and_multi(self):
lexer = IncrementalLexer("""
"\"[^\"]*\"":str
"[a-z]+":var
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text = TextNode(Terminal("\"abc\rdef\""))
bos.insert_after(text)
lexer.relex(text)
assert bos.next_term.lookup == "str"
assert bos.next_term == mk_multitextnode(
[Terminal("\"abc"),
Terminal("\r"),
Terminal("def\"")])
bos.next_term.children[0].symbol.name = "ab\"c"
lexer.relex(bos.next_term)
assert bos.next_term == TextNode(Terminal("ab"))
assert bos.next_term.next_term == mk_multitextnode(
[Terminal("\"c"),
Terminal("\r"),
Terminal("def\"")])
def test_normal_and_multi_to_normal_and_multi(self):
lexer = IncrementalLexer("""
"\"[a-z\r\x80]*\"":str
"[a-z]+":var
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
n1 = TextNode(Terminal("ab"))
n2 = mk_multitextnode(
[Terminal("cd\"e"),
Terminal("\r"),
Terminal("fg\"")])
bos.insert_after(n1)
n1.insert_after(n2)
lexer.relex(n1)
assert bos.next_term.symbol.name == "abcd"
assert bos.next_term is n1
assert bos.next_term.next_term.lookup == "str"
assert bos.next_term.next_term == mk_multitextnode(
[Terminal("\"e"),
Terminal("\r"), Terminal("fg\"")])
assert bos.next_term.next_term is n2
assert bos.next_term.next_term.next_term is eos
def test_lookahead(self):
lexer = IncrementalLexer("""
"aaa":aaa
"a":a
"b":b
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text = TextNode(Terminal("baab"))
bos.insert_after(text)
lexer.relex(text)
assert ast.parent.children[1].symbol.name == "b"
assert ast.parent.children[2].symbol.name == "a"
assert ast.parent.children[3].symbol.name == "a"
assert ast.parent.children[4].symbol.name == "b"
ast.parent.children[
1].symbol = None # Check that lookback doesn't overreach
ast.parent.children[3].symbol.name = "aa"
lexer.relex(ast.parent.children[3])
assert ast.parent.children[2].symbol.name == "aaa"
assert ast.parent.children[3].symbol.name == "aa"
assert ast.parent.children[3].deleted is True
assert ast.parent.children[4].symbol.name == "b"
def test_backwards_lexing(self):
lexer = IncrementalLexer("""
"::=":doublecolon
"=":equal
":":singlecolon
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text = TextNode(Terminal(":"))
bos.insert_after(text)
lexer.relex(text)
assert bos.next_term.symbol.name == ":"
assert bos.next_term.lookup == "singlecolon"
assert text.lookahead == 1
text.symbol.name = "::"
lexer.relex(text)
assert text.lookahead == 2
text2 = text.next_term
assert text2.lookback == 1
assert bos.next_term.symbol.name == ":"
assert bos.next_term.next_term.symbol.name == ":"
text2.symbol.name = ":="
lexer.relex(text2)
assert bos.next_term.symbol.name == "::="
assert isinstance(bos.next_term.next_term, EOS)
def test_simple3(self):
ast = AST()
ast.init()
new = TextNode(Terminal('"""'))
ast.parent.children[0].insert_after(new)
it = self.lexer.get_token_iter(new)
assert next(it) == ('""', "dstring", 2, [TextNode(Terminal('"""'))], -1)
def test_relex2(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
new = TextNode(Terminal("1"))
bos.insert_after(new)
self.relex(new)
node = bos.next_term
assert node.symbol == Terminal("1")
new.symbol.name = "1+"
self.relex(new)
node = bos.next_term
assert node.symbol == Terminal("1")
node = node.next_term
assert node.symbol == Terminal("+")
node.symbol.name = "+2"
self.relex(node)
node = bos.next_term
assert node.symbol == Terminal("1")
node = node.next_term
assert node.symbol == Terminal("+")
node = node.next_term
assert node.symbol == Terminal("2")
def test_simple(self):
ast = AST()
ast.init()
new = TextNode(Terminal('"""abc"""'))
ast.parent.children[0].insert_after(new)
it = self.lexer.get_token_iter(new)
assert it.next() == ('"""abc"""', "MLS", 0,
[TextNode(Terminal('"""abc"""'))], 0)
def test_multitoken_reuse1(self):
lexer = IncrementalLexer("""
"\"[a-z\r\x80]*\"":str
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text = TextNode(Terminal("\"abc\rdef\""))
bos.insert_after(text)
lexer.relex(text)
assert bos.next_term.lookup == "str"
assert bos.next_term == mk_multitextnode(
[Terminal("\"abc"),
Terminal("\r"),
Terminal("def\"")])
assert bos.next_term.children[0] is text
bos.next_term.children[
2].symbol.name = "de\rf\"" # insert another newline
child0 = bos.next_term.children[0]
child1 = bos.next_term.children[1]
child2 = bos.next_term.children[2]
mt = bos.next_term
lexer.relex(bos.next_term)
assert bos.next_term == mk_multitextnode([
Terminal("\"abc"),
Terminal("\r"),
Terminal("de"),
Terminal("\r"),
Terminal("f\"")
])
# test if nodes within a MultiTextNode are reused
assert bos.next_term.children[0] is child0
assert bos.next_term.children[1] is child1
assert bos.next_term.children[2] is child2
child3 = bos.next_term.children[3]
child4 = bos.next_term.children[4]
assert child0.prev_term is None
assert child0.next_term is child1
assert child1.prev_term is child0
assert child1.next_term is child2
assert child2.prev_term is child1
assert child2.next_term is child3
assert child3.prev_term is child2
assert child3.next_term is child4
assert child4.prev_term is child3
assert child4.next_term is None
assert bos.next_term is mt # reused the MultiTextNode
def test_lexingerror(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
new = TextNode(Terminal("1b"))
bos.insert_after(new)
it = self.lexer.get_token_iter(new)
assert next(it) == ("1", "INT", 1, [TextNode(Terminal("1b"))], -1)
with pytest.raises(LexingError):
next(it)
def test_lookback(self):
lexer = IncrementalLexer("""
"\"\"\"[^\"]*\"\"\"":triplestring
"\"[^\"]*\"":string
"[a-z]+":var
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text1 = TextNode(Terminal("abcd"))
text2 = TextNode(Terminal("ef"))
text3 = TextNode(Terminal("ghij"))
text4 = TextNode(Terminal("k"))
text5 = TextNode(Terminal("lm"))
text6 = TextNode(Terminal("nopqr"))
text7 = TextNode(Terminal("stu"))
bos.insert_after(text1)
text1.insert_after(text2)
text2.insert_after(text3)
text3.insert_after(text4)
text4.insert_after(text5)
text5.insert_after(text6)
text6.insert_after(text7)
lexer.relexed.add(text1)
lexer.relexed.add(text2)
lexer.relexed.add(text3)
lexer.relexed.add(text4)
lexer.relexed.add(text5)
lexer.relexed.add(text6)
text1.lookahead = 7
text2.lookahead = 7
text3.lookahead = 1
text4.lookahead = 0
text5.lookahead = 0
text6.lookahead = 3
text7.lookahead = 0
lexer.update_lookback(text1, text1)
assert text1.lookback == 0
assert text2.lookback == 1
assert text3.lookback == 2
assert text4.lookback == 3
assert text5.lookback == 3
assert text6.lookback == 0
assert text7.lookback == 1
def test_token_iter(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
new = TextNode(Terminal("1+2*3"))
bos.insert_after(new)
next_token = self.lexer.lexer.get_token_iter(new).next
assert next_token() == ("1", "INT", 1, [TextNode(Terminal("1+2*3"))])
assert next_token() == ("+", "plus", 0, [TextNode(Terminal("1+2*3"))])
assert next_token() == ("2", "INT", 1, [TextNode(Terminal("1+2*3"))])
assert next_token() == ("*", "mul", 0, [TextNode(Terminal("1+2*3"))])
assert next_token() == ("3", "INT", 1, [TextNode(Terminal("1+2*3"))])
def test_relex_newline(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
new1 = TextNode(Terminal("1+2\r3+4"))
bos.insert_after(new1)
self.relex(new1)
assert ast.parent.symbol == Nonterminal("Root")
assert isinstance(ast.parent.children[0], BOS)
assert isinstance(ast.parent.children[-1], EOS)
node = bos.next_term; assert node.symbol == Terminal("1")
node = node.next_term; assert node.symbol == Terminal("+")
node = node.next_term; assert node.symbol == Terminal("2")
node = node.next_term; assert node.symbol == Terminal("\r")
node = node.next_term; assert node.symbol == Terminal("3")
node = node.next_term; assert node.symbol == Terminal("+")
node = node.next_term; assert node.symbol == Terminal("4")
def test_relex2(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
new = TextNode(Terminal("1"))
bos.insert_after(new)
self.relex(new)
node = bos.next_term; assert node.symbol == Terminal("1")
new.symbol.name = "1+"
self.relex(new)
node = bos.next_term; assert node.symbol == Terminal("1")
node = node.next_term; assert node.symbol == Terminal("+")
node.symbol.name = "+2"
self.relex(node)
node = bos.next_term; assert node.symbol == Terminal("1")
node = node.next_term; assert node.symbol == Terminal("+")
node = node.next_term; assert node.symbol == Terminal("2")
def test_relex_return(self):
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text = TextNode(Terminal("123\r"))
bos.insert_after(text)
self.relex(text)
last_return = eos.prev_term
assert last_return.symbol.name == "\r"
assert last_return.lookup == "<return>"
new_number = TextNode(Terminal("3"))
last_return.insert_after(new_number)
self.relex(new_number)
new = TextNode(Terminal("\r"))
last_return.insert_after(new)
self.relex(new)
assert new.symbol == Terminal("\r")
assert new.lookup == "<return>"
def test_lookahead(self):
lexer = IncrementalLexer("""
"aaa":aaa
"a":a
"b":b
""")
ast = AST()
ast.init()
bos = ast.parent.children[0]
eos = ast.parent.children[1]
text = TextNode(Terminal("baab"))
bos.insert_after(text)
lexer.relex(text)
assert ast.parent.children[1].symbol.name == "b"
assert ast.parent.children[2].symbol.name == "a"
assert ast.parent.children[3].symbol.name == "a"
assert ast.parent.children[4].symbol.name == "b"
ast.parent.children[1].symbol = None
ast.parent.children[3].symbol.name = "aa"
lexer.relex(ast.parent.children[3])
assert ast.parent.children[2].symbol.name == "aaa"
assert ast.parent.children[3].symbol.name == "b"
