Fix parsing of '!=' after an identifier (#4834)

fixes #4815

spec/compiler/lexer/lexer_spec.cr

@@ -451,6 +451,17 @@ describe "Lexer" do
     token.type.should eq(:"/=")
   end
 
+  it "lexes != after identifier (#4815)" do
+    lexer = Lexer.new("some_method!=5")
+    token = lexer.next_token
+    token.type.should eq(:IDENT)
+    token.value.should eq("some_method")
+    token = lexer.next_token
+    token.type.should eq(:"!=")
+    token = lexer.next_token
+    token.type.should eq(:NUMBER)
+  end
+
   assert_syntax_error "'\\uFEDZ'", "expected hexadecimal character in unicode escape"
   assert_syntax_error "'\\u{}'", "expected hexadecimal character in unicode escape"
   assert_syntax_error "'\\u{110000}'", "invalid unicode codepoint (too large)"

spec/compiler/parser/parser_spec.cr

@@ -153,6 +153,10 @@ describe "Parser" do
   it_parses "def foo(n); foo(n -1); end", Def.new("foo", ["n".arg], "foo".call(Call.new("n".var, "-", 1.int32)))
   it_parses "def type(type); end", Def.new("type", ["type".arg])
 
+  # #4815
+  assert_syntax_error "def foo!=; end", "unexpected token: !="
+  assert_syntax_error "def foo?=(x); end", "unexpected token: ?"
+
   it_parses "def self.foo\n1\nend", Def.new("foo", body: 1.int32, receiver: "self".var)
   it_parses "def self.foo()\n1\nend", Def.new("foo", body: 1.int32, receiver: "self".var)
   it_parses "def self.foo=\n1\nend", Def.new("foo=", body: 1.int32, receiver: "self".var)

src/compiler/crystal/syntax/lexer.cr

@@ -1214,8 +1214,7 @@ module Crystal
       while ident_part?(current_char)
         next_char
       end
-      case current_char
-      when '!', '?'
+      if (current_char == '?' || current_char == '!') && peek_next_char != '='
         next_char
       end
       @token.type = :IDENT