Fix class selector with colon (#464)

This is related to the following:

#458
#411

I decided to push the "cleaning" to the lexer, but I think
for more complex escaping rules, we may need to push back to
Elixir.

lib/floki/selector/parser.ex

@@ -54,7 +54,7 @@ defmodule Floki.Selector.Parser do
   end
 
   defp do_parse([{:hash, _, id} | t], selector) do
-    do_parse(t, %{selector | id: to_string(id) |> String.replace("\\.", ".")})
+    do_parse(t, %{selector | id: to_string(id)})
   end
 
   defp do_parse([{:class, _, class} | t], selector) do

src/floki_selector_lexer.xrl

@@ -1,6 +1,7 @@
 Definitions.
 
-IDENTIFIER = [-A-Za-z0-9_]+(\\\.[-A-Za-z0-9_]+)*
+ESCAPED = \\[:.]
+IDENTIFIER = [-A-Za-z0-9_]+(({ESCAPED})?[-A-Za-z0-9_]+)*
 QUOTED = (\"[^"]*\"|\'[^']*\')
 PARENTESIS = \([^)]*\)
 INT = [0-9]+
@@ -18,9 +19,8 @@ Rules.
 {QUOTED}                             : {token, {quoted, TokenLine, remove_wrapper(TokenChars)}}.
 {ATTRIBUTE_IDENTIFIER}               : {token, {attribute_identifier, TokenLine, TokenChars}}.
 {SYMBOL}                             : {token, {TokenChars, TokenLine}}.
-#{IDENTIFIER}                        : {token, {hash, TokenLine, tail(TokenChars)}}.
-\.{IDENTIFIER}                       : {token, {class, TokenLine, tail(TokenChars)}}.
-\.{IDENTIFIER}\\:{IDENTIFIER}        : {token, {class, TokenLine, tail(TokenChars)}}.
+#{IDENTIFIER}                        : {token, {hash, TokenLine, unescape_inside_id_name(tail(TokenChars))}}.
+\.{IDENTIFIER}                       : {token, {class, TokenLine, unescape_inside_class_name(tail(TokenChars))}}.
 \:{NOT}\(                            : {token, {pseudo_not, TokenLine}}.
 \:{IDENTIFIER}                       : {token, {pseudo, TokenLine, tail(TokenChars)}}.
 \({INT}\)                            : {token, {pseudo_class_int, TokenLine, list_to_integer(remove_wrapper(TokenChars))}}.
@@ -51,3 +51,9 @@ remove_wrapper(Chars) ->
 
 tail([_|T]) ->
   T.
+
+unescape_inside_class_name(Chars) ->
+  lists:flatten(string:replace(Chars, "\\:", ":", all)).
+
+unescape_inside_id_name(Chars) ->
+  lists:flatten(string:replace(Chars, "\\.", ".", all)).

test/floki/selector/parser_test.exs

@@ -26,7 +26,13 @@ defmodule Floki.Selector.ParserTest do
     tokens = tokenize("a.xs\\:red-500")
 
     assert Parser.parse(tokens) == [
-             %Selector{type: "a", classes: ["xs\\:red-500"], pseudo_classes: []}
+             %Selector{type: "a", classes: ["xs:red-500"], pseudo_classes: []}
+           ]
+
+    tokens = tokenize("a.xs\\:red-500\\:big")
+
+    assert Parser.parse(tokens) == [
+             %Selector{type: "a", classes: ["xs:red-500:big"], pseudo_classes: []}
            ]
   end
 

test/floki_test.exs

@@ -630,6 +630,31 @@ defmodule FlokiTest do
     assert Floki.find(document!(@html), class_selector) == []
   end
 
+  test "find elements with colon in class names" do
+    html =
+      document!(
+        html_body("""
+        <div class="w-56 flex justify-end astro-SCKKX6R4"></div>
+        <div class="m-auto max-w-7xl px-4 pt-12 pb-20 flex flex-col xl:flex-row space-y-16
+        md:space-y-20 xl:space-y-0"></div>
+        <section class="flex flex-col xl:flex-row"></section>
+        """)
+      )
+
+    assert Floki.find(html, ".xl\\:flex-row.md\\:space-y-20") == [
+             {
+               "div",
+               [
+                 {
+                   "class",
+                   "m-auto max-w-7xl px-4 pt-12 pb-20 flex flex-col xl:flex-row space-y-16\nmd:space-y-20 xl:space-y-0"
+                 }
+               ],
+               []
+             }
+           ]
+  end
+
   # Floki.find/2 - Tag name
 
   test "select elements by tag name" do