Handle shorthand syntax in explanations

When performing a performance test on https://github.com/kddnewton/syntax_tree/blob/50f9fc1b815156bdde53f13d4f250d9522f15e64/test/syntax_tree_test.rb I noticed a strange output. The explain syntax indicated that it was missing a bracket when I had only deleted an `end`: 

```
$ time DEAD_END_TIMEOUT=10 ./exe/dead_end spec/fixtures/syntax_tree.rb.txt
--> /Users/rschneeman/Documents/projects/dead_end/spec/fixtures/syntax_tree.rb.txt
Unmatched `]', missing `[' ?
Unmatched keyword, missing `end' ?
     6  class SyntaxTree < Ripper
   170    def self.parse(source)
   174    end
   176    private
❯  754    def on_args_add(arguments, argument)
❯  776    class ArgsAddBlock
❯  810    end
  9233  end
DEAD_END_TIMEOUT=10 ./exe/dead_end spec/fixtures/syntax_tree.rb.txt  1.63s user 0.07s system 99% cpu 1.703 total
```

The cause was this line:

```
        node.is_a?(Op) && %w[| ||].include?(node.value) &&
```

With the `%w[]` syntax the lexer tokenizes `%w[` which (since it doesn't match `[` is not counted.

This commit corrects that logic by explicitly looking for the last character of these events.

Author: schneems

CHANGELOG.md

@@ -1,5 +1,7 @@
 ## HEAD (unreleased)
 
+- Fix explanation involving shorthand syntax for literals like `%w[]` and `%Q{}` (https://github.com/zombocom/dead_end/pull/116)
+
 ## 3.0.2
 
 - Fix windows filename detection (https://github.com/zombocom/dead_end/pull/114)

lib/dead_end/left_right_lex_count.rb

@@ -61,6 +61,17 @@ def count_lex(lex)
         # ^^^
         # Means it's a string or a symbol `"{"` rather than being
         # part of a data structure (like a hash) `{ a: b }`
+        # ignore it.
+      when :on_words_beg, :on_symbos_beg, :on_qwords_beg,
+           :on_qsymbols_beg, :on_regexp_beg, :on_tstring_beg
+        # ^^^
+        # Handle shorthand syntaxes like `%Q{ i am a string }`
+        #
+        # The start token will be the full thing `%Q{` but we
+        # need to count it as if it's a `{`. Any token
+        # can be used
+        char = lex.token[-1]
+        @count_for_char[char] += 1 if @count_for_char.key?(char)
       when :on_embexpr_beg
         # ^^^
         # Embedded string expressions like `"#{foo} <-embed"`

spec/unit/explain_syntax_spec.rb

@@ -4,6 +4,18 @@
 
 module DeadEnd
   RSpec.describe "ExplainSyntax" do
+    it "handles %w[]" do
+      source = <<~EOM
+         node.is_a?(Op) && %w[| ||].include?(node.value) &&
+      EOM
+
+      explain = ExplainSyntax.new(
+        code_lines: CodeLine.from_source(source)
+      ).call
+
+      expect(explain.missing).to eq([])
+    end
+
     it "doesn't falsely identify strings or symbols as critical chars" do
       source = <<~EOM
         a = ['(', '{', '[', '|']