From c48148ea2a2b87709f813ab259c988303ffe69f1 Mon Sep 17 00:00:00 2001
From: Wes Oldenbeuving <wes@narnach.com>
Date: Mon, 28 Feb 2022 10:48:10 +0100
Subject: [PATCH] Refactor: condense `Tokenizer#tokenize_urls!`

- Extracted `maybe_parse_url` to encapsulate that Strings matched by
  gsub might not in fact be valid URls.
- Condensed the `var = (uri part).to_s; var.tr!()` logic required due to
  `String#tr!` not returning `self` in case of a no-op to put it in a
  `tap {}` block instead. I'm not in love with the solution, but it's
  a minor improvement over the previous one. One line now matches to one
  part of the url.
---
 lib/groupie/tokenizer.rb | 29 ++++++++++++++++++-----------
 1 file changed, 18 insertions(+), 11 deletions(-)

diff --git a/lib/groupie/tokenizer.rb b/lib/groupie/tokenizer.rb
index 77caa87..ba3528c 100644
--- a/lib/groupie/tokenizer.rb
+++ b/lib/groupie/tokenizer.rb
@@ -50,17 +50,13 @@ def strip_html_tags!
     # Intelligently split URLs into their component parts
     def tokenize_urls!
       @raw.gsub!(%r{http[\w\-\#:/_.?&=]+}) do |url|
-        uri = URI.parse(url)
-      rescue URI::InvalidURIError
-        url
-      else
-        path = uri.path.to_s
-        path.tr!('/_\-', ' ')
-        query = uri.query.to_s
-        query.tr!('?=&#_\-', ' ')
-        fragment = uri.fragment.to_s
-        fragment.tr!('#_/\-', ' ')
-        "#{uri.scheme} #{uri.host} #{path} #{query} #{fragment}"
+        maybe_parse_url(url) do |uri|
+          path = uri.path.tap { |str| str&.tr!('/_\-', ' ') }
+          query = uri.query.tap { |str| str&.tr!('?=&#_\-', ' ') }
+          fragment = uri.fragment.tap { |str| str&.tr!('#_/\-', ' ') }
+
+          "#{uri.scheme} #{uri.host} #{path} #{query} #{fragment}"
+        end
       end
     end
 
@@ -74,5 +70,16 @@ def remove_interpunction!(str)
       str.gsub!(/\A['"]+|[!,."']+\Z/, '')
       str
     end
+
+    # Sometimes a String looks like a URL, but it's not.
+    # This method attempts to parse the input string into a URI.
+    # If it's successful, yield it to the block and return its response.
+    # In case of failure, return the original string.
+    def maybe_parse_url(input)
+      uri = URI.parse(input)
+      yield uri
+    rescue URI::InvalidURIError
+      input
+    end
   end
 end