davideisinger.com

commit 91b2863e34f4bf0ea593931ba250a1a16643d3dc
parent fd86fe4bcab63c1394a7b88af841407c3a32bd52
Author: David Eisinger <[email protected]>
Date:   Tue, 30 Jan 2024 10:27:00 -0500

Unescape HTML titles in archive script

Diffstat:
M
bin/archive
 | 
13
++++++++-----

1 file changed, 8 insertions(+), 5 deletions(-)
diff --git a/bin/archive b/bin/archive
@@ -1,5 +1,6 @@
 #!/usr/bin/env ruby
 
+require "cgi"
 require "uri"
 require "digest"
 require "time"
@@ -18,11 +19,13 @@ urls.each do |url|
   text_content = `w3m -dump -T text/html -o display_link_number=1 #{url}`
 
   begin
-    title = page_content
-      .scan(/<title[^>]*>(.*?)<\/title>/mi)
-      .first
-      .first
-      .strip
+    title = CGI.unescapeHTML(
+      page_content
+        .scan(/<title[^>]*>(.*?)<\/title>/mi)
+        .first
+        .first
+        .strip
+      )
   rescue => ex
     warn "Title error (#{ex}; #{url})"
     exit 1