<guid> matching of entries is done case-insensitively, should be case-sensitive

[Dan-Q]

I have an RSS feed which summarises URLs produced by a URL-shortening service. The of each entry in the feed is the shortened URL, each of which is unique. However, FreshRSS discards those where the guid matches that of an existing guid, even where they differ in case.

E.g. the following feed:

<?xml version="1.0" encoding="UTF-8" ?>
<rss version='2.0' xmlns:atom='http://www.w3.org/2005/Atom'>
  <channel>
    <item>
      <title>An Oral History of ‘Leisure Suit Larry’ – MEL Magazine</title>
      <description></description>
      <link>https:&#x2F;&#x2F;melmagazine.com&#x2F;an-oral-history-of-leisure-suit-larry-ef41bc374802</link>
      <guid>https://url-shortening-service/Mw</guid>
      <pubDate>Sat, 04 Aug 2018 18:52:06 -0000</pubDate>
    </item>
    <item>
      <title>The Web That Never Was - Dylan Beattie - YouTube</title>
      <description></description>
      <link>https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=j51Fmn4JVwU</link>
      <guid>https://url-shortening-service/MW</guid>
      <pubDate>Sat, 04 Aug 2018 13:14:21 -0000</pubDate>
    </item>
  </channel>
</trr>

When imported into FreshRSS produces only ONE entry in the database: it appears that the guids are being compared in a case-insensitive manner.

This is incorrect behaviour because (with the exception of the protocol and domain name parts) URLs, which are often used as guids, are NOT (necessarily) case-sensitive.

[Frenzie]

Does the behavior differ if you add isPermaLink="true"?

At a glance I only see a whole bunch of characters being stripped out (cf. #1335) but perhaps it's down to SimplePie.

[Alkarex]

@Dan-Q Thanks for the bug report.

In FreshRSS, GUID is currently case-insensitive only when using MySQL (COLLATE utf8mb4_unicode_ci), and should already be case-sensitive when using SQLite (COLLATE BINARY) or PostgreSQL.

• Neither RSS 1.0 / Dublin Core nor RSS 2.0 specification are explicit on the matter.
• GUIDs in general are not case-sensitive (typically a long integer serialised as an hexadecimal string).
• But the Atom specification is clear on the fact that it must be case-sensitive:

Processors MUST compare atom:id elements on a character-by-character basis (in a case-sensitive fashion)

So this is something that indeed requires fixing on FreshRSS side.

[Alkarex]

@Dan-Q Could you please provide an URL for such a feed?

[Alkarex]

Patch available #2078 (not tested yet - feedback much welcome)

[Alkarex]

Fixed by #2078