news_parse_ex

commit: 18972a2a8d9a2c95a56ab13e5ee1d3c0dcafb3db
parent 1f3340d3604fa37fa5f2812c7bfa87b5ae856ba7
Author: Haelwenn (lanodan) Monnier <contact@hacktivis.me>
Date:   Sun,  5 Feb 2023 11:47:42 +0100

Switch from :calendar to :timex to preserve TZ info

Diffstat:
M
lib/rss2_0.ex
16
+++++++++++++++-
M
mix.exs
2
+-
M
test/news_parse_ex_test.exs
21
++++++++++++++++++++-

3 files changed, 36 insertions(+), 3 deletions(-)
diff --git a/lib/rss2_0.ex b/lib/rss2_0.ex
@@ -29,13 +29,27 @@ defmodule NewsParseEx.RSS2_0 do
   defp get_entry_link(frag), do: XML.string_from_xpath(~s{/item/link}, frag)
   defp get_entry_id(frag), do: XML.string_from_xpath(~s{/item/guid}, frag)
 
+  defp parse_datetime(_str, []) do
+    :error
+  end
+
+  defp parse_datetime(str, [format | rest]) when is_bitstring(str) do
+    with {:ok, _} = dt <- Timex.parse(str, format) do
+      dt
+    else
+      _e -> parse_datetime(str, rest)
+    end
+  end
+
   defp get_entry_published(frag) do
     with {:ok, pubDate} <- XML.string_from_xpath(~s{/item/pubDate}, frag) do
-      Calendar.DateTime.Parse.rfc822_utc(pubDate)
+      # RFC1123 is because of SourceHut
+      parse_datetime(pubDate, ["{RFC822}", "{RFC1123}"])
     end
   end
 
   defp get_feed_entry(frag) do
+    # FIXME: Don't just shove errors away
     %{}
     |> Maps.put_if_ok(:title, get_entry_title(frag))
     |> Maps.put_if_ok(:id, get_entry_id(frag))
diff --git a/mix.exs b/mix.exs
@@ -19,6 +19,6 @@ defmodule NewsParseEx.MixProject do
   end
 
   defp deps do
-    [{:calendar, "~> 1.0.0"}]
+    [{:timex, "~> 3.0"}]
   end
 end
diff --git a/test/news_parse_ex_test.exs b/test/news_parse_ex_test.exs
@@ -1,3 +1,7 @@
+# NewsParseEx: RSS/Atom parser
+# Copyright © 2022-2023 Haelwenn (lanodan) Monnier <contact+news_parse_ex@hacktivis.me>
+# SPDX-License-Identifier: AGPL-3.0-only
+
 defmodule NewsParseExTest do
   use ExUnit.Case, async: true
 
@@ -131,13 +135,28 @@ defmodule NewsParseExTest do
 
       assert(length(parsed.entries) == 20)
 
+      # Sat, 30 May 2020 19:17:12 -0600
+      # #DateTime<2020-05-30 19:17:12-06:00 -06 Etc/UTC-6>
+      published = %DateTime{
+        year: 2020,
+        month: 05,
+        day: 30,
+        hour: 19,
+        minute: 17,
+        second: 12,
+        std_offset: 0,
+        utc_offset: -6 * 60 * 60,
+        time_zone: "Etc/UTC-6",
+        zone_abbr: "-06"
+      }
+
       assert(
         Enum.at(parsed.entries, 0) == %{
           id: "https://git.sr.ht/~kaniini/pkgconf/refs/pkgconf-1.7.3",
           title: "pkgconf-1.7.3",
           description: "pkgconf 1.7.3.",
           link: "https://git.sr.ht/~kaniini/pkgconf/refs/pkgconf-1.7.3",
-          published: ~U[2020-05-31 01:17:12Z]
+          published: published
         }
       )
     end