news_parse_ex

commit: 1f3340d3604fa37fa5f2812c7bfa87b5ae856ba7
parent 96f57d300aefb7dec1edc982e423634dc0b67fbf
Author: Haelwenn (lanodan) Monnier <contact@hacktivis.me>
Date:   Sat,  4 Feb 2023 12:16:53 +0100

Split Atom and RSS2.0 in two modules

Diffstat:
A
lib/atom.ex
61
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
M
lib/news_parse_ex.ex
110
+++++++------------------------------------------------------------------------
A
lib/rss2_0.ex
66
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

3 files changed, 136 insertions(+), 101 deletions(-)
diff --git a/lib/atom.ex b/lib/atom.ex
@@ -0,0 +1,61 @@
+# NewsParseEx: RSS/Atom parser
+# Copyright © 2022-2023 Haelwenn (lanodan) Monnier <contact+news_parse_ex@hacktivis.me>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule NewsParseEx.Atom do
+  alias NewsParseEx.XML
+
+  defp get_feed_title(doc) do
+    with {:ok, type} <- XML.string_from_xpath(~s[/feed/title/@type], doc) do
+      get_feed_title(doc, type)
+    end
+  end
+
+  defp get_feed_title(doc, "html") do
+    title = :xmerl_xpath.string('/feed/title/child::node()', doc)
+    {:ok, {:html, title}}
+  end
+
+  defp get_feed_title(doc, "xhtml") do
+    title = :xmerl_xpath.string('/feed/title/*[local-name(.)="div"]', doc)
+    {:ok, {:xhtml, title}}
+  end
+
+  defp get_feed_title(doc, _) do
+    XML.string_from_xpath(~s[/feed/title/text()], doc)
+  end
+
+  defp get_feed_id(doc), do: XML.string_from_xpath(~s[/feed/id/text()], doc)
+
+  defp get_feed_last_update(doc) do
+    {:ok, updated} = XML.string_from_xpath(~s[/feed/updated/text()], doc)
+    DateTime.from_iso8601(updated)
+  end
+
+  defp get_feed_description(_doc), do: {:ok, nil}
+
+  defp get_feed_entries(_doc) do
+    # FIXME
+    {:ok, []}
+  end
+
+  def parse(doc) do
+    with {_, {:ok, id}} <- {:id, get_feed_id(doc)},
+         {_, {:ok, last_update, _tz_offset}} <-
+           {:last_update, get_feed_last_update(doc)},
+         {_, {:ok, title}} <- {:title, get_feed_title(doc)},
+         {_, {:ok, description}} <- {:desc, get_feed_description(doc)},
+         {_, {:ok, entries}} <- {:entries, get_feed_entries(doc)} do
+      data = %{
+        :parser => NewsParseEx.Atom,
+        :title => title,
+        :description => description,
+        :id => id,
+        :last_update => last_update,
+        :entries => entries
+      }
+
+      {:ok, data}
+    end
+  end
+end
diff --git a/lib/news_parse_ex.ex b/lib/news_parse_ex.ex
@@ -4,132 +4,40 @@
 
 defmodule NewsParseEx do
   alias NewsParseEx.XML
-  alias NewsParseEx.Maps
 
-  defp get_feed_type(doc) do
+  defp get_feed_parser(doc) do
     with {:ok, root_name} <- XML.string_from_xpath(~s[name()], doc) do
-      get_feed_type(doc, root_name)
+      get_feed_parser(doc, root_name)
     end
   end
 
-  defp get_feed_type(doc, "feed") do
+  defp get_feed_parser(doc, "feed") do
     with {:ok, namespace} <- XML.string_from_xpath(~s{/feed/namespace::*[name()='']}, doc) do
       if namespace == "http://www.w3.org/2005/Atom" do
-        {:ok, :atom}
+        {:ok, NewsParseEx.Atom}
       else
         {:error, "Atom feed with wrong root namespace: #{namespace}"}
       end
     end
   end
 
-  defp get_feed_type(doc, "rss") do
+  defp get_feed_parser(doc, "rss") do
     with {:ok, version} <- XML.string_from_xpath(~s{/rss/@version}, doc) do
       case version do
-        "2.0" -> {:ok, :rss2_0}
+        "2.0" -> {:ok, NewsParseEx.RSS2_0}
         version -> {:error, "RSS with unknown version: #{version}"}
       end
     end
   end
 
-  defp get_feed_type(_doc, root_name) do
+  defp get_feed_parser(_doc, root_name) do
     {:error, "XML root isn't <feed> but <#{root_name}>"}
   end
 
-  defp get_feed_title(doc, :rss2_0), do: XML.string_from_xpath(~s[/rss/channel/title/text()], doc)
-
-  defp get_feed_title(doc, :atom) do
-    with {:ok, type} <- XML.string_from_xpath(~s[/feed/title/@type], doc) do
-      get_feed_title(doc, :atom, type)
-    end
-  end
-
-  defp get_feed_title(doc, :atom, "html") do
-    title = :xmerl_xpath.string('/feed/title/child::node()', doc)
-    {:ok, {:html, title}}
-  end
-
-  defp get_feed_title(doc, :atom, "xhtml") do
-    title = :xmerl_xpath.string('/feed/title/*[local-name(.)="div"]', doc)
-    {:ok, {:xhtml, title}}
-  end
-
-  defp get_feed_title(doc, :atom, _) do
-    XML.string_from_xpath(~s[/feed/title/text()], doc)
-  end
-
-  defp get_feed_id(doc, :atom), do: XML.string_from_xpath(~s[/feed/id/text()], doc)
-  defp get_feed_id(doc, :rss2_0), do: XML.string_from_xpath(~s[/rss/channel/link/text()], doc)
-
-  defp get_feed_last_update(doc, :atom) do
-    {:ok, updated} = XML.string_from_xpath(~s[/feed/updated/text()], doc)
-    DateTime.from_iso8601(updated)
-  end
-
-  defp get_feed_last_update(_doc, :rss2_0), do: {:ok, nil, nil}
-
-  defp get_feed_description(_doc, :atom), do: {:ok, nil}
-
-  defp get_feed_description(doc, :rss2_0),
-    do: XML.string_from_xpath(~s[/rss/channel/description/text()], doc)
-
-  defp get_feed_entries(_doc, :atom) do
-    # FIXME
-    {:ok, []}
-  end
-
-  defp get_feed_entries(doc, :rss2_0) do
-    items = :xmerl_xpath.string('/rss/channel/item', doc)
-
-    if length(items) != 0 do
-      entries = Enum.map(items, &get_feed_entry(&1, :rss2_0))
-      {:ok, entries}
-    else
-      {:ok, []}
-    end
-  end
-
-  defp get_entry_title(frag, :rss2_0), do: XML.string_from_xpath(~s{/item/title}, frag)
-
-  defp get_entry_description(frag, :rss2_0),
-    do: XML.string_from_xpath(~s{/item/description}, frag)
-
-  defp get_entry_link(frag, :rss2_0), do: XML.string_from_xpath(~s{/item/link}, frag)
-  defp get_entry_id(frag, :rss2_0), do: XML.string_from_xpath(~s{/item/guid}, frag)
-
-  defp get_entry_published(frag, :rss2_0) do
-    with {:ok, pubDate} <- XML.string_from_xpath(~s{/item/pubDate}, frag) do
-      Calendar.DateTime.Parse.rfc822_utc(pubDate)
-    end
-  end
-
-  defp get_feed_entry(frag, :rss2_0) do
-    %{}
-    |> Maps.put_if_ok(:title, get_entry_title(frag, :rss2_0))
-    |> Maps.put_if_ok(:id, get_entry_id(frag, :rss2_0))
-    |> Maps.put_if_ok(:link, get_entry_link(frag, :rss2_0))
-    |> Maps.put_if_ok(:published, get_entry_published(frag, :rss2_0))
-    |> Maps.put_if_ok(:description, get_entry_description(frag, :rss2_0))
-  end
-
   def parse(str) when is_bitstring(str) do
     with {_, {:ok, doc}} <- {:parse, XML.parse_document(str)},
-         {_, {:ok, feed_type}} <- {:type, get_feed_type(doc)},
-         {_, {:ok, id}} <- {:id, get_feed_id(doc, feed_type)},
-         {_, {:ok, last_update, _tz_offset}} <-
-           {:last_update, get_feed_last_update(doc, feed_type)},
-         {_, {:ok, title}} <- {:title, get_feed_title(doc, feed_type)},
-         {_, {:ok, description}} <- {:desc, get_feed_description(doc, feed_type)},
-         {_, {:ok, entries}} <- {:entries, get_feed_entries(doc, feed_type)} do
-      data = %{
-        :type => feed_type,
-        :title => title,
-        :description => description,
-        :id => id,
-        :last_update => last_update,
-        :entries => entries
-      }
-
-      {:ok, data}
+         {_, {:ok, parser}} <- {:parser, get_feed_parser(doc)} do
+      parser.parse(doc)
     end
   end
 end
diff --git a/lib/rss2_0.ex b/lib/rss2_0.ex
@@ -0,0 +1,66 @@
+# NewsParseEx: RSS/Atom parser
+# Copyright © 2022-2023 Haelwenn (lanodan) Monnier <contact+news_parse_ex@hacktivis.me>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule NewsParseEx.RSS2_0 do
+  alias NewsParseEx.Maps
+  alias NewsParseEx.XML
+
+  defp get_feed_title(doc), do: XML.string_from_xpath(~s[/rss/channel/title/text()], doc)
+  defp get_feed_id(doc), do: XML.string_from_xpath(~s[/rss/channel/link/text()], doc)
+  defp get_feed_last_update(_doc), do: {:ok, nil, nil}
+
+  defp get_feed_description(doc),
+    do: XML.string_from_xpath(~s[/rss/channel/description/text()], doc)
+
+  defp get_feed_entries(doc) do
+    items = :xmerl_xpath.string('/rss/channel/item', doc)
+
+    if length(items) != 0 do
+      entries = Enum.map(items, &get_feed_entry(&1))
+      {:ok, entries}
+    else
+      {:ok, []}
+    end
+  end
+
+  defp get_entry_title(frag), do: XML.string_from_xpath(~s{/item/title}, frag)
+  defp get_entry_description(frag), do: XML.string_from_xpath(~s{/item/description}, frag)
+  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 get_entry_published(frag) do
+    with {:ok, pubDate} <- XML.string_from_xpath(~s{/item/pubDate}, frag) do
+      Calendar.DateTime.Parse.rfc822_utc(pubDate)
+    end
+  end
+
+  defp get_feed_entry(frag) do
+    %{}
+    |> Maps.put_if_ok(:title, get_entry_title(frag))
+    |> Maps.put_if_ok(:id, get_entry_id(frag))
+    |> Maps.put_if_ok(:link, get_entry_link(frag))
+    |> Maps.put_if_ok(:published, get_entry_published(frag))
+    |> Maps.put_if_ok(:description, get_entry_description(frag))
+  end
+
+  def parse(doc) do
+    with {_, {:ok, id}} <- {:id, get_feed_id(doc)},
+         {_, {:ok, last_update, _tz_offset}} <-
+           {:last_update, get_feed_last_update(doc)},
+         {_, {:ok, title}} <- {:title, get_feed_title(doc)},
+         {_, {:ok, description}} <- {:desc, get_feed_description(doc)},
+         {_, {:ok, entries}} <- {:entries, get_feed_entries(doc)} do
+      data = %{
+        :parser => NewsParseEx.RSS2_0,
+        :title => title,
+        :description => description,
+        :id => id,
+        :last_update => last_update,
+        :entries => entries
+      }
+
+      {:ok, data}
+    end
+  end
+end