Add a scraper for Pillowfort.social

Closes #5

lib/philomena/scrapers.ex

@@ -1,6 +1,7 @@
 defmodule Philomena.Scrapers do
   @scrapers [
     Philomena.Scrapers.Deviantart,
+    Philomena.Scrapers.Pillowfort,
     Philomena.Scrapers.Twitter,
     Philomena.Scrapers.Tumblr,
     Philomena.Scrapers.Raw

lib/philomena/scrapers/pillowfort.ex

@@ -0,0 +1,48 @@
+defmodule Philomena.Scrapers.Pillowfort do
+  @url_regex ~r|\Ahttps?://www.pillowfort.social/posts/(\d+)/?|
+
+  @spec can_handle?(URI.t(), String.t()) :: true | false
+  def can_handle?(_uri, url) do
+    String.match?(url, @url_regex)
+  end
+
+  def scrape(_uri, url) do
+    api_response!(url)
+    |> extract_data()
+  end
+
+  defp extract_data(post) do
+    if Enum.count(post["media"]) > 0 do
+      images =
+        post["media"]
+        |> Enum.filter(fn(x) -> x["media_type"] == "picture" and x["url"] end)
+        |> Enum.map(
+          &%{
+            url: &1["url"],
+            camo_url: Camo.Image.image_url(&1["url"])
+          }
+        )
+
+      %{
+        source_url: "https://www.pillowfort.social/posts/#{post["id"]}",
+        author_name: post["username"],
+        description: HtmlSanitizeEx.strip_tags(post["content"]),
+        images: images
+      }
+    end
+  end
+
+  def api_response!(url) do
+    [post_id] = Regex.run(@url_regex, url, capture: :all_but_first)
+
+    api_url =
+      "https://www.pillowfort.social/posts/#{post_id}/json/"
+
+    url = "https://www.pillowfort.social/posts/#{post_id}"
+
+    Philomena.Http.get!(api_url, [{"Accept", "application/json"}])
+    |> Map.get(:body)
+    |> Jason.decode!()
+    |> Map.put("url", url)
+  end
+end

mix.exs

@@ -69,7 +69,8 @@ defmodule Philomena.MixProject do
       {:mint, "~> 1.1"},
       {:libcluster, "~> 3.2"},
       {:exq, "~> 0.13"},
-      {:dialyxir, "~> 1.0", only: :dev, runtime: false}
+      {:dialyxir, "~> 1.0", only: :dev, runtime: false},
+      {:html_sanitize_ex, "~> 1.3.0-rc3"}
     ]
   end
 

mix.lock

@@ -30,6 +30,7 @@
   "gen_smtp": {:hex, :gen_smtp, "0.15.0", "9f51960c17769b26833b50df0b96123605a8024738b62db747fece14eb2fbfcc", [:rebar3], [], "hexpm", "29bd14a88030980849c7ed2447b8db6d6c9278a28b11a44cafe41b791205440f"},
   "gettext": {:hex, :gettext, "0.18.0", "406d6b9e0e3278162c2ae1de0a60270452c553536772167e2d701f028116f870", [:mix], [], "hexpm", "c3f850be6367ebe1a08616c2158affe4a23231c70391050bf359d5f92f66a571"},
   "hackney": {:hex, :hackney, "1.16.0", "5096ac8e823e3a441477b2d187e30dd3fff1a82991a806b2003845ce72ce2d84", [:rebar3], [{:certifi, "2.5.2", [hex: :certifi, repo: "hexpm", optional: false]}, {:idna, "6.0.1", [hex: :idna, repo: "hexpm", optional: false]}, {:metrics, "1.0.1", [hex: :metrics, repo: "hexpm", optional: false]}, {:mimerl, "~>1.1", [hex: :mimerl, repo: "hexpm", optional: false]}, {:parse_trans, "3.3.0", [hex: :parse_trans, repo: "hexpm", optional: false]}, {:ssl_verify_fun, "1.1.6", [hex: :ssl_verify_fun, repo: "hexpm", optional: false]}], "hexpm", "3bf0bebbd5d3092a3543b783bf065165fa5d3ad4b899b836810e513064134e18"},
+  "html_sanitize_ex": {:hex, :html_sanitize_ex, "1.3.0", "f005ad692b717691203f940c686208aa3d8ffd9dd4bb3699240096a51fa9564e", [:mix], [{:mochiweb, "~> 2.15", [hex: :mochiweb, repo: "hexpm", optional: false]}], "hexpm", "abfb393ad888d57700f4d0f119c2643c8a9d98856f9b8a92001be7efad1419d6"},
   "httpoison": {:hex, :httpoison, "1.7.0", "abba7d086233c2d8574726227b6c2c4f6e53c4deae7fe5f6de531162ce9929a0", [:mix], [{:hackney, "~> 1.16", [hex: :hackney, repo: "hexpm", optional: false]}], "hexpm", "975cc87c845a103d3d1ea1ccfd68a2700c211a434d8428b10c323dc95dc5b980"},
   "idna": {:hex, :idna, "6.0.1", "1d038fb2e7668ce41fbf681d2c45902e52b3cb9e9c77b55334353b222c2ee50c", [:rebar3], [{:unicode_util_compat, "0.5.0", [hex: :unicode_util_compat, repo: "hexpm", optional: false]}], "hexpm", "a02c8a1c4fd601215bb0b0324c8a6986749f807ce35f25449ec9e69758708122"},
   "inet_cidr": {:hex, :inet_cidr, "1.0.4", "a05744ab7c221ca8e395c926c3919a821eb512e8f36547c062f62c4ca0cf3d6e", [:mix], [], "hexpm", "64a2d30189704ae41ca7dbdd587f5291db5d1dda1414e0774c29ffc81088c1bc"},
@@ -39,6 +40,7 @@
   "mime": {:hex, :mime, "1.3.1", "30ce04ab3175b6ad0bdce0035cba77bba68b813d523d1aac73d9781b4d193cf8", [:mix], [], "hexpm", "6cbe761d6a0ca5a31a0931bf4c63204bceb64538e664a8ecf784a9a6f3b875f1"},
   "mimerl": {:hex, :mimerl, "1.2.0", "67e2d3f571088d5cfd3e550c383094b47159f3eee8ffa08e64106cdf5e981be3", [:rebar3], [], "hexpm", "f278585650aa581986264638ebf698f8bb19df297f66ad91b18910dfc6e19323"},
   "mint": {:hex, :mint, "1.1.0", "1fd0189edd9e3ffdbd7fcd8bc3835902b987a63ec6c4fd1aa8c2a56e2165f252", [:mix], [{:castore, "~> 0.1.0", [hex: :castore, repo: "hexpm", optional: true]}], "hexpm", "5bfd316c3789340b682d5679a8116bcf2112e332447bdc20c1d62909ee45f48d"},
+  "mochiweb": {:hex, :mochiweb, "2.20.1", "e4dbd0ed716f076366ecf62ada5755a844e1d95c781e8c77df1d4114be868cdf", [:rebar3], [], "hexpm", "d1aeee7870470d2fa9eae0b3d5ab6c33801aa2d82b10e9dade885c5c921b36aa"},
   "neotoma": {:hex, :neotoma, "1.7.3", "d8bd5404b73273989946e4f4f6d529e5c2088f5fa1ca790b4dbe81f4be408e61", [:rebar], [], "hexpm", "2da322b9b1567ffa0706a7f30f6bbbde70835ae44a1050615f4b4a3d436e0f28"},
   "nimble_parsec": {:hex, :nimble_parsec, "0.5.3", "def21c10a9ed70ce22754fdeea0810dafd53c2db3219a0cd54cf5526377af1c6", [:mix], [], "hexpm", "589b5af56f4afca65217a1f3eb3fee7e79b09c40c742fddc1c312b3ac0b3399f"},
   "parse_trans": {:hex, :parse_trans, "3.3.0", "09765507a3c7590a784615cfd421d101aec25098d50b89d7aa1d66646bc571c1", [:rebar3], [], "hexpm", "17ef63abde837ad30680ea7f857dd9e7ced9476cdd7b0394432af4bfc241b960"},