fix posted date parsing for legacy scanned posts

app/lib/domain/fa/parser/submission_parser_helper.rb

@@ -114,7 +114,13 @@ class Domain::Fa::Parser::SubmissionParserHelper < Domain::Fa::Parser::Base
       when VERSION_0, VERSION_1
         idx = elem_idx_after_text_match(info_children, /Posted/)
         child = info_children[idx..idx + 5].find { |ic| ic.name == "span" }
-        child.try(:[], "title").try(:strip)
+        date_str = child.try(:[], "title").try(:strip)
+        if date_str
+          DateTime.strptime(
+            date_str.gsub(/(\d+)(st|nd|rd|th)/, '\1'),
+            "%b %d, %Y %I:%M %p",
+          )
+        end
       when VERSION_2
         date_str = @elem.css(".popup_date").first["title"]
         # e.g. 'Feb 18, 2023 06:47 PM'
@@ -246,7 +252,7 @@ class Domain::Fa::Parser::SubmissionParserHelper < Domain::Fa::Parser::Base
       case @page_version
       when VERSION_0
         elem_after_text_match(info_children, /Resolution/).try(:text).try(
-          :strip
+          :strip,
         )
       when VERSION_1
         idx = elem_idx_after_text_match(info_children, /Resolution/)

app/lib/domain/fa/post_factor_calculator.rb

@@ -8,13 +8,13 @@ class Domain::Fa::PostFactorCalculator
   end
 
   def fit
-    limit = 100_000_000
+    limit = 10_000_000
     dataset =
       measure(
-        ->(r) do
+        proc do |r|
           r && "loaded #{r.length.to_s.bold} favs" ||
-            "loading up to #{limit} favs"
-        end
+            "loading up to #{limit.to_s.bold} favs"
+        end,
       ) { Domain::Fa::Fav.all.limit(limit).pluck(:user_id, :post_id).to_a }
 
     measure("convert to hash") do
@@ -51,13 +51,13 @@ class Domain::Fa::PostFactorCalculator
       .each_slice(20_000) do |chunk|
         total += chunk.size
         measure(
-          " -> wrote chunk of #{chunk.size.to_s.bold} - (#{total.to_s.bold} total)"
+          " -> wrote chunk of #{chunk.size.to_s.bold} - (#{total.to_s.bold} total)",
         ) do
           Domain::Fa::PostFactor.upsert_all(
             chunk,
             unique_by: :post_id,
             update_only: factors_col_name,
-            returning: :id
+            returning: :id,
           )
         end
       end

spec/lib/domain/fa/parser/legacy_page_spec.rb

@@ -115,20 +115,20 @@ describe Domain::Fa::Parser::Page do
     assert_equal(
       {
         href: "/gallery/feretta/folder/60236/Tale-of-Tails",
-        title: "Tale of Tails"
+        title: "Tale of Tails",
       },
-      folders[1]
+      folders[1],
     )
     assert_equal(
       {
         href: "/gallery/feretta/folder/60234/Illustrations",
-        title: "Illustrations"
+        title: "Illustrations",
       },
-      folders[2]
+      folders[2],
     )
     assert_equal(
       { href: "/gallery/feretta/folder/229520/Marketing", title: "Marketing" },
-      folders.last
+      folders.last,
     )
   end
 
@@ -177,7 +177,7 @@ describe Domain::Fa::Parser::Page do
     assert_equal "Miles-DF", sub.artist
     assert_equal "//d.facdn.net/art/miles-df/1455889648/1455889648.miles-df_miles-df_stream.jpg",
                  sub.full_res_img
-    assert_equal "Feb 19th, 2016 08:47 AM", sub.posted_date
+    assert_equal DateTime.parse("Feb 19, 2016 08:47 AM"), sub.posted_date
     assert_equal "All", sub.category
     assert_equal "All", sub.theme
     assert_equal "Unspecified / Any", sub.species
@@ -201,7 +201,7 @@ describe Domain::Fa::Parser::Page do
     assert_equal 82, sub.num_views
     assert_equal "Male", sub.gender
     assert_equal "Dragon (Other)", sub.species
-    assert_equal "Feb 24th, 2016 12:51 AM", sub.posted_date
+    assert_equal DateTime.parse("Feb 24, 2016 12:51 AM"), sub.posted_date
     assert_equal %w[kobold frog hunting male frog mighty hunter],
                  sub.keywords_array
   end
@@ -219,7 +219,7 @@ describe Domain::Fa::Parser::Page do
     assert sub.description_html =~ /Forgot to put this up/
     assert_equal "//d.facdn.net/art/feretta/1479650817/1479650802.feretta_161004_space_vixen_hop_desktop.jpg",
                  sub.full_res_img
-    assert_equal "Nov 20th, 2016 09:06 AM", sub.posted_date
+    assert_equal DateTime.parse("Nov 20, 2016 09:06 AM"), sub.posted_date
     assert_equal "Artwork (Digital)", sub.category
     assert_equal "Miscellaneous", sub.theme
     assert_equal "Canid - Vulpine", sub.species
@@ -247,14 +247,14 @@ describe Domain::Fa::Parser::Page do
     assert_equal({ href: "/scraps/salkitten/", title: "Scraps" }, folders[0])
     assert_equal(
       { href: "/gallery/salkitten/folder/51051/7up-Pup", title: "7up Pup" },
-      folders[2]
+      folders[2],
     )
     assert_equal(
       {
         href: "/gallery/salkitten/folder/93190/Animations",
-        title: "Animations"
+        title: "Animations",
       },
-      folders.last
+      folders.last,
     )
   end