bandcamp-discover 0.6.1 → 0.7.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c2f1f8a4daa65012813a3a09bd62db1afc77e51751f14d9f69fa39ecfb144dab
-  data.tar.gz: f0311426fe47b54f71c722869395a9b435f63b13085e3c6779e0b82d40e81a77
+  metadata.gz: 0f788095223220ca96084269592bcfacc5ffcd11bdc4ccc27c43312f9a4f019e
+  data.tar.gz: d8428295e7b665d2ae448f34f4a84b74bd1290357499d6266e6859fe9b60b7fd
 SHA512:
-  metadata.gz: ff76b0b9b44909e568340ad18db61ed52af685391e9607ea88478bb23ba8feb346a96504a9624b9aed68b5c17b2f074bf38a2c9d26dde9f215c36b803a360bbc
-  data.tar.gz: 503083930726ae15941c4cbcf63a7783aed3c921c4f25af1745e1ef684481743f69efd4d87cae53c0ec572df95fdcd7ac197ac69291a59b9e9f4f2a94c9a688a
+  metadata.gz: 97041bbe3141a01e73712b5fa8f1a175ce437ad8ec6f9c38a09447c21e828af10390a248a508d63048447c982b2ac27fcef24bd1f82f285421537aaeb0acbd5a
+  data.tar.gz: cd751e7ed3dd432b66e3988c89277cf6b91363193dd3a174f5803a2d34e620ad0ec0a2ad347b4c8636c19d263fedde9beb114d9a7594f7182922bc91bca35fb2

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bandcamp-discover (0.6.1)
+    bandcamp-discover (0.7.1)
       async
       base64
       concurrent-ruby

data/README.rdoc

@@ -17,7 +17,9 @@ accepts demos. The model and both prompts can be set by the host app:
 Before the bio is asked, the /music grid is read: a page whose releases carry
 no artist credit other than the owner is an artist, not a label, and is
 rejected without a model call. The credits that remain are appended to the
-bio for the model and returned as +artists+ in the scrape result.
+bio for the model and returned as +artists+ in the scrape result; each
+entry in +albums+ carries its own credit as +artist+ (nil when the release
+is credited to the page owner).
 
 Without a token, +label?+ falls back to matching /label|platform|records/i
 and +accepts_demos?+ is always false.

data/lib/bandcamp-discover/analyzer.rb

@@ -43,7 +43,13 @@ module BandcampDiscover
         extras: {response_format: {type: "json_object"}}
       )
 
-      JSON.parse(response["choices"][0]["message"]["content"])["answer"]&.to_s&.downcase == "true"
+      answer(response["choices"][0]["message"]["content"])["answer"]&.to_s&.downcase == "true"
+    end
+
+    # response_format is advisory on OpenRouter: Anthropic models return the
+    # object inside a ```json fence, and some prepend a sentence.
+    def answer(content)
+      JSON.parse(content[/\{.*\}/m] || content)
     end
 
     # The bio alone cannot tell one person releasing under aliases from a

data/lib/bandcamp-discover/scrapers/music.rb

@@ -31,7 +31,7 @@ module BandcampDiscover
           items.map do |item|
             {
               url: absolute(item.query_selector("a")[:href]),
-              credit: item.query_selector(".artist-override")&.inner_text
+              credit: item.query_selector(".artist-override")&.inner_text&.strip
             }
           end
         end
@@ -41,6 +41,9 @@ module BandcampDiscover
         Roster.new(band_name: band_name, credits: grid.map { _1[:credit] }, releases: grid.size)
       end
 
+      # Each album carries the credit its grid item showed: the artist's name,
+      # or nil when the release is credited to the page owner. The grid is the
+      # only place the credit appears without another page load.
       def albums(grid)
         guarded do
           semaphore = Async::Semaphore.new(@max_tasks)
@@ -53,12 +56,12 @@ module BandcampDiscover
 
               puts "done scraping #{item[:url]}"
 
-              album
+              [item, album]
             end
           end.map(&:wait)
 
-          albums.map! do |album_url, album_title, album_tags, album_player|
-            {url: album_url, title: album_title, tags: album_tags, player_url: album_player}
+          albums.map! do |item, (album_url, album_title, album_tags, album_player)|
+            {url: album_url, title: album_title, tags: album_tags, player_url: album_player, artist: item[:credit]}
           end
 
           [albums, normalize_tally(albums.map { _1[:tags] }.flatten.tally)]

data/lib/bandcamp-discover/version.rb

@@ -1,3 +1,3 @@
 module BandcampDiscover
-  VERSION = "0.6.1"
+  VERSION = "0.7.1"
 end

data/test/analyzer_test.rb

@@ -23,12 +23,12 @@ module OpenRouter
 
   class Client
     class << self
-      attr_accessor :requests, :answer
+      attr_accessor :requests, :answer, :raw
     end
 
     def complete(messages, model:, extras:)
       self.class.requests << {messages: messages, model: model, extras: extras}
-      {"choices" => [{"message" => {"content" => JSON.generate("answer" => self.class.answer)}}]}
+      {"choices" => [{"message" => {"content" => self.class.raw || JSON.generate("answer" => self.class.answer)}}]}
     end
   end
 end
@@ -41,6 +41,7 @@ class AnalyzerTest < Minitest::Test
     OpenRouter.configuration.access_token = "token"
     OpenRouter::Client.requests = []
     OpenRouter::Client.answer = true
+    OpenRouter::Client.raw = nil
   end
 
   def teardown
@@ -99,6 +100,18 @@ class AnalyzerTest < Minitest::Test
     assert_equal "", OpenRouter::Client.requests.last[:messages].last[:content]
   end
 
+  def test_reads_the_answer_out_of_a_markdown_fence
+    OpenRouter::Client.raw = "```json\n{\"answer\": false}\n```"
+
+    refute Analyzer.new("bio").label?
+  end
+
+  def test_reads_the_answer_out_of_surrounding_prose
+    OpenRouter::Client.raw = "Sure. {\"answer\": true} Hope that helps."
+
+    assert Analyzer.new("bio").label?
+  end
+
   def test_parses_the_answer
     OpenRouter::Client.answer = false
 

data/test/music_test.rb

@@ -0,0 +1,46 @@
+require_relative "test_helper"
+require "bandcamp-discover/scrapers/music"
+
+# The album pages are stubbed; what is under test is how the grid's credits
+# travel into the album hashes.
+class MusicTest < Minitest::Test
+  Music = BandcampDiscover::Scrapers::Music
+
+  FakeBrowser = Struct.new(:pages) do
+    def new_page = nil
+  end
+
+  FakeAlbum = Struct.new(:url) do
+    def scrape = [url, "Title of #{url}", ["drone"], nil]
+  end
+
+  def setup
+    @music = Music.new(url: "https://kranky.bandcamp.com/music", browser: FakeBrowser.new, max_tasks: 2)
+  end
+
+  def test_each_album_carries_the_credit_from_its_grid_item
+    grid = [
+      {url: "https://kranky.bandcamp.com/album/one", credit: "Helen"},
+      {url: "https://kranky.bandcamp.com/album/two", credit: nil}
+    ]
+
+    albums, tags = BandcampDiscover::Scrapers::Album.stub(:new, ->(url:, **) { FakeAlbum.new(url) }) do
+      Sync { @music.albums(grid) }
+    end
+
+    assert_equal ["Helen", nil], albums.map { _1[:artist] }
+    assert_equal ["https://kranky.bandcamp.com/album/one", "https://kranky.bandcamp.com/album/two"], albums.map { _1[:url] }
+    assert_equal({"drone" => 1.0}, tags)
+  end
+
+  def test_a_timed_out_album_still_keeps_its_credit
+    grid = [{url: "https://kranky.bandcamp.com/album/gone", credit: "Helen"}]
+    timed_out = Struct.new(:url) { def scrape = nil }
+
+    albums, _tags = BandcampDiscover::Scrapers::Album.stub(:new, ->(url:, **) { timed_out.new(url) }) do
+      Sync { @music.albums(grid) }
+    end
+
+    assert_equal [{url: nil, title: nil, tags: nil, player_url: nil, artist: "Helen"}], albums
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: bandcamp-discover
 version: !ruby/object:Gem::Version
-  version: 0.6.1
+  version: 0.7.1
 platform: ruby
 authors:
 - Julian RUbisch
@@ -206,6 +206,7 @@ files:
 - test/analyzer_test.rb
 - test/default_test.rb
 - test/discover_test.rb
+- test/music_test.rb
 - test/roster_test.rb
 - test/test_helper.rb
 homepage: https://julianrubisch.at