dl-core 0.1.0 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: cf4c6083775de6f5cfb392c154831eab236fca2b9981dd0e43acd72c96c52d0f
-  data.tar.gz: 87ed15f963c4e8e6c571ec7ad9f2a10f9f4c79b1badb166911e4c106b7da22b2
+  metadata.gz: '0344548899c858f4c63e4d2ac6e9b08ad7ce0e0546469357bbde4bd9520cb8ae'
+  data.tar.gz: 9a11d71acc23886b25970deb87ffebd940c5204606e46764c8463c69aedafef7
 SHA512:
-  metadata.gz: f080be0fd4f22e3c96bc3d7c2581bf4cd4bdf6e95bc85f5cc501c0c5be0b379513e4a0ac96f509152a96bb074c556dc7956f2a902becc98220d14ba99c9cee31
-  data.tar.gz: af8f7747b957236d540661d1440e3b7eddd78f83f04a4017ec62da54498602a55e1f947bb43e60b8e9283ef96e2b8c1558ed7c790398fe4c0e66aa19c15d2cb5
+  metadata.gz: 3e001006d0aef5ef9991e21771e581bcb4e575c7d7bb995cf958e359bb8e8087f4dbca9fb614a377ecab9c8e0476ce08ec1a0a2ca39a19be3b33abed6810e87a
+  data.tar.gz: c4dc9749d8a1c2b666fd1521b6efa0e3c8cdf60ba344f71dd476dfcbcbad50c8987850de98125de7ec686081225fc53a380e37224192edd1fc3e72a6059bf6e9

data/CHANGELOG.md

@@ -1,3 +1,7 @@
+## [0.2.0]
+
+- `DL::Core::PaperFolder`, extracted from arxiv-dl and hal-dl: a paper's folder is flat while one version (v1) is archived and uses `v<N>/` folders once there are several; `unflatten!` moves a flat version into its `v<N>/` folder and calls `moved_into(destination)`, which a gem can override (arxiv-dl rewrites `html/` links there).
+
 ## [0.1.0]
 
 First version, extracted from arxiv-dl 0.3.0 and jstor-dl 0.1.0.

data/README.md

@@ -10,6 +10,7 @@ Shared foundation for the `<site>-dl` gems that build offline archives: [arxiv-d
 | `DL::Core::Error`, `DL::Core::HTTPError` | Base error for every gem; `HTTPError` for non-success responses (`status`, `url`)               |
 | `DL::Core::Author`                     | `name` and `affiliations`                                                                         |
 | `DL::Core::Slug`                       | Title → URL-safe slug, truncated at a word boundary                                               |
+| `DL::Core::PaperFolder`                | A paper's folder: flat while one version is archived, `v<N>/` folders once there are several; override `moved_into` to fix up files moved one level deeper |
 | `DL::Core::Sidecar::YAML`, `JSON`      | `metadata.yaml` / `metadata.json` from any metadata `Data` object                                 |
 | `DL::Core::Sidecar::Markdown`          | `metadata.md`: YAML frontmatter from the metadata, plus a body each gem writes                    |
 | `DL::Core::CLI`                        | The shared command line: targets as args or `--input FILE\|-`, `-p`, `--rate-limit`, `-v`, `-q`, per-target errors, exit 1 on any failure |

data/lib/dl/core/paper_folder.rb

@@ -0,0 +1,78 @@
+require 'date'
+require 'fileutils'
+require 'yaml'
+
+module DL
+  module Core
+    # One paper's folder. A paper with a single archived version is kept flat
+    # (files directly in the folder); once a second version arrives, each
+    # version gets its own v<N>/ folder.
+    #
+    # A gem whose files need fixing after moving one level deeper (arxiv-dl's
+    # relative links in html/) subclasses this and overrides moved_into.
+    class PaperFolder
+      VERSION_FOLDER = /\Av\d+(\.partial)?\z/
+
+      attr_reader :path
+
+      def initialize path
+        @path = path
+      end
+
+      def archived? version
+        Dir.exist? location_of(version)
+      end
+
+      def location_of version
+        return path if flat_version == version
+
+        version_path version
+      end
+
+      # where a new download of `version` goes; call unflatten! first if the folder is flat
+      def destination_for version
+        return path if version == 1 && !versioned?
+
+        version_path version
+      end
+
+      # moves the flat version into v<N>/ so another version can sit beside it
+      def unflatten!
+        return if flat_version.nil?
+
+        destination = version_path flat_version
+        FileUtils.mkdir_p destination
+
+        # metadata.yaml marks the folder as flat, so it moves last
+        children = Dir.children(path) - [File.basename(destination), metadata_filename]
+        children.each { FileUtils.mv File.join(path, it), destination }
+        moved_into destination
+        FileUtils.mv File.join(path, metadata_filename), destination
+      end
+
+      # called after a flat version's files moved into `destination`, one level deeper
+      def moved_into(_destination) = nil
+
+      private
+
+      def flat_version
+        metadata_path = File.join path, metadata_filename
+        return unless File.exist? metadata_path
+
+        ::YAML.safe_load_file(metadata_path, permitted_classes: [Date]).fetch 'version'
+      end
+
+      def versioned?
+        Dir.exist?(path) && Dir.children(path).any? { VERSION_FOLDER.match? it }
+      end
+
+      def version_path version
+        File.join path, "v#{version}"
+      end
+
+      def metadata_filename
+        Sidecar::YAML::FILENAME
+      end
+    end
+  end
+end

data/lib/dl/core/version.rb

@@ -1,5 +1,5 @@
 module DL
   module Core
-    VERSION = '0.1.0'.freeze
+    VERSION = '0.2.0'.freeze
   end
 end

data/lib/dl/core.rb

@@ -3,6 +3,7 @@ require_relative 'core/cli'
 require_relative 'core/client'
 require_relative 'core/error'      # before errors below that subclass Error
 require_relative 'core/http_error' # after error
+require_relative 'core/paper_folder'
 require_relative 'core/sidecar/json'
 require_relative 'core/sidecar/markdown'
 require_relative 'core/sidecar/yaml'

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: dl-core
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.2.0
 platform: ruby
 authors:
 - Shane Becker
@@ -70,6 +70,7 @@ files:
 - lib/dl/core/client.rb
 - lib/dl/core/error.rb
 - lib/dl/core/http_error.rb
+- lib/dl/core/paper_folder.rb
 - lib/dl/core/sidecar/json.rb
 - lib/dl/core/sidecar/markdown.rb
 - lib/dl/core/sidecar/yaml.rb