bolognese 0.6.2 → 0.6.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 4c34a24e63743974fc6a09dca95d8dbdef85f592
-  data.tar.gz: a12199e846b5c1598e181126d6c20f9c40341a79
+  metadata.gz: 019af5df5c06f424579440313912c3841dd717ae
+  data.tar.gz: 9cd1726770f9685335c7bc5afb7b053e618a1c4e
 SHA512:
-  metadata.gz: b4fab385944d35dd6da5b1c39732e3196f88dc1c673e83ecc0c815256f66c2a8dd902fe6ea7eb9e8ff5f9dc0105a1cca3f17a68524a98b16e94ffbc99696202b
-  data.tar.gz: 75783a83adabd791461eefb29b5810f636b848c1578daeecff977a650820339ef3bb62cb718e1f9201e7c1d619ccbe5d18f8414a65e5f6897c88ababd737d118
+  metadata.gz: 35f06b9c7e9e3960a54614fdc9c16da563da9cf9496b95279c5cb4258cc042d8fcd67214147f857f242e967be4eff83cb57e0143edbf2931e22e33572779ed20
+  data.tar.gz: 429b51e24ff0db4669d6f0d2a6e6c7aee00e7a7d35ba6e2b09f784b1da63533c1bcf83f4bd8907c03802073c53391d3cf71fd547abb201ee95d69775bd13a0ff

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (0.6.2)
+    bolognese (0.6.3)
       activesupport (~> 4.2, >= 4.2.5)
       bibtex-ruby (~> 4.1)
       builder (~> 3.2, >= 3.2.2)

data/README.md

@@ -4,24 +4,58 @@
 
 # Bolognese
 
-Ruby gem and command-line utility for conversion of DOI metadata from and to [schema.org](https://schema.org) in JSON-LD.
+Ruby gem and command-line utility for conversion of DOI metadata from and to different metadata formats, including [schema.org](https://schema.org).
 
 ## Features
 
-* convert [Crossref XML](https://support.crossref.org/hc/en-us/articles/214936283-UNIXREF-query-output-format) to schema.org/JSON-LD
-* convert [DataCite XML](http://schema.datacite.org/) to schema.org/JSON-LD
-* fetch schema.org/JSON-LD from a URL
-* convert schema.org/JSON-LD to [DataCite XML](http://schema.datacite.org/)
-* convert Crossref XML to [DataCite XML](http://schema.datacite.org/)
-* convert Crossref XML, DataCite XML or schema.org/JSON-LD to BibTeX
-* convert BibTeX to DataCite XML or schema.org/JSON-LD
-
-Conversion to Crossref XML is not yet supported.
+Bolognese reads and writes these metadata formats.
+
+<table border="2" cellspacing="0" cellpadding="6" rules="groups" frame="hsides" class="table table-bordered table-striped">
+  <thead>
+    <tr>
+      <th>Format</th>
+      <th>Name</th>
+      <th>Content Type</th>
+      <th>Read</th>
+      <th>Write</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td><a href='https://www.crossref.org/schema/documentation/unixref1.1/unixref1.1.html'>CrossRef Unixref XML</a></td>
+      <td>crossref</td>
+      <td>application/vnd.crossref.unixref+xml</td>
+      <td>Yes</td>
+      <td>No</td>
+   </tr>
+    <tr>
+      <td><a href='https://schema.datacite.org/'>DataCite XML</a></td>
+      <td>datacite</td>
+      <td>application/vnd.datacite.datacite+xml</td>
+      <td>Yes</td>
+      <td>Yes</td>
+  </tr>
+    <tr>
+        <td><a href='http://schema.org/'>Schema.org in JSON-LD</a></td>
+        <td>schema_org</td>
+        <td>application/vnd.schemaorg.ld+json</td>
+        <td>Yes</td>
+        <td>Yes</td>
+    </tr>
+    <tr>
+        <td><a href='http://en.wikipedia.org/wiki/BibTeX'>BibTeX</a></td>
+        <td>bibtex</td>
+        <td>application/x-bibtex</td>
+        <td>Yes</td>
+        <td>Yes</td>
+    </tr>
+  </tbody>
+</table>
 
 ## Installation
 
-The usual way with Bundler: add the following to your `Gemfile` to install the
-current version of the gem:
+Requires Ruby 2.2 or later. Then add the following to your `Gemfile` to install the
+latest version:
 
 ```ruby
 gem 'bolognese'
@@ -37,14 +71,41 @@ gem install bolognese
 
 ## Commands
 
-The `bolognese` commands understand URLs and DOIs as arguments. The `--as` command
-line flag sets the format, either `crossref`, `datacite`, or `schema_org` (default).
+Run the `bolognese` command with either an identifier (DOI or URL) or filename:
+
+```
+bolognese https://doi.org/10.7554/elife.01567
+```
+
+```
+bolognese example.xml
+```
+
+Bolognese can read BibTeX files (file extension `.bib`) and Crossref or DataCite
+XML files (file extension `.xml`).
+
+The input format (e.g. Crossref XML or BibteX) is automatically detected, but
+you can also provide the format with the `--from` or `-f` flag. The supported
+input formats are listed in the table above.
+
+The output format is determined by the `--to` or `-t` flag, and defaults to `schema_org`.
+
+Show all commands with `bolognese help`:
+
+```
+Commands:
+  bolognese --version, -v   # print the version
+  bolognese help [COMMAND]  # Describe available commands or one specific command
+  bolognese open file       # read metadata from file
+  bolognese read id         # read metadata for ID
+```
 
 ## Examples
 
 Read Crossref XML:
+
 ```
-bolognese read https://doi.org/10.7554/elife.01567 --as crossref
+bolognese https://doi.org/10.7554/elife.01567 -t crossref
 
 <?xml version="1.0" encoding="UTF-8"?>
 <doi_records>
@@ -167,7 +228,7 @@ bolognese read https://doi.org/10.7554/elife.01567 --as crossref
 
 Convert Crossref XML to schema.org/JSON-LD:
 ```
-bolognese read https://doi.org/10.7554/elife.01567
+bolognese https://doi.org/10.7554/elife.01567
 
 {
     "@context": "http://schema.org",
@@ -364,7 +425,7 @@ bolognese read https://doi.org/10.7554/elife.01567
 
 Convert Crossref XML to DataCite XML:
 ```
-bolognese read https://doi.org/10.7554/elife.01567 --as datacite
+bolognese https://doi.org/10.7554/elife.01567 -t datacite
 
 <?xml version="1.0" encoding="UTF-8"?>
 <resource xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns="http://datacite.org/schema/kernel-4" xsi:schemaLocation="http://datacite.org/schema/kernel-4 http://schema.datacite.org/meta/kernel-4/metadata.xsd">
@@ -458,7 +519,7 @@ bolognese read https://doi.org/10.7554/elife.01567 --as datacite
 Convert Crossref XML to BibTeX:
 
 ```
-bolognese read https://doi.org/10.7554/elife.01567 --as bibtex
+bolognese https://doi.org/10.7554/elife.01567 -t bibtex
 
 @article{https://doi.org/10.7554/elife.01567,
   doi = {10.7554/eLife.01567},
@@ -472,7 +533,7 @@ bolognese read https://doi.org/10.7554/elife.01567 --as bibtex
 
 Read DataCite XML:
 ```
-bolognese read 10.5061/DRYAD.8515 --as datacite
+bolognese 10.5061/DRYAD.8515 -t datacite
 
 <?xml version="1.0" encoding="UTF-8"?>
 <resource
@@ -543,7 +604,7 @@ bolognese read 10.5061/DRYAD.8515 --as datacite
 
 Convert DataCite XML to schema.org/JSON-LD:
 ```sh
-bolognese read 10.5061/DRYAD.8515
+bolognese 10.5061/DRYAD.8515
 
 {
     "@context": "http://schema.org",
@@ -614,7 +675,7 @@ bolognese read 10.5061/DRYAD.8515
 
 Convert DataCite XML to schema version 4.0:
 ```
-bolognese read 10.5061/DRYAD.8515 --as datacite --schema_version http://datacite.org/schema/kernel-4
+bolognese 10.5061/DRYAD.8515 -t datacite --schema_version http://datacite.org/schema/kernel-4
 
 <?xml version="1.0" encoding="UTF-8"?>
 <resource xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns="http://datacite.org/schema/kernel-4" xsi:schemaLocation="http://datacite.org/schema/kernel-4 http://schema.datacite.org/meta/kernel-4/metadata.xsd">
@@ -697,7 +758,7 @@ bolognese read 10.5061/DRYAD.8515 --as datacite --schema_version http://datacite
 Convert DataCite XML to BibTeX:
 
 ```
-bolognese read 10.5061/DRYAD.8515 --as bibtex
+bolognese 10.5061/DRYAD.8515 -t bibtex
 
 @misc{https://doi.org/10.5061/dryad.8515,
   doi = {10.5061/DRYAD.8515},
@@ -712,7 +773,7 @@ bolognese read 10.5061/DRYAD.8515 --as bibtex
 Convert schema.org/JSON-LD to DataCite XML:
 
 ```
-bolognese read https://blog.datacite.org/eating-your-own-dog-food --as datacite
+bolognese https://blog.datacite.org/eating-your-own-dog-food -t datacite
 
 <?xml version="1.0" encoding="UTF-8"?>
 <resource xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns="http://datacite.org/schema/kernel-4" xsi:schemaLocation="http://datacite.org/schema/kernel-4 http://schema.datacite.org/meta/kernel-4/metadata.xsd">
@@ -763,7 +824,7 @@ bolognese read https://blog.datacite.org/eating-your-own-dog-food --as datacite
 Convert schema.org/JSON-LD to BibTeX:
 
 ```
-bolognese read https://blog.datacite.org/eating-your-own-dog-food --as bibtex
+bolognese https://blog.datacite.org/eating-your-own-dog-food -t bibtex
 
 @article{https://doi.org/10.5438/4k3m-nyvg,
   doi = {10.5438/4k3m-nyvg},

data/bin/bolognese

@@ -2,4 +2,8 @@
 
 require File.expand_path("../../lib/bolognese", __FILE__)
 
-Bolognese::CLI.start
+if (ARGV & %w(--version -v help --help)).empty?
+  Bolognese::CLI.start(ARGV.dup.unshift("convert"))
+else
+  Bolognese::CLI.start
+end

data/lib/bolognese/bibtex.rb

@@ -95,7 +95,11 @@ module Bolognese
     end
 
     def description
-      metadata.note.to_s.presence
+      metadata.field?(:abstract) && metadata.abstract.to_s.presence
+    end
+
+    def license
+      metadata.field?(:copyright) && metadata.copyright.to_s.presence
     end
   end
 end

data/lib/bolognese/cli.rb

@@ -10,10 +10,8 @@ module Bolognese
     include Bolognese::DoiUtils
     include Bolognese::Utils
 
-    default_task :open
-
     def self.exit_on_failure?
-      true
+      false
     end
 
     # from http://stackoverflow.com/questions/22809972/adding-a-version-option-to-a-ruby-thor-cli
@@ -24,30 +22,31 @@ module Bolognese
       puts Bolognese::VERSION
     end
 
-    desc "read id", "read metadata for ID"
-    method_option :as, default: "schema_org"
-    method_option :schema_version
-    def read(id)
-      id = normalize_id(id)
-      provider = find_provider(id)
-
-      set_metadata(id: id, provider: provider, as: options[:as], schema_version: options[:schema_version])
-    end
-
-    desc "open file", "read metadata from file"
-    method_option :as, default: "schema_org"
+    desc "", "convert metadata"
+    method_option :from, aliases: "-f"
+    method_option :to, aliases: "-t", default: "schema_org"
     method_option :schema_version
-    def open(file)
-      ext = File.extname(file)
-      unless %w(.bib).include? ext
-        $stderr.puts "File type #{ext} not supported"
-        exit 1
+    def convert(input)
+      id = normalize_id(input)
+
+      if id.present?
+        from = options[:from] || find_from_format(id: id)
+      else
+        ext = File.extname(input)
+        if %w(.bib .xml).include? ext
+          string = IO.read(input)
+        else
+          $stderr.puts "File type #{ext} not supported"
+          exit 1
+        end
+        from = options[:from] || find_from_format(string: string, ext: ext)
       end
 
-      string = IO.read(file)
-      provider = "bibtex"
+      to = options[:to] || "schema_org"
 
-      set_metadata(string: string, provider: provider, as: options[:as], schema_version: options[:schema_version])
+      write(id: id, string: string, from: from, to: to, schema_version: options[:schema_version])
     end
+
+    default_task :convert
   end
 end

data/lib/bolognese/metadata.rb

@@ -15,7 +15,8 @@ module Bolognese
     attr_reader :id, :raw, :provider, :schema_version, :license, :citation,
       :additional_type, :alternate_name, :url, :version, :keywords, :editor,
       :page_start, :page_end, :date_modified, :language, :spatial_coverage,
-      :content_size, :funder, :journal, :bibtex_type
+      :content_size, :funder, :journal, :bibtex_type, :date_created, :has_part,
+      :publisher, :contributor, :schema_version
 
     alias_method :datacite, :as_datacite
 
@@ -23,30 +24,10 @@ module Bolognese
       date_published && date_published[0..3]
     end
 
-    def date_created
-
-    end
-
     def pagination
       [page_start, page_end].compact.join("-").presence
     end
 
-    def has_part
-
-    end
-
-    def publisher
-
-    end
-
-    def contributor
-
-    end
-
-    def schema_version
-
-    end
-
     def author_string
       author.map { |a| [a["familyName"], a["givenName"]].join(", ") }
             .join(" and ").presence
@@ -96,6 +77,7 @@ module Bolognese
         url: url,
         author: author_string,
         keywords: keywords,
+        language: language,
         title: name,
         journal: journal,
         pages: pagination,

data/lib/bolognese/utils.rb

@@ -1,22 +1,51 @@
 module Bolognese
   module Utils
-    def set_metadata(id: nil, string: nil, provider: nil, **options)
-      output = options[:as] || "schema_org"
 
-      if provider.present?
-        p = case provider
-            when "crossref" then Crossref.new(id: id)
-            when "datacite" then Datacite.new(id: id, schema_version: options[:schema_version])
+    def find_from_format(id: nil, string: nil, ext: nil)
+      if id.present?
+        find_from_format_by_id(id)
+      elsif string.present?
+        find_from_format_by_string(string, ext: ext)
+      end
+    end
+
+    def find_from_format_by_id(id)
+      id = normalize_id(id)
+
+      if /\A(?:(http|https):\/\/(dx\.)?doi.org\/)?(doi:)?(10\.\d{4,5}\/.+)\z/.match(id)
+        get_doi_ra(id).fetch("id", nil)
+      elsif /\A(?:(http|https):\/\/orcid\.org\/)?(\d{4}-\d{4}-\d{4}-\d{3}[0-9X]+)\z/.match(id)
+        "orcid"
+      else
+        "schema_org"
+      end
+    end
+
+    def find_from_format_by_string(string, options={})
+      if options[:ext] == ".bib"
+        "bibtex"
+      elsif Maremma.from_xml(string).dig("doi_records", "doi_record", "crossref")
+        "crossref"
+      elsif Maremma.from_xml(string).dig("resource", "xmlns") == "http://datacite.org/schema/kernel-4"
+        "datacite"
+      end
+    end
+
+    def write(id: nil, string: nil, from: nil, to: nil, **options)
+      if from.present?
+        p = case from
+            when "crossref" then Crossref.new(id: id, string: string)
+            when "datacite" then Datacite.new(id: id, string: string, schema_version: options[:schema_version])
             when "bibtex" then Bibtex.new(string: string)
             else SchemaOrg.new(id: id)
             end
 
-        puts p.send(output)
+        puts p.send(to)
       else
         puts "not implemented"
       end
     end
-    
+
     def orcid_from_url(url)
       Array(/\Ahttp:\/\/orcid\.org\/(.+)/.match(url)).last
     end
@@ -61,22 +90,10 @@ module Bolognese
       end
     end
 
-    def find_provider(id)
-      id = normalize_id(id)
-
-      if /\A(?:(http|https):\/\/(dx\.)?doi.org\/)?(doi:)?(10\.\d{4,5}\/.+)\z/.match(id)
-        get_doi_ra(id).fetch("id", nil)
-      elsif /\A(?:(http|https):\/\/orcid\.org\/)?(\d{4}-\d{4}-\d{4}-\d{3}[0-9X]+)\z/.match(id)
-        "orcid"
-      else
-        "schema_org"
-      end
-    end
-
     def normalize_id(id)
       return nil unless id.present?
 
-      normalize_doi(id) || PostRank::URI.clean(id)
+      normalize_doi(id) || Addressable::URI.parse(id).host && PostRank::URI.clean(id)
     end
 
     def normalize_orcid(orcid)

data/lib/bolognese/version.rb

@@ -1,3 +1,3 @@
 module Bolognese
-  VERSION = "0.6.2"
+  VERSION = "0.6.3"
 end

data/spec/bibtex_spec.rb

@@ -17,6 +17,8 @@ describe Bolognese::Bibtex, vcr: true do
                                     {"@type"=>"Person", "givenName"=>"Ioannis", "familyName"=>"Xenarios"},
                                     {"@type"=>"Person", "givenName"=>"Christian S", "familyName"=>"Hardtke"}])
       expect(subject.name).to eq("Automated quantitative histology reveals vascular morphodynamics during Arabidopsis hypocotyl secondary growth")
+      expect(subject.description).to start_with("Among various advantages, their small size makes model organisms preferred subjects of investigation.")
+      expect(subject.license).to eq("http://creativecommons.org/licenses/by/3.0/")
       expect(subject.date_published).to eq("2014")
       expect(subject.is_part_of).to eq("@type"=>"Periodical", "name"=>"eLife", "issn"=>"2050-084X")
     end
@@ -28,6 +30,7 @@ describe Bolognese::Bibtex, vcr: true do
       datacite = Maremma.from_xml(subject.as_datacite).fetch("resource", {})
       expect(datacite.dig("resourceType", "resourceTypeGeneral")).to eq("Text")
       expect(datacite.dig("titles", "title")).to eq("Automated quantitative histology reveals vascular morphodynamics during Arabidopsis hypocotyl secondary growth")
+      expect(datacite.dig("descriptions", "description", "__content__")).to start_with("Among various advantages, their small size makes model organisms preferred subjects of investigation.")
       expect(datacite.dig("creators", "creator").count).to eq(5)
       expect(datacite.dig("creators", "creator").first).to eq("creatorName"=>"Sankar, Martial", "givenName"=>"Martial", "familyName"=>"Sankar")
     end