bolognese 0.8.5 → 0.8.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: e7ecdd8495d5c59b4745c4e474f00fdfb43bc689
-  data.tar.gz: c067a911205b6ca30f82ceee7579a3b7e2cfa45f
+  metadata.gz: 7649099d379d9ce1513465ff2d9f012139542c2a
+  data.tar.gz: 476243eda3bd701c7677e36c7f49692c202347f6
 SHA512:
-  metadata.gz: 08628597e24777b6116c7cd6106206e528211444daab260c609fa3f8beb08fd56fc0eba55db6f2d262686601fc9fa190f34cf4f6a0baa8f0607367f22fed7ea3
-  data.tar.gz: 62470519d4523414df33e4b546db7d41057af0f68c62e4b8af537d97c281f972b549a88b02d81660688b0f7418c523a2666d0e8014cdcdf15b43af5908a68443
+  metadata.gz: 35d1ec8c50ddb12e3fd155b020bc0c4b542740b57c553ccf05739f0832eecef7c0fadc099ce17be355c42b24cd7711146540fd53d37fb995d1e2caefc3b169a7
+  data.tar.gz: a1ac1c3e04f106c0161d6e6fcc2f7a21a7691ee77f0c1f4f66a309758d5a6f20e3555fe8b7a4384df564d518a1df17f62ad6085a3d09fde54bef8e31edb7517a

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bolognese (0.8.5)
+    bolognese (0.8.7)
       activesupport (~> 4.2, >= 4.2.5)
       bibtex-ruby (~> 4.1)
       builder (~> 3.2, >= 3.2.2)
@@ -21,12 +21,13 @@ GEM
       minitest (~> 5.1)
       thread_safe (~> 0.3, >= 0.3.4)
       tzinfo (~> 1.1)
-    addressable (2.3.8)
+    addressable (2.5.1)
+      public_suffix (~> 2.0, >= 2.0.2)
     bibtex-ruby (4.4.3)
       latex-decode (~> 0.0)
     builder (3.2.3)
-    codeclimate-test-reporter (1.0.6)
-      simplecov
+    codeclimate-test-reporter (1.0.8)
+      simplecov (<= 0.13)
     colorize (0.8.1)
     crack (0.4.3)
       safe_yaml (~> 1.0.0)
@@ -64,8 +65,8 @@ GEM
     namae (0.10.2)
     nokogiri (1.6.8.1)
       mini_portile2 (~> 2.1.0)
-    oj (2.18.2)
-    postrank-uri (1.0.20)
+    oj (2.18.5)
+    postrank-uri (1.0.22)
       addressable (>= 2.3.0, < 2.6)
       nokogiri (>= 1.6.1, < 1.8)
       public_suffix (>= 2.0.0, < 2.1)
@@ -98,9 +99,9 @@ GEM
     simplecov-html (0.10.0)
     thor (0.19.4)
     thread_safe (0.3.6)
-    tzinfo (1.2.2)
+    tzinfo (1.2.3)
       thread_safe (~> 0.1)
-    unicode (0.4.4.2)
+    unicode (0.4.4.4)
     vcr (3.0.3)
     webmock (1.24.6)
       addressable (>= 2.3.6)
@@ -123,4 +124,4 @@ DEPENDENCIES
   webmock (~> 1.22, >= 1.22.3)
 
 BUNDLED WITH
-   1.12.5
+   1.14.6

data/README.md

@@ -39,7 +39,7 @@ Bolognese reads and/or writes these metadata formats:
     </tr>
     <tr>
       <td><a href='https://api.datacite.org/'>DataCite JSON</a></td>
-      <td>datacite</td>
+      <td>datacite_json</td>
       <td>application/vnd.datacite+json</td>
       <td>Yes</td>
       <td>Yes</td>
@@ -51,6 +51,13 @@ Bolognese reads and/or writes these metadata formats:
       <td>Yes</td>
       <td>Yes</td>
     </tr>
+    <tr>
+      <td><a href='http://gsl-nagoya-u.net/http/pub/citeproc-doc.html'>Citeproc JSON</a></td>
+      <td>citeproc</td>
+      <td>application/vnd.citationstyles.csl+json</td>
+      <td>Yes</td>
+      <td>Yes</td>
+    </tr>
     <tr>
       <td><a href='https://codemeta.github.io/'>Codemeta</a></td>
       <td>codemeta</td>
@@ -65,6 +72,13 @@ Bolognese reads and/or writes these metadata formats:
       <td>Yes</td>
       <td>Yes</td>
     </tr>
+    <tr>
+      <td><a href='http://en.wikipedia.org/wiki/RIS_(file_format)'>RIS</a></td>
+      <td>ris</td>
+      <td>application/x-research-info-systems</td>
+      <td>Yes</td>
+      <td>Yes</td>
+    </tr>
   </tbody>
 </table>
 

data/lib/bolognese.rb

@@ -14,6 +14,8 @@ require "bolognese/datacite_json"
 require "bolognese/schema_org"
 require "bolognese/codemeta"
 require "bolognese/bibtex"
+require "bolognese/ris"
+require "bolognese/citeproc"
 require "bolognese/orcid"
 require "bolognese/cli"
 require "bolognese/string"

data/lib/bolognese/bibtex.rb

@@ -26,6 +26,44 @@ module Bolognese
       "WebSite" => "misc"
     }
 
+    BIB_TO_CP_TRANSLATIONS = {
+      "article" => "article-journal"
+    }
+
+    BIB_TO_RIS_TRANSLATIONS = {
+      "article" => "JOUR",
+      "book" => "BOOK",
+      "inbook" => "CHAP",
+      "inproceedings" => "CPAPER",
+      "manual" => nil,
+      "misc" => "GEN",
+      "phdthesis" => "THES",
+      "proceedings" => "CONF",
+      "techreport" => "RPRT",
+      "unpublished" => "UNPD"
+    }
+
+    SO_TO_RIS_TRANSLATIONS = {
+      "Article" => "",
+      "AudioObject" => "song",
+      "Blog" => nil,
+      "BlogPosting" => "BLOG",
+      "Collection" => nil,
+      "CreativeWork" => "GEN",
+      "DataCatalog" => "CTLG",
+      "Dataset" => "DATA",
+      "Event" => nil,
+      "ImageObject" => "FIGURE",
+      "Movie" => "MPCT",
+      "PublicationIssue" => nil,
+      "ScholarlyArticle" => "JOUR",
+      "Service" => nil,
+      "SoftwareSourceCode" => "COMP",
+      "VideoObject" => "VIDEO",
+      "WebPage" => "ELEC",
+      "WebSite" => nil
+    }
+
     def initialize(string: nil)
       @raw = string
     end
@@ -46,6 +84,14 @@ module Bolognese
       BIB_TO_SO_TRANSLATIONS[metadata.type.to_s] || "ScholarlyArticle"
     end
 
+    def citeproc_type
+      BIB_TO_CP_TRANSLATIONS[metadata.type.to_s] || "misc"
+    end
+
+    def ris_type
+      BIB_TO_RIS_TRANSLATIONS[metadata.type.to_s] || "GEN"
+    end
+
     def resource_type_general
       SO_TO_DC_TRANSLATIONS[type]
     end
@@ -75,9 +121,11 @@ module Bolognese
     end
 
     def container_title
-      metadata.journal.to_s.presence || metadata.pubisher.to_s
+      metadata.journal.to_s.presence || metadata.publisher.to_s
     end
 
+    alias_method :journal, :container_title
+
     def date_published
       metadata.date.to_s.presence
     end
@@ -86,6 +134,14 @@ module Bolognese
       metadata.year.to_s.presence
     end
 
+    def volume
+      metadata.volume.to_s.presence
+    end
+
+    def pagination
+      metadata.pages.to_s.presence
+    end
+
     def is_part_of
       if metadata.journal.present?
         { "type" => "Periodical",

data/lib/bolognese/citeproc.rb

@@ -0,0 +1,116 @@
+module Bolognese
+  class Citeproc < Metadata
+
+    SO_TO_DC_RELATION_TYPES = {
+      "citation" => "References",
+      "sameAs" => "IsIdenticalTo",
+      "isPartOf" => "IsPartOf",
+      "hasPart" => "HasPart",
+      "isPredecessor" => "IsPreviousVersionOf",
+      "isSuccessor" => "IsNewVersionOf"
+    }
+
+    def initialize(string: nil)
+      @raw = string
+    end
+
+    def metadata
+      @metadata ||= raw.present? ? Maremma.from_json(raw) : {}
+    end
+
+    def exists?
+      metadata.present?
+    end
+
+    def valid?
+      citeproc.present?
+    end
+
+    def doi
+      metadata.fetch("DOI", nil)
+    end
+
+    def id
+      normalize_id(metadata.fetch("id", nil))
+    end
+
+    def url
+      normalize_id(metadata.fetch("URL", nil))
+    end
+
+    def resource_type_general
+      SO_TO_DC_TRANSLATIONS[type]
+    end
+
+    def type
+      CP_TO_SO_TRANSLATIONS[citeproc_type] || "CreativeWork"
+    end
+
+    def citeproc_type
+      metadata.fetch("type", nil)
+    end
+
+    def additional_type
+      metadata.fetch("additionalType", nil)
+    end
+
+    def bibtex_type
+      Bolognese::Bibtex::SO_TO_BIB_TRANSLATIONS[type] || "misc"
+    end
+
+    def ris_type
+      CP_TO_RIS_TRANSLATIONS[type] || "GEN"
+    end
+
+    def title
+      metadata.fetch("title", nil)
+    end
+
+    def author
+      authors = from_citeproc(Array.wrap(metadata.fetch("author", nil)))
+      get_authors(authors)
+    end
+
+    def editor
+      editors = from_citeproc(Array.wrap(metadata.fetch("editor", nil)))
+      get_authors(editors)
+    end
+
+    def description
+      { "text" => metadata.fetch("abstract", nil) }
+    end
+
+    def version
+      metadata.fetch("version", nil)
+    end
+
+    def keywords
+      metadata.fetch("categories", nil)
+    end
+
+    def date_published
+      issued = metadata.fetch("issued", nil)
+      get_date_from_date_parts(issued)
+    end
+
+    def publication_year
+      date_published[0..3].to_i.presence
+    end
+
+    def is_part_of
+      if container_title.present?
+        { "type" => "Periodical",
+          "name" => container_title,
+          "issn" => metadata.fetch("ISSN", nil) }.compact
+      end
+    end
+
+    def publisher
+      metadata.fetch("publisher", nil)
+    end
+
+    def container_title
+      metadata.fetch("container-title", nil)
+    end
+  end
+end

data/lib/bolognese/codemeta.rb

@@ -48,6 +48,14 @@ module Bolognese
       metadata.fetch("@type", nil)
     end
 
+    def citeproc_type
+      "computer_program"
+    end
+
+    def ris_type
+      SO_TO_RIS_TRANSLATIONS[type] || "GEN"
+    end
+
     def additional_type
       metadata.fetch("additionalType", nil)
     end
@@ -87,7 +95,7 @@ module Bolognese
     end
 
     def keywords
-      Array(metadata.fetch("tags", nil)).join(", ").presence
+      Array.wrap(metadata.fetch("tags", nil)).join(", ").presence
     end
 
     def date_created

data/lib/bolognese/crossref.rb

@@ -102,6 +102,10 @@ module Bolognese
       metadata.dig("crossref", "journal", "journal_metadata").presence || {}
     end
 
+    def journal_issue
+      metadata.dig("crossref", "journal", "journal_issue").presence || {}
+    end
+
     def bibliographic_metadata
       metadata.dig("crossref", "journal", "journal_article").presence ||
       metadata.dig("crossref", "conference", "conference_paper").presence ||
@@ -121,6 +125,14 @@ module Bolognese
       CR_TO_SO_TRANSLATIONS[additional_type] || "ScholarlyArticle"
     end
 
+    def citeproc_type
+      CR_TO_CP_TRANSLATIONS[additional_type] || "article-journal"
+    end
+
+    def ris_type
+      CR_TO_RIS_TRANSLATIONS[additional_type] || "JOUR"
+    end
+
     def additional_type
       if metadata.dig("crossref", "journal").present?
         metadata.dig("crossref", "journal").keys.last.camelize
@@ -204,6 +216,14 @@ module Bolognese
       Time.parse(metadata.fetch("timestamp", "")).utc.iso8601
     end
 
+    def volume
+      journal_issue.dig("journal_volume", "volume")
+    end
+
+    def issue
+      journal_issue.dig("issue")
+    end
+
     def page_start
       bibliographic_metadata.dig("pages", "first_page")
     end

data/lib/bolognese/datacite.rb

@@ -66,6 +66,14 @@ module Bolognese
       DC_TO_SO_TRANSLATIONS[resource_type_general.to_s.dasherize] || "CreativeWork"
     end
 
+    def citeproc_type
+      DC_TO_CP_TRANSLATIONS[resource_type_general.to_s.dasherize] || "other"
+    end
+
+    def ris_type
+      SO_TO_RIS_TRANSLATIONS[resource_type_general.to_s.dasherize] || "GEN"
+    end
+
     def additional_type
       metadata.fetch("resourceType", {}).fetch("__content__", nil) ||
       metadata.fetch("resourceType", {}).fetch("resourceTypeGeneral", nil)
@@ -187,7 +195,7 @@ module Bolognese
     end
 
     def publication_year
-      metadata.fetch("publicationYear")
+      metadata.fetch("publicationYear", nil)
     end
 
     def language

data/lib/bolognese/datacite_json.rb

@@ -35,6 +35,14 @@ module Bolognese
       DC_TO_SO_TRANSLATIONS[resource_type_general.to_s.dasherize] || "CreativeWork"
     end
 
+    def citeproc_type
+      DC_TO_CP_TRANSLATIONS[resource_type_general.to_s.dasherize] || "other"
+    end
+
+    def ris_type
+      SO_TO_RIS_TRANSLATIONS[resource_type_general.to_s.dasherize] || "GEN"
+    end
+
     def additional_type
       metadata.fetch("resource-type", nil)
     end
@@ -60,7 +68,7 @@ module Bolognese
     end
 
     def keywords
-      metadata.fetch("subject", nil)
+      Array.wrap(metadata.fetch("subject", nil)).join(", ").presence
     end
 
     def author

data/lib/bolognese/datacite_utils.rb

@@ -67,7 +67,7 @@ module Bolognese
     end
 
     def insert_person(xml, person, type)
-      person_name = person["name"].presence || [person["familyName"], person["givenName"]].compact.join(", ")
+      person_name = person["familyName"].present? ? [person["familyName"], person["givenName"]].compact.join(", ") : person["name"]
 
       xml.send(type + "Name", person_name)
       xml.givenName(person["givenName"]) if person["givenName"].present?

data/lib/bolognese/metadata.rb

@@ -29,6 +29,51 @@ module Bolognese
       "Other" => "CreativeWork"
     }
 
+    DC_TO_CP_TRANSLATIONS = {
+      "Audiovisual" => "motion_picture",
+      "Collection" => nil,
+      "Dataset" => "dataset",
+      "Event" => nil,
+      "Image" => "graphic",
+      "InteractiveResource" => nil,
+      "Model" => nil,
+      "PhysicalObject" => nil,
+      "Service" => nil,
+      "Software" => "computer_program",
+      "Sound" => "song",
+      "Text" => "report",
+      "Workflow" => nil,
+      "Other" => nil
+    }
+
+    CR_TO_CP_TRANSLATIONS = {
+      "proceedings" => nil,
+      "reference-book" => nil,
+      "journal-issue" => nil,
+      "proceedings-article" => "paper-conference",
+      "other" => nil,
+      "dissertation" => "thesis",
+      "dataset" => "dataset",
+      "edited-book" => "book",
+      "journal-article" => "article-journal",
+      "journal" => nil,
+      "report" => "report",
+      "book-series" => nil,
+      "report-series" => nil,
+      "book-track" => nil,
+      "standard" => nil,
+      "book-section" => "chapter",
+      "book-part" => nil,
+      "book" => "book",
+      "book-chapter" => "chapter",
+      "standard-series" => nil,
+      "monograph" => "book",
+      "component" => nil,
+      "reference-entry" => "entry-dictionary",
+      "journal-volume" => nil,
+      "book-set" => nil
+    }
+
     SO_TO_DC_TRANSLATIONS = {
       "Article" => "Text",
       "AudioObject" => "Sound",
@@ -50,6 +95,100 @@ module Bolognese
       "WebSite" => "Text"
     }
 
+    SO_TO_CP_TRANSLATIONS = {
+      "Article" => "",
+      "AudioObject" => "song",
+      "Blog" => "report",
+      "BlogPosting" => "post-weblog",
+      "Collection" => nil,
+      "CreativeWork" => nil,
+      "DataCatalog" => "dataset",
+      "Dataset" => "dataset",
+      "Event" => nil,
+      "ImageObject" => "graphic",
+      "Movie" => "motion_picture",
+      "PublicationIssue" => nil,
+      "ScholarlyArticle" => "article-journal",
+      "Service" => nil,
+      "SoftwareSourceCode" => "computer_program",
+      "VideoObject" => "broadcast",
+      "WebPage" => "webpage",
+      "WebSite" => "webpage"
+    }
+
+    CP_TO_SO_TRANSLATIONS = {
+      "song" => "AudioObject",
+      "post-weblog" => "BlogPosting",
+      "dataset" => "Dataset",
+      "graphic" => "ImageObject",
+      "motion_picture" => "Movie",
+      "article-journal" => "ScholarlyArticle",
+      "computer_program" => "SoftwareSourceCode",
+      "broadcast" => "VideoObject",
+      "webpage" => "WebPage"
+    }
+
+    SO_TO_RIS_TRANSLATIONS = {
+      "Article" => nil,
+      "AudioObject" => nil,
+      "Blog" => nil,
+      "BlogPosting" => "BLOG",
+      "Collection" => nil,
+      "CreativeWork" => "GEN",
+      "DataCatalog" => "CTLG",
+      "Dataset" => "DATA",
+      "Event" => nil,
+      "ImageObject" => "FIGURE",
+      "Movie" => "MPCT",
+      "PublicationIssue" => nil,
+      "ScholarlyArticle" => "JOUR",
+      "Service" => nil,
+      "SoftwareSourceCode" => "COMP",
+      "VideoObject" => "VIDEO",
+      "WebPage" => "ELEC",
+      "WebSite" => nil
+    }
+
+    CR_TO_RIS_TRANSLATIONS = {
+      "proceedings" => "CONF",
+      "reference-book" => "BOOK",
+      "journal-issue" => nil,
+      "proceedings-article" => "CPAPER",
+      "other" => "GEN",
+      "dissertation" => "THES",
+      "dataset" => "DATA",
+      "edited-book" => "BOOK",
+      "journal-article" => "JOUR",
+      "journal" => nil,
+      "report" => nil,
+      "book-series" => nil,
+      "report-series" => nil,
+      "book-track" => nil,
+      "standard" => nil,
+      "book-section" => "CHAP",
+      "book-part" => "CHAP",
+      "book" => "BOOK",
+      "book-chapter" => "CHAP",
+      "standard-series" => nil,
+      "monograph" => "BOOK",
+      "component" => nil,
+      "reference-entry" => "DICT",
+      "journal-volume" => nil,
+      "book-set" => nil
+    }
+
+    CP_TO_RIS_TRANSLATIONS = {
+      "post-weblog" => "BLOG",
+      "dataset" => "DATA",
+      "graphic" => "FIGURE",
+      "book" => "BOOK",
+      "motion_picture" => "MPCT",
+      "article-journal" => "JOUR",
+      "computer_program" => "COMP",
+      "broadcast" => "MPCT",
+      "webpage" => "ELEC"
+    }
+
     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,
@@ -62,7 +201,7 @@ module Bolognese
       :is_referenced_by, :references, :is_documented_by, :documents,
       :is_compiled_by, :compiles, :is_variant_form_of, :is_original_form_of,
       :is_reviewed_by, :reviews, :is_derived_from, :is_source_of, :format,
-      :related_identifier, :reverse
+      :related_identifier, :reverse, :citeproc_type, :ris_type, :volume, :issue
 
     def publication_year
       date_published && date_published[0..3]
@@ -156,6 +295,30 @@ module Bolognese
       JSON.pretty_generate hsh
     end
 
+    def citeproc
+      hsh = {
+        "type" => citeproc_type,
+        "id" => id,
+        "categories" => keywords.present? ? keywords.split(", ") : nil,
+        "language" => language,
+        "author" => to_citeproc(author),
+        "editor" => to_citeproc(editor),
+        "issued" => get_date_parts(date_published),
+        "submitted" => get_date_parts(date_submitted),
+        "abstract" => description.is_a?(Hash) ? description.fetch("text", nil) : description,
+        "container-title" => journal,
+        "DOI" => doi,
+        "issue" => issue,
+        "page" => pagination,
+        "publisher" => publisher,
+        "title" => title,
+        "URL" => url,
+        "version" => version,
+        "volume" => volume
+      }.compact
+      JSON.pretty_generate hsh
+    end
+
     def codemeta
       hsh = {
         "@context" => id.present? ? "https://raw.githubusercontent.com/codemeta/codemeta/master/codemeta.jsonld" : nil,
@@ -193,5 +356,26 @@ module Bolognese
       }.compact
       BibTeX::Entry.new(bib).to_s
     end
+
+    def ris
+      {
+        "TY" => ris_type,
+        "T1" => title,
+        "T2" => container_title,
+        "AU" => to_ris(author),
+        "DO" => doi,
+        "UR" => url,
+        "AB" => description.present? ? description["text"] : nil,
+        "KW" => keywords.to_s.split(", ").presence,
+        "PY" => publication_year,
+        "PB" => publisher,
+        "AN" => alternate_name.present? ? alternate_name["name"] : nil,
+        "LA" => language,
+        "VL" => volume,
+        "IS" => issue,
+        "SP" => pagination,
+        "ER" => ""
+      }.compact.map { |k, v| v.is_a?(Array) ? v.map { |vi| "#{k} - #{vi}" }.join("\r\n") : "#{k} - #{v}" }.join("\r\n")
+    end
   end
 end