briard 2.8.1 → 2.9.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: debbaf7e340fa77df735b4443f7bf7120842908f9223250ce905e398c4754df5
-  data.tar.gz: fb782cc48aeaa3435ef4b9eb4004de44786ea3dbb257c653ce62213e55ab6612
+  metadata.gz: 630bdfe771252795f1d396dc2eb3da9fce76c3bf8c0281347be8c8e97e5ae1f4
+  data.tar.gz: e80734a4a4163ce8a6ff581aa9d49034ddff4b961953ab8740566d09e2886e73
 SHA512:
-  metadata.gz: dd60f155303d1e52ec8acf01ae8b648cc4625a10e52473f613a1aa34ed0b7f8b4f7af42dff2938f7b35bfdfd8c62bc0ae1e58e17a799d816470c0743df945118
-  data.tar.gz: 056d24ffef47ed0a3f04a5222464a507b18a00b1f573e437ac10c8de2dd89b581da7ae507528b23039f5c7a7ead79c042b566b9d37bf259482ac59f104cdcec3
+  metadata.gz: 5dfce207ab441f813d69fb79fb0d354823234a8fced4566703de4cdb67f4522c99b4836e1e7d5abeedb7863ad82f082019d2a2289b019ddf56048b8afe959ecf
+  data.tar.gz: 07dfbff46d904dc090d640d8e8d26591d3d88ec6d841e6f781c2804946e679fc03c7d0407943e516af158d34a6e5ec0379d2f5d8f2799efad8c57d7b77ebbba9

data/CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## [v2.8.2](https://github.com/front-matter/briard/tree/v2.8.2) (2022-11-23)
+
+[Full Changelog](https://github.com/front-matter/briard/compare/v2.8.0...v2.8.2)
+
 ## [v2.8.0](https://github.com/front-matter/briard/tree/v2.8.0) (2022-11-22)
 
 [Full Changelog](https://github.com/front-matter/briard/compare/v2.6.5...v2.8.0)

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    briard (2.8.1)
+    briard (2.9.0)
       activesupport (>= 4.2.5, < 8.0)
       base32-url (>= 0.5.0, < 1)
       benchmark_methods (~> 0.7)
@@ -18,6 +18,7 @@ PATH
       gender_detector (~> 0.1.2)
       iso8601 (~> 0.9.1)
       json-ld-preloaded (~> 3.1, >= 3.1.3)
+      json_schemer (~> 0.2.23)
       jsonlint (~> 0.3.0)
       loofah (~> 2.19)
       maremma (>= 4.9.7, < 5)
@@ -75,6 +76,8 @@ GEM
       scanf (~> 1.0)
       sxp (~> 1.2)
       unicode-types (~> 1.7)
+    ecma-re-validator (0.4.0)
+      regexp_parser (~> 2.2)
     edtf (3.1.0)
       activesupport (>= 3.0, < 8.0)
     excon (0.71.1)
@@ -99,6 +102,7 @@ GEM
     haml (5.2.2)
       temple (>= 0.8.0)
       tilt
+    hana (1.3.7)
     hashdiff (1.0.1)
     htmlentities (4.3.4)
     i18n (1.12.0)
@@ -116,6 +120,11 @@ GEM
     json-ld-preloaded (3.2.2)
       json-ld (~> 3.2)
       rdf (~> 3.2)
+    json_schemer (0.2.23)
+      ecma-re-validator (~> 0.3)
+      hana (~> 1.3)
+      regexp_parser (~> 2.0)
+      uri_template (~> 0.7)
     jsonlint (0.3.0)
       oj (~> 3)
       optimist (~> 3)
@@ -245,6 +254,7 @@ GEM
     unicode-display_width (2.3.0)
     unicode-types (1.8.0)
     unicode_utils (1.4.0)
+    uri_template (0.7.0)
     vcr (3.0.3)
     webmock (3.18.1)
       addressable (>= 2.8.0)

data/briard.gemspec

@@ -35,6 +35,7 @@ Gem::Specification.new do |s|
   s.add_dependency 'iso8601', '~> 0.9.1'
   s.add_dependency 'json-ld-preloaded', '~> 3.1', '>= 3.1.3'
   s.add_dependency 'jsonlint', '~> 0.3.0'
+  s.add_dependency 'json_schemer', '~> 0.2.23'
   s.add_dependency 'loofah', '~> 2.19'
   s.add_dependency 'maremma', '>= 4.9.7', '< 5'
   s.add_dependency 'namae', '~> 1.0'

data/lib/briard/metadata.rb

@@ -18,7 +18,26 @@ module Briard
       id = normalize_id(options[:input], options)
       ra = nil
 
-      if options[:input].present? && File.exist?(options[:input])
+      if id.present?
+        @from = options[:from] || find_from_format(id: id)
+
+        # mEDRA, KISTI, JaLC and OP DOIs are found in the Crossref index
+        case @from
+        when 'medra'
+          ra = 'mEDRA'
+        when 'kisti'
+          ra = 'KISTI'
+        when 'jalc'
+          ra = 'JaLC'
+        when 'op'
+          ra = 'OP'
+        end
+
+        # generate name for method to call dynamically
+        hsh = @from.present? ? send("get_#{@from}", id: id, **options) : {}
+        string = hsh.fetch('string', nil)
+
+      elsif options[:input].present? && File.exist?(options[:input])
         filename = File.basename(options[:input])
         ext = File.extname(options[:input])
         if %w[.bib .ris .xml .json .cff].include?(ext)
@@ -35,29 +54,11 @@ module Briard
             'content_url' => options[:content_url]
           }
           string = File.read(options[:input])
-          @from = options[:from] || find_from_format(string: string, filename: filename, ext: ext)
+          @from = options[:from] || find_from_format(string: string, ext: ext)
         else
           warn "File type #{ext} not supported"
           exit 1
         end
-      elsif id.present?
-        @from = options[:from] || find_from_format(id: id)
-
-        # mEDRA, KISTI, JaLC and OP DOIs are found in the Crossref index
-        case @from
-        when 'medra'
-          ra = 'mEDRA'
-        when 'kisti'
-          ra = 'KISTI'
-        when 'jalc'
-          ra = 'JaLC'
-        when 'op'
-          ra = 'OP'
-        end
-
-        # generate name for method to call dynamically
-        hsh = @from.present? ? send("get_#{@from}", id: id, **options) : {}
-        string = hsh.fetch('string', nil)
       else
         hsh = {
           'url' => options[:url],
@@ -162,9 +163,11 @@ module Briard
       exists? && errors.nil?
     end
 
-    # validate against DataCite schema, unless already errors in the reader
+    # Catch errors in the reader
+    # Then validate against JSON schema for internal metadata format
+    # Then validate against DataCite schema, unless already errors in the reader
     def errors
-      meta.fetch('errors', nil) || datacite_errors(xml: datacite, schema_version: schema_version)
+      meta.fetch('errors', nil) || json_schema_errors || datacite_errors(xml: datacite, schema_version: schema_version)
     end
 
     def descriptions

data/lib/briard/metadata_utils.rb

@@ -4,6 +4,7 @@ require_relative 'doi_utils'
 require_relative 'author_utils'
 require_relative 'crossref_utils'
 require_relative 'datacite_utils'
+require_relative 'schema_utils'
 require_relative 'utils'
 
 require_relative 'readers/bibtex_reader'
@@ -42,6 +43,7 @@ module Briard
     include Briard::AuthorUtils
     include Briard::CrossrefUtils
     include Briard::DataciteUtils
+    include Briard::SchemaUtils
     include Briard::Utils
 
     include Briard::Readers::BibtexReader

data/lib/briard/readers/crossref_json_reader.rb

@@ -7,8 +7,10 @@ module Briard
           return { 'string' => nil, 'state' => 'not_found' } unless id.present?
 
           url = crossref_api_url(id, options)
-          response = Maremma.get(url, raw: true)
-          string = response.body.fetch('data', nil)
+          response = Maremma.get(url)
+          return { 'string' => nil, 'state' => 'not_found' } if response.body.dig('data', 'status') != 'ok'
+          
+          string = response.body.dig('data', 'message').to_json
 
           { 'string' => string }
         end
@@ -21,9 +23,8 @@ module Briard
   
           read_options = ActiveSupport::HashWithIndifferentAccess.new(options.except(:doi, :id, :url,
                                                                                      :sandbox, :validate, :ra))
-  
-          meta = string.present? ? Maremma.from_json(string).fetch('message', {}) : {}
-  
+          meta = string.present? ? Maremma.from_json(string) : {}
+
           resource_type = meta.fetch('type', nil)
           resource_type = resource_type.present? ? resource_type.underscore.camelcase : nil
 

data/lib/briard/readers/datacite_reader.rb

@@ -310,7 +310,7 @@ module Briard
           'container' => set_container(meta),
           'publisher' => parse_attributes(meta.fetch('publisher', nil),
                                           first: true).to_s.strip.presence,
-          'agency' => 'datacite',
+          'agency' => 'DataCite',
           'funding_references' => funding_references,
           'dates' => dates,
           'publication_year' => parse_attributes(meta.fetch('publicationYear', nil),

data/lib/briard/schema_utils.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+require 'json_schemer'
+require 'pathname'
+
+module Briard
+  module SchemaUtils
+    JSON_SCHEMA = schema = File.read(File.expand_path('../../resources/json-schema/briard_schema.json', __dir__))
+    
+    def json_schema_errors
+      schemer = JSONSchemer.schema(JSON_SCHEMA)
+      errors = schemer.validate(self.meta).to_a
+      errors.map {|err| JSONSchemer::Errors.pretty err }.presence
+    end
+  end
+end

data/lib/briard/utils.rb

@@ -490,10 +490,12 @@ module Briard
     def find_from_format(id: nil, string: nil, ext: nil, filename: nil)
       if id.present?
         find_from_format_by_id(id)
-      elsif ext.present?
-        find_from_format_by_filename(filename) || find_from_format_by_ext(string, ext: ext)
+      elsif string.present? && ext.present?
+        find_from_format_by_ext(string, ext: ext)
       elsif string.present?
         find_from_format_by_string(string)
+      elsif filename.present?
+        find_from_format_by_filename(filename)
       else
         'datacite'
       end
@@ -574,6 +576,8 @@ module Briard
         'codemeta'
       elsif Maremma.from_json(string).to_h.dig('schema-version').to_s.start_with?('http://datacite.org/schema/kernel')
         'datacite_json'
+      elsif Maremma.from_json(string).to_h.dig('source') == ('Crossref')
+        'crossref_json'
       elsif Maremma.from_json(string).to_h.dig('types').present? && Maremma.from_json(string).to_h.dig('publication_year').present?
         'crosscite'
       elsif Maremma.from_json(string).to_h.dig('issued', 'date-parts').present?

data/lib/briard/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Briard
-  VERSION = '2.8.1'
+  VERSION = '2.9.0'
 end

data/resources/json-schema/briard_schema.json

@@ -0,0 +1,367 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "title": "Briard Schema",
+  "definitions": {
+    "affiliations": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "properties": {
+          "affiliation": {
+            "type": "string"
+          }
+        },
+        "required": ["affiliation"]
+      },
+      "uniqueItems": true
+    },
+    "contributorType": {
+      "type": "string",
+      "enum": [
+        "ContactPerson",
+        "DataCollector",
+        "DataCurator",
+        "DataManager",
+        "Distributor",
+        "Editor",
+        "HostingInstitution",
+        "Producer",
+        "ProjectLeader",
+        "ProjectManager",
+        "ProjectMember",
+        "RegistrationAgency",
+        "RegistrationAuthority",
+        "RelatedPerson",
+        "Researcher",
+        "ResearchGroup",
+        "RightsHolder",
+        "Sponsor",
+        "Supervisor",
+        "WorkPackageLeader",
+        "Other"
+      ]
+    },
+    "descriptionType": {
+      "type": "string",
+      "enum": [
+        "Abstract",
+        "Methods",
+        "SeriesInformation",
+        "TableOfContents",
+        "TechnicalInfo",
+        "Other"
+      ]
+    },
+    "funderIdentifierType": {
+      "type": "string",
+      "enum": ["ISNI", "GRID", "Crossref Funder ID", "Other"]
+    },
+    "nameIdentifiers": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "properties": {
+          "nameIdentifier": {
+            "type": "string"
+          },
+          "nameIdentifierScheme": {
+            "type": "string"
+          },
+          "schemeURI": {
+            "type": "string",
+            "format": "uri"
+          }
+        },
+        "required": ["nameIdentifier", "nameIdentifierScheme"]
+      },
+      "uniqueItems": true
+    },
+    "nameType": {
+      "type": "string",
+      "enum": ["Organizational", "Personal"]
+    },
+    "noEmptyString": {
+        "type": "string",
+        "minLength": 1
+    },
+    "resourceTypeGeneral": {
+      "type": "string",
+      "enum": [
+        "Audiovisual",
+        "Collection",
+        "DataPaper",
+        "Dataset",
+        "Event",
+        "Image",
+        "InteractiveResource",
+        "Model",
+        "PhysicalObject",
+        "Service",
+        "Software",
+        "Sound",
+        "Text",
+        "Workflow",
+        "Other"
+      ]
+    },
+    "titleType": {
+      "type": "string",
+      "enum": ["AlternativeTitle", "Subtitle", "TranslatedTitle", "Other"]
+    }
+  },
+  "prefixItems": {
+    "id": {
+      "type": "string",
+      "format": "uri"
+    },
+    "types": {
+      "type": "object",
+      "properties": {
+        "resourceType": {
+          "type": "string"
+        },
+        "resourceTypeGeneral": {
+          "$ref": "#/definitions/resourceTypeGeneral"
+        },
+        "schemaOrg": {
+          "type": "string"
+        },
+        "citeproc": {
+          "type": "string"
+        },
+        "bibtex": {
+          "type": "string"
+        },
+        "ris": {
+          "type": "string"
+        }
+      },
+      "required": [
+        "resourceType",
+        "resourceTypeGeneral",
+        "schemaOrg",
+        "citeproc",
+        "bibtex",
+        "ris"
+      ]
+    },
+    "doi": {
+      "type": "string",
+      "format": "uri"
+    },
+    "url": {
+      "type": "string",
+      "format": "uri"
+    },
+    "titles": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "properties": {
+          "title": {
+            "type": "string"
+          },
+          "titleType": {
+            "$ref": "#/definitions/titleType"
+          },
+          "lang": {
+            "type": "string"
+          }
+        },
+        "required": ["title"]
+      },
+      "minItems": 1,
+      "uniqueItems": true
+    },
+    "creators": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "properties": {
+          "name": {
+            "type": "string"
+          },
+          "nameType": {
+            "$ref": "#/definitions/nameType"
+          },
+          "givenName": {
+            "type": "string"
+          },
+          "familyName": {
+            "type": "string"
+          },
+          "nameIdentifiers": {
+            "$ref": "#/definitions/nameIdentifiers"
+          },
+          "affiliations": {
+            "$ref": "#/definitions/affiliations"
+          },
+          "lang": {
+            "type": "string"
+          }
+        },
+        "required": ["name"]
+      },
+      "minItems": 1,
+      "uniqueItems": true
+    },
+    "contributors": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "properties": {
+          "contributorType": {
+            "$ref": "#/definitions/contributorType"
+          },
+          "name": {
+            "type": "string"
+          },
+          "nameType": {
+            "$ref": "#/definitions/nameType"
+          },
+          "givenName": {
+            "type": "string"
+          },
+          "familyName": {
+            "type": "string"
+          },
+          "nameIdentifiers": {
+            "$ref": "#/definitions/nameIdentifiers"
+          },
+          "affiliations": {
+            "$ref": "#/definitions/affiliations"
+          },
+          "lang": {
+            "type": "string"
+          }
+        },
+        "required": ["contributorType", "name"]
+      },
+      "uniqueItems": true
+    },
+    "funding_references": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "properties": {
+          "funderName": {
+            "type": "string"
+          },
+          "funderIdentifier": {
+            "type": "string"
+          },
+          "funderIdentifierType": {
+            "$ref": "#/definitions/funderIdentifierType"
+          },
+          "awardNumber": {
+            "type": "string"
+          },
+          "awardURI": {
+            "type": "string",
+            "format": "uri"
+          },
+          "awardTitle": {
+            "type": "string"
+          }
+        },
+        "required": ["funderName"]
+      },
+      "uniqueItems": true
+    },
+    "publisher": {
+      "$ref": "#/definitions/noEmptyString"
+    },
+    "publication_year": {
+      "$ref": "#/definitions/noEmptyString"
+    },
+    "language": {
+      "type": "string",
+      "$comment": "Primary language of the resource. Allowed values are taken from  IETF BCP 47, ISO 639-1 language codes."
+    },
+    "sizes": {
+      "type": "array",
+      "items": {
+        "type": "string"
+      },
+      "uniqueItems": true
+    },
+    "formats": {
+      "type": "array",
+      "items": {
+        "type": "string"
+      },
+      "uniqueItems": true
+    },
+    "version_info": {
+      "type": "string"
+    },
+    "rights_list": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "properties": {
+          "rights": {
+            "type": "string"
+          },
+          "rightsURI": {
+            "type": "string",
+            "format": "uri"
+          },
+          "rightsIdentifier": {
+            "type": "string"
+          },
+          "rightsIdentifierScheme": {
+            "type": "string"
+          },
+          "schemeURI": {
+            "type": "string",
+            "format": "uri"
+          },
+          "lang": {
+            "type": "string"
+          }
+        }
+      },
+      "uniqueItems": true
+    },
+    "descriptions": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "properties": {
+          "description": {
+            "$ref": "#/definitions/noEmptyString"
+          },
+          "descriptionType": {
+            "$ref": "#/definitions/descriptionType"
+          },
+          "lang": {
+            "type": "string"
+          }
+        },
+        "required": ["description", "descriptionType"]
+      },
+      "uniqueItems": true
+    },
+    "agency": {
+      "type": "string",
+      "enum": [ "Crossref", "DataCite", "JaLC", "KISTI", "mEDRA", "OP"]
+    },
+    "schema_version": {
+      "type": "string",
+      "const": "http://datacite.org/schema/kernel-4"
+    },
+    "state": {
+      "type": "string",
+      "enum": [ "draft", "findable", "registered", "not_found"]
+    }
+  },
+  "required": [
+    "id",
+    "creators",
+    "titles",
+    "publisher",
+    "publication_year",
+    "types"
+  ]
+}