briard 2.0

data/spec/doi_utils_spec.rb

@@ -0,0 +1,314 @@
+# frozen_string_literal: true
+
+require 'spec_helper'
+
+describe Briard::Metadata, vcr: true do
+  let(:input) { "https://doi.org/10.1101/097196" }
+
+  subject { Briard::Metadata.new(input: input, from: "crossref") }
+
+  context "doi resolver" do
+    it "doi" do
+      doi = "10.5061/DRYAD.8515"
+      response = subject.doi_resolver(doi)
+      expect(response).to eq("https://doi.org/")
+    end
+
+    it "doi with protocol" do
+      doi = "doi:10.5061/DRYAD.8515"
+      response = subject.doi_resolver(doi)
+      expect(response).to eq("https://doi.org/")
+    end
+
+    it "https url" do
+      doi = "https://doi.org/10.5061/dryad.8515"
+      response = subject.doi_resolver(doi)
+      expect(response).to eq("https://doi.org/")
+    end
+
+    it "dx.doi.org url" do
+      doi = "http://dx.doi.org/10.5061/dryad.8515"
+      response = subject.doi_resolver(doi)
+      expect(response).to eq("https://doi.org/")
+    end
+
+    it "stage resolver" do
+      doi = "https://handle.stage.datacite.org/10.5061/dryad.8515"
+      response = subject.doi_resolver(doi)
+      expect(response).to eq("https://handle.stage.datacite.org/")
+    end
+
+    it "stage resolver http" do
+      doi = "http://handle.stage.datacite.org/10.5061/dryad.8515"
+      response = subject.doi_resolver(doi)
+      expect(response).to eq("https://handle.stage.datacite.org/")
+    end
+
+    it "force stage resolver" do
+      doi = "https://doi.org/10.5061/dryad.8515"
+      response = subject.doi_resolver(doi, sandbox: true)
+      expect(response).to eq("https://handle.stage.datacite.org/")
+    end
+  end
+
+  context "doi_api_url" do
+    it "doi" do
+      doi = "10.5061/DRYAD.8515"
+      response = subject.doi_api_url(doi)
+      expect(response).to eq("https://api.datacite.org/dois/10.5061/dryad.8515?include=media,client")
+    end
+
+    it "doi with protocol" do
+      doi = "doi:10.5061/DRYAD.8515"
+      response = subject.doi_api_url(doi)
+      expect(response).to eq("https://api.datacite.org/dois/10.5061/dryad.8515?include=media,client")
+    end
+
+    it "https url" do
+      doi = "https://doi.org/10.5061/dryad.8515"
+      response = subject.doi_api_url(doi)
+      expect(response).to eq("https://api.datacite.org/dois/10.5061/dryad.8515?include=media,client")
+    end
+
+    it "dx.doi.org url" do
+      doi = "http://dx.doi.org/10.5061/dryad.8515"
+      response = subject.doi_api_url(doi)
+      expect(response).to eq("https://api.datacite.org/dois/10.5061/dryad.8515?include=media,client")
+    end
+
+    it "test resolver" do
+      doi = "https://handle.stage.datacite.org/10.5061/dryad.8515"
+      response = subject.doi_api_url(doi)
+      expect(response).to eq("https://api.stage.datacite.org/dois/10.5061/dryad.8515?include=media,client")
+    end
+
+    it "test resolver http" do
+      doi = "http://handle.stage.datacite.org/10.5061/dryad.8515"
+      response = subject.doi_api_url(doi)
+      expect(response).to eq("https://api.stage.datacite.org/dois/10.5061/dryad.8515?include=media,client")
+    end
+
+    it "force test resolver" do
+      doi = "https://doi.org/10.5061/dryad.8515"
+      response = subject.doi_api_url(doi, sandbox: true)
+      expect(response).to eq("https://api.stage.datacite.org/dois/10.5061/dryad.8515?include=media,client")
+    end
+  end
+
+  context "normalize doi" do
+    it "doi" do
+      doi = "10.5061/DRYAD.8515"
+      response = subject.normalize_doi(doi)
+      expect(response).to eq("https://doi.org/10.5061/dryad.8515")
+    end
+
+    it "doi with protocol" do
+      doi = "doi:10.5061/DRYAD.8515"
+      response = subject.normalize_doi(doi)
+      expect(response).to eq("https://doi.org/10.5061/dryad.8515")
+    end
+
+    it "SICI doi" do
+      doi = "10.1890/0012-9658(2006)87[2832:tiopma]2.0.co;2"
+      response = subject.normalize_doi(doi)
+      expect(response).to eq("https://doi.org/10.1890/0012-9658(2006)87%5B2832:tiopma%5D2.0.co;2")
+    end
+
+    it "https url" do
+      doi = "https://doi.org/10.5061/dryad.8515"
+      response = subject.normalize_doi(doi)
+      expect(response).to eq("https://doi.org/10.5061/dryad.8515")
+    end
+
+    it "dx.doi.org url" do
+      doi = "http://dx.doi.org/10.5061/dryad.8515"
+      response = subject.normalize_doi(doi)
+      expect(response).to eq("https://doi.org/10.5061/dryad.8515")
+    end
+
+    it "not valid doi prefix" do
+      doi = "https://doi.org/20.5061/dryad.8515"
+      response = subject.normalize_doi(doi)
+      expect(response).to be_nil
+    end
+
+    it "doi prefix with string" do
+      doi = "https://doi.org/10.506X/dryad.8515"
+      response = subject.normalize_doi(doi)
+      expect(response).to be_nil
+    end
+
+    it "doi prefix too long" do
+      doi = "https://doi.org/10.506123/dryad.8515"
+      response = subject.normalize_doi(doi)
+      expect(response).to be_nil
+    end
+
+    it "doi from url without doi proxy" do
+      doi = "https://handle.net/10.5061/dryad.8515"
+      response = subject.normalize_doi(doi)
+      expect(response).to be_nil
+    end
+
+    it "url with one slash" do
+      doi = "https:/doi.org/10.5061/dryad.8515"
+      response = subject.normalize_doi(doi)
+      expect(response).to eq("https://doi.org/10.5061/dryad.8515")
+    end
+
+    it "doi from datacite sandbox" do
+      doi = "https://handle.stage.datacite.org/10.5438/55e5-t5c0"
+      response = subject.normalize_doi(doi)
+      expect(response).to eq("https://handle.stage.datacite.org/10.5438/55e5-t5c0")
+    end
+
+    it "doi force datacite sandbox" do
+      doi = "10.5438/55e5-t5c0"
+      response = subject.normalize_doi(doi, sandbox: true)
+      expect(response).to eq("https://handle.stage.datacite.org/10.5438/55e5-t5c0")
+    end
+  end
+
+  context "doi_from_url" do
+    it "url" do
+      doi = subject.doi_from_url("https://doi.org/10.5061/dryad.8515")
+      expect(doi).to eq("10.5061/dryad.8515")
+    end
+
+    it "doi" do
+      doi = subject.doi_from_url("10.5061/dryad.8515")
+      expect(doi).to eq("10.5061/dryad.8515")
+    end
+
+    it "doi with special characters" do
+      doi = subject.doi_from_url("10.5067/terra+aqua/ceres/cldtyphist_l3.004")
+      expect(doi).to eq("10.5067/terra+aqua/ceres/cldtyphist_l3.004")
+    end
+
+    it "not a doi" do
+      doi = subject.doi_from_url("https://doi.org/10.5061")
+      expect(doi).to be nil
+    end
+
+    it "sandbox url" do
+      doi = subject.doi_from_url("https://handle.stage.datacite.org/10.5438/55e5-t5c0")
+      expect(doi).to eq("10.5438/55e5-t5c0")
+    end
+  end
+
+  context "doi registration agency" do
+    it "datacite" do
+      doi = "https://doi.org/10.5061/dryad.8515"
+      response = subject.get_doi_ra(doi)
+      expect(response).to eq("DataCite")
+    end
+
+    it "crossref" do
+      doi = "10.1371/journal.pone.0000030"
+      response = subject.get_doi_ra(doi)
+      expect(response).to eq("Crossref")
+    end
+
+    it "medra" do
+      doi = "https://doi.org/10.1392/roma081203"
+      response = subject.get_doi_ra(doi)
+      expect(response).to eq("mEDRA")
+    end
+
+    it "kisti" do
+      doi = "https://doi.org/10.5012/bkcs.2013.34.10.2889"
+      response = subject.get_doi_ra(doi)
+      expect(response).to eq("KISTI")
+    end
+
+    it "jalc" do
+      doi = "https://doi.org/10.11367/grsj1979.12.283"
+      response = subject.get_doi_ra(doi)
+      expect(response).to eq("JaLC")
+    end
+
+    it "op" do
+      doi = "https://doi.org/10.2791/81962"
+      response = subject.get_doi_ra(doi)
+      expect(response).to eq("OP")
+    end
+
+    it "not a valid prefix" do
+      doi = "https://doi.org/10.a/dryad.8515x"
+      response = subject.get_doi_ra(doi)
+      expect(response).to be_nil
+    end
+
+    it "not found" do
+      doi = "https://doi.org/10.99999/dryad.8515x"
+      response = subject.get_doi_ra(doi)
+      expect(response).to be_nil
+    end
+  end
+
+  context "validate doi" do
+    it "doi" do
+      doi = "10.5061/dryad.8515"
+      response = subject.validate_doi(doi)
+      expect(response).to eq("10.5061/dryad.8515")
+    end
+  end
+
+  context "validate funder doi" do
+    it "funder doi" do
+      doi = "10.13039/501100001711"
+      response = subject.validate_funder_doi(doi)
+      expect(response).to eq("https://doi.org/10.13039/501100001711")
+    end
+
+    it "funder doi without prefix" do
+      doi = "501100001711"
+      response = subject.validate_funder_doi(doi)
+      expect(response).to eq("https://doi.org/10.13039/501100001711")
+    end
+
+    it "non-funder doi" do
+      doi = "10.5061/dryad.8515"
+      response = subject.validate_funder_doi(doi)
+      expect(response).to be_nil
+    end
+
+    it { expect(subject.validate_funder_doi("10.13039/100000050")).to eq "https://doi.org/10.13039/100000050" }
+    it { expect(subject.validate_funder_doi("10.13039/100006492")).to eq "https://doi.org/10.13039/100006492" }
+    it { expect(subject.validate_funder_doi('http://handle.test.datacite.org/10.13039/100000080')).to eq "https://doi.org/10.13039/100000080" }
+    it { expect(subject.validate_funder_doi('https://doi.org/10.13039/100000001')).to eq "https://doi.org/10.13039/100000001" }
+    it { expect(subject.validate_funder_doi('http://doi.org/10.13039/501100001711')).to eq "https://doi.org/10.13039/501100001711" }
+    it { expect(subject.validate_funder_doi('https://dx.doi.org/10.13039/501100001711')).to eq "https://doi.org/10.13039/501100001711" }
+    it { expect(subject.validate_funder_doi('doi:10.13039/501100001711')).to eq "https://doi.org/10.13039/501100001711" }
+    it { expect(subject.validate_funder_doi('10.13039/501100001711')).to eq "https://doi.org/10.13039/501100001711" }
+    it { expect(subject.validate_funder_doi('501100001711')).to eq "https://doi.org/10.13039/501100001711" }
+    it { expect(subject.validate_funder_doi("https://doi.org/10.13039/5monkeymonkey")).to be_nil }
+    it { expect(subject.validate_funder_doi('10.13039/5monkeymonkey')).to be_nil }
+  end
+
+  context "validate prefix" do
+    it "doi" do
+      doi = "10.5061/dryad.8515"
+      response = subject.validate_prefix(doi)
+      expect(response).to eq("10.5061")
+    end
+
+    it "doi with protocol" do
+      doi = "doi:10.5061/dryad.8515"
+      response = subject.validate_prefix(doi)
+      expect(response).to eq("10.5061")
+    end
+
+    it "doi as url" do
+      doi = "https://doi.org/10.5061/dryad.8515"
+      response = subject.validate_prefix(doi)
+      expect(response).to eq("10.5061")
+    end
+
+    it "only prefix" do
+      doi = "10.5061"
+      response = subject.validate_prefix(doi)
+      expect(response).to eq("10.5061")
+    end
+  end
+end

data/spec/find_from_format_spec.rb

@@ -0,0 +1,114 @@
+# frozen_string_literal: true
+
+require 'spec_helper'
+require 'briard/cli'
+
+describe Briard::CLI do
+  context "find_from_format_by_id", vcr: true do
+    let(:subject) do
+      described_class.new
+    end
+
+    it "crossref" do
+      id = "https://doi.org/10.7554/eLife.01567"
+      expect(subject.find_from_format_by_id(id)).to eq("crossref")
+    end
+
+    it "datacite" do
+      id = "https://doi.org/10.5061/DRYAD.8515"
+      expect(subject.find_from_format_by_id(id)).to eq("datacite")
+    end
+
+    it "medra" do
+      id = "https://doi.org/10.1392/roma081203"
+      expect(subject.find_from_format_by_id(id)).to eq("medra")
+    end
+
+    it "kisti" do
+      id = "https://doi.org/10.5012/bkcs.2013.34.10.2889"
+      expect(subject.find_from_format_by_id(id)).to eq("kisti")
+    end
+
+    it "jalc" do
+      id = "https://doi.org/10.11367/grsj1979.12.283"
+      expect(subject.find_from_format_by_id(id)).to eq("jalc")
+    end
+
+    it "op" do
+      id = "https://doi.org/10.2791/81962"
+      expect(subject.find_from_format_by_id(id)).to eq("op")
+    end
+
+    it "codemeta" do
+      id = "https://github.com/datacite/maremma"
+      expect(subject.find_from_format_by_id(id)).to eq("codemeta")
+    end
+
+    it "npm" do
+      id = "https://github.com/datacite/bracco/blob/master/package.json"
+      expect(subject.find_from_format_by_id(id)).to eq("npm")
+    end
+
+    it "schema_org" do
+      id = "https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/GAOC03"
+      expect(subject.find_from_format_by_id(id)).to eq("schema_org")
+    end
+  end
+
+  context "find_from_format_by_filename" do
+    let(:subject) do
+      described_class.new
+    end
+
+    it "npm" do
+      filename = "package.json"
+      expect(subject.find_from_format_by_filename(filename)).to eq("npm")
+    end
+  end
+
+  context "find_from_format_by_string" do
+    let(:subject) do
+      described_class.new
+    end
+
+    it "crossref" do
+      string = IO.read(fixture_path + 'crossref.xml').strip
+      expect(subject.find_from_format_by_string(string)).to eq("crossref")
+    end
+
+    it "datacite" do
+      string = IO.read(fixture_path + 'datacite.xml').strip
+      expect(subject.find_from_format_by_string(string)).to eq("datacite")
+    end
+
+    it "datacite namespaced" do
+      string = IO.read(fixture_path + 'ns0.xml').strip
+      expect(subject.find_from_format_by_string(string)).to eq("datacite")
+    end
+
+    it "codemeta" do
+      string = IO.read(fixture_path + 'codemeta.json').strip
+      expect(subject.find_from_format_by_string(string)).to eq("codemeta")
+    end
+
+    it "schema_org" do
+      string = IO.read(fixture_path + 'schema_org_topmed.json').strip
+      expect(subject.find_from_format_by_string(string)).to eq("schema_org")
+    end
+
+    it "ris" do
+      string = IO.read(fixture_path + 'crossref.ris').strip
+      expect(subject.find_from_format_by_string(string)).to eq("ris")
+    end
+
+    it "ris from pure" do
+      string = IO.read(fixture_path + 'pure.ris').strip
+      expect(subject.find_from_format_by_string(string)).to eq("ris")
+    end
+
+    it "bibtex" do
+      string = IO.read(fixture_path + 'crossref.bib').strip
+      expect(subject.find_from_format_by_string(string)).to eq("bibtex")
+    end
+  end
+end

data/spec/fixtures/aida.json

@@ -0,0 +1,82 @@
+{
+  "@context": "http://schema.org",
+  "@type": "Dataset",
+  "@id": "https://doi.org/10.23698/aida/drov",
+  "name": "Ovary data from the Visual Sweden project DROID",
+  "about": "Pathology",
+  "url": "https://doi.aida.medtech4health.se/10.23698/aida/drov",
+  "author": [{
+    "@type": "Person",
+    "@id": "https://orcid.org/0000-0003-1298-517X",
+    "givenName": "Karin",
+    "familyName": "Lindman",
+    "name": "Karin Lindman"
+  }, {
+    "@type": "Person",
+    "@id": "https://orcid.org/",
+    "givenName": "Jerónimo",
+    "familyName": "Rose",
+    "name": "Jerónimo F. Rose"
+  }, {
+    "@type": "Person",
+    "@id": "https://orcid.org/",
+    "givenName": "Martin",
+    "familyName": "Lindvall",
+    "name": "Martin Lindvall"
+  }, {
+    "@type": "Person",
+    "@id": "https://orcid.org/0000-0001-7250-234X",
+    "givenName": "Caroline",
+    "familyName": "Bivik Stadler",
+    "name": "Caroline Bivik Stadler"
+  }],
+  "publisher": {
+    "@type": "Organization",
+    "name": "AIDA"
+  },
+  "copyrightYear": 2019,
+  "copyrightHolder": [{
+    "@type": "Organization",
+    "name": "Linköping University",
+    "url": "https://liu.se/"
+  }, {
+    "@type": "Person",
+    "@id": "https://orcid.org/0000-0002-9368-0177",
+    "givenName": "Claes",
+    "familyName": "Lundström",
+    "name": "Claes Lundström"
+  }],
+  "provider": [{
+    "@type": "Person",
+    "@id": "https://orcid.org/0000-0002-0128-870X",
+    "name": "Karin Lindman",
+    "email": "Karin.Lindman@regionostergotland.se"
+  }, {
+    "@type": "Person",
+    "@id": "https://orcid.org/0000-0002-9368-0177",
+    "name": "Claes Lundstrom",
+    "email": "claes.lundstrom@liu.se"
+  }, {
+    "@type": "Person",
+    "@id": "https://orcid.org/0000-0001-7250-234X",
+    "givenName": "Caroline",
+    "familyName": "Bivik Stadler",
+    "name": "Caroline Bivik Stadler"
+  }, {
+    "@type": "Person",
+    "@id": "https://orcid.org/0000-0001-6443-3604",
+    "name": "Joel Hedlund",
+    "email": "joel.hedlund@liu.se"
+  }],
+  "dateCreated": "2019-01-09",
+  "datePublished": "2019-01-09",
+  "dateModified": "2019-01-09",
+  "keywords": "pathology, whole slide imaging, annotated",
+  "version": "1.0",
+  "description": "This dataset consists of 174 WSI ovary whole slide images (WSI): 158\nmalignant and 16 benign. Eight of the most common, histological definable\ntumour types were annotated: high grade serous carcinoma (HGSC), low grade\nserous carcinoma (LGSC), clear cell carcinoma (CC), endometrioid\nadenocarcinoma (EN), metastastic serous carcinoma (MS), metastatic other\n(MO), serous borderline tumor (SB) and mucinous borderline tumor (MB). Also\nnormal ovarian tissue were annotated. 11258 separate annotations were made.\nFor the benign structures only the epithelial structures, stroma and support\ntissue were annotated.\n",
+  "license": {
+    "name": "Restricted access",
+    "id": "https://datasets.aida.medtech4health.se/10.23698/aida/drov#license"
+  },
+  "citation": null
+}

data/spec/fixtures/cgimp_package.json

@@ -0,0 +1,18 @@
+{
+  "name": "fullstack_app",
+  "version": "1.0.0",
+  "main": "index.js",
+  "scripts": {
+    "start": "concurrently \"sudo -i service elasticsearch start\" \"cd backend && node --max_old_space_size=8000 server.js\" \"cd client && npm start\""
+  },
+  "keywords": [],
+  "author": "",
+  "license": "ISC",
+  "dependencies": {
+    "concurrently": "^4.1.0",
+    "d3": "^5.9.2",
+    "monk": "^6.0.6"
+  },
+  "devDependencies": {},
+  "description": ""
+}

data/spec/fixtures/cit_package.json

@@ -0,0 +1,19 @@
+{
+  "name": "CITapp",
+  "description": "Concealed Information Test app",
+  "version": "1.1.0",
+  "scripts": {
+    "test": "karma start test/karma.conf.js --single-run"
+  },
+  "devDependencies": {
+    "jasmine-core": "*",
+    "karma": "*",
+    "karma-chrome-launcher": "^2.2.0",
+    "karma-cli": "*",
+    "karma-fixture": "^0.2.6",
+    "karma-html2js-preprocessor": "*",
+    "karma-jasmine": "*",
+    "karma-phantomjs-launcher": "*",
+    "phantomjs-prebuilt": "*"
+  }
+}

data/spec/fixtures/citeproc-no-author.json

@@ -0,0 +1,26 @@
+{
+    "type": "post-weblog",
+    "id": "https://doi.org/10.5438/4k3m-nyvg",
+    "DOI": "10.5438/4k3m-nyvg",
+    "URL": "https://blog.datacite.org/eating-your-own-dog-food",
+    "title": "Eating your own Dog Food",
+    "container-title": "DataCite Blog",
+    "publisher": "DataCite",
+    "abstract": "Eating your own dog food is a slang term to describe that an organization should itself use the products and services it provides. For DataCite this means that we should use DOIs with appropriate metadata and strategies for long-term preservation for...",
+    "categories": [
+        "Phylogeny",
+        "Malaria",
+        "Parasites",
+        "Taxonomy",
+        "Mitochondrial genome",
+        "Africa",
+        "Plasmodium"
+    ],
+    "issued": {
+        "date-parts": [
+            [
+                2016, 12, 20
+            ]
+        ]
+    }
+}

data/spec/fixtures/citeproc-no-categories.json

@@ -0,0 +1,21 @@
+{
+    "type": "post-weblog",
+    "id": "https://doi.org/10.5072/4k3m-nyvg",
+    "DOI": "10.5072/4k3m-nyvg",
+    "URL": "https://blog.datacite.org/eating-your-own-dog-food",
+    "title": "Eating your own Dog Food",
+    "container-title": "DataCite Blog",
+    "publisher": "DataCite",
+    "abstract": "Eating your own dog food is a slang term to describe that an organization should itself use the products and services it provides. For DataCite this means that we should use DOIs with appropriate metadata and strategies for long-term preservation for...",
+    "issued": {
+        "date-parts": [
+            [
+                2016, 12, 20
+            ]
+        ]
+    },
+    "author": [{
+        "family": "Fenner",
+        "given": "Martin"
+    }]
+}

data/spec/fixtures/citeproc.json

@@ -0,0 +1,30 @@
+{
+    "type": "post-weblog",
+    "id": "https://doi.org/10.5438/4k3m-nyvg",
+    "DOI": "10.5438/4k3m-nyvg",
+    "URL": "https://blog.datacite.org/eating-your-own-dog-food",
+    "title": "Eating your own Dog Food",
+    "container-title": "DataCite Blog",
+    "publisher": "DataCite",
+    "abstract": "Eating your own dog food is a slang term to describe that an organization should itself use the products and services it provides. For DataCite this means that we should use DOIs with appropriate metadata and strategies for long-term preservation for...",
+    "categories": [
+        "Phylogeny",
+        "Malaria",
+        "Parasites",
+        "Taxonomy",
+        "Mitochondrial genome",
+        "Africa",
+        "Plasmodium"
+    ],
+    "issued": {
+        "date-parts": [
+            [
+                2016, 12, 20
+            ]
+        ]
+    },
+    "author": [{
+        "family": "Fenner",
+        "given": "Martin"
+    }]
+}