Merge branch 'main' of github.com:caltechlibrary/irdmtools into gh-pages

.gitignore

@@ -116,3 +116,4 @@ t.json
 t.keys
 t.txt
 man/*
+sandbox/*

CITATION.cff

@@ -14,7 +14,7 @@ authors:
 
 
 repository-code: "https://github.com/caltechlibrary/irdmtools"
-version: 0.0.76
+version: 0.0.78
 license-url: "https://caltechlibrary.github.io/irdmtools/LICENSE"
 keywords: [ "institutional repository", "data
 management", "Invenio", "Invenio-RDM" ]

Makefile

@@ -78,9 +78,9 @@ installer.sh: .FORCE
 test: $(PACKAGE)
 	#go test -timeout 120h
 	go test -test.v -run Test01Config
-	go test -test.v -run Test01Query
-	go test -timeout 2h -ids testdata/test_record_ids.json -run Test02GetRecord
-	go test -timeout 2h -ids testdata/test_record_ids.json -run Test03Harvest
+	#go test -test.v -run Test01Query
+	#go test -timeout 2h -ids testdata/test_record_ids.json -run Test02GetRecord
+	#go test -timeout 2h -ids testdata/test_record_ids.json -run Test03Harvest
 	go test -timeout 2h -run Test01GetRecordIds
 	go test -timeout 2h -run Test01GetModifiedIds
 
@@ -117,6 +117,7 @@ install: build
 	@echo ""
 	@echo "Make sure $(PREFIX)/bin is in your PATH"
 	@echo "Installing man page in $(PREFIX)/man"
+	@mkdir -p $(PREFIX)/man/man1
 	@for FNAME in $(MAN_PAGES); do if [ -f "./man/man1/$${FNAME}" ]; then cp -v "./man/man1/$${FNAME}" "$(PREFIX)/man/man1/$${FNAME}"; fi; done
 	@echo ""
 	@echo "Make sure $(PREFIX)/man is in your MANPATH"

README.md

@@ -41,7 +41,7 @@ This tools take an EPrint record in a dataset collection and returns an abbrevia
 - An Invenio RDM deployment
 - To building the Go based software and documentation
     - git
-    - Go >= 1.20.7
+    - Go >= 1.22.1
     - Make (e.g. GNU Make)
     - Pandoc >= 3
 - For harvesting content

about.html

@@ -24,7 +24,7 @@
 
 <section>
 <h1 id="about-this-software">About this software</h1>
-<h2 id="irdmtools-0.0.76">irdmtools 0.0.76</h2>
+<h2 id="irdmtools-0.0.78">irdmtools 0.0.78</h2>
 <h3 id="authors">Authors</h3>
 <ul>
 <li>R. S. Doiel</li>
@@ -60,7 +60,7 @@ <h3 id="operating-systems">Operating Systems</h3>
 </ul>
 <h3 id="software-requiremets">Software Requiremets</h3>
 <ul>
-<li>Go &gt;= 1.22</li>
+<li>Go &gt;= 1.22.1</li>
 <li>PostgreSQL &gt;= 16</li>
 <li>PostgREST &gt;= 12</li>
 <li>Pandoc &gt;= 3</li>

about.md

@@ -14,7 +14,7 @@ authors:
     orcid: "https://orcid.org/0000-0001-9266-5146"
 
 repository-code: "https://github.com/caltechlibrary/irdmtools"
-version: 0.0.76
+version: 0.0.78
 license-url: "https://caltechlibrary.github.io/irdmtools/LICENSE"
 keywords: [ "institutional repository", "data
 management", "Invenio", "Invenio-RDM" ]
@@ -24,7 +24,7 @@ management", "Invenio", "Invenio-RDM" ]
 About this software
 ===================
 
-## irdmtools 0.0.76
+## irdmtools 0.0.78
 
 ### Authors
 
@@ -59,7 +59,7 @@ systems. Current implementation targets Invenio-RDM.
 
 ### Software Requiremets
 
-- Go >= 1.22
+- Go >= 1.22.1
 - PostgreSQL >= 16
 - PostgREST >= 12
 - Pandoc >= 3

citation.go

@@ -7,8 +7,8 @@ import (
 	"strings"
 
 	// Caltech Library Packages
-	"github.com/caltechlibrary/simplified"
 	"github.com/caltechlibrary/eprinttools"
+	"github.com/caltechlibrary/simplified"
 )
 
 // irdmtools provides a means of turning an EPrint or RDM record into a datastructure suitable
@@ -139,7 +139,6 @@ type Citation struct {
 	// Chapters from book
 	Chapters string `json:"chapters,omitempty" xml:"chapters,omitempty" yaml:"chapters,omitempty"`
 
-
 	// Series/SeriesNumber values from CaltechAUTHORS (mapped from custom fields)
 	Series       string `json:"series,omitempty" xml:"series,omitempty" yaml:"series,omitempty"`
 	SeriesNumber string `json:"series_number,omitempty" xml:"series_number,omitempty" yaml:"series_number,omitempty"`
@@ -151,7 +150,6 @@ type Citation struct {
 	// Pages
 	Pages string `json:"pages,omitempty" xml:"pages,omitempty" yaml:"pages,omitempty"`
 
-
 	// ThesisDegree for thesis types
 	ThesisDegree string `json:"thesis_degree,omitempty" xml:"thesis_degree,omitempty" yaml:"thesis_degree,omitempty"`
 
@@ -255,19 +253,19 @@ func (cite *Citation) CrosswalkRecord(cName string, cID string, citeUsingURL str
 	if rec.Files != nil {
 		// In RDM the "default preview" is the primary document (e.g. article, thesis, etc) of record.
 		defaultPreview := rec.Files.DefaultPreview
-        for _, entry := range rec.Files.Entries {
+		for _, entry := range rec.Files.Entries {
 			// Handle the case where default preview isn't set and treat the first file as the one
 			// serving as a default.
-            if defaultPreview == "" {
-               defaultPreview = entry.Key
-            }
-            if defaultPreview == entry.Key {
-               cite.PrimaryObject = map[string]interface{}{
-                   "basename": defaultPreview,
-                   "url": fmt.Sprintf("%s/records/%s/files/%s", repoURL, rec.ID, defaultPreview),
-               }
-            }
-        }
+			if defaultPreview == "" {
+				defaultPreview = entry.Key
+			}
+			if defaultPreview == entry.Key {
+				cite.PrimaryObject = map[string]interface{}{
+					"basename": defaultPreview,
+					"url":      fmt.Sprintf("%s/records/%s/files/%s", repoURL, rec.ID, defaultPreview),
+				}
+			}
+		}
 	}
 
 	// Now crosswalk the rest of the citation from the simplified record.
@@ -481,7 +479,6 @@ func (ca *CitationAgent) ToString() string {
 	return fmt.Sprintf("%s, %s", ca.FamilyName, ca.LivedName)
 }
 
-
 // CrosswalkEPrint takes an eprinttools.EPrint record and return maps the values into the Citation.
 func (cite *Citation) CrosswalkEPrint(cName string, cID string, citeUsingURL string, eprint *eprinttools.EPrint) error {
 	// map repository required fields, everything else is derived from crosswalk
@@ -504,7 +501,6 @@ func (cite *Citation) CrosswalkEPrint(cName string, cID string, citeUsingURL str
 		}
 	}
 
-
 	// from the eprint table
 	cite.Title = eprint.Title
 	cite.Type = eprint.Type
@@ -526,7 +522,7 @@ func (cite *Citation) CrosswalkEPrint(cName string, cID string, citeUsingURL str
 	cite.ISSN = eprint.ISSN
 	cite.DOI = eprint.DOI
 	cite.PMCID = eprint.PMCID
-	
+
 	if eprint.ThesisType != "" {
 		cite.ThesisType = eprint.ThesisType
 	}
@@ -611,9 +607,9 @@ func (cite *Citation) CrosswalkEPrint(cName string, cID string, citeUsingURL str
 				cite.Contributor = append(cite.Contributor, agent)
 			}
 		}
-	}	
+	}
 
-	// map in Thesis Adivors 
+	// map in Thesis Adivors
 	if eprint.ThesisAdvisor.Length() > 0 {
 		for i := 0; i < eprint.ThesisAdvisor.Length(); i++ {
 			creator := eprint.ThesisAdvisor.IndexOf(i)

citation_test.go

@@ -27,16 +27,16 @@ func TestCitationCrosswalkRecord(t *testing.T) {
 	}
 
 	item := &Citation{}
-	if err := item.CrosswalkRecord("rdm_website", "10.5281-inveniordm.1234", "", rec); err != nil {
+	if err := item.CrosswalkRecord("rdm_website", "10.5281-inveniordm.1234", "", "", rec); err != nil {
 		t.Error(err)
 	}
 	expectedS := "rdm_website"
-	if item.Repository != expectedS {
-		t.Errorf("incorrect repository, expected %q, got %q", expectedS, item.Repository)
+	if item.Collection != expectedS {
+		t.Errorf("incorrect repository, expected %q, got %q", expectedS, item.Collection)
 	}
 	expectedS = "10.5281-inveniordm.1234"
-	if item.RepositoryRecordID != expectedS {
-		t.Errorf("incorrect repo. rec. id, expected %q, got %q", expectedS, item.RepositoryRecordID)
+	if item.CollectionID != expectedS {
+		t.Errorf("incorrect repo. rec. id, expected %q, got %q", expectedS, item.CollectionID)
 	}
 	expectedS = "InvenioRDM"
 	if item.Title != expectedS {
@@ -208,17 +208,17 @@ func TestCrosswalkCreatorToCitationAgent(t *testing.T) {
 	contributorList := []*simplified.Creator{
 		&simplified.Creator{
 			PersonOrOrg: &simplified.PersonOrOrg{
-				Name: "Nielsen, Lars Holm",
+				Name:       "Nielsen, Lars Holm",
 				FamilyName: "Nielsen",
-				GivenName: "Lars Holm",
-				Type: "person",
+				GivenName:  "Lars Holm",
+				Type:       "person",
 				Identifiers: []*simplified.Identifier{
 					&simplified.Identifier{
-						Scheme: "orcid",
+						Scheme:     "orcid",
 						Identifier: "0000-0001-8135-3489",
 					},
 					&simplified.Identifier{
-						Scheme: "clpid",
+						Scheme:     "clpid",
 						Identifier: "Nielsen-Lars-Holm",
 					},
 				},
@@ -228,7 +228,7 @@ func TestCrosswalkCreatorToCitationAgent(t *testing.T) {
 			},
 			Affiliations: []*simplified.Affiliation{
 				&simplified.Affiliation{
-					ID: "01ggx415",
+					ID:   "01ggx415",
 					Name: "CERN",
 				},
 			},
@@ -238,8 +238,8 @@ func TestCrosswalkCreatorToCitationAgent(t *testing.T) {
 		&CitationAgent{
 			FamilyName: "Nielsen",
 			LivedName:  "Lars Holm",
-			ORCID:  "0000-0001-8135-3489",
-			CLpid:  "Nielsen-Lars-Holm",
+			ORCID:      "0000-0001-8135-3489",
+			CLpid:      "Nielsen-Lars-Holm",
 		},
 	}
 	expectedRole := "editor"

cmd/doi2rdm/doi2rdm.go

@@ -54,7 +54,7 @@ var (
 
 # SYNOPSIS
 
-{app_name} [OPTIONS] [OPTIONS_YAML] crossref|datacite DOI
+{app_name} [OPTIONS] [OPTIONS_YAML] [crossref|datacite] DOI
 
 # DESCRIPTION
 
@@ -92,26 +92,35 @@ migrating content from CrossRef to RDM.
 
 # EXAMPLES
 
+Save the default YAML options to a file. You can customize this to match your
+vocabulary requirements in your RDM deployment.
+
+~~~
+	{app_name} -show-yaml >options.yaml
+~~~
+
 Example generating a JSON document for a single DOI. The resulting
 text file is called "article.json". In this example "options.yaml"
-is the configuration file for setup for your RDM instance.
+is the configuration file for setup for your RDM instance. It'll first
+check CrossRef then DataCite.
 
 ~~~
-	{app_name} options.yaml crossref "10.1021/acsami.7b15651" >article.json
+	{app_name} options.yaml "10.1021/acsami.7b15651" >article.json
 ~~~
 
 Check to see the difference from the saved "article.json" and
-the current metadata retrieved from CrossRef.
+the current metadata retrieved from CrossRef or DataCite.
 
 ~~~
-	{app_name} -diff article.json crossref doi2rdm.yaml "10.1021/acsami.7b15651
+	{app_name} -diff article.json options.yaml "10.1021/acsami.7b15651"
 ~~~
 
-Save the default YAML options to a file. 
+Example getting metadata for an arXiv record from DataCite
 
 ~~~
-	{app_name} -show-yaml >options.yaml
+	{app_name} options.yaml "arXiv:2312.07215"
 ~~~
+
 `
 )
 
@@ -166,12 +175,15 @@ func main() {
 		app.Cfg.Debug = false
 	}
 
-	if len(args) < 2 {
-		fmt.Fprintln(eout, "expected a 'crossref' or 'datacite' and single DOI on the command line")
-		os.Exit(1)
-	}
 	optionsFName, dataSource, doi := "", "", ""
-	if len(args) > 2 {
+	if len(args) < 1 {
+		fmt.Fprintln(eout, "expected a least a single DOI on the command line")
+		os.Exit(1)
+	} else if len(args) == 1 {
+		optionsFName, dataSource, doi = "", "", args[0]
+	} else if len(args) == 2 {
+		optionsFName, dataSource, doi = args[0], "", args[1]
+	} else if len(args) > 2 {
 		optionsFName, dataSource, doi = args[0], args[1], args[2]
 	} else {
 		dataSource, doi = args[0], args[1]
@@ -182,14 +194,15 @@ func main() {
 				fmt.Fprintf(eout, "%s\n", err)
 				os.Exit(1)
 			}
-			/*
 		case "datacite":
 			if err := app.RunDataCiteToRdm(in, out, eout, optionsFName, doi, diffFName); err != nil {
 				fmt.Fprintf(eout, "%s\n", err)
 				os.Exit(1)
 			}
-			*/
 		default:
-			fmt.Fprintf(eout, "%q is not supported service to retrive DOI metadata\n", dataSource)
+			if err := app.RunDoiToRdmCombined(in, out, eout, optionsFName, doi, diffFName); err != nil {
+				fmt.Fprintf(eout, "%s\n", err)
+				os.Exit(1)
+			}
 	}
 }

cmd/rdm2eprint/rdm2eprint.go

@@ -95,6 +95,9 @@ specified by C_NAME.
 -pipeline
 : read from standard input and write crosswalk to standard out.
 
+-latest
+: only convert record(s) if latest version.
+
 # EXAMPLE
 
 Example generating a EPRINT JSON document from RDM would use the following
@@ -146,6 +149,7 @@ func main() {
 	releaseDate := irdmtools.ReleaseDate
 	releaseHash := irdmtools.ReleaseHash
 	fmtHelp := irdmtools.FmtHelp
+	latestVersions := false
 
 	showHelp, showVersion, showLicense := false, false, false
 	configFName, debug, asXML := "", false, false
@@ -159,6 +163,7 @@ func main() {
 	flag.StringVar(&idsFName, "ids", idsFName, "read ids from a file")
 	flag.StringVar(&cName, "harvest", cName, "harvest JSON eprint records into the dataset collection.")
 	flag.BoolVar(&pipeline, "pipeline", pipeline, "read from standard input, crosswalk and write to standard out")
+	flag.BoolVar(&latestVersions, "latest", latestVersions, "only convert record if the latest version")
 
 	flag.Parse()
 	rdmids := flag.Args()
@@ -196,20 +201,20 @@ func main() {
 		os.Exit(1)
 	}
 	if cName != "" {
-		if err := app.RunHarvest(os.Stdin, os.Stdout, os.Stderr, cName, rdmids); err != nil {
+		if err := app.RunHarvest(os.Stdin, os.Stdout, os.Stderr, cName, rdmids, latestVersions); err != nil {
 			fmt.Fprintf(os.Stderr, "%s\n", err)
 			os.Exit(1)
 		}
 		os.Exit(0)
 	}
 	if pipeline {
-		if err := app.RunPipeline(os.Stdin, os.Stdout, os.Stderr, asXML); err != nil {
+		if err := app.RunPipeline(os.Stdin, os.Stdout, os.Stderr, asXML, latestVersions); err != nil {
 			fmt.Fprintf(os.Stderr, "%s\n", err)
 			os.Exit(1)
 		}
 		os.Exit(0)
 	}
-	if err := app.Run(os.Stdin, os.Stdout, os.Stderr, rdmids, asXML); err != nil {
+	if err := app.Run(os.Stdin, os.Stdout, os.Stderr, rdmids, asXML, latestVersions); err != nil {
 		fmt.Fprintf(os.Stderr, "%s\n", err)
 		os.Exit(1)
 	}