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

CITATION.cff

@@ -14,7 +14,7 @@ authors:
 
 
 repository-code: "https://github.com/caltechlibrary/irdmtools"
-version: 0.0.68
+version: 0.0.70
 license-url: "https://caltechlibrary.github.io/irdmtools/LICENSE"
 keywords: [ "institutional repository", "data
 management", "Invenio", "Invenio-RDM" ]

TODO.html

@@ -89,6 +89,12 @@ <h2 id="bugs">Bugs</h2>
 </ul>
 <h2 id="next">Next</h2>
 <ul class="task-list">
+<li><label><input type="checkbox" />Integrate a YAML options file into
+doi2rdm so that we can easily map our customized mapings via
+configuration instead of hard coding them.</label></li>
+<li><label><input type="checkbox" />irdmtools Go based commands should
+not use the RDM JSON API, they should all be used to directly access the
+Postgres database</label></li>
 <li><label><input type="checkbox" checked="" />rdmutil get_all_ids needs
 a get_all_stale_ids counterpart, see issue #68 (implemented
 get_record_versions”`</label></li>

TODO.md

@@ -36,6 +36,8 @@ Bugs
 Next
 ----
 
+- [ ] Integrate a YAML options file into doi2rdm so that we can easily map our customized mapings via configuration instead of hard coding them.
+- [ ] irdmtools Go based commands should not use the RDM JSON API, they should all be used to directly access the Postgres database
 - [x] rdmutil get_all_ids needs a get_all_stale_ids counterpart, see issue #68 (implemented get_record_versions"`
 - [x] add put_record to rdmutil, actually done as many steps, new_record, new_draft, update_draft, ...
 - [x] Implement a CrossRef to Invenio RDM record

about.html

@@ -24,7 +24,7 @@
 
 <section>
 <h1 id="about-this-software">About this software</h1>
-<h2 id="irdmtools-0.0.68">irdmtools 0.0.68</h2>
+<h2 id="irdmtools-0.0.70">irdmtools 0.0.70</h2>
 <h3 id="authors">Authors</h3>
 <ul>
 <li>R. S. Doiel</li>

about.md

@@ -14,7 +14,7 @@ authors:
     orcid: "https://orcid.org/0000-0001-9266-5146"
 
 repository-code: "https://github.com/caltechlibrary/irdmtools"
-version: 0.0.68
+version: 0.0.70
 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 0.0.68
+## irdmtools 0.0.70
 
 ### Authors
 

cmd/doi2rdm/doi2rdm.go

@@ -54,7 +54,7 @@ var (
 
 # SYNOPSIS
 
-{app_name} [OPTIONS] DOI
+{app_name} [OPTIONS] [OPTIONS_YAML] DOI
 
 # DESCRIPTION
 
@@ -64,6 +64,15 @@ suitable for import into Invenio RDM. The DOI can be in either their
 canonical form or URL form (e.g. "10.1021/acsami.7b15651" or
 "https://doi.org/10.1021/acsami.7b15651").
 
+# OPTIONS_YAML
+
+{app_name} can use an YAML options file to set the behavior of the
+crosswalk from CrossRef to RDM. This replaces many of the options
+previously required in prior implementations of this tool. See all the
+default options setting use the `+"`"+`-show-yaml`+"`"+` command line
+options. You can save this to disk, modify it, then use them for
+migrating content from CrossRef to RDM.
+
 # OPTIONS
 
 -help
@@ -78,38 +87,31 @@ canonical form or URL form (e.g. "10.1021/acsami.7b15651" or
 -diff JSON_FILENAME
 : compare the JSON_FILENAME contents with record generated from CrossRef works record
 
--dot-initials
-: Add period to initials in given name
-
--download
-: attempt to download the digital object if object URL provided
-
--mailto
-: (string) set the mailto value for CrossRef API access (default "[email protected]")
-
--resource-map
-: Use this two column CSV file (no header row) to map resource types in CrossRef to RDM
-
--contributor-map
-: Use this two column CSV file (no header row) to map contributor types from CrossRef (e.g.
-"author", "translator", "editor", "chair") to RDM roles.
+-show-yaml
+: This will display the default YAML configuration file. You can save this and customize to suit your needs.
 
 # EXAMPLES
 
 Example generating a JSON document for a single DOI. The resulting
-text file is called "article.json".
+text file is called "article.json". In this example "options.yaml"
+is the configuration file for setup for your RDM instance.
 
 ~~~
-	{app_name} "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.
 
 ~~~
-	{app_name} -diff article.json "10.1021/acsami.7b15651
+	{app_name} -diff article.json doi2rdm.yaml "10.1021/acsami.7b15651
 ~~~
 
+Save the default YAML options to a file. 
+
+~~~
+	{app_name} -show-yaml >options.yaml
+~~~
 `
 )
 
@@ -123,19 +125,13 @@ func main() {
 	fmtHelp := irdmtools.FmtHelp
 
 	showHelp, showVersion, showLicense := false, false, false
-	debug, downloadDocument := false, false
-	dotInitials := false
-	mailTo, diffFName := "", ""
-	resourceTypeFName, contributorTypeFName := "", ""
+	debug, showYAML := false, false
+	diffFName := ""
 	flag.BoolVar(&showHelp, "help", false, "display help")
 	flag.BoolVar(&showVersion, "version", false, "display version")
 	flag.BoolVar(&showLicense, "license", false, "display license")
+	flag.BoolVar(&showYAML, "show-yaml", false, "display the YAML configuration")
 	flag.StringVar(&diffFName, "diff", diffFName, "compare the JSON file with the current record generated from CrossRef")
-	flag.BoolVar(&dotInitials, "dot-initials", dotInitials, "Add period to initials in given name")
-	flag.BoolVar(&downloadDocument, "download", downloadDocument, "attempt to download the digital object if object URL provided")
-	flag.StringVar(&mailTo, "mailto", mailTo, "set the mail to value for CrossRef API access")
-	flag.StringVar(&resourceTypeFName, "resource-map", resourceTypeFName, "Use this CSV to map resource types from CrossREF to RDM")
-	flag.StringVar(&contributorTypeFName, "contributor-map", contributorTypeFName, "Use this CSV to map contributor types from CrossREF to RDM")
 	flag.BoolVar(&debug, "debug", debug, "display additional info to stderr")
 	flag.Parse()
 	args := flag.Args()
@@ -153,39 +149,30 @@ func main() {
 		fmt.Fprintf(os.Stdout, "%s\n", irdmtools.LicenseText)
 		os.Exit(0)
 	}
+	if showYAML {
+		fmt.Fprintf(os.Stdout, "%s\n", irdmtools.DefaultDoi2RdmOptionsYAML)
+		os.Exit(0)
+	}
 	// Create a appity object
 	app := new(irdmtools.Doi2Rdm)
 	app.Cfg = new(irdmtools.Config)
-	options := map[string]string{}
-	if resourceTypeFName != "" {
-		options["resource-map"] = resourceTypeFName
-	}
-	if contributorTypeFName != "" {
-		options["contributor-map"] = contributorTypeFName
-	}
-	if diffFName != "" {
-		options["diff"] = diffFName
-	}
-	if dotInitials {
-		options["dot_initials"] = "true"
-	}
-	if downloadDocument {
-		options["download_document"] = "true"
-	}
-	if mailTo != "" {
-		options["mailto"] = mailTo
-	}
 	if debug {
 		app.Cfg.Debug = true
 	} else {
 		app.Cfg.Debug = false
 	}
 
-	if len(args) != 1 {
+	if len(args) < 1 {
 		fmt.Fprintln(os.Stderr, "expected a single DOI on the command line")
 		os.Exit(1)
 	}
-	if err := app.Run(os.Stdin, os.Stdout, os.Stderr, options, args[0]); err != nil {
+	optionsFName, doi := "", ""
+	if len(args) > 1 {
+		optionsFName, doi = args[0], args[1]
+	} else {
+		doi = args[0]
+	}
+	if err := app.Run(os.Stdin, os.Stdout, os.Stderr, optionsFName, doi, diffFName); err != nil {
 		fmt.Fprintf(os.Stderr, "%s\n", err)
 		os.Exit(1)
 	}

codemeta.json

@@ -4,10 +4,10 @@
     "license": "https://caltechlibrary.github.io/irdmtools/LICENSE",
     "codeRepository": "https://github.com/caltechlibrary/irdmtools",
     "dateCreated": "2022-10-27",
-	"dateRelease": "2024-01-16",
+	"dateRelease": "2024-01-24",
     "issueTracker": "https://github.com/caltechlibrary/irdmtools/issues",
     "name": "irdmtools",
-    "version": "0.0.68",
+    "version": "0.0.70",
     "description": "Tools for working with institutional repositories and data management systems. Current implementation targets Invenio-RDM.",
     "applicationCategory": "library science",
     "releaseNotes": "Proof of concept and refinements, piloting in production",

crossref.go

@@ -12,9 +12,9 @@ import (
 	"github.com/caltechlibrary/simplified"
 )
 
-func QueryCrossRefWork(cfg *Config, doi string, mailTo string, dotInitials bool, downloadDocument bool) (*crossrefapi.Works, error) {
+func QueryCrossRefWork(cfg *Config, doi string, options *Doi2RdmOptions) (*crossrefapi.Works, error) {
 	appName := path.Base(os.Args[0])
-	client, err := crossrefapi.NewCrossRefClient(appName, mailTo)
+	client, err := crossrefapi.NewCrossRefClient(appName, options.MailTo)
 	if err != nil {
 		return nil, err
 	}
@@ -29,23 +29,6 @@ func QueryCrossRefWork(cfg *Config, doi string, mailTo string, dotInitials bool,
 	return works, nil
 }
 
-// normalizeCrossRefType converts content type from CrossRef
-// to Authors (e.g. "journal-article" to "publication-article")
-func normalizeCrossRefType(s string) string {
-	//FIXME: Ideally this should take a resource type map.
-	switch strings.ToLower(s) {
-	//	case "proceedings-article":
-	//		//FIXME: this mapping may not be correct, was book_section in EPrints CaltechAUTHORS
-	//		return "publication-section"
-	case "journal-article":
-		return "publication-article"
-	case "book-chapter":
-		return "publication-section"
-	default:
-		return s
-	}
-}
-
 // getResourceType retrives the resource type from works.message.type
 // runs normalize
 func getResourceType(work *crossrefapi.Works) string {
@@ -468,17 +451,37 @@ func getApproved(work *crossrefapi.Works) *simplified.DateType {
 	return nil
 }
 
+// normalizePublisherName will check the publisher DOI and ISSN to see if we have
+// a preferred name in our options. If so it will return that.
+func normalizePublisherName(val string, work *crossrefapi.Works, options *Doi2RdmOptions) string {
+	doi := getDOI(work)
+	if doi != "" {
+		doiPrefix, _ := DoiPrefix(val)
+		if value, ok := options.DoiPrefixPublishers[doiPrefix]; ok {
+			return value
+		}
+	}
+	for _, issn := range getISSNs(work) {
+		if issn != "" {
+			if value, ok := options.ISSNPublishers[issn]; ok {
+				return value
+			}
+		}
+	}
+	return val
+}
+
 // CrosswalkCrossRefWork takes a Works object from the CrossRef API
 // and maps the fields into an simplified Record struct return a
 // new struct or error.
-func CrosswalkCrossRefWork(cfg *Config, work *crossrefapi.Works, resourceTypeMap map[string]string, contributorTypeMap map[string]string) (*simplified.Record, error) {
+func CrosswalkCrossRefWork(cfg *Config, work *crossrefapi.Works, options *Doi2RdmOptions) (*simplified.Record, error) {
 	if work == nil {
 		return nil, fmt.Errorf("crossref api works not populated")
 	}
 	rec := new(simplified.Record)
 	// .message.type -> .record.metadata.resource_type (via controlled vocabulary)
 	if value := getResourceType(work); value != "" {
-		if err := SetResourceType(rec, value, resourceTypeMap); err != nil {
+		if err := SetResourceType(rec, value, options.ResourceTypes); err != nil {
 			return nil, err
 		}
 	}
@@ -517,6 +520,8 @@ func CrosswalkCrossRefWork(cfg *Config, work *crossrefapi.Works, resourceTypeMap
 		}
 	}
 	if value := getPublisher(work); value != "" {
+		// FIXME: Setting the publisher name is going to be normalized via DOI prefix, maybe ISSN?
+		value := normalizePublisherName(value, work, options)
 		if err := SetPublisher(rec, value); err != nil {
 			return nil, err
 		}