fix: relatedItem/location has text children

map URLs into location/url
and physical places into location/physicalLocation

Author: phette23

collection-export/strict-mods.js

@@ -76,15 +76,18 @@ const ELEMENT_NAMES = {
     LANGUAGE_OF_CATALOGING: 'languageOfCataloging',
     LANGUAGE_TERM: 'languageTerm',
     LIST: 'list',
+    LOCATION: 'location',
     NAME_PART: 'namePart',
     NOTE: 'note',
     NUMBER: 'number',
+    PHYSICAL_LOCATION: 'physicalLocation',
     PLACE_TERM: 'placeTerm',
     SUBJECT: 'subject',
     TEXT: 'text',
     TITLE_INFO: 'titleInfo',
     TITLE: 'title',
     TOPIC: 'topic',
+    URL: 'url',
 }
 
 // Attributes
@@ -743,6 +746,79 @@ export function reorderCopyInformationChildren(doc) {
     return doc
 }
 
+/**
+ * Wrap location text content with appropriate child elements
+ * - URLs go into <url> elements
+ * - Physical locations go into <physicalLocation> elements
+ * Uses Node's URL constructor to validate URLs
+ *
+ * @param {Document} doc - XML DOM document
+ * @returns {Document} Modified document
+ */
+export function wrapLocationTextContent(doc) {
+    if (!doc) {
+        return doc
+    }
+
+    // Find all location elements that might have direct text content
+    const locationElements = safeSelect('//location', doc)
+
+    for (let location of locationElements) {
+        // Only process if element has direct text content (not already wrapped)
+        let textContent = ''
+        let hasDirectText = false
+        
+        for (let node of location.childNodes) {
+            if (node.nodeType === 3) { // TEXT_NODE
+                const text = node.nodeValue.trim()
+                if (text) {
+                    hasDirectText = true
+                    textContent = text
+                    break
+                }
+            }
+        }
+
+        if (!hasDirectText) {
+            continue
+        }
+
+        // Determine if text is a URL or physical location
+        let isURL = false
+        try {
+            // Use Node's URL constructor to validate URL
+            // This handles various URL formats including http, https, ftp, etc.
+            new URL(textContent)
+            isURL = true
+        } catch {
+            // Not a valid URL, treat as physical location
+            isURL = false
+        }
+
+        // Create appropriate child element
+        const childElementName = isURL ? ELEMENT_NAMES.URL : ELEMENT_NAMES.PHYSICAL_LOCATION
+        const child = doc.createElement(childElementName)
+
+        // Move text content to child element
+        while (location.firstChild) {
+            if (location.firstChild.nodeType === 3) { // TEXT_NODE
+                child.appendChild(location.firstChild)
+            } else {
+                // If there's already an element child, don't process this location
+                // (it means the text is mixed with elements, which shouldn't happen)
+                break
+            }
+        }
+
+        // Only add the child if it has content
+        if (child.textContent.trim()) {
+            location.appendChild(child)
+        }
+    }
+
+    return doc
+}
+
 /**
  * Fix nonstandard mods/name/subNameWrapper elements, several operations:
  * - affiliation -> name/affiliation ("CCAC")
@@ -908,6 +984,9 @@ export function toStrictMODS(xmlString) {
     // Wrap title elements that are direct children of relatedItem with titleInfo
     wrapElement(doc, XPATH_CONTEXTS.RELATEDITEM, ELEMENT_NAMES.TITLE, ELEMENT_NAMES.TITLE_INFO)
 
+    // Wrap location text content with url or physicalLocation
+    wrapLocationTextContent(doc)
+
     // Wrap language text content with languageTerm and move authority attribute
     wrapTextWithChild(doc, XPATH_CONTEXTS.LANGUAGE, ELEMENT_NAMES.LANGUAGE_TERM, [ATTRIBUTES.AUTHORITY])
 

collection-export/strict-mods.test.js

@@ -3,7 +3,7 @@ import { describe, it } from 'mocha'
 import xpath from 'xpath'
 import { DOMParser as xmldom } from '@xmldom/xmldom'
 
-import { removeBadNameUsageAttrs, unwrapSimpleElement, fixTitleAttributes, unwrapDateCreated, unwrapDateOther, renameElement, removeElement, removeEmptyElements, removeAttribute, convertAuthorityElement, moveClassificationToSubject, wrapElement, wrapTextWithChild, moveAndRenameElement, convertNamePartDate, convertSubNameWrapper, wrapCopyInformation, convertSpeakerReleaseDetail, toStrictMODS } from './strict-mods.js'
+import { removeBadNameUsageAttrs, unwrapSimpleElement, fixTitleAttributes, unwrapDateCreated, unwrapDateOther, renameElement, removeElement, removeEmptyElements, removeAttribute, convertAuthorityElement, moveClassificationToSubject, wrapElement, wrapTextWithChild, moveAndRenameElement, convertNamePartDate, convertSubNameWrapper, wrapCopyInformation, wrapLocationTextContent, convertSpeakerReleaseDetail, toStrictMODS } from './strict-mods.js'
 
 // Test fixtures
 const fixtures = {
@@ -2090,6 +2090,134 @@ describe('Strict MODS Conversion', () => {
         })
     })
 
+    describe('wrapLocationTextContent', () => {
+        const locationTextFixtures = {
+            urlInRelatedItem: {
+                input: `<xml><mods>
+                    <relatedItem type="isReferencedBy">
+                        <titleInfo><title>Related Resource</title></titleInfo>
+                        <location>https://vault.cca.edu/items/9d019022-72ce-4774-9e2a-0c315c14f1d1/1/</location>
+                    </relatedItem>
+                </mods></xml>`,
+                expected: `<xml><mods>
+                    <relatedItem type="isReferencedBy">
+                        <titleInfo><title>Related Resource</title></titleInfo>
+                        <location><url>https://vault.cca.edu/items/9d019022-72ce-4774-9e2a-0c315c14f1d1/1/</url></location>
+                    </relatedItem>
+                </mods></xml>`
+            },
+            physicalLocationInRelatedItem: {
+                input: `<xml><mods>
+                    <relatedItem type="otherVersion">
+                        <titleInfo><title>Publication</title></titleInfo>
+                        <location>CCA/C Archives / Archives Publications / Catalogs:Reference Copies / 1971-1974</location>
+                    </relatedItem>
+                </mods></xml>`,
+                expected: `<xml><mods>
+                    <relatedItem type="otherVersion">
+                        <titleInfo><title>Publication</title></titleInfo>
+                        <location><physicalLocation>CCA/C Archives / Archives Publications / Catalogs:Reference Copies / 1971-1974</physicalLocation></location>
+                    </relatedItem>
+                </mods></xml>`
+            },
+            httpUrlInLocation: {
+                input: `<xml><mods>
+                    <location>http://www.example.com/resource</location>
+                </mods></xml>`,
+                expected: `<xml><mods>
+                    <location><url>http://www.example.com/resource</url></location>
+                </mods></xml>`
+            },
+            wikipediaUrl: {
+                input: `<xml><mods>
+                    <location>https://en.wikipedia.org/wiki/Wikipedia:Meetup/Oakland/ArtandFeminism_2015</location>
+                </mods></xml>`,
+                expected: `<xml><mods>
+                    <location><url>https://en.wikipedia.org/wiki/Wikipedia:Meetup/Oakland/ArtandFeminism_2015</url></location>
+                </mods></xml>`
+            },
+            alreadyWrapped: {
+                input: `<xml><mods>
+                    <location>
+                        <physicalLocation>Oakland Campus</physicalLocation>
+                    </location>
+                </mods></xml>`,
+                expected: `<xml><mods>
+                    <location>
+                        <physicalLocation>Oakland Campus</physicalLocation>
+                    </location>
+                </mods></xml>`
+            },
+            emptyLocation: {
+                input: `<xml><mods>
+                    <location></location>
+                </mods></xml>`,
+                expected: `<xml><mods>
+                    <location></location>
+                </mods></xml>`
+            },
+            whitespaceOnly: {
+                input: `<xml><mods>
+                    <location>   </location>
+                </mods></xml>`,
+                expected: `<xml><mods>
+                    <location>   </location>
+                </mods></xml>`
+            }
+        }
+
+        it('should wrap URL text content in <url> element', () => {
+            const result = normalizeXML(toStrictMODS(locationTextFixtures.urlInRelatedItem.input))
+            const expected = normalizeXML(`<mods xmlns="http://www.loc.gov/mods/v3"><relatedItem type="isReferencedBy"><titleInfo><title>Related Resource</title></titleInfo><location><url>https://vault.cca.edu/items/9d019022-72ce-4774-9e2a-0c315c14f1d1/1/</url></location></relatedItem></mods>`)
+
+            assert.strictEqual(result, expected)
+        })
+
+        it('should wrap physical location text content in <physicalLocation> element', () => {
+            const result = normalizeXML(toStrictMODS(locationTextFixtures.physicalLocationInRelatedItem.input))
+            const expected = normalizeXML(`<mods xmlns="http://www.loc.gov/mods/v3"><relatedItem type="otherVersion"><titleInfo><title>Publication</title></titleInfo><location><physicalLocation>CCA/C Archives / Archives Publications / Catalogs:Reference Copies / 1971-1974</physicalLocation></location></relatedItem></mods>`)
+
+            assert.strictEqual(result, expected)
+        })
+
+        it('should wrap HTTP URLs in <url> element', () => {
+            const result = normalizeXML(toStrictMODS(locationTextFixtures.httpUrlInLocation.input))
+            const expected = normalizeXML(`<mods xmlns="http://www.loc.gov/mods/v3"><location><url>http://www.example.com/resource</url></location></mods>`)
+
+            assert.strictEqual(result, expected)
+        })
+
+        it('should wrap Wikipedia URLs in <url> element', () => {
+            const result = normalizeXML(toStrictMODS(locationTextFixtures.wikipediaUrl.input))
+            const expected = normalizeXML(`<mods xmlns="http://www.loc.gov/mods/v3"><location><url>https://en.wikipedia.org/wiki/Wikipedia:Meetup/Oakland/ArtandFeminism_2015</url></location></mods>`)
+
+            assert.strictEqual(result, expected)
+        })
+
+        it('should not modify already wrapped location', () => {
+            const result = normalizeXML(toStrictMODS(locationTextFixtures.alreadyWrapped.input))
+            const expected = normalizeXML(`<mods xmlns="http://www.loc.gov/mods/v3"><location><physicalLocation>Oakland Campus</physicalLocation></location></mods>`)
+
+            assert.strictEqual(result, expected)
+        })
+
+        it('should not modify empty location', () => {
+            const result = normalizeXML(toStrictMODS(locationTextFixtures.emptyLocation.input))
+            // Empty elements are removed by removeEmptyElements()
+            const expected = normalizeXML(`<xml/>`)
+
+            assert.strictEqual(result, expected)
+        })
+
+        it('should not modify location with only whitespace', () => {
+            const result = normalizeXML(toStrictMODS(locationTextFixtures.whitespaceOnly.input))
+            // Empty elements are removed by removeEmptyElements()
+            const expected = normalizeXML(`<xml/>`)
+
+            assert.strictEqual(result, expected)
+        })
+    })
+
     describe('toStrictMODS', () => {
         it('should extract mods element and add namespace by default', () => {
             const input = `<xml><mods>