ochre-sdk 1.0.68 → 1.0.70

package/dist/constants.mjs

@@ -30,6 +30,9 @@ const XML_ARRAY_TAGS = [
 	"reference",
 	"coord",
 	"area",
+	"Page",
+	"TextBlock",
+	"TextLine",
 	"footnote",
 	"language",
 	"section",

package/dist/fetchers/gallery.mjs

@@ -73,8 +73,9 @@ async function fetchGallery(parameters, options) {
 		const { success, issues, output } = v.safeParse(XMLGalleryData, data);
 		if (!success) throw createSchemaValidationError("Failed to parse gallery XML", issues);
 		restoreXMLMetadata(output, data);
+		const languages = resolveGalleryLanguages(output, requestedLanguages);
 		return {
-			gallery: parseGallery(output, { languages: resolveGalleryLanguages(output, requestedLanguages) }),
+			gallery: parseGallery(output, { languages }),
 			error: null,
 			detailedError: null
 		};

package/dist/fetchers/website-metadata.mjs

@@ -16,7 +16,8 @@ function parseWebsiteMetadata(data, options) {
 	const identification = parseIdentification(websiteTree.identification, options);
 	const websiteName = identification.label.getText().trim();
 	const metadataDescription = (parseStringLike(rawOchre.metadata.description) ?? "").trim();
-	const reader = websitePresentationReader(parseSimplifiedProperties(websiteTree.properties, options));
+	const properties = parseSimplifiedProperties(websiteTree.properties, options);
+	const reader = websitePresentationReader(properties);
 	const webpage = websiteTree.items?.resource?.[0] ?? null;
 	const webpageTitle = webpage != null && "identification" in webpage ? parseIdentification(webpage.identification, options).label : null;
 	return {
@@ -154,8 +155,9 @@ async function fetchWebsiteMetadata(abbreviation, options) {
 		const { success, issues, output } = v.safeParse(XMLWebsiteData, data);
 		if (!success) throw createSchemaValidationError("Failed to parse website metadata XML", issues);
 		restoreXMLMetadata(output, data);
+		const metadataLanguages = parseMetadataLanguages(output.result.ochre);
 		return {
-			websiteMetadata: parseWebsiteMetadata(output, { languages: resolveLanguages(requestedLanguages, parseMetadataLanguages(output.result.ochre)) }),
+			websiteMetadata: parseWebsiteMetadata(output, { languages: resolveLanguages(requestedLanguages, metadataLanguages) }),
 			error: null,
 			detailedError: null
 		};

package/dist/getters.mjs

@@ -73,12 +73,10 @@ function clonePropertyValues(values) {
 				content: value.content
 			});
 			break;
-		case "boolean":
-			clonedValues.push({
-				...value,
-				content: value.content
-			});
-			break;
+		case "boolean": clonedValues.push({
+			...value,
+			content: value.content
+		});
 	}
 	return clonedValues;
 }

package/dist/index.d.mts

@@ -1,6 +1,6 @@
 import { MultilingualOptions, MultilingualString, MultilingualStringEntries, MultilingualStringEntry, MultilingualStringInput, MultilingualStringJSON, MultilingualStringObject, MultilingualStringText } from "./parsers/multilingual.mjs";
 import { AccordionWebBlock, ContextTree, ContextTreeFilterLevel, ContextTreeLevel, ContextTreeLevelItem, ProtectedWebsite, Scope, Style, StylesheetCategory, StylesheetItem, WebAccordionItem, WebBlock, WebBlockByLayout, WebBlockItem, WebBlockLayout, WebElement, WebElementComponent, WebElementComponentName, WebElementComponentOf, WebElementOf, WebImage, WebSidebar, WebTitle, Webpage, Website, WebsiteMetadata, WebsitePropertyQuery, WebsitePropertyQueryNode, WebsiteSegment, WebsiteType } from "./types/website.mjs";
-import { AnyBibliography, AnyConcept, AnyItem, AnyPeriod, AnyPerson, AnyPropertyValue, AnyPropertyVariable, AnyResource, AnySet, AnySpatialUnit, AnyText, AnyTree, BaseItem, BaseItemLink, BelongsTo, Bibliography, BibliographyEntryInfo, BibliographyItemLink, BibliographySourceDocument, Concept, ConceptItemLink, ContainedItemCategory, ContainedItemCategoryFromOption, ContainedItemCategoryOption, Context, ContextItem, ContextItemCategory, ContextNode, Coordinates, CoordinatesSource, DictionaryUnitItemLink, EmbeddedBibliography, EmbeddedConcept, EmbeddedItem, EmbeddedPeriod, EmbeddedPerson, EmbeddedPropertyValue, EmbeddedPropertyVariable, EmbeddedResource, EmbeddedSet, EmbeddedSpatialUnit, EmbeddedText, EmbeddedTree, Event, Gallery, Heading, HeadingItemCategory, Identification, Image, ImageMap, ImageMapArea, Interpretation, Item, ItemCategory, ItemCategoryFromOption, ItemCategoryOption, ItemCategoryWithEmbeddedItems, ItemContainerCategory, ItemLink, ItemLinkCategory, ItemLinks, ItemPayloadKind, ItemProperty, ItemWithoutEmbeddedItems, LanguageCodes, License, Metadata, Note, Observation, Period, PeriodItemLink, Person, PersonItemLink, Prettify, Property, PropertyLike, PropertyRelation, PropertyValue, PropertyValueContent, PropertyValueDataType, PropertyValueItemLink, PropertyValueQueryItem, PropertyVariable, PropertyVariableItemLink, Query, QueryGroup, QueryLeaf, QueryablePropertyValueDataType, RecursiveItemCategory, Resource, ResourceItemLink, Section, Set, SetAttributeValueQueryItem, SetBibliography, SetConcept, SetItem, SetItemCategory, SetItemLink, SetItemProperty, SetItemSimplifiedProperty, SetItemsSort, SetItemsSortDirection, SetPeriod, SetResource, SetSpatialUnit, SetTree, SimplifiedProperty, SpatialUnit, SpatialUnitItemLink, Text, TextItemLink, TopLevelItem, Tree, TreeItemCategory, TreeItemLink } from "./types/index.mjs";
+import { AnyBibliography, AnyConcept, AnyItem, AnyPeriod, AnyPerson, AnyPropertyValue, AnyPropertyVariable, AnyResource, AnySet, AnySpatialUnit, AnyText, AnyTree, BaseItem, BaseItemLink, BelongsTo, Bibliography, BibliographyEntryInfo, BibliographyItemLink, BibliographySourceDocument, Concept, ConceptItemLink, ContainedItemCategory, ContainedItemCategoryFromOption, ContainedItemCategoryOption, Context, ContextItem, ContextItemCategory, ContextNode, Coordinates, CoordinatesSource, DictionaryUnitItemLink, EmbeddedBibliography, EmbeddedConcept, EmbeddedItem, EmbeddedPeriod, EmbeddedPerson, EmbeddedPropertyValue, EmbeddedPropertyVariable, EmbeddedResource, EmbeddedSet, EmbeddedSpatialUnit, EmbeddedText, EmbeddedTree, Event, Gallery, Heading, HeadingItemCategory, Identification, Image, ImageMap, ImageMapArea, Interpretation, Item, ItemCategory, ItemCategoryFromOption, ItemCategoryOption, ItemCategoryWithEmbeddedItems, ItemContainerCategory, ItemLink, ItemLinkCategory, ItemLinks, ItemPayloadKind, ItemProperty, ItemWithoutEmbeddedItems, LanguageCodes, License, Metadata, Note, Observation, OcrPage, OcrPoint, OcrTextBlock, OcrTextLine, OcrWord, Period, PeriodItemLink, Person, PersonItemLink, Prettify, Property, PropertyLike, PropertyRelation, PropertyValue, PropertyValueContent, PropertyValueDataType, PropertyValueItemLink, PropertyValueQueryItem, PropertyVariable, PropertyVariableItemLink, Query, QueryGroup, QueryLeaf, QueryablePropertyValueDataType, RecursiveItemCategory, Resource, ResourceItemLink, Section, Set, SetAttributeValueQueryItem, SetBibliography, SetConcept, SetItem, SetItemCategory, SetItemLink, SetItemProperty, SetItemSimplifiedProperty, SetItemsSort, SetItemsSortDirection, SetPeriod, SetResource, SetSpatialUnit, SetTree, SimplifiedProperty, SpatialUnit, SpatialUnitItemLink, Text, TextItemLink, TopLevelItem, Tree, TreeItemCategory, TreeItemLink } from "./types/index.mjs";
 import { fetchGallery } from "./fetchers/gallery.mjs";
 import { fetchItemChildren } from "./fetchers/item-children.mjs";
 import { fetchItemLinks } from "./fetchers/item-links.mjs";
@@ -11,4 +11,4 @@ import { fetchWebsiteMetadata } from "./fetchers/website-metadata.mjs";
 import { fetchWebsite } from "./fetchers/website.mjs";
 import { PropertyOptions, filterProperties, getLeafPropertyValues, getPropertyByVariableLabel, getPropertyByVariableLabelAndValue, getPropertyByVariableLabelAndValueContent, getPropertyByVariableLabelAndValueContents, getPropertyByVariableLabelAndValues, getPropertyByVariableUuid, getPropertyValueByVariableLabel, getPropertyValueByVariableUuid, getPropertyValueContentByVariableLabel, getPropertyValueContentByVariableUuid, getPropertyValueContentsByVariableUuid, getPropertyValuesByVariableLabel, getPropertyValuesByVariableUuid, getUniqueProperties, getUniquePropertyVariableLabels } from "./getters.mjs";
 import { DEFAULT_PAGE_SIZE, flattenItemProperties } from "./helpers.mjs";
-export { type AccordionWebBlock, type AnyBibliography, type AnyConcept, type AnyItem, type AnyPeriod, type AnyPerson, type AnyPropertyValue, type AnyPropertyVariable, type AnyResource, type AnySet, type AnySpatialUnit, type AnyText, type AnyTree, type BaseItem, type BaseItemLink, type BelongsTo, type Bibliography, type BibliographyEntryInfo, type BibliographyItemLink, type BibliographySourceDocument, type Concept, type ConceptItemLink, type ContainedItemCategory, type ContainedItemCategoryFromOption, type ContainedItemCategoryOption, type Context, type ContextItem, type ContextItemCategory, type ContextNode, type ContextTree, type ContextTreeFilterLevel, type ContextTreeLevel, type ContextTreeLevelItem, type Coordinates, type CoordinatesSource, DEFAULT_PAGE_SIZE, type DictionaryUnitItemLink, type EmbeddedBibliography, type EmbeddedConcept, type EmbeddedItem, type EmbeddedPeriod, type EmbeddedPerson, type EmbeddedPropertyValue, type EmbeddedPropertyVariable, type EmbeddedResource, type EmbeddedSet, type EmbeddedSpatialUnit, type EmbeddedText, type EmbeddedTree, type Event, type Gallery, type Heading, type HeadingItemCategory, type Identification, type Image, type ImageMap, type ImageMapArea, type Interpretation, type Item, type ItemCategory, type ItemCategoryFromOption, type ItemCategoryOption, type ItemCategoryWithEmbeddedItems, type ItemContainerCategory, type ItemLink, type ItemLinkCategory, type ItemLinks, type ItemPayloadKind, type ItemProperty, type ItemWithoutEmbeddedItems, type LanguageCodes, type License, type Metadata, type MultilingualOptions, MultilingualString, type MultilingualStringEntries, type MultilingualStringEntry, type MultilingualStringInput, type MultilingualStringJSON, type MultilingualStringObject, type MultilingualStringText, type Note, type Observation, type Period, type PeriodItemLink, type Person, type PersonItemLink, type Prettify, type Property, type PropertyLike, PropertyOptions, type PropertyRelation, type PropertyValue, type PropertyValueContent, type PropertyValueDataType, type PropertyValueItemLink, type PropertyValueQueryItem, type PropertyVariable, type PropertyVariableItemLink, type ProtectedWebsite, type Query, type QueryGroup, type QueryLeaf, type QueryablePropertyValueDataType, type RecursiveItemCategory, type Resource, type ResourceItemLink, type Scope, type Section, type Set, type SetAttributeValueQueryItem, type SetBibliography, type SetConcept, type SetItem, type SetItemCategory, type SetItemLink, type SetItemProperty, type SetItemSimplifiedProperty, type SetItemsSort, type SetItemsSortDirection, type SetPeriod, type SetResource, type SetSpatialUnit, type SetTree, type SimplifiedProperty, type SpatialUnit, type SpatialUnitItemLink, type Style, type StylesheetCategory, type StylesheetItem, type Text, type TextItemLink, type TopLevelItem, type Tree, type TreeItemCategory, type TreeItemLink, type WebAccordionItem, type WebBlock, type WebBlockByLayout, type WebBlockItem, type WebBlockLayout, type WebElement, type WebElementComponent, type WebElementComponentName, type WebElementComponentOf, type WebElementOf, type WebImage, type WebSidebar, type WebTitle, type Webpage, type Website, type WebsiteMetadata, type WebsitePropertyQuery, type WebsitePropertyQueryNode, type WebsiteSegment, type WebsiteType, defineLanguages, fetchGallery, fetchItem, fetchItemChildren, fetchItemLinks, fetchSetItems, fetchSetPropertyValues, fetchWebsite, fetchWebsiteMetadata, filterProperties, flattenItemProperties, getLeafPropertyValues, getPropertyByVariableLabel, getPropertyByVariableLabelAndValue, getPropertyByVariableLabelAndValueContent, getPropertyByVariableLabelAndValueContents, getPropertyByVariableLabelAndValues, getPropertyByVariableUuid, getPropertyValueByVariableLabel, getPropertyValueByVariableUuid, getPropertyValueContentByVariableLabel, getPropertyValueContentByVariableUuid, getPropertyValueContentsByVariableUuid, getPropertyValuesByVariableLabel, getPropertyValuesByVariableUuid, getUniqueProperties, getUniquePropertyVariableLabels, withLanguages };
+export { type AccordionWebBlock, type AnyBibliography, type AnyConcept, type AnyItem, type AnyPeriod, type AnyPerson, type AnyPropertyValue, type AnyPropertyVariable, type AnyResource, type AnySet, type AnySpatialUnit, type AnyText, type AnyTree, type BaseItem, type BaseItemLink, type BelongsTo, type Bibliography, type BibliographyEntryInfo, type BibliographyItemLink, type BibliographySourceDocument, type Concept, type ConceptItemLink, type ContainedItemCategory, type ContainedItemCategoryFromOption, type ContainedItemCategoryOption, type Context, type ContextItem, type ContextItemCategory, type ContextNode, type ContextTree, type ContextTreeFilterLevel, type ContextTreeLevel, type ContextTreeLevelItem, type Coordinates, type CoordinatesSource, DEFAULT_PAGE_SIZE, type DictionaryUnitItemLink, type EmbeddedBibliography, type EmbeddedConcept, type EmbeddedItem, type EmbeddedPeriod, type EmbeddedPerson, type EmbeddedPropertyValue, type EmbeddedPropertyVariable, type EmbeddedResource, type EmbeddedSet, type EmbeddedSpatialUnit, type EmbeddedText, type EmbeddedTree, type Event, type Gallery, type Heading, type HeadingItemCategory, type Identification, type Image, type ImageMap, type ImageMapArea, type Interpretation, type Item, type ItemCategory, type ItemCategoryFromOption, type ItemCategoryOption, type ItemCategoryWithEmbeddedItems, type ItemContainerCategory, type ItemLink, type ItemLinkCategory, type ItemLinks, type ItemPayloadKind, type ItemProperty, type ItemWithoutEmbeddedItems, type LanguageCodes, type License, type Metadata, type MultilingualOptions, MultilingualString, type MultilingualStringEntries, type MultilingualStringEntry, type MultilingualStringInput, type MultilingualStringJSON, type MultilingualStringObject, type MultilingualStringText, type Note, type Observation, type OcrPage, type OcrPoint, type OcrTextBlock, type OcrTextLine, type OcrWord, type Period, type PeriodItemLink, type Person, type PersonItemLink, type Prettify, type Property, type PropertyLike, PropertyOptions, type PropertyRelation, type PropertyValue, type PropertyValueContent, type PropertyValueDataType, type PropertyValueItemLink, type PropertyValueQueryItem, type PropertyVariable, type PropertyVariableItemLink, type ProtectedWebsite, type Query, type QueryGroup, type QueryLeaf, type QueryablePropertyValueDataType, type RecursiveItemCategory, type Resource, type ResourceItemLink, type Scope, type Section, type Set, type SetAttributeValueQueryItem, type SetBibliography, type SetConcept, type SetItem, type SetItemCategory, type SetItemLink, type SetItemProperty, type SetItemSimplifiedProperty, type SetItemsSort, type SetItemsSortDirection, type SetPeriod, type SetResource, type SetSpatialUnit, type SetTree, type SimplifiedProperty, type SpatialUnit, type SpatialUnitItemLink, type Style, type StylesheetCategory, type StylesheetItem, type Text, type TextItemLink, type TopLevelItem, type Tree, type TreeItemCategory, type TreeItemLink, type WebAccordionItem, type WebBlock, type WebBlockByLayout, type WebBlockItem, type WebBlockLayout, type WebElement, type WebElementComponent, type WebElementComponentName, type WebElementComponentOf, type WebElementOf, type WebImage, type WebSidebar, type WebTitle, type Webpage, type Website, type WebsiteMetadata, type WebsitePropertyQuery, type WebsitePropertyQueryNode, type WebsiteSegment, type WebsiteType, defineLanguages, fetchGallery, fetchItem, fetchItemChildren, fetchItemLinks, fetchSetItems, fetchSetPropertyValues, fetchWebsite, fetchWebsiteMetadata, filterProperties, flattenItemProperties, getLeafPropertyValues, getPropertyByVariableLabel, getPropertyByVariableLabelAndValue, getPropertyByVariableLabelAndValueContent, getPropertyByVariableLabelAndValueContents, getPropertyByVariableLabelAndValues, getPropertyByVariableUuid, getPropertyValueByVariableLabel, getPropertyValueByVariableUuid, getPropertyValueContentByVariableLabel, getPropertyValueContentByVariableUuid, getPropertyValueContentsByVariableUuid, getPropertyValuesByVariableLabel, getPropertyValuesByVariableUuid, getUniqueProperties, getUniquePropertyVariableLabels, withLanguages };

package/dist/parsers/index.mjs

@@ -430,6 +430,45 @@ function parseImageMap(rawImageMap) {
 		height: rawImageMap.height
 	};
 }
+function parseOcrVertices(rawVertices) {
+	const vertices = [];
+	if (rawVertices == null) return vertices;
+	for (const match of rawVertices.matchAll(/\(\s*(-?[\d.]+)\s*,\s*(-?[\d.]+)\s*\)/g)) {
+		const x = Number(match[1]);
+		const y = Number(match[2]);
+		vertices.push({
+			x: Number.isNaN(x) ? 0 : x,
+			y: Number.isNaN(y) ? 0 : y
+		});
+	}
+	return vertices;
+}
+function parseOcrPage(rawPage) {
+	const blocks = Array.from(rawPage.TextBlock ?? [], (rawBlock) => ({ lines: Array.from(rawBlock.TextLine ?? [], (rawLine) => {
+		const words = Array.from(rawLine.string ?? [], (rawWord) => ({
+			content: rawWord.CONTENT,
+			x: rawWord.HPOS,
+			y: rawWord.VPOS,
+			width: rawWord.WIDTH,
+			height: rawWord.HEIGHT,
+			vertices: parseOcrVertices(rawWord.VERTICES)
+		}));
+		return {
+			content: Array.from(words, (word) => word.content).join(" "),
+			words
+		};
+	}) }));
+	return {
+		number: rawPage.n ?? null,
+		fileName: rawPage.fileName ?? null,
+		width: rawPage.WIDTH ?? null,
+		height: rawPage.HEIGHT ?? null,
+		blocks
+	};
+}
+function parseOcr(rawOcr) {
+	return Array.from(rawOcr?.Page ?? [], (rawPage) => parseOcrPage(rawPage));
+}
 function parseNote(rawNote, options) {
 	const authors = Array.from(rawNote.authors?.author ?? [], (author) => parsePerson(author, options));
 	const content = rawNote.content == null ? multilingualFromText(parseXMLString(rawNote), options) : parseRequiredContentLike(rawNote, options);
@@ -635,11 +674,7 @@ function parseSetItemHierarchy(hierarchy, options, categories) {
 		case "text":
 			items.push(parseText(entry.item, options));
 			break;
-		case "set":
-			items.push(parseSetSet(entry.item, options));
-			break;
-		case "dictionaryUnit":
-		case "heading": break;
+		case "set": items.push(parseSetSet(entry.item, options));
 	}
 	return items;
 }
@@ -838,10 +873,7 @@ function parseLinks(rawLinks, options) {
 		case "set":
 			links.push(parseSetItemLink(entry.item, options));
 			break;
-		case "dictionaryUnit":
-			links.push(parseDictionaryUnitItemLink(entry.item, options));
-			break;
-		case "heading": break;
+		case "dictionaryUnit": links.push(parseDictionaryUnitItemLink(entry.item, options));
 	}
 	return links;
 }
@@ -1130,6 +1162,7 @@ function parseResource(rawResource, options) {
 		image: parseImage(rawResource.image, options),
 		document: parseContentLike(rawResource.document, options),
 		imageMap: parseImageMap(rawResource.imagemap),
+		ocr: parseOcr(rawResource.ocr),
 		coordinates: parseCoordinates(rawResource.coordinates, options),
 		periods: parsePeriodList(rawResource.periods, options),
 		links: parseLinks(rawResource.links, options),
@@ -1194,7 +1227,8 @@ function resolveDefaultLanguage(rawOchre, languages) {
 	return firstLanguage;
 }
 function parseMetadataPublisher(rawPublisher) {
-	return parseStringLike(Array.isArray(rawPublisher) ? rawPublisher[0] : rawPublisher) ?? "";
+	const publisher = Array.isArray(rawPublisher) ? rawPublisher[0] : rawPublisher;
+	return parseStringLike(publisher) ?? "";
 }
 function parseMetadata(rawOchre, options, defaultLanguage) {
 	const metadataOptions = options;

package/dist/parsers/string.mjs

@@ -129,9 +129,7 @@ function createMDXComponent(variant, properties) {
 		case "documentLink":
 			returnString = `<ExternalLink${createMDXStringAttribute("href", `https://ochre.lib.uchicago.edu/ochre/v2/ochre.php?uuid=${uuid}&load`)}${createMDXStringAttribute("content", tooltipContent)}>${text}</ExternalLink>`;
 			break;
-		case "tooltipSpan":
-			returnString = `<TooltipSpan${createMDXStringAttribute("content", content)}>${text}</TooltipSpan>`;
-			break;
+		case "tooltipSpan": returnString = `<TooltipSpan${createMDXStringAttribute("content", content)}>${text}</TooltipSpan>`;
 	}
 	return returnString;
 }
@@ -407,18 +405,16 @@ function renderRichTextItem(item, linkString, contentItem, options) {
 				result += component;
 				break;
 			}
-			default:
-				result += link.publicationDateTime != null ? createInternalLinkComponent({
-					uuid: getLinkStringProperty(link, "uuid"),
-					text: linkString,
-					content: contentText,
-					annotationMetadata
-				}) : createMDXComponent("tooltipSpan", {
-					uuid: getLinkStringProperty(link, "uuid"),
-					text: linkString,
-					content: contentText
-				});
-				break;
+			default: result += link.publicationDateTime != null ? createInternalLinkComponent({
+				uuid: getLinkStringProperty(link, "uuid"),
+				text: linkString,
+				content: contentText,
+				annotationMetadata
+			}) : createMDXComponent("tooltipSpan", {
+				uuid: getLinkStringProperty(link, "uuid"),
+				text: linkString,
+				content: contentText
+			});
 		}
 		else if (link.publicationDateTime != null) {
 			const component = createInternalLinkComponent({
@@ -472,11 +468,12 @@ function parseXMLContent(item, options) {
 }
 function parseXMLContentItem(contentItem, options) {
 	const rawMDXBlocks = [];
-	let serializedRichText = serializeMDXContent(parseNestedStringItems(contentItem.string, contentItem, {
+	const richText = parseNestedStringItems(contentItem.string, contentItem, {
 		...options,
 		rendering: "rich",
 		rawMDXBlocks
-	}));
+	});
+	let serializedRichText = serializeMDXContent(richText);
 	for (const [index, rawMDXBlock] of rawMDXBlocks.entries()) serializedRichText = serializedRichText.replaceAll(`${RAW_MDX_BLOCK_PLACEHOLDER_PREFIX}${index}${RAW_MDX_BLOCK_PLACEHOLDER_SUFFIX}`, () => rawMDXBlock);
 	return {
 		text: parseNestedStringItems(contentItem.string, contentItem, {

package/dist/parsers/website/index.mjs

@@ -60,7 +60,8 @@ function collectWebsitePageSlugs(resources, options, slugPrefix, pageSlugsByUuid
 			collectWebsitePageSlugs(resource.resource, options, slugPrefix, pageSlugsByUuid, segmentSlugPrefix);
 			continue;
 		}
-		if (websitePresentationReader(resource.properties != null ? parseSimplifiedProperties(resource.properties, options) : []).value("presentation") === "page") {
+		const resourceProperties = resource.properties != null ? parseSimplifiedProperties(resource.properties, options) : [];
+		if (websitePresentationReader(resourceProperties).value("presentation") === "page") {
 			const slug = cleanWebsitePageSlug(resource.slug);
 			if (slug == null) throw new Error(`Slug not found for page (${formatXMLWebsiteResourceMetadata(resource)})`, { cause: resource });
 			const pageSlug = prefixSlug(slug, slugPrefix);
@@ -812,7 +813,8 @@ function parseWebTitle(properties, identification, overrides) {
 function parseWebElement(elementResource, options, context) {
 	const identification = parseIdentification(elementResource.identification, options);
 	const elementProperties = elementResource.properties?.property ? parseSimplifiedProperties(elementResource.properties, options) : [];
-	const properties = parseWebElementProperties(websitePresentationReader(websitePresentationReader(elementProperties).requiredProperty("presentation", `Presentation property not found for element (${formatXMLWebsiteResourceMetadata(elementResource)})`).properties).requiredProperty("component", `Component property not found for element (${formatXMLWebsiteResourceMetadata(elementResource)})`), elementResource, options, context);
+	const presentationProperty = websitePresentationReader(elementProperties).requiredProperty("presentation", `Presentation property not found for element (${formatXMLWebsiteResourceMetadata(elementResource)})`);
+	const properties = parseWebElementProperties(websitePresentationReader(presentationProperty.properties).requiredProperty("component", `Component property not found for element (${formatXMLWebsiteResourceMetadata(elementResource)})`), elementResource, options, context);
 	const cssStyles = parseResponsiveCssStyles(elementProperties);
 	const title = parseWebTitle(elementProperties, identification, {
 		isNameDisplayed: [
@@ -980,7 +982,8 @@ function parseSidebar(resources, options, context) {
 		mobile: []
 	};
 	const sidebarResource = resources.find((resource) => {
-		const resourceReader = websitePresentationReader(resource.properties ? parseSimplifiedProperties(resource.properties, options) : []);
+		const resourceProperties = resource.properties ? parseSimplifiedProperties(resource.properties, options) : [];
+		const resourceReader = websitePresentationReader(resourceProperties);
 		return resourceReader.value("presentation") === "element" && resourceReader.nestedByValue("presentation", "element").value("component") === "sidebar";
 	});
 	if (sidebarResource != null) {
@@ -995,7 +998,8 @@ function parseSidebar(resources, options, context) {
 		cssStyles.mobile = parsedCssStyles.mobile;
 		const sidebarResources = sidebarResource.resource ? normalizeWebsiteResources(sidebarResource.resource) : [];
 		for (const resource of sidebarResources) {
-			const resourceType = websitePresentationReader(resource.properties ? parseSimplifiedProperties(resource.properties, options) : []).value("presentation");
+			const resourceProperties = resource.properties ? parseSimplifiedProperties(resource.properties, options) : [];
+			const resourceType = websitePresentationReader(resourceProperties).value("presentation");
 			if (resourceType === null) continue;
 			switch (resourceType) {
 				case "element": {
@@ -1025,7 +1029,8 @@ function parseWebAccordionItem(elementResource, childResources, options, context
 	const trigger = parseWebElement(elementResource, options, context);
 	const items = [];
 	for (const resource of childResources) {
-		const resourceType = websitePresentationReader(resource.properties ? parseSimplifiedProperties(resource.properties, options) : []).value("presentation");
+		const resourceProperties = resource.properties ? parseSimplifiedProperties(resource.properties, options) : [];
+		const resourceType = websitePresentationReader(resourceProperties).value("presentation");
 		if (resourceType === null) continue;
 		switch (resourceType) {
 			case "element": {
@@ -1131,7 +1136,8 @@ function parseWebBlock(blockResource, options, context) {
 	const isSupportingAccordionItems = returnBlock.properties.default.layout === "accordion" || returnBlock.properties.tablet?.layout === "accordion" || returnBlock.properties.mobile?.layout === "accordion";
 	const blockItems = [];
 	for (const resource of blockResources) {
-		const resourceReader = websitePresentationReader(resource.properties ? parseSimplifiedProperties(resource.properties, options) : []);
+		const resourceProperties = resource.properties ? parseSimplifiedProperties(resource.properties, options) : [];
+		const resourceReader = websitePresentationReader(resourceProperties);
 		const resourceType = resourceReader.value("presentation");
 		if (resourceType === null) continue;
 		switch (resourceType) {
@@ -1165,7 +1171,8 @@ function parseWebBlock(blockResource, options, context) {
 * @returns Parsed WebsiteProperties object
 */
 function parseWebsiteProperties(properties, websiteTree, sidebar, options, parent) {
-	const websiteReader = websitePresentationReader(parseSimplifiedProperties({ property: properties }, options)).nested("presentation");
+	const mainProperties = parseSimplifiedProperties({ property: properties }, options);
+	const websiteReader = websitePresentationReader(mainProperties).nested("presentation");
 	const returnProperties = {
 		type: websiteReader.valueOr("webUI", parent?.type ?? "traditional"),
 		status: websiteReader.valueOr("status", parent?.status ?? "development"),
@@ -1197,10 +1204,7 @@ function parseWebsiteProperties(properties, websiteTree, sidebar, options, paren
 		},
 		sidebar: sidebar ?? parent?.sidebar ?? null,
 		itemPage: {
-			mainContent: {
-				isDisplayed: parent?.itemPage.mainContent.isDisplayed ?? true,
-				isHeaderDisplayed: parent?.itemPage.mainContent.isHeaderDisplayed ?? true
-			},
+			isMainContentDisplayed: parent?.itemPage.isMainContentDisplayed ?? true,
 			description: {
 				isDisplayed: parent?.itemPage.description.isDisplayed ?? true,
 				isHeaderDisplayed: parent?.itemPage.description.isHeaderDisplayed ?? true
@@ -1223,10 +1227,7 @@ function parseWebsiteProperties(properties, websiteTree, sidebar, options, paren
 				isDisplayed: parent?.itemPage.periods.isDisplayed ?? true,
 				isHeaderDisplayed: parent?.itemPage.periods.isHeaderDisplayed ?? true
 			},
-			properties: {
-				isDisplayed: parent?.itemPage.properties.isDisplayed ?? true,
-				isHeaderDisplayed: parent?.itemPage.properties.isHeaderDisplayed ?? true
-			},
+			isPropertiesDisplayed: parent?.itemPage.isPropertiesDisplayed ?? true,
 			bibliography: {
 				isDisplayed: parent?.itemPage.bibliography.isDisplayed ?? true,
 				isHeaderDisplayed: parent?.itemPage.bibliography.isHeaderDisplayed ?? true
@@ -1268,13 +1269,11 @@ function parseWebsiteProperties(properties, websiteTree, sidebar, options, paren
 	const itemPageReader = websiteReader.nestedByValue("page-type", "item-page");
 	if (itemPageReader.size > 0) {
 		for (const [key, slug] of [
-			["mainContent", "main-content"],
 			["description", "description"],
 			["document", "document"],
 			["notes", "notes"],
 			["events", "events"],
 			["periods", "periods"],
-			["properties", "properties"],
 			["bibliography", "bibliography"]
 		]) {
 			const section = returnProperties.itemPage[key];

package/dist/types/index.d.mts

@@ -262,6 +262,47 @@ type ImageMap = {
   width: number;
   height: number;
 };
+/**
+ *  Vertex of an OCR word's bounding polygon in OCHRE
+ */
+type OcrPoint = {
+  x: number;
+  y: number;
+};
+/**
+ *  Word recognized within an OCR text line in OCHRE
+ */
+type OcrWord = {
+  content: string;
+  x: number;
+  y: number;
+  width: number;
+  height: number;
+  vertices: Array<OcrPoint>;
+};
+/**
+ *  Text line of an OCR text block in OCHRE
+ */
+type OcrTextLine = {
+  content: string;
+  words: Array<OcrWord>;
+};
+/**
+ *  Text block of an OCR page in OCHRE
+ */
+type OcrTextBlock = {
+  lines: Array<OcrTextLine>;
+};
+/**
+ *  OCR page of a resource in OCHRE
+ */
+type OcrPage = {
+  number: number | null;
+  fileName: string | null;
+  width: number | null;
+  height: number | null;
+  blocks: Array<OcrTextBlock>;
+};
 /**
  *  Note in OCHRE
  */
@@ -680,6 +721,7 @@ type Resource<T extends LanguageCodes = LanguageCodes, U extends ItemPayloadKind
   image: Image<T> | null;
   document: MultilingualString<T> | null;
   imageMap: ImageMap | null;
+  ocr: Array<OcrPage>;
   coordinates: Array<Coordinates<T>>;
   periods: Array<Period<T, "embedded">>;
   links: ItemLinks<T>;
@@ -876,4 +918,4 @@ type QueryGroup = {
  */
 type Query = QueryLeaf | QueryGroup;
 //#endregion
-export { AnyBibliography, AnyConcept, AnyItem, AnyPeriod, AnyPerson, AnyPropertyValue, AnyPropertyVariable, AnyResource, AnySet, AnySpatialUnit, AnyText, AnyTree, BaseItem, BaseItemLink, BelongsTo, Bibliography, BibliographyEntryInfo, BibliographyItemLink, BibliographySourceDocument, Concept, ConceptItemLink, ContainedItemCategory, ContainedItemCategoryFromOption, ContainedItemCategoryOption, Context, ContextItem, ContextItemCategory, ContextNode, Coordinates, CoordinatesSource, DictionaryUnitItemLink, EmbeddedBibliography, EmbeddedConcept, EmbeddedItem, EmbeddedPeriod, EmbeddedPerson, EmbeddedPropertyValue, EmbeddedPropertyVariable, EmbeddedResource, EmbeddedSet, EmbeddedSpatialUnit, EmbeddedText, EmbeddedTree, Event, Gallery, Heading, HeadingItemCategory, Identification, Image, ImageMap, ImageMapArea, Interpretation, Item, ItemCategory, ItemCategoryFromOption, ItemCategoryOption, ItemCategoryWithEmbeddedItems, ItemContainerCategory, ItemLink, ItemLinkCategory, ItemLinks, ItemPayloadKind, ItemProperty, ItemWithoutEmbeddedItems, LanguageCodes, License, Metadata, Note, Observation, Period, PeriodItemLink, Person, PersonItemLink, Prettify, Property, PropertyLike, PropertyRelation, PropertyValue, PropertyValueContent, PropertyValueDataType, PropertyValueItemLink, PropertyValueQueryItem, PropertyVariable, PropertyVariableItemLink, Query, QueryGroup, QueryLeaf, QueryablePropertyValueDataType, RecursiveItemCategory, Resource, ResourceItemLink, Section, Set, SetAttributeValueQueryItem, SetBibliography, SetConcept, SetItem, SetItemCategory, SetItemLink, SetItemProperty, SetItemSimplifiedProperty, SetItemsSort, SetItemsSortDirection, SetPeriod, SetResource, SetSpatialUnit, SetTree, SimplifiedProperty, SpatialUnit, SpatialUnitItemLink, Text, TextItemLink, TopLevelItem, Tree, TreeItemCategory, TreeItemLink };
+export { AnyBibliography, AnyConcept, AnyItem, AnyPeriod, AnyPerson, AnyPropertyValue, AnyPropertyVariable, AnyResource, AnySet, AnySpatialUnit, AnyText, AnyTree, BaseItem, BaseItemLink, BelongsTo, Bibliography, BibliographyEntryInfo, BibliographyItemLink, BibliographySourceDocument, Concept, ConceptItemLink, ContainedItemCategory, ContainedItemCategoryFromOption, ContainedItemCategoryOption, Context, ContextItem, ContextItemCategory, ContextNode, Coordinates, CoordinatesSource, DictionaryUnitItemLink, EmbeddedBibliography, EmbeddedConcept, EmbeddedItem, EmbeddedPeriod, EmbeddedPerson, EmbeddedPropertyValue, EmbeddedPropertyVariable, EmbeddedResource, EmbeddedSet, EmbeddedSpatialUnit, EmbeddedText, EmbeddedTree, Event, Gallery, Heading, HeadingItemCategory, Identification, Image, ImageMap, ImageMapArea, Interpretation, Item, ItemCategory, ItemCategoryFromOption, ItemCategoryOption, ItemCategoryWithEmbeddedItems, ItemContainerCategory, ItemLink, ItemLinkCategory, ItemLinks, ItemPayloadKind, ItemProperty, ItemWithoutEmbeddedItems, LanguageCodes, License, Metadata, Note, Observation, OcrPage, OcrPoint, OcrTextBlock, OcrTextLine, OcrWord, Period, PeriodItemLink, Person, PersonItemLink, Prettify, Property, PropertyLike, PropertyRelation, PropertyValue, PropertyValueContent, PropertyValueDataType, PropertyValueItemLink, PropertyValueQueryItem, PropertyVariable, PropertyVariableItemLink, Query, QueryGroup, QueryLeaf, QueryablePropertyValueDataType, RecursiveItemCategory, Resource, ResourceItemLink, Section, Set, SetAttributeValueQueryItem, SetBibliography, SetConcept, SetItem, SetItemCategory, SetItemLink, SetItemProperty, SetItemSimplifiedProperty, SetItemsSort, SetItemsSortDirection, SetPeriod, SetResource, SetSpatialUnit, SetTree, SimplifiedProperty, SpatialUnit, SpatialUnitItemLink, Text, TextItemLink, TopLevelItem, Tree, TreeItemCategory, TreeItemLink };

package/dist/types/website.d.mts

@@ -155,10 +155,7 @@ type Website<T extends LanguageCodes = LanguageCodes> = {
     };
     sidebar: WebSidebar<T> | null;
     itemPage: {
-      mainContent: {
-        isDisplayed: boolean;
-        isHeaderDisplayed: boolean;
-      };
+      isMainContentDisplayed: boolean;
       description: {
         isDisplayed: boolean;
         isHeaderDisplayed: boolean;
@@ -181,10 +178,7 @@ type Website<T extends LanguageCodes = LanguageCodes> = {
         isDisplayed: boolean;
         isHeaderDisplayed: boolean;
       };
-      properties: {
-        isDisplayed: boolean;
-        isHeaderDisplayed: boolean;
-      };
+      isPropertiesDisplayed: boolean;
       bibliography: {
         isDisplayed: boolean;
         isHeaderDisplayed: boolean;

package/dist/xml/schemas.mjs

@@ -220,6 +220,24 @@ const XMLImageMap = v.object({
 	width: XMLNumber,
 	height: XMLNumber
 }, "XMLImageMap: Shape error");
+const XMLOcrString = v.object({
+	HPOS: XMLNumber,
+	VPOS: XMLNumber,
+	WIDTH: XMLNumber,
+	HEIGHT: XMLNumber,
+	CONTENT: v.string("XMLOcrString: CONTENT is string and required"),
+	VERTICES: v.optional(v.string("XMLOcrString: VERTICES is string and optional"))
+}, "XMLOcrString: Shape error");
+const XMLOcrTextLine = v.object({ string: v.optional(v.array(XMLOcrString, "XMLOcrTextLine: string is array of XMLOcrString")) }, "XMLOcrTextLine: Shape error");
+const XMLOcrTextBlock = v.object({ TextLine: v.optional(v.array(XMLOcrTextLine, "XMLOcrTextBlock: TextLine is array of XMLOcrTextLine")) }, "XMLOcrTextBlock: Shape error");
+const XMLOcrPage = v.object({
+	n: XMLOptionalNumber,
+	fileName: v.optional(v.string("XMLOcrPage: fileName is string and optional")),
+	WIDTH: XMLOptionalNumber,
+	HEIGHT: XMLOptionalNumber,
+	TextBlock: v.optional(v.array(XMLOcrTextBlock, "XMLOcrPage: TextBlock is array of XMLOcrTextBlock"))
+}, "XMLOcrPage: Shape error");
+const XMLOcr = v.object({ Page: v.optional(v.array(XMLOcrPage, "XMLOcr: Page is array of XMLOcrPage")) }, "XMLOcr: Shape error");
 const XMLNote = v.object({
 	content: v.optional(XMLContent.entries.content),
 	payload: v.optional(v.string("XMLNote: payload is string and optional")),
@@ -640,6 +658,7 @@ const XMLResource = v.object({
 	width: XMLOptionalNumber,
 	image: v.optional(XMLImage),
 	imagemap: v.optional(XMLImageMap),
+	ocr: v.optional(XMLOcr),
 	document: v.optional(XMLContent),
 	coordinates: v.optional(XMLCoordinates),
 	periods: v.optional(v.object({ period: v.array(XMLPeriod) })),
@@ -791,6 +810,7 @@ const XMLWebsiteResource = v.lazy(() => v.object({
 	width: XMLOptionalNumber,
 	image: v.optional(XMLImage),
 	imagemap: v.optional(XMLImageMap),
+	ocr: v.optional(XMLOcr),
 	document: v.optional(XMLContent),
 	coordinates: v.optional(XMLCoordinates),
 	periods: v.optional(v.object({ period: v.array(XMLPeriod) })),

package/dist/xml/types.d.mts

@@ -175,6 +175,30 @@ type XMLImageMap = {
   width: XMLNumber;
   height: XMLNumber;
 };
+type XMLOcrString = {
+  HPOS: XMLNumber;
+  VPOS: XMLNumber;
+  WIDTH: XMLNumber;
+  HEIGHT: XMLNumber;
+  CONTENT: string;
+  VERTICES?: string;
+};
+type XMLOcrTextLine = {
+  string?: Array<XMLOcrString>;
+};
+type XMLOcrTextBlock = {
+  TextLine?: Array<XMLOcrTextLine>;
+};
+type XMLOcrPage = {
+  n?: XMLNumber;
+  fileName?: string;
+  WIDTH?: XMLNumber;
+  HEIGHT?: XMLNumber;
+  TextBlock?: Array<XMLOcrTextBlock>;
+};
+type XMLOcr = {
+  Page?: Array<XMLOcrPage>;
+};
 type XMLNote = Partial<XMLContent> & XMLString & {
   noteNo?: XMLNumber;
   title?: string;
@@ -556,6 +580,7 @@ type XMLResource = XMLBaseItem & {
   width?: XMLNumber;
   image?: XMLImage;
   imagemap?: XMLImageMap;
+  ocr?: XMLOcr;
   document?: XMLContent;
   coordinates?: XMLCoordinates;
   periods?: {
@@ -913,4 +938,4 @@ type XMLWebsiteData = {
   };
 };
 //#endregion
-export { XMLBaseItem, XMLBibliography, XMLBoolean, XMLConcept, XMLContent, XMLContext, XMLContextGroup, XMLContextItem, XMLContextValue, XMLCoordinate, XMLCoordinates, XMLCoordinatesSource, XMLData, XMLDataItem, XMLDictionaryUnit, XMLEmptyContext, XMLEvent, XMLGallery, XMLGalleryData, XMLHeading, XMLHeadingItemCategory, XMLIdentification, XMLImage, XMLImageMap, XMLImageMapArea, XMLInterpretation, XMLItemCategory, XMLItemLinks, XMLItemLinksData, XMLLicense, XMLLink, XMLLinkedBaseItem, XMLLinkedBibliography, XMLLinkedConcept, XMLLinkedPeriod, XMLLinkedPerson, XMLLinkedPropertyValue, XMLLinkedPropertyVariable, XMLLinkedResource, XMLLinkedSet, XMLLinkedSpatialUnit, XMLLinkedText, XMLLinkedTree, XMLMetadata, XMLNote, XMLNumber, XMLObservation, XMLPeriod, XMLPerson, XMLProperty, XMLPropertyRelation, XMLPropertyValue, XMLPropertyVariable, XMLRecursiveItemCategory, XMLResource, XMLSection, XMLSet, XMLSetItems, XMLSetItemsData, XMLSimplifiedProperty, XMLSpatialUnit, XMLString, XMLText, XMLTree, XMLWebsiteContext, XMLWebsiteContextItem, XMLWebsiteContextLevel, XMLWebsiteData, XMLWebsiteFilterContext, XMLWebsiteFilterContextItem, XMLWebsiteOptions, XMLWebsiteProperties, XMLWebsiteResource, XMLWebsiteResourceGroup, XMLWebsiteResourceItem, XMLWebsiteScope, XMLWebsiteSegment, XMLWebsiteStyle, XMLWebsiteTree };
+export { XMLBaseItem, XMLBibliography, XMLBoolean, XMLConcept, XMLContent, XMLContext, XMLContextGroup, XMLContextItem, XMLContextValue, XMLCoordinate, XMLCoordinates, XMLCoordinatesSource, XMLData, XMLDataItem, XMLDictionaryUnit, XMLEmptyContext, XMLEvent, XMLGallery, XMLGalleryData, XMLHeading, XMLHeadingItemCategory, XMLIdentification, XMLImage, XMLImageMap, XMLImageMapArea, XMLInterpretation, XMLItemCategory, XMLItemLinks, XMLItemLinksData, XMLLicense, XMLLink, XMLLinkedBaseItem, XMLLinkedBibliography, XMLLinkedConcept, XMLLinkedPeriod, XMLLinkedPerson, XMLLinkedPropertyValue, XMLLinkedPropertyVariable, XMLLinkedResource, XMLLinkedSet, XMLLinkedSpatialUnit, XMLLinkedText, XMLLinkedTree, XMLMetadata, XMLNote, XMLNumber, XMLObservation, XMLOcr, XMLOcrPage, XMLOcrString, XMLOcrTextBlock, XMLOcrTextLine, XMLPeriod, XMLPerson, XMLProperty, XMLPropertyRelation, XMLPropertyValue, XMLPropertyVariable, XMLRecursiveItemCategory, XMLResource, XMLSection, XMLSet, XMLSetItems, XMLSetItemsData, XMLSimplifiedProperty, XMLSpatialUnit, XMLString, XMLText, XMLTree, XMLWebsiteContext, XMLWebsiteContextItem, XMLWebsiteContextLevel, XMLWebsiteData, XMLWebsiteFilterContext, XMLWebsiteFilterContextItem, XMLWebsiteOptions, XMLWebsiteProperties, XMLWebsiteResource, XMLWebsiteResourceGroup, XMLWebsiteResourceItem, XMLWebsiteScope, XMLWebsiteSegment, XMLWebsiteStyle, XMLWebsiteTree };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ochre-sdk",
-  "version": "1.0.68",
+  "version": "1.0.70",
   "type": "module",
   "license": "MIT",
   "description": "Node.js library for working with OCHRE (Online Cultural and Historical Research Environment) data",
@@ -53,10 +53,10 @@
   "devDependencies": {
     "@antfu/eslint-config": "^9.2.0",
     "@types/node": "^24.13.3",
-    "bumpp": "^12.0.0",
+    "bumpp": "^12.1.1",
     "eslint": "^10.8.0",
-    "knip": "^6.29.0",
-    "oxfmt": "^0.60.0",
+    "knip": "^6.31.0",
+    "oxfmt": "^0.61.0",
     "tsdown": "^0.22.14",
     "typescript": "^6.0.3",
     "vitest": "^4.1.10"