ochre-sdk 1.0.69 → 1.0.71

package/dist/constants.mjs

@@ -30,6 +30,11 @@ const XML_ARRAY_TAGS = [
 	"reference",
 	"coord",
 	"area",
+	"Page",
+	"TextBlock",
+	"TextLine",
+	"ocrItem",
+	"ocrMatch",
 	"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/ocr-matches.d.mts

@@ -0,0 +1,44 @@
+import { OcrMatch } from "../types/index.mjs";
+import { FetchFunction } from "../parsers/helpers.mjs";
+//#region src/fetchers/ocr-matches.d.ts
+/**
+ * Fetches the locations of OCR text matches within OCHRE resources
+ *
+ * Matching mirrors the `ocr` Set item query target, so the same value and match
+ * mode that selected an item will locate its hits. `matchCountsByUuid` reports
+ * the untruncated count, which can exceed the returned matches when
+ * `maxMatchesPerItem` caps them.
+ *
+ * @param parameters - The parameters for the fetch
+ * @param parameters.uuids - The resource UUIDs to search, typically from a filtered Set item fetch
+ * @param parameters.value - The search value
+ * @param parameters.matchMode - Whether to match loosely (stemming and wildcards) or on whole OCR words, defaults to "includes"
+ * @param parameters.isCaseSensitive - Whether matching is case sensitive, defaults to false
+ * @param parameters.maxMatchesPerItem - The cap on returned matches per requested UUID, defaults to 50
+ * @param options - Options for the fetch
+ * @param options.fetch - The fetch function to use
+ * @returns The OCR matches, or null if the fetch/parse fails
+ */
+declare function fetchOcrMatches(parameters: {
+  uuids: Array<string>;
+  value: string;
+  matchMode?: "includes" | "exact";
+  isCaseSensitive?: boolean;
+  maxMatchesPerItem?: number;
+}, options?: {
+  fetch?: FetchFunction;
+}): Promise<{
+  matches: Array<OcrMatch>;
+  matchesByUuid: Record<string, Array<OcrMatch>>;
+  matchCountsByUuid: Record<string, number>;
+  error: null;
+  detailedError: null;
+} | {
+  matches: null;
+  matchesByUuid: null;
+  matchCountsByUuid: null;
+  error: string;
+  detailedError: string;
+}>;
+//#endregion
+export { fetchOcrMatches };

package/dist/fetchers/ocr-matches.mjs

@@ -0,0 +1,134 @@
+import { XML_PARSER_OPTIONS } from "../constants.mjs";
+import { createSchemaValidationError, getErrorOutput, stringLiteral } from "../utilities.mjs";
+import { buildOcrTermQueryExpressions } from "../query.mjs";
+import { ocrMatchesParametersSchema } from "../schemas.mjs";
+import { restoreXMLMetadata } from "../xml/metadata.mjs";
+import { parseOcrMatches } from "../parsers/index.mjs";
+import { XMLOcrMatchesData } from "../xml/schemas.mjs";
+import * as v from "valibot";
+import { XMLParser } from "fast-xml-parser";
+//#region src/fetchers/ocr-matches.ts
+/**
+* Build an XQuery string to fetch OCR match locations from the OCHRE API
+*
+* Each OCR word is matched with `cts:contains` against the same per-term CTS
+* queries the Set item filter compiles, so hit locations always agree with what
+* the filter matched — including stemming and wildcards, which cannot be
+* reproduced outside MarkLogic.
+*
+* @param parameters - The parameters for the fetch
+* @param parameters.uuids - The resource UUIDs to search
+* @param parameters.termQueryExpressions - One CTS query expression per search term, in word order
+* @param parameters.maxMatchesPerItem - The cap on returned matches per requested UUID
+* @returns An XQuery string
+*/
+function buildXQuery(parameters) {
+	const { uuids, termQueryExpressions, maxMatchesPerItem } = parameters;
+	return `xquery version "1.0-ml";
+
+declare variable $uuids := (${Array.from(uuids, (uuid) => stringLiteral(uuid)).join(", ")});
+
+declare variable $termQueries := (
+  ${termQueryExpressions.join(",\n  ")}
+);
+
+declare variable $termCount := ${termQueryExpressions.length};
+
+<ochre>{
+  <ocrMatches>{
+    for $uuid in $uuids
+    let $matches :=
+      for $page in doc($uuid)//ocr/Page
+      let $words := $page//TextLine/string
+      let $wordCount := count($words)
+      let $resourceUuid := string($page/ancestor::resource[1]/@uuid)
+      for $word at $index in $words
+      where $index + $termCount - 1 le $wordCount
+        and (every $offset in (1 to $termCount)
+             satisfies cts:contains($words[$index + $offset - 1], $termQueries[$offset]))
+      return <ocrMatch resourceUuid="{$resourceUuid}">{
+        $page/@n, $page/@fileName, $page/@WIDTH, $page/@HEIGHT,
+        subsequence($words, $index, $termCount)
+      }</ocrMatch>
+    return <ocrItem uuid="{$uuid}" matchCount="{count($matches)}">{
+      subsequence($matches, 1, ${maxMatchesPerItem})
+    }</ocrItem>
+  }</ocrMatches>
+}</ochre>`;
+}
+/**
+* Fetches the locations of OCR text matches within OCHRE resources
+*
+* Matching mirrors the `ocr` Set item query target, so the same value and match
+* mode that selected an item will locate its hits. `matchCountsByUuid` reports
+* the untruncated count, which can exceed the returned matches when
+* `maxMatchesPerItem` caps them.
+*
+* @param parameters - The parameters for the fetch
+* @param parameters.uuids - The resource UUIDs to search, typically from a filtered Set item fetch
+* @param parameters.value - The search value
+* @param parameters.matchMode - Whether to match loosely (stemming and wildcards) or on whole OCR words, defaults to "includes"
+* @param parameters.isCaseSensitive - Whether matching is case sensitive, defaults to false
+* @param parameters.maxMatchesPerItem - The cap on returned matches per requested UUID, defaults to 50
+* @param options - Options for the fetch
+* @param options.fetch - The fetch function to use
+* @returns The OCR matches, or null if the fetch/parse fails
+*/
+async function fetchOcrMatches(parameters, options) {
+	try {
+		const { uuids, value, matchMode, isCaseSensitive, maxMatchesPerItem } = v.parse(ocrMatchesParametersSchema, parameters);
+		const termQueryExpressions = buildOcrTermQueryExpressions({
+			value,
+			matchMode,
+			isCaseSensitive
+		});
+		if (termQueryExpressions.length === 0) return {
+			matches: [],
+			matchesByUuid: {},
+			matchCountsByUuid: {},
+			error: null,
+			detailedError: null
+		};
+		const response = await (options?.fetch ?? fetch)("https://ochre.lib.uchicago.edu/ochre/v2/ochre.php?xquery&xsl=none&lang=\"*\"", {
+			method: "POST",
+			body: buildXQuery({
+				uuids,
+				termQueryExpressions,
+				maxMatchesPerItem
+			}),
+			headers: { "Content-Type": "application/xquery" }
+		});
+		if (!response.ok) throw new Error(`OCHRE API responded with status: ${response.status}`, { cause: response.statusText });
+		const dataRaw = await response.text();
+		const data = new XMLParser(XML_PARSER_OPTIONS).parse(dataRaw);
+		const { success, issues, output } = v.safeParse(XMLOcrMatchesData, data);
+		if (!success) throw createSchemaValidationError("Failed to parse OCHRE OCR matches", issues);
+		restoreXMLMetadata(output, data);
+		const rawOcrItems = output.result.ochre.ocrMatches?.ocrItem ?? [];
+		const matches = parseOcrMatches(rawOcrItems);
+		const matchesByUuid = {};
+		const matchCountsByUuid = {};
+		for (const uuid of uuids) {
+			matchesByUuid[uuid] = [];
+			matchCountsByUuid[uuid] = 0;
+		}
+		for (const rawOcrItem of rawOcrItems) matchCountsByUuid[rawOcrItem.uuid] = rawOcrItem.matchCount;
+		for (const match of matches) matchesByUuid[match.uuid]?.push(match);
+		return {
+			matches,
+			matchesByUuid,
+			matchCountsByUuid,
+			error: null,
+			detailedError: null
+		};
+	} catch (error) {
+		return {
+			matches: null,
+			matchesByUuid: null,
+			matchCountsByUuid: null,
+			...getErrorOutput(error, "Unknown error")
+		};
+	}
+}
+//#endregion
+export { fetchOcrMatches };

package/dist/fetchers/set/items.mjs

@@ -1,10 +1,10 @@
 import { BELONGS_TO_COLLECTION_UUID, DEFAULT_LANGUAGES, XML_PARSER_OPTIONS } from "../../constants.mjs";
 import { createSchemaValidationError, getErrorOutput, stringLiteral } from "../../utilities.mjs";
+import { buildAndCtsQueryExpression, buildBelongsToCollectionQueryExpression, buildQueryPlan } from "../../query.mjs";
 import { iso639_3Schema, setItemsParametersSchema } from "../../schemas.mjs";
 import { restoreXMLMetadata } from "../../xml/metadata.mjs";
 import { parseSetItems } from "../../parsers/index.mjs";
 import { XMLSetItemsData } from "../../xml/schemas.mjs";
-import { buildAndCtsQueryExpression, buildBelongsToCollectionQueryExpression, buildQueryPlan } from "../../query.mjs";
 import * as v from "valibot";
 import { XMLParser } from "fast-xml-parser";
 //#region src/fetchers/set/items.ts
@@ -138,8 +138,8 @@ function buildXQuery(parameters) {
 	const { queries, sort, setScopeUuids, belongsToCollectionScopeUuids, page, pageSize } = parameters;
 	const startPosition = (page - 1) * pageSize + 1;
 	const setScopeDeclaration = `declare variable $setScopeUuids := (${setScopeUuids.map((uuid) => stringLiteral(uuid)).join(", ")});`;
-	const baseItemsExpression = "doc()/ochre/set[@uuid = $setScopeUuids]/items/*";
 	const compiledQueryPlan = buildQueryPlan({ queries });
+	const baseItemsExpression = `doc()/ochre/set[@uuid = $setScopeUuids]/items/*${compiledQueryPlan.itemPredicates}`;
 	const itemsQueryExpressions = [];
 	const belongsToCollectionQueryExpression = buildBelongsToCollectionQueryExpression(belongsToCollectionScopeUuids, BELONGS_TO_COLLECTION_UUID);
 	if (compiledQueryPlan.queryExpression != null) itemsQueryExpressions.push(compiledQueryPlan.queryExpression);
@@ -148,8 +148,10 @@ function buildXQuery(parameters) {
 	const orderedItemsClause = buildOrderedItemsClause(sort);
 	const xqueryDeclarations = ["xquery version \"1.0-ml\";", setScopeDeclaration];
 	if (compiledQueryPlan.prolog !== "") xqueryDeclarations.push(compiledQueryPlan.prolog);
-	const itemsClause = itemsQueryExpression == null ? `let $items := ${baseItemsExpression}` : `let $query := ${itemsQueryExpression}
-  let $items := cts:search(${baseItemsExpression}, $query)`;
+	const letClauses = Array.from(compiledQueryPlan.ocrBindings, (binding) => `let ${binding.name} := ${binding.expression}`);
+	if (itemsQueryExpression == null) letClauses.push(`let $items := ${baseItemsExpression}`);
+	else letClauses.push(`let $query := ${itemsQueryExpression}`, `let $items := cts:search(${baseItemsExpression}, $query)`);
+	const itemsClause = letClauses.join("\n  ");
 	return `${xqueryDeclarations.join("\n\n")}
 
 <ochre>{

package/dist/fetchers/set/property-values.mjs

@@ -1,9 +1,9 @@
 import { BELONGS_TO_COLLECTION_UUID, DEFAULT_LANGUAGES, XML_PARSER_OPTIONS } from "../../constants.mjs";
 import { createSchemaValidationError, getErrorOutput, stringLiteral } from "../../utilities.mjs";
 import { MultilingualString } from "../../parsers/multilingual.mjs";
+import { buildAndCtsQueryExpression, buildBelongsToCollectionQueryExpression, buildQueryPlan } from "../../query.mjs";
 import { setPropertyValuesParametersSchema } from "../../schemas.mjs";
 import { parseXMLContent } from "../../parsers/string.mjs";
-import { buildAndCtsQueryExpression, buildBelongsToCollectionQueryExpression, buildQueryPlan } from "../../query.mjs";
 import * as v from "valibot";
 import { XMLParser } from "fast-xml-parser";
 //#region src/fetchers/set/property-values.ts
@@ -190,8 +190,8 @@ const responseSchema = v.object({ result: v.object({ ochre: v.object({
 function buildXQuery(parameters) {
 	const { setScopeUuids, belongsToCollectionScopeUuids, queries, propertyFacetSelectors, attributes, isLimitedToLeafPropertyValues } = parameters;
 	const setScopeDeclaration = `declare variable $setScopeUuids := (${setScopeUuids.map((uuid) => stringLiteral(uuid)).join(", ")});`;
-	const baseItemsExpression = "doc()/ochre/set[@uuid = $setScopeUuids]/items/*";
 	const compiledQueryPlan = buildQueryPlan({ queries: getItemFilterQueriesFromPropertyValueQueries(queries) });
+	const baseItemsExpression = `doc()/ochre/set[@uuid = $setScopeUuids]/items/*${compiledQueryPlan.itemPredicates}`;
 	const itemsQueryExpressions = [];
 	const belongsToCollectionQueryExpression = buildBelongsToCollectionQueryExpression(belongsToCollectionScopeUuids, BELONGS_TO_COLLECTION_UUID);
 	if (compiledQueryPlan.queryExpression != null) itemsQueryExpressions.push(compiledQueryPlan.queryExpression);
@@ -404,8 +404,10 @@ let $period-values :=
   )`);
 		returnedSequences.push("$period-values");
 	}
-	const itemsClause = itemsQueryExpression == null ? `let $items := ${baseItemsExpression}` : `let $query := ${itemsQueryExpression}
-  let $items := cts:search(${baseItemsExpression}, $query)`;
+	const letClauses = Array.from(compiledQueryPlan.ocrBindings, (binding) => `let ${binding.name} := ${binding.expression}`);
+	if (itemsQueryExpression == null) letClauses.push(`let $items := ${baseItemsExpression}`);
+	else letClauses.push(`let $query := ${itemsQueryExpression}`, `let $items := cts:search(${baseItemsExpression}, $query)`);
+	const itemsClause = letClauses.join("\n  ");
 	return `${xqueryDeclarations.join("\n\n")}
 
 <ochre>{

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/helpers.d.mts

@@ -7,6 +7,10 @@ type FlattenedItem<U, T extends LanguageCodes> = Omit<U, "properties"> & {
  * The default page size to use for fetching paginated items
  */
 declare const DEFAULT_PAGE_SIZE = 48;
+/**
+ * The default cap on OCR matches returned per requested item
+ */
+declare const DEFAULT_MAX_OCR_MATCHES_PER_ITEM = 50;
 /**
  * Flatten the properties of an item
  * @param item - The item whose properties to flatten
@@ -14,4 +18,4 @@ declare const DEFAULT_PAGE_SIZE = 48;
  */
 declare function flattenItemProperties<U extends ItemCategory = ItemCategory, V extends ContainedItemCategory<U> = ContainedItemCategory<U>, T extends LanguageCodes = LanguageCodes, W extends ItemPayloadKind = "topLevel">(item: Item<U, V, T, W>): FlattenedItem<Item<U, V, T, W>, T>;
 //#endregion
-export { DEFAULT_PAGE_SIZE, flattenItemProperties };
+export { DEFAULT_MAX_OCR_MATCHES_PER_ITEM, DEFAULT_PAGE_SIZE, flattenItemProperties };

package/dist/helpers.mjs

@@ -5,6 +5,10 @@ import { flattenProperties } from "./utilities.mjs";
 */
 const DEFAULT_PAGE_SIZE = 48;
 /**
+* The default cap on OCR matches returned per requested item
+*/
+const DEFAULT_MAX_OCR_MATCHES_PER_ITEM = 50;
+/**
 * Flatten the properties of an item
 * @param item - The item whose properties to flatten
 * @returns The item with the properties flattened
@@ -30,4 +34,4 @@ function flattenItemProperties(item) {
 	};
 }
 //#endregion
-export { DEFAULT_PAGE_SIZE, flattenItemProperties };
+export { DEFAULT_MAX_OCR_MATCHES_PER_ITEM, DEFAULT_PAGE_SIZE, flattenItemProperties };

package/dist/index.d.mts

@@ -1,14 +1,15 @@
 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, OcrMatch, 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";
 import { defineLanguages, fetchItem, withLanguages } from "./fetchers/item.mjs";
+import { fetchOcrMatches } from "./fetchers/ocr-matches.mjs";
 import { fetchSetItems } from "./fetchers/set/items.mjs";
 import { fetchSetPropertyValues } from "./fetchers/set/property-values.mjs";
 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 };
+import { DEFAULT_MAX_OCR_MATCHES_PER_ITEM, 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_MAX_OCR_MATCHES_PER_ITEM, 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 OcrMatch, 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, fetchOcrMatches, 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/index.mjs

@@ -1,12 +1,13 @@
 import { 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";
+import { DEFAULT_MAX_OCR_MATCHES_PER_ITEM, DEFAULT_PAGE_SIZE, flattenItemProperties } from "./helpers.mjs";
 import { MultilingualString } from "./parsers/multilingual.mjs";
 import { fetchGallery } from "./fetchers/gallery.mjs";
 import { fetchItemChildren } from "./fetchers/item-children.mjs";
 import { fetchItemLinks } from "./fetchers/item-links.mjs";
 import { defineLanguages, fetchItem, withLanguages } from "./fetchers/item.mjs";
+import { fetchOcrMatches } from "./fetchers/ocr-matches.mjs";
 import { fetchSetItems } from "./fetchers/set/items.mjs";
 import { fetchSetPropertyValues } from "./fetchers/set/property-values.mjs";
 import { fetchWebsiteMetadata } from "./fetchers/website-metadata.mjs";
 import { fetchWebsite } from "./fetchers/website.mjs";
-export { DEFAULT_PAGE_SIZE, MultilingualString, 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 { DEFAULT_MAX_OCR_MATCHES_PER_ITEM, DEFAULT_PAGE_SIZE, MultilingualString, defineLanguages, fetchGallery, fetchItem, fetchItemChildren, fetchItemLinks, fetchOcrMatches, 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.d.mts

@@ -1,12 +1,19 @@
 import { Webpage } from "../types/website.mjs";
-import { Bibliography, ContainedItemCategoryFromOption, ContainedItemCategoryOption, Gallery, Identification, Item, ItemCategory, ItemCategoryFromOption, ItemCategoryOption, ItemContainerCategory, ItemLinks, Metadata, Note, Person, Property, Resource, SetItem, SetItemCategory, SimplifiedProperty } from "../types/index.mjs";
-import { XMLBibliography, XMLData, XMLDataItem, XMLGalleryData, XMLIdentification, XMLItemLinks, XMLLink, XMLMetadata, XMLNote, XMLPerson, XMLProperty, XMLResource, XMLSetItems, XMLSimplifiedProperty } from "../xml/types.mjs";
+import { Bibliography, ContainedItemCategoryFromOption, ContainedItemCategoryOption, Gallery, Identification, Item, ItemCategory, ItemCategoryFromOption, ItemCategoryOption, ItemContainerCategory, ItemLinks, Metadata, Note, OcrMatch, Person, Property, Resource, SetItem, SetItemCategory, SimplifiedProperty } from "../types/index.mjs";
+import { XMLBibliography, XMLData, XMLDataItem, XMLGalleryData, XMLIdentification, XMLItemLinks, XMLLink, XMLMetadata, XMLNote, XMLOcrMatchItem, XMLPerson, XMLProperty, XMLResource, XMLSetItems, XMLSimplifiedProperty } from "../xml/types.mjs";
 import { ParserOptions, getParserOptions, parseStringLike } from "./helpers.mjs";
 //#region src/parsers/index.d.ts
 type RawOchre = XMLData["result"]["ochre"];
 type ResourceViewParser<T extends ReadonlyArray<string>> = (view: XMLResource["view"], context: Pick<Resource<T, "topLevel">, "belongsTo" | "metadata">) => Webpage<T> | null;
 type SetItemCategoryFromCategories<T extends ReadonlyArray<SetItemCategory> | undefined> = T extends ReadonlyArray<infer U> ? Extract<U, SetItemCategory> : SetItemCategory;
 declare function parseIdentification<T extends ReadonlyArray<string>>(rawIdentification: XMLIdentification, options: ParserOptions<T>): Identification<T>;
+/**
+ * Parse OCR matches returned by the OCHRE API
+ * @param rawOcrItems - The raw OCR match items
+ * @returns The parsed OCR matches, in request order
+ * @internal
+ */
+declare function parseOcrMatches(rawOcrItems: Array<XMLOcrMatchItem>): Array<OcrMatch>;
 declare function parseNotes<T extends ReadonlyArray<string>>(rawNotes: {
   note: Array<XMLNote>;
 } | undefined, options: ParserOptions<T>): Array<Note<T>>;
@@ -61,4 +68,4 @@ declare function parseItem(rawData: XMLData, options: {
   parseResourceView?: ResourceViewParser<ReadonlyArray<string>>;
 }): Item<ItemCategory, SetItemCategory, ReadonlyArray<string>>;
 //#endregion
-export { type ParserOptions, RawOchre, getParserOptions, parseBibliographyList, parseGallery, parseIdentification, parseItem, parseLinkedItems, parseLinks, parseMetadata, parseMetadataLanguages, parseNotes, parsePersonList, parseProperties, parseSetItems, parseSimplifiedProperties, parseStringLike, resolveDefaultLanguage, resolveLanguages };
+export { type ParserOptions, RawOchre, getParserOptions, parseBibliographyList, parseGallery, parseIdentification, parseItem, parseLinkedItems, parseLinks, parseMetadata, parseMetadataLanguages, parseNotes, parseOcrMatches, parsePersonList, parseProperties, parseSetItems, parseSimplifiedProperties, parseStringLike, resolveDefaultLanguage, resolveLanguages };

package/dist/parsers/index.mjs

@@ -430,6 +430,79 @@ 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 parseOcrWords(rawWords) {
+	return Array.from(rawWords ?? [], (rawWord) => ({
+		content: rawWord.CONTENT,
+		x: rawWord.HPOS,
+		y: rawWord.VPOS,
+		width: rawWord.WIDTH,
+		height: rawWord.HEIGHT,
+		vertices: parseOcrVertices(rawWord.VERTICES)
+	}));
+}
+function joinOcrWordContents(words) {
+	return Array.from(words, (word) => word.content).join(" ");
+}
+function parseOcrPage(rawPage) {
+	const blocks = Array.from(rawPage.TextBlock ?? [], (rawBlock) => ({ lines: Array.from(rawBlock.TextLine ?? [], (rawLine) => {
+		const words = parseOcrWords(rawLine.string);
+		return {
+			content: joinOcrWordContents(words),
+			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));
+}
+/**
+* Parse OCR matches returned by the OCHRE API
+* @param rawOcrItems - The raw OCR match items
+* @returns The parsed OCR matches, in request order
+* @internal
+*/
+function parseOcrMatches(rawOcrItems) {
+	const matches = [];
+	for (const rawOcrItem of rawOcrItems) {
+		const rawMatches = rawOcrItem.ocrMatch ?? [];
+		for (const rawMatch of rawMatches) {
+			const words = parseOcrWords(rawMatch.string);
+			matches.push({
+				uuid: rawOcrItem.uuid,
+				resourceUuid: rawMatch.resourceUuid ?? null,
+				page: {
+					number: rawMatch.n ?? null,
+					fileName: rawMatch.fileName ?? null,
+					width: rawMatch.WIDTH ?? null,
+					height: rawMatch.HEIGHT ?? null
+				},
+				content: joinOcrWordContents(words),
+				words
+			});
+		}
+	}
+	return matches;
+}
 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 +708,7 @@ 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 +907,7 @@ 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 +1196,7 @@ 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 +1261,8 @@ 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;
@@ -1331,4 +1399,4 @@ function parseItem(rawData, options) {
 	};
 }
 //#endregion
-export { getParserOptions, parseBibliographyList, parseGallery, parseIdentification, parseItem, parseLinkedItems, parseLinks, parseMetadata, parseMetadataLanguages, parseNotes, parsePersonList, parseProperties, parseSetItems, parseSimplifiedProperties, parseStringLike, resolveDefaultLanguage, resolveLanguages };
+export { getParserOptions, parseBibliographyList, parseGallery, parseIdentification, parseItem, parseLinkedItems, parseLinks, parseMetadata, parseMetadataLanguages, parseNotes, parseOcrMatches, parsePersonList, parseProperties, parseSetItems, parseSimplifiedProperties, parseStringLike, resolveDefaultLanguage, resolveLanguages };

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, {