ochre-sdk 1.0.72 → 1.0.74

package/dist/fetchers/website-metadata.mjs

@@ -1,5 +1,5 @@
 import { XML_PARSER_OPTIONS } from "../constants.mjs";
-import { createSchemaValidationError, getErrorOutput, stringLiteral } from "../utilities.mjs";
+import { SUPPLEMENTAL_XQUERY_PROLOG, createSchemaValidationError, getErrorOutput, omitSupplemental, stringLiteral } from "../utilities.mjs";
 import { iso639_3Schema } from "../schemas.mjs";
 import { restoreXMLMetadata } from "../xml/metadata.mjs";
 import { parseStringLike } from "../parsers/helpers.mjs";
@@ -41,6 +41,8 @@ function parseWebsiteMetadata(data, options) {
 function buildXQuery(parameters) {
 	return String.raw`xquery version "1.0-ml";
 
+${SUPPLEMENTAL_XQUERY_PROLOG}
+
 declare function local:resource-items($resources) {
   for $resource in $resources
   return
@@ -131,8 +133,10 @@ return
     $website/@belongsTo,
     $website/@publicationDateTime,
     $website/@languages,
-    $website/metadata,
-    local:metadata-tree($website/tree[1], $target-slug, "")
+    ${omitSupplemental(`(
+      $website/metadata,
+      local:metadata-tree($website/tree[1], $target-slug, "")
+    )`)}
   }</ochre>`;
 }
 async function fetchWebsiteMetadata(abbreviation, options) {

package/dist/fetchers/website.mjs

@@ -1,12 +1,12 @@
 import { XML_PARSER_OPTIONS } from "../constants.mjs";
-import { createSchemaValidationError, getErrorOutput } from "../utilities.mjs";
+import { SUPPLEMENTAL_XQUERY_PROLOG, createSchemaValidationError, getErrorOutput, omitSupplemental, stringLiteral } from "../utilities.mjs";
 import { restoreXMLMetadata } from "../xml/metadata.mjs";
 import { XMLWebsiteData } from "../xml/schemas.mjs";
 import { parseWebsite } from "../parsers/website/index.mjs";
 import * as v from "valibot";
 import { XMLParser } from "fast-xml-parser";
 //#region src/fetchers/website.ts
-async function validateWebsiteCredentials(uuid, credentials, fetcher) {
+async function areWebsiteCredentialsValid(uuid, credentials, fetcher) {
 	const security = typeof credentials === "string" ? { validate: credentials } : {
 		validate: credentials.password,
 		userOCHRE: credentials.username
@@ -21,6 +21,20 @@ async function validateWebsiteCredentials(uuid, credentials, fetcher) {
 	})).ok;
 }
 /**
+* Build an XQuery string to fetch a website tree document by abbreviation.
+*
+* @param abbreviation - The lowercased website abbreviation to match
+* @returns An XQuery string
+*/
+function buildXQuery(abbreviation) {
+	return `xquery version "1.0-ml";
+
+${SUPPLEMENTAL_XQUERY_PROLOG}
+
+for $ochre in collection("ochre/tree")/ochre[tree/identification/abbreviation/content/string = ${stringLiteral(abbreviation)}]
+return element ochre { $ochre/@*, ${omitSupplemental("$ochre/node()")} }`;
+}
+/**
 * Fetches and parses a website configuration from the OCHRE API.
 *
 * For password-protected or OCHRE-credential-protected websites, if no credentials
@@ -32,8 +46,11 @@ async function validateWebsiteCredentials(uuid, credentials, fetcher) {
 async function fetchWebsite(abbreviation, options) {
 	try {
 		const fetcher = options?.fetch ?? fetch;
-		const cleanAbbreviation = abbreviation.trim().toLocaleLowerCase("en-US");
-		const response = await fetcher(`https://ochre.lib.uchicago.edu/ochre/v2/ochre.php?xquery=${encodeURIComponent(`collection('ochre/tree')/ochre[tree/identification/abbreviation/content/string='${cleanAbbreviation}']`)}&xsl=none&lang="*"`);
+		const response = await fetcher("https://ochre.lib.uchicago.edu/ochre/v2/ochre.php?xquery&xsl=none&lang=\"*\"", {
+			method: "POST",
+			body: buildXQuery(abbreviation.trim().toLocaleLowerCase("en-US")),
+			headers: { "Content-Type": "application/xquery" }
+		});
 		if (!response.ok) throw new Error("Failed to fetch website", { cause: response.statusText });
 		const dataRaw = await response.text();
 		const data = new XMLParser(XML_PARSER_OPTIONS).parse(dataRaw);
@@ -52,7 +69,7 @@ async function fetchWebsite(abbreviation, options) {
 				error: null,
 				detailedError: null
 			};
-			if (!await validateWebsiteCredentials(website.uuid, options.credentials, fetcher)) throw new Error("Invalid credentials for protected website");
+			if (!await areWebsiteCredentialsValid(website.uuid, options.credentials, fetcher)) throw new Error("Invalid credentials for protected website");
 		}
 		return {
 			website,

package/dist/getters.d.mts

@@ -4,9 +4,13 @@ import { LanguageCodes, Property, PropertyLike, PropertyValueContent, SetItemPro
  * Options for property search operations.
  */
 type PropertyOptions = {
-  /** Whether to recursively search through nested properties. */
+  /**
+    Whether to recursively search through nested properties.
+    */
   includeNestedProperties?: boolean;
-  /** Whether to limit property values to leaf values. */
+  /**
+    Whether to limit property values to leaf values.
+    */
   limitToLeafPropertyValues?: boolean;
 };
 type PropertyContent<T extends LanguageCodes> = PropertyValueContent<T>["content"];

package/dist/helpers.d.mts

@@ -7,10 +7,6 @@ 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
@@ -18,4 +14,4 @@ declare const DEFAULT_MAX_OCR_MATCHES_PER_ITEM = 50;
  */
 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_MAX_OCR_MATCHES_PER_ITEM, DEFAULT_PAGE_SIZE, flattenItemProperties };
+export { DEFAULT_PAGE_SIZE, flattenItemProperties };

package/dist/helpers.mjs

@@ -5,10 +5,6 @@ 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
@@ -34,4 +30,4 @@ function flattenItemProperties(item) {
 	};
 }
 //#endregion
-export { DEFAULT_MAX_OCR_MATCHES_PER_ITEM, DEFAULT_PAGE_SIZE, flattenItemProperties };
+export { DEFAULT_PAGE_SIZE, flattenItemProperties };

package/dist/index.d.mts

@@ -1,15 +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, 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 { 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, OcrString, 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 { fetchItemOcrData } from "./fetchers/item-ocr-data.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_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 };
+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 OcrString, 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, fetchItemOcrData, 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

@@ -1,13 +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_MAX_OCR_MATCHES_PER_ITEM, DEFAULT_PAGE_SIZE, flattenItemProperties } from "./helpers.mjs";
+import { 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 { fetchItemOcrData } from "./fetchers/item-ocr-data.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_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 };
+export { DEFAULT_PAGE_SIZE, MultilingualString, defineLanguages, fetchGallery, fetchItem, fetchItemChildren, fetchItemLinks, fetchItemOcrData, 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

@@ -1,19 +1,12 @@
 import { Webpage } from "../types/website.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 { 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 { 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>>;
@@ -68,4 +61,4 @@ 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, parseOcrMatches, parsePersonList, parseProperties, parseSetItems, parseSimplifiedProperties, parseStringLike, resolveDefaultLanguage, resolveLanguages };
+export { type ParserOptions, RawOchre, getParserOptions, parseBibliographyList, parseGallery, parseIdentification, parseItem, parseLinkedItems, parseLinks, parseMetadata, parseMetadataLanguages, parseNotes, parsePersonList, parseProperties, parseSetItems, parseSimplifiedProperties, parseStringLike, resolveDefaultLanguage, resolveLanguages };

package/dist/parsers/index.mjs

@@ -430,79 +430,6 @@ 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);
@@ -1196,7 +1123,6 @@ 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),
@@ -1399,4 +1325,4 @@ function parseItem(rawData, options) {
 	};
 }
 //#endregion
-export { getParserOptions, parseBibliographyList, parseGallery, parseIdentification, parseItem, parseLinkedItems, parseLinks, parseMetadata, parseMetadataLanguages, parseNotes, parseOcrMatches, parsePersonList, parseProperties, parseSetItems, parseSimplifiedProperties, parseStringLike, resolveDefaultLanguage, resolveLanguages };
+export { getParserOptions, parseBibliographyList, parseGallery, parseIdentification, parseItem, parseLinkedItems, parseLinks, parseMetadata, parseMetadataLanguages, parseNotes, parsePersonList, parseProperties, parseSetItems, parseSimplifiedProperties, parseStringLike, resolveDefaultLanguage, resolveLanguages };

package/dist/parsers/multilingual.d.mts

@@ -26,11 +26,17 @@ type MultilingualStringEntries<T extends ReadonlyArray<string> = ReadonlyArray<s
  * Options for creating and working with multilingual strings
  */
 type MultilingualOptions = {
-  /** Default language to use for fallbacks */
+  /**
+    Default language to use for fallbacks
+    */
   defaultLanguage?: string;
-  /** Available languages for this string */
+  /**
+    Available languages for this string
+    */
   availableLanguages?: ReadonlyArray<string>;
-  /** Alias values carried by OCHRE as zxx content */
+  /**
+    Alias values carried by OCHRE as zxx content
+    */
   aliases?: ReadonlyArray<string>;
 };
 type MultilingualContent<T extends ReadonlyArray<string>> = Partial<Record<T[number], ReadonlyArray<MultilingualStringEntry>>>;
@@ -78,7 +84,9 @@ declare class MultilingualString<T extends ReadonlyArray<string> = ReadonlyArray
   /**
    * Create a new multilingual string from an object of language codes to text.
    */
-  /** @internal */
+  /**
+    @internal
+    */
   constructor(init: MultilingualStringInternalInit<T>);
   constructor(content: MultilingualStringObject<T>, languages: T, options?: MultilingualOptions);
   constructor(content?: Partial<Record<string, MultilingualStringInput>>, languages?: undefined, options?: MultilingualOptions);

package/dist/parsers/website/index.mjs

@@ -907,17 +907,18 @@ function parseWebpage(webpageResource, options, context, slugPrefix) {
 		returnWebpage.properties.isNavbarSearchBarDisplayed = pageReader.valueOr("navbar-search-bar-displayed", true);
 		const redirectValue = pageReader.valueNode("redirect-to");
 		const redirectTarget = parseWebsiteLinkTarget(redirectValue, context);
-		if (redirectTarget != null) if (redirectValue?.href == null && redirectValue?.uuid != null) returnWebpage.properties.redirect = {
-			type: "page",
-			slug: redirectTarget,
-			uuid: redirectValue.uuid
-		};
-		else returnWebpage.properties.redirect = {
-			type: "url",
-			href: redirectTarget,
-			isExternal: redirectTarget.startsWith("http")
-		};
-		else if (redirectValue?.uuid != null) returnWebpage.properties.redirect = {
+		if (redirectTarget != null) {
+			if (redirectValue?.href == null && redirectValue?.uuid != null) returnWebpage.properties.redirect = {
+				type: "page",
+				slug: redirectTarget,
+				uuid: redirectValue.uuid
+			};
+			else returnWebpage.properties.redirect = {
+				type: "url",
+				href: redirectTarget,
+				isExternal: redirectTarget.startsWith("http")
+			};
+		} else if (redirectValue?.uuid != null) returnWebpage.properties.redirect = {
 			type: "item",
 			uuid: redirectValue.uuid,
 			pageType: "item"
@@ -1015,7 +1016,7 @@ function parseSidebar(resources, options, context) {
 			}
 		}
 	}
-	if (items.length > 0 && title != null) returnSidebar = {
+	if (title != null && items.length > 0) returnSidebar = {
 		isDisplayed: true,
 		items,
 		title,

package/dist/query.d.mts

@@ -1,40 +1,32 @@
 import { Query } from "./types/index.mjs";
 //#region src/query.d.ts
+declare function buildBelongsToCollectionQueryExpression(belongsToCollectionScopeUuids: Array<string>, belongsToCollectionPropertyVariableUuid: string): string | null;
 /**
- * Error message for OCR queries nested inside an OR group
- * @internal
- */
-declare const OCR_DISJUNCTION_ERROR_MESSAGE = "OCR queries cannot be nested inside an OR group because they are resolved by a document join instead of a CTS query";
-type OcrUuidBinding = {
-  name: string;
-  expression: string;
-};
-declare function buildAndCtsQueryExpression(queryExpressions: Array<string>): string | null;
-/**
- * Compile one CTS query expression per OCR search term, in word order
+ * Compile a query tree into the XQuery `let` clauses that bind `$items` to the
+ * matching Set items
  *
- * Highlighting matches each OCR word against these same per-term expressions,
- * so hit locations always agree with what the filter matched.
- * @internal
- */
-declare function buildOcrTermQueryExpressions(parameters: {
-  value: string;
-  matchMode: "includes" | "exact";
-  isCaseSensitive: boolean;
-}): Array<string>;
-/**
- * Whether a query tree nests an OCR leaf inside an OR group
- * @internal
+ * Most queries compile to a single `cts:search` over the Set item projections.
+ * An `ocr` leaf cannot: the projections drop the `<ocr>` layer, so it resolves
+ * to a search over the Resource documents whose matching UUIDs are joined back
+ * in as an item path predicate. Path predicates only ever AND, so an `ocr` leaf
+ * that sits under an `or` becomes its own arm of a node union instead, and one
+ * that sits under an `and` alongside a union becomes an intersection.
+ *
+ * The searchable path has to stay inline in `cts:search`: binding it to a
+ * variable first makes every query XDMP-UNSEARCHABLE.
+ * @param parameters - The parameters for the compilation
+ * @param parameters.queries - Recursive query tree to compile, if any
+ * @param parameters.baseItemsExpression - The inline XQuery path selecting the items to search
+ * @param parameters.scopeQueryExpression - An optional CTS query ANDed into every compiled search
+ * @returns The prolog declaring the query helpers, and the `let` clauses binding `$items`
  */
-declare function hasOcrQueryInDisjunction(query: Query, isInDisjunction?: boolean): boolean;
-declare function buildBelongsToCollectionQueryExpression(belongsToCollectionScopeUuids: Array<string>, belongsToCollectionPropertyVariableUuid: string): string | null;
 declare function buildQueryPlan(parameters: {
   queries: Query | null;
+  baseItemsExpression: string;
+  scopeQueryExpression?: string | null;
 }): {
   prolog: string;
-  queryExpression: string | null;
-  ocrBindings: Array<OcrUuidBinding>;
-  itemPredicates: string;
+  itemsClause: string;
 };
 //#endregion
-export { OCR_DISJUNCTION_ERROR_MESSAGE, buildAndCtsQueryExpression, buildBelongsToCollectionQueryExpression, buildOcrTermQueryExpressions, buildQueryPlan, hasOcrQueryInDisjunction };
+export { buildBelongsToCollectionQueryExpression, buildQueryPlan };