ochre-sdk 0.22.3 → 0.22.5

package/dist/index.mjs

@@ -2167,54 +2167,121 @@ const CTS_INCLUDES_STOP_WORDS = new Set([
 	"the",
 	"to"
 ]);
+const CTS_INCLUDES_TOKEN_WORD_REGEX = /^\p{L}+$/u;
+const CTS_INCLUDES_TOKEN_REGEX = /[\p{L}\p{N}*?]+/gu;
 const CTS_INCLUDES_TOKEN_SPLIT_REGEX = /\W+/u;
-const CONTENT_TARGET_CONTAINER_ELEMENTS = {
-	title: "identification",
-	description: "description",
-	image: "image",
-	periods: "period",
-	bibliography: "bibliography"
+const CONTENT_TARGET_CONTENT_ELEMENT_PATHS = {
+	title: [
+		"identification",
+		"label",
+		"content"
+	],
+	description: ["description", "content"],
+	image: [
+		"image",
+		"identification",
+		"label",
+		"content"
+	],
+	periods: [
+		"periods",
+		"period",
+		"identification",
+		"label",
+		"content"
+	],
+	bibliography: [
+		"bibliographies",
+		"bibliography",
+		"identification",
+		"label",
+		"content"
+	]
 };
 function tokenizeIncludesSearchValue(params) {
 	const { value, isCaseSensitive } = params;
-	const rawTerms = (isCaseSensitive ? value : value.toLowerCase()).split(CTS_INCLUDES_TOKEN_SPLIT_REGEX);
+	const rawTerms = (isCaseSensitive ? value : value.toLowerCase()).match(CTS_INCLUDES_TOKEN_REGEX) ?? [];
 	const terms = [];
 	for (const term of rawTerms) {
+		if (term.includes("*") || term.includes("?")) {
+			if (term.replaceAll("*", "").replaceAll("?", "") !== "") terms.push(term);
+			continue;
+		}
 		const normalizedTerm = term.toLowerCase();
 		if (normalizedTerm !== "" && !CTS_INCLUDES_STOP_WORDS.has(normalizedTerm)) terms.push(term);
 	}
 	return terms;
 }
+function tokenizeExactPhraseSearchValue(params) {
+	const { value, isCaseSensitive } = params;
+	const rawTerms = (isCaseSensitive ? value : value.toLowerCase()).split(CTS_INCLUDES_TOKEN_SPLIT_REGEX);
+	const terms = [];
+	for (const term of rawTerms) if (term !== "") terms.push(term);
+	return terms;
+}
+function hasWildcardCharacters(value) {
+	return value.includes("*") || value.includes("?");
+}
+function getWildcardStrippedValue(value) {
+	return value.replaceAll("*", "").replaceAll("?", "");
+}
+function shouldUseStemmedTextSearch(value) {
+	const wildcardStrippedValue = getWildcardStrippedValue(value);
+	return wildcardStrippedValue.length >= 3 && CTS_INCLUDES_TOKEN_WORD_REGEX.test(wildcardStrippedValue);
+}
 function buildCtsMatchOptionsExpression(params) {
-	const { isCaseSensitive } = params;
-	return `(${[
+	const { matchMode, isCaseSensitive, queryFamily, language, isWildcarded } = params;
+	const { isStemmed } = params;
+	const options = [
 		isCaseSensitive ? "case-sensitive" : "case-insensitive",
-		"diacritic-insensitive",
-		"punctuation-insensitive",
-		"whitespace-insensitive",
-		"unstemmed",
-		"unwildcarded"
-	].map((option) => stringLiteral(option)).join(", ")})`;
+		matchMode === "exact" ? "diacritic-sensitive" : "diacritic-insensitive",
+		matchMode === "exact" ? "punctuation-sensitive" : "punctuation-insensitive",
+		matchMode === "exact" ? "whitespace-sensitive" : "whitespace-insensitive"
+	];
+	if (matchMode === "exact") options.push("unstemmed", "unwildcarded");
+	else if (queryFamily === "text") {
+		options.push(isStemmed ? "stemmed" : "unstemmed", isWildcarded ? "wildcarded" : "unwildcarded");
+		if (isStemmed && language != null && language !== "") options.push(`lang=${language}`);
+	} else options.push("unstemmed", isWildcarded ? "wildcarded" : "unwildcarded");
+	return `(${options.map((option) => stringLiteral(option)).join(", ")})`;
 }
 function buildCtsWordQueryExpression(params) {
-	const { value, matchMode, isCaseSensitive } = params;
+	const { value, matchMode, isCaseSensitive, queryFamily, language } = params;
+	const isWildcarded = matchMode === "includes" && hasWildcardCharacters(value);
+	const isStemmed = matchMode === "includes" && queryFamily === "text" && !isWildcarded && shouldUseStemmedTextSearch(value);
 	return `cts:word-query(${stringLiteral(value)}, ${buildCtsMatchOptionsExpression({
 		matchMode,
-		isCaseSensitive
+		isCaseSensitive,
+		queryFamily,
+		language,
+		isWildcarded,
+		isStemmed
 	})})`;
 }
 function buildCtsElementWordQueryExpression(params) {
-	const { elementName, value, matchMode, isCaseSensitive } = params;
+	const { elementName, value, matchMode, isCaseSensitive, queryFamily, language } = params;
+	const isWildcarded = matchMode === "includes" && hasWildcardCharacters(value);
+	const isStemmed = matchMode === "includes" && queryFamily === "text" && !isWildcarded && shouldUseStemmedTextSearch(value);
 	return `cts:element-word-query(xs:QName("${elementName}"), ${stringLiteral(value)}, ${buildCtsMatchOptionsExpression({
 		matchMode,
-		isCaseSensitive
+		isCaseSensitive,
+		queryFamily,
+		language,
+		isWildcarded,
+		isStemmed
 	})})`;
 }
 function buildCtsElementAttributeWordQueryExpression(params) {
-	const { elementName, attributeName, value, matchMode, isCaseSensitive } = params;
+	const { elementName, attributeName, value, matchMode, isCaseSensitive, queryFamily, language } = params;
+	const isWildcarded = matchMode === "includes" && hasWildcardCharacters(value);
+	const isStemmed = matchMode === "includes" && queryFamily === "text" && !isWildcarded && shouldUseStemmedTextSearch(value);
 	return `cts:element-attribute-word-query(xs:QName("${elementName}"), xs:QName("${attributeName}"), ${stringLiteral(value)}, ${buildCtsMatchOptionsExpression({
 		matchMode,
-		isCaseSensitive
+		isCaseSensitive,
+		queryFamily,
+		language,
+		isWildcarded,
+		isStemmed
 	})})`;
 }
 function buildCtsElementValueQueryExpression(params) {
@@ -2236,21 +2303,36 @@ function buildPlainElementAttributeValueQueryExpression(params) {
 	return `cts:element-attribute-value-query(xs:QName("${elementName}"), xs:QName("${attributeName}"), ${stringLiteral(value)})`;
 }
 function buildSearchableContentTextQueryExpression(params) {
-	const { value, matchMode, isCaseSensitive } = params;
-	if (matchMode === "exact") return buildCtsWordQueryExpression({
-		value,
-		matchMode,
-		isCaseSensitive
-	});
+	const { value, matchMode, isCaseSensitive, language } = params;
+	if (matchMode === "exact") {
+		const phraseTerms = tokenizeExactPhraseSearchValue({
+			value,
+			isCaseSensitive
+		});
+		if (phraseTerms.length > 1) return buildAndCtsQueryExpressionInternal(phraseTerms.map((term) => buildCtsWordQueryExpression({
+			value: term,
+			matchMode,
+			isCaseSensitive
+		})));
+		return buildCtsWordQueryExpression({
+			value,
+			matchMode,
+			isCaseSensitive
+		});
+	}
 	return buildOrCtsQueryExpressionInternal([buildCtsElementWordQueryExpression({
 		elementName: "string",
 		value,
 		matchMode,
-		isCaseSensitive
+		isCaseSensitive,
+		queryFamily: "text",
+		language
 	}), buildCtsWordQueryExpression({
 		value,
 		matchMode,
-		isCaseSensitive
+		isCaseSensitive,
+		queryFamily: "text",
+		language
 	})]);
 }
 function buildNestedElementQuery(elementNames, queryExpression) {
@@ -2308,7 +2390,8 @@ function buildValueContentInnerQuery(params) {
 	return buildNestedElementQuery(["content"], buildAndCtsQueryExpressionInternal([buildContentLanguageQuery(language), buildSearchableContentTextQueryExpression({
 		value,
 		matchMode,
-		isCaseSensitive
+		isCaseSensitive,
+		language
 	})]));
 }
 function buildValueDirectTextInnerQuery(params) {
@@ -2321,7 +2404,8 @@ function buildValueDirectTextInnerQuery(params) {
 		elementName: "value",
 		value,
 		matchMode,
-		isCaseSensitive
+		isCaseSensitive,
+		queryFamily: "raw"
 	});
 	return buildAndCtsQueryExpressionInternal([buildNotCtsQueryExpression(buildNestedElementQuery(["content"], "cts:true-query()")), directTextQuery]);
 }
@@ -2338,7 +2422,8 @@ function buildValueRawValueInnerQuery(params) {
 		attributeName: "rawValue",
 		value,
 		matchMode,
-		isCaseSensitive
+		isCaseSensitive,
+		queryFamily: "raw"
 	});
 }
 function buildNotesQueryExpression(params) {
@@ -2357,20 +2442,24 @@ function buildNotesQueryExpression(params) {
 		attributeName: "title",
 		value,
 		matchMode,
-		isCaseSensitive
+		isCaseSensitive,
+		queryFamily: "text",
+		language
 	}), buildSearchableContentTextQueryExpression({
 		value,
 		matchMode,
-		isCaseSensitive
+		isCaseSensitive,
+		language
 	})])]));
 }
 function buildContentTargetQueryExpression(params) {
 	const { target, value, matchMode, isCaseSensitive, language } = params;
-	const containerElement = CONTENT_TARGET_CONTAINER_ELEMENTS[target];
-	return buildNestedElementQuery([containerElement], buildAndCtsQueryExpressionInternal([buildContentLanguageQuery(language), buildSearchableContentTextQueryExpression({
+	const contentElementPath = CONTENT_TARGET_CONTENT_ELEMENT_PATHS[target];
+	return buildNestedElementQuery(contentElementPath, buildAndCtsQueryExpressionInternal([buildContentLanguageQuery(language), buildSearchableContentTextQueryExpression({
 		value,
 		matchMode,
-		isCaseSensitive
+		isCaseSensitive,
+		language
 	})]));
 }
 function buildPropertyQueryExpression(params) {
@@ -2379,6 +2468,12 @@ function buildPropertyQueryExpression(params) {
 	if (propertyVariable != null) propertyQueryExpressions.unshift(buildPropertyLabelQuery(propertyVariable));
 	return buildNestedElementQuery(["properties", "property"], buildAndCtsQueryExpressionInternal(propertyQueryExpressions));
 }
+function buildPropertyPresenceQueryExpression(params) {
+	return buildPropertyQueryExpression({
+		propertyVariable: params.propertyVariable,
+		queryExpression: "cts:true-query()"
+	});
+}
 function buildPropertyStringQueryExpression(params) {
 	const { propertyVariable, value, matchMode, isCaseSensitive, language } = params;
 	return buildPropertyQueryExpression({
@@ -2408,35 +2503,35 @@ function buildPropertyScalarQueryExpression(params) {
 }
 function buildPropertyAllQueryExpression(params) {
 	const { query, value, matchMode } = params;
-	if (matchMode === "exact") return buildPropertyQueryExpression({
-		propertyVariable: query.propertyVariable,
-		queryExpression: buildNestedElementQuery(["value"], buildAndCtsQueryExpressionInternal([buildValueNotIdRefQuery(), buildCtsWordQueryExpression({
-			value,
-			matchMode,
-			isCaseSensitive: query.isCaseSensitive
-		})]))
-	});
-	return buildPropertyQueryExpression({
+	if (matchMode === "includes") return buildPropertyQueryExpression({
 		propertyVariable: query.propertyVariable,
 		queryExpression: buildNestedElementQuery(["value"], buildAndCtsQueryExpressionInternal([buildValueNotIdRefQuery(), buildOrCtsQueryExpressionInternal([
-			buildValueRawValueInnerQuery({
+			buildValueContentInnerQuery({
+				language: query.language,
 				value,
 				matchMode,
 				isCaseSensitive: query.isCaseSensitive
 			}),
-			buildValueDirectTextInnerQuery({
+			buildValueRawValueInnerQuery({
 				value,
 				matchMode,
 				isCaseSensitive: query.isCaseSensitive
 			}),
-			buildValueContentInnerQuery({
-				language: query.language,
+			buildValueDirectTextInnerQuery({
 				value,
 				matchMode,
 				isCaseSensitive: query.isCaseSensitive
 			})
 		])]))
 	});
+	return buildPropertyQueryExpression({
+		propertyVariable: query.propertyVariable,
+		queryExpression: buildNestedElementQuery(["value"], buildAndCtsQueryExpressionInternal([buildValueNotIdRefQuery(), buildCtsWordQueryExpression({
+			value,
+			matchMode,
+			isCaseSensitive: query.isCaseSensitive
+		})]))
+	});
 }
 function buildPropertyIdRefQueryExpression(params) {
 	const { propertyVariable, value } = params;
@@ -2451,8 +2546,8 @@ function buildPropertyIdRefQueryExpression(params) {
 }
 function buildPropertyDateRangeQueryExpression(query) {
 	const rangeQueryExpressions = [];
-	if (query.from != null) rangeQueryExpressions.push(`cts:element-attribute-range-query(xs:QName("value"), xs:QName("rawValue"), ">=", xs:${query.dataType}(${stringLiteral(query.from)}))`);
-	if (query.to != null) rangeQueryExpressions.push(`cts:element-attribute-range-query(xs:QName("value"), xs:QName("rawValue"), "<=", xs:${query.dataType}(${stringLiteral(query.to)}))`);
+	if (query.from != null) rangeQueryExpressions.push(`cts:element-attribute-range-query(xs:QName("value"), xs:QName("rawValue"), ">=", ${stringLiteral(query.from)})`);
+	if (query.to != null) rangeQueryExpressions.push(`cts:element-attribute-range-query(xs:QName("value"), xs:QName("rawValue"), "<=", ${stringLiteral(query.to)})`);
 	return buildPropertyQueryExpression({
 		propertyVariable: query.propertyVariable,
 		queryExpression: buildNestedElementQuery(["value"], buildAndCtsQueryExpressionInternal(rangeQueryExpressions))
@@ -2543,6 +2638,7 @@ function buildLeafValueQueryExpression(params) {
 	}
 }
 function buildLeafQueryExpression(query) {
+	if (query.target === "property" && query.dataType !== "date" && query.dataType !== "dateTime" && !("value" in query) && query.propertyVariable != null) return buildPropertyPresenceQueryExpression({ propertyVariable: query.propertyVariable });
 	if (query.target === "property" && (query.dataType === "date" || query.dataType === "dateTime") && query.value == null) return buildPropertyDateRangeQueryExpression(query);
 	const searchValue = getLeafSearchValue(query);
 	if (searchValue == null) throw new Error("Missing searchable value for query leaf");
@@ -2555,6 +2651,11 @@ function buildLeafQueryExpression(query) {
 		value: searchValue,
 		isCaseSensitive: query.isCaseSensitive
 	});
+	const fullValueQueryExpression = buildLeafValueQueryExpression({
+		query,
+		value: searchValue,
+		matchMode: "exact"
+	});
 	if (terms.length === 0) return "cts:false-query()";
 	const termQueryExpressions = [];
 	for (const term of terms) termQueryExpressions.push(buildLeafValueQueryExpression({
@@ -2562,7 +2663,9 @@ function buildLeafQueryExpression(query) {
 		value: term,
 		matchMode: "includes"
 	}));
-	return buildAndCtsQueryExpressionInternal(termQueryExpressions);
+	const tokenizedQueryExpression = buildAndCtsQueryExpressionInternal(termQueryExpressions);
+	if (terms.length === 1) return tokenizedQueryExpression;
+	return buildOrCtsQueryExpressionInternal([fullValueQueryExpression, tokenizedQueryExpression]);
 }
 function getGroupableIncludesValue(query) {
 	if (query.matchMode !== "includes" || query.isNegated === true) return null;
@@ -2603,11 +2706,18 @@ function getCompatibleIncludesGroupLeaves(query) {
 	return leafQueries;
 }
 function buildIncludesGroupMember(query) {
-	return { buildTermQuery: (term) => buildLeafValueQueryExpression({
-		query,
-		value: term,
-		matchMode: "includes"
-	}) };
+	return {
+		buildTermQuery: (term) => buildLeafValueQueryExpression({
+			query,
+			value: term,
+			matchMode: "includes"
+		}),
+		buildFullValueQuery: (value) => buildLeafValueQueryExpression({
+			query,
+			value,
+			matchMode: "exact"
+		})
+	};
 }
 function buildIncludesGroupQueryExpression(queries) {
 	const firstQuery = queries[0];
@@ -2621,12 +2731,16 @@ function buildIncludesGroupQueryExpression(queries) {
 	if (terms.length === 0) return "cts:false-query()";
 	const members = queries.map((query) => buildIncludesGroupMember(query));
 	const perTermQueryExpressions = [];
+	const fullValueFieldQueryExpressions = [];
+	for (const member of members) fullValueFieldQueryExpressions.push(member.buildFullValueQuery(groupValue));
 	for (const term of terms) {
 		const fieldQueryExpressions = [];
 		for (const member of members) fieldQueryExpressions.push(member.buildTermQuery(term));
 		perTermQueryExpressions.push(buildOrCtsQueryExpressionInternal(fieldQueryExpressions));
 	}
-	return buildAndCtsQueryExpressionInternal(perTermQueryExpressions);
+	const tokenizedGroupQueryExpression = buildAndCtsQueryExpressionInternal(perTermQueryExpressions);
+	if (terms.length === 1) return tokenizedGroupQueryExpression;
+	return buildOrCtsQueryExpressionInternal([buildOrCtsQueryExpressionInternal(fullValueFieldQueryExpressions), tokenizedGroupQueryExpression]);
 }
 function buildQueryNode(query) {
 	if (isQueryLeaf(query)) {
@@ -2959,6 +3073,23 @@ function getPropertyVariableUuidsFromQueries(queries) {
 	}
 	return [...propertyVariableUuids];
 }
+function getItemFilterQueriesFromPropertyValueQueries(queries) {
+	if (queries == null) return null;
+	if ("target" in queries) {
+		if (queries.target !== "property") return queries;
+		if (queries.dataType === "date" || queries.dataType === "dateTime") return queries;
+		return "value" in queries && queries.value != null ? queries : null;
+	}
+	const filteredChildren = [];
+	const childQueries = "and" in queries ? queries.and : queries.or;
+	for (const childQuery of childQueries) {
+		const filteredChildQuery = getItemFilterQueriesFromPropertyValueQueries(childQuery);
+		if (filteredChildQuery != null) filteredChildren.push(filteredChildQuery);
+	}
+	if (filteredChildren.length === 0) return null;
+	if (filteredChildren.length === 1) return filteredChildren[0] ?? null;
+	return "and" in queries ? { and: filteredChildren } : { or: filteredChildren };
+}
 /**
 * Schema for a single property value query item in the OCHRE API response
 */
@@ -3040,7 +3171,7 @@ function buildXQuery(params) {
 	if (setScopeUuids.length > 0) setScopeFilter = `/set[(${setScopeUuids.map((uuid) => `@uuid="${uuid}"`).join(" or ")})]/items/*`;
 	const propertyVariableFilters = getPropertyVariableUuidsFromQueries(queries).map((uuid) => `@uuid="${uuid}"`).join(" or ");
 	const baseItemsExpression = `doc()/ochre${setScopeFilter}`;
-	const compiledQueryPlan = buildQueryPlan({ queries });
+	const compiledQueryPlan = buildQueryPlan({ queries: getItemFilterQueriesFromPropertyValueQueries(queries) });
 	const itemsQueryExpressions = [];
 	const belongsToCollectionQueryExpression = buildBelongsToCollectionQueryExpression(belongsToCollectionScopeUuids, BELONGS_TO_COLLECTION_UUID);
 	if (compiledQueryPlan.queryExpression != null) itemsQueryExpressions.push(compiledQueryPlan.queryExpression);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ochre-sdk",
-  "version": "0.22.3",
+  "version": "0.22.5",
   "type": "module",
   "license": "MIT",
   "description": "Node.js library for working with OCHRE (Online Cultural and Historical Research Environment) data",