@matdata/yasr 5.20.3 → 6.1.0

package/src/plugins/response/uriUtils.ts

@@ -0,0 +1,117 @@
+/**
+ * Standalone helpers for the Response plugin's Ctrl+Click URI exploration feature.
+ *
+ * These functions are intentionally free of any DOM or CodeMirror dependencies so
+ * that they can be unit-tested in isolation.
+ */
+
+// Matches angle-bracket IRIs as they appear in Turtle / N-Triples / TriG, e.g. `<http://example.org/foo>`
+const ANGLE_IRI_RE = /<([^<>\s"{}|\\^`]+)>/g;
+// Matches bare or quoted URIs as they appear in JSON / XML / CSV responses, e.g. `"http://example.org/foo"`
+const BARE_URI_RE = /(?:https?|urn|ftp|mailto):[^\s<>"'`{}|\\^[\]]+/g;
+// Matches Turtle/SPARQL prefix declarations such as `@prefix ex: <...> .` and `PREFIX ex: <...>`.
+const PREFIX_DECL_RE = /^\s*(?:@prefix|PREFIX)\s+(\w*):\s*<[^>]+>\s*\.?\s*$/i;
+
+/**
+ * Remove trailing punctuation that is commonly adjacent to a URI in a serialized
+ * response (e.g. the `.` terminating a Turtle statement) but is not part of the URI.
+ */
+function sanitizeUri(uri: string): string {
+  return uri.replace(/[.,;)\]]+$/, "").trim();
+}
+
+/**
+ * Find the URI located at the given character offset within a serialized SPARQL response.
+ *
+ * Angle-bracket IRIs (Turtle/N-Triples style) take precedence over bare URIs so that a
+ * click inside `<http://…>` returns the full IRI rather than the embedded `http://…` run.
+ *
+ * @param text   The full response text as shown in the response view.
+ * @param offset The character offset (0-based) of the click within `text`.
+ * @returns The URI at that offset, or `undefined` when the offset is not on a URI.
+ */
+export function extractUriAtOffset(text: string, offset: number): string | undefined {
+  if (!text || offset < 0 || offset >= text.length) return undefined;
+
+  let match: RegExpExecArray | null;
+
+  ANGLE_IRI_RE.lastIndex = 0;
+  while ((match = ANGLE_IRI_RE.exec(text)) !== null) {
+    const start = match.index;
+    const end = start + match[0].length;
+    if (offset >= start && offset <= end) {
+      const uri = sanitizeUri(match[1]);
+      return uri.length > 0 ? uri : undefined;
+    }
+  }
+
+  BARE_URI_RE.lastIndex = 0;
+  while ((match = BARE_URI_RE.exec(text)) !== null) {
+    const start = match.index;
+    const end = start + match[0].length;
+    if (offset >= start && offset <= end) {
+      const uri = sanitizeUri(match[0]);
+      return uri.length > 0 ? uri : undefined;
+    }
+  }
+
+  return undefined;
+}
+
+/**
+ * Build a CONSTRUCT query that retrieves all triples where the given URI is the subject.
+ *
+ * @param uri The URI to look up as a subject.
+ * @returns A SPARQL `CONSTRUCT` query string.
+ */
+export function buildSubjectOfQuery(uri: string): string {
+  return `CONSTRUCT { <${uri}> ?p ?o } WHERE { <${uri}> ?p ?o }`;
+}
+
+/**
+ * Build a CONSTRUCT query that retrieves all triples where the given URI is the object.
+ *
+ * @param uri The URI to look up as an object.
+ * @returns A SPARQL `CONSTRUCT` query string.
+ */
+export function buildObjectOfQuery(uri: string): string {
+  return `CONSTRUCT { ?s ?p <${uri}> } WHERE { ?s ?p <${uri}> } LIMIT 1000`;
+}
+
+/**
+ * Remove duplicate prefix declarations from a text block.
+ *
+ * Prefixes are considered duplicates by label (case-insensitive), and can be
+ * deduplicated against already-rendered content in the response viewer.
+ */
+export function stripDuplicatePrefixDeclarations(text: string, existingText = ""): string {
+  if (!text) return text;
+
+  const seenLabels = new Set<string>();
+  const addSeenLabels = (source: string) => {
+    for (const line of source.split("\n")) {
+      const match = line.match(PREFIX_DECL_RE);
+      if (!match) continue;
+      seenLabels.add(match[1].toLowerCase());
+    }
+  };
+
+  addSeenLabels(existingText);
+
+  const deduplicatedLines: string[] = [];
+  for (const line of text.split("\n")) {
+    const match = line.match(PREFIX_DECL_RE);
+    if (!match) {
+      deduplicatedLines.push(line);
+      continue;
+    }
+
+    const label = match[1].toLowerCase();
+    if (seenLabels.has(label)) continue;
+
+    seenLabels.add(label);
+    deduplicatedLines.push(line);
+  }
+
+  return deduplicatedLines.join("\n");
+}