@riboseinc/anafero-cli 0.0.66 → 0.0.67

package/build-site.mjs

@@ -313648,12 +313648,48 @@ function getTextContent(graph, subject, partPredicate) {
 // ../anafero/search.mts
 init_cjs_shim();
 function preprocessStringForIndexing(text9) {
-  return text9.normalize("NFKD").replace(/\p{Diacritic}/gu, "").trim();
+  return text9.normalize("NFKD").replace(/\p{Diacritic}/gu, "").replace(/[\p{P}$+<=>^`|~]/gu, " ").replace(/[\u200B-\u200D\uFEFF]/g, "").trim();
 }
-function extractRelationsForIndexing(uri, graph, isDefinedSubject) {
-  return graph.filter(
-    ([s2, p3, o2]) => p3 === "hasPart" && (s2 === uri || s2 === ROOT_SUBJECT) && !o2.startsWith("data:") && (!isURIString(o2) || !isDefinedSubject(o2))
+function extractRelationsForIndexing(uri, graph, isIndexable, isAlreadyIndexed, _seen, _log) {
+  const seen = _seen ?? /* @__PURE__ */ new Set();
+  seen.add(uri);
+  const nonData = graph.filter(
+    ([, , o2]) => !o2.startsWith("data:")
   );
+  const immediateGraph = nonData.filter(
+    ([s2, ,]) => s2 === uri || s2 === ROOT_SUBJECT
+  );
+  const references = immediateGraph.filter(
+    ([, , o2]) => isURIString(o2)
+  );
+  const indexable = immediateGraph.filter(
+    ([s2, p3, o2]) => !isURIString(o2) && isIndexable([s2, p3, o2])
+  ).map(([, , o2]) => o2).filter((o2) => o2.trim() !== "");
+  for (const [, , o2] of references) {
+    if (_log) {
+      console.debug(
+        "search: processing triple object",
+        { o: o2, isAlreadyIndexed: isAlreadyIndexed(o2) }
+      );
+    }
+    if (!isAlreadyIndexed(o2) && !seen.has(o2)) {
+      indexable.push(...extractRelationsForIndexing(
+        o2,
+        graph,
+        isIndexable,
+        isAlreadyIndexed,
+        seen,
+        _log
+      ));
+    }
+  }
+  if (_log) {
+    console.debug(
+      "search: obtained indexable from graph",
+      { uri, graph, indexable }
+    );
+  }
+  return indexable;
 }
 
 // ../anafero-gui/loader.mts
@@ -317656,10 +317692,19 @@ ${inject.head ?? ""}`;
     );
     pathProgress({ state: "updating artifacts" });
     if (content?.content) {
+      let isIndexable2 = function(rel) {
+        return rel[1] === "hasPart";
+      }, isIndexed2 = function(uri) {
+        return reader.exists(uri) && describedResourceIDs.has(uri);
+      };
+      var isIndexable = isIndexable2, isIndexed = isIndexed2;
       if (meta.primaryLanguageID) {
         allLanguages.add(meta.primaryLanguageID);
       }
       pathProgress({ state: "indexing page resource" });
+      const describedResourceIDs = gatherDescribedResourcesFromJsonifiedProseMirrorNode(
+        content.content.contentDoc
+      );
       resourceMap[path3] = resourceURI;
       resourceGraph.push([resourceURI, "isDefinedBy", `${path3}/resource.json`]);
       resourceDescriptions[resourceURI] = resourceMeta;
@@ -317668,13 +317713,15 @@ ${inject.head ?? ""}`;
         title: preprocessStringForIndexing(meta.labelInPlainText),
         lang: meta.primaryLanguageID || "",
         body: preprocessStringForIndexing(
-          extractRelationsForIndexing(resourceURI, relations, reader.exists).join("").trim()
+          extractRelationsForIndexing(
+            resourceURI,
+            relations,
+            isIndexable2,
+            isIndexed2
+          ).join("\n").trim()
         )
       };
       pathProgress({ state: "indexing on-page subresources" });
-      const describedResourceIDs = gatherDescribedResourcesFromJsonifiedProseMirrorNode(
-        content.content.contentDoc
-      );
       for (const inPageResourceID of describedResourceIDs) {
         if (reader.exists(inPageResourceID)) {
           const pathWithFragment = `${path3}#${encodeURIComponent(inPageResourceID)}`;
@@ -317684,13 +317731,19 @@ ${inject.head ?? ""}`;
           resourceGraph.push([inPageResourceID, "isDefinedBy", `${path3}/resource.json`]);
           resourceDescriptions[inPageResourceID] = meta2;
           if (inPageResourceID !== resourceURI) {
+            const body = preprocessStringForIndexing(
+              extractRelationsForIndexing(
+                inPageResourceID,
+                graph2,
+                isIndexable2,
+                isIndexed2
+              ).join("\n").trim()
+            );
             searchableResources.resources[inPageResourceID] = {
               name: inPageResourceID,
               title: preprocessStringForIndexing(meta2.labelInPlainText),
               lang: meta2.primaryLanguageID || "",
-              body: preprocessStringForIndexing(
-                extractRelationsForIndexing(inPageResourceID, graph2, reader.exists).join("").trim()
-              )
+              body
             };
           }
         } else {

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@riboseinc/anafero-cli",
   "type": "module",
-  "version": "0.0.66",
+  "version": "0.0.67",
   "packageManager": "yarn@4.5.0",
   "bin": {
     "build-site": "build-site.mjs"