@hraness/wordcell 0.22.2 → 0.22.4

package/dist/index.js

@@ -167,7 +167,7 @@ import {
   publishVault,
   renderPublishReportText,
   serializeSiteFile
-} from "./index-hmw17zaa.js";
+} from "./index-pnt9wr59.js";
 import {
   MAX_NOTE_UTF8_BYTES,
   MAX_SCANNED_NOTES,
@@ -245,7 +245,7 @@ import {
   parseSitePostingsV1,
   parseSiteTermsV1
 } from "./index-66pshdtx.js";
-import"./index-3agn8scn.js";
+import"./index-23m6bbjt.js";
 import {
   auditAgentGuideRepository,
   auditAgentGuideSource,

package/dist/publish-reader/reader.js

@@ -757,22 +757,58 @@ function comparePublishScores(left, right) {
   return Number(right.identity) - Number(left.identity) || right.score - left.score || left.i - right.i;
 }
 function publishSnippet(text, query, fallback, windowBytes = 160) {
-  const normalized = publishNormalize(text);
-  let offset = query.normalized === "" ? -1 : normalized.indexOf(query.normalized);
-  if (offset < 0) {
+  if (!Number.isSafeInteger(windowBytes) || windowBytes < 0)
+    throw new RangeError("Snippet window must be a nonnegative safe byte count.");
+  const display = text.normalize("NFC");
+  const normalized = publishNormalize(display);
+  let match = query.normalized === "" ? -1 : normalized.indexOf(query.normalized);
+  if (match < 0) {
     for (const term of query.terms) {
-      offset = normalized.indexOf(term);
-      if (offset >= 0)
+      match = normalized.indexOf(term);
+      if (match >= 0)
         break;
     }
   }
-  if (offset < 0)
+  if (match < 0)
     return fallback;
-  const half = Math.floor(windowBytes / 2);
-  const start = Math.max(0, offset - half);
-  const end = Math.min(text.length, offset + half);
-  const snippet = text.slice(start, end).replace(/\s+/gu, " ").trim();
-  return `${start > 0 ? "…" : ""}${snippet}${end < text.length ? "…" : ""}`;
+  let offset = 0, normalizedOffset = 0;
+  for (const character of display) {
+    const width = character.toLocaleLowerCase("en-US").length;
+    if (normalizedOffset + width > match)
+      break;
+    normalizedOffset += width;
+    offset += character.length;
+  }
+  const edge = windowBytes >= 6 ? "…" : "";
+  const budget = windowBytes - 2 * publishUtf8Bytes(edge);
+  const previous = (end2) => {
+    const last = display.charCodeAt(end2 - 1);
+    return end2 > 1 && last >= 56320 && last <= 57343 && display.charCodeAt(end2 - 2) >= 55296 && display.charCodeAt(end2 - 2) <= 56319 ? end2 - 2 : end2 - 1;
+  };
+  let start = offset, end = offset, used = 0;
+  while (start > 0) {
+    const before = previous(start), bytes = publishUtf8Bytes(display.slice(before, start));
+    if (used + bytes > Math.floor(budget / 2))
+      break;
+    start = before;
+    used += bytes;
+  }
+  while (end < display.length) {
+    const character = String.fromCodePoint(display.codePointAt(end) ?? 0), bytes = publishUtf8Bytes(character);
+    if (used + bytes > budget)
+      break;
+    end += character.length;
+    used += bytes;
+  }
+  while (start > 0) {
+    const before = previous(start), bytes = publishUtf8Bytes(display.slice(before, start));
+    if (used + bytes > budget)
+      break;
+    start = before;
+    used += bytes;
+  }
+  const snippet = display.slice(start, end).replace(/\s+/gu, " ").trim();
+  return `${start > 0 ? edge : ""}${snippet}${end < display.length ? edge : ""}`;
 }
 
 // src/publish-theme.ts
@@ -924,7 +960,7 @@ function resultHref(base, slug) {
 }
 function snippetFor(index, doc, query) {
   const hydrated = index.notes.get(doc.s);
-  const text = hydrated?.text ?? doc.x;
+  const text = hydrated?.text;
   return text === undefined ? doc.p : publishSnippet(text, query, doc.p);
 }
 function appendMarked(parent, text, terms) {

package/dist/publish-search.js

@@ -23,7 +23,7 @@ import {
   publishSnippet,
   publishUtf8Bytes,
   scorePublishDocument
-} from "./index-3agn8scn.js";
+} from "./index-23m6bbjt.js";
 import"./index-z1w83f81.js";
 export {
   scorePublishDocument,

package/dist/publish.js

@@ -10,14 +10,14 @@ import {
   publishVault,
   renderPublishReportText,
   serializeSiteFile
-} from "./index-hmw17zaa.js";
+} from "./index-pnt9wr59.js";
 import"./index-233z9wmn.js";
 import"./index-d13v9ckt.js";
 import"./index-48pz4jpc.js";
 import"./index-06c9ctr6.js";
 import"./index-4knsp9qj.js";
 import"./index-66pshdtx.js";
-import"./index-3agn8scn.js";
+import"./index-23m6bbjt.js";
 import"./index-5vwpzb5a.js";
 import"./index-x3fthpsc.js";
 import"./index-qbssx940.js";

package/dist/{support-p452p9c4.js → support-m3t917ev.js}

@@ -2,7 +2,7 @@
 import {
   parseArguments as parseArguments2,
   parseUrlMetadataArguments
-} from "./index-bw2r37z9.js";
+} from "./index-9h7zh0wt.js";
 import"./index-bcknqxrq.js";
 import {
   parsePdfArguments
@@ -42,14 +42,14 @@ import"./index-tf4wzjew.js";
 import"./index-1gwbassd.js";
 import"./index-vnybgywh.js";
 import"./index-1xxnjn0d.js";
-import"./index-hmw17zaa.js";
+import"./index-pnt9wr59.js";
 import"./index-233z9wmn.js";
 import"./index-d13v9ckt.js";
 import"./index-48pz4jpc.js";
 import"./index-06c9ctr6.js";
 import"./index-4knsp9qj.js";
 import"./index-66pshdtx.js";
-import"./index-3agn8scn.js";
+import"./index-23m6bbjt.js";
 import"./index-hya40gb2.js";
 import"./index-5vwpzb5a.js";
 import"./index-x3fthpsc.js";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hraness/wordcell",
-  "version": "0.22.2",
+  "version": "0.22.4",
   "description": "A local Markdown knowledge base with superpowers. Save decisions, sources, and plans as files you own, then recover them through exact and semantic search, typed relationships, backlinks, and Git history.",
   "license": "MIT",
   "type": "module",
@@ -479,8 +479,8 @@
     "check": "bun run check:worker && bun run check:effect && bun run check:workflows && bun run check:portfolio-inventory && bun run check:installed-command-docs && bun run typecheck && bun run build && bun run test:package && bun run test && bun run url-metadata:tool:check",
     "prepack": "bun run check",
     "check:effect": "bun run scripts/check-effect-policy.ts",
-    "kb:search": "bunx --bun --package https://github.com/hraness/wordcell/releases/download/v0.22.1/hraness-wordcell-0.22.1.tgz wordcell search --root kb --mode exact --rerank typesafe --rerank-limit 25",
-    "kb:search:local": "bunx --bun --package https://github.com/hraness/wordcell/releases/download/v0.22.1/hraness-wordcell-0.22.1.tgz wordcell search --root kb --mode exact",
+    "kb:search": "bunx --bun --package https://github.com/hraness/wordcell/releases/download/v0.22.3/hraness-wordcell-0.22.3.tgz wordcell search --root kb --mode exact --rerank typesafe --rerank-limit 25",
+    "kb:search:local": "bunx --bun --package https://github.com/hraness/wordcell/releases/download/v0.22.3/hraness-wordcell-0.22.3.tgz wordcell search --root kb --mode exact",
     "check:worker": "tsc -p worker/tsconfig.json --noEmit"
   },
   "dependencies": {

package/skills/wordcell/SKILL.md

@@ -56,7 +56,7 @@ missing:
 ```sh
 command -v wordcell >/dev/null 2>&1 || {
   command -v bun >/dev/null 2>&1 || exit 1
-  bun add --global --ignore-scripts https://github.com/hraness/wordcell/releases/download/v0.22.2/hraness-wordcell-0.22.2.tgz
+  bun add --global --ignore-scripts https://github.com/hraness/wordcell/releases/download/v0.22.4/hraness-wordcell-0.22.4.tgz
 }
 wordcell --help
 ```

package/src/publish-index.ts

@@ -1,6 +1,7 @@
 import { canonicalJson } from "@hraness/oh";
 
 import type { Note } from "./graph.js";
+import { projectMarkdownText } from "./publish-markdown.js";
 import {
   WORDCELL_SITE_DOCS_FORMAT_V1,
   WORDCELL_SITE_LIMITS_V1,
@@ -38,8 +39,8 @@ function bounded(value: string, maximumBytes: number): { text: string; truncated
 }
 
 function previewText(note: Note): string {
-  const basis = note.summary !== "" ? note.summary : note.searchableText;
-  const collapsed = basis.replace(/\s+/gu, " ").trim();
+  const display = projectMarkdownText(note.content, note.properties["description"]);
+  const collapsed = display.preview.replace(/\s+/gu, " ").trim();
   return bounded(collapsed, WORDCELL_SITE_LIMITS_V1.docPreviewBytes).text;
 }