@hraness/sponge 0.3.4 → 0.3.5

package/dist/src/annotations/quote.d.ts

@@ -20,6 +20,7 @@ export declare function textQuote(source: string, start: number, end: number): T
 export declare function locateQuote(source: string, quote: string): TextQuote | undefined;
 /** A stored selector still names its exact text in `source`. */
 export declare function anchorHolds(source: string, anchor: TextQuote): boolean;
+/** Word count: space-separated tokens, with each run of Han or kana counted as half its characters, rounded up. */
 export declare function words(value: string): number;
 /**
  * Every occurrence of `quote` in `source` ignoring inline Markdown and case, within optional raw bounds.

package/dist/src/annotations/quote.js

@@ -107,9 +107,19 @@ export function anchorHolds(source, anchor) {
         anchor.end > anchor.start &&
         source.slice(anchor.start, anchor.end) === anchor.exact);
 }
+/** Han and kana are written without spaces; about two characters make a word. */
+const unspaced = /[\p{Script=Han}\p{Script=Hiragana}\p{Script=Katakana}]/gu;
+/** Word count: space-separated tokens, with each run of Han or kana counted as half its characters, rounded up. */
 export function words(value) {
     const t = foldText(value);
-    return t ? t.split(' ').length : 0;
+    if (!t)
+        return 0;
+    let n = 0;
+    for (const token of t.split(' ')) {
+        const cjk = token.match(unspaced)?.length ?? 0;
+        n += cjk ? Math.ceil(cjk / 2) + (token.replace(unspaced, '').match(/[\p{L}\p{N}]/u) ? 1 : 0) : 1;
+    }
+    return n;
 }
 /**
  * Rendered-text projection for snapping a near-verbatim quote: drops inline Markdown syntax (link targets, brackets,

package/dist/src/digest/pipeline.js

@@ -317,10 +317,13 @@ export function snapQuote(source, text, within) {
 export function trimQuote(text, max = L.quoteWords) {
     if (words(text) <= max)
         return text;
-    const ends = [...text.matchAll(/\S+/g)].map((m) => m.index + m[0].length);
-    let n = Math.min(max, ends.length), out = text;
+    // Cut after a space-separated token or, inside unspaced Han and kana, after any character.
+    const ends = [
+        ...new Set([...text.matchAll(/\S+/g), ...text.matchAll(/[\p{Script=Han}\p{Script=Hiragana}\p{Script=Katakana}]/gu)].map((m) => m.index + m[0].length)),
+    ].sort((a, b) => a - b);
+    let n = ends.length, out = text;
     do {
-        out = text.slice(0, ends[n - 1]).replace(/[\s,;:(\[\u2014\u2013-]+$/, '');
+        out = text.slice(0, ends[n - 1]).replace(/[\s,;:(\[\u2014\u2013\u3001\uff0c-]+$/, '');
         n--;
     } while (n > 0 && words(out) > max);
     return out;

package/dist/src/digest/schema.js

@@ -49,8 +49,12 @@ const keys = (v, allowed) => Object.keys(v).every((k) => allowed.includes(k));
 const str = (v, max = 2000) => typeof v === 'string' && v.trim().length > 0 && v.length <= max && !/[\u0000-\u0008\u000b-\u001f\u007f]/.test(v);
 /** v1: sentence terminators followed by more text; abbreviations are rare enough in a 155-character lead to reject. */
 const sentencesV1 = (v) => foldText(v).split(/(?<=[.!?])\s+(?=\S)/).length;
-/** v2: a terminator (and closing quotes) followed by a capital or digit, the same split Jungle's first-sentence lead uses, so `e.g. this` stays one sentence. */
-const sentences = (v) => foldText(v).split(/(?<=[.!?]["')]*)\s+(?=["(]?[A-Z0-9])/).length;
+/**
+ * v2: a terminator (and closing quotes) followed by a capital or digit, the split Jungle's first-sentence lead uses, so
+ * `e.g. this` stays one sentence. A period after an initial or a common title does not end a sentence: `Lucas C.
+ * Cordeiro`, `F. Zakaria`, `U.S. Senate`, `Dr. Lind`.
+ */
+const sentences = (v) => foldText(v).split(/(?<=(?:(?<!\b(?:[A-Z]|Mr|Mrs|Ms|Dr|Prof|St|Jr|Sr|vs|No|Inc|Ltd|Co))\.|[!?])["')]*)\s+(?=["(]?[A-Z0-9])/).length;
 /** Parse an untrusted digest (v1 or v2) into its shape, or report why not. */
 export function parseDigest(value) {
     const issues = [], bad = (path, code = 'shape') => {

package/native/runtime-manifest.json

@@ -27,14 +27,14 @@
     }
   ],
   "build": {
-    "asrSourceSha256": "ca98fb111b0d2b26ca697a6c0d55d76f7d2b5b752bff8fc053ae19389d16a52c",
+    "asrSourceSha256": "3b32718d4626aff97f2c3e043c9fd020f237adc7f7e5508e7d7b99a555762982",
     "clangVersion": "Apple clang version 21.0.0 (clang-2100.1.1.101)",
-    "embeddingSourceSha256": "af6d6c01934d64427f2a9cb36f83edaea372f6bf2e522166b9e86e3b15f72784",
+    "embeddingSourceSha256": "66f008ea29050d114b0df74337124bac8f6cdba9732e257cc70ad85eb9ebee6f",
     "swiftVersion": "Apple Swift version 6.3.3 (swiftlang-6.3.3.1.3 clang-2100.1.1.101)"
   },
   "minimumMacOS": "26.0",
-  "packageVersion": "0.3.4",
+  "packageVersion": "0.3.5",
   "platform": "darwin",
   "schemaVersion": 1,
-  "sourceCommit": "4e42216e86da07224ec81512c2efa0a710e38500"
+  "sourceCommit": "49b3f7f100cdbcbf5f52418d28c53dc13f198424"
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hraness/sponge",
-  "version": "0.3.4",
+  "version": "0.3.5",
   "description": "A local library for captured articles and verified offline assets.",
   "repository": {
     "type": "git",