@hraness/sponge 0.3.1 → 0.3.2

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

@@ -7,6 +7,8 @@ export interface TextQuote {
     start: number;
     end: number;
 }
+/** Spoken text without transcript cues, whitespace collapsed: what a reader quotes from a transcript. */
+export declare function stripCues(raw: string): string;
 export declare function foldText(value: string): string;
 /** Every occurrence of `quote` in `source`, compared after folding. */
 export declare function locateAll(source: string, quote: string): {

package/dist/src/annotations/quote.js

@@ -1,9 +1,20 @@
 const context = 32;
 const graphemes = new Intl.Segmenter(undefined, { granularity: 'grapheme' });
 const escapable = /^[\\`*_{}[\]()#+\-.!|>~<"']$/;
+/** A transcript cue at a line start: `- [00:02] ` (clipped captions) or `[00:00:02] ` (Sponge transcripts). */
+const cue = /^[ \t]*(?:[-*+][ \t]+)?\[(?:\d{1,2}:)?\d{1,2}:\d{2}\][ \t]+/;
+/** Spoken text without transcript cues, whitespace collapsed: what a reader quotes from a transcript. */
+export function stripCues(raw) {
+    return raw
+        .split('\n')
+        .map((line) => line.replace(cue, ''))
+        .join('\n')
+        .replace(/\s+/g, ' ')
+        .trim();
+}
 /**
  * Fold the differences a reader cannot see: Markdown escapes, curly quotes, dashes, runs of
- * whitespace and Unicode normalization. Folding works on grapheme clusters, each NFC-normalized;
+ * whitespace, line-start transcript cues and Unicode normalization. Folding works on grapheme clusters, each NFC-normalized;
  * every folded code unit of a cluster maps to the cluster's raw bounds, so `map[i]` and `ends[i]`
  * bound the raw text of folded unit i and matches map back exactly.
  */
@@ -19,6 +30,17 @@ function fold(raw) {
     };
     for (let c = 0; c < clusters.length; c++) {
         const { segment, index } = clusters[c];
+        // A transcript cue reads as the space between caption lines, so a quote of the speech spans them.
+        const marker = (index === 0 || raw[index - 1] === '\n') && cue.exec(raw.slice(index, index + 32));
+        if (marker) {
+            const stop = index + marker[0].length;
+            if (!space && text.length)
+                emit(' ', index, stop);
+            space = true;
+            while (c + 1 < clusters.length && clusters[c + 1].index < stop)
+                c++;
+            continue;
+        }
         // An escaped character maps to its backslash, so a match starting there keeps the escape.
         let ch = segment, end = index + segment.length;
         if (segment === '\\' && c + 1 < clusters.length && escapable.test(clusters[c + 1].segment)) {
@@ -98,6 +120,12 @@ function rendered(raw) {
     for (let i = 0; i < raw.length; i++) {
         let ch = raw[i], at = i;
         if (lineStart) {
+            const c = cue.exec(raw.slice(i, i + 32));
+            if (c) {
+                i += c[0].length - 1;
+                lineStart = false;
+                continue;
+            }
             const m = /^[ \t]*(?:>[ \t]?|#{1,6}[ \t]+|[-*+][ \t]+|\d{1,3}[.)][ \t]+)*/.exec(raw.slice(i, i + 64));
             if (m[0].length) {
                 i += m[0].length - 1;

package/dist/src/cli/commands.js

@@ -791,5 +791,5 @@ export const usageOf = (name) => command(name)
     .join('\n');
 export function commandHelp(name) {
     const c = command(name), flags = Object.keys(c.flags);
-    return `sponge ${c.name} — ${c.summary}\n\nUsage:\n${c.usage.map(row).join('').replace(/^  /gm, '  sponge ')}${flags.length ? `\nFlags: ${flags.map((f) => '--' + f).join(' ')}\n` : ''}\nGlobal flags: --library /absolute/path --json --debug\n`;
+    return `sponge ${c.name} — ${c.summary}\n\nUsage:\n${c.usage.map(([synopsis, description]) => row([`sponge ${synopsis}`, description])).join('')}${flags.length ? `\nFlags: ${flags.map((f) => '--' + f).join(' ')}\n` : ''}\nGlobal flags: --library /absolute/path --json --debug\n`;
 }

package/dist/src/digest/pipeline.js

@@ -1,4 +1,4 @@
-import { foldText, locateAll, locateLoose, words } from '../annotations/quote.js';
+import { foldText, locateAll, locateLoose, stripCues, words } from '../annotations/quote.js';
 import { checkDigest, digestLimits as L, forms, parseDigest, } from './schema.js';
 import { SpongeError } from '../errors.js';
 import { sha256 } from '../util/hash.js';
@@ -97,7 +97,9 @@ function lastSentenceEnd(window) {
         best = m.index + m[0].length;
     return best;
 }
-const display = (p) => `[${p.id}] ${collapse(p.text)}`;
+const display = (p) => `[${p.id}] ${stripCues(p.text)}`;
+/** A raw source span as a quote: transcript cues inside it are dropped, so the quote reads as the speech. */
+const spoken = (raw) => (raw.includes('\n') && stripCues(raw) !== collapse(raw) ? stripCues(raw) : raw);
 /** Pack passages into sections of at most `budget` display characters, preferring heading boundaries past half a budget. */
 export function sectionsOf(list, budget) {
     const out = [];
@@ -295,7 +297,7 @@ export function snapQuote(source, text, within) {
     const folded = locateAll(source, text);
     if (folded.length) {
         const m = pick(folded);
-        return { text: m ? source.slice(m.start, m.end) : text, match: 'folded' };
+        return { text: m ? spoken(source.slice(m.start, m.end)) : text, match: 'folded' };
     }
     const loose = (within && pick(locateLoose(source, text, within))) || pick(locateLoose(source, text));
     if (!loose)
@@ -306,7 +308,7 @@ export function snapQuote(source, text, within) {
         start--;
     while (end < source.length && /[*_`]/.test(source[end]))
         end++;
-    return { text: source.slice(start, end), match: 'snapped' };
+    return { text: spoken(source.slice(start, end)), match: 'snapped' };
 }
 /**
  * A located quote over the word limit keeps its leading words: a prefix of a source span is still verbatim. Trailing

package/dist/src/errors.d.ts

@@ -16,6 +16,8 @@ declare const parents: {
     readonly 'not-digest': "unsupported";
     readonly malformed: "invalid-input";
     readonly 'no-capture': "invalid-input";
+    readonly 'parent-missing': "invalid-input";
+    readonly 'target-exists': "invalid-input";
 };
 export type DetailCode = keyof typeof parents;
 export type SpongeCode = ErrorCode | DetailCode;

package/dist/src/errors.js

@@ -30,6 +30,8 @@ const parents = {
     'not-digest': 'unsupported',
     malformed: 'invalid-input',
     'no-capture': 'invalid-input',
+    'parent-missing': 'invalid-input',
+    'target-exists': 'invalid-input',
 };
 export const indexCodes = [
     'invalid-index',

package/dist/src/portability/files.js

@@ -112,8 +112,18 @@ export function copyVerified(source, target, record) {
 export function newDirectory(path) {
     if (resolve(path) !== path)
         throw new SpongeError('invalid-input');
-    safeDirectory(dirname(path));
-    mkdirSync(path, { mode: 0o700 });
+    try {
+        safeDirectory(dirname(path));
+        mkdirSync(path, { mode: 0o700 });
+    }
+    catch (error) {
+        const code = error.code;
+        if (code === 'ENOENT')
+            throw new SpongeError('parent-missing', { cause: error });
+        if (code === 'EEXIST')
+            throw new SpongeError('target-exists', { cause: error });
+        throw error;
+    }
     const stat = lstatSync(path);
     return { root: path, device: stat.dev, inode: stat.ino };
 }

package/native/runtime-manifest.json

@@ -27,14 +27,14 @@
     }
   ],
   "build": {
-    "asrSourceSha256": "94b9303afcf9304e695c37411cef065e9cc45a31e0e51cd171819cbfc9ee3955",
+    "asrSourceSha256": "8bd77ebffabae786902304eacb983fc950f70c981fb1538696517d85052c9407",
     "clangVersion": "Apple clang version 21.0.0 (clang-2100.1.1.101)",
-    "embeddingSourceSha256": "eed7e110bc888af99e0d6e614275cea7a7dbaa56f88d3de9bdd04cc9ff4cfa45",
+    "embeddingSourceSha256": "2b0cd2b9c52c9dee578a3be50a443d528d5204d47f3f9ba750f6a1f37a191e78",
     "swiftVersion": "Apple Swift version 6.3.3 (swiftlang-6.3.3.1.3 clang-2100.1.1.101)"
   },
   "minimumMacOS": "26.0",
-  "packageVersion": "0.3.1",
+  "packageVersion": "0.3.2",
   "platform": "darwin",
   "schemaVersion": 1,
-  "sourceCommit": "b8975aff1dc06e398c3f78811e76f4ff875729a1"
+  "sourceCommit": "e9eb0613365859c5809a05e8ad909504ac6e8d66"
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hraness/sponge",
-  "version": "0.3.1",
+  "version": "0.3.2",
   "description": "A local library for captured articles and verified offline assets.",
   "repository": {
     "type": "git",