@yegor256/dogent 0.12.2 → 0.12.4

package/package.json

@@ -40,7 +40,7 @@
     "lint": "eslint .",
     "test": "mocha 'test/**/*.js' --timeout 60000"
   },
-  "version": "0.12.2",
+  "version": "0.12.4",
   "dependencies": {
     "minimist": "^1.2.8",
     "pretty-ms": "^7.0.1"

package/src/dogent.js

@@ -73,11 +73,17 @@ const scan = () => {
   return [];
 };
 const scanned = scan();
-scanned.forEach((file) => process.stderr.write(`Scanning ${file}\n`));
+const bodies = new Map(scanned.map((file) => [file, fs.readFileSync(file, 'utf8')]));
+scanned.forEach((file) => {
+  const body = bodies.get(file);
+  const lines = body === '' ? 0 : body.split('\n').length - (body.endsWith('\n') ? 1 : 0);
+  const bytes = Buffer.byteLength(body);
+  process.stderr.write(`Scanning ${file} (${lines} lines, ${bytes} bytes)\n`);
+});
 const checks = rules();
 process.stderr.write(`${scanned.length} files scanned, ${checks.length} rules applied\n`);
 const documents = scanned.map(
-  (file) => new Markdown(file, fs.readFileSync(file, 'utf8')).document()
+  (file) => new Markdown(file, bodies.get(file)).document()
 );
 const started = Date.now();
 const suppressed = args.suppress();

package/src/rules/description-triggers.js

@@ -16,6 +16,9 @@ const Region = require('../region');
  * short or that never names a trigger with the word "when". Its prompt
  * hands the deeper judgement to the AI oracle, which weighs whether the
  * description truly names the situations and phrases that activate it.
+ * A quoted example phrase is the strongest form of that naming, so a
+ * deterministic guard then vetoes any oracle flag on a description that
+ * holds both "when" and a quoted phrase, however the model rules.
  */
 class DescriptionTriggers {
   constructor() {
@@ -26,7 +29,7 @@ class DescriptionTriggers {
     return 'Name the concrete situations and user phrases that should activate the skill in its description, so the loader knows exactly when to invoke it.';
   }
   prompt() {
-    return `${this.id}: in a SKILL.md, flag a description that is too short or fails to name the concrete situations and user phrases that should activate the skill, even when it contains the word "when"`;
+    return `${this.id}: in a SKILL.md, flag a description that is too short or fails to name the concrete situations and user phrases that should activate the skill, even when it contains the word "when"; a description that quotes an example user phrase, such as "file this bug", already names a trigger in its strongest form and must never be flagged`;
   }
   violations(document) {
     const uri = document.uri();
@@ -56,6 +59,30 @@ class DescriptionTriggers {
     }
     return [];
   }
+  suppress(violation, document) {
+    if (violation.rule !== this.id) {
+      return false;
+    }
+    const value = this.description(document);
+    return /\bwhen\b/iu.test(value) && this.quoted(value);
+  }
+  description(document) {
+    const pairs = document.walk({
+      header: () => [],
+      prose: () => [],
+      snippet: () => [],
+      bullets: () => [],
+      frontmatter: (keys) => keys
+    });
+    const found = pairs.filter((pair) => pair.key === 'description');
+    if (found.length === 0) {
+      return '';
+    }
+    return found[0].value.trim();
+  }
+  quoted(value) {
+    return /["'‘’“”][^"'‘’“”]+["'‘’“”]/u.test(value);
+  }
   flag(message, row, uri) {
     return new Violation(this.id, 'warning', message, new Region(uri, row, 1));
   }

package/src/rules/description-voice.js

@@ -20,7 +20,11 @@ const Region = require('../region');
  * Distinct from description-triggers, which checks that a "when" clause
  * exists, and from description-length, which checks the size; this one
  * checks the grammatical voice. Its prompt hands subtler voice
- * judgement to the AI oracle.
+ * judgement to the AI oracle. A deterministic guard then drops any
+ * oracle flag that lands off the frontmatter description row, or on that
+ * row when the deterministic pronoun check finds nothing, so the oracle
+ * stays bound to the same single line and vocabulary the standalone
+ * checker enforces.
  */
 class DescriptionVoice {
   constructor() {
@@ -38,6 +42,23 @@ class DescriptionVoice {
     if (uri.replace(/^.*\//u, '') !== 'SKILL.md') {
       return [];
     }
+    const pair = this.description(document);
+    if (pair === null) {
+      return [];
+    }
+    return this.judge(pair, uri);
+  }
+  suppress(violation, document) {
+    if (violation.rule !== this.id) {
+      return false;
+    }
+    const pair = this.description(document);
+    if (pair === null || violation.spot.line() !== pair.row) {
+      return true;
+    }
+    return this.judge(pair, document.uri()).length === 0;
+  }
+  description(document) {
     const pairs = document.walk({
       header: () => [],
       prose: () => [],
@@ -46,10 +67,7 @@ class DescriptionVoice {
       frontmatter: (keys) => keys
     });
     const found = pairs.filter((pair) => pair.key === 'description');
-    if (found.length === 0) {
-      return [];
-    }
-    return this.judge(found[0], uri);
+    return found.length === 0 ? null : found[0];
   }
   judge(pair, uri) {
     const text = pair.value.replace(/use when.*$/isu, '');

package/src/rules/no-articles.js

@@ -35,6 +35,21 @@ class NoArticles {
       frontmatter: () => []
     });
   }
+  suppress(violation, document) {
+    if (violation.rule !== this.id) {
+      return false;
+    }
+    return this.headers(document).has(violation.spot.line());
+  }
+  headers(document) {
+    return new Set(document.walk({
+      header: (text, line) => [line],
+      prose: () => [],
+      snippet: () => [],
+      bullets: () => [],
+      frontmatter: () => []
+    }));
+  }
   scan(text, line, uri) {
     const found = [];
     const masked = mask(text);

package/src/rules/untrusted.js

@@ -44,7 +44,7 @@ class Untrusted {
   }
   scan(text, line, uri) {
     const masked = mask(text);
-    const verb = /\b(?:read|fetch|open|follow|execute)\b/iu;
+    const verb = /(?<!-)\b(?:read|fetch|open|follow|execute)\b(?!-)/iu;
     const source = /\b(?:page|url|link|email|file|issue|output|comment)\b/iu;
     const guard = /\b(?:as data|do not follow|treat as untrusted|inside delimiters|untrusted)\b/iu;
     if (!verb.test(masked) || !source.test(masked) || guard.test(masked)) {

package/src/version.js

@@ -9,8 +9,8 @@
  * Version.
  *
  * The current release of dogent, replaced on every release by rultor.
- * The default `0.12.2` marks an unreleased build straight from source.
+ * The default `0.12.4` marks an unreleased build straight from source.
  */
-const version = '0.12.2';
+const version = '0.12.4';
 
 module.exports = version;