@lokascript/semantic 1.0.0

package/dist/languages/ar.d.ts

@@ -0,0 +1,78 @@
+import { B as BaseTokenizer, T as TokenStream, a as TokenKind, L as LanguageProfile } from '../types-C4dcj53L.js';
+
+/**
+ * Arabic Tokenizer
+ *
+ * Tokenizes Arabic hyperscript input.
+ * Arabic is challenging because:
+ * - Right-to-left (RTL) text direction
+ * - Prefix prepositions that attach to words (بـ, لـ, كـ)
+ * - Root-pattern morphology
+ * - CSS selectors are LTR islands within RTL text
+ */
+
+declare class ArabicTokenizer extends BaseTokenizer {
+    readonly language = "ar";
+    readonly direction: "rtl";
+    constructor();
+    tokenize(input: string): TokenStream;
+    classifyToken(token: string): TokenKind;
+    /**
+     * Try to match an Arabic preposition.
+     * Attaches prepositionValue metadata for disambiguation in pattern matching.
+     */
+    private tryPreposition;
+    /**
+     * Try to extract a proclitic (conjunction or preposition) that's attached to the following word.
+     *
+     * Arabic proclitics attach directly to words without space:
+     * - والنقر → و + النقر (and + the-click)
+     * - فالتبديل → ف + التبديل (then + the-toggle)
+     * - بالنقر → ب + النقر (with + the-click)
+     * - ولالنقر → و + ل + النقر (and + to + the-click)
+     *
+     * This enables:
+     * - Polysyndetic coordination: A وB وC
+     * - Attached prepositions: بالنقر (with-the-click)
+     * - Multi-proclitic sequences: ولالنقر (and-to-the-click)
+     *
+     * Returns null if:
+     * - Not a proclitic character/sequence
+     * - Proclitic is standalone (followed by space)
+     * - Remaining word is too short (< 2 chars, to avoid false positives)
+     * - Full word is a recognized keyword (e.g., بدل should NOT be split to ب + دل)
+     *
+     * @see NATIVE_REVIEW_NEEDED.md for implementation rationale
+     */
+    private tryProclitic;
+    /**
+     * Extract an Arabic word.
+     * Uses morphological normalization to handle prefix/suffix variations.
+     * Attaches metadata for temporal markers (formality, confidence).
+     */
+    private extractArabicWord;
+    /**
+     * Extract an ASCII word.
+     */
+    private extractAsciiWord;
+    /**
+     * Extract a number, including Arabic time unit suffixes.
+     * Arabic allows space between number and unit.
+     */
+    private extractArabicNumber;
+}
+/**
+ * Singleton instance.
+ */
+declare const arabicTokenizer: ArabicTokenizer;
+
+/**
+ * Arabic Language Profile
+ *
+ * VSO word order, prepositions, RTL (right-to-left), space-separated.
+ * Features root-based morphology and rich verb conjugation.
+ */
+
+declare const arabicProfile: LanguageProfile;
+
+export { arabicProfile, arabicTokenizer };