@lokascript/semantic 1.0.0 → 1.1.0

package/dist/languages/qu.js

@@ -929,6 +929,7 @@ var QUECHUA_EXTRAS = [
   { native: "tayta", normalized: "parent" },
   // Events
   { native: "llikllay", normalized: "click" },
+  { native: "\xF1itiy", normalized: "click" },
   { native: "click", normalized: "click" },
   { native: "yaykuy", normalized: "input" },
   { native: "llave uray", normalized: "keydown" },
@@ -1002,8 +1003,15 @@ var QuechuaTokenizer = class extends BaseTokenizer {
         }
         const selectorToken = this.trySelector(input, pos);
         if (selectorToken) {
-          tokens.push(selectorToken);
-          pos = selectorToken.position.end;
+          const selectorWithSuffix = this.splitSelectorSuffix(selectorToken);
+          if (selectorWithSuffix.length === 2) {
+            tokens.push(selectorWithSuffix[0]);
+            tokens.push(selectorWithSuffix[1]);
+            pos = selectorWithSuffix[1].position.end;
+          } else {
+            tokens.push(selectorToken);
+            pos = selectorToken.position.end;
+          }
           continue;
         }
       }
@@ -1046,6 +1054,12 @@ var QuechuaTokenizer = class extends BaseTokenizer {
         }
       }
       if (isQuechuaLetter(input[pos])) {
+        const multiWordToken = this.tryMultiWordKeyword(input, pos);
+        if (multiWordToken) {
+          tokens.push(multiWordToken);
+          pos = multiWordToken.position.end;
+          continue;
+        }
         const wordToken = this.extractWord(input, pos);
         if (wordToken) {
           tokens.push(wordToken);
@@ -1084,6 +1098,67 @@ var QuechuaTokenizer = class extends BaseTokenizer {
     }
     return null;
   }
+  /**
+   * Split a selector token if it has a Quechua suffix attached.
+   * E.g., ".openta" -> [".open", "-ta"]
+   * Returns array with 1 token (no suffix) or 2 tokens (selector + suffix)
+   */
+  splitSelectorSuffix(selectorToken) {
+    const text = selectorToken.value;
+    for (const suffix of SUFFIXES) {
+      if (text.toLowerCase().endsWith(suffix)) {
+        const baseEnd = text.length - suffix.length;
+        const base = text.slice(0, baseEnd);
+        const suffixPart = text.slice(baseEnd);
+        const baseToken = createToken(
+          base,
+          "selector",
+          createPosition(selectorToken.position.start, selectorToken.position.start + baseEnd)
+        );
+        const suffixToken = createToken(
+          suffixPart,
+          "particle",
+          createPosition(selectorToken.position.start + baseEnd, selectorToken.position.end)
+        );
+        return [baseToken, suffixToken];
+      }
+    }
+    return [selectorToken];
+  }
+  /**
+   * Try to match multi-word keywords that should be treated as a single unit.
+   * E.g., "mana qhawachiy" (not focus = blur)
+   */
+  tryMultiWordKeyword(input, pos) {
+    const multiWordKeywords = [
+      { pattern: "mana qhawachiy", normalized: "blur" },
+      { pattern: "mana qhaway", normalized: "blur" },
+      { pattern: "mana riqsisqa", normalized: "undefined" },
+      { pattern: "mana waqtalla", normalized: "async" },
+      { pattern: "\xF1awpaq kaq", normalized: "previous" },
+      { pattern: "aswan qayllaqa", normalized: "closest" },
+      { pattern: "llave uray", normalized: "keydown" },
+      { pattern: "llave hawa", normalized: "keyup" },
+      { pattern: "mausiri yayku", normalized: "mouseover" },
+      { pattern: "mausiri lluqsi", normalized: "mouseout" },
+      { pattern: "waranqa sikundu", normalized: "ms" }
+    ];
+    const inputLower = input.toLowerCase();
+    for (const { pattern, normalized } of multiWordKeywords) {
+      if (inputLower.slice(pos, pos + pattern.length) === pattern) {
+        const endPos = pos + pattern.length;
+        if (endPos >= input.length || isWhitespace(input[endPos]) || !isQuechuaLetter(input[endPos])) {
+          return createToken(
+            input.slice(pos, endPos),
+            "keyword",
+            createPosition(pos, endPos),
+            normalized
+          );
+        }
+      }
+    }
+    return null;
+  }
   extractWord(input, startPos) {
     let pos = startPos;
     let word = "";