resuml 1.12.0 → 1.12.1

package/dist/index.cjs

@@ -762,7 +762,71 @@ var init_en = __esm({
         "process",
         "robust",
         "consistent",
-        "operations"
+        "operations",
+        // URL/email/domain fragments
+        "http",
+        "https",
+        "www",
+        "com",
+        "org",
+        "net",
+        "mailto",
+        // Resume/YAML schema field names (in case raw YAML is pasted)
+        "name",
+        "keywords",
+        "highlights",
+        "startdate",
+        "enddate",
+        "website",
+        "profiles",
+        "basics",
+        "position",
+        "institution",
+        "studytype",
+        "fluency",
+        "issuer",
+        "network",
+        "username",
+        "countrycode",
+        "region",
+        // Generic nouns that aren't skills
+        "product",
+        "company",
+        "service",
+        "services",
+        "platform",
+        "solutions",
+        "ability",
+        "opportunity",
+        "candidate",
+        "applicant",
+        "position",
+        "salary",
+        "compensation",
+        "benefits",
+        "perks",
+        "bonus",
+        "development",
+        "management",
+        "knowledge",
+        "modern",
+        "advanced",
+        "practices",
+        "nice",
+        "technologies",
+        "technology",
+        "frameworks",
+        "framework",
+        "tools",
+        "data",
+        "based",
+        "contribute",
+        "contributions",
+        "migration",
+        "leading",
+        "source",
+        "visit",
+        "join"
       ]
     };
     en_default = en;
@@ -1333,8 +1397,20 @@ var init_genericChecks = __esm({
 });
 
 // src/ats/jdMatcher.ts
+function stripNoise(text) {
+  return text.replace(/https?:\/\/[^\s]+/gi, " ").replace(/www\.[^\s]+/gi, " ").replace(/[\w.+-]+@[\w.-]+\.[a-z]{2,}/gi, " ").replace(/(?:^|\s)\/[\w/.-]+/g, " ").replace(/\b[a-z]+[A-Z][a-zA-Z]*\b/g, (match2) => {
+    return match2.replace(/([a-z])([A-Z])/g, "$1 $2");
+  });
+}
 function tokenize(text, stopWords) {
-  return text.toLowerCase().replace(/[^a-zA-Z0-9äöüßÄÖÜàáâãéèêëíìîïóòôõúùûüñç\s/+-]/g, " ").split(/\s+/).filter((word) => word.length > 2 && !stopWords.has(word));
+  return text.toLowerCase().replace(/[^a-zA-Z0-9äöüßÄÖÜàáâãéèêëíìîïóòôõúùûüñç\s/+-]/g, " ").split(/\s+/).filter((word) => {
+    if (word.length <= 2) return false;
+    if (stopWords.has(word)) return false;
+    if (word.startsWith("//") || word.startsWith("http")) return false;
+    if (/^\d+$/.test(word)) return false;
+    if (/^[/+-]+$/.test(word)) return false;
+    return true;
+  });
 }
 function simpleStem(word, language) {
   if (language === "de") {
@@ -1496,12 +1572,17 @@ function extractBrandNames(text) {
 function extractKeywords(text, language, maxKeywords = 30) {
   const langData = getLanguageData(language);
   const stopWords = new Set(langData.stopWords);
-  const compoundTerms = extractCompoundTerms(text);
+  const cleanText = stripNoise(text);
+  const compoundTerms = extractCompoundTerms(cleanText);
   const brandNames = extractBrandNames(text);
-  const { requirementText, otherText } = splitJdSections(text);
-  const reqTokens = tokenize(requirementText, stopWords).filter((t) => !brandNames.has(t));
-  const otherTokens = tokenize(otherText, stopWords).filter((t) => !brandNames.has(t));
-  const allTokens = [...reqTokens, ...reqTokens, ...reqTokens, ...otherTokens];
+  const { requirementText } = splitJdSections(cleanText);
+  const hasRequirementSections = requirementText.trim().length > 0;
+  let allTokens;
+  if (hasRequirementSections) {
+    allTokens = tokenize(requirementText, stopWords).filter((t) => !brandNames.has(t));
+  } else {
+    allTokens = tokenize(cleanText, stopWords).filter((t) => !brandNames.has(t));
+  }
   const stemmed = allTokens.map((t) => simpleStem(t, language));
   const tf = buildTfMap(stemmed);
   const stemToOriginal = /* @__PURE__ */ new Map();
@@ -1515,7 +1596,7 @@ function extractKeywords(text, language, maxKeywords = 30) {
   const compoundWordSet = new Set(compoundsFlat);
   const singleKeywords = [...tf.entries()].filter(([stem]) => stem.length > 2).filter(([stem]) => {
     const original = stemToOriginal.get(stem) || stem;
-    if (compoundWordSet.has(original) && !reqTokens.includes(original)) {
+    if (compoundWordSet.has(original) && !allTokens.includes(original)) {
       return false;
     }
     return true;