resuml 1.11.0 → 1.12.1

package/dist/index.cjs

@@ -481,6 +481,7 @@ var init_en = __esm({
       ],
       pronouns: ["i", "me", "my", "mine", "myself", "we", "our", "ours"],
       stopWords: [
+        // Articles & determiners
         "a",
         "an",
         "the",
@@ -551,7 +552,281 @@ var init_en = __esm({
         "such",
         "than",
         "too",
-        "very"
+        "very",
+        // Pronouns & possessives (also checked by pronoun check, but filter from JD keywords)
+        "you",
+        "your",
+        "yours",
+        "yourself",
+        "we",
+        "our",
+        "ours",
+        "ourselves",
+        "they",
+        "them",
+        "their",
+        "theirs",
+        "he",
+        "she",
+        "his",
+        "her",
+        "hers",
+        "who",
+        "whom",
+        "whose",
+        "which",
+        "what",
+        "where",
+        "when",
+        "how",
+        "why",
+        // Common JD filler words (not meaningful for skill matching)
+        "able",
+        "also",
+        "across",
+        "already",
+        "always",
+        "among",
+        "any",
+        "apply",
+        "become",
+        "believe",
+        "best",
+        "bring",
+        "change",
+        "come",
+        "committed",
+        "company",
+        "comfortable",
+        "critical",
+        "current",
+        "day",
+        "desired",
+        "either",
+        "end",
+        "ensure",
+        "environment",
+        "equal",
+        "even",
+        "excellent",
+        "exciting",
+        "exceptional",
+        "expected",
+        "experience",
+        "fast",
+        "field",
+        "find",
+        "first",
+        "focused",
+        "follow",
+        "get",
+        "give",
+        "go",
+        "going",
+        "good",
+        "great",
+        "group",
+        "grow",
+        "growing",
+        "growth",
+        "help",
+        "here",
+        "high",
+        "highly",
+        "ideal",
+        "impact",
+        "important",
+        "include",
+        "includes",
+        "including",
+        "industry",
+        "interested",
+        "job",
+        "join",
+        "just",
+        "keep",
+        "key",
+        "know",
+        "large",
+        "latest",
+        "lead",
+        "level",
+        "like",
+        "location",
+        "long",
+        "look",
+        "looking",
+        "love",
+        "make",
+        "many",
+        "much",
+        "must",
+        "need",
+        "new",
+        "next",
+        "offer",
+        "one",
+        "only",
+        "open",
+        "opportunity",
+        "order",
+        "others",
+        "own",
+        "pace",
+        "part",
+        "partner",
+        "passionate",
+        "people",
+        "per",
+        "play",
+        "plus",
+        "position",
+        "preferred",
+        "provide",
+        "put",
+        "qualifications",
+        "quickly",
+        "range",
+        "related",
+        "required",
+        "requirements",
+        "requirement",
+        "responsible",
+        "responsibilities",
+        "responsibility",
+        "result",
+        "right",
+        "role",
+        "run",
+        "same",
+        "see",
+        "seek",
+        "seeking",
+        "set",
+        "several",
+        "since",
+        "skills",
+        "someone",
+        "start",
+        "state",
+        "still",
+        "strong",
+        "success",
+        "successful",
+        "support",
+        "sure",
+        "take",
+        "team",
+        "then",
+        "there",
+        "thing",
+        "think",
+        "through",
+        "time",
+        "together",
+        "top",
+        "truly",
+        "try",
+        "two",
+        "type",
+        "use",
+        "used",
+        "using",
+        "value",
+        "want",
+        "way",
+        "well",
+        "while",
+        "within",
+        "without",
+        "work",
+        "working",
+        "world",
+        "would",
+        "year",
+        "years",
+        // Section headers & structural words (not technical skills)
+        "description",
+        "overview",
+        "summary",
+        "duties",
+        "bachelor",
+        "bachelors",
+        "master",
+        "masters",
+        "degree",
+        "phd",
+        "minimum",
+        "preferred",
+        "implement",
+        "process",
+        "robust",
+        "consistent",
+        "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;
@@ -1122,8 +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") {
@@ -1166,20 +1453,163 @@ function buildTfMap(tokens) {
   }
   return tf;
 }
+function splitJdSections(text) {
+  const lines = text.split("\n");
+  const reqPatterns = /^(required|requirements?|minimum|preferred|qualifications?|must[\s-]have|nice[\s-]to[\s-]have|what you.?ll|what we.?re looking|skills|technical|you.?ll need|responsibilities)/i;
+  let inReqSection = false;
+  const reqLines = [];
+  const otherLines = [];
+  for (const line of lines) {
+    const trimmed = line.trim();
+    if (reqPatterns.test(trimmed.replace(/[:#*-]/g, "").trim())) {
+      inReqSection = true;
+    } else if (/^(about|summary|who we are|our (company|team|mission)|description|overview|benefits|perks|compensation|salary)/i.test(trimmed.replace(/[:#*-]/g, "").trim())) {
+      inReqSection = false;
+    }
+    if (inReqSection) {
+      reqLines.push(line);
+    } else {
+      otherLines.push(line);
+    }
+  }
+  return {
+    requirementText: reqLines.join("\n"),
+    otherText: otherLines.join("\n")
+  };
+}
+function extractCompoundTerms(text) {
+  const patterns = [
+    /\b(machine\s+learning)\b/gi,
+    /\b(deep\s+learning)\b/gi,
+    /\b(computer\s+vision)\b/gi,
+    /\b(natural\s+language\s+processing)\b/gi,
+    /\b(data\s+pipelines?)\b/gi,
+    /\b(data\s+models?)\b/gi,
+    /\b(data\s+engineering)\b/gi,
+    /\b(data\s+structures?)\b/gi,
+    /\b(data\s+quality)\b/gi,
+    /\b(data\s+flows?)\b/gi,
+    /\b(data\s+orchestration)\b/gi,
+    /\b(data\s+warehou\w+)\b/gi,
+    /\b(synthetic\s+data)\b/gi,
+    /\b(ci\s*\/?\s*cd)\b/gi,
+    /\b(rest\s+api)\b/gi,
+    /\b(open\s+source)\b/gi,
+    /\b(human[\s-]+in[\s-]+the[\s-]+loop)\b/gi,
+    /\b(self[\s-]+service)\b/gi,
+    /\b(agentic\s+workflows?)\b/gi,
+    /\b(distributed\s+systems?)\b/gi,
+    /\b(cloud\s+infrastructure)\b/gi,
+    /\b(micro\s*services?)\b/gi,
+    /\b(full[\s-]+stack)\b/gi,
+    /\b(front[\s-]*end)\b/gi,
+    /\b(back[\s-]*end)\b/gi,
+    /\b(sql\s*\/?\s*nosql)\b/gi
+  ];
+  const found = [];
+  for (const pattern of patterns) {
+    const matches = text.matchAll(pattern);
+    for (const m of matches) {
+      const term = m[1]?.toLowerCase().replace(/\s+/g, " ").trim();
+      if (term && !found.includes(term)) {
+        found.push(term);
+      }
+    }
+  }
+  return found;
+}
+function extractBrandNames(text) {
+  const brands = /* @__PURE__ */ new Set();
+  const brandPatterns = [
+    /\bat\s+([A-Z][a-zA-Z]+(?:\s+[A-Z][a-zA-Z]+)*)/g,
+    /(?:^|\.\s+)([A-Z][a-zA-Z]+(?:\s+[A-Z][a-zA-Z]+)*)\s+(?:is|are|was|has|Inc|Corp|Ltd|GmbH)/g,
+    /\b(?:join(?:ing)?|about)\s+([A-Z][a-zA-Z]+)/g
+  ];
+  for (const pattern of brandPatterns) {
+    const matches = text.matchAll(pattern);
+    for (const m of matches) {
+      const name = m[1]?.toLowerCase();
+      if (name) {
+        for (const word of name.split(/\s+/)) {
+          if (word.length > 2) brands.add(word);
+        }
+      }
+    }
+  }
+  const knownBrands = [
+    "apple",
+    "google",
+    "meta",
+    "facebook",
+    "amazon",
+    "microsoft",
+    "netflix",
+    "uber",
+    "airbnb",
+    "twitter",
+    "linkedin",
+    "spotify",
+    "stripe",
+    "shopify",
+    "iphone",
+    "ipad",
+    "mac",
+    "macbook",
+    "airpods",
+    "android",
+    "windows",
+    "alexa",
+    "siri",
+    "cortana",
+    "gmail",
+    "chrome",
+    "safari",
+    "firefox"
+  ];
+  for (const b of knownBrands) brands.add(b);
+  return brands;
+}
 function extractKeywords(text, language, maxKeywords = 30) {
   const langData = getLanguageData(language);
   const stopWords = new Set(langData.stopWords);
-  const tokens = tokenize(text, stopWords);
-  const stemmed = tokens.map((t) => simpleStem(t, language));
+  const cleanText = stripNoise(text);
+  const compoundTerms = extractCompoundTerms(cleanText);
+  const brandNames = extractBrandNames(text);
+  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();
-  for (let i = 0; i < tokens.length; i++) {
+  for (let i = 0; i < allTokens.length; i++) {
     const stem = stemmed[i] ?? "";
     if (!stemToOriginal.has(stem)) {
-      stemToOriginal.set(stem, tokens[i] ?? "");
+      stemToOriginal.set(stem, allTokens[i] ?? "");
     }
   }
-  return [...tf.entries()].filter(([stem]) => stem.length > 2).sort((a, b) => b[1] - a[1]).slice(0, maxKeywords).map(([stem]) => stemToOriginal.get(stem) || stem);
+  const compoundsFlat = compoundTerms.join(" ").split(/\s+/);
+  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) && !allTokens.includes(original)) {
+      return false;
+    }
+    return true;
+  }).sort((a, b) => b[1] - a[1]).slice(0, maxKeywords - compoundTerms.length).map(([stem]) => stemToOriginal.get(stem) || stem);
+  const seen = /* @__PURE__ */ new Set();
+  const keywords = [];
+  for (const term of [...compoundTerms, ...singleKeywords]) {
+    if (!seen.has(term)) {
+      seen.add(term);
+      keywords.push(term);
+    }
+  }
+  return keywords.slice(0, maxKeywords);
 }
 function matchJobDescription(resume, jobDescription, language = "en") {
   const langData = getLanguageData(language);
@@ -1192,11 +1622,24 @@ function matchJobDescription(resume, jobDescription, language = "en") {
   const matched = [];
   const missing = [];
   for (const keyword of jdKeywords) {
-    const stem = simpleStem(keyword, language);
-    if (resumeStems.has(stem) || resumeTokenSet.has(keyword.toLowerCase())) {
-      matched.push(keyword);
+    if (keyword.includes(" ")) {
+      const parts = keyword.split(/\s+/);
+      const allPartsMatch = parts.every((part) => {
+        const stem = simpleStem(part, language);
+        return resumeStems.has(stem) || resumeTokenSet.has(part.toLowerCase());
+      });
+      if (allPartsMatch) {
+        matched.push(keyword);
+      } else {
+        missing.push(keyword);
+      }
     } else {
-      missing.push(keyword);
+      const stem = simpleStem(keyword, language);
+      if (resumeStems.has(stem) || resumeTokenSet.has(keyword.toLowerCase())) {
+        matched.push(keyword);
+      } else {
+        missing.push(keyword);
+      }
     }
   }
   const matchPercentage = jdKeywords.length > 0 ? Math.round(matched.length / jdKeywords.length * 100) : 0;