@pranavraut033/ats-checker 1.3.0 → 1.3.2

package/README.md

@@ -52,9 +52,12 @@ import { analyzeResume } from "@pranavraut033/ats-checker";
 const result = analyzeResume({
   resumeText: `
     Software Engineer with 5 years of experience.
-    Skills: JavaScript, TypeScript, React, Node.js, SQL
-    Experience: Senior Engineer at ExampleCorp (Jan 2020 - Present)
-    Education: B.S. Computer Science
+    Skills
+    JavaScript, TypeScript, React, Node.js, SQL
+    Experience
+    Senior Engineer at ExampleCorp (Jan 2020 - Present)
+    Education
+    B.S. Computer Science
   `,
   jobDescription: `
     Frontend engineer role. Must have React, TypeScript, accessibility best practices.
@@ -63,11 +66,11 @@ const result = analyzeResume({
   config: { referenceDate: "2026-01-01" }, // freeze clock for reproducible scores
 });
 
-console.log(result.score);            // e.g. 72.45
+console.log(result.score);            // 44.44
 console.log(result.matchedSkills);    // ["javascript", "node", "react", "typescript"]
-console.log(result.missingSkills);    // ["accessibility best practices", "graphql"]
+console.log(result.missingSkills);    // ["accessibility", "frontend", "graphql"]
 console.log(result.experienceGap);    // 0 (requirement met)
-console.log(result.suggestions);      // ["Add GraphQL to your skills section", ...]
+console.log(result.suggestions);      // ["Highlight these required skills: accessibility, frontend, graphql", ...]
 ```
 
 ---
@@ -102,6 +105,10 @@ console.log(result.suggestions);      // ["Add GraphQL to your skills section",
 
 The `keywords` sub-score is a **weighted** coverage ratio, not a flat count: each JD keyword gets a weight from its location (required > preferred > body text) and frequency, so missing a required keyword drops the score more than missing one mentioned once in the body.
 
+> **Caveat — malformed/copy-pasted JD text:** required/preferred detection scans each line for literal trigger phrases (`required`, `must`, `nice to have`, `preferred`). Job postings copy-pasted from a wrapped/columned source sometimes split words across line breaks (e.g. `"Nice to\n\nhaveExperience..."`), which breaks these phrases across two lines and silently drops them into the unweighted body-keyword bucket instead of required/preferred. Skill keywords themselves (e.g. `react`, `python/fastapi`) are still picked up via the whole-text token scan and unaffected. If a JD looks oddly broken, paste it through a plain-text cleanup pass first, or expect required/preferred weighting to under-count.
+
+The `education` sub-score normalizes degree abbreviations on both sides to a canonical level (`bachelor`, `master`, `phd`, `mba`, `associate`) before comparing — so a resume listing "B.S. Computer Science" satisfies a JD requiring "Bachelor's degree".
+
 ---
 
 ## Configuration
@@ -288,6 +295,23 @@ if (result.warnings.length) {
 }
 ```
 
+### OCR fallback for scanned PDFs
+
+`extractTextFromPDF` accepts an optional `ocrFallback` that's only invoked when the text layer comes back too short (default threshold: 100 chars). The OCR engine and its dependency are entirely your choice — the core library never bundles one:
+
+```typescript
+const resumeText = await extractTextFromPDF(bytes, {
+  ocrFallback: async (data) => {
+    // bring your own OCR engine, e.g. tesseract.js or a cloud OCR API
+    const { recognize } = await import("tesseract.js");
+    const { data: { text } } = await recognize(data, "eng");
+    return text;
+  },
+});
+```
+
+If `ocrFallback` throws or returns text that isn't longer than the text-layer result, `extractTextFromPDF` silently keeps the original result — OCR failures never break the deterministic extraction path.
+
 ---
 
 ## LLM Integration (deprecated)

package/dist/index.cjs

@@ -282,15 +282,38 @@ var LEVEL_RANK = {
   fluent: 5,
   native: 6,
   "native speaker": 6,
-  bilingual: 6
+  bilingual: 6,
+  // German
+  grundkenntnisse: 1,
+  gering: 2,
+  gut: 3,
+  fortgeschritten: 4,
+  flie\u00DFend: 5,
+  muttersprache: 6,
+  muttersprachler: 6,
+  // French
+  "d\xE9butant": 1,
+  "\xE9l\xE9mentaire": 1,
+  "limit\xE9": 2,
+  "interm\xE9diaire": 3,
+  "avanc\xE9": 4,
+  courant: 5,
+  natif: 6,
+  "langue maternelle": 6,
+  bilingue: 6
 };
 var LANGUAGE_GROUP = KNOWN_LANGUAGES.join("|");
 var LEVEL_GROUP = Object.keys(LEVEL_RANK).sort((a, b) => b.length - a.length).map((l) => l.replace(/\s+/g, "\\s+")).join("|");
+var BOUNDARY_START = "(?:^|(?<=[^a-z\xE0-\xFF]))";
+var BOUNDARY_END = "(?:$|(?=[^a-z\xE0-\xFF]))";
 var LANGUAGE_LEVEL_RE = new RegExp(
-  `\\b(${LANGUAGE_GROUP})\\b(?:\\s*[\\(:\\-]?\\s*(${LEVEL_GROUP}|[abc][12]))?`,
+  `\\b(${LANGUAGE_GROUP})\\b(?:\\s*[\\(:\\-]?\\s*(${BOUNDARY_START}(?:${LEVEL_GROUP})${BOUNDARY_END}|[abc][12]))?`,
+  "gi"
+);
+var LEVEL_BEFORE_LANGUAGE_RE = new RegExp(
+  `${BOUNDARY_START}(${LEVEL_GROUP})${BOUNDARY_END}\\s+(?:in\\s+)?(${LANGUAGE_GROUP})\\b`,
   "gi"
 );
-var LEVEL_BEFORE_LANGUAGE_RE = new RegExp(`\\b(${LEVEL_GROUP})\\s+(?:in\\s+)?(${LANGUAGE_GROUP})\\b`, "gi");
 function canonicalLanguage(name) {
   const lower = name.toLowerCase();
   return LANGUAGE_ALIASES[lower] ?? lower;
@@ -340,9 +363,9 @@ function diffLanguages(resumeLanguages, requiredLanguages) {
 
 // src/core/parser/jd.parser.ts
 var DEGREE_VARIANTS = [
-  [/\b(?:bachelor(?:'s)?|b\.s\.?|bs\.?|bsc\.?)\b/i, "bachelor"],
-  [/\b(?:master(?:'s)?|m\.s\.?|ms\.?|msc\.?)\b/i, "master"],
-  [/\b(?:phd|ph\.d\.?|doctorate)\b/i, "phd"],
+  [/\b(?:bachelor(?:'s)?|b\.s\.?|bs\.?|bsc\.?|licence)\b/i, "bachelor"],
+  [/\b(?:master(?:'s)?|m\.s\.?|ms\.?|msc\.?|diplom)\b/i, "master"],
+  [/\b(?:phd|ph\.d\.?|doctorate|doktor|doctorat)\b/i, "phd"],
   [/\bmba\b/i, "mba"],
   [/\bassociate(?:'s)?\b/i, "associate"]
 ];
@@ -365,16 +388,15 @@ function extractPreferredSkills(lines) {
   return preferred;
 }
 function extractRoleKeywords(text) {
-  const roleMatch = text.match(/(engineer|developer|manager|scientist|analyst|designer|architect)/i);
-  const titleTokens = roleMatch ? roleMatch[0].split(/\s+/) : [];
-  return unique(tokenize(titleTokens.join(" ") || text.split(/\n/)[0] || ""));
+  const roleMatches = text.match(/(engineer|developer|manager|scientist|analyst|designer|architect|director|consultant|lead|vp)/gi) ?? [];
+  const fallback = roleMatches.length === 0 ? [text.split(/\n/)[0] ?? ""] : [];
+  return unique(tokenize([...roleMatches, ...fallback].join(" ")));
 }
 function extractMinExperience(text) {
-  const match = text.match(/(\d{1,2})\+?\s+(?:years|yrs)/i);
-  if (match) {
-    return Number.parseInt(match[1], 10);
-  }
-  return void 0;
+  const match = text.match(/(\d{1,2})\+?\s*(?:years?|yrs\.?|jahre?|ans?|années?)/i);
+  if (!match) return void 0;
+  const parsed = Number.parseInt(match[1], 10);
+  return parsed <= 60 ? parsed : void 0;
 }
 var SURFACE_TOKEN_RE = /[a-z0-9][a-z0-9.#+\-/]*[a-z0-9#+]/gi;
 function collectKeywordSurfaceForms(rawText, aliases) {
@@ -388,7 +410,16 @@ function collectKeywordSurfaceForms(rawText, aliases) {
   }
   return surfaceForms;
 }
-function extractEducationRequirements(text) {
+var LANG_SECTION_RE = /^\s*(?:languages?|sprache|langue)s?\s*[:\-–—]?\s*/i;
+var LANG_REQUIREMENT_HINT_RE = /\b(fluent|required|must|need|speak|proficient|native|conversational|intermediate|advanced|professional|[abc][12])\b/i;
+function isLanguageRequired(lang, jobDescription) {
+  return splitLines(jobDescription).some((line) => {
+    const lower = line.toLowerCase();
+    if (!lower.includes(lang.name)) return false;
+    return LANG_SECTION_RE.test(line) || LANG_REQUIREMENT_HINT_RE.test(line);
+  });
+}
+function extractDegreeLevels(text) {
   const found = /* @__PURE__ */ new Set();
   for (const [pattern, canonical] of DEGREE_VARIANTS) {
     if (pattern.test(text)) found.add(canonical);
@@ -426,11 +457,13 @@ function parseJobDescription(jobDescription, config) {
     roleKeywords,
     keywords,
     minExperienceYears: extractMinExperience(jobDescription),
-    educationRequirements: extractEducationRequirements(jobDescription),
+    educationRequirements: extractDegreeLevels(jobDescription),
     keywordSurfaceForms: collectKeywordSurfaceForms(jobDescription, config.skillAliases),
-    // ponytail: any language mention in the JD is treated as a requirement — good enough until
-    // JDs that merely *reference* a language (not require it) show up as false positives.
-    requiredLanguages: parseLanguageMentions(jobDescription)
+    // A language only counts as required if its mention carries a requirement/level cue
+    // or sits in a "Languages:" line — plain references ("our Berlin office") don't count.
+    requiredLanguages: parseLanguageMentions(jobDescription).filter(
+      (lang) => isLanguageRequired(lang, jobDescription)
+    )
   };
 }
 
@@ -459,11 +492,37 @@ var MONTHS = {
   nov: 11,
   november: 11,
   dec: 12,
-  december: 12
+  december: 12,
+  // German
+  januar: 1,
+  j\u00E4nner: 1,
+  februar: 2,
+  m\u00E4rz: 3,
+  maerz: 3,
+  mai: 5,
+  juni: 6,
+  juli: 7,
+  oktober: 10,
+  dezember: 12,
+  // French
+  janvier: 1,
+  f\u00E9vrier: 2,
+  fevrier: 2,
+  mars: 3,
+  avril: 4,
+  juin: 6,
+  juillet: 7,
+  ao\u00FBt: 8,
+  aout: 8,
+  septembre: 9,
+  octobre: 10,
+  novembre: 11,
+  d\u00E9cembre: 12,
+  decembre: 12
 };
 function parseDateToken(raw) {
   const cleaned = raw.trim().toLowerCase();
-  const monthMatch = cleaned.match(/([a-z]{3,9})\s*(\d{4})/i);
+  const monthMatch = cleaned.match(/([a-zà-ÿ]{3,9})\s*(\d{4})/i);
   if (monthMatch) {
     const monthName = monthMatch[1].toLowerCase();
     const year = Number.parseInt(monthMatch[2], 10);
@@ -495,14 +554,14 @@ function monthsBetween(start, end) {
 function parseDateRange(text, referenceDate) {
   const normalized = text.trim();
   const rangeMatch = normalized.match(
-    /(\d{1,2}\/\d{4}|[A-Za-z]{3,9}\s+\d{4}|\d{4})\s*(?:-|to|–|—)\s*(Present|Current|Now|\d{1,2}\/\d{4}|[A-Za-z]{3,9}\s+\d{4}|\d{4})/i
+    /(\d{1,2}\/\d{4}|[A-Za-zà-ÿ]{3,9}\s+\d{4}|\d{4})\s*(?:-|to|through|until|bis|jusqu'à|à|–|—)\s*(Present|Current|Now|Aktuell|Heute|Actuellement|Présent|\d{1,2}\/\d{4}|[A-Za-zà-ÿ]{3,9}\s+\d{4}|\d{4})/i
   );
   if (!rangeMatch) {
     return null;
   }
   const startToken = parseDateToken(rangeMatch[1]);
   const endRaw = rangeMatch[2];
-  const isPresent = /present|current|now/i.test(endRaw);
+  const isPresent = /present|current|now|aktuell|heute|actuellement|présent|actuel/i.test(endRaw);
   const endToken = isPresent ? void 0 : parseDateToken(endRaw);
   if (!startToken) {
     return null;
@@ -556,12 +615,21 @@ function sumExperienceYears(ranges) {
 
 // src/core/parser/resume.parser.ts
 var SECTION_ALIASES = {
-  summary: ["summary", "profile", "about"],
-  experience: ["experience", "work experience", "professional experience", "employment"],
-  skills: ["skills", "technical skills", "technologies"],
-  education: ["education", "academics", "academic background"],
-  projects: ["projects", "portfolio"],
-  certifications: ["certifications", "licenses"]
+  summary: ["summary", "profile", "about", "zusammenfassung", "profil", "r\xE9sum\xE9", "\xE0 propos"],
+  experience: [
+    "experience",
+    "work experience",
+    "professional experience",
+    "employment",
+    "erfahrung",
+    "berufserfahrung",
+    "exp\xE9rience",
+    "exp\xE9rience professionnelle"
+  ],
+  skills: ["skills", "technical skills", "technologies", "f\xE4higkeiten", "kenntnisse", "comp\xE9tences"],
+  education: ["education", "academics", "academic background", "ausbildung", "formation", "\xE9tudes"],
+  projects: ["projects", "portfolio", "projekte", "projets"],
+  certifications: ["certifications", "licenses", "zertifizierungen", "certifications professionnelles"]
 };
 var STRONG_VERBS = [
   "led",
@@ -639,7 +707,9 @@ function extractSections(text) {
 }
 function parseSkills(sectionContent, aliases) {
   if (!sectionContent) return [];
-  const raw = sectionContent.split(/[,;\n]/).map((skill) => skill.trim()).filter(Boolean);
+  const hasBullets = /[•·‣▪○●◦]/.test(sectionContent);
+  const normalized = hasBullets ? sectionContent.replace(/\n/g, " ") : sectionContent;
+  const raw = normalized.split(/[,;\n]|[•·‣▪○●◦]/).map((skill) => skill.trim().replace(/^[-•·‣▪○●◦\s]+|[-•·‣▪○●◦\s]+$/g, "").trim()).filter(Boolean);
   return normalizeSkills(raw, aliases);
 }
 function parseActionVerbs(text) {
@@ -672,7 +742,7 @@ function parseExperience(sectionContent, referenceDate) {
       }
       continue;
     }
-    const titleMatch = line.match(/^(Senior|Lead|Principal|Staff|Software|Full\s*Stack|Frontend|Backend|Engineer|Developer|Manager|Analyst)[^,-]*/i);
+    const titleMatch = line.match(/^(Senior|Lead|Principal|Staff|VP|Director|Consultant|Architect|Software|Full\s*Stack|Frontend|Backend|Engineer|Developer|Manager|Analyst)[^,-]*/i);
     if (titleMatch) {
       const title = titleMatch[0].trim();
       jobTitles.push(title.toLowerCase());
@@ -710,7 +780,8 @@ function parseResume(resumeText, config) {
     const textToScan = sections.summary ?? normalizedText;
     const yearsMatch = textToScan.match(/(\d{1,2})\+?\s*years?/i);
     if (yearsMatch) {
-      totalExperienceYears = Number.parseInt(yearsMatch[1], 10);
+      const parsed = Number.parseInt(yearsMatch[1], 10);
+      totalExperienceYears = parsed <= 60 ? parsed : 0;
     }
   }
   const requiredSections = ["summary", "experience", "skills", "education"];
@@ -946,10 +1017,9 @@ function scoreEducation(resume, job) {
   if (job.educationRequirements.length === 0) {
     return 100;
   }
-  const resumeEducationText = resume.educationEntries.join(" ");
-  const normalizedEducation = resumeEducationText.toLowerCase();
+  const resumeDegreeLevels = extractDegreeLevels(resume.educationEntries.join(" "));
   const matched = job.educationRequirements.filter(
-    (requirement) => normalizedEducation.includes(requirement.toLowerCase())
+    (requirement) => resumeDegreeLevels.includes(requirement)
   );
   if (matched.length === 0) {
     return 0;