@pranavraut033/ats-checker 1.1.1 → 1.2.0

package/README.md

@@ -19,6 +19,7 @@ Zero-dependency TypeScript library that scores a resume against a job descriptio
 - **Explainable** — breakdown by category (skills / experience / keywords / education) plus matched and missing skill/keyword lists
 - **Configurable** — adjust weights, add skill aliases, define custom penalty rules
 - **Zero dependencies** — core library has no runtime deps; ships ESM + CJS
+- **PDF input** — optional `/pdf` subpath extracts resume text from a PDF buffer (requires `pdfjs-dist` peer dep)
 - **Built-in profiles** — software engineer, data scientist, product manager out of the box
 
 ---
@@ -184,6 +185,41 @@ const result = analyzeResume({
 
 ---
 
+## PDF Input
+
+Extract text from a PDF resume before passing it to `analyzeResume`. This uses `pdfjs-dist` as an optional peer dependency — the core library stays zero-dep.
+
+```bash
+npm install pdfjs-dist
+```
+
+```typescript
+import { extractTextFromPDF } from "@pranavraut033/ats-checker/pdf";
+import { analyzeResume } from "@pranavraut033/ats-checker";
+import { readFileSync } from "fs";
+
+const bytes = readFileSync("resume.pdf");
+const resumeText = await extractTextFromPDF(bytes);
+
+const result = analyzeResume({ resumeText, jobDescription: "..." });
+```
+
+`extractTextFromPDF` accepts a `Uint8Array` or `ArrayBuffer` and returns a plain `string`. Works in Node.js and the browser (text-layer PDFs only).
+
+**Multi-column layouts are handled automatically.** The extractor uses glyph x/y coordinates to detect column boundaries and process each column independently, so a two-column resume parses cleanly without interleaved text.
+
+For PDFs that can't be recovered — scanned/image resumes or exports with no text layer — `analyzeResume` surfaces an actionable message in `result.warnings`. Always check it after PDF input:
+
+```typescript
+const result = analyzeResume({ resumeText, jobDescription: "..." });
+if (result.warnings.length) {
+  console.warn("Parsing issues:", result.warnings);
+  // e.g. "Almost no text was extracted — the resume may be a scanned/image PDF."
+}
+```
+
+---
+
 ## LLM Integration (deprecated)
 
 `analyzeResumeAsync` accepts an optional `llm` config that rewrites suggestion text via a caller-supplied LLM client. **This path is deprecated** — scores and breakdowns are never touched by LLM. Prefer calling `analyzeResume` and running your own LLM pass on `result.suggestions` if you want AI-enhanced wording.

package/dist/index.js

@@ -541,6 +541,16 @@ function parseResume(resumeText, config) {
   }
   const requiredSections = ["summary", "experience", "skills", "education"];
   const warnings = [];
+  const lineCount = splitLines(resumeText).length;
+  if (resumeText.trim().length < 100) {
+    warnings.push(
+      "Almost no text was extracted \u2014 the resume may be a scanned/image PDF. Upload a text-based PDF or paste the text directly."
+    );
+  } else if (lineCount <= 2) {
+    warnings.push(
+      "Resume text has no line breaks \u2014 the PDF layout likely didn't export cleanly (common with multi-column designs). Export as a single-column PDF or paste plain text for accurate parsing."
+    );
+  }
   for (const section of requiredSections) {
     if (!detected.includes(section)) {
       warnings.push(`${section} section not detected`);
@@ -920,6 +930,11 @@ var SuggestionEngine = class {
         "Strengthen bullet points with impact verbs (led, built, improved, delivered)."
       );
     }
+    if (input.resume.detectedSections.length < 2 && input.resume.raw.trim().length > 300) {
+      suggestions.push(
+        "Your resume may use a multi-column layout. Export as a single-column PDF or paste plain text \u2014 most ATS systems and this parser work best with a linear layout."
+      );
+    }
     return { suggestions, warnings };
   }
 };