test: deployment + API-route smoke scripts (deterministic PDF, two-parser)

scripts/deploy_smoke.py

@@ -0,0 +1,78 @@
+"""Deployment smoke test against the LIVE HF endpoint (deterministic pipeline).
+
+Waits for the Space to be up, POSTs a real JD + the default résumé to /api/generate
+(v1), decodes the returned PDF, and parses it with BOTH pdftotext and pdfplumber.
+No LLM dependency — the deterministic pipeline must return a real PDF.
+"""
+import base64
+import os
+import re
+import sys
+import time
+
+import requests
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+BASE = "https://saitejatirunagari-jaa-ats-tool.hf.space"
+OUT = os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))),
+                   "data", "acceptance", "out")
+os.makedirs(OUT, exist_ok=True)
+
+
+def _wait_up(tries=30, delay=15):
+    for i in range(tries):
+        try:
+            r = requests.get(f"{BASE}/api/health", timeout=20)
+            if r.status_code == 200:
+                print(f"[health] up after {i*delay}s: {r.text[:120]}")
+                return True
+        except Exception as e:
+            print(f"[health] try {i}: {str(e)[:60]}")
+        time.sleep(delay)
+    return False
+
+
+def main():
+    print("Waiting for deployment (HF rebuild)…")
+    if not _wait_up():
+        print("DEPLOY SMOKE: endpoint never healthy"); return
+    jd = re.sub(r"SOURCE_URL:.*", "",
+                open(os.path.join(os.path.dirname(OUT), "jd1_stripe_payments.txt"),
+                     encoding="utf-8").read(), count=1).strip()
+    data = {"jd_text": jd, "job_title": "Product Manager, Payments",
+            "company": "Stripe", "version": "v1"}
+    print("POST /api/generate …")
+    try:
+        r = requests.post(f"{BASE}/api/generate", data=data, timeout=180)
+    except Exception as e:
+        print("DEPLOY SMOKE: request failed:", str(e)[:200]); return
+    print("HTTP", r.status_code)
+    try:
+        j = r.json()
+    except Exception:
+        print("non-JSON response:", r.text[:300]); return
+    print("keys:", sorted(list(j.keys()))[:20])
+    print("status:", j.get("status"), "| source:", j.get("source"),
+          "| pct:", j.get("external_coverage_pct") or j.get("pct"))
+    pdfb = j.get("pdf_b64")
+    if not pdfb:
+        print("DEPLOY SMOKE: no pdf_b64. error:", j.get("error"), j.get("detail"))
+        return
+    path = os.path.join(OUT, "deploy_smoke.pdf")
+    with open(path, "wb") as f:
+        f.write(base64.b64decode(pdfb))
+    print("saved:", path, os.path.getsize(path), "bytes")
+    # two-parser parse of the DOWNLOADED PDF
+    from src.pdf_validate import _extract_pdf_text, _pdftotext_extract
+    a = _extract_pdf_text(path) or ""
+    b = _pdftotext_extract(path) or ""
+    print(f"two-parser: pdfplumber={len(a)}ch pdftotext={len(b)}ch")
+    for kw in ["stakeholder", "roadmap", "cross-functional", "product"]:
+        print(f"  {kw}: pdfplumber={kw in a.lower()} pdftotext={kw in b.lower()}")
+    print("DEPLOY SMOKE: PASSED (endpoint returned a real, two-parser-readable PDF)"
+          if a and b else "DEPLOY SMOKE: PDF not parseable by both")
+
+
+if __name__ == "__main__":
+    main()