Create analyzer.py

analyzer.py

@@ -0,0 +1,65 @@
+import traceback
+from loguru import logger
+from .analysis_helpers import analyze_url, analyze_text
+from .orchestrator import ORCH
+
+def on_analyze(news_text: str = "", news_url: str = "", image_url: str = "", run_serp: bool = False):
+    """
+    Analyze either a text or a URL (mutually exclusive).
+    """
+    try:
+        if news_url and news_text:
+            return {"error": "Please provide either a URL or text — not both."}, "", [], {}
+
+        if news_url:
+            article_text, headline, qa_fallback_note = analyze_url(news_url, run_serp)
+            url = news_url
+        elif news_text:
+            article_text, headline, qa_fallback_note = analyze_text(news_text)
+            url = None
+        else:
+            return {"error": "No input provided."}, "", [], {}
+
+        claim = ""
+        report = ORCH.run(
+            claim_text=claim,
+            article_text=article_text,
+            url=url,
+            image_url=image_url or None,
+            run_serpapi=run_serp,
+        )
+
+        extracted_claims = [r.get("claim") for r in report.get("reports", [])]
+        qa_text = ""
+        if report.get("reports"):
+            qa_text = report["reports"][0].get("qa_summary", "") or qa_fallback_note
+
+        phishing_tag = extract_phishing_tag(report)
+        if report.get("reports"):
+            report["reports"][0]["phishing_tag"] = phishing_tag
+
+        phish = report.get("reports", [{}])[0].get("phishing_analysis", {}) or {}
+        phish["phishing_tag"] = phishing_tag
+
+        return report, qa_text, extracted_claims, phish
+
+    except Exception:
+        logger.exception("on_analyze failed")
+        return {"error": traceback.format_exc()}, "", [], {}
+
+
+def extract_phishing_tag(report):
+    summary_phish_flag = report.get("summary", {}).get("phishing_flag")
+    if summary_phish_flag is True:
+        return "Unsafe"
+    elif summary_phish_flag is False:
+        return "Safe"
+
+    first_phish = report.get("reports", [{}])[0].get("phishing_analysis", {}) or {}
+    sb = (first_phish.get("safe_browsing") or {})
+    vt = (first_phish.get("virustotal") or {})
+    if sb.get("safe") is False or vt.get("safe") is False:
+        return "Unsafe"
+    elif sb.get("safe") is True and vt.get("safe") is True:
+        return "Safe"
+    return "Unknown"