Add PII redaction report and audit log improvements

Author: marcoderoni

audit_log.jsonl

@@ -1 +1,2 @@
 {"timestamp": "2026-04-21T23:35:47.632246", "contract": "Lic_Online_TOMA_v062223_CH_CA_ENG.pdf", "overall_score": "RED", "missing_clauses": ["Gross Negligence / Wilful Misconduct Carve-out"], "governing_law": "Swiss law without reference to its confli", "notice_period": "Not detected", "hash": "f2eb3a487f9aa3945ea19a4ed68986e358eb0e37b1d1a0c00eb4b5c9aa60b2ce"}
+{"timestamp": "2026-04-21T23:41:05.008016", "contract": "Lic_Online_TOMA_v062223_CH_CA_ENG.pdf", "overall_score": "RED", "missing_clauses": ["Gross Negligence / Wilful Misconduct Carve-out"], "governing_law": "Swiss law without reference to its confli", "notice_period": "Not detected", "hash": "80c23387048a6a995ed11b26876bb2f7a8c65987a617a0835ddf04345ca7263f"}

main.py

@@ -43,12 +43,12 @@ def scan_contract(path: str, rules: dict) -> dict:
     print(f"   → {len(clauses)} sezioni rilevate")
 
     # Sanitize PII
+    pii_mapping = {}
     try:
-        from scanner.sanitizer import sanitize, desanitize
+        from scanner.sanitizer import sanitize
         text, pii_mapping = sanitize(text)
         print(f"   → PII sanitizzato: {len(pii_mapping)} entità redatte")
     except Exception as e:
-        pii_mapping = {}
         print(f"   → PII sanitization skipped: {e}")
 
     print("   → Estrazione metadati...")
@@ -57,8 +57,14 @@ def scan_contract(path: str, rules: dict) -> dict:
     print("   → Analisi clausole...")
     analysis = analyze(text, clauses, rules)
 
+    # Build PII summary
+    pii_summary = {"total_entities": len(pii_mapping), "breakdown": {}}
+    for placeholder in pii_mapping.keys():
+        entity_type = placeholder.split("_")[0].replace("[", "")
+        pii_summary["breakdown"][entity_type] = pii_summary["breakdown"].get(entity_type, 0) + 1
+
     print("   → Generazione report Word...")
-    report_path = generate_report(name, metadata, analysis, OUTPUT_DIR)
+    report_path = generate_report(name, metadata, analysis, OUTPUT_DIR, pii_summary=pii_summary)
 
     # Audit log
     try:

scanner/reporter.py

@@ -1,4 +1,4 @@
-# Copyright (c) 2025 Marco De Roni. All rights reserved.
+# Copyright (c) 2026 Marco De Roni. All rights reserved.
 # Licensed under the MIT License — see LICENSE file for details.
 
 import os
@@ -46,7 +46,8 @@ def generate_report(
     contract_name: str,
     metadata: dict,
     analysis: dict,
-    output_dir: str
+    output_dir: str,
+    pii_summary: dict = None
 ) -> str:
     """Genera un report Word professionale."""
 
@@ -68,6 +69,30 @@ def generate_report(
 
     doc.add_paragraph()
 
+    # --- PII Redaction Summary ---
+    if pii_summary and pii_summary.get("total_entities", 0) > 0:
+        add_heading(doc, "Privacy & PII Redaction Summary", level=1)
+        total = pii_summary.get("total_entities", 0)
+        breakdown = pii_summary.get("breakdown", {})
+
+        doc.add_paragraph(
+            f"Before analysis, {total} sensitive entities were automatically "
+            f"redacted and restored in this report."
+        )
+
+        if breakdown:
+            table = doc.add_table(rows=1, cols=2)
+            table.style = "Table Grid"
+            table.rows[0].cells[0].text = "Entity Type"
+            table.rows[0].cells[1].text = "Count"
+            for entity_type, count in sorted(breakdown.items()):
+                row = table.add_row().cells
+                row[0].text = entity_type
+                row[1].text = str(count)
+
+        doc.add_paragraph()
+
+    # --- Overall Risk ---
     overall = analysis.get("overall_score", "UNKNOWN")
     add_heading(doc, "Overall Risk Assessment", level=1)
     add_colored_paragraph(
@@ -77,6 +102,7 @@ def generate_report(
     )
     doc.add_paragraph()
 
+    # --- Metadata ---
     add_heading(doc, "Contract Metadata", level=1)
     table = doc.add_table(rows=1, cols=2)
     table.style = "Table Grid"
@@ -101,6 +127,7 @@ def generate_report(
 
     doc.add_paragraph()
 
+    # --- Missing Clauses ---
     add_heading(doc, "Missing Clauses", level=1)
     missing = analysis.get("missing_clauses", [])
     if missing:
@@ -114,6 +141,7 @@ def generate_report(
 
     doc.add_paragraph()
 
+    # --- Clause Risk Analysis ---
     add_heading(doc, "Clause Risk Analysis", level=1)
     findings = analysis.get("findings", {})