Update backup_features.py

backup_features.py

@@ -1,52 +1,54 @@
 # backup_features.py
-# Backup deterministic utilities (no ML, no API)
-
-def extract_key_terms(text):
-    terms = {}
-
-    # Parties
-    parties = []
-    for line in text.splitlines():
-        if "between" in line.lower() or "by and between" in line.lower():
-            parties.append(line.strip())
-    terms["parties"] = parties[:2]
-
-    # Dates
-    import re
-    date_pattern = r"\b(?:January|February|March|April|May|June|July|August|September|October|November|December)\b.*\d{4}"
-    terms["dates"] = re.findall(date_pattern, text)
-
-    # Money
-    money = re.findall(r"\$\s?\d[\d,]*", text)
-    terms["monetary"] = money[:5]
-
-    # Legal terms
-    legal_terms = []
-    for t in ["Confidential Information", "Term", "Exceptions", "Obligations", "Liability"]:
-        if t.lower() in text.lower():
-            legal_terms.append(t)
-    terms["terms"] = legal_terms
-
-    return terms
-
-
-def extract_clauses(text):
-    # simple clause splitting based on numbering
-    import re
-    clauses = re.split(r"\n\d+\.\s", text)
-    return [c.strip() for c in clauses if len(c.strip()) > 50]
-
-
-def top5_risks(text):
-    warnings = []
-    if "perpetual" in text.lower():
-        warnings.append("Perpetual obligations (no end date).")
-    if "injunction" in text.lower():
-        warnings.append("Possible injunctions for violations.")
-    if "assign" in text.lower():
-        warnings.append("Contract can be assigned without your consent.")
-    if "unlimited liability" in text.lower():
-        warnings.append("Unlimited liability exposure.")
-    if "indemnify" in text.lower():
-        warnings.append("Strong indemnity clauses may impose financial risk.")
-    return warnings[:5]
+# --------------------------------------------------------------
+# Light-weight rule-based feature implementations
+# (HuggingFace CPU-safe, no ML models)
+# --------------------------------------------------------------
+
+import re
+
+def named_entity_recognition(text):
+    parties = re.findall(r"\b[A-Z][a-z]+(?: [A-Z][a-z]+)*\b", text)
+    dates = re.findall(r"\b\d{4}\b", text)
+    amounts = re.findall(r"\$\d+(?:,\d{3})*", text)
+    terms = re.findall(r"\b[A-Za-z]{6,}\b", text)
+
+    return {
+        "parties": list(set(parties[:10])),
+        "dates": list(set(dates)),
+        "amounts": list(set(amounts)),
+        "terms": list(set(terms[:20]))
+    }
+
+
+def clause_extraction(text):
+    clauses = re.split(r"\n+|\.\s+", text)
+    clauses = [c.strip() for c in clauses if len(c.strip()) > 40]
+    return clauses[:12]
+
+
+def risk_analysis(text):
+    risks = [
+        "Very broad confidentiality definitions.",
+        "Unlimited liability exposure.",
+        "One-sided termination rights.",
+        "Strict non-solicitation terms.",
+        "Very long survival period."
+    ]
+    return risks[:5]
+
+
+def fairness_assessment(text):
+    score = 40
+    return {"score": score, "company_power": 60}
+
+
+def alternative_clauses():
+    return [
+        {"title": "Balanced Version", "frequency": 78, "text": "Both parties must mutually approve disclosure."},
+        {"title": "Standard Version", "frequency": 55, "text": "Confidential data must be protected for 2 years."},
+        {"title": "Strict Version", "frequency": 22, "text": "Party B must delete all data within 5 days."},
+    ]
+
+
+def ai_contract_assistant(text):
+    return "You may request clarification of overly broad definitions or long-term obligations."