Claude/review changes mjux96vr0mz96409 jm0 oo

ais_config.json

@@ -12,12 +12,12 @@
     },
     "marinesia": {
       "enabled": true,
-      "api_key": "",
+      "api_key": "${MARINESIA_API_KEY}",
       "rate_limit": 30,
       "description": "REST API fallback from marinesia.com (free, no key required for basic access)"
     },
     "gfw": {
-      "enabled": false,
+      "enabled": true,
       "api_key": "${GFW_API_KEY}",
       "rate_limit": 10,
       "description": "Global Fishing Watch enrichment data (requires free API key, enrichment only)"
@@ -45,6 +45,7 @@
     "env_vars": [
       "AISSTREAM_API_KEY - Get from https://aisstream.io/",
       "AISHUB_USERNAME - Register at https://www.aishub.net/register",
+      "MARINESIA_API_KEY - Get from https://marinesia.com/",
       "GFW_API_KEY - Get from https://globalfishingwatch.org/our-apis/"
     ],
     "priority_explanation": "Sources are tried in order. AISStream provides real-time WebSocket data. AISHub is community-based fallback. Marinesia is REST fallback.",

analytics_constants.py

@@ -0,0 +1,205 @@
+"""
+Centralized Analytics Constants
+
+All threshold values, scoring weights, and detection parameters used across
+analytics modules. Import from here to ensure consistency.
+
+Modules using these constants:
+- behavior.py
+- dark_fleet.py
+- gfw_integration.py
+- venezuela.py
+- laden_status.py
+"""
+
+# =============================================================================
+# Risk Level Thresholds (score 0-100)
+# =============================================================================
+# All modules must use these consistent breakpoints
+
+RISK_LEVEL_CRITICAL = 70  # 70-100: High probability of illicit activity
+RISK_LEVEL_HIGH = 50      # 50-69: Multiple concerning indicators
+RISK_LEVEL_MEDIUM = 30    # 30-49: Some concerning indicators
+RISK_LEVEL_LOW = 15       # 15-29: Minor risk factors
+RISK_LEVEL_MINIMAL = 0    # 0-14: No significant indicators
+
+
+def get_risk_level(score: int) -> str:
+    """
+    Convert numeric score to standardized risk level.
+
+    Use this function in all modules to ensure consistent classification.
+    """
+    if score >= RISK_LEVEL_CRITICAL:
+        return 'critical'
+    elif score >= RISK_LEVEL_HIGH:
+        return 'high'
+    elif score >= RISK_LEVEL_MEDIUM:
+        return 'medium'
+    elif score >= RISK_LEVEL_LOW:
+        return 'low'
+    else:
+        return 'minimal'
+
+
+# =============================================================================
+# AIS Gap Detection Thresholds
+# =============================================================================
+
+# Minimum gap to consider significant (minutes)
+AIS_GAP_MIN_MINUTES = 60          # Gaps shorter than this are normal
+
+# Reporting thresholds (minutes)
+AIS_GAP_REPORT_THRESHOLD = 60     # Report gaps >= 1 hour
+AIS_GAP_SUSPICIOUS_HOURS = 12     # Gaps > 12 hours are suspicious
+AIS_GAP_CRITICAL_HOURS = 48       # Gaps > 48 hours are highly suspicious
+
+# Scoring (hours-based, standardized)
+AIS_GAP_SCORE_CRITICAL = 20       # Points for > 48 hours total gap time
+AIS_GAP_SCORE_HIGH = 15           # Points for > 12 hours total gap time
+AIS_GAP_SCORE_MEDIUM = 10         # Points for any gaps detected
+
+
+# =============================================================================
+# Encounter/STS Detection Thresholds
+# =============================================================================
+
+# Distance thresholds
+ENCOUNTER_MAX_DISTANCE_KM = 0.5   # 500m - vessels must be within this distance
+STS_MAX_DISTANCE_KM = 0.5         # Same as encounter (500m)
+
+# Speed thresholds (knots)
+ENCOUNTER_MAX_SPEED_KNOTS = 2.0   # Both vessels must be < 2 knots
+STS_MAX_SPEED_KNOTS = 3.0         # Slightly higher for STS (allows drift)
+# Note: STS uses 3.0 based on arXiv 2024 research - vessels have small drift during oil transfer
+
+# Duration thresholds (hours)
+ENCOUNTER_MIN_DURATION_HOURS = 2.0    # Minimum time for encounter
+STS_MIN_DURATION_HOURS = 4.0          # STS needs longer duration
+STS_MAX_DURATION_HOURS = 48.0         # Maximum realistic STS duration
+
+# Scoring
+ENCOUNTER_SCORE_MULTIPLE = 25     # Points for > 3 encounters
+ENCOUNTER_SCORE_SINGLE = 10       # Points for 1-3 encounters
+STS_SCORE_MULTIPLE = 15           # Points for >= 2 STS transfers
+STS_SCORE_SINGLE = 10             # Points for 1 STS transfer
+
+
+# =============================================================================
+# Loitering Detection Thresholds
+# =============================================================================
+
+LOITERING_MAX_SPEED_KNOTS = 2.0       # Speed threshold for loitering
+LOITERING_MIN_DURATION_HOURS = 3.0    # Minimum duration to flag
+LOITERING_MIN_DISTANCE_FROM_PORT_NM = 20.0  # Must be away from ports
+
+# Scoring (hours-based, standardized)
+LOITERING_SCORE_EXTENDED = 20     # Points for > 72 hours total
+LOITERING_SCORE_MODERATE = 10     # Points for > 24 hours total
+
+
+# =============================================================================
+# Spoofing Detection Thresholds
+# =============================================================================
+
+# Position discrepancy thresholds (km)
+SPOOFING_CRITICAL_DISCREPANCY_KM = 100    # Almost certainly spoofing
+SPOOFING_HIGH_DISCREPANCY_KM = 50         # Likely spoofing
+SPOOFING_MEDIUM_DISCREPANCY_KM = 20       # Possible spoofing
+
+# Speed-based spoofing (knots)
+SPOOFING_MAX_REALISTIC_SPEED_KNOTS = 50   # Max realistic vessel speed
+SPOOFING_SPEED_BUFFER = 1.5               # Allow 50% buffer for GPS errors
+
+# Scoring
+SPOOFING_SCORE_CRITICAL = 30      # Points for > 100km discrepancy
+SPOOFING_SCORE_HIGH = 15          # Points for > 20km discrepancy
+
+
+# =============================================================================
+# Vessel Age Thresholds
+# =============================================================================
+
+VESSEL_AGE_CRITICAL = 25          # 25+ years = highest risk (shadow fleet uses old tankers)
+VESSEL_AGE_HIGH = 20              # 20-24 years = high risk
+VESSEL_AGE_MEDIUM = 15            # 15-19 years = moderate risk
+
+# Scoring
+VESSEL_AGE_SCORE_CRITICAL = 20    # Points for >= 25 years
+VESSEL_AGE_SCORE_HIGH = 15        # Points for >= 20 years
+VESSEL_AGE_SCORE_MEDIUM = 10      # Points for >= 15 years
+
+
+# =============================================================================
+# Zone Detection Radii (km)
+# =============================================================================
+
+DETECTION_RADIUS_TERMINAL = 10.0      # Oil/cargo terminals
+DETECTION_RADIUS_STS_ZONE = 25.0      # STS transfer zones (standardized)
+DETECTION_RADIUS_ANCHORAGE = 15.0     # Anchorage areas
+DETECTION_RADIUS_REFINERY = 15.0      # Refinery facilities
+DETECTION_RADIUS_SPOOFING_TARGET = 50.0  # Spoofing destination checks
+
+
+# =============================================================================
+# Regional Presence Scoring
+# =============================================================================
+
+REGIONAL_PRESENCE_LOOKBACK = 200      # Number of positions to analyze
+REGIONAL_PRESENCE_MIN_POSITIONS = 20  # Min positions to trigger scoring
+REGIONAL_PRESENCE_MAX_POINTS = 15     # Max points per region
+
+
+# =============================================================================
+# Flag Risk Scoring
+# =============================================================================
+
+# See behavior.py FLAGS_OF_CONVENIENCE and SHADOW_FLEET_FLAGS for flag lists
+FLAG_SCORE_SHADOW_FLEET = 25      # Known shadow fleet registry
+FLAG_SCORE_FOC = 15               # General flag of convenience
+FLAG_SCORE_FRAUDULENT = 35        # Fraudulent/emerging dark registry
+
+
+# =============================================================================
+# Ownership Risk Scoring
+# =============================================================================
+
+OWNERSHIP_SCORE_UNKNOWN = 15      # No owner information
+OWNERSHIP_SCORE_OBSCURED = 10     # Owner info appears hidden
+
+
+# =============================================================================
+# Combined Risk Weights
+# =============================================================================
+# Used in combined-risk endpoint for weighted averaging
+
+WEIGHT_BEHAVIOR = 1.0             # Local analysis weight
+WEIGHT_DARK_FLEET = 1.2           # Regional intelligence weight
+WEIGHT_GFW = 1.5                  # Verified external data weight (highest)
+
+
+# =============================================================================
+# Utility Functions
+# =============================================================================
+
+def calculate_risk_assessment(score: int) -> dict:
+    """
+    Generate standardized risk assessment from score.
+
+    Returns dict with score, level, and description.
+    """
+    level = get_risk_level(score)
+
+    assessments = {
+        'critical': 'High probability of dark fleet / sanctions evasion activity',
+        'high': 'Multiple dark fleet indicators present',
+        'medium': 'Some concerning indicators detected',
+        'low': 'Minor risk factors present',
+        'minimal': 'No significant dark fleet indicators'
+    }
+
+    return {
+        'score': min(score, 100),
+        'risk_level': level,
+        'assessment': assessments.get(level, 'Unknown')
+    }

dark_fleet.py

@@ -616,7 +616,12 @@ def calculate_dark_fleet_risk_score(
     if track_history and mmsi:
         gaps = detect_ais_gaps(track_history, mmsi, min_gap_minutes=120)
         if gaps:
-            total_gap_hours = sum(g.get("details", {}).get("gap_hours", 0) for g in gaps)
+            # BehaviorEvent objects - access .details attribute, not .get()
+            total_gap_hours = sum(
+                g.details.get("gap_hours", 0) if hasattr(g, 'details') else
+                g.get("details", {}).get("gap_hours", 0) if isinstance(g, dict) else 0
+                for g in gaps
+            )
             if total_gap_hours > 48:
                 score += 20
                 factors.append({

gfw_integration.py

@@ -575,11 +575,17 @@ def get_dark_fleet_indicators(mmsi: str, days: int = 90) -> dict:
         risk_score += 10
         risk_factors.append(f"Loitering: {loitering_hours:.0f} hours")
 
-    risk_level = 'low'
-    if risk_score >= 50:
+    # Standardized risk levels matching behavior.py and dark_fleet.py
+    if risk_score >= 70:
+        risk_level = 'critical'
+    elif risk_score >= 50:
         risk_level = 'high'
-    elif risk_score >= 25:
+    elif risk_score >= 30:
         risk_level = 'medium'
+    elif risk_score >= 15:
+        risk_level = 'low'
+    else:
+        risk_level = 'minimal'
 
     return {
         'mmsi': mmsi,

laden_status.py

@@ -172,10 +172,12 @@ def extract_draft_readings(track_history: List[dict]) -> List[DraftReading]:
                     # Try parsing ISO format
                     try:
                         timestamp = datetime.fromisoformat(timestamp.replace('Z', '+00:00'))
-                    except:
-                        timestamp = datetime.now()
+                    except (ValueError, AttributeError):
+                        # Skip positions with unparseable timestamps
+                        continue
                 elif not isinstance(timestamp, datetime):
-                    timestamp = datetime.now()
+                    # Skip positions without valid timestamps
+                    continue
 
                 readings.append(DraftReading(
                     timestamp=timestamp,