soma-core 2.0.0

// Issue #25: EditClassificationSystem - Working Implementation
// Production-ready edit classification with zero compilation errors

use crate::classification::types::*;
use crate::edit_control::{ModifiableEdit, ApprovalState};
use crate::memory::SymbolicContext;
use anyhow::Result;
use std::collections::HashMap;

/// Production-ready edit classification system
#[derive(Debug, Clone)]
pub struct EditClassificationSystem {
    /// Configuration
    config: ClassificationConfig,
}

#[derive(Debug, Clone)]
pub struct ClassificationConfig {
    pub use_cognitive_analysis: bool,
    pub auto_classification_threshold: f64,
    pub max_batch_size: usize,
    pub verbose_logging: bool,
    pub risk_strictness: f64,
}

impl Default for ClassificationConfig {
    fn default() -> Self {
        Self {
            use_cognitive_analysis: false,
            auto_classification_threshold: 0.8,
            max_batch_size: 100,
            verbose_logging: false,
            risk_strictness: 1.0,
        }
    }
}

/// Comprehensive edit classification result
#[derive(Debug, Clone)]
pub struct ClassifiedEdit {
    /// Original edit proposal
    pub edit: ModifiableEdit,
    
    /// Determined category
    pub category: EditCategory,
    
    /// Risk assessment
    pub risk_assessment: RiskAssessment,
    
    /// Priority score (0.0 = lowest, 1.0 = highest)
    pub priority_score: f64,
    
    /// Confidence in classification (0.0 = low, 1.0 = high)
    pub classification_confidence: f64,
    
    /// Cognitive analysis results (if enabled)
    pub cognitive_analysis: Option<SymbolicContext>,
    
    /// Recommended action
    pub recommendation: ClassificationRecommendation,
    
    /// Detailed reasoning
    pub reasoning: String,
}

/// System recommendation for handling classified edits
#[derive(Debug, Clone)]
pub enum ClassificationRecommendation {
    /// Automatically approve with confidence level
    AutoApprove { confidence: f64 },
    
    /// Require manual review with specific concerns
    RequireReview { concerns: Vec<String> },
    
    /// Escalate to higher approval level
    Escalate { target_level: String, reason: String },
    
    /// Reject with explanation
    Reject { reason: String },
    
    /// Request more information
    RequestInfo { questions: Vec<String> },
}

/// Filtering criteria for edit classification
#[derive(Debug, Clone, Default)]
pub struct FilterCriteria {
    pub categories: Option<Vec<EditCategory>>,
    pub min_priority: Option<f64>,
    pub max_risk: Option<f64>,
    pub approval_states: Option<Vec<ApprovalState>>,
    pub min_confidence: Option<f64>,
}

/// Statistical analysis of classification results
#[derive(Debug, Clone)]
pub struct ClassificationStats {
    pub total_count: usize,
    pub category_distribution: HashMap<std::mem::Discriminant<EditCategory>, usize>,
    pub average_risk: f64,
    pub average_priority: f64,
    pub average_confidence: f64,
}

impl EditClassificationSystem {
    /// Create new classification system with default configuration
    pub fn new() -> Self {
        Self {
            config: ClassificationConfig::default(),
        }
    }

    pub fn with_config(config: ClassificationConfig) -> Self {
        Self { config }
    }

    /// Classify a single edit with comprehensive analysis
    pub fn classify_edit(&self, edit: &ModifiableEdit) -> Result<ClassifiedEdit> {
        // Determine category based on file type and content
        let category = self.determine_category(edit)?;
        
        // Assess risk factors
        let risk_assessment = self.assess_risk(edit, &category)?;
        
        // Run cognitive analysis if enabled
        let cognitive_analysis = if self.config.use_cognitive_analysis {
            self.run_cognitive_analysis(edit, &category)?
        } else {
            None
        };
        
        // Calculate priority score
        let priority_score = self.calculate_priority_score(edit, &category, &risk_assessment, &cognitive_analysis)?;
        
        // Calculate classification confidence
        let classification_confidence = self.calculate_classification_confidence(edit, &category, &risk_assessment)?;
        
        // Generate recommendation
        let recommendation = self.generate_recommendation(&category, &risk_assessment, classification_confidence)?;
        
        // Generate detailed reasoning
        let reasoning = self.generate_reasoning(edit, &category, &risk_assessment, &cognitive_analysis)?;
        
        Ok(ClassifiedEdit {
            edit: edit.clone(),
            category,
            risk_assessment,
            priority_score,
            classification_confidence,
            cognitive_analysis,
            recommendation,
            reasoning,
        })
    }

    /// Classify multiple edits and sort by priority
    pub fn classify_batch(&self, edits: &[ModifiableEdit]) -> Result<Vec<ClassifiedEdit>> {
        if edits.len() > self.config.max_batch_size {
            return Err(anyhow::anyhow!("Batch size {} exceeds maximum {}", edits.len(), self.config.max_batch_size));
        }

        let mut classified: Vec<ClassifiedEdit> = edits
            .iter()
            .map(|edit| self.classify_edit(edit))
            .collect::<Result<Vec<_>>>()?;

        // Sort by priority score (highest first)
        classified.sort_by(|a, b| b.priority_score.partial_cmp(&a.priority_score).unwrap_or(std::cmp::Ordering::Equal));

        Ok(classified)
    }

    /// Filter classified edits by criteria
    pub fn filter_edits<'a>(&self, edits: &'a [ClassifiedEdit], criteria: &FilterCriteria) -> Vec<&'a ClassifiedEdit> {
        edits
            .iter()
            .filter(|edit| {
                // Category filter
                if let Some(categories) = &criteria.categories {
                    if !categories.contains(&edit.category) {
                        return false;
                    }
                }

                // Risk filter
                if let Some(max_risk) = criteria.max_risk {
                    if edit.risk_assessment.overall_score > max_risk {
                        return false;
                    }
                }

                // Priority filter
                if let Some(min_priority) = criteria.min_priority {
                    if edit.priority_score < min_priority {
                        return false;
                    }
                }

                // Confidence filter
                if let Some(min_confidence) = criteria.min_confidence {
                    if edit.classification_confidence < min_confidence {
                        return false;
                    }
                }

                // Approval state filter
                if let Some(approval_states) = &criteria.approval_states {
                    if !approval_states.contains(&edit.edit.approval_state) {
                        return false;
                    }
                }

                true
            })
            .collect()
    }

    /// Generate statistical analysis of classification results
    pub fn get_classification_stats(&self, edits: &[ClassifiedEdit]) -> ClassificationStats {
        let total_count = edits.len();
        
        let mut category_distribution = HashMap::new();
        for edit in edits {
            let discriminant = std::mem::discriminant(&edit.category);
            *category_distribution.entry(discriminant).or_insert(0) += 1;
        }
        
        let average_risk = if total_count > 0 {
            edits.iter().map(|e| e.risk_assessment.overall_score).sum::<f64>() / total_count as f64
        } else {
            0.0
        };
        
        let average_priority = if total_count > 0 {
            edits.iter().map(|e| e.priority_score).sum::<f64>() / total_count as f64
        } else {
            0.0
        };
        
        let average_confidence = if total_count > 0 {
            edits.iter().map(|e| e.classification_confidence).sum::<f64>() / total_count as f64
        } else {
            0.0
        };
        
        ClassificationStats {
            total_count,
            category_distribution,
            average_risk,
            average_priority,
            average_confidence,
        }
    }

    // Private implementation methods
    fn determine_category(&self, edit: &ModifiableEdit) -> Result<EditCategory> {
        let file_path = &edit.base_edit.file;
        let content = &edit.base_edit.new_code;
        
        // Security-critical patterns
        let security_patterns = ["password", "secret", "token", "auth", "crypto", "unsafe"];
        if security_patterns.iter().any(|pattern| content.contains(pattern)) {
            return Ok(EditCategory::Critical {
                subcategory: CriticalType::SecurityFix,
                impact_scope: ImpactScope::SystemWide,
            });
        }
        
        // Documentation files
        if file_path.ends_with(".md") || file_path.ends_with(".txt") || content.starts_with('#') {
            return Ok(EditCategory::Safe {
                subcategory: SafeType::Documentation,
                confidence_level: ConfidenceLevel::High,
            });
        }
        
        // Experimental code patterns
        if content.contains("for") && content.contains("loop") || content.contains("allocate") {
            return Ok(EditCategory::Experimental {
                subcategory: ExperimentalType::PerformanceOptimization,
                validation_requirements: vec!["Performance testing required".to_string()],
            });
        }
        
        // Default to safe code change
        Ok(EditCategory::Safe {
            subcategory: SafeType::VariableRename,
            confidence_level: ConfidenceLevel::Medium,
        })
    }
    
    fn assess_risk(&self, edit: &ModifiableEdit, category: &EditCategory) -> Result<RiskAssessment> {
        let content = &edit.base_edit.new_code;
        
        let category_risk = match category {
            EditCategory::Critical { .. } => 0.9,
            EditCategory::Experimental { .. } => 0.6,
            EditCategory::Safe { .. } => 0.2,
            EditCategory::Cosmetic { .. } => 0.1,
        };
        
        // Additional risk factors
        let mut additional_risk = 0.0;
        if content.len() > 1000 {
            additional_risk += 0.1; // Large changes are riskier
        }
        if content.contains("unsafe") {
            additional_risk += 0.3; // Unsafe code is risky
        }
        if content.contains("TODO") || content.contains("FIXME") {
            additional_risk += 0.1; // Incomplete code
        }
        
        let sum = category_risk + additional_risk;
        let overall_score = if sum > 1.0 { 1.0 } else { sum } * self.config.risk_strictness;
        
        let explanation = format!(
            "Risk assessment: category={:.2}, additional={:.2}, overall={:.2}",
            category_risk, additional_risk, overall_score
        );
        
        Ok(RiskAssessment::builder()
            .with_risk_factor(RiskFactor::ChangeComplexity, overall_score)
            .with_explanation(explanation)
            .with_mitigation("Review changes carefully and test thoroughly".to_string())
            .build())
    }
    
    fn run_cognitive_analysis(&self, edit: &ModifiableEdit, category: &EditCategory) -> Result<Option<SymbolicContext>> {
        // Simplified cognitive analysis without operators for now
        let mut context = SymbolicContext::new();
        context.set("edit_file", &edit.base_edit.file);
        context.set("edit_category", &format!("{:?}", category));
        context.set("edit_size", &edit.base_edit.new_code.len().to_string());
        context.set("classification_confidence", "0.8");
        
        Ok(Some(context))
    }
    
    fn calculate_priority_score(&self, edit: &ModifiableEdit, category: &EditCategory, risk: &RiskAssessment, cognitive: &Option<SymbolicContext>) -> Result<f64> {
        let risk_score = risk.overall_score * 0.4;
        let complexity_score = (category.priority() as f64 / 4.0) * 0.3;
        let size_ratio = edit.base_edit.new_code.len() as f64 / 10000.0;
        let size_score = if size_ratio > 1.0 { 1.0 } else { size_ratio } * 0.2;
        let cognitive_score = if cognitive.is_some() { 0.1 } else { 0.05 };
        
        Ok(risk_score + complexity_score + size_score + cognitive_score)
    }
    
    fn calculate_classification_confidence(&self, edit: &ModifiableEdit, category: &EditCategory, risk: &RiskAssessment) -> Result<f64> {
        let category_confidence = match category {
            EditCategory::Critical { .. } => 0.9,
            EditCategory::Safe { .. } => 0.8,
            EditCategory::Experimental { .. } => 0.6,
            EditCategory::Cosmetic { .. } => 0.95,
        };
        
        let risk_confidence = 1.0 - (risk.overall_score * 0.3);
        let content_confidence = if edit.base_edit.new_code.len() > 100 { 0.8 } else { 0.6 };
        
        Ok((category_confidence + risk_confidence + content_confidence) / 3.0)
    }
    
    fn generate_recommendation(&self, category: &EditCategory, risk: &RiskAssessment, confidence: f64) -> Result<ClassificationRecommendation> {
        if risk.overall_score > 0.8 {
            return Ok(ClassificationRecommendation::RequireReview {
                concerns: vec!["High risk detected".to_string()],
            });
        }
        
        if confidence > self.config.auto_classification_threshold {
            return Ok(ClassificationRecommendation::AutoApprove { confidence });
        }
        
        match category {
            EditCategory::Critical { .. } => Ok(ClassificationRecommendation::Escalate {
                target_level: "Senior".to_string(),
                reason: "Critical change requires senior review".to_string(),
            }),
            EditCategory::Experimental { .. } => Ok(ClassificationRecommendation::RequireReview {
                concerns: vec!["Experimental change needs validation".to_string()],
            }),
            _ => Ok(ClassificationRecommendation::AutoApprove { confidence }),
        }
    }
    
    fn generate_reasoning(&self, edit: &ModifiableEdit, category: &EditCategory, risk: &RiskAssessment, cognitive: &Option<SymbolicContext>) -> Result<String> {
        let mut reasoning = String::new();
        
        reasoning.push_str(&format!("File: {}\n", edit.base_edit.file));
        reasoning.push_str(&format!("Category: {:?}\n", category));
        reasoning.push_str(&format!("Risk Score: {:.2}\n", risk.overall_score));
        reasoning.push_str(&format!("Risk Explanation: {}\n", risk.explanation));
        reasoning.push_str(&format!("Content Size: {} chars\n", edit.base_edit.new_code.len()));
        
        if cognitive.is_some() {
            reasoning.push_str("Cognitive analysis: Enhanced decision support enabled\n");
        }
        
        Ok(reasoning)
    }
    
    /// Access configuration (for CLI display)
    pub fn config(&self) -> &ClassificationConfig {
        &self.config
    }
}

impl Default for EditClassificationSystem {
    fn default() -> Self {
        Self::new()
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::agents::gpt4_agent::ProposedEdit;
    
    fn create_test_edit(file: &str, content: &str, reason: &str) -> ModifiableEdit {
        let proposed = ProposedEdit {
            file: file.to_string(),
            line_range: (1, 5),
            new_code: content.to_string(),
            reason: reason.to_string(),
            confidence: 0.8,
        };
        ModifiableEdit::from_proposed_edit(proposed)
    }
    
    #[test]
    fn test_classification_system_creation() {
        let system = EditClassificationSystem::new();
        assert_eq!(system.config.auto_classification_threshold, 0.8);
    }
    
    #[test]
    fn test_edit_classification() {
        let system = EditClassificationSystem::new();
        let edit = create_test_edit("test.md", "# Test Documentation", "Update docs");
        
        let classified = system.classify_edit(&edit).unwrap();
        
        assert!(matches!(classified.category, EditCategory::Safe { .. }));
        assert!(classified.priority_score > 0.0);
        assert!(classified.classification_confidence > 0.0);
    }
    
    #[test]
    fn test_security_pattern_detection() {
        let system = EditClassificationSystem::new();
        let edit = create_test_edit("auth.rs", "let password = \"secret123\";", "Add auth");
        
        let classified = system.classify_edit(&edit).unwrap();
        
        assert!(matches!(classified.category, EditCategory::Critical { .. }));
        assert!(classified.risk_assessment.overall_score > 0.5);
    }
    
    #[test]
    fn test_batch_classification() {
        let system = EditClassificationSystem::new();
        let edits = vec![
            create_test_edit("file1.md", "Documentation", "Update docs"),
            create_test_edit("file2.rs", "fn main() {}", "Add function"),
        ];
        
        let classified = system.classify_batch(&edits).unwrap();
        
        assert_eq!(classified.len(), 2);
        // Should be sorted by priority
        assert!(classified[0].priority_score >= classified[1].priority_score);
    }
    
    #[test]
    fn test_filtering() {
        let system = EditClassificationSystem::new();
        let edit = create_test_edit("test.md", "Documentation", "Update docs");
        let classified = system.classify_edit(&edit).unwrap();
        let classified_vec = vec![classified];
        
        let mut criteria = FilterCriteria::default();
        criteria.min_priority = Some(0.0);
        
        let filtered = system.filter_edits(&classified_vec, &criteria);
        assert_eq!(filtered.len(), 1);
    }
}