Add smart review summary, symbol index, and feedback loop

Author: haasonsaas

.diffscope.yml.example

@@ -11,6 +11,13 @@ min_confidence: 0.0       # Drop comments below this confidence (0.0-1.0)
 review_profile: balanced  # balanced | chill | assertive
 review_instructions: |
   Prioritize security and correctness issues. Avoid stylistic comments unless they impact maintainability.
+smart_review_summary: true   # Include AI-generated PR summary in smart-review output
+smart_review_diagram: false  # Include Mermaid diagram in PR summary (smart-review)
+symbol_index: true           # Build repo symbol index for cross-file context
+symbol_index_max_files: 500
+symbol_index_max_bytes: 200000
+symbol_index_max_locations: 5
+feedback_path: ".diffscope.feedback.json"
 
 # API configuration (optional - can use environment variables)
 # api_key: your-api-key-here

README.md

@@ -196,6 +196,13 @@ min_confidence: 0.0       # Drop comments below this confidence (0.0-1.0)
 review_profile: balanced  # balanced | chill | assertive
 review_instructions: |
   Prioritize security and correctness issues. Avoid stylistic comments unless they impact maintainability.
+smart_review_summary: true   # Include AI-generated PR summary in smart-review output
+smart_review_diagram: false  # Include Mermaid diagram in PR summary (smart-review)
+symbol_index: true           # Build repo symbol index for cross-file context
+symbol_index_max_files: 500
+symbol_index_max_bytes: 200000
+symbol_index_max_locations: 5
+feedback_path: ".diffscope.feedback.json"
 system_prompt: "Focus on security vulnerabilities, performance issues, and best practices"
 openai_use_responses: true  # Use OpenAI Responses API (recommended) instead of chat completions
 
@@ -274,6 +281,25 @@ Apache-2.0 License. See [LICENSE](LICENSE) for details.
 🚨 **Critical Issues:** 1
 📁 **Files Analyzed:** 3
 
+## 🧾 PR Summary
+
+**Add auth safeguards** (Fix)
+
+### Key Changes
+
+- Harden auth query handling
+- Add route-level guards
+- Introduce safe defaults for user lookups
+
+### Diagram
+
+```mermaid
+flowchart TD
+  A[Request] --> B[Auth Guard]
+  B --> C[DB Query]
+  C --> D[Response]
+```
+
 ## 🧭 Change Walkthrough
 
 - `src/auth.py` (modified; +12, -3)
@@ -413,6 +439,9 @@ temperature: 0.1  # Low for consistent reviews
 max_tokens: 4000
 min_confidence: 0.2
 review_profile: assertive
+smart_review_summary: true
+smart_review_diagram: true
+symbol_index: true
 
 system_prompt: |
   You are reviewing Python code for a production FastAPI application.
@@ -607,6 +636,20 @@ Planned support for responding to pull request comments with interactive command
 @diffscope help                   # Show all commands
 ```
 
+### ✅ Feedback Loop (Reduce Repeated False Positives)
+
+Use the feedback store to suppress comments you’ve already reviewed:
+
+```bash
+# Reject comments from a prior JSON review output
+diffscope feedback --reject review.json
+
+# Accept comments (keeps a record, and removes them from suppress list)
+diffscope feedback --accept review.json
+```
+
+The feedback file defaults to `.diffscope.feedback.json` and can be configured in `.diffscope.yml`.
+
 ### 📊 PR Summary Generation
 
 Generate executive summaries for pull requests:

src/config.rs

@@ -29,6 +29,27 @@ pub struct Config {
     #[serde(default)]
     pub review_instructions: Option<String>,
 
+    #[serde(default = "default_true")]
+    pub smart_review_summary: bool,
+
+    #[serde(default)]
+    pub smart_review_diagram: bool,
+
+    #[serde(default = "default_true")]
+    pub symbol_index: bool,
+
+    #[serde(default = "default_symbol_index_max_files")]
+    pub symbol_index_max_files: usize,
+
+    #[serde(default = "default_symbol_index_max_bytes")]
+    pub symbol_index_max_bytes: usize,
+
+    #[serde(default = "default_symbol_index_max_locations")]
+    pub symbol_index_max_locations: usize,
+
+    #[serde(default = "default_feedback_path")]
+    pub feedback_path: PathBuf,
+
     pub system_prompt: Option<String>,
     pub api_key: Option<String>,
     pub base_url: Option<String>,
@@ -101,6 +122,13 @@ impl Default for Config {
             min_confidence: default_min_confidence(),
             review_profile: None,
             review_instructions: None,
+            smart_review_summary: true,
+            smart_review_diagram: false,
+            symbol_index: true,
+            symbol_index_max_files: default_symbol_index_max_files(),
+            symbol_index_max_bytes: default_symbol_index_max_bytes(),
+            symbol_index_max_locations: default_symbol_index_max_locations(),
+            feedback_path: default_feedback_path(),
             system_prompt: None,
             api_key: None,
             base_url: None,
@@ -166,6 +194,16 @@ impl Config {
             self.max_tokens = default_max_tokens();
         }
 
+        if self.symbol_index_max_files == 0 {
+            self.symbol_index_max_files = default_symbol_index_max_files();
+        }
+        if self.symbol_index_max_bytes == 0 {
+            self.symbol_index_max_bytes = default_symbol_index_max_bytes();
+        }
+        if self.symbol_index_max_locations == 0 {
+            self.symbol_index_max_locations = default_symbol_index_max_locations();
+        }
+
         if !self.min_confidence.is_finite() {
             self.min_confidence = default_min_confidence();
         } else if !(0.0..=1.0).contains(&self.min_confidence) {
@@ -292,6 +330,22 @@ fn default_min_confidence() -> f32 {
     0.0
 }
 
+fn default_symbol_index_max_files() -> usize {
+    500
+}
+
+fn default_symbol_index_max_bytes() -> usize {
+    200_000
+}
+
+fn default_symbol_index_max_locations() -> usize {
+    5
+}
+
+fn default_feedback_path() -> PathBuf {
+    PathBuf::from(".diffscope.feedback.json")
+}
+
 fn default_true() -> bool {
     true
 }

src/core/comment.rs

@@ -5,6 +5,8 @@ use std::path::PathBuf;
 
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct Comment {
+    #[serde(default)]
+    pub id: String,
     pub file_path: PathBuf,
     pub line_number: usize,
     pub content: String,
@@ -139,8 +141,10 @@ impl CommentSynthesizer {
             .clone()
             .unwrap_or_else(|| Self::determine_fix_effort(&raw.content, &category));
         let code_suggestion = Self::generate_code_suggestion(&raw);
+        let id = Self::generate_comment_id(&raw.file_path, &raw.content, &category);
 
         Ok(Some(Comment {
+            id,
             file_path: raw.file_path,
             line_number: raw.line_number,
             content: raw.content,
@@ -154,6 +158,10 @@ impl CommentSynthesizer {
         }))
     }
 
+    fn generate_comment_id(file_path: &PathBuf, content: &str, category: &Category) -> String {
+        compute_comment_id(file_path, content, category)
+    }
+
     fn determine_severity(content: &str) -> Severity {
         let lower = content.to_lowercase();
         if lower.contains("error") || lower.contains("critical") {
@@ -416,6 +424,47 @@ impl CommentSynthesizer {
     }
 }
 
+pub fn compute_comment_id(file_path: &PathBuf, content: &str, category: &Category) -> String {
+    let normalized = normalize_content(content);
+    let key = format!("{}|{:?}|{}", file_path.display(), category, normalized);
+    let hash = fnv1a64(key.as_bytes());
+    format!("cmt_{:016x}", hash)
+}
+
+fn fnv1a64(bytes: &[u8]) -> u64 {
+    let mut hash: u64 = 0xcbf29ce484222325;
+    for &byte in bytes {
+        hash ^= byte as u64;
+        hash = hash.wrapping_mul(0x100000001b3);
+    }
+    hash
+}
+
+fn normalize_content(content: &str) -> String {
+    let mut normalized = String::new();
+    let mut last_space = false;
+
+    for ch in content.chars() {
+        let ch = if ch.is_ascii_digit() {
+            '#'
+        } else {
+            ch.to_ascii_lowercase()
+        };
+
+        if ch.is_whitespace() {
+            if !last_space {
+                normalized.push(' ');
+                last_space = true;
+            }
+        } else {
+            normalized.push(ch);
+            last_space = false;
+        }
+    }
+
+    normalized.trim().to_string()
+}
+
 #[derive(Debug)]
 pub struct RawComment {
     pub file_path: PathBuf,

src/core/context.rs

@@ -5,6 +5,7 @@ use std::collections::HashSet;
 use std::path::Path;
 use std::path::PathBuf;
 
+use crate::core::SymbolIndex;
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct LLMContextChunk {
     pub file_path: PathBuf,
@@ -176,6 +177,39 @@ impl ContextFetcher {
 
         Ok(chunks)
     }
+
+    pub async fn fetch_related_definitions_with_index(
+        &self,
+        file_path: &PathBuf,
+        symbols: &[String],
+        index: &SymbolIndex,
+        max_locations: usize,
+    ) -> Result<Vec<LLMContextChunk>> {
+        let mut chunks = Vec::new();
+
+        if symbols.is_empty() {
+            return Ok(chunks);
+        }
+
+        for symbol in symbols {
+            if let Some(locations) = index.lookup(symbol) {
+                for location in locations.iter().take(max_locations) {
+                    if &location.file_path == file_path {
+                        continue;
+                    }
+                    let snippet = truncate_with_notice(location.snippet.clone(), MAX_CONTEXT_CHARS);
+                    chunks.push(LLMContextChunk {
+                        file_path: location.file_path.clone(),
+                        content: snippet,
+                        context_type: ContextType::Definition,
+                        line_range: Some(location.line_range),
+                    });
+                }
+            }
+        }
+
+        Ok(chunks)
+    }
 }
 
 fn merge_ranges(lines: &[(usize, usize)]) -> Vec<(usize, usize)> {

src/core/mod.rs

@@ -8,13 +8,15 @@ pub mod interactive;
 pub mod pr_summary;
 pub mod prompt;
 pub mod smart_review_prompt;
+pub mod symbol_index;
 
 pub use changelog::ChangelogGenerator;
 pub use comment::{Comment, CommentSynthesizer};
 pub use commit_prompt::CommitPromptBuilder;
 pub use context::{ContextFetcher, ContextType, LLMContextChunk};
 pub use diff_parser::{DiffParser, UnifiedDiff};
 pub use git::GitIntegration;
-pub use pr_summary::PRSummaryGenerator;
+pub use pr_summary::{PRSummaryGenerator, SummaryOptions};
 pub use prompt::PromptBuilder;
 pub use smart_review_prompt::SmartReviewPromptBuilder;
+pub use symbol_index::SymbolIndex;