use std::collections::HashMap;
use std::sync::Arc;
use tracing::info;
use crate::storage::MemoryStorage;
use mr_common::types::{Memory, MemoryScope, MemoryType};
use super::PhaseResult;
#[allow(dead_code)]
pub struct CrossProjectExtractor {
storage: Arc<dyn MemoryStorage>,
batch_size: usize,
batch_interval_ms: u64,
}
impl CrossProjectExtractor {
pub fn new(storage: Arc<dyn MemoryStorage>, batch_size: usize, batch_interval_ms: u64) -> Self {
Self {
storage,
batch_size,
batch_interval_ms,
}
}
pub async fn execute(&self) -> PhaseResult {
info!(target: "dream", "Dream Phase 1 started: CrossProjectExtract");
let memories = match self.storage.list(10000).await {
Ok(m) => m,
Err(e) => return PhaseResult::err("CrossProjectExtract", e.to_string()),
};
if memories.is_empty() {
info!(target: "dream", "Dream Phase 1 skipped: no memories found");
return PhaseResult::ok("CrossProjectExtract", 0, 0);
}
let mut tag_project_count: HashMap<String, usize> = HashMap::new();
for memory in &memories {
let project_key = memory
.project_id
.map(|p| p.to_string())
.unwrap_or_else(|| "global".to_string());
for tag in &memory.tags {
let key = format!("{}:{}", project_key, tag);
*tag_project_count.entry(key).or_insert(0) += 1;
}
}
let mut cross_project_tags: HashMap<String, usize> = HashMap::new();
for key in tag_project_count.keys() {
let parts: Vec<&str> = key.splitn(2, ':').collect();
if parts.len() == 2 {
let tag = parts[1];
*cross_project_tags.entry(tag.to_string()).or_insert(0) += 1;
}
}
let cross_project_tags: Vec<String> = cross_project_tags
.iter()
.filter(|(_, count)| **count > 1)
.map(|(tag, _)| tag.clone())
.collect();
let processed = memories.len();
if cross_project_tags.is_empty() {
info!(target: "dream", "Dream Phase 1 completed: no cross-project tags found");
return PhaseResult::ok("CrossProjectExtract", processed, 0);
}
info!(
target: "dream",
"Dream Phase 1: found {} cross-project tags",
cross_project_tags.len()
);
let summary_content = format!(
"Cross-project common themes: {}",
cross_project_tags.join(", ")
);
let global_memory = Memory::new(summary_content, MemoryType::Knowledge)
.with_tags(vec!["cross-project".to_string()])
.with_scope(MemoryScope::Global);
let created = match self.storage.add(&global_memory).await {
Ok(_) => {
info!(target: "dream", "Dream Phase 1: created global memory for cross-project themes");
1
}
Err(e) => {
tracing::warn!("Failed to create cross-project memory: {}", e);
0
}
};
info!(
target: "dream",
"Dream Phase 1 completed: processed {} memories, created {}",
processed, created
);
PhaseResult::ok("CrossProjectExtract", processed, created)
}
}
#[cfg(test)]
mod tests {
use super::*;
use crate::storage::{MemoryStore, RocksDBStore};
use mr_common::MemoryType;
use tempfile::tempdir;
#[tokio::test]
async fn test_cross_project_extractor_empty() {
let dir = tempdir().unwrap();
let rocksdb = RocksDBStore::open(dir.path()).unwrap();
let storage = Arc::new(MemoryStore::new(std::sync::Arc::new(rocksdb)));
let extractor = CrossProjectExtractor::new(storage, 100, 10);
let result = extractor.execute().await;
assert!(result.success);
assert_eq!(result.processed_count, 0);
}
#[tokio::test]
async fn test_cross_project_extractor_with_memories() {
let dir = tempdir().unwrap();
let rocksdb = RocksDBStore::open(dir.path()).unwrap();
let storage = Arc::new(MemoryStore::new(std::sync::Arc::new(rocksdb)));
let mut m1 = Memory::new("memory 1".to_string(), MemoryType::Knowledge);
m1.tags = vec!["rust".to_string()];
m1.project_id = Some(uuid::Uuid::new_v4());
storage.save(&m1).await.unwrap();
let mut m2 = Memory::new("memory 2".to_string(), MemoryType::Knowledge);
m2.tags = vec!["rust".to_string()];
m2.project_id = Some(uuid::Uuid::new_v4());
storage.save(&m2).await.unwrap();
let extractor = CrossProjectExtractor::new(storage, 100, 10);
let result = extractor.execute().await;
assert!(result.success);
assert_eq!(result.processed_count, 2);
assert_eq!(result.created_count, 1);
}
}