compaction.h

Go to the documentation of this file.

// SPDX-License-Identifier: Apache-2.0
#pragma once
 
#include <entropic/types/config.h>
#include <entropic/types/message.h>
 
#include <string>
#include <unordered_map>
#include <vector>
 
namespace entropic {
 
class TokenCounter {
public:
    explicit TokenCounter(int max_tokens);
 
    int count_message(const Message& msg) const;
 
    int count_messages(const std::vector<Message>& messages) const;
 
    float usage_percent(const std::vector<Message>& messages) const;
 
    void clear_cache();
 
    int max_tokens; 
 
private:
    static int count_text(const std::string& text);
 
    mutable std::unordered_map<const void*, int> cache_; 
};
 
struct CompactionResult {
    bool compacted = false;        
    int old_token_count = 0;       
    int new_token_count = 0;       
    std::string summary;           
    int preserved_messages = 0;    
    int messages_summarized = 0;   
    std::vector<Message> messages; 
 
    /* v1.9.9 additions */
    std::string identity;              
    std::string compactor_source;      
    bool custom_compactor_used = false; 
};
 
class CompactionManager {
public:
    CompactionManager(const CompactionConfig& config, TokenCounter& counter);
 
    CompactionResult check_and_compact(
        std::vector<Message>& messages,
        bool force = false,
        const std::string& conversation_id = "");
 
    CompactionResult compact_messages(
        const std::vector<Message>& messages);
 
    void set_storage(const struct StorageInterface* storage);
 
    CompactionConfig config; 
    TokenCounter& counter;   
 
private:
    std::vector<Message> compact(
        const std::vector<Message>& messages,
        std::string& summary,
        int& stripped_count);
 
    static std::string structured_summary(
        const std::vector<Message>& messages);
 
    static std::string extract_original_task(
        const std::vector<Message>& messages);
 
    static std::vector<std::pair<std::string, std::string>>
    extract_tool_log(const std::vector<Message>& messages);
 
    static std::string format_summary(
        const std::string& summary,
        int message_count);
 
    void save_snapshot(const std::string& conversation_id,
                       const std::vector<Message>& messages);
 
    const struct StorageInterface* storage_ = nullptr; 
};
 
} // namespace entropic