// ChatIPC := Chat Incremental Pattern Constructor

#include <algorithm>
#include <atomic>
#include <cctype>
#include <cinttypes>
#include <cstring>
#include <fstream>
#include <iostream>
#include <iterator>
#include <map>
#include <mutex>
#include <optional>
#include <sstream>
#include <stdexcept>
#include <string>
#include <thread>
#include <unordered_map>
#include <unordered_set>
#include <vector>

#ifdef _OPENMP
#include <omp.h>
#else
inline int omp_get_max_threads(){ return 1; }
inline int omp_get_thread_num(){ return 0; }
#endif

extern unsigned char dictionary_json[];      // provide dictionary.cpp to embed dictionary JSON bytes
extern unsigned int dictionary_json_len;

// --------------------------- Short utility functions ----------------------

static inline bool is_space(char c){ return std::isspace(static_cast<unsigned char>(c)) != 0; }
static inline char to_low(char c){ return static_cast<char>(std::tolower(static_cast<unsigned char>(c))); }
static inline void safe_flush(std::ostream &os){ os.flush(); }

// Tokenize by whitespace
static std::vector<std::string> tokenize_whitespace(const std::string &s){
    std::istringstream iss(s);
    std::vector<std::string> out;
    std::string t;
    while (iss >> t) out.push_back(t);
    return out;
}

// Tokenize by non-alphanumeric characters (for definitions)
static std::vector<std::string> tokenize_non_alnum(const std::string &s){
    std::vector<std::string> out; std::string cur;
    for (char ch : s){
        if (std::isalnum(static_cast<unsigned char>(ch)) || ch=='-' || ch=='\''){
            cur.push_back(to_low(ch));
        } else {
            if (!cur.empty()){ out.push_back(cur); cur.clear(); }
        }
    }
    if (!cur.empty()) out.push_back(cur);
    return out;
}

// --------------------------- String interning (short methods) --------------

struct StringInterner {
    std::unordered_set<std::string> pool;
    std::mutex m;
    const std::string* intern(const std::string &s){
        std::lock_guard<std::mutex> lk(m);
        auto it = pool.find(s);
        if (it != pool.end()) return &*it;
        auto pr = pool.insert(s);
        return &*pr.first;
    }
};

// ---------- Global parsed dictionary (populated once in main) ----------
static std::unordered_map<std::string,std::string> g_raw_dict;

static std::unordered_set<std::string> def_tokens_from_text(const std::string &s){
    auto toks = tokenize_non_alnum(s);
    return std::unordered_set<std::string>(toks.begin(), toks.end());
}

// --------------------------- Knowledge base (short methods) --------------
using StrPtr = const std::string*;
struct PtrHash { size_t operator()(StrPtr p) const noexcept { return std::hash<std::string>()(*p); } };
struct PtrEq   { bool operator()(StrPtr a, StrPtr b) const noexcept { return *a == *b; } };

using NextSet = std::vector<StrPtr>;

struct KnowledgeBase {
    StringInterner interner;
    std::unordered_map<StrPtr, NextSet, PtrHash, PtrEq> next;
    std::mutex m;

    // def-index: for each interned word pointer -> list of interned tokens (definition expansion)
    std::unordered_map<StrPtr, std::vector<StrPtr>, PtrHash, PtrEq> def_index;
    std::mutex def_m;
    int def_depth = 0;

    void add_pair_interned(StrPtr k, StrPtr v){
        std::lock_guard<std::mutex> lk(m);
        auto &vec = next[k];
        for (auto p : vec) if (*p == *v) return;
        vec.push_back(v);
    }

    // set def depth; if changed, drop previously computed def expansions
    void set_def_depth(int D){
        std::lock_guard<std::mutex> lk(def_m);
        if (D != def_depth){
            def_index.clear();
            def_depth = D;
        }
    }

    // compute definition expansion for a single interned word (if needed)
    void ensure_def_for_interned(StrPtr wp){
        // quick no-op checks
        if (wp == nullptr) return;
        if (def_depth <= 0) return;

        // double-checked locking
        {
            std::lock_guard<std::mutex> lk(def_m);
            if (def_index.find(wp) != def_index.end()) return;
        }

        // compute expansion using global parsed dictionary g_raw_dict
        std::unordered_set<std::string> acc;
        std::vector<std::string> frontier;
        auto it_raw = g_raw_dict.find(*wp);
        if (it_raw != g_raw_dict.end()){
            auto toks = def_tokens_from_text(it_raw->second);
            for (auto &t : toks){
                if (acc.insert(t).second) frontier.push_back(t);
            }
        }

        for (int depth = 1; depth < def_depth && !frontier.empty(); ++depth){
            std::vector<std::string> nextf;
            for (auto &w : frontier){
                auto it2 = g_raw_dict.find(w);
                if (it2 == g_raw_dict.end()) continue;
                auto toks2 = def_tokens_from_text(it2->second);
                for (auto &t : toks2){
                    if (acc.insert(t).second) nextf.push_back(t);
                }
            }
            frontier.swap(nextf);
        }

        // intern all accumulated tokens and store pointers
        std::vector<StrPtr> out;
        out.reserve(acc.size());
        for (auto &s : acc){
            out.push_back(interner.intern(s));
        }

        // store atomically (prevent double insertion)
        {
            std::lock_guard<std::mutex> lk(def_m);
            // another thread may have inserted meanwhile; do not overwrite
            if (def_index.find(wp) == def_index.end()){
                def_index.emplace(wp, std::move(out));
            }
        }
    }

    // existing public add_pair but now ensure def-expansion is built immediately
    void add_pair(const std::string &k, const std::string &v){
        StrPtr kp = interner.intern(k);
        StrPtr vp = interner.intern(v);
        // ensure definition expansion for both words as soon as they are seen
        ensure_def_for_interned(kp);
        ensure_def_for_interned(vp);
        add_pair_interned(kp, vp);
    }

    std::optional<NextSet> lookup_by_string(const std::string &k) const {
        for (auto &pr : next) if (*pr.first == k) return pr.second;
        return std::nullopt;
    }
    std::optional<NextSet> lookup_by_ptr(StrPtr k) const {
        auto it = next.find(k);
        if (it==next.end()) return std::nullopt;
        return it->second;
    }
};

// thread-safe snapshot of kb.def_index as string-based def-index
static std::unordered_map<std::string,std::unordered_set<std::string>>
snapshot_def_index(KnowledgeBase &kb){
    std::unordered_map<std::string,std::unordered_set<std::string>> out;
    std::lock_guard<std::mutex> lk(kb.def_m);
    out.reserve(kb.def_index.size());
    for (auto &pr : kb.def_index){
        std::unordered_set<std::string> s;
        s.reserve(pr.second.size());
        for (auto p : pr.second) s.insert(*p);
        out.emplace(*pr.first, std::move(s));
    }
    return out;
}

// --------------------------- Small JSON parse helpers ----------------------

static inline bool json_valid_index(size_t i, size_t n){ return i < n; }

static std::string parse_quoted_string(const std::string &text, size_t &i){
    std::string out;
    if (!json_valid_index(i, text.size()) || text[i] != '"') throw std::runtime_error("expected '\"'");
    ++i;
    while (json_valid_index(i, text.size())){
        char c = text[i++];
        if (c == '"') break;
        if (c == '\\'){
            if (!json_valid_index(i, text.size())) break;
            char e = text[i++];
            if (e=='n') out.push_back('\n');
            else if (e=='t') out.push_back('\t');
            else out.push_back(e);
        } else out.push_back(c);
    }
    return out;
}

static void skip_spaces(const std::string &s, size_t &i){
    while (json_valid_index(i, s.size()) && is_space(s[i])) ++i;
}

// Very small JSON-like parser tailored to dictionary_json structure
static std::unordered_map<std::string,std::string> parse_dictionary_json(){
    std::unordered_map<std::string,std::string> dict;
    if (dictionary_json_len == 0) return dict;
    std::string text; text.reserve(dictionary_json_len + 1);
    for (unsigned int b=0; b < dictionary_json_len; ++b) text.push_back(static_cast<char>(dictionary_json[b]));
    size_t i = 0;
    skip_spaces(text,i);
    if (!json_valid_index(i,text.size()) || text[i] != '{') return dict;
    ++i;
    while (true){
        skip_spaces(text,i);
        if (!json_valid_index(i,text.size())) break;
        if (text[i] == '}'){ ++i; break; }
        std::string key = parse_quoted_string(text,i);
        skip_spaces(text,i);
        if (!json_valid_index(i,text.size()) || text[i] != ':') break;
        ++i;
        skip_spaces(text,i);
        std::string val;
        if (json_valid_index(i,text.size()) && text[i] == '"') val = parse_quoted_string(text,i);
        else {
            size_t start = i;
            while (json_valid_index(i,text.size()) && text[i] != ',' && text[i] != '}') ++i;
            val = text.substr(start, i-start);
        }
        dict.emplace(std::move(key), std::move(val));
        skip_spaces(text,i);
        if (json_valid_index(i,text.size()) && text[i] == ','){ ++i; continue; }
        if (json_valid_index(i,text.size()) && text[i] == '}'){ ++i; break; }
    }
    return dict;
}

// --------------------------- Similarity helpers (very small) ----------------

static double jaccard_similarity(const std::unordered_set<std::string> &A,
                                 const std::unordered_set<std::string> &B)
{
    if (A.empty() && B.empty()) return 1.0;
    size_t inter = 0;
    if (A.size() < B.size()){
        for (const auto &x : A) if (B.count(x)) ++inter;
    } else {
        for (const auto &x : B) if (A.count(x)) ++inter;
    }
    size_t uni = A.size() + B.size() - inter;
    if (uni == 0) return 0.0;
    return static_cast<double>(inter) / static_cast<double>(uni);
}

static std::unordered_set<std::string>
aggregate_sets(const std::vector<std::string> &tokens,
               const std::unordered_map<std::string,std::unordered_set<std::string>> &def_index)
{
    std::unordered_set<std::string> agg;
    for (auto &t : tokens){
        agg.insert(t);
        auto it = def_index.find(t);
        if (it != def_index.end()){
            for (auto &d : it->second) agg.insert(d);
        }
    }
    return agg;
}

// --------------------------- Candidate selection (short funcs) ---------------

static std::string best_candidate_by_similarity(const NextSet &cands,
    const std::vector<std::string> &prompt_toks,
    const std::vector<std::string> &resp_toks,
    const std::unordered_map<std::string,std::unordered_set<std::string>> &def_index,
    const std::unordered_map<std::string,int> &recent_counts,
    double repeat_penalty)
{
    if (cands.empty()) return std::string();
    if (cands.size() == 1) return *cands[0];

    auto agg = aggregate_sets(prompt_toks, def_index);
    for (auto &r : resp_toks){
        auto it = def_index.find(r);
        if (it != def_index.end()) for (auto &d : it->second) agg.insert(d);
    }

    double best = -1e9;
    std::string best_tok;
    size_t M = cands.size();
    std::vector<double> scores(M, 0.0);

    #pragma omp parallel for schedule(static)
    for (ptrdiff_t i=0;i<static_cast<ptrdiff_t>(M);++i){
        std::unordered_set<std::string> candset;
        candset.insert(*cands[(size_t)i]);
        auto it = def_index.find(*cands[(size_t)i]);
        if (it != def_index.end()) for (auto &d : it->second) candset.insert(d);
        double s = jaccard_similarity(agg, candset);
        scores[(size_t)i] = s;
    }

    for (size_t i=0;i<M;++i){
        const std::string &tok = *cands[i];
        double s = scores[i];
        auto rc_it = recent_counts.find(tok);
        int cnt = (rc_it==recent_counts.end()? 0 : rc_it->second);
        double adjusted = s - repeat_penalty * static_cast<double>(cnt);
        if (adjusted > best || (adjusted == best && tok < best_tok)){
            best = adjusted;
            best_tok = tok;
        }
    }
    return best_tok;
}

// --------------------------- Response constructor (short units) ---------------

static std::vector<std::string> construct_response(KnowledgeBase &kb,
                                                  const std::vector<std::string> &prompt_toks,
                                                  size_t maxlen,
                                                  const std::unordered_map<std::string,std::unordered_set<std::string>> &def_index,
                                                  double repeat_penalty)
{
    std::vector<std::string> resp;
    if (prompt_toks.empty() || maxlen == 0) return resp;
    std::unordered_map<std::string,int> recent_counts;

    auto would_create_2_cycle = [&](const std::string &cand) -> bool {
        if (resp.size() < 2) return false;
        // check alternation: X Y X Y ... then candidate == X and last == Y
        const std::string &last = resp.back();
        const std::string &prev = resp[resp.size()-2];
        return (cand == prev && last == resp[resp.size()-3 < resp.size() ? resp.size()-3 : 0]);
        // this is a cheap conservative check; main guard is repeat_penalty + single-candidate rule
    };

    std::string last_printed;
    for (size_t step=0; step<maxlen; ++step){
        NextSet candidates;
        bool found = false;
        if (step==0){
            for (ssize_t p = static_cast<ssize_t>(prompt_toks.size())-1; p>=0; --p){
                auto opt = kb.lookup_by_string(prompt_toks[(size_t)p]);
                if (opt){ candidates = *opt; found = true; break; }
            }
        } else {
            auto opt = kb.lookup_by_string(last_printed);
            if (opt){ candidates = *opt; found = true; }
            else {
                for (ssize_t p = static_cast<ssize_t>(prompt_toks.size())-1; p>=0; --p){
                    auto opt2 = kb.lookup_by_string(prompt_toks[(size_t)p]);
                    if (opt2){ candidates = *opt2; found = true; break; }
                }
            }
        }
        if (!found || candidates.empty()) break;

        // If only one candidate and it already appeared, stop to avoid 1-cycle.
        if (candidates.size()==1){
            std::string only = *candidates[0];
            if (recent_counts[only] > 0) break;
            resp.push_back(only);
            recent_counts[only] += 1;
            last_printed = only;
            std::cout << only << ' ' << std::flush;   // print immediately
            continue;
        }

        // choose best with repeat penalty
        std::string chosen = best_candidate_by_similarity(candidates, prompt_toks, resp, def_index, recent_counts, repeat_penalty);
        if (chosen.empty()) break;

        // cheap 2-cycle avoider: if this would continue a trivial alternation, stop
        if (would_create_2_cycle(chosen)) break;

        resp.push_back(chosen);
        recent_counts[chosen] += 1;
        last_printed = chosen;
        std::cout << chosen << ' ' << std::flush;     // print immediately
    }
    return resp;
}

// --------------------------- Learning from files (short) -------------------

static void learn_from_file(KnowledgeBase &kb, const std::string &fname){
    std::ifstream ifs(fname);
    if (!ifs) return;
    std::string tok;
    std::string prev;
    bool have_prev = false;
    while (ifs >> tok){
        if (have_prev) kb.add_pair(prev, tok);
        prev = tok; have_prev = true;
    }
}

static void learn_files_parallel(KnowledgeBase &kb, const std::vector<std::string> &files){
    #pragma omp parallel for schedule(dynamic)
    for (ptrdiff_t i=0;i<static_cast<ptrdiff_t>(files.size());++i) learn_from_file(kb, files[(size_t)i]);
}

// --------------------------- Serialization (short functions) ----------------

static void save_kb_binary(const KnowledgeBase &kb, const std::string &fname){
    std::ofstream ofs(fname, std::ios::binary);
    if (!ofs) throw std::runtime_error("cannot open save file");

    // interned strings snapshot (must include all tokens used by def_index)
    std::vector<const std::string*> interned;
    interned.reserve(kb.interner.pool.size());
    for (auto &s : kb.interner.pool) interned.push_back(&s);

    uint64_t N = interned.size();
    ofs.write(reinterpret_cast<const char*>(&N), sizeof(N));
    for (auto p : interned){
        uint64_t L = p->size();
        ofs.write(reinterpret_cast<const char*>(&L), sizeof(L));
        ofs.write(p->data(), static_cast<std::streamsize>(L));
    }

    // edges
    uint64_t E = kb.next.size();
    ofs.write(reinterpret_cast<const char*>(&E), sizeof(E));
    for (auto &pr : kb.next){
        const std::string &key = *pr.first;
        auto it = std::find_if(interned.begin(), interned.end(), [&](const std::string* s){ return *s == key; });
        if (it == interned.end()) throw std::runtime_error("save index error");
        uint64_t key_idx = static_cast<uint64_t>(std::distance(interned.begin(), it));
        ofs.write(reinterpret_cast<const char*>(&key_idx), sizeof(key_idx));
        uint64_t M = pr.second.size();
        ofs.write(reinterpret_cast<const char*>(&M), sizeof(M));
        for (auto nxt : pr.second){
            auto it2 = std::find_if(interned.begin(), interned.end(), [&](const std::string* s){ return *s == *nxt; });
            if (it2 == interned.end()) throw std::runtime_error("save index error2");
            uint64_t v_idx = static_cast<uint64_t>(std::distance(interned.begin(), it2));
            ofs.write(reinterpret_cast<const char*>(&v_idx), sizeof(v_idx));
        }
    }

    // --- write definition expansion section ---
    uint64_t D = static_cast<uint64_t>(kb.def_depth);
    ofs.write(reinterpret_cast<const char*>(&D), sizeof(D));

    // def entries: number of keys with a stored expansion
    uint64_t K = kb.def_index.size();
    ofs.write(reinterpret_cast<const char*>(&K), sizeof(K));
    for (auto &pr : kb.def_index){
        // key index
        const std::string &key = *pr.first;
        auto it = std::find_if(interned.begin(), interned.end(), [&](const std::string* s){ return *s == key; });
        if (it == interned.end()) throw std::runtime_error("save def index error");
        uint64_t key_idx = static_cast<uint64_t>(std::distance(interned.begin(), it));
        ofs.write(reinterpret_cast<const char*>(&key_idx), sizeof(key_idx));

        // number of tokens
        uint64_t M = pr.second.size();
        ofs.write(reinterpret_cast<const char*>(&M), sizeof(M));
        for (auto tokp : pr.second){
            auto it2 = std::find_if(interned.begin(), interned.end(), [&](const std::string* s){ return *s == *tokp; });
            if (it2 == interned.end()) throw std::runtime_error("save def token index error");
            uint64_t v_idx = static_cast<uint64_t>(std::distance(interned.begin(), it2));
            ofs.write(reinterpret_cast<const char*>(&v_idx), sizeof(v_idx));
        }
    }

    safe_flush(ofs);
}

static void load_kb_binary(KnowledgeBase &kb, const std::string &fname, int cli_dict_depth){
    std::ifstream ifs(fname, std::ios::binary);
    if (!ifs) throw std::runtime_error("cannot open load file");

    uint64_t N;
    ifs.read(reinterpret_cast<char*>(&N), sizeof(N));
    std::vector<std::string> strings; strings.reserve((size_t)N);
    for (uint64_t i=0;i<N;++i){
        uint64_t L; ifs.read(reinterpret_cast<char*>(&L), sizeof(L));
        std::string s; s.resize((size_t)L);
        ifs.read(&s[0], static_cast<std::streamsize>(L));
        strings.push_back(std::move(s));
    }
    std::vector<StrPtr> ptrs; ptrs.reserve(strings.size());
    for (auto &s : strings) ptrs.push_back(kb.interner.intern(s));

    uint64_t E; ifs.read(reinterpret_cast<char*>(&E), sizeof(E));
    for (uint64_t i=0;i<E;++i){
        uint64_t key_idx; ifs.read(reinterpret_cast<char*>(&key_idx), sizeof(key_idx));
        uint64_t M; ifs.read(reinterpret_cast<char*>(&M), sizeof(M));
        StrPtr key_ptr = ptrs.at((size_t)key_idx);
        NextSet vec; vec.reserve((size_t)M);
        for (uint64_t j=0;j<M;++j){
            uint64_t v_idx; ifs.read(reinterpret_cast<char*>(&v_idx), sizeof(v_idx));
            vec.push_back(ptrs.at((size_t)v_idx));
        }
        kb.next.emplace(key_ptr, std::move(vec));
    }

    // read def-expansion section (new-format)
    uint64_t file_def_depth;
    ifs.read(reinterpret_cast<char*>(&file_def_depth), sizeof(file_def_depth));
    uint64_t K; ifs.read(reinterpret_cast<char*>(&K), sizeof(K));
    // populate kb.def_index from file
    {
        std::lock_guard<std::mutex> lk(kb.def_m);
        kb.def_index.clear();
        kb.def_depth = static_cast<int>(file_def_depth);
    }
    for (uint64_t i=0;i<K;++i){
        uint64_t key_idx; ifs.read(reinterpret_cast<char*>(&key_idx), sizeof(key_idx));
        uint64_t M; ifs.read(reinterpret_cast<char*>(&M), sizeof(M));
        std::vector<StrPtr> tokens; tokens.reserve((size_t)M);
        for (uint64_t j=0;j<M;++j){
            uint64_t v_idx; ifs.read(reinterpret_cast<char*>(&v_idx), sizeof(v_idx));
            tokens.push_back(ptrs.at((size_t)v_idx));
        }
        kb.def_index.emplace(ptrs.at((size_t)key_idx), std::move(tokens));
    }

    // If CLI requested a different dict depth, clear and recompute expansion for loaded words only
    if (cli_dict_depth != kb.def_depth){
        kb.set_def_depth(cli_dict_depth);
        // --- build deduplicated union of "words present" = saved strings (ptrs) ∪ KB words (keys and neighbors)
        std::vector<StrPtr> targets;
        targets.reserve(ptrs.size() + kb.next.size()*2);

        {
            std::unordered_set<StrPtr, PtrHash, PtrEq> seen;
            // include all strings from the saved file
            for (auto p : ptrs) {
                if (seen.insert(p).second) targets.push_back(p);
            }
            // include all words present in KB edges (keys and their neighbors)
            for (auto &pr : kb.next) {
                if (seen.insert(pr.first).second) targets.push_back(pr.first);
                for (auto v : pr.second) {
                    if (seen.insert(v).second) targets.push_back(v);
                }
            }
        }

        // --- recompute definition expansion for each target in parallel
        #pragma omp parallel for schedule(dynamic)
        for (ptrdiff_t i = 0; i < static_cast<ptrdiff_t>(targets.size()); ++i) {
            kb.ensure_def_for_interned(targets[(size_t)i]);
        }
    }
}

// --------------------------- CLI + Interactive loop (shorters) -----------

static void print_usage(const char *p){
    std::cout << "Usage: " << p << " [--maxlen N] [--save FILE] [--load-kb FILE] [--dict-depth D] [--learn f1 f2 ...] [--repeat-penalty P] [--help]\n";
    std::cout << "  --maxlen N          Maximum number of tokens constructed in a response.\n";
    std::cout << "  --save FILE         Save the knowledge-base and dictionary expansions to a binary file.\n";
    std::cout << "  --load-kb FILE      Load a previously saved knowledge-base (and dictionary expansions) from a binary file.\n";
    std::cout << "  --dict-depth D      Depth of dictionary-definition expansion used during learning.\n";
    std::cout << "  --learn f1 f2 ...   Learn from one or more text files to update the knowledge base.\n";
    std::cout << "  --repeat-penalty P  Penalize repeated tokens during response generation (higher values discourage repetition).\n";
    std::cout << "  --help              Show command-line interface options for ChatIPC usage.\n";
}

int main(int argc, char **argv){
    size_t maxlen = 100;
    std::string savefile;
    std::string load_txt;
    std::string load_kb;
    int dict_depth = 2;
    double repeat_penalty = 0.7; // default λ
    std::vector<std::string> learn_files;

    for (int i=1;i<argc;++i){
        std::string a = argv[i];
        if (a=="--help"){ print_usage(argv[0]); return 0; }
        if (a=="--maxlen" && i+1<argc){ maxlen = std::stoul(argv[++i]); continue; }
        if (a=="--save" && i+1<argc){ savefile = argv[++i]; continue; }
        if (a=="--load-kb" && i+1<argc){ load_kb = argv[++i]; continue; }
        if (a=="--dict-depth" && i+1<argc){ dict_depth = std::stoi(argv[++i]); continue; }
        if (a=="--repeat-penalty" && i+1<argc){ repeat_penalty = std::stod(argv[++i]); continue; }
        if (a=="--learn"){
            ++i;
            for (; i<argc; ++i){
                if (!argv[i]) break;
                std::string s = argv[i];
                if (!s.empty() && s[0]=='-'){ --i; break; }
                learn_files.push_back(s);
            }
            continue;
        }
        learn_files.push_back(a);
    }

    KnowledgeBase kb;

    // parse the embedded dictionary once for use by per-word expansion
    g_raw_dict = parse_dictionary_json();
    // set KB def depth (clears any previous expansion)
    kb.set_def_depth(dict_depth);


    if (!load_kb.empty()){
        try { load_kb_binary(kb, load_kb, dict_depth); std::cerr << "Loaded KB: " << load_kb << "\n"; }
        catch (const std::exception &e){ std::cerr << "Load KB error: " << e.what() << "\n"; }
    }

    if (!learn_files.empty()){
        std::cerr << "Learning from file/s (" << learn_files.size() << ") using threads=" << omp_get_max_threads() << "\n";
        learn_files_parallel(kb, learn_files);
    }

    std::string line;
    std::cout << "Ready. Enter prompts.\n";
    while (std::cout << "> " , std::getline(std::cin, line)){
        if (line.empty()){ std::cout << "\n"; continue; }
        auto prompt_toks = tokenize_whitespace(line);
        for (size_t i=1;i<prompt_toks.size();++i) kb.add_pair(prompt_toks[i-1], prompt_toks[i]);
        auto def_index = snapshot_def_index(kb);
        auto resp = construct_response(kb, prompt_toks, maxlen, def_index, repeat_penalty);
        std::cout << "\n";
        if (!resp.empty()){
            std::vector<std::string> combined = prompt_toks;
            combined.insert(combined.end(), resp.begin(), resp.end());
            for (size_t i=1;i<combined.size();++i) kb.add_pair(combined[i-1], combined[i]);
        }
        if (!savefile.empty()){
            try { save_kb_binary(kb, savefile); std::cerr << "Saved KB: " << savefile << "\n"; }
            catch (const std::exception &e){ std::cerr << "Save KB error: " << e.what() << "\n"; }
        }
    }

    if (!savefile.empty()){
        try { save_kb_binary(kb, savefile); std::cerr << "Saved KB: " << savefile << "\n"; }
        catch (const std::exception &e){ std::cerr << "Save KB error: " << e.what() << "\n"; }
    }

    return 0;
}