archivmail/internal/index/xapian_wrapper.cpp

#include "xapian_wrapper.h"
#include <xapian.h>
#include <cstring>
#include <cstdlib>
#include <string>
#include <sstream>
#include <stdexcept>

struct XapianDB {
    Xapian::WritableDatabase* wdb;
    Xapian::Database*         rdb;
    bool writable;
};

static char* dup_error(const std::string& msg) {
    char* s = (char*)malloc(msg.size() + 1);
    if (s) memcpy(s, msg.c_str(), msg.size() + 1);
    return s;
}

extern "C" {

XapianDB* xapian_open(const char* path, int writable, char** err) {
    try {
        XapianDB* db = new XapianDB{nullptr, nullptr, (bool)writable};
        if (writable) {
            db->wdb = new Xapian::WritableDatabase(path, Xapian::DB_CREATE_OR_OPEN);
        } else {
            db->rdb = new Xapian::Database(path);
        }
        return db;
    } catch (const std::exception& e) {
        if (err) *err = dup_error(e.what());
        return nullptr;
    }
}

void xapian_close(XapianDB* db) {
    if (!db) return;
    if (db->wdb) { db->wdb->close(); delete db->wdb; }
    if (db->rdb) { db->rdb->close(); delete db->rdb; }
    delete db;
}

int xapian_index(XapianDB* db, const char* id, const char* from,
                  const char* to, const char* subject, const char* body,
                  long long timestamp, int has_attachment, char** err) {
    try {
        Xapian::Document doc;
        Xapian::TermGenerator gen;
        gen.set_document(doc);
        gen.set_stemmer(Xapian::Stem("en"));

        // Prefix-indexed fields for filtering
        gen.index_text(from,    1, "XF");
        gen.index_text(to,      1, "XT");
        gen.index_text(subject, 1, "XS");

        // Free-text indexed fields
        gen.index_text(subject);
        gen.increase_termpos();
        gen.index_text(body);
        gen.increase_termpos();
        gen.index_text(from);
        gen.increase_termpos();
        gen.index_text(to);

        // Boolean term for attachment filter
        if (has_attachment) {
            doc.add_boolean_term("XHA");
        }

        // Store timestamp for date range queries (value slot 0)
        doc.add_value(0, Xapian::sortable_serialise((double)timestamp));

        // Store ID as document data
        doc.set_data(id);
        doc.add_boolean_term(std::string("Q") + id);

        db->wdb->replace_document(std::string("Q") + id, doc);
        db->wdb->commit();
        return 0;
    } catch (const std::exception& e) {
        if (err) *err = dup_error(e.what());
        return -1;
    }
}

int xapian_delete(XapianDB* db, const char* id, char** err) {
    try {
        db->wdb->delete_document(std::string("Q") + id);
        db->wdb->commit();
        return 0;
    } catch (const std::exception& e) {
        if (err) *err = dup_error(e.what());
        return -1;
    }
}

char* xapian_search(XapianDB* db, const char* query_str,
                     const char* from_filter, const char* own_email,
                     const char* to_filter,
                     long long date_from, long long date_to,
                     int offset, int limit,
                     int sort_mode, int has_attachment,
                     char** err) {
    try {
        Xapian::Database& xdb = db->wdb ? (Xapian::Database&)*db->wdb : *db->rdb;
        Xapian::Enquire enquire(xdb);

        Xapian::Query main_query;

        // Full-text query
        if (query_str && query_str[0] != '\0') {
            Xapian::QueryParser qp;
            qp.set_database(xdb);
            qp.set_stemmer(Xapian::Stem("en"));
            qp.set_stemming_strategy(Xapian::QueryParser::STEM_SOME);
            qp.add_prefix("from",    "XF");
            qp.add_prefix("to",      "XT");
            qp.add_prefix("subject", "XS");
            main_query = qp.parse_query(query_str,
                Xapian::QueryParser::FLAG_DEFAULT |
                Xapian::QueryParser::FLAG_PARTIAL);
        } else {
            main_query = Xapian::Query::MatchAll;
        }

        // From filter
        if (from_filter && from_filter[0] != '\0') {
            Xapian::QueryParser qp;
            qp.set_database(xdb);
            Xapian::Query fq = qp.parse_query(from_filter,
                Xapian::QueryParser::FLAG_DEFAULT, "XF");
            main_query = Xapian::Query(Xapian::Query::OP_AND, main_query, fq);
        }

        // OwnEmail filter: (from=own OR to=own)
        if (own_email && own_email[0] != '\0') {
            Xapian::QueryParser qp;
            qp.set_database(xdb);
            Xapian::Query fq = qp.parse_query(own_email,
                Xapian::QueryParser::FLAG_DEFAULT, "XF");
            Xapian::Query tq = qp.parse_query(own_email,
                Xapian::QueryParser::FLAG_DEFAULT, "XT");
            Xapian::Query owq(Xapian::Query::OP_OR, fq, tq);
            main_query = Xapian::Query(Xapian::Query::OP_AND, main_query, owq);
        }

        // To filter
        if (to_filter && to_filter[0] != '\0') {
            Xapian::QueryParser qp;
            qp.set_database(xdb);
            Xapian::Query tq = qp.parse_query(to_filter,
                Xapian::QueryParser::FLAG_DEFAULT, "XT");
            main_query = Xapian::Query(Xapian::Query::OP_AND, main_query, tq);
        }

        // Date range
        if (date_from > 0 || date_to > 0) {
            double lo = date_from > 0 ? (double)date_from : 0.0;
            double hi = date_to   > 0 ? (double)date_to   : 1e18;
            Xapian::Query drq(Xapian::Query::OP_VALUE_RANGE, 0,
                Xapian::sortable_serialise(lo),
                Xapian::sortable_serialise(hi));
            main_query = Xapian::Query(Xapian::Query::OP_AND, main_query, drq);
        }

        // Attachment filter
        if (has_attachment == 1) {
            Xapian::Query aq("XHA");
            main_query = Xapian::Query(Xapian::Query::OP_AND, main_query, aq);
        } else if (has_attachment == -1) {
            Xapian::Query aq("XHA");
            main_query = Xapian::Query(Xapian::Query::OP_AND_NOT, main_query, aq);
        }

        enquire.set_query(main_query);

        // Sort mode: 0=relevance, 1=date_desc, 2=date_asc
        if (sort_mode == 2) {
            enquire.set_sort_by_value(0, false); // date ascending
        } else if (sort_mode == 0 && query_str && query_str[0] != '\0') {
            // relevance: default BM25 ranking (no explicit sort)
        } else {
            enquire.set_sort_by_value(0, true); // date descending (default)
        }

        // Get total count
        Xapian::MSet all = enquire.get_mset(0, xdb.get_doccount());
        int total = (int)all.get_matches_estimated();

        // Get page
        Xapian::MSet mset = enquire.get_mset(offset, limit);

        std::ostringstream json;
        json << "{\"total\":" << total << ",\"hits\":[";
        bool first = true;
        for (auto it = mset.begin(); it != mset.end(); ++it) {
            if (!first) json << ",";
            first = false;
            std::string id = it.get_document().get_data();
            double score   = it.get_weight();
            json << "{\"id\":\"" << id << "\",\"score\":" << score << "}";
        }
        json << "]}";

        std::string result = json.str();
        char* out = (char*)malloc(result.size() + 1);
        memcpy(out, result.c_str(), result.size() + 1);
        return out;

    } catch (const std::exception& e) {
        if (err) *err = dup_error(e.what());
        return nullptr;
    }
}

void xapian_free_string(char* s) {
    free(s);
}

} // extern "C"