isotree 0.1.4 → 0.1.5

data/vendor/cereal/include/cereal/version.hpp

@@ -0,0 +1,52 @@
+/*! \file version.hpp
+    \brief Macros to detect cereal version
+
+    These macros can assist in determining the version of cereal. Be
+    warned that cereal is not guaranteed to be compatible across
+    different versions. For more information on releases of cereal,
+    see https://github.com/USCiLab/cereal/releases.
+
+    \ingroup utility */
+/*
+  Copyright (c) 2018, Shane Grant
+  All rights reserved.
+
+  Redistribution and use in source and binary forms, with or without
+  modification, are permitted provided that the following conditions are met:
+      * Redistributions of source code must retain the above copyright
+        notice, this list of conditions and the following disclaimer.
+      * Redistributions in binary form must reproduce the above copyright
+        notice, this list of conditions and the following disclaimer in the
+        documentation and/or other materials provided with the distribution.
+      * Neither the name of cereal nor the
+        names of its contributors may be used to endorse or promote products
+        derived from this software without specific prior written permission.
+
+  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+  ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+  WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+  DISCLAIMED. IN NO EVENT SHALL RANDOLPH VOORHIES OR SHANE GRANT BE LIABLE FOR ANY
+  DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+  (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+  LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+  ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+  (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#ifndef CEREAL_VERSION_HPP_
+#define CEREAL_VERSION_HPP_
+
+//! The major version
+#define CEREAL_VERSION_MAJOR 1
+//! The minor version
+#define CEREAL_VERSION_MINOR 3
+//! The patch version
+#define CEREAL_VERSION_PATCH 0
+
+//! The full version as a single number
+#define CEREAL_VERSION (CEREAL_VERSION_MAJOR * 10000 \
+                        + CEREAL_VERSION_MINOR * 100 \
+                        + CEREAL_VERSION_PATCH)
+
+#endif // CEREAL_VERSION_HPP_

data/vendor/isotree/LICENSE

@@ -1,6 +1,6 @@
 BSD 2-Clause License
 
-Copyright (c) 2019, David Cortes
+Copyright (c) 2020, David Cortes
 All rights reserved.
 
 Redistribution and use in source and binary forms, with or without

data/vendor/isotree/README.md

@@ -42,7 +42,8 @@ There's already many available implementations of isolation forests for both Pyt
 * Uses exact formula (not approximation as others do) for harmonic numbers at lower sample and remainder sizes.
 * Can fit trees incrementally to user-provided data samples.
 * Produces serializable model objects with reasonable file sizes.
-* Fast and multi-threaded C++ code. Can be wrapped in languages other than Python and R.
+* Can translate the generated trees into SQL statements.
+* Fast and multi-threaded C++ code. Can be wrapped in languages other than Python/R/Ruby.
 
 (Note that categoricals, NAs, and density-like sample weights, are treated heuristically with different options as there is no single logical extension of the original idea to them, and having them present might degrade performance/accuracy for regular numerical non-missing observations)
 

data/vendor/isotree/src/RcppExports.cpp

@@ -50,8 +50,8 @@ BEGIN_RCPP
 END_RCPP
 }
 // fit_model
-Rcpp::List fit_model(Rcpp::NumericVector X_num, Rcpp::IntegerVector X_cat, Rcpp::IntegerVector ncat, Rcpp::NumericVector Xc, Rcpp::IntegerVector Xc_ind, Rcpp::IntegerVector Xc_indptr, Rcpp::NumericVector sample_weights, Rcpp::NumericVector col_weights, size_t nrows, size_t ncols_numeric, size_t ncols_categ, size_t ndim, size_t ntry, Rcpp::CharacterVector coef_type, bool coef_by_prop, bool with_replacement, bool weight_as_sample, size_t sample_size, size_t ntrees, size_t max_depth, bool limit_depth, bool penalize_range, bool calc_dist, bool standardize_dist, bool sq_dist, bool calc_depth, bool standardize_depth, bool weigh_by_kurt, double prob_pick_by_gain_avg, double prob_split_by_gain_avg, double prob_pick_by_gain_pl, double prob_split_by_gain_pl, double min_gain, Rcpp::CharacterVector cat_split_type, Rcpp::CharacterVector new_cat_action, Rcpp::CharacterVector missing_action, bool all_perm, bool build_imputer, bool output_imputations, size_t min_imp_obs, Rcpp::CharacterVector depth_imp, Rcpp::CharacterVector weigh_imp_rows, int random_seed, int nthreads);
-RcppExport SEXP _isotree_fit_model(SEXP X_numSEXP, SEXP X_catSEXP, SEXP ncatSEXP, SEXP XcSEXP, SEXP Xc_indSEXP, SEXP Xc_indptrSEXP, SEXP sample_weightsSEXP, SEXP col_weightsSEXP, SEXP nrowsSEXP, SEXP ncols_numericSEXP, SEXP ncols_categSEXP, SEXP ndimSEXP, SEXP ntrySEXP, SEXP coef_typeSEXP, SEXP coef_by_propSEXP, SEXP with_replacementSEXP, SEXP weight_as_sampleSEXP, SEXP sample_sizeSEXP, SEXP ntreesSEXP, SEXP max_depthSEXP, SEXP limit_depthSEXP, SEXP penalize_rangeSEXP, SEXP calc_distSEXP, SEXP standardize_distSEXP, SEXP sq_distSEXP, SEXP calc_depthSEXP, SEXP standardize_depthSEXP, SEXP weigh_by_kurtSEXP, SEXP prob_pick_by_gain_avgSEXP, SEXP prob_split_by_gain_avgSEXP, SEXP prob_pick_by_gain_plSEXP, SEXP prob_split_by_gain_plSEXP, SEXP min_gainSEXP, SEXP cat_split_typeSEXP, SEXP new_cat_actionSEXP, SEXP missing_actionSEXP, SEXP all_permSEXP, SEXP build_imputerSEXP, SEXP output_imputationsSEXP, SEXP min_imp_obsSEXP, SEXP depth_impSEXP, SEXP weigh_imp_rowsSEXP, SEXP random_seedSEXP, SEXP nthreadsSEXP) {
+Rcpp::List fit_model(Rcpp::NumericVector X_num, Rcpp::IntegerVector X_cat, Rcpp::IntegerVector ncat, Rcpp::NumericVector Xc, Rcpp::IntegerVector Xc_ind, Rcpp::IntegerVector Xc_indptr, Rcpp::NumericVector sample_weights, Rcpp::NumericVector col_weights, size_t nrows, size_t ncols_numeric, size_t ncols_categ, size_t ndim, size_t ntry, Rcpp::CharacterVector coef_type, bool coef_by_prop, bool with_replacement, bool weight_as_sample, size_t sample_size, size_t ntrees, size_t max_depth, bool limit_depth, bool penalize_range, bool calc_dist, bool standardize_dist, bool sq_dist, bool calc_depth, bool standardize_depth, bool weigh_by_kurt, double prob_pick_by_gain_avg, double prob_split_by_gain_avg, double prob_pick_by_gain_pl, double prob_split_by_gain_pl, double min_gain, Rcpp::CharacterVector cat_split_type, Rcpp::CharacterVector new_cat_action, Rcpp::CharacterVector missing_action, bool all_perm, bool build_imputer, bool output_imputations, size_t min_imp_obs, Rcpp::CharacterVector depth_imp, Rcpp::CharacterVector weigh_imp_rows, int random_seed, bool handle_interrupt, int nthreads);
+RcppExport SEXP _isotree_fit_model(SEXP X_numSEXP, SEXP X_catSEXP, SEXP ncatSEXP, SEXP XcSEXP, SEXP Xc_indSEXP, SEXP Xc_indptrSEXP, SEXP sample_weightsSEXP, SEXP col_weightsSEXP, SEXP nrowsSEXP, SEXP ncols_numericSEXP, SEXP ncols_categSEXP, SEXP ndimSEXP, SEXP ntrySEXP, SEXP coef_typeSEXP, SEXP coef_by_propSEXP, SEXP with_replacementSEXP, SEXP weight_as_sampleSEXP, SEXP sample_sizeSEXP, SEXP ntreesSEXP, SEXP max_depthSEXP, SEXP limit_depthSEXP, SEXP penalize_rangeSEXP, SEXP calc_distSEXP, SEXP standardize_distSEXP, SEXP sq_distSEXP, SEXP calc_depthSEXP, SEXP standardize_depthSEXP, SEXP weigh_by_kurtSEXP, SEXP prob_pick_by_gain_avgSEXP, SEXP prob_split_by_gain_avgSEXP, SEXP prob_pick_by_gain_plSEXP, SEXP prob_split_by_gain_plSEXP, SEXP min_gainSEXP, SEXP cat_split_typeSEXP, SEXP new_cat_actionSEXP, SEXP missing_actionSEXP, SEXP all_permSEXP, SEXP build_imputerSEXP, SEXP output_imputationsSEXP, SEXP min_imp_obsSEXP, SEXP depth_impSEXP, SEXP weigh_imp_rowsSEXP, SEXP random_seedSEXP, SEXP handle_interruptSEXP, SEXP nthreadsSEXP) {
 BEGIN_RCPP
     Rcpp::RObject rcpp_result_gen;
     Rcpp::RNGScope rcpp_rngScope_gen;
@@ -98,8 +98,9 @@ BEGIN_RCPP
     Rcpp::traits::input_parameter< Rcpp::CharacterVector >::type depth_imp(depth_impSEXP);
     Rcpp::traits::input_parameter< Rcpp::CharacterVector >::type weigh_imp_rows(weigh_imp_rowsSEXP);
     Rcpp::traits::input_parameter< int >::type random_seed(random_seedSEXP);
+    Rcpp::traits::input_parameter< bool >::type handle_interrupt(handle_interruptSEXP);
     Rcpp::traits::input_parameter< int >::type nthreads(nthreadsSEXP);
-    rcpp_result_gen = Rcpp::wrap(fit_model(X_num, X_cat, ncat, Xc, Xc_ind, Xc_indptr, sample_weights, col_weights, nrows, ncols_numeric, ncols_categ, ndim, ntry, coef_type, coef_by_prop, with_replacement, weight_as_sample, sample_size, ntrees, max_depth, limit_depth, penalize_range, calc_dist, standardize_dist, sq_dist, calc_depth, standardize_depth, weigh_by_kurt, prob_pick_by_gain_avg, prob_split_by_gain_avg, prob_pick_by_gain_pl, prob_split_by_gain_pl, min_gain, cat_split_type, new_cat_action, missing_action, all_perm, build_imputer, output_imputations, min_imp_obs, depth_imp, weigh_imp_rows, random_seed, nthreads));
+    rcpp_result_gen = Rcpp::wrap(fit_model(X_num, X_cat, ncat, Xc, Xc_ind, Xc_indptr, sample_weights, col_weights, nrows, ncols_numeric, ncols_categ, ndim, ntry, coef_type, coef_by_prop, with_replacement, weight_as_sample, sample_size, ntrees, max_depth, limit_depth, penalize_range, calc_dist, standardize_dist, sq_dist, calc_depth, standardize_depth, weigh_by_kurt, prob_pick_by_gain_avg, prob_split_by_gain_avg, prob_pick_by_gain_pl, prob_split_by_gain_pl, min_gain, cat_split_type, new_cat_action, missing_action, all_perm, build_imputer, output_imputations, min_imp_obs, depth_imp, weigh_imp_rows, random_seed, handle_interrupt, nthreads));
     return rcpp_result_gen;
 END_RCPP
 }
@@ -245,19 +246,58 @@ BEGIN_RCPP
     return rcpp_result_gen;
 END_RCPP
 }
+// model_to_sql
+Rcpp::ListOf<Rcpp::CharacterVector> model_to_sql(SEXP model_R_ptr, bool is_extended, Rcpp::CharacterVector numeric_colanmes, Rcpp::CharacterVector categ_colnames, Rcpp::ListOf<Rcpp::CharacterVector> categ_levels, bool output_tree_num, bool single_tree, size_t tree_num, int nthreads);
+RcppExport SEXP _isotree_model_to_sql(SEXP model_R_ptrSEXP, SEXP is_extendedSEXP, SEXP numeric_colanmesSEXP, SEXP categ_colnamesSEXP, SEXP categ_levelsSEXP, SEXP output_tree_numSEXP, SEXP single_treeSEXP, SEXP tree_numSEXP, SEXP nthreadsSEXP) {
+BEGIN_RCPP
+    Rcpp::RObject rcpp_result_gen;
+    Rcpp::RNGScope rcpp_rngScope_gen;
+    Rcpp::traits::input_parameter< SEXP >::type model_R_ptr(model_R_ptrSEXP);
+    Rcpp::traits::input_parameter< bool >::type is_extended(is_extendedSEXP);
+    Rcpp::traits::input_parameter< Rcpp::CharacterVector >::type numeric_colanmes(numeric_colanmesSEXP);
+    Rcpp::traits::input_parameter< Rcpp::CharacterVector >::type categ_colnames(categ_colnamesSEXP);
+    Rcpp::traits::input_parameter< Rcpp::ListOf<Rcpp::CharacterVector> >::type categ_levels(categ_levelsSEXP);
+    Rcpp::traits::input_parameter< bool >::type output_tree_num(output_tree_numSEXP);
+    Rcpp::traits::input_parameter< bool >::type single_tree(single_treeSEXP);
+    Rcpp::traits::input_parameter< size_t >::type tree_num(tree_numSEXP);
+    Rcpp::traits::input_parameter< int >::type nthreads(nthreadsSEXP);
+    rcpp_result_gen = Rcpp::wrap(model_to_sql(model_R_ptr, is_extended, numeric_colanmes, categ_colnames, categ_levels, output_tree_num, single_tree, tree_num, nthreads));
+    return rcpp_result_gen;
+END_RCPP
+}
+// model_to_sql_with_select_from
+Rcpp::CharacterVector model_to_sql_with_select_from(SEXP model_R_ptr, bool is_extended, Rcpp::CharacterVector numeric_colanmes, Rcpp::CharacterVector categ_colnames, Rcpp::ListOf<Rcpp::CharacterVector> categ_levels, Rcpp::CharacterVector table_from, Rcpp::CharacterVector select_as, int nthreads);
+RcppExport SEXP _isotree_model_to_sql_with_select_from(SEXP model_R_ptrSEXP, SEXP is_extendedSEXP, SEXP numeric_colanmesSEXP, SEXP categ_colnamesSEXP, SEXP categ_levelsSEXP, SEXP table_fromSEXP, SEXP select_asSEXP, SEXP nthreadsSEXP) {
+BEGIN_RCPP
+    Rcpp::RObject rcpp_result_gen;
+    Rcpp::RNGScope rcpp_rngScope_gen;
+    Rcpp::traits::input_parameter< SEXP >::type model_R_ptr(model_R_ptrSEXP);
+    Rcpp::traits::input_parameter< bool >::type is_extended(is_extendedSEXP);
+    Rcpp::traits::input_parameter< Rcpp::CharacterVector >::type numeric_colanmes(numeric_colanmesSEXP);
+    Rcpp::traits::input_parameter< Rcpp::CharacterVector >::type categ_colnames(categ_colnamesSEXP);
+    Rcpp::traits::input_parameter< Rcpp::ListOf<Rcpp::CharacterVector> >::type categ_levels(categ_levelsSEXP);
+    Rcpp::traits::input_parameter< Rcpp::CharacterVector >::type table_from(table_fromSEXP);
+    Rcpp::traits::input_parameter< Rcpp::CharacterVector >::type select_as(select_asSEXP);
+    Rcpp::traits::input_parameter< int >::type nthreads(nthreadsSEXP);
+    rcpp_result_gen = Rcpp::wrap(model_to_sql_with_select_from(model_R_ptr, is_extended, numeric_colanmes, categ_colnames, categ_levels, table_from, select_as, nthreads));
+    return rcpp_result_gen;
+END_RCPP
+}
 
 static const R_CallMethodDef CallEntries[] = {
     {"_isotree_deserialize_IsoForest", (DL_FUNC) &_isotree_deserialize_IsoForest, 1},
     {"_isotree_deserialize_ExtIsoForest", (DL_FUNC) &_isotree_deserialize_ExtIsoForest, 1},
     {"_isotree_deserialize_Imputer", (DL_FUNC) &_isotree_deserialize_Imputer, 1},
     {"_isotree_check_null_ptr_model", (DL_FUNC) &_isotree_check_null_ptr_model, 1},
-    {"_isotree_fit_model", (DL_FUNC) &_isotree_fit_model, 44},
+    {"_isotree_fit_model", (DL_FUNC) &_isotree_fit_model, 45},
     {"_isotree_fit_tree", (DL_FUNC) &_isotree_fit_tree, 35},
     {"_isotree_predict_iso", (DL_FUNC) &_isotree_predict_iso, 15},
     {"_isotree_dist_iso", (DL_FUNC) &_isotree_dist_iso, 16},
     {"_isotree_impute_iso", (DL_FUNC) &_isotree_impute_iso, 10},
     {"_isotree_get_n_nodes", (DL_FUNC) &_isotree_get_n_nodes, 3},
     {"_isotree_append_trees_from_other", (DL_FUNC) &_isotree_append_trees_from_other, 5},
+    {"_isotree_model_to_sql", (DL_FUNC) &_isotree_model_to_sql, 9},
+    {"_isotree_model_to_sql_with_select_from", (DL_FUNC) &_isotree_model_to_sql_with_select_from, 8},
     {NULL, NULL, 0}
 };
 

data/vendor/isotree/src/Rwrapper.cpp

@@ -22,7 +22,7 @@
 *     [9] Cortes, David. "Imputing missing values with unsupervised random trees." arXiv preprint arXiv:1911.06646 (2019).
 * 
 *     BSD 2-Clause License
-*     Copyright (c) 2019, David Cortes
+*     Copyright (c) 2020, David Cortes
 *     All rights reserved.
 *     Redistribution and use in source and binary forms, with or without
 *     modification, are permitted provided that the following conditions are met:
@@ -42,6 +42,7 @@
 *     OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 *     OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */
+#ifdef _FOR_R
 
 #include <Rcpp.h>
 // [[Rcpp::plugins(cpp11)]]
@@ -69,52 +70,48 @@ Rcpp::RawVector serialize_cpp_obj(T *model_outputs)
         oarchive(*model_outputs);
     }
     ss.seekg(0, ss.end);
-    Rcpp::RawVector retval(ss.tellg());
+    /* Checking for potential integer overflows */
+    std::stringstream::pos_type vec_size = ss.tellg();
+    if (vec_size <= 0) {
+        Rcpp::Rcerr << "Error: model is too big to serialize, resulting object will not be usable.\n" << std::endl;
+        return Rcpp::RawVector();
+    }
+    Rcpp::RawVector retval((size_t) vec_size);
     ss.seekg(0, ss.beg);
     ss.read(reinterpret_cast<char*>(&retval[0]), retval.size());
     return retval;
 }
 
-// [[Rcpp::export]]
-SEXP deserialize_IsoForest(Rcpp::RawVector src)
+template <class T>
+SEXP deserialize_cpp_obj(Rcpp::RawVector src)
 {
     std::stringstream ss;
     ss.write(reinterpret_cast<char*>(&src[0]), src.size());
     ss.seekg(0, ss.beg);
-    std::unique_ptr<IsoForest> model_outputs = std::unique_ptr<IsoForest>(new IsoForest());
+    std::unique_ptr<T> model_outputs = std::unique_ptr<T>(new T());
     {
         cereal::BinaryInputArchive iarchive(ss);
         iarchive(*model_outputs);
     }
-    return Rcpp::XPtr<IsoForest>(model_outputs.release(), true);
+    return Rcpp::XPtr<T>(model_outputs.release(), true);
+}
+
+// [[Rcpp::export]]
+SEXP deserialize_IsoForest(Rcpp::RawVector src)
+{
+    return deserialize_cpp_obj<IsoForest>(src);
 }
 
 // [[Rcpp::export]]
 SEXP deserialize_ExtIsoForest(Rcpp::RawVector src)
 {
-    std::stringstream ss;
-    ss.write(reinterpret_cast<char*>(&src[0]), src.size());
-    ss.seekg(0, ss.beg);
-    std::unique_ptr<ExtIsoForest> model_outputs = std::unique_ptr<ExtIsoForest>(new ExtIsoForest());
-    {
-        cereal::BinaryInputArchive iarchive(ss);
-        iarchive(*model_outputs);
-    }
-    return Rcpp::XPtr<ExtIsoForest>(model_outputs.release(), true);
+    return deserialize_cpp_obj<ExtIsoForest>(src);
 }
 
 // [[Rcpp::export]]
 SEXP deserialize_Imputer(Rcpp::RawVector src)
 {
-    std::stringstream ss;
-    ss.write(reinterpret_cast<char*>(&src[0]), src.size());
-    ss.seekg(0, ss.beg);
-    std::unique_ptr<Imputer> imputer = std::unique_ptr<Imputer>(new Imputer());
-    {
-        cereal::BinaryInputArchive iarchive(ss);
-        iarchive(*imputer);
-    }
-    return Rcpp::XPtr<Imputer>(imputer.release(), true);
+    return deserialize_cpp_obj<Imputer>(src);
 }
 
 // [[Rcpp::export]]
@@ -139,7 +136,7 @@ double* set_R_nan_as_C_nan(double *x, size_t n, int nthreads)
     for (size_t_for i = 0; i < n; i++)
         if (isnan(x[i]))
             x[i] = NAN;
-    return &x[0];
+    return x;
 }
 
 // [[Rcpp::export]]
@@ -157,7 +154,7 @@ Rcpp::List fit_model(Rcpp::NumericVector X_num, Rcpp::IntegerVector X_cat, Rcpp:
                      Rcpp::CharacterVector missing_action, bool all_perm,
                      bool build_imputer, bool output_imputations, size_t min_imp_obs,
                      Rcpp::CharacterVector depth_imp, Rcpp::CharacterVector weigh_imp_rows,
-                     int random_seed, int nthreads)
+                     int random_seed, bool handle_interrupt, int nthreads)
 {
     double*     numeric_data_ptr    =  NULL;
     int*        categ_data_ptr      =  NULL;
@@ -303,7 +300,7 @@ Rcpp::List fit_model(Rcpp::NumericVector X_num, Rcpp::IntegerVector X_cat, Rcpp:
                 cat_split_type_C, new_cat_action_C,
                 all_perm, imputer_ptr.get(), min_imp_obs,
                 depth_imp_C, weigh_imp_rows_C, output_imputations,
-                (uint64_t) random_seed, nthreads);
+                (uint64_t) random_seed, handle_interrupt, nthreads);
 
     if (ret_val == EXIT_FAILURE)
     {
@@ -313,11 +310,19 @@ Rcpp::List fit_model(Rcpp::NumericVector X_num, Rcpp::IntegerVector X_cat, Rcpp:
     if (calc_dist && sq_dist)
         tmat_to_dense(tmat_ptr, dmat_ptr, nrows, !standardize_dist);
 
+    bool serialization_failed = false;
     Rcpp::RawVector serialized_obj;
     if (ndim == 1)
         serialized_obj  =  serialize_cpp_obj(model_ptr.get());
     else
         serialized_obj  =  serialize_cpp_obj(ext_model_ptr.get());
+    if (!serialized_obj.size()) serialization_failed = true;
+    if (serialization_failed) {
+        if (ndim == 1)
+            model_ptr.reset();
+        else
+            ext_model_ptr.reset();
+    }
 
     Rcpp::List outp = Rcpp::List::create(
                 Rcpp::_["serialized_obj"] = serialized_obj,
@@ -326,18 +331,33 @@ Rcpp::List fit_model(Rcpp::NumericVector X_num, Rcpp::IntegerVector X_cat, Rcpp:
                 Rcpp::_["dmat"]      = dmat
         );
 
-    if (ndim == 1)
-        outp["model_ptr"]   =  Rcpp::XPtr<IsoForest>(model_ptr.release(), true);
-    else
-        outp["model_ptr"]   =  Rcpp::XPtr<ExtIsoForest>(ext_model_ptr.release(), true);
+    if (!serialization_failed)
+    {
+        if (ndim == 1)
+            outp["model_ptr"]   =  Rcpp::XPtr<IsoForest>(model_ptr.release(), true);
+        else
+            outp["model_ptr"]   =  Rcpp::XPtr<ExtIsoForest>(ext_model_ptr.release(), true);
+    } else
+        outp["model_ptr"] = R_NilValue;
 
-    if (build_imputer)
+    if (build_imputer && !serialization_failed)
     {
         outp["imputer_ser"] =  serialize_cpp_obj(imputer_ptr.get());
-        outp["imputer_ptr"] =  Rcpp::XPtr<Imputer>(imputer_ptr.release(), true);
+        if (!Rf_xlength(outp["imputer_ser"]))
+        {
+            serialization_failed = true;
+            imputer_ptr.reset();
+            if (ndim == 1)
+                model_ptr.reset();
+            else
+                ext_model_ptr.reset();
+            outp["imputer_ptr"]  =  R_NilValue;
+            outp["model_ptr"]    =  R_NilValue;
+        } else
+            outp["imputer_ptr"] =  Rcpp::XPtr<Imputer>(imputer_ptr.release(), true);
     }
 
-    if (output_imputations)
+    if (output_imputations && !serialization_failed)
     {
         outp["imputed_num"] = Rcpp::NumericVector(Xcpp.begin(), Xcpp.end());
         outp["imputed_cat"] = X_cat;
@@ -517,21 +537,25 @@ void predict_iso(SEXP model_R_ptr, Rcpp::NumericVector outp, Rcpp::IntegerVector
 
     if (X_cat.size())
     {
-        categ_data_ptr   =  &X_cat[0];
+        categ_data_ptr    =  &X_cat[0];
     }
 
-    if (Xc.size())
+    if (Xc_indptr.size())
     {
-        Xc_ptr         =  &Xc[0];
-        Xc_ind_ptr     =  &Xc_ind[0];
-        Xc_indptr_ptr  =  &Xc_indptr[0];
+        if (Xc.size())
+            Xc_ptr         =  &Xc[0];
+        if (Xc_ind.size())
+            Xc_ind_ptr     =  &Xc_ind[0];
+        Xc_indptr_ptr      =  &Xc_indptr[0];
     }
 
-    if (Xr.size())
+    if (Xr_indptr.size())
     {
-        Xr_ptr         =  &Xr[0];
-        Xr_ind_ptr     =  &Xr_ind[0];
-        Xr_indptr_ptr  =  &Xr_indptr[0];
+        if (Xr.size())
+            Xr_ptr         =  &Xr[0];
+        if (Xr_ind.size())
+            Xr_ind_ptr     =  &Xr_ind[0];
+        Xr_indptr_ptr      =  &Xr_indptr[0];
     }
 
     if (tree_num.size())
@@ -539,7 +563,7 @@ void predict_iso(SEXP model_R_ptr, Rcpp::NumericVector outp, Rcpp::IntegerVector
         tree_num_ptr = &tree_num[0];
     }
 
-    double* depths_ptr    =  &outp[0];
+    double* depths_ptr =  &outp[0];
 
     IsoForest*     model_ptr      =  NULL;
     ExtIsoForest*  ext_model_ptr  =  NULL;
@@ -592,10 +616,12 @@ void dist_iso(SEXP model_R_ptr, Rcpp::NumericVector tmat, Rcpp::NumericVector dm
         categ_data_ptr    =  &X_cat[0];
     }
 
-    if (Xc.size())
+    if (Xc_indptr.size())
     {
-        Xc_ptr         =  &Xc[0];
-        Xc_ind_ptr     =  &Xc_ind[0];
+        if (Xc.size())
+            Xc_ptr         =  &Xc[0];
+        if (Xc_ind.size())
+            Xc_ind_ptr     =  &Xc_ind[0];
         Xc_indptr_ptr  =  &Xc_indptr[0];
     }
 
@@ -654,10 +680,12 @@ Rcpp::List impute_iso(SEXP model_R_ptr, SEXP imputer_R_ptr, bool is_extended,
         categ_data_ptr    =  &X_cat[0];
     }
 
-    if (Xr.size())
+    if (Xr_indptr.size())
     {
-        Xr_ptr         =  &Xr[0];
-        Xr_ind_ptr     =  &Xr_ind[0];
+        if (Xr.size())
+            Xr_ptr         =  &Xr[0];
+        if (Xr_ind.size())
+            Xr_ind_ptr     =  &Xr_ind[0];
         Xr_indptr_ptr  =  &Xr_indptr[0];
     }
 
@@ -681,7 +709,7 @@ Rcpp::List impute_iso(SEXP model_R_ptr, SEXP imputer_R_ptr, bool is_extended,
                           *imputer_ptr);
 
     return Rcpp::List::create(
-                Rcpp::_["X_num"] = Xr.size()? Xr : X_num,
+                Rcpp::_["X_num"] = (Xr.size())? (Xr) : (X_num),
                 Rcpp::_["X_cat"] = X_cat
             );
 }
@@ -760,3 +788,65 @@ Rcpp::List append_trees_from_other(SEXP model_R_ptr, SEXP other_R_ptr,
 
     return out;
 }
+
+// [[Rcpp::export]]
+Rcpp::ListOf<Rcpp::CharacterVector> model_to_sql(SEXP model_R_ptr, bool is_extended,
+                                                 Rcpp::CharacterVector numeric_colanmes,
+                                                 Rcpp::CharacterVector categ_colnames,
+                                                 Rcpp::ListOf<Rcpp::CharacterVector> categ_levels,
+                                                 bool output_tree_num, bool single_tree, size_t tree_num,
+                                                 int nthreads)
+{
+    IsoForest*     model_ptr      =  NULL;
+    ExtIsoForest*  ext_model_ptr  =  NULL;
+    if (is_extended)
+        ext_model_ptr  =  static_cast<ExtIsoForest*>(R_ExternalPtrAddr(model_R_ptr));
+    else
+        model_ptr      =  static_cast<IsoForest*>(R_ExternalPtrAddr(model_R_ptr));
+
+    std::vector<std::string> numeric_colanmes_cpp = Rcpp::as<std::vector<std::string>>(numeric_colanmes);
+    std::vector<std::string> categ_colanmes_cpp = Rcpp::as<std::vector<std::string>>(categ_colnames);
+    std::vector<std::vector<std::string>> categ_levels_cpp = Rcpp::as<std::vector<std::vector<std::string>>>(categ_levels);
+
+    std::vector<std::string> res = generate_sql(model_ptr, ext_model_ptr,
+                                                numeric_colanmes_cpp,
+                                                categ_colanmes_cpp,
+                                                categ_levels_cpp,
+                                                output_tree_num, true, single_tree, tree_num,
+                                                nthreads);
+    Rcpp::List out(res.size());
+    for (size_t ix = 0; ix < res.size(); ix++)
+        out[ix] = Rcpp::CharacterVector(res[ix]);
+    return out;
+}
+
+// [[Rcpp::export]]
+Rcpp::CharacterVector model_to_sql_with_select_from(SEXP model_R_ptr, bool is_extended,
+                                                    Rcpp::CharacterVector numeric_colanmes,
+                                                    Rcpp::CharacterVector categ_colnames,
+                                                    Rcpp::ListOf<Rcpp::CharacterVector> categ_levels,
+                                                    Rcpp::CharacterVector table_from,
+                                                    Rcpp::CharacterVector select_as,
+                                                    int nthreads)
+{
+    IsoForest*     model_ptr      =  NULL;
+    ExtIsoForest*  ext_model_ptr  =  NULL;
+    if (is_extended)
+        ext_model_ptr  =  static_cast<ExtIsoForest*>(R_ExternalPtrAddr(model_R_ptr));
+    else
+        model_ptr      =  static_cast<IsoForest*>(R_ExternalPtrAddr(model_R_ptr));
+
+    std::vector<std::string> numeric_colanmes_cpp = Rcpp::as<std::vector<std::string>>(numeric_colanmes);
+    std::vector<std::string> categ_colanmes_cpp = Rcpp::as<std::vector<std::string>>(categ_colnames);
+    std::vector<std::vector<std::string>> categ_levels_cpp = Rcpp::as<std::vector<std::vector<std::string>>>(categ_levels);
+    std::string table_from_cpp = Rcpp::as<std::string>(table_from);
+    std::string select_as_cpp = Rcpp::as<std::string>(select_as);
+
+    return generate_sql_with_select_from(model_ptr, ext_model_ptr,
+                                         table_from_cpp, select_as_cpp,
+                                         numeric_colanmes_cpp, categ_colanmes_cpp,
+                                         categ_levels_cpp,
+                                         true, nthreads);
+}
+
+#endif /* _FOR_R */