Bump version

* Avoid importing datatable.
* Fix #5363.

CMakeLists.txt

@@ -1,5 +1,5 @@
 cmake_minimum_required(VERSION 3.12)
-project(xgboost LANGUAGES CXX C VERSION 1.0.0)
+project(xgboost LANGUAGES CXX C VERSION 1.0.2)
 include(cmake/Utils.cmake)
 list(APPEND CMAKE_MODULE_PATH "${xgboost_SOURCE_DIR}/cmake/modules")
 cmake_policy(SET CMP0022 NEW)

Jenkinsfile

@@ -273,6 +273,7 @@ def TestPythonCPU() {
     def docker_binary = "docker"
     sh """
     ${dockerRun} ${container_type} ${docker_binary} tests/ci_build/test_python.sh cpu
+    ${dockerRun} ${container_type} ${docker_binary} tests/ci_build/test_python.sh cpu-py35
     """
     deleteDir()
   }

cmake/Python_version.in

@@ -1 +1 @@
-@xgboost_VERSION_MAJOR@.@xgboost_VERSION_MINOR@.@xgboost_VERSION_PATCH@rc2
+@xgboost_VERSION_MAJOR@.@xgboost_VERSION_MINOR@.@xgboost_VERSION_PATCH@

doc/model.schema

@@ -195,12 +195,22 @@
   "properties": {
     "version": {
       "type": "array",
-      "const": [
-        1,
-        0,
-        0
+      "items": [
+        {
+          "type": "number",
+          "const": 1
+        },
+        {
+          "type": "number",
+          "minimum": 0
+        },
+        {
+          "type": "number",
+          "minimum": 0
+        }
       ],
-      "additionalItems": false
+      "minItems": 3,
+      "maxItems": 3
     },
     "learner": {
       "type": "object",

doc/tutorials/saving_model.rst

@@ -195,7 +195,9 @@ You can load it back to the model generated by same version of XGBoost by:
 
   bst.load_config(config)
 
-This way users can study the internal representation more closely.
+This way users can study the internal representation more closely.  Please note that some
+JSON generators make use of locale dependent floating point serialization methods, which
+is not supported by XGBoost.
 
 ************
 Future Plans

include/xgboost/version_config.h

@@ -6,6 +6,6 @@
 
 #define XGBOOST_VER_MAJOR 1
 #define XGBOOST_VER_MINOR 0
-#define XGBOOST_VER_PATCH 0
+#define XGBOOST_VER_PATCH 1
 
 #endif  // XGBOOST_VERSION_CONFIG_H_

jvm-packages/pom.xml

@@ -6,7 +6,7 @@
 
     <groupId>ml.dmlc</groupId>
     <artifactId>xgboost-jvm_2.12</artifactId>
-    <version>1.0.0-RC2</version>
+    <version>1.0.0</version>
     <packaging>pom</packaging>
     <name>XGBoost JVM Package</name>
     <description>JVM Package for XGBoost</description>

jvm-packages/xgboost4j-example/pom.xml

@@ -6,10 +6,10 @@
     <parent>
         <groupId>ml.dmlc</groupId>
         <artifactId>xgboost-jvm_2.12</artifactId>
-        <version>1.0.0-RC2</version>
+        <version>1.0.0</version>
     </parent>
     <artifactId>xgboost4j-example_2.12</artifactId>
-    <version>1.0.0-RC2</version>
+    <version>1.0.0</version>
     <packaging>jar</packaging>
     <build>
         <plugins>
@@ -26,7 +26,7 @@
         <dependency>
             <groupId>ml.dmlc</groupId>
             <artifactId>xgboost4j-spark_${scala.binary.version}</artifactId>
-            <version>1.0.0-RC2</version>
+            <version>1.0.0</version>
         </dependency>
         <dependency>
             <groupId>org.apache.spark</groupId>
@@ -37,7 +37,7 @@
         <dependency>
             <groupId>ml.dmlc</groupId>
             <artifactId>xgboost4j-flink_${scala.binary.version}</artifactId>
-            <version>1.0.0-RC2</version>
+            <version>1.0.0</version>
         </dependency>
         <dependency>
             <groupId>org.apache.commons</groupId>

jvm-packages/xgboost4j-flink/pom.xml

@@ -6,10 +6,10 @@
     <parent>
         <groupId>ml.dmlc</groupId>
         <artifactId>xgboost-jvm_2.12</artifactId>
-        <version>1.0.0-RC2</version>
+        <version>1.0.0</version>
     </parent>
     <artifactId>xgboost4j-flink_2.12</artifactId>
-    <version>1.0.0-RC2</version>
+    <version>1.0.0</version>
     <build>
         <plugins>
             <plugin>
@@ -26,7 +26,7 @@
         <dependency>
             <groupId>ml.dmlc</groupId>
             <artifactId>xgboost4j_${scala.binary.version}</artifactId>
-            <version>1.0.0-RC2</version>
+            <version>1.0.0</version>
         </dependency>
         <dependency>
             <groupId>org.apache.commons</groupId>

jvm-packages/xgboost4j-spark/pom.xml

@@ -6,7 +6,7 @@
     <parent>
         <groupId>ml.dmlc</groupId>
         <artifactId>xgboost-jvm_2.12</artifactId>
-        <version>1.0.0-RC2</version>
+        <version>1.0.0</version>
     </parent>
     <artifactId>xgboost4j-spark_2.12</artifactId>
     <build>
@@ -24,7 +24,7 @@
         <dependency>
             <groupId>ml.dmlc</groupId>
             <artifactId>xgboost4j_${scala.binary.version}</artifactId>
-            <version>1.0.0-RC2</version>
+            <version>1.0.0</version>
         </dependency>
         <dependency>
             <groupId>org.apache.spark</groupId>

jvm-packages/xgboost4j/pom.xml

@@ -6,10 +6,10 @@
     <parent>
         <groupId>ml.dmlc</groupId>
         <artifactId>xgboost-jvm_2.12</artifactId>
-        <version>1.0.0-RC2</version>
+        <version>1.0.0</version>
     </parent>
     <artifactId>xgboost4j_2.12</artifactId>
-    <version>1.0.0-RC2</version>
+    <version>1.0.0</version>
     <packaging>jar</packaging>
 
     <dependencies>

python-package/xgboost/VERSION

@@ -1 +1 @@
-1.0.0rc2
+1.0.2

python-package/xgboost/__init__.py

@@ -5,6 +5,8 @@ Contributors: https://github.com/dmlc/xgboost/blob/master/CONTRIBUTORS.md
 """
 
 import os
+import sys
+import warnings
 
 from .core import DMatrix, Booster
 from .training import train, cv
@@ -19,6 +21,12 @@ try:
 except ImportError:
     pass
 
+if sys.version_info[:2] == (3, 5):
+    warnings.warn(
+        'Python 3.5 support is deprecated; XGBoost will require Python 3.6+ in the near future. ' +
+        'Consider upgrading to Python 3.6+.',
+        FutureWarning)
+
 VERSION_FILE = os.path.join(os.path.dirname(__file__), 'VERSION')
 with open(VERSION_FILE) as f:
     __version__ = f.read().strip()

python-package/xgboost/compat.py

@@ -79,6 +79,14 @@ else:
 # END NUMPY PATHLIB ATTRIBUTION
 ###############################################################################
 
+
+def lazy_isinstance(instance, module, name):
+    '''Use string representation to identify a type.'''
+    module = type(instance).__module__ == module
+    name = type(instance).__name__ == name
+    return module and name
+
+
 # pandas
 try:
     from pandas import DataFrame, Series
@@ -95,27 +103,6 @@ except ImportError:
     pandas_concat = None
     PANDAS_INSTALLED = False
 
-# dt
-try:
-    # Workaround for #4473, compatibility with dask
-    if sys.__stdin__ is not None and sys.__stdin__.closed:
-        sys.__stdin__ = None
-    import datatable
-
-    if hasattr(datatable, "Frame"):
-        DataTable = datatable.Frame
-    else:
-        DataTable = datatable.DataTable
-    DT_INSTALLED = True
-except ImportError:
-
-    # pylint: disable=too-few-public-methods
-    class DataTable(object):
-        """ dummy for datatable.DataTable """
-
-    DT_INSTALLED = False
-
-
 # cudf
 try:
     from cudf import DataFrame as CUDF_DataFrame

python-package/xgboost/core.py

@@ -19,9 +19,9 @@ import scipy.sparse
 
 from .compat import (
     STRING_TYPES, DataFrame, MultiIndex, Int64Index, py_str,
-    PANDAS_INSTALLED, DataTable,
-    CUDF_INSTALLED, CUDF_DataFrame, CUDF_Series, CUDF_MultiIndex,
-    os_fspath, os_PathLike)
+    PANDAS_INSTALLED, CUDF_INSTALLED,
+    CUDF_DataFrame, CUDF_Series, CUDF_MultiIndex,
+    os_fspath, os_PathLike, lazy_isinstance)
 from .libpath import find_lib_path
 
 # c_bst_ulong corresponds to bst_ulong defined in xgboost/c_api.h
@@ -319,7 +319,8 @@ DT_TYPE_MAPPER2 = {'bool': 'i', 'int': 'int', 'real': 'float'}
 def _maybe_dt_data(data, feature_names, feature_types,
                    meta=None, meta_type=None):
     """Validate feature names and types if data table"""
-    if not isinstance(data, DataTable):
+    if (not lazy_isinstance(data, 'datatable', 'Frame') and
+            not lazy_isinstance(data, 'datatable', 'DataTable')):
         return data, feature_names, feature_types
 
     if meta and data.shape[1] > 1:
@@ -470,7 +471,7 @@ class DMatrix(object):
             self._init_from_csc(data)
         elif isinstance(data, np.ndarray):
             self._init_from_npy2d(data, missing, nthread)
-        elif isinstance(data, DataTable):
+        elif lazy_isinstance(data, 'datatable', 'Frame'):
             self._init_from_dt(data, nthread)
         elif hasattr(data, "__cuda_array_interface__"):
             self._init_from_array_interface(data, missing, nthread)
@@ -1052,7 +1053,7 @@ class Booster(object):
             _check_call(
                 _LIB.XGBoosterUnserializeFromBuffer(self.handle, ptr, length))
             self.__dict__.update(state)
-        elif isinstance(model_file, (STRING_TYPES, os_PathLike)):
+        elif isinstance(model_file, (STRING_TYPES, os_PathLike, bytearray)):
             self.load_model(model_file)
         elif model_file is None:
             pass
@@ -1512,7 +1513,8 @@ class Booster(object):
         return ctypes2buffer(cptr, length.value)
 
     def load_model(self, fname):
-        """Load the model from a file, local or as URI.
+        """Load the model from a file or bytearray. Path to file can be local
+        or as an URI.
 
         The model is loaded from an XGBoost format which is universal among the
         various XGBoost interfaces. Auxiliary attributes of the Python Booster
@@ -1530,6 +1532,12 @@ class Booster(object):
             # from URL.
             _check_call(_LIB.XGBoosterLoadModel(
                 self.handle, c_str(os_fspath(fname))))
+        elif isinstance(fname, bytearray):
+            buf = fname
+            length = c_bst_ulong(len(buf))
+            ptr = (ctypes.c_char * len(buf)).from_buffer(buf)
+            _check_call(_LIB.XGBoosterLoadModelFromBuffer(self.handle, ptr,
+                                                          length))
         else:
             raise TypeError('Unknown file type: ', fname)
 

python-package/xgboost/sklearn.py

@@ -434,8 +434,8 @@ class XGBModel(XGBModelBase):
                 self.classes_ = np.array(v)
                 continue
             if k == 'type' and type(self).__name__ != v:
-                msg = f'Current model type: {type(self).__name__}, ' + \
-                    f'type of model in file: {v}'
+                msg = 'Current model type: {}, '.format(type(self).__name__) + \
+                      'type of model in file: {}'.format(v)
                 raise TypeError(msg)
             if k == 'type':
                 continue

python-package/xgboost/training.py

@@ -38,7 +38,7 @@ def _train_internal(params, dtrain,
 
     _params = dict(params) if isinstance(params, list) else params
 
-    if 'num_parallel_tree' in _params and params[
+    if 'num_parallel_tree' in _params and _params[
             'num_parallel_tree'] is not None:
         num_parallel_tree = _params['num_parallel_tree']
         nboost //= num_parallel_tree

src/common/json.cc

@@ -2,6 +2,7 @@
  * Copyright (c) by Contributors 2019
  */
 #include <cctype>
+#include <locale>
 #include <sstream>
 #include <limits>
 #include <cmath>
@@ -692,47 +693,23 @@ Json JsonReader::ParseBoolean() {
   return Json{JsonBoolean{result}};
 }
 
-// This is an ad-hoc solution for writing numeric value in standard way.  We need to add
-// a locale independent way of writing stream like `std::{from, to}_chars' from C++-17.
-// FIXME(trivialfis): Remove this.
-class GlobalCLocale {
-  std::locale ori_;
-
- public:
-  GlobalCLocale() : ori_{std::locale()} {
-    std::string const name {"C"};
-    try {
-      std::locale::global(std::locale(name.c_str()));
-    } catch (std::runtime_error const& e) {
-      LOG(FATAL) << "Failed to set locale: " << name;
-    }
-  }
-  ~GlobalCLocale() {
-    std::locale::global(ori_);
-  }
-};
-
 Json Json::Load(StringView str) {
-  GlobalCLocale guard;
   JsonReader reader(str);
   Json json{reader.Load()};
   return json;
 }
 
 Json Json::Load(JsonReader* reader) {
-  GlobalCLocale guard;
   Json json{reader->Load()};
   return json;
 }
 
 void Json::Dump(Json json, std::ostream *stream, bool pretty) {
-  GlobalCLocale guard;
   JsonWriter writer(stream, pretty);
   writer.Save(json);
 }
 
 void Json::Dump(Json json, std::string* str, bool pretty) {
-  GlobalCLocale guard;
   std::stringstream ss;
   JsonWriter writer(&ss, pretty);
   writer.Save(json);

tests/ci_build/Dockerfile.cpu

@@ -3,6 +3,7 @@ ARG CMAKE_VERSION=3.12
 
 # Environment
 ENV DEBIAN_FRONTEND noninteractive
+SHELL ["/bin/bash", "-c"]   # Use Bash as shell
 
 # Install all basic requirements
 RUN \
@@ -19,10 +20,16 @@ ENV PATH=/opt/python/bin:$PATH
 
 ENV GOSU_VERSION 1.10
 
-# Install Python packages
+# Create new Conda environment with Python 3.5
+RUN conda create -n py35 python=3.5 && \
+    source activate py35 && \
+    pip install numpy pytest scipy scikit-learn pandas matplotlib wheel kubernetes urllib3 graphviz && \
+    source deactivate
+
+# Install Python packages in default env
 RUN \
     pip install pyyaml cpplint pylint astroid sphinx numpy scipy pandas matplotlib sh \
-    		recommonmark guzzle_sphinx_theme mock breathe matplotlib graphviz \
+    		recommonmark guzzle_sphinx_theme mock breathe graphviz \
 		pytest scikit-learn wheel kubernetes urllib3 jsonschema boto3 && \
     pip install https://h2o-release.s3.amazonaws.com/datatable/stable/datatable-0.7.0/datatable-0.7.0-cp37-cp37m-linux_x86_64.whl && \
     pip install "dask[complete]"

tests/ci_build/test_python.sh

@@ -5,31 +5,35 @@ set -x
 suite=$1
 
 # Install XGBoost Python package
-wheel_found=0
-for file in python-package/dist/*.whl
-do
-  if [ -e "${file}" ]
+function install_xgboost {
+  wheel_found=0
+  for file in python-package/dist/*.whl
+  do
+    if [ -e "${file}" ]
+    then
+      pip install --user "${file}"
+      wheel_found=1
+      break  # need just one
+    fi
+  done
+  if [ "$wheel_found" -eq 0 ]
   then
-    pip install --user "${file}"
-    wheel_found=1
-    break  # need just one
+    pushd .
+    cd python-package
+    python setup.py install --user
+    popd
   fi
-done
-if [ "$wheel_found" -eq 0 ]
-then
-  pushd .
-  cd python-package
-  python setup.py install --user
-  popd
-fi
+}
 
 # Run specified test suite
 case "$suite" in
   gpu)
+    install_xgboost
     pytest -v -s --fulltrace -m "not mgpu" tests/python-gpu
     ;;
 
   mgpu)
+    install_xgboost
     pytest -v -s --fulltrace -m "mgpu" tests/python-gpu
     cd tests/distributed
     ./runtests-gpu.sh
@@ -39,17 +43,25 @@ case "$suite" in
 
   cudf)
     source activate cudf_test
+    install_xgboost
     pytest -v -s --fulltrace -m "not mgpu" tests/python-gpu/test_from_columnar.py tests/python-gpu/test_from_cupy.py
     ;;
 
   cpu)
+    install_xgboost
     pytest -v -s --fulltrace tests/python
     cd tests/distributed
     ./runtests.sh
     ;;
 
+  cpu-py35)
+    source activate py35
+    install_xgboost
+    pytest -v -s --fulltrace tests/python
+    ;;
+
   *)
-    echo "Usage: $0 {gpu|mgpu|cudf|cpu}"
+    echo "Usage: $0 {gpu|mgpu|cudf|cpu|cpu-py35}"
     exit 1
     ;;
 esac

tests/cpp/common/test_version.cc

@@ -54,7 +54,7 @@ TEST(Version, Basic) {
 
   ptr = 0;
   v = std::stoi(str, &ptr);
-  ASSERT_EQ(v, XGBOOST_VER_MINOR) << "patch: " << v;;
+  ASSERT_EQ(v, XGBOOST_VER_PATCH) << "patch: " << v;;
 
   str = str.substr(ptr);
   ASSERT_EQ(str.size(), 0);

tests/python/test_basic.py

@@ -35,6 +35,11 @@ def captured_output():
 
 
 class TestBasic(unittest.TestCase):
+    def test_compat(self):
+        from xgboost.compat import lazy_isinstance
+        a = np.array([1, 2, 3])
+        assert lazy_isinstance(a, 'numpy', 'ndarray')
+        assert not lazy_isinstance(a, 'numpy', 'dataframe')
 
     def test_basic(self):
         dtrain = xgb.DMatrix(dpath + 'agaricus.txt.train')

tests/python/test_basic_models.py

@@ -5,6 +5,7 @@ import os
 import json
 import testing as tm
 import pytest
+import locale
 
 dpath = 'demo/data/'
 dtrain = xgb.DMatrix(dpath + 'agaricus.txt.train')
@@ -299,7 +300,15 @@ class TestModels(unittest.TestCase):
         assert float(config['learner']['objective'][
             'reg_loss_param']['scale_pos_weight']) == 0.5
 
+        buf = bst.save_raw()
+        from_raw = xgb.Booster()
+        from_raw.load_model(buf)
+
+        buf_from_raw = from_raw.save_raw()
+        assert buf == buf_from_raw
+
     def test_model_json_io(self):
+        loc = locale.getpreferredencoding(False)
         model_path = 'test_model_json_io.json'
         parameters = {'tree_method': 'hist', 'booster': 'gbtree'}
         j_model = json_model(model_path, parameters)
@@ -313,6 +322,7 @@ class TestModels(unittest.TestCase):
         assert isinstance(j_model['learner'], dict)
 
         os.remove(model_path)
+        assert locale.getpreferredencoding(False) == loc
 
     @pytest.mark.skipif(**tm.no_json_schema())
     def test_json_schema(self):

tests/python/test_with_sklearn.py

@@ -34,7 +34,8 @@ def test_binary_classification():
     kf = KFold(n_splits=2, shuffle=True, random_state=rng)
     for cls in (xgb.XGBClassifier, xgb.XGBRFClassifier):
         for train_index, test_index in kf.split(X, y):
-            xgb_model = cls(random_state=42).fit(X[train_index], y[train_index])
+            clf = cls(random_state=42)
+            xgb_model = clf.fit(X[train_index], y[train_index], eval_metric=['auc', 'logloss'])
             preds = xgb_model.predict(X[test_index])
             labels = y[test_index]
             err = sum(1 for i in range(len(preds))

tests/python/testing.py

@@ -1,5 +1,5 @@
 # coding: utf-8
-from xgboost.compat import SKLEARN_INSTALLED, PANDAS_INSTALLED, DT_INSTALLED
+from xgboost.compat import SKLEARN_INSTALLED, PANDAS_INSTALLED
 from xgboost.compat import CUDF_INSTALLED, DASK_INSTALLED
 
 
@@ -19,7 +19,9 @@ def no_pandas():
 
 
 def no_dt():
-    return {'condition': not DT_INSTALLED,
+    import importlib.util
+    spec = importlib.util.find_spec('datatable')
+    return {'condition': spec is None,
             'reason': 'Datatable is not installed.'}