Merge branch 'master' into Rinterface7

.appveyor.yml

@@ -3,7 +3,7 @@ version: 3.3.2.99.{build}
 image: Visual Studio 2015
 platform: x64
 configuration:  # a trick to construct a build matrix with multiple Python versions
-  - 3.7
+  - '3.7'
 
 # only build pull requests and
 # commits to 'master'

.ci/test.sh

@@ -36,8 +36,15 @@ cd $BUILD_DIRECTORY
 
 if [[ $TASK == "check-docs" ]] || [[ $TASK == "check-links" ]]; then
     cd $BUILD_DIRECTORY/docs
-    conda install -q -y -n $CONDA_ENV -c conda-forge doxygen rstcheck
-    pip install --user -r requirements.txt
+    conda env update \
+        -n $CONDA_ENV \
+        --file ./env.yml || exit -1
+    conda install \
+        -q \
+        -y \
+        -n $CONDA_ENV \
+            doxygen \
+            rstcheck || exit -1
     # check reStructuredText formatting
     cd $BUILD_DIRECTORY/python-package
     rstcheck --report warning $(find . -type f -name "*.rst") || exit -1

.gitignore

@@ -346,6 +346,7 @@ instance/
 # Sphinx documentation
 docs/_build/
 docs/pythonapi/
+*.flag
 
 # Doxygen documentation
 docs/doxyoutput/

.readthedocs.yaml

@@ -1,10 +1,12 @@
 version: 2
+build:
+  os: "ubuntu-20.04"
+  tools:
+    python: "miniconda3-4.7"
+conda:
+  environment: docs/env.yml
 formats:
   - pdf
-python:
-  version: 3
-  install:
-    - requirements: docs/requirements.txt
 sphinx:
   builder: html
   configuration: docs/conf.py

.vsts-ci.yml

@@ -41,6 +41,7 @@ jobs:
         PYTHON_VERSION: '3.7'
       bdist:
         TASK: bdist
+        PYTHON_VERSION: '3.8'
       inference:
         TASK: if-else
       mpi_source:
@@ -52,7 +53,7 @@ jobs:
         METHOD: source
         # on Ubuntu 14.04, gpu_source build segfaults with newer version of Python
         # (and newer version of scipy as a consequence)
-        PYTHON_VERSION: 3.7
+        PYTHON_VERSION: '3.7'
       swig:
         TASK: swig
   steps:

R-package/R/lgb.Booster.R

@@ -26,108 +26,90 @@ Booster <- R6::R6Class(
                           modelfile = NULL,
                           model_str = NULL) {
 
-      # Create parameters and handle
       handle <- NULL
 
-      # Attempts to create a handle for the dataset
-      try({
-
-        # Check if training dataset is not null
-        if (!is.null(train_set)) {
-          # Check if training dataset is lgb.Dataset or not
-          if (!lgb.is.Dataset(train_set)) {
-            stop("lgb.Booster: Can only use lgb.Dataset as training data")
-          }
-          train_set_handle <- train_set$.__enclos_env__$private$get_handle()
-          params <- utils::modifyList(params, train_set$get_params())
-          params_str <- lgb.params2str(params = params)
-          # Store booster handle
-          handle <- .Call(
-            LGBM_BoosterCreate_R
-            , train_set_handle
-            , params_str
-          )
-
-          # Create private booster information
-          private$train_set <- train_set
-          private$train_set_version <- train_set$.__enclos_env__$private$version
-          private$num_dataset <- 1L
-          private$init_predictor <- train_set$.__enclos_env__$private$predictor
-
-          # Check if predictor is existing
-          if (!is.null(private$init_predictor)) {
-
-            # Merge booster
-            .Call(
-              LGBM_BoosterMerge_R
-              , handle
-              , private$init_predictor$.__enclos_env__$private$handle
-            )
-
-          }
-
-          # Check current iteration
-          private$is_predicted_cur_iter <- c(private$is_predicted_cur_iter, FALSE)
+      if (!is.null(train_set)) {
 
-        } else if (!is.null(modelfile)) {
+        if (!lgb.is.Dataset(train_set)) {
+          stop("lgb.Booster: Can only use lgb.Dataset as training data")
+        }
+        train_set_handle <- train_set$.__enclos_env__$private$get_handle()
+        params <- utils::modifyList(params, train_set$get_params())
+        params_str <- lgb.params2str(params = params)
+        # Store booster handle
+        handle <- .Call(
+          LGBM_BoosterCreate_R
+          , train_set_handle
+          , params_str
+        )
 
-          # Do we have a model file as character?
-          if (!is.character(modelfile)) {
-            stop("lgb.Booster: Can only use a string as model file path")
-          }
+        # Create private booster information
+        private$train_set <- train_set
+        private$train_set_version <- train_set$.__enclos_env__$private$version
+        private$num_dataset <- 1L
+        private$init_predictor <- train_set$.__enclos_env__$private$predictor
 
-          modelfile <- path.expand(modelfile)
+        if (!is.null(private$init_predictor)) {
 
-          # Create booster from model
-          handle <- .Call(
-            LGBM_BoosterCreateFromModelfile_R
-            , modelfile
+          # Merge booster
+          .Call(
+            LGBM_BoosterMerge_R
+            , handle
+            , private$init_predictor$.__enclos_env__$private$handle
           )
 
-        } else if (!is.null(model_str)) {
+        }
 
-          # Do we have a model_str as character/raw?
-          if (!is.raw(model_str) && !is.character(model_str)) {
-            stop("lgb.Booster: Can only use a character/raw vector as model_str")
-          }
+        # Check current iteration
+        private$is_predicted_cur_iter <- c(private$is_predicted_cur_iter, FALSE)
 
-          # Create booster from model
-          handle <- .Call(
-            LGBM_BoosterLoadModelFromString_R
-            , model_str
-          )
+      } else if (!is.null(modelfile)) {
 
-        } else {
+        # Do we have a model file as character?
+        if (!is.character(modelfile)) {
+          stop("lgb.Booster: Can only use a string as model file path")
+        }
 
-          # Booster non existent
-          stop(
-            "lgb.Booster: Need at least either training dataset, "
-            , "model file, or model_str to create booster instance"
-          )
+        modelfile <- path.expand(modelfile)
 
-        }
+        # Create booster from model
+        handle <- .Call(
+          LGBM_BoosterCreateFromModelfile_R
+          , modelfile
+        )
 
-      })
+      } else if (!is.null(model_str)) {
 
-      # Check whether the handle was created properly if it was not stopped earlier by a stop call
-      if (isTRUE(lgb.is.null.handle(x = handle))) {
+        # Do we have a model_str as character/raw?
+        if (!is.raw(model_str) && !is.character(model_str)) {
+          stop("lgb.Booster: Can only use a character/raw vector as model_str")
+        }
 
-        stop("lgb.Booster: cannot create Booster handle")
+        # Create booster from model
+        handle <- .Call(
+          LGBM_BoosterLoadModelFromString_R
+          , model_str
+        )
 
       } else {
 
-        # Create class
-        class(handle) <- "lgb.Booster.handle"
-        private$handle <- handle
-        private$num_class <- 1L
-        .Call(
-          LGBM_BoosterGetNumClasses_R
-          , private$handle
-          , private$num_class
+        # Booster non existent
+        stop(
+          "lgb.Booster: Need at least either training dataset, "
+          , "model file, or model_str to create booster instance"
         )
 
       }
 
+      class(handle) <- "lgb.Booster.handle"
+      private$handle <- handle
+      private$num_class <- 1L
+      .Call(
+        LGBM_BoosterGetNumClasses_R
+        , private$handle
+        , private$num_class
+      )
+
       self$params <- params
 
       return(invisible(NULL))

R-package/R/lgb.cv.R

@@ -137,7 +137,6 @@ lgb.cv <- function(params = list()
   early_stopping_rounds <- params[["early_stopping_round"]]
 
   # extract any function objects passed for objective or metric
-  params <- lgb.check.obj(params = params)
   fobj <- NULL
   if (is.function(params$objective)) {
     fobj <- params$objective

R-package/R/lgb.train.R

@@ -105,7 +105,6 @@ lgb.train <- function(params = list(),
   early_stopping_rounds <- params[["early_stopping_round"]]
 
   # extract any function objects passed for objective or metric
-  params <- lgb.check.obj(params = params)
   fobj <- NULL
   if (is.function(params$objective)) {
     fobj <- params$objective

R-package/R/utils.R

@@ -117,64 +117,6 @@ lgb.check_interaction_constraints <- function(interaction_constraints, column_na
 
 }
 
-lgb.check.obj <- function(params) {
-
-  # List known objectives in a vector
-  OBJECTIVES <- c(
-    "regression"
-    , "regression_l1"
-    , "regression_l2"
-    , "mean_squared_error"
-    , "mse"
-    , "l2_root"
-    , "root_mean_squared_error"
-    , "rmse"
-    , "mean_absolute_error"
-    , "mae"
-    , "quantile"
-    , "huber"
-    , "fair"
-    , "poisson"
-    , "binary"
-    , "lambdarank"
-    , "multiclass"
-    , "softmax"
-    , "multiclassova"
-    , "multiclass_ova"
-    , "ova"
-    , "ovr"
-    , "xentropy"
-    , "cross_entropy"
-    , "xentlambda"
-    , "cross_entropy_lambda"
-    , "mean_absolute_percentage_error"
-    , "mape"
-    , "gamma"
-    , "tweedie"
-    , "rank_xendcg"
-    , "xendcg"
-    , "xe_ndcg"
-    , "xe_ndcg_mart"
-    , "xendcg_mart"
-  )
-
-  if (is.null(params$objective)) {
-    stop("lgb.check.obj: objective should be a character or a function")
-  }
-
-  if (is.character(params$objective)) {
-
-    if (!(params$objective %in% OBJECTIVES)) {
-
-      stop("lgb.check.obj: objective name error should be one of (", paste0(OBJECTIVES, collapse = ", "), ")")
-
-    }
-
-  }
-
-  return(params)
-
-}
 
 # [description]
 #     Take any character values from eval and store them in params$metric.

R-package/tests/testthat/test_basic.R

@@ -520,6 +520,22 @@ test_that("lgb.cv() respects showsd argument", {
   expect_identical(evals_no_showsd[["eval_err"]], list())
 })
 
+test_that("lgb.cv() raises an informative error for unrecognized objectives", {
+  dtrain <- lgb.Dataset(
+    data = train$data
+    , label = train$label
+  )
+  expect_error({
+    bst <- lgb.cv(
+      data = dtrain
+      , params = list(
+        objective_type = "not_a_real_objective"
+        , verbosity = VERBOSITY
+      )
+    )
+  }, regexp = "Unknown objective type name: not_a_real_objective")
+})
+
 test_that("lgb.cv() respects parameter aliases for objective", {
   nrounds <- 3L
   nfold <- 4L
@@ -663,6 +679,22 @@ test_that("lgb.train() works as expected with multiple eval metrics", {
   )
 })
 
+test_that("lgb.train() raises an informative error for unrecognized objectives", {
+  dtrain <- lgb.Dataset(
+    data = train$data
+    , label = train$label
+  )
+  expect_error({
+    bst <- lgb.train(
+      data = dtrain
+      , params = list(
+        objective_type = "not_a_real_objective"
+        , verbosity = VERBOSITY
+      )
+    )
+  }, regexp = "Unknown objective type name: not_a_real_objective")
+})
+
 test_that("lgb.train() respects parameter aliases for objective", {
   nrounds <- 3L
   dtrain <- lgb.Dataset(