tighten up TF & TFP version dependencies

greta-dev · Oct 24, 2018 · be76c70 · be76c70
1 parent f85b93b
commit be76c70
Show file tree

Hide file tree

Showing 5 changed files with 96 additions and 56 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -24,8 +24,8 @@ License: Apache License 2.0
 URL: https://github.com/greta-dev/greta
 BugReports: https://github.com/greta-dev/greta/issues
 SystemRequirements: Python (>= 2.7.0) with header files and shared library;
-    TensorFlow (>= 1.8; https://www.tensorflow.org/);
-    Tensorflow Probability (>=0.0.1; https://github.com/tensorflow/probability)
+    TensorFlow (>= 1.10; https://www.tensorflow.org/);
+    Tensorflow Probability (>=0.3.0; https://www.tensorflow.org/probability/)
 Encoding: UTF-8
 LazyData: true
 Depends:

diff --git a/NAMESPACE b/NAMESPACE
@@ -217,6 +217,7 @@ importFrom(grDevices,col2rgb)
 importFrom(grDevices,colorRampPalette)
 importFrom(progress,progress_bar)
 importFrom(reticulate,conda_binary)
+importFrom(reticulate,py_available)
 importFrom(reticulate,py_module_available)
 importFrom(reticulate,py_set_attr)
 importFrom(stats,na.omit)

diff --git a/R/dag_class.R b/R/dag_class.R
@@ -1,5 +1,6 @@
 #' @importFrom reticulate py_set_attr
 #' @importFrom tensorflow dict
+#' @importFrom R6 R6Class
 
 # create dag class
 dag_class <- R6Class(

diff --git a/R/package.R b/R/package.R
@@ -19,9 +19,6 @@
 #'
 #' @docType package
 #' @importFrom tensorflow tf
-#' @importFrom reticulate py_module_available
-#' @importFrom R6 R6Class
-#' @importFrom grDevices colorRampPalette
 #' @examples
 #' \dontrun{
 #' # a simple Bayesian regression model for the iris data

diff --git a/R/utils.R b/R/utils.R
@@ -53,6 +53,22 @@ have_virtualenv <- function () {
 
 }
 
+#' @importFrom reticulate py_available
+have_python <- function () {
+  tryCatch(reticulate::py_available(initialize = TRUE),
+                        error = function(e) FALSE)
+}
+
+#' @importFrom reticulate py_module_available
+have_tfp <- function () {
+  reticulate::py_module_available("tensorflow_probability")
+}
+
+#' @importFrom reticulate py_module_available
+have_tf <- function () {
+  reticulate::py_module_available("tensorflow")
+}
+
 # check tensorflow and tensorflow-probability are installed and have valid
 # versions. error, warn, or message if not and (if not an error) return an
 # invisible logical saying whether it is valid
@@ -65,93 +81,117 @@ check_tf_version <- function(alert = c("none",
                                        "startup")) {
 
   alert <- match.arg(alert)
-  text <- NULL
+
+  py_available <- TRUE
   tf_available <- TRUE
   tfp_available <- TRUE
 
-  # check TF installation
-  if (!reticulate::py_module_available("tensorflow")) {
+  # check python installation
+  if (!have_python()) {
 
-    text <- "TensorFlow isn't installed"
-    tf_available <- FALSE
+    text <- paste0("\n\ngreta requires Python and several Python packages ",
+                  "to be installed, but no Python installation was detected.\n",
+                  "You can install Python directly from ",
+                  "https://www.python.org/downloads/ ",
+                  "or with the Anaconda distribution from ",
+                  "https://www.anaconda.com/download/")
 
-  } else {
+    py_available <- tf_available <- tfp_available <- FALSE
 
-    tf_version <- tf$`__version__`
-    tf_version_valid <- utils::compareVersion("1.8", tf_version) != 1
+  }
 
-    if (!tf_version_valid) {
-      text <- paste0("you have TensorFlow version ", tf_version)
+  if (py_available) {
+
+    text <- NULL
+
+    # check TF installation
+    if (!have_tf()) {
+
+      text <- "TensorFlow isn't installed"
       tf_available <- FALSE
-    }
 
-  }
+    } else {
 
-  # check TFP installation
-  if (!reticulate::py_module_available("tensorflow_probability")) {
+      tf_version <- tf$`__version__`
+      tf_version_valid <- utils::compareVersion("1.10.0", tf_version) != 1
 
-    text <- paste0(text,
-                   ifelse(is.null(text), "", " and "),
-                   "TensorFlow Probability isn't installed")
-    tfp_available <- FALSE
+      if (!tf_version_valid) {
+        text <- paste0("you have TensorFlow version ", tf_version)
+        tf_available <- FALSE
+      }
 
-  } else {
+    }
 
-    pkg <- reticulate::import("pkg_resources")
-    tfp_version <- pkg$get_distribution("tensorflow_probability")$version
-    tfp_version_valid <- utils::compareVersion("0.3.0", tfp_version) != 1
+    # check TFP installation
+    if (!have_tfp()) {
 
-    if (!tfp_version_valid) {
-      text <- paste0("you have TensorFlow Probability version ", tfp_version)
+      text <- paste0(text,
+                     ifelse(is.null(text), "", " and "),
+                     "TensorFlow Probability isn't installed")
       tfp_available <- FALSE
+
+    } else {
+
+      pkg <- reticulate::import("pkg_resources")
+      tfp_version <- pkg$get_distribution("tensorflow_probability")$version
+      tfp_version_valid <- utils::compareVersion("0.3.0", tfp_version) != 1
+
+      if (!tfp_version_valid) {
+        text <- paste0("you have TensorFlow Probability version ", tfp_version)
+        tfp_available <- FALSE
+      }
+
     }
 
-  }
+    # if there was a problem, append the solution
+    if (!tf_available | !tfp_available) {
 
-  if (!is.null(text)) {
+      # conda-specific installation instructions, to handle conda not having TFP
+      if (have_conda() & !have_virtualenv()) {
 
-    # conda-specific installation instructions, to handle conda not having TFP
-    if (have_conda() & !have_virtualenv()) {
+        tf_install <- tfp_install <- ""
 
-      tf_install <- tfp_install <- ""
+        if (!tf_available | !tfp_available) {
+          tf_install <- '    install_tensorflow(method = "conda")\n'
+        }
 
-      if (!tf_available | !tfp_available) {
-        tf_install <- '    install_tensorflow(method = "conda")\n'
-      }
+        if (!tfp_available) {
+          tfp_install <- paste0('   reticulate::conda_install("r-tensorflow", ',
+                                '"tensorflow-probability", pip = TRUE)\n')
+        }
+
+        install <- paste(tf_install, tfp_install, collapse = "\n")
 
-      if (!tfp_available) {
-        tfp_install <- paste0('   reticulate::conda_install("r-tensorflow", ',
-                              '"tensorflow-probability", pip = TRUE)\n')
+      } else {
+        # non-conda installation instructions
+        install <- sprintf("install_tensorflow(%s) ",
+                           ifelse(tfp_available,
+                                  "",
+                                  "extra_packages = \"tensorflow-probability\""))
       }
 
-      install <- paste(tf_install, tfp_install, collapse = "\n")
+      # combine the problem and solution messages
+      text <- paste0("\n\ngreta requires TensorFlow (>=1.10.0) ",
+                     "and Tensorflow Probability (>=0.3.0), ",
+                     "but ", text, ". Use:\n\n",
+                     install,
+                     "\nto install the latest version.",
+                     "\n\n")
 
-    } else {
-      # non-conda installation instructions
-      install <- sprintf("install_tensorflow(%s) ",
-                         ifelse(tfp_available,
-                                "",
-                                "extra_packages = \"tensorflow-probability\""))
     }
 
-    # combine the problem and solution messages
-    text <- paste0("\n\ngreta requires TensorFlow (>=1.8) or higher ",
-                   "and Tensorflow Probability (>=0.3.0), ",
-                   "but ", text, ". Use:\n\n",
-                   install,
-                   "\nto install the latest version.",
-                   "\n\n")
+  }
 
+  if (!is.null(text)) {
     switch(alert,
            error = stop(text, call. = FALSE),
            warn = warning(text, call. = FALSE),
            message = message(text),
            startup = packageStartupMessage(text),
            none = NULL)
-
   }
 
-  invisible(tf_available & tfp_available)
+  invisible(py_available & tf_available & tfp_available)
 
 }
 
@@ -816,6 +856,7 @@ dummy_array_module <- module(flatten_rowwise,
 # returning a colour linearly interpolated between black, the colour and white,
 # so that values close to 0.5 match the base colour, values close to 0 are
 # nearer black, and values close to 1 are nearer white
+#' @importFrom grDevices colorRampPalette
 palettize <- function(base_colour) {
   pal <- colorRampPalette(c("#000000", base_colour, "#ffffff"))
   function(val) {