Merge pull request #26 from mrkn/rewrite

Rewrite toward version 1.0

Author: mrkn

.travis.yml

@@ -9,16 +9,20 @@ rvm:
   - 2.1.10
 
 env:
+  global:
+  - PYCALL_DEBUG_FIND_LIBPYTHON=1
+  matrix:
   - PYTHON=python
   - PYTHON=python3 LIBPYTHON=wrong_value
-  - LIBPYTHON=/usr/lib/libpython3.2mu.so.1
+  - LIBPYTHON=/opt/python/3.5.3/lib/libpython3.5m.so
 
 addons:
   apt:
     packages:
       - python3
       - python3-dev
       - python3-all
+      - python3-all-dev
 
 before_install:
   - gem update --system
@@ -28,3 +32,10 @@ before_script:
   - bundle exec rake clobber compile
   - echo === python investigator.py ===
   - python lib/pycall/python/investigator.py
+  - python3 lib/pycall/python/investigator.py
+  - pip install numpy
+  - pip3 install numpy
+
+matrix:
+  allow_failures:
+  - env: PYTHON=python  # Ignore failed on python 2.7

CHANGES.md

@@ -0,0 +1,32 @@
+# The change history of PyCall
+
+## master
+
+* Rewrite almost all fundamental parts of PyCall as C extension.
+
+* PyCall now calls `Py_DecRef` in the finalizer of `PyCall::PyPtr`.
+
+* Change the system of object mapping between Python and Ruby, drastically.
+  Now PyCall does not have `PyObject` class for wrapper objects.
+  Instead, PyCall generally makes `Object` instances and extends them by
+  `PyObjectWrapper` module.
+  But for Python module objects, PyCall makes anonymous `Module` instances 
+  that are extended by `PyObjectWrapper` module.
+  Moreover for Python type objects, PyCall makes `Class` instances and extends
+  them by `PyTypeObjectWrapper` module.
+
+* Change `PyCall.eval` to be a wrapper of `__builtins__.eval` in Python.
+  This means that `filename:` and `input_type:` parameters are dropped.
+  Instead, two new parameters `globals:` and `locals:` are introduced.
+  `globals:` is used for specifying a dictionary that is the global
+  namespace referred by the evaluated expression.
+  `locals:` is used for specifying a mapping object that is the local
+  namespace referred by the evaluated expression.
+
+* Add `PyCall.exec` for the replacement of the former `PyCall.eval`
+  with `input_type: :file`.
+  It has `globals:` and `locals:` parameters for the same meaning as
+  the new `PyCall.eval` described above.
+
+* Drop `PyCall.wrap_ruby_callable` and `PyCall.wrap_ruby_object` always
+  craetes a callable Python object taht has an ID of the given Ruby object.

Gemfile

@@ -2,8 +2,3 @@ source 'https://rubygems.org'
 
 # Specify your gem's dependencies in pycall.gemspec
 gemspec
-
-group :development do
-  gem 'guard-rspec', require: false
-  gem 'terminal-notifier-guard', require: false
-end

Rakefile

@@ -1,10 +1,29 @@
-require "bundler/gem_tasks"
-require "rspec/core/rake_task"
+require "bundler"
+Bundler::GemHelper.install_tasks
+
+require "rake"
 require "rake/extensiontask"
+require "rspec/core/rake_task"
 
 Dir[File.expand_path('../tasks/**/*.rake', __FILE__)].each {|f| load f }
 
-Rake::ExtensionTask.new('pycall/pyptr')
+gem_spec = eval(File.read('pycall.gemspec'))
+Rake::ExtensionTask.new('pycall', gem_spec) do |ext|
+  ext.lib_dir = File.join(*['lib', ENV['FAT_DIR']].compact)
+  ext.cross_compile = true
+  ext.cross_platform = %w[x86-mingw32 x64-mingw32]
+  ext.cross_compiling do |s|
+    s.files.concat %w[lib/2.2/pycall.so lib/2.3/pycall.so lib/2.4/pycall.so]
+  end
+end
+
+desc "Compile binaries for mingw platform using rake-compiler-dock"
+task 'build:mingw' do
+  require 'rake_compiler_dock'
+  RakeCompilerDock.sh "bundle && rake cross native gem RUBY_CC_VERSION=2.1.6:2.2.2:2.3.0:2.4.0"
+end
+
 RSpec::Core::RakeTask.new(:spec)
 
 task :default => :spec
+task spec: :compile

appveyor.yml

@@ -6,10 +6,6 @@ environment:
     PYTHONDIR: "C:\\Python27"
     PYTHON: "C:\\Python27\\python.exe"
 
-  - ruby_version: "21"
-    PYTHONDIR: "C:\\Python33"
-    PYTHON: "C:\\Python33\\python.exe"
-
   - ruby_version: "21"
     PYTHONDIR: "C:\\Python34"
     PYTHON: "C:\\Python34\\python.exe"
@@ -27,10 +23,6 @@ environment:
     PYTHONDIR: "C:\\Python27-x64"
     PYTHON: "C:\\Python27-x64\\python.exe"
 
-  - ruby_version: "21-x64"
-    PYTHONDIR: "C:\\Python33-x64"
-    PYTHON: "C:\\Python33-x64\\python.exe"
-
   - ruby_version: "21-x64"
     PYTHONDIR: "C:\\Python34-x64"
     PYTHON: "C:\\Python34-x64\\python.exe"
@@ -48,10 +40,6 @@ environment:
     PYTHONDIR: "C:\\Python27"
     PYTHON: "C:\\Python27\\python.exe"
 
-  - ruby_version: "22"
-    PYTHONDIR: "C:\\Python33"
-    PYTHON: "C:\\Python33\\python.exe"
-
   - ruby_version: "22"
     PYTHONDIR: "C:\\Python34"
     PYTHON: "C:\\Python34\\python.exe"
@@ -69,10 +57,6 @@ environment:
     PYTHONDIR: "C:\\Python27-x64"
     PYTHON: "C:\\Python27-x64\\python.exe"
 
-  - ruby_version: "22-x64"
-    PYTHONDIR: "C:\\Python33-x64"
-    PYTHON: "C:\\Python33-x64\\python.exe"
-
   - ruby_version: "22-x64"
     PYTHONDIR: "C:\\Python34-x64"
     PYTHON: "C:\\Python34-x64\\python.exe"
@@ -90,10 +74,6 @@ environment:
     PYTHONDIR: "C:\\Python27"
     PYTHON: "C:\\Python27\\python.exe"
 
-  - ruby_version: "23"
-    PYTHONDIR: "C:\\Python33"
-    PYTHON: "C:\\Python33\\python.exe"
-
   - ruby_version: "23"
     PYTHONDIR: "C:\\Python34"
     PYTHON: "C:\\Python34\\python.exe"
@@ -111,10 +91,6 @@ environment:
     PYTHONDIR: "C:\\Python27-x64"
     PYTHON: "C:\\Python27-x64\\python.exe"
 
-  - ruby_version: "23-x64"
-    PYTHONDIR: "C:\\Python33-x64"
-    PYTHON: "C:\\Python33-x64\\python.exe"
-
   - ruby_version: "23-x64"
     PYTHONDIR: "C:\\Python34-x64"
     PYTHON: "C:\\Python34-x64\\python.exe"
@@ -144,13 +120,19 @@ build: off
 install:
   - "SET PATH=%PYTHONDIR%;%PYTHONDIR%\\Scripts;%PATH%"
   - "SET PATH=C:\\Ruby%ruby_version%\\bin;%PATH%"
-  - bundle install
+  - "bundle install"
+  - "pip install numpy"
 
 before_test:
   - "bundle exec rake -rdevkit clobber compile"
   - ECHO "=== python investigator.py ==="
   - "python lib\\pycall\\python\\investigator.py"
 
 test_script:
-  - "SET DEBUG_FIND_LIBPYTHON=1"
+  - "SET PYCALL_DEBUG_FIND_LIBPYTHON=1"
   - rake
+
+matrix:
+  allow_failures:
+  - PYTHONDIR: "C:\\Python27"
+  - PYTHONDIR: "C:\\Python27-x64"

examples/classifier_comparison.rb

@@ -34,102 +34,102 @@
 ]
 
 classifiers = [
-  KNeighborsClassifier.(3),
-  SVC.(kernel: 'linear', C: 0.025),
-  SVC.(gamma: 2, C: 1),
-  DecisionTreeClassifier.(max_depth: 5),
-  RandomForestClassifier.(max_depth: 5, n_estimators: 10, max_features: 1),
-  AdaBoostClassifier.(),
-  GaussianNB.(),
-  LinearDiscriminantAnalysis.(),
-  QuadraticDiscriminantAnalysis.()
+  KNeighborsClassifier.new(3),
+  SVC.new(kernel: 'linear', C: 0.025),
+  SVC.new(gamma: 2, C: 1),
+  DecisionTreeClassifier.new(max_depth: 5),
+  RandomForestClassifier.new(max_depth: 5, n_estimators: 10, max_features: 1),
+  AdaBoostClassifier.new(),
+  GaussianNB.new(),
+  LinearDiscriminantAnalysis.new(),
+  QuadraticDiscriminantAnalysis.new()
 ]
 
-x, y = make_classification.(
+x, y = *make_classification(
   n_features: 2,
   n_redundant: 0,
   n_informative: 2,
   random_state: 1,
   n_clusters_per_class: 1
 )
 
-np.random.seed.(42)
-x += 2 * np.random.random_sample.(x.shape)
-linearly_separable = PyCall.tuple(x, y)
+np.random.seed(42)
+x += 2 * np.random.random_sample(x.shape)
+linearly_separable = PyCall.tuple([x, y]) # FIXME: allow PyCall.tuple(x, y)
 
 datasets = [
-  make_moons.(noise: 0.3, random_state: 0),
-  make_circles.(noise: 0.2, factor: 0.5, random_state: 1),
+  make_moons(noise: 0.3, random_state: 0),
+  make_circles(noise: 0.2, factor: 0.5, random_state: 1),
   linearly_separable
 ]
 
-fig = plt.figure.(figsize: PyCall.tuple(27, 9))
+fig = plt.figure(figsize: [27, 9])
 i = 1
-all = PyCall.slice(nil)
+all = PyCall::Slice.all
 datasets.each do |ds|
-  x, y = ds
-  x = StandardScaler.().fit_transform.(x)
-  x_train, x_test, y_train, y_test = train_test_split.(x, y, test_size: 0.4)
+  x, y = *ds
+  x = StandardScaler.new.fit_transform(x)
+  x_train, x_test, y_train, y_test = train_test_split(x, y, test_size: 0.4)
 
-  x_min, x_max = np.min.(x[all, 0]) - 0.5, np.max.(x[all, 0]) + 0.5
-  y_min, y_max = np.min.(x[all, 1]) - 0.5, np.max.(x[all, 1]) + 0.5
+  x_min, x_max = np.min(x[all, 0]) - 0.5, np.max(x[all, 0]) + 0.5
+  y_min, y_max = np.min(x[all, 1]) - 0.5, np.max(x[all, 1]) + 0.5
 
-  xx, yy = np.meshgrid.(
-    np.linspace.(x_min, x_max, ((x_max - x_min)/h).round),
-    np.linspace.(y_min, y_max, ((y_max - y_min)/h).round),
+  xx, yy = np.meshgrid(
+    np.linspace(x_min, x_max, ((x_max - x_min)/h).round),
+    np.linspace(y_min, y_max, ((y_max - y_min)/h).round),
   )
-  mesh_points = np.dstack.(PyCall.tuple(xx.ravel.(), yy.ravel.()))[0, all, all]
+  mesh_points = np.dstack(PyCall.tuple([xx.ravel(), yy.ravel()]))[0, all, all]
 
   # just plot the dataset first
-  cm = plt.cm.RdBu
-  cm_bright = mplc.ListedColormap.(["#FF0000", "#0000FF"])
-  ax = plt.subplot.(datasets.length, classifiers.length + 1, i)
+  cm = plt.cm.__dict__[:RdBu]
+  cm_bright = mplc.ListedColormap.new(["#FF0000", "#0000FF"])
+  ax = plt.subplot(datasets.length, classifiers.length + 1, i)
   # plot the training points
-  ax.scatter.(x_train[all, 0], x_train[all, 1], c: y_train, cmap: cm_bright)
+  ax.scatter(x_train[all, 0], x_train[all, 1], c: y_train, cmap: cm_bright)
   # and testing points
-  ax.scatter.(x_test[all, 0], x_test[all, 1], c: y_test, cmap: cm_bright, alpha: 0.6)
+  ax.scatter(x_test[all, 0], x_test[all, 1], c: y_test, cmap: cm_bright, alpha: 0.6)
 
-  ax.set_xlim.(np.min.(xx), np.max.(xx))
-  ax.set_ylim.(np.min.(yy), np.max.(yy))
-  ax.set_xticks.(PyCall.tuple())
-  ax.set_yticks.(PyCall.tuple())
+  ax.set_xlim(np.min(xx), np.max(xx))
+  ax.set_ylim(np.min(yy), np.max(yy))
+  ax.set_xticks(PyCall.tuple())
+  ax.set_yticks(PyCall.tuple())
   i += 1
 
   # iterate over classifiers
   names.zip(classifiers).each do |name, clf|
-    ax = plt.subplot.(datasets.length, classifiers.length + 1, i)
-    clf.fit.(x_train, y_train)
-    scor = clf.score.(x_test, y_test)
+    ax = plt.subplot(datasets.length, classifiers.length + 1, i)
+    clf.fit(x_train, y_train)
+    scor = clf.score(x_test, y_test)
 
     # Plot the decision boundary.  For that, we will assign a color to each
     # point in the mesh [x_min, x_max]x[y_min, y_max]
     begin
       # not implemented for some
-      z = clf.decision_function.(mesh_points)
+      z = clf.decision_function(mesh_points)
     rescue
-      z = clf.predict_proba.(mesh_points)[all, 1]
+      z = clf.predict_proba(mesh_points)[all, 1]
     end
 
     # Put the result into a color plot
-    z = z.reshape.(xx.shape)
-    ax.contourf.(xx, yy, z, cmap: cm, alpha: 0.8)
+    z = z.reshape(xx.shape)
+    ax.contourf(xx, yy, z, cmap: cm, alpha: 0.8)
 
     # Plot also the training points
-    ax.scatter.(x_train[all, 0], x_train[all, 1], c: y_train, cmap: cm_bright)
+    ax.scatter(x_train[all, 0], x_train[all, 1], c: y_train, cmap: cm_bright)
     # and testing points
-    ax.scatter.(x_test[all, 0], x_test[all, 1], c: y_test, cmap: cm_bright, alpha: 0.6)
+    ax.scatter(x_test[all, 0], x_test[all, 1], c: y_test, cmap: cm_bright, alpha: 0.6)
 
-    ax.set_xlim.(np.min.(xx), np.max.(xx))
-    ax.set_ylim.(np.min.(yy), np.max.(yy))
-    ax.set_xticks.(PyCall.tuple())
-    ax.set_yticks.(PyCall.tuple())
-    ax.set_title.(name)
+    ax.set_xlim(np.min(xx), np.max(xx))
+    ax.set_ylim(np.min(yy), np.max(yy))
+    ax.set_xticks(PyCall.tuple())
+    ax.set_yticks(PyCall.tuple())
+    ax.set_title(name)
 
-    ax.text.(np.max.(xx) - 0.3, np.min.(yy) + 0.3, "%.2f" % scor, size: 15, horizontalalignment: 'right')
+    ax.text(np.max(xx) - 0.3, np.min(yy) + 0.3, "%.2f" % scor, size: 15, horizontalalignment: 'right')
 
     i += 1
   end
 end
 
-fig.subplots_adjust.(left: 0.02, right: 0.98)
-plt.show.()
+fig.subplots_adjust(left: 0.02, right: 0.98)
+plt.show()

examples/hist.rb

@@ -10,23 +10,23 @@
 pyimport 'matplotlib.mlab', as: 'mlab'
 pyimport 'matplotlib.pyplot', as: 'plt'
 
-np.random.seed.(0)
+np.random.seed(0)
 
 mu = 100
 sigma = 15
-x = mu + sigma * np.random.randn.(437)
+x = mu + sigma * np.random.randn(437)
 
 num_bins = 50
 
-fig, ax = plt.subplots.()
+fig, ax = *plt.subplots
 
-n, bins, patches = ax.hist.(x, num_bins, normed: 1)
+n, bins, patches = *ax.hist(x, num_bins, normed: 1)
 
-y = mlab.normpdf.(bins, mu, sigma)
-ax.plot.(bins, y, '--')
-ax.set_xlabel.('Smarts')
-ax.set_ylabel.('Probability density')
-ax.set_title.('Histogram of IQ: $\mu=100$, $\sigma=15$')
+y = mlab.normpdf(bins, mu, sigma)
+ax.plot(bins, y, '--')
+ax.set_xlabel('Smarts')
+ax.set_ylabel('Probability density')
+ax.set_title('Histogram of IQ: $\mu=100$, $\sigma=15$')
 
-fig.tight_layout.()
-plt.show.()
+fig.tight_layout()
+plt.show()