chupa-text-decomposer-webkit 1.0.3 → 1.0.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 8517897f7ddbaa789f62ac121f66f875eb6a1dfe
-  data.tar.gz: 3783319afd02620126d691e9b4c3cce5fa04d8e4
+  metadata.gz: 0522e247250744ce0a2df3317cc3226948f152a6
+  data.tar.gz: 1162b7e926467f66728378fb4b64cb6928afef96
 SHA512:
-  metadata.gz: 8c097c1800d2cbafb859216bc2218ceab2f2005ae8b0c2f2d82eaf42d6c326a1d48411c14e14fdafd2ce54d5d603a658557e0fbfdd04031f02494e3d0c749a12
-  data.tar.gz: 9465dd3692b81e47fdb5d10147a66f4cc6744561db76b14157d66bcb7f2e3793e0ea4777cd1c0e6b16d7ffda6dc3784dc9f0e8f57faa04432e68e7f1867afa92
+  metadata.gz: 7f7030778edcfe73c3a7ef5211c48d7218efc7e10f38cbf6f4a6c9a1e3374c202501669503ded21b5db483b18e48b3406a0890fbb725859d3aca6338a59cd3c2
+  data.tar.gz: 0b6aa0901239e94e35d65d70054dcd5d0d58e584034f17208d989ffc51848f3782069b04e45890fb7661697511480d4d1f43fefb9565ff948c5d05cbf5a5b4a2

data/chupa-text-decomposer-webkit.gemspec

@@ -22,7 +22,7 @@ end
 
 Gem::Specification.new do |spec|
   spec.name = "chupa-text-decomposer-webkit"
-  spec.version = "1.0.3"
+  spec.version = "1.0.4"
   spec.homepage = "https://github.com/ranguba/chupa-text-decomposer-webkit"
   spec.authors = ["Kouhei Sutou"]
   spec.email = ["kou@clear-code.com"]

data/doc/text/news.md

@@ -1,5 +1,13 @@
 # News
 
+## 1.0.4: 2017-07-12
+
+### Improvements
+
+  * Added more logs.
+
+  * Supported proxy.
+
 ## 1.0.3: 2017-07-11
 
 ### Improvements

data/lib/chupa-text/decomposers/webkit.rb

@@ -19,7 +19,15 @@ require "webkit2-gtk"
 module ChupaText
   module Decomposers
     class WebKit < Decomposer
+      module LogTag
+        private
+        def log_tag
+          "[decomposer][webkit]"
+        end
+      end
+
       include Loggable
+      include LogTag
 
       registry.register("webkit", self)
 
@@ -50,126 +58,163 @@ module ChupaText
       end
 
       def decompose(data)
-        data.screenshot = create_screenshot(data.source)
+        screenshoter = Screenshoter.new(data)
+        screenshoter.run
         data[AVAILABLE_ATTRIBUTE_NAME] = !data.screenshot.nil?
         yield(data)
       end
 
-      private
-      def create_screenshot(data)
-        screenshot = nil
-
-        @@view_context ||= WebKit2Gtk::WebContext.new(ephemeral: true)
-        view = WebKit2Gtk::WebView.new(context: @@view_context)
-        window = Gtk::OffscreenWindow.new
-        window.set_default_size(800, 600)
-        window.add(view)
-        window.show_all
-
-        status = {
-          finished: false,
-          screenshot: nil,
-        }
-        prepare_screenshot(data, view, status)
-        debug do
-          "#{log_tag}[load][html] #{data.uri}"
-        end
-        view.load_html(data.body, data.uri.to_s)
+      class Screenshoter
+        include Loggable
+        include LogTag
 
-        main_context = GLib::MainContext.default
-        timeout(compute_timeout_second, view, main_context, status) do
-          until status[:finished]
-            main_context.iteration(true)
-          end
+        def initialize(data)
+          @data = data
+          @@view_context ||= create_view_context
+          @main_loop = GLib::MainLoop.new(nil, false)
+          @timeout_second = compute_timeout_second
         end
 
-        window.destroy
+        def run
+          view = WebKit2Gtk::WebView.new(context: @@view_context)
+          window = Gtk::OffscreenWindow.new
+          window.set_default_size(800, 600)
+          window.add(view)
+          window.show_all
 
-        status[:screenshot]
-      end
+          setup_callbacks(view)
 
-      def prepare_screenshot(data, view, status)
-        view.signal_connect("load-changed") do |_, load_event|
-          debug do
-            "#{log_tag}[load][#{load_event.nick}] #{view.uri}"
+          timeout(view) do
+            debug do
+              "#{log_tag}[load][HTML] #{@data.uri}"
+            end
+            view.load_html(@data.source.body, @data.source.uri.to_s)
+            @main_loop.run
           end
 
-          case load_event
-          when WebKit2Gtk::LoadEvent::FINISHED
-            view.get_snapshot(:full_document, :none) do |_, result|
-              status[:finished] = true
-              snapshot_surface = view.get_snapshot_finish(result)
-              unless snapshot_surface.width.zero?
-                png = convert_snapshot_surface_to_png(data, snapshot_surface)
-                status[:screenshot] = Screenshot.new("image/png",
-                                                     [png].pack("m*"),
-                                                     "base64")
-              end
+          window.destroy
+        end
+
+        private
+        def create_view_context
+          context = WebKit2Gtk::WebContext.new(ephemeral: true)
+          http_proxy = ENV["http_proxy"]
+          https_proxy = ENV["https_proxy"]
+          ftp_proxy = ENV["ftp_proxy"]
+          if http_proxy or https_proxy or ftp_proxy
+            proxy_settings = WebKit2Gtk::NetworkProxySettings.new
+            if http_proxy
+              proxy_settings.add_proxy_for_scheme("http", http_proxy)
+            end
+            if https_proxy
+              proxy_settings.add_proxy_for_scheme("https", https_proxy)
+            end
+            if ftp_proxy
+              proxy_settings.add_proxy_for_scheme("ftp", ftp_proxy)
             end
+            context.set_network_proxy_settings(:custom, proxy_settings)
           end
+          context
         end
-        view.signal_connect("load-failed") do |_, _, failed_uri, error|
-          status[:finished] = true
-          error do
-            message = "failed to load URI: #{failed_uri}: "
-            message << "#{error.class}(#{error.code}): #{error.message}"
-            "#{log_tag}[load][failed] #{message}"
+
+        def setup_callbacks(view)
+          view.signal_connect("load-changed") do |_, load_event|
+            debug do
+              "#{log_tag}[load][#{load_event.nick}] #{view.uri}"
+            end
+
+            case load_event
+            when WebKit2Gtk::LoadEvent::FINISHED
+              debug do
+                "#{log_tag}[screenshot][start] #{view.uri}"
+              end
+              view.get_snapshot(:full_document, :none) do |_, result|
+                @main_loop.quit
+                snapshot_surface = view.get_snapshot_finish(result)
+                debug do
+                  size = "#{snapshot_surface.width}x#{snapshot_surface.height}"
+                  "#{log_tag}[screenshot][finish] #{view.uri}: #{size}"
+                end
+                unless snapshot_surface.width.zero?
+                  png = convert_snapshot_surface_to_png(snapshot_surface)
+                  @data.screenshot = Screenshot.new("image/png",
+                                                    [png].pack("m*"),
+                                                    "base64")
+                end
+              end
+            end
+          end
+          view.signal_connect("load-failed") do |_, _, failed_uri, error|
+            @main_loop.quit
+            error do
+              message = "failed to load URI: #{failed_uri}: "
+              message << "#{error.class}(#{error.code}): #{error.message}"
+              "#{log_tag}[load][failed] #{message}"
+            end
+            true
           end
-          true
         end
-      end
 
-      def convert_snapshot_surface_to_png(data, snapshot_surface)
-        screenshot_width, screenshot_height = data.expected_screenshot_size
+        def convert_snapshot_surface_to_png(snapshot_surface)
+          screenshot_width, screenshot_height = @data.expected_screenshot_size
 
-        screenshot_surface = Cairo::ImageSurface.new(:argb32,
-                                                     screenshot_width,
-                                                     screenshot_height)
-        context = Cairo::Context.new(screenshot_surface)
-        context.set_source_color(:white)
-        context.paint
+          screenshot_surface = Cairo::ImageSurface.new(:argb32,
+                                                       screenshot_width,
+                                                       screenshot_height)
+          context = Cairo::Context.new(screenshot_surface)
+          context.set_source_color(:white)
+          context.paint
 
-        ratio = screenshot_width.to_f / snapshot_surface.width
-        context.scale(ratio, ratio)
-        context.set_source(snapshot_surface)
-        context.paint
+          ratio = screenshot_width.to_f / snapshot_surface.width
+          context.scale(ratio, ratio)
+          context.set_source(snapshot_surface)
+          context.paint
 
-        png = StringIO.new
-        screenshot_surface.write_to_png(png)
-        png.string
-      end
+          png = StringIO.new
+          screenshot_surface.write_to_png(png)
+          png.string
+        end
+
+        def timeout(view)
+          timeout_id = GLib::Timeout.add_seconds(@timeout_second) do
+            timeout_id = nil
+            error do
+              message = "timeout to load URI: #{@timeout_second}s: #{view.uri}"
+              message << ": loading" if view.loading?
+              "#{log_tag}[load][timeout] #{message}"
+            end
+            if view.loading?
+              view.signal_connect("close") do
+                @main_loop.quit
+                error do
+                  "#{log_tag}[load][closed] #{view.uri}"
+                end
+              end
+              view.try_close
+            else
+              @main_loop.quit
+            end
+            GLib::Source::REMOVE
+          end
 
-      def timeout(second, view, main_context, status)
-        timeout_id = nil
-        timeout_source = GLib::Timeout.source_new_seconds(second)
-        timeout_source.set_callback do
-          timeout_id = nil
-          status[:finished] = true
-          error do
-            message = "timeout to load URI: #{second}s: #{view.uri}"
-            "#{log_tag}[load][timeout] #{message}"
+          begin
+            yield
+          ensure
+            GLib::Source.remove(timeout_id) if timeout_id
           end
-          GLib::Source::REMOVE
         end
-        timeout_id = timeout_source.attach(main_context)
-        yield
-        GLib::Source.remove(timeout_id) if timeout_id
-      end
 
-      def compute_timeout_second
-        default_timeout = 5
-        timeout_string =
-          ENV["CHUPA_TEXT_DECOMPOSER_WEBKIT_TIMEOUT"] || default_timeout.to_s
-        begin
-          Integer(timeout_string)
-        rescue ArgumentError
-          default_timeout
+        def compute_timeout_second
+          default_timeout = 5
+          timeout_string =
+            ENV["CHUPA_TEXT_DECOMPOSER_WEBKIT_TIMEOUT"] || default_timeout.to_s
+          begin
+            Integer(timeout_string)
+          rescue ArgumentError
+            default_timeout
+          end
         end
       end
-
-      def log_tag
-        "[decomposer][webkit]"
-      end
     end
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: chupa-text-decomposer-webkit
 version: !ruby/object:Gem::Version
-  version: 1.0.3
+  version: 1.0.4
 platform: ruby
 authors:
 - Kouhei Sutou
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2017-07-11 00:00:00.000000000 Z
+date: 2017-07-12 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: chupa-text