Add RemoteInput.open

lib/remote_input.rb

@@ -1,6 +1,80 @@
-require_relative "remote_input/downloader"
 require_relative "remote_input/cache-path"
+require_relative "remote_input/downloader"
 require_relative "remote_input/zip-extractor"
 
-module RemoteInput
+class RemoteInput
+  class << self
+    def open(...)
+      input = new(...)
+      if block_given?
+        begin
+          yield(input)
+        ensure
+          input.close
+        end
+      else
+        input
+      end
+    end
+  end
+
+  def initialize(url,
+                 *fallback_urls,
+                 encoding: nil,
+                 internal_encoding: nil,
+                 external_encoding: nil,
+                 **http_options)
+    @url = URI(url)
+    @encoding = encoding
+    @internal_encoding = internal_encoding
+    @external_encoding = external_encoding
+    @downloader = Downloader.new(url, *fallback_urls, **http_options)
+    @cache_path = nil
+    @local_file = nil
+    @closed = false
+  end
+
+  def read(maxlen=nil, out_string=nil)
+    local_file.read(maxlen, out_string)
+  end
+
+  def close
+    @local_file.close if @local_file and not @local_file.closed?
+    @closed = true
+  end
+
+  def clear_cache
+    cache_path.remove
+  end
+
+  private
+
+  def path
+    cache_path.base_dir + File.basename(normalize_path)
+  end
+
+  def cache_path
+    return @cache_path if @cache_path
+    dirname = File.dirname(normalize_path).delete_suffix("/")
+    cache_id = "#{@url.host}#{dirname}".tr("/", "-")
+    @cache_path = CachePath.new(cache_id)
+  end
+
+  def normalize_path
+    url_path = @url.path
+    url_path = "/" if url_path.empty?
+    url_path += "data" if url_path.end_with?("/")
+    url_path
+  end
+
+  def local_file
+    raise IOError, "closed stream" if @closed
+    return @local_file if @local_file
+    @downloader.download(path)
+    options = {}
+    options[:encoding] = @encoding if @encoding
+    options[:internal_encoding] = @internal_encoding if @internal_encoding
+    options[:external_encoding] = @external_encoding if @external_encoding
+    @local_file = path.open(**options)
+  end
 end

lib/remote_input/cache-path.rb

@@ -1,4 +1,4 @@
-module RemoteInput
+class RemoteInput
   class CachePath
     def initialize(id)
       @id = id

lib/remote_input/downloader.rb

@@ -9,7 +9,7 @@
 require_relative "error"
 require_relative "version"
 
-module RemoteInput
+class RemoteInput
   class Downloader
     class TooManyRedirects < Error; end
 

lib/remote_input/error.rb

@@ -1,4 +1,4 @@
-module RemoteInput
+class RemoteInput
   class Error < StandardError
   end
 end

lib/remote_input/version.rb

@@ -1,3 +1,3 @@
-module RemoteInput
+class RemoteInput
   VERSION = "0.0.0"
 end

lib/remote_input/zip-extractor.rb

@@ -1,6 +1,6 @@
 require 'zip'
 
-module RemoteInput
+class RemoteInput
   class ZipExtractor
     def initialize(path)
       @path = path

test/test-remote-input.rb

@@ -0,0 +1,99 @@
+require_relative "helper"
+
+class RemoteInputTest < Test::Unit::TestCase
+  def test_read_once
+    open_input do |input|
+      input.send(:path).parent.mkpath
+      input.send(:path).write("12")
+      assert_equal("12", input.read)
+    end
+  end
+
+  def test_read_twice
+    open_input do |input|
+      input.send(:path).parent.mkpath
+      input.send(:path).write("12")
+      assert_equal("1", input.read(1))
+      assert_equal("2", input.read(1))
+    end
+  end
+
+  def test_read_with_encoding
+    open_input(encoding: "Windows-31J:UTF-8") do |input|
+      input.send(:path).parent.mkpath
+      input.send(:path).write("入力", encoding: "Windows-31J:UTF-8")
+      assert_equal("入力", input.read)
+    end
+  end
+
+  def test_read_after_close
+    open_input do |input|
+      input.close
+      assert_raise(IOError.new("closed stream")) do
+        input.read
+      end
+    end
+  end
+
+  def test_open_with_block
+    opened_input = nil
+    open_input do |input|
+      opened_input = input
+    end
+    assert_raise(IOError.new("closed stream")) do
+      opened_input.read
+    end
+  end
+
+  def test_open_with_block_raised
+    opened_input = nil
+    assert_raise(RuntimeError.new("error in block")) do
+      open_input do |input|
+        opened_input = input
+        raise "error in block"
+      end
+    end
+    assert_raise(IOError.new("closed stream")) do
+      opened_input.read
+    end
+  end
+
+  data("no path",       ["/example.com/data",     "https://example.com"])
+  data("root",          ["/example.com/data",     "https://example.com/"])
+  data("file",          ["/example.com/file",     "https://example.com/file"])
+  data("query",         ["/example.com/file",     "https://example.com/file?a=b"])
+  data("directory",     ["/example.com-a/data",   "https://example.com/a/"])
+  data("nested file",   ["/example.com-a/file",   "https://example.com/a/file"])
+  data("deeply nested", ["/example.com-a-b/file", "https://example.com/a/b/file"])
+  def test_path(data)
+    expected, url = data
+    RemoteInput.open(url) do |input|
+      assert do
+        input.send(:path).to_s.end_with?(expected)
+      end
+    end
+  end
+
+  def test_clear_cache
+    open_input do |input|
+      input.send(:path).parent.mkpath
+      input.send(:path).write("1")
+      input.clear_cache
+      assert do
+        not input.send(:path).parent.exist?
+      end
+    end
+  end
+
+  private
+
+  def open_input(**options)
+    input = RemoteInput.open("https://example.com/file", **options)
+    begin
+      yield(input)
+    ensure
+      input.close
+      input.clear_cache
+    end
+  end
+end