data_collector 0.22.0 → 0.24.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e8333bf619cbafbac678351542ff0680abafd8d02fbe79ef286a1998eb85d3e4
-  data.tar.gz: bac65514f7f0314004feec335b20bc512e09ec1c62526321eefd61e7b98beece
+  metadata.gz: 7745e79eb3836ab3c469cc5da39f395d42c144940b98115c96058fb01f8a629c
+  data.tar.gz: 77aeb246a6a23477d07195c020091999825af21c9639c0a9679017daefeea9e9
 SHA512:
-  metadata.gz: d4b60f51ba32d710f89b396379847785c36d522fbfa190eb3a27527dea53e87a277e7f7d814e01a0639f8c64705b40b4621a21f19b0aa3afb1d8ee4aa56edd86
-  data.tar.gz: 1566edaa8ca58544d5747cd86b73b9f1d66ed52f684aecba62c4ceb695a096cb7b1f5063e973b42eafda65732bd6f5664210bd53ec83489f9c3b5d5b9d233935
+  metadata.gz: e48596ac6e5fc14be89c2aadc50416558fa4a08594d28cccee630c1157dc365a556999c82d303e4669c94e1db88d6b0cf3f044730d0057586220a6c3172b72a6
+  data.tar.gz: 2112dfa9191e8aa948317a16d581a9ea487327030cb013b43937757c144c33ca4fc975faee91fd6731d8f043ed0cefd5ecc28a562351c5de9b96196167871e36

data/lib/data_collector/input/dir.rb

@@ -9,8 +9,14 @@ module DataCollector
       end
 
       def run(should_block = false, &block)
-        @listener.start
-        sleep if should_block
+        @listener.start unless running?
+        if block_given?
+          while should_block && !paused?
+            yield
+          end
+        else
+          sleep if should_block && running? && !paused?
+        end
       end
 
       def running?
@@ -20,7 +26,10 @@ module DataCollector
       private
 
       def create_listener
-        @listener ||= Listen.to("#{@uri.host}#{@uri.path}", @options) do |modified, added, _|
+        absolute_path = File.absolute_path("#{@uri.host}#{@uri.path}")
+        raise DataCollector::Error, "#{@uri.to_s} not found" unless File.exist?(absolute_path)
+
+        @listener ||= Listen.to(absolute_path, @options) do |modified, added, _|
           files = added | modified
           files.each do |filename|
             handle_on_message(@input, @output, filename)

data/lib/data_collector/input/generic.rb

@@ -4,7 +4,7 @@ module DataCollector
   class Input
     class Generic
       def initialize(uri, options = {})
-        @uri = uri
+        @uri = URI(URI.decode_uri_component(uri.to_s)) #"#{uri.scheme}://#{URI.decode_uri_component(uri.host)}#{URI.decode_uri_component(uri.path)}"
         @options = options
         @running = false
 

data/lib/data_collector/input.rb

@@ -28,6 +28,10 @@ module DataCollector
     def from_uri(source, options = {})
       source = CGI.unescapeHTML(source)
       @logger.info("Reading #{source}")
+      raise DataCollector::Error 'from_uri expects a scheme like file:// of https://' unless source =~ /:\/\//
+
+      scheme, path = source.split('://')
+      source="#{scheme}://#{URI.encode_uri_component(path)}"
       uri = URI(source)
       begin
         data = nil
@@ -37,11 +41,12 @@ module DataCollector
         when 'https'
           data = from_https(uri, options)
         when 'file'
-          if File.directory?("#{uri.host}/#{uri.path}")
-            raise DataCollector::Error, "#{uri.host}/#{uri.path} not found" unless File.exist?("#{uri.host}/#{uri.path}")
+          absolute_path = File.absolute_path("#{URI.decode_uri_component("#{uri.host}#{uri.path}")}")
+          if File.directory?(absolute_path)
+            #raise DataCollector::Error, "#{uri.host}/#{uri.path} not found" unless File.exist?("#{uri.host}/#{uri.path}")
             return from_dir(uri, options)
           else
-            raise DataCollector::Error, "#{uri.host}/#{uri.path} not found" unless File.exist?("#{uri.host}/#{uri.path}")
+            #            raise DataCollector::Error, "#{uri.host}/#{uri.path} not found" unless File.exist?("#{uri.host}/#{uri.path}")
             data = from_file(uri, options)
           end
         when /amqp/
@@ -75,6 +80,7 @@ module DataCollector
     end
 
     def from_https(uri, options = {})
+      uri = URI.decode_uri_component("#{uri.to_s}")
       data = nil
       if options.with_indifferent_access.include?(:logging) && options.with_indifferent_access[:logging]
         HTTP.default_options = HTTP::Options.new(features: { logging: { logger: @logger } })
@@ -154,7 +160,9 @@ module DataCollector
 
     def from_file(uri, options = {})
       data = nil
-      absolute_path = File.absolute_path("#{uri.host}#{uri.path}")
+      uri = normalize_uri(uri)
+      absolute_path = File.absolute_path(uri)
+      raise DataCollector::Error, "#{uri.to_s} not found" unless File.exist?("#{absolute_path}")
       unless options.has_key?('raw') && options['raw'] == true
         @raw = data = File.read("#{absolute_path}")
         case File.extname(absolute_path)
@@ -226,5 +234,8 @@ module DataCollector
       return file_type
     end
 
+    def normalize_uri(uri)
+      "#{URI.decode_uri_component(uri.host)}#{URI.decode_uri_component(uri.path)}"
+    end
   end
 end

data/lib/data_collector/version.rb

@@ -1,4 +1,4 @@
 # encoding: utf-8
 module DataCollector
-  VERSION = "0.22.0"
+  VERSION = "0.24.0"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: data_collector
 version: !ruby/object:Gem::Version
-  version: 0.22.0
+  version: 0.24.0
 platform: ruby
 authors:
 - Mehmet Celik