moss_ruby 1.0.0 → 1.0.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 661f6ad9b42d35f82f5717e20785c4437c672c7a
-  data.tar.gz: 32f0decb5024d8baf1afb356b341da5a7e98b959
+  metadata.gz: de73a58e506ce9d6bd301468837e0015172bd9e8
+  data.tar.gz: 6a42a22ed34ef12ee88520f98679b3b05c67ba72
 SHA512:
-  metadata.gz: 0c2439a3d832bbf53f15e20c67707b2beb4ff16a4b03c412bc8af9ac4fc1548d586eb8b26d1d05ebf312ed2e9d83502af8c5424d1314a87f0ae2175bf8bcc79a
-  data.tar.gz: d2e039a786dce1a571bcc1b7e60882acc3502194794ceeacd4ae33326883f82a2bbcfae910e44d5a29fa9c5dd181dd60db19eed52fca35c2c5ff8883ca737c18
+  metadata.gz: 542d17634cb116fee0856d008be17594256d0f5440f776284a94ca93ffaa655fdf2279bd65019142aeff1610373b8f6091644f789f6e6f0a74adc83535ea4a11
+  data.tar.gz: 0f72cea79daa7112c91f094c12d28ba27a2b2bf73f2ca7fe0a601db04717d21e47b793ebd8a7608253048fda4ec69103a1adbe0bd562a3daa2183fd3710c72cc

data/lib/moss_ruby.rb

@@ -68,14 +68,25 @@ class MossRuby
 		content = IO.read(filename)
 		size = content.length
 
-		moss_server.write "file #{id} #{@options[:language]} #{size} #{file}\n"
-		moss_server.write content
+		if size > 0 
+			moss_server.write "file #{id} #{@options[:language]} #{size} #{file}\n"
+			moss_server.write content
+		end
+		puts "File upload size #{filename} #{size}"
+		begin
+			line = moss_server.read_nonblock(100)
+			puts "--------------------------------------------"
+			puts " From MOSS: #{line}"
+			puts "--------------------------------------------"
+		rescue IO::WaitReadable
+			#expected
+		end
 	end
 
-	def check(files_dict)
+	def check(files_dict, callback=nil)
 		# Chech that the files_dict contains valid filenames
 		files_dict[:base_files].each do |file_search|
-			if Dir[file_search].length == 0
+			if Dir.glob(file_search, File::FNM_CASEFOLD).length == 0
 				raise "Unable to locate base file(s) matching #{file_search}"
 			end
 		end
@@ -85,15 +96,17 @@ class MossRuby
 		end
 
 		files_dict[:files].each do |file_search|
-			if Dir[file_search].length == 0
+			if Dir.glob(file_search, File::FNM_CASEFOLD).length == 0
 				raise "Unable to locate base file(s) matching #{file_search}"
 			end
 		end
 
 		# Connect to the server
+		callback.call('Connecting to MOSS') unless callback.nil?
 		moss_server = TCPSocket.new @server, @port
 		begin
 			# Send header details
+			callback.call(' - Sending configuration details') unless callback.nil?
 			moss_server.write "moss #{@userid}\n"
 			moss_server.write "directory #{@options[:directory_submission] ? 1 : 0 }\n"
 			moss_server.write "X #{@options[:experimental_server] ? 1 : 0}\n"
@@ -103,26 +116,43 @@ class MossRuby
 			# Send language option
 			moss_server.write "language #{@options[:language]}\n"
 
+			callback.call(' - Checking language') unless callback.nil?
 			line = moss_server.gets
 			if line.strip() != "yes"
 				moss_server.write "end\n"
 				raise "Invalid language option."
 			end
 
-			files_dict[:base_files].each do |file_search|
-				Dir[file_search].each do |file|
+			count = 1
+			processing = files_dict[:base_files]
+			processing.each do |file_search|
+				callback.call(" - Sending base files #{count} of #{processing.count} - #{file_search}") unless callback.nil?
+				files = Dir.glob(file_search, File::FNM_CASEFOLD)
+				file_count = 1
+				files.each do |file|
+					callback.call("   - Base file #{file_count} of #{files.count} - #{file}") unless callback.nil?
 					upload_file moss_server, file
+					file_count += 1
 				end
+				count += 1
 			end
 
 			idx = 1
-			files_dict[:files].each do |file_search|
-				Dir[file_search].each do |file|
+			count = 1
+			processing = files_dict[:files]
+			processing.each do |file_search|
+				callback.call(" - Sending files #{count} of #{processing.count} - #{file_search}") unless callback.nil?
+				files = Dir.glob(file_search, File::FNM_CASEFOLD)
+				file_count = 1
+				files.each do |file|
+					callback.call("   - File #{idx} = #{file_count} of #{files.count} - #{file}") unless callback.nil?
 					upload_file moss_server, file, idx
 					idx += 1
+					file_count += 1
 				end
 			end
 
+			callback.call(" - Waiting for server response") unless callback.nil?
 			moss_server.write "query 0 #{@options[:comment]}\n"
 
 			result = moss_server.gets
@@ -134,17 +164,29 @@ class MossRuby
 		end
 	end
 
-	def extract_results(uri)
+	def extract_results(uri, min_pct = 10, callback = nil)
 		result = Array.new
 		begin
-			match = 0
+			match = -1
 			match_file = Array.new
 			data = Array.new
-			while true
+			to_fetch = get_matches(uri, min_pct, callback)
+			to_fetch.each do |id|
+				match += 1
+				callback.call("Checking match #{match + 1} (id #{id})") unless callback.nil?
+
 				# read the two match files
-				match_top = open("#{uri}/match#{match}-top.html").read
-				match_file[0] = open("#{uri}/match#{match}-0.html").read
-				match_file[1] = open("#{uri}/match#{match}-1.html").read
+				match_top = open("#{uri}/match#{id}-top.html").read
+
+				callback.call(" - checking match #{match} percents") unless callback.nil?
+				top = read_pcts match_top
+
+				next if Integer(top[:pct0]) < min_pct && Integer(top[:pct1]) < min_pct
+
+				callback.call(" - fetching #{match} html") unless callback.nil?
+
+				match_file[0] = open("#{uri}/match#{id}-0.html").read
+				match_file[1] = open("#{uri}/match#{id}-1.html").read
 
 				# puts match_top
 				# puts "---FILE0\n\n"
@@ -152,10 +194,12 @@ class MossRuby
 				# puts "---FILE1\n\n"
 				# puts match_file[1]
 
+				callback.call(" - extracting data for #{match}") unless callback.nil?
+
 				data[0] = read_data match_file[0]
 				data[1] = read_data match_file[1]
-				top = read_pcts match_top
-
+				
+				callback.call(" - adding #{match} result") unless callback.nil?
 				result << [ 
 					{
 						filename: 	data[0][:filename], 
@@ -168,8 +212,6 @@ class MossRuby
 						pct:  		Integer(top[:pct1])						
 					}
 				]
-
-				match += 1
 			end
 		rescue OpenURI::HTTPError
 			#end when there are no more matches -- indicated by 404 when accessing matches-n-top.html
@@ -180,6 +222,31 @@ class MossRuby
 
 	private
 
+# <tr><td><a href="http://moss.stanford.edu/results/916994582/match0.html">./done/COS10009-5/4936175/27258/ (51%)</a>
+#     </td><td><a href="http://moss.stanford.edu/results/916994582/match0.html">./done/COS10009-5/4928431/34630/ (51%)</a>
+# </td><td align="right">80
+# </td></tr>
+
+	def get_matches(uri, min_pct, callback)
+		result = Array.new
+		begin
+			callback.call(" - Reading match data") unless callback.nil?
+			page = open("#{uri}").read
+			regex = /<TR><TD><A HREF=".*?match(?<match_id>\d+).html">.*\((?<pct0>\d+)%\)<\/A>\n.*?<TD><A.*?((?<pct0>\d+)%\))/i
+			# puts "scanning page"
+			page.scan(regex).each do | match | 
+				id, pct0, pct1 = match
+				# puts "#{id}, #{pct0}, #{pct1}"
+				if Integer(pct0) >= min_pct || Integer(pct1) >= min_pct
+					result << id
+				end
+			end
+			callback.call(" - Found #{result.count} match with at least #{min_pct}% similar") unless callback.nil?
+		rescue
+		end
+		result
+	end
+
 	def strip_a(html)
 		html.gsub(/<A.*?>.*?<\/A>/, '')
 	end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: moss_ruby
 version: !ruby/object:Gem::Version
-  version: 1.0.0
+  version: 1.0.8
 platform: ruby
 authors:
 - Andrew Cain
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-10-01 00:00:00.000000000 Z
+date: 2015-06-26 00:00:00.000000000 Z
 dependencies: []
 description: Moss-ruby is an unofficial ruby gem for the Moss system for Detecting
   Software Plagiarism (http://theory.stanford.edu/~aiken/moss/)