CSV-datagen 0.0.1

data/README.txt

@@ -0,0 +1,17 @@
+This file contain all information about "CSV data generator" project.
+
+1. Overview
+2. Format of configuration file.
+
+Configuration file contain lines that present parameters of generated csv file. 
+[out,<output filename>]
+[lines,<number of lines in the output file>]
+[headers,<true if output file should contain headers row>]
+[field specification]
+
+2.1 Field specification
+
+integer - int,<name>,<max value> - the generated number will be between 0..max and had the given name.
+string - string,<name>,<length> - the generated string had the given name and length.
+float - float,<name>,<max>,<decimal> - the generated number had the given name, will be between 0..max and had decimal digits after comma.
+datetime - dtime,<name>,<date format>,<time format> . The date format is a string where user can use the following acronyms - DD for days, MM for months, YYYY or YY for years. The delimiter should be the '/' sign. The time format is "12" for twelve time format or "24" for twenty fours time format.

data/bin/gen-csv.rb

@@ -0,0 +1,40 @@
+#This file contain main script
+require 'lib\gen-data'
+require 'lib\read-params'
+require 'lib\version'
+require 'faster_csv'
+
+if $0 == __FILE__
+	puts 'Csv data generator version:'+GenCSV::Version
+	reader = GenCSV::ParamsReader.new('param.ini')
+	datagen = DataGenerator.new
+	csv = FasterCSV.open(reader.out, "w")
+	if reader.headers
+		# out header row
+		headers = []
+		reader.fields.each do | field |
+			headers.push(field[:name])
+		end
+		csv << headers
+	end
+	reader.lines.times do
+		# generate array of data
+		row = []
+		reader.fields.each do | field |
+			element = ''
+			if field[:type] == 'int' 
+				element = datagen.gen_int(field[:max]).to_s
+			 end
+			if field[:type] == 'string'
+				element = datagen.gen_string(field[:length]).to_s
+			end
+			if field[:type] == 'float'
+				element = datagen.gen_float(field[:max], field[:decimal]).to_s
+			end
+			row.push(element)	
+		end
+		csv << row
+	end
+	csv.close
+	puts "Data generation complete"
+end

data/lib/gen-data.rb

@@ -0,0 +1,122 @@
+#Class that generate all types of data
+
+class DataGenerator
+	def gen_int( interval )
+		rand(interval)
+	end
+	
+	def gen_string( len )
+		chars = ("a".."z").to_a  + ("A".."Z").to_a + ("0".."9").to_a
+		newstr = ""
+		1.upto(len) { |i| newstr << chars[rand(chars.size-1)] }
+		return newstr
+	end
+	
+	def raise_to_power( a, i )# a raise to i power
+		res = 1
+		i.times do
+			res *= a
+		end
+		return res
+	end
+
+	def gen_float( interval, drob = 0 )
+		c = rand( interval )
+		d = 0
+		if drob != 0
+			d = rand( raise_to_power( 10, drob ) )
+		end
+		number = (c.to_s + '.' + d.to_s).to_f
+	end
+	
+	def gen_bool
+		i = gen_int( 1000 )
+		res = false
+		if i > 500 
+			res = true
+		end
+		return res
+	end
+	
+	def gen_date( format)# return string created by the given format
+		 fa = format.split('/')
+		 del = '/'
+		 if fa.size == 1
+			fa = format.split(':')
+			del = ':'
+		 end
+		 if fa.size == 1 
+			puts 'date has unrecognized format'
+			return ''
+		end
+		res = ''
+		fa.each do | var |
+			var = var.downcase
+			case var
+				when 'd' , 'dd'
+					res += gen_int( 28 ).to_s
+				when 'm' , 'mm'
+					res += gen_int( 12 ).to_s
+				when 'yy'
+					res += gen_int( 100 ).to_s
+				when 'yyyy'
+					res += ( gen_int( 150 ) + 1900).to_s
+				else
+					puts 'date has unrecognized format'
+					return ''
+			end
+			res += del
+		end
+		res = res[0..-2]
+	end
+	
+	def gen_time( format)
+		fa = format.split(':')
+		if fa.size == 1
+			puts 'time has unrecognized format'
+			return ''
+		end
+		is24 = true
+		if fa[0] == '12'
+			is24 = false
+		end
+		if fa[0] == '12' or fa[0] == '24'
+			fa.shift
+		end
+		res = ''		
+		del = ':'
+		h = 0
+		fa.each do | var |
+			var = var.downcase
+			case var
+				when 'h'
+					if is24
+						res += gen_int(24).to_s
+					else
+						h = gen_int(12)
+						res += h.to_s
+					end
+				when 'm'
+					res += gen_int(60).to_s
+				when 'ms'
+					res += gen_int(1000).to_s
+			else
+				puts 'time has unrecognized format'
+				return ''
+			end
+			res += del
+		end		
+		res = res[0..-2]
+		if not is24 
+			if h > 12
+				res += ' PM'
+			else
+				res += ' AM'
+			end
+		end
+		return res
+	end
+	
+	def gen_weekday
+	end
+end

data/lib/read-params.rb

@@ -0,0 +1,129 @@
+#Class that read params from settings file
+
+module GenCSV
+
+DEFAULT_OUT_FILE = "out-data.csv"
+DEFAULT_LINES_COUNT = 100
+DEFAULT_HEADERS = false
+
+class ParamsReader
+	attr_reader :out #out file name
+	attr_reader :lines #lines count for generated csv
+	attr_reader :fields #array that contain information about fields and their params
+	attr_reader :headers # true if output file should contain headers row
+
+	def initialize( inifile ) #inifile - path to file, that contain description of csv file
+		content = File.open(inifile).readlines 
+		content = content.collect do | param | # remove all ignored symbols
+			param.strip
+		end
+		
+		if content.size == 0
+			@out = DEFAULT_OUT_FILE
+			@lines = DEFAULT_LINES_COUNT
+			@headers = DEFAULT_HEADERS
+			@fields = []
+			return ''
+		end
+		# get out file name
+		p1 = content[0].split(',')
+		if p1[0] == 'out' && p1.length == 2
+			@out = p1[1]
+			content.shift
+		else
+			@out = DEFAULT_OUT_FILE
+		end
+		
+		#get lines count
+		p2 = content[0].split(',')
+		if p2[0] == 'lines' && p2.length == 2
+			@lines = p2[1].to_i
+			content.shift
+		else
+			@lines = DEFAULT_LINES_COUNT
+		end
+		
+		#get headers
+		p3 = content[0].split(',')
+		if p3[0] == "headers" && p3.length == 2
+			@headers = true
+			content.shift
+		else
+			@headers = DEFAULT_HEADERS
+		end
+		
+		#get fields specification
+		@fields = []
+		content.each do | p |
+			pp = p.split(',')
+			if pp[0] == 'int'
+				pp.shift
+				@fields.push(parse_int( pp ))
+			end
+			if pp[0] == 'string'
+				pp.shift
+				@fields.push(parse_string( pp ))
+			end
+			if pp[0] == 'float'
+				pp.shift
+				@fields.push(parse_float( pp ))
+			end
+			if pp[0] == 'dtime'
+				pp.shift
+				@fields.push(parse_dtime( pp ))
+			end
+			if pp[0] == 'tdate'
+				pp.shift
+				@fields.push(parse_tdate( pp ))
+			end
+
+		end
+
+	end
+	
+	def parse_int( pp ) # return hash with all information about field
+		field = {}
+		field[:type] = 'int'
+		field[:name] = pp[0]
+		field[:max] = pp[1].to_i
+		return field
+	end
+	
+	def parse_string( pp )
+		field = {}
+		field[:type] = 'string'
+		field[:name] = pp[0]
+		field[:length] = pp[1].to_i
+		return field
+	end
+	
+	def parse_float( pp )
+		field = {}
+		field[:type] = 'float'
+		field[:name] = pp[0]
+		field[:max] = pp[1].to_i
+		field[:decimal] = pp[2].to_i
+		return field
+	end
+	
+	def parse_dtime( pp )
+		field = {}
+		field[:type] = 'dtime'
+		field[:name] = pp[0]
+		field[:dformat] = pp[1]
+		field[:tformat] = pp[2]
+		return field
+	end
+	
+	def parse_tdate( pp )
+		field = {}
+		field[:type] = 'dtime'
+		field[:name] = pp[0]
+		field[:tformat] = pp[1]
+		field[:dformat] = pp[2]
+		return field
+	end
+	
+end
+
+end

data/lib/version.rb

@@ -0,0 +1,3 @@
+module GenCSV
+	Version = "0.0.5"
+end

data/test/test-gen.rb

@@ -0,0 +1,57 @@
+#This file contain test cases for all working classes
+require 'test\unit'
+require 'lib\gen-data'
+
+class DataGeneratorTestCase < Test::Unit::TestCase
+	def setup
+		@gen = DataGenerator.new
+	end
+	def test_int_generation
+		first = @gen.gen_int(1000)
+		second = @gen.gen_int(1000)
+		assert_not_equal(first, second)
+	end
+	
+	def test_string_generation
+		first = @gen.gen_string( 20 )
+		second = @gen.gen_string( 20 )
+		assert_not_equal(first, second)
+	end
+	
+	def test_float_generation
+		first = @gen.gen_float( 1000, 4)
+		second = @gen.gen_float( 1000, 4)
+		assert_not_equal(first, second)
+	end
+
+
+	def test_bool_generation
+		i = 0
+		10.times do 
+			if @gen.gen_bool == true
+				i += 1
+			end
+		end
+		res = false
+		if i >=3 && i <= 70
+			res = true
+		end
+		assert_equal(true, res)
+	end
+	
+	def test_data_generation
+		assert_match(/[0-9]+\/[0-9]+\/[0-9]+/,@gen.gen_date("DD/MM/YY"))
+		assert_match(/[0-9]+\/[0-9]+\/[0-9][0-9][0-9][0-9]/,@gen.gen_date("DD/MM/YYYY"))
+		assert_match(/[0-9]+\:[0-9]+\:[0-9]+/,@gen.gen_date("DD:MM:YY"))
+		assert_match(/[0-9]+\:[0-9]+\:[0-9][0-9][0-9][0-9]/,@gen.gen_date("DD:MM:YYYY"))
+		assert_equal("",@gen.gen_date("h:m:ms"))
+		assert_equal("",@gen.gen_date("1;2;3"))
+	end
+	
+	def test_time_generation
+		assert_match(/[0-9]+\:[0-9]+\:[0-9]+\s[AM,PM]/,@gen.gen_time("12:h:m:ms"))
+		assert_match(/[0-9]+\:[0-9]+\:[0-9]+/,@gen.gen_time("h:m:ms"))
+		assert_equal("",@gen.gen_time("hh:mm:ms"))
+		assert_equal("",@gen.gen_time("12/h/m/ms"))
+	end
+end

data/test/test-read-params.rb

@@ -0,0 +1,31 @@
+#test app
+require 'test\unit'
+require 'lib\read-params'
+
+#sample
+class ParamsReadTestCase < Test::Unit::TestCase
+
+	def test_first # use for test including file 'param.ini'
+		reader = GenCSV::ParamsReader.new('.\param.ini')
+		assert_equal("out-data.csv", reader.out)
+		assert_equal(1000, reader.lines)
+		assert_equal(true, reader.headers)
+		assert_equal(5, reader.fields.size)
+	end
+	
+	def test_default_1_params
+		reader = GenCSV::ParamsReader.new('.\default-1.ini')
+		assert_equal(GenCSV::DEFAULT_OUT_FILE, reader.out)
+		assert_equal(GenCSV::DEFAULT_LINES_COUNT, reader.lines)
+		assert_equal(GenCSV::DEFAULT_HEADERS, reader.headers)
+		assert_equal(0, reader.fields.size)
+	end
+
+	def test_default_2_params
+		reader = GenCSV::ParamsReader.new('.\default-2.ini')
+		assert_equal(GenCSV::DEFAULT_OUT_FILE, reader.out)
+		assert_equal(GenCSV::DEFAULT_LINES_COUNT, reader.lines)
+		assert_equal(GenCSV::DEFAULT_HEADERS, reader.headers)
+		assert_equal(1, reader.fields.size)		
+	end
+end

metadata

@@ -0,0 +1,51 @@
+--- !ruby/object:Gem::Specification 
+rubygems_version: 0.9.2
+specification_version: 1
+name: CSV-datagen
+version: !ruby/object:Gem::Version 
+  version: 0.0.1
+date: 2007-03-29 00:00:00 +04:00
+summary: A free tool for generating random data for testing purposes.
+require_paths: 
+- lib
+email: dmtmax@gmail.com
+homepage: http://blogbydmt/blogspot.com
+rubyforge_project: 
+description: 
+autorequire: csvdatagen
+default_executable: 
+bindir: bin
+has_rdoc: false
+required_ruby_version: !ruby/object:Gem::Version::Requirement 
+  requirements: 
+  - - ">"
+    - !ruby/object:Gem::Version 
+      version: 0.0.0
+  version: 
+platform: ruby
+signing_key: 
+cert_chain: 
+post_install_message: 
+authors: 
+- Dmitry Maksimov
+files: 
+- bin/gen-csv.rb
+- lib/gen-data.rb
+- lib/read-params.rb
+- lib/version.rb
+- README.txt
+test_files: 
+- test/test-gen.rb
+- test/test-read-params.rb
+rdoc_options: []
+
+extra_rdoc_files: 
+- README.txt
+executables: []
+
+extensions: []
+
+requirements: []
+
+dependencies: []
+