presume 0.0.2

data/spec/classifide_spec.rb

@@ -0,0 +1,17 @@
+require "spec_helper"
+
+describe Classifide do
+
+	classifide_params = {:phone => nil, :email => nil, :address => nil, :section => nil, :professions => "Project Coordinator ", :companies => "Thorns and Thistles Ltd.", :schools => nil, :dates => "September 2012 – January 2014", :dates_2 => nil, :number_of_words => 6, :many_words => true, :verbs => nil, :cities => "Toronto, Ontario", :text => "Project Coordinator, Thorns and Thistles Ltd., Toronto, Ontario, September 2012 – January 2014", :type => "header_x"}
+	let(:classifide) {Classifide.new(classifide_params)}
+
+	classifide_params_2 = {:phone => nil, :email => nil, :address => nil, :section => "of Toronto School of Continuing Education", :professions => nil, :companies => nil, :schools => "University of Toronto School of Continuing", :dates => "Summer 2013", :dates_2 => nil, :cities => nil, :number_of_words => 8, :many_words => true, :verbs => nil, :text => "University of Toronto School of Continuing EducationSummer 2013", :type => "header_x"}
+	
+	let(:classifide_2) {Classifide.new(classifide_params_2)}
+
+	it "checks more words than" do
+		expect(classifide.more_words_than?(5)).to be_truthy
+		expect(classifide.more_words_than?(10)).to be_falsy
+	end
+
+end

data/spec/header_spec.rb

@@ -0,0 +1,73 @@
+require "spec_helper"
+
+describe Header do
+
+	classifide_params = {:phone => nil, :email => nil, :address => nil, :section => nil, :professions => "Project Coordinator ", :companies => nil, :schools => nil, :dates => "September 2012 – January 2014", :cities => nil, :number_of_words => 7, :many_words => true, :verbs => nil, :text => "Project Coordinator, Thorns and Thistles Ltd., Toronto, Ontario, September 2012 – January 2014", :type => "header_x"}
+	classifide = Classifide.new(classifide_params)	
+	let(:header) {Header.new(classifide)}
+	
+	it "finds start time text" do
+		expect(header.start_time_text?).to eq("September 2012")
+	
+	end
+
+	it "finds start year" do
+		expect(header.start_year).to eq("2012")
+	
+	end
+
+	it "finds start time number" do
+		expect(header.start_time_number).to eq(2012 + (8/12).to_f)
+	end
+	
+
+	it "finds end time text" do
+		expect(header.end_time_text?).to eq("January 2014")
+	end
+
+	it "finds end year" do
+		expect(header.end_year).to eq("2014")
+	end
+
+	it "finds end time number" do
+		expect(header.end_time_number).to eq(2014 + (0/12).to_f)
+	end
+
+	it "finds end time number" do
+		header.instance_variable_set("@dates", 'September 2014 to present')
+		puts header.dates
+		expect(header.start_time_text?).to eq("September 2014")
+		expect(header.end_time_number).to eq(2015 + (6/12).to_f)
+	end
+
+	it "checks if end time text is current job" do
+		expect(header.current?).to be_falsy
+	end
+
+	it "removes date from text" do
+		header.remove_date
+		expect(header.clean_profession).to eq("Project Coordinator, Thorns and Thistles Ltd., Toronto, Ontario, ")
+	end
+
+	it "removes city from text" do
+		header.remove_date
+		header.remove_city
+		expect(header.clean_profession).to eq("Project Coordinator, Thorns and Thistles Ltd., , ")
+	end
+
+	it "removes companies" do
+		header.remove_date
+		header.remove_city
+		header.remove_institution
+		expect(header.clean_profession).to eq("Project Coordinator, , , ")
+	end
+
+	it "removes companies" do
+		header.remove_date
+		header.remove_city
+		header.remove_institution
+		header.remove_all_but_profession
+		expect(header.clean_profession).to eq("Project Coordinator")
+	end
+
+end

data/spec/presume_spec.rb

@@ -0,0 +1,72 @@
+require 'spec_helper'
+
+describe Presume do
+	
+	doc = Docx::Document.open('sample_resume.docx').text
+    
+	let(:presume){Presume.new(doc, "Sample Person")}
+
+  searchable_params = {"Marketing Coordinator" => 1, "Project Coordinator" => 1}
+
+  skill_params = {"present investment proposals" => 0, "clear communication" => 0}
+
+  	it "should return sections" do
+	    expect(presume.sections).to_not be_nil
+  	end
+
+  	it "should return section ids" do
+	    expect(presume.get_sections_id.kind_of?(Array)).to be_truthy
+  	end
+
+  	it "should return section info" do
+	    expect(presume.get_sections_info.kind_of?(Array)).to be_truthy
+  	end
+
+  	it "should return header ids" do
+	    expect(presume.get_headers_id.kind_of?(Array)).to be_truthy
+  	end
+
+  	it "should return header info" do
+	    expect(presume.get_headers_info.kind_of?(Array)).to be_truthy
+  	end
+
+  	it "should return children" do
+	    expect(presume.sections[0].children.length).to be < presume.all_types[0].length
+  	end
+
+    it "should get matched searchables" do
+      presume.instance_variable_set("@searchables", Searchables.new(searchable_params))
+      presume.instance_variable_set("@matched_searchables", [])
+      presume.headers.values.each do |header|
+        presume.instance_variable_get("@searchables").all.values.each do |searchable|
+          puts header.text
+          puts header.dates
+          puts searchable.duration
+          puts header.duration
+
+          if header.duration > searchable.duration
+            puts "duration is greater"
+            puts searchable.regex
+            if !header.text[searchable.regex].nil?
+              puts "match"
+              presume.instance_variable_set("@matched_searchables", presume.instance_variable_get("@matched_searchables")) + [[searchable,header]]
+            end
+          end
+        end
+      end
+      presume.instance_variable_get("@matched_searchables")
+    end
+
+  	it "should find similar searchable and time" do
+    	expect(presume.searchables?(searchable_params).class.name).to eq("Array") 
+      expect(presume.instance_variable_get("@matched_searchables").length).to eq(1)
+      expect(presume.instance_variable_get("@matched_searchables")[0][0].raw_name).to eq("Project Coordinator")
+    end
+
+  	it "should find similar skills and times" do
+      expect(presume.skills?(skill_params).class.name).to eq("Array")
+      expect(presume.instance_variable_get("@matched_searchables").length).to eq(1)
+      expect(presume.instance_variable_get("@matched_searchables")[0][1].text).to eq("Assisted in the financial reporting, pro forma analysis, and presentation of investment proposals to International stakeholders, leasing agents and project managers in Cantonese, Mandarin and English using Excel and Power,Point. ")
+  	end
+
+end

data/spec/resume_builder_spec.rb

@@ -0,0 +1,213 @@
+require 'spec_helper'
+
+describe ResumeBuilder do
+	
+  doc = Docx::Document.open('sample_resume.docx').text
+	
+	resume_classifier = ResumeClassifier.new(doc, "Sample Person", "fake_presume_object")
+	resume_classifier.classify
+
+	let(:resume_builder){ResumeBuilder.new(resume_classifier.classifide_lines)}
+
+    it 'should be properly setup' do
+      expect(resume_classifier.classifide_lines.length).to be > 10
+      expect(resume_builder.classifides.length).to be > 10
+    end
+
+  	it "it should complete first pass" do
+  		resume_builder.first_pass
+  		expect(resume_builder.classifides[0].type).to eq("section")
+  	end
+
+    it "it should complete second pass" do
+      resume_builder.first_pass
+      resume_builder.second_pass
+      expect(resume_builder.classifides[0].type).to eq("section")
+      resume_builder.classifides.values.each do |classifide|
+        puts ("#{classifide.type} | #{classifide.text} | #{classifide.professions} | #{classifide.many_words?} ")
+      end
+    end
+
+    it "it should build resume" do
+      resume_builder.first_pass
+      resume_builder.second_pass
+      resume_builder.reset_header_x_start
+      expect(resume_builder.instance_variable_get("@header_x_start")).to eq(true)
+      resume_builder.instance_variable_get("@length").times do |n|
+            resume_builder.set_classifide(n)
+            @classifide = resume_builder.instance_variable_get("@classifide")
+            expect(@classifide.id).to eq(n)
+            unless @classifide.type?
+                unless @classifide.type == "name" or @classifide.type == "email" or @classifide.type == "phone" or @classifide.type == "address"
+                    resume_builder.send(@classifide.type + "_build")
+                end
+            end 
+        end            
+    end
+
+    it "should build section" do
+      resume_builder.first_pass
+      resume_builder.second_pass
+      resume_builder.instance_variable_get("@length").times do |n|
+            resume_builder.set_classifide(n)
+            @classifide = resume_builder.instance_variable_get("@classifide")
+            
+            unless @classifide.type?
+                unless @classifide.type == "name" or @classifide.type == "email" or @classifide.type == "phone" or @classifide.type == "address"
+                    if @classifide.type == "section"
+                      resume_builder.send(@classifide.type + "_build")   
+                      expect(resume_builder.instance_variable_get("@header_number")).to be_nil
+                      expect(resume_builder.instance_variable_get("@section_number")).to eq(n)
+                      expect(resume_builder.instance_variable_get("@header_x_start")).to be_truthy
+                      expect(resume_builder.sections.values.length).to be > 0
+                      expect(resume_builder.all_types.length).to be > 0
+                    end
+                end
+            end 
+        end            
+      
+    end
+
+    it "should build header" do
+      resume_builder.first_pass
+      resume_builder.second_pass
+      resume_builder.instance_variable_get("@length").times do |n|
+            resume_builder.set_classifide(n)
+            @classifide = resume_builder.instance_variable_get("@classifide")
+
+            if resume_builder.all_types.values.nil?
+              check_all_types_length = 0
+            else  
+              check_all_types_length = resume_builder.all_types.values.length
+            end
+
+            if (parent_types =resume_builder.all_types[resume_builder.section_number?]).nil?
+              check_all_types_parent_length = 0
+            else  
+              check_all_types_parent_length = parent_types.length
+            end
+
+            if (headers = resume_builder.headers.values).nil?
+              check_headers_length = 0
+            else  
+              check_headers_length = headers.length
+            end
+
+            
+            unless @classifide.type?
+                unless @classifide.type == "name" or @classifide.type == "email" or @classifide.type == "phone" or @classifide.type == "address"
+                    if @classifide.type == "header"
+                      resume_builder.send(@classifide.type + "_build")   
+                      expect(resume_builder.instance_variable_get("@header_number")).to eq(n)
+                      expect(resume_builder.instance_variable_get("@header_x_start")).to be_truthy
+                      expect(resume_builder.bullets.values.length).to be > check_headers_length
+                      expect(resume_builder.all_types.values.length).to be > check_all_types_length
+                      expect(resume_builder.all_types[resume_builder.section_number?].length).to be > check_all_types_parent_length
+                    end
+                end
+            end 
+        end            
+      
+    end
+
+    it "should build bullet" do
+      resume_builder.first_pass
+      resume_builder.second_pass
+      resume_builder.instance_variable_get("@length").times do |n|
+            resume_builder.set_classifide(n)
+            @classifide = resume_builder.instance_variable_get("@classifide")
+
+            if resume_builder.all_types.values.nil?
+              check_all_types_length = 0
+            else  
+              check_all_types_length = resume_builder.all_types.values.length
+            end
+
+            if (parent_types =resume_builder.all_types[resume_builder.header_number?]).nil?
+              check_all_types_parent_length = 0
+            else  
+              check_all_types_parent_length = parent_types.length
+            end
+
+            if (bullets = resume_builder.bullets.values).nil?
+              check_bullets_length = 0
+            else  
+              check_bullets_length = bullets.length
+            end
+
+            unless @classifide.type?
+                unless @classifide.type == "name" or @classifide.type == "email" or @classifide.type == "phone" or @classifide.type == "address"
+                    if @classifide.type == "bullet"
+                      resume_builder.send(@classifide.type + "_build")
+                      expect(resume_builder.instance_variable_get("@header_x_start")).to be_truthy
+                      expect(resume_builder.bullets.values.length).to be > check_bullets_length
+                      expect(resume_builder.all_types.values.length).to be > check_all_types_length
+                      expect(resume_builder.all_types[resume_builder.header_number?].length).to be > check_all_types_parent_length
+                    end
+                end
+            end 
+        end            
+      
+    end
+
+    it "should build header_x" do
+      resume_builder.first_pass
+      resume_builder.second_pass
+      resume_builder.instance_variable_get("@length").times do |n|
+            resume_builder.set_classifide(n)
+            @classifide = resume_builder.instance_variable_get("@classifide")
+            
+            if resume_builder.all_types.values.nil?
+              check_all_types_length = 0
+            else  
+              check_all_types_length = resume_builder.all_types.values.length
+            end
+
+            if (parent_types =resume_builder.all_types[resume_builder.section_number?]).nil?
+              check_all_types_parent_length = 0
+            else  
+              check_all_types_parent_length = parent_types.length
+            end
+
+            if (headers = resume_builder.headers.values).nil?
+              check_headers_length = 0
+            else  
+              check_headers_length = headers.length
+            end            
+            
+            unless @classifide.type?
+                unless @classifide.type == "name" or @classifide.type == "email" or @classifide.type == "phone" or @classifide.type == "address"
+                    if @classifide.type == "header_x"
+                      resume_builder.send(@classifide.type + "_build")
+                      
+                      if resume_builder.classifide_after == "header_x"
+                        if resume_builder.classifide_after == "header_x" and resume_builder.instance_variable_get("@header_x_start") == "almost"
+                        else
+                          expect(resume_builder.instance_variable_get("@header_x_start")).to be_falsy
+                        end
+                      
+                        expect(resume_builder.bullets.values.length).to be > check_headers_length
+                        expect(resume_builder.all_types.values.length).to be > check_all_types_length
+                        expect(resume_builder.all_types[resume_builder.section_number?].length).to be > check_all_types_parent_length
+                      end                      
+                    end
+                end
+            end 
+        end            
+      
+    end
+
+    it "should add to all_types" do
+      allow(resume_builder).to receive(:classifide) {"This is a fake classifide"}
+      resume_builder.all_types
+      expect(resume_builder.all_types.values.length).to eq(0)
+      resume_builder.add_to_all_types(1)
+      expect(resume_builder.all_types.values.length).to eq(1)
+      resume_builder.add_to_all_types(1)
+      expect(resume_builder.all_types.values.length).to eq(1)
+      expect(resume_builder.all_types[1].length).to eq(2)
+    end
+
+
+  	
+end

data/spec/resume_classifier_spec.rb

@@ -0,0 +1,140 @@
+require "spec_helper"
+require "benchmark"
+
+describe ResumeClassifier do
+
+
+	doc = Docx::Document.open('sample_resume.docx').text
+    
+
+
+	let(:resume_classifier){ResumeClassifier.new(doc, "Sample Person", "fake_presume_object")}
+
+  	it "should return resume text!" do
+	    expect(resume_classifier.text).to_not be_nil
+  	end
+
+  	it "separates tabbed words" do
+  		resume_classifier.instance_variable_set(:@text, "University of ChicagoJune 2008")
+  		resume_classifier.separate_tabbed_words
+  		expect(resume_classifier.text).to eq("University of Chicago,June 2008")
+  	end
+
+  	it "removes extra spaces" do
+  		resume_classifier.instance_variable_set(:@text, "University of Chicago      June 2008")
+  		resume_classifier.remove_extra_spaces
+  		expect(resume_classifier.text).to eq("University of Chicago  ,,,,June 2008")
+  	end
+
+
+	it "removes blank lines" do
+		
+		resume_classifier.instance_variable_set("@lines", ["line with code", "          "])
+		
+		time = Benchmark.measure do
+			resume_classifier.remove_blanks
+		end
+		puts time
+
+		expect(resume_classifier.lines).to eq(["line with code"])
+		
+	end
+
+	it "checks number of lines" do
+		resume_classifier.instance_variable_set("@lines", ["line with code", "          "])
+		
+		time = Benchmark.measure do
+			resume_classifier.number_of_lines
+		end
+		puts time
+
+		expect(resume_classifier.number_of_lines).to eq(2)
+	end
+
+	
+  	it "splits text by new lines" do
+  		time = Benchmark.measure do
+			resume_classifier.split_text
+		end
+  		puts time
+
+	    expect(resume_classifier.lines).to_not be_nil
+  	end
+	
+	it "removes cities and dates from line" do
+		resume_classifier.instance_variable_set("@line", "line with, Toronto, Ontario Sep 2009")
+		expect(resume_classifier.remove_dates_and_cities).to eq("line with,  ")
+	end
+
+	it "counts words without date and city" do
+		resume_classifier.instance_variable_set("@line", "line with, Toronto, Ontario Sep 2009")
+		expect(resume_classifier.number_of_words).to eq(2)
+	end
+
+  	it "sets number of lines" do
+  		resume_classifier.set_line_number(0)
+  		expect(resume_classifier.instance_variable_get(:@line_number)).to eq(0)
+  	end
+
+  	it "sets number of lines" do
+  		resume_classifier.set_line(0)
+  		puts resume_classifier.instance_variable_get(:@lines)
+  		expect(resume_classifier.instance_variable_get(:@line)).to eq("Professional Summary")
+  	end
+
+  	it "checks classifications" do
+  		resume_classifier.split_text
+  		expect(resume_classifier.number_of_lines).to be > 10
+
+  		resume_classifier.number_of_lines.times do |n|
+
+	  		resume_classifier.set_line_number(n)
+	  		expect(resume_classifier.instance_variable_get(:@line_number)).to eq(n)
+	  		resume_classifier.set_line(n)
+	  		time = Benchmark.measure do
+	  			resume_classifier.check_classifications
+	  		end
+
+	  		puts time
+
+	  	end
+  	end
+
+  	it "merges to classifide lines" do
+  		resume_classifier.split_text
+  		expect(resume_classifier.number_of_lines).to be > 10
+
+  		resume_classifier.number_of_lines.times do |n|
+
+	  		resume_classifier.set_line_number(n)
+	  		expect(resume_classifier.instance_variable_get(:@line_number)).to eq(n)
+	  		resume_classifier.set_line(n)
+
+	  		resume_classifier.check_classifications
+	  		time = Benchmark.measure do
+	  			resume_classifier.merge_to_classifide_lines
+	  			expect(resume_classifier.classifide_lines.length).to eq(n + 1)
+	  		end
+	  		puts time
+	  	end
+  	end
+
+=begin  		
+
+	it "classifies lines by resume components" do
+
+		time = Benchmark.measure do
+			resume_classifier.classify
+		end
+  		puts time
+		
+	    expect(resume_classifier.classifide_lines[0].section).to eq("Professional Summary")
+	    expect(resume_classifier.classifide_lines[1].phone).to be_nil
+  	end
+
+  	it "sorts classified lines into sections, headers, and bullets" do
+		
+  	end
+=end
+
+end