search_in_file 1.0.0 → 1.0.1

data/lib/search_in_file.rb

@@ -1,29 +1,36 @@
 require "search_in_file/version"
-require 'find'
+require "search_in_file/settings"
 
 require "file_parsers/txt_parser"
 require "file_parsers/pdf_parser"
-require "file_parsers/doc_parser"
+require "file_parsers/doc_parser" if Gem.win_platform?
 require "file_parsers/docx_parser"
 
+require "find"
+
 module SearchInFile
 
-  # search files by phrase
-  def self.search( dir_path, term )
+  def self.search( path, term )
+    is_document?( path ) ? search_in_file( path, term ) : search_in_directory( path, term )
+  end
+
+  def self.search_in_directory( path, term )
     results = []
-    each_file_in( dir_path ) do |path|  
-      # read file
-      parser_class    = Object.const_get( "#{File.extname( path )[1..-1].capitalize}Parser" )
-      file_content    = parser_class.new.read_file( path )
-      file_paragraphs = file_content.split(/\tor\n|\n/)
-      # search for phrase
-      paragraphs = []
-      file_paragraphs.each{ |p| paragraphs << p if p.include?(term) }
-      results << {file: path, paragraphs: paragraphs} if !paragraphs.empty?
+    each_file_in( path ) do |f_path| 
+      f_result = search_in_file( f_path, term )
+      results = results + f_result if !f_result.empty?
     end
     results
   end
 
+  def self.search_in_file( f_path, term )
+    term_paragraphs = []
+    file_paragraphs = paragraphs_of( f_path )
+    # search for phrase
+    file_paragraphs.each{ |p| term_paragraphs << p if p.include?(term) }
+    term_paragraphs.empty? ? [] : [{file: f_path, paragraphs: term_paragraphs}]    
+  end
+
   def self.content_of file
     class_name   = "#{extname( file )[1..-1].capitalize}Parser"
     parser_class = Object.const_get( class_name )
@@ -31,7 +38,7 @@ module SearchInFile
   end
 
   def self.paragraphs_of file
-    content_of( file ).split(/\tor\n|\n/)
+    content_of( file ).split(/\r/)
   end
 
   def self.find_by_type_in d_path, f_type
@@ -59,8 +66,7 @@ module SearchInFile
   end
 
   def self.is_document? f_name
-    f_type = extname( f_name )
-    ['.doc', '.docx', '.pdf', '.txt'].include? f_type
+    Settings.supported_docs.include?( extname(f_name) )
   end    
 
 end

data/lib/search_in_file/settings.rb

@@ -0,0 +1,17 @@
+require 'rbconfig'
+
+module Settings
+
+  def self.supported_docs
+    Gem.win_platform? ? ['.doc', '.docx', '.pdf', '.txt'] : ['.docx', '.pdf', '.txt']
+  end
+
+  def self.test_count
+    1
+  end
+
+  def self.fixtures_path
+    File.expand_path('../../..', __FILE__) + '/spec/fixtures'
+  end
+
+end

data/lib/search_in_file/version.rb

@@ -1,3 +1,3 @@
 module SearchInFile
-  VERSION = "1.0.0"
+  VERSION = "1.0.1"
 end

data/search_in_file.gemspec

@@ -9,7 +9,7 @@ Gem::Specification.new do |spec|
   spec.authors       = ["bmalets"]
   spec.email         = ["b.malets@gmail.com"]
   spec.summary       = %q{Search files with phrase by directory path}
-  spec.description   = %q{Search files with phrase by directory path}
+  spec.description   = %q{Search word or phrase in .pdf,.txt,.doc and .docx files by directory path}
   spec.homepage      = "https://github.com/bmalets/search_in_file/blob/master/README.md"
   spec.license       = "MIT"
 

data/spec/specs/search_engine_spec.rb

@@ -4,23 +4,55 @@ describe "File Search System" do
 
   describe "Search engine tests" do
 
-    10.times do |number|
-      it "Search phrase in directory number #{number}" do
+    Settings.test_count.times do |number|
 
-        dir_path  = File.expand_path('../..', __FILE__) + '/fixtures'
+      it "Search phrase in directory or in file number #{number}" do
+        dir_path  = Settings.fixtures_path
         term      = 'Often needs to retrieve documents on its'
         results   = SearchInFile.search( dir_path, term )
 
-        results.class == 'Array'
-        results.count == 9
+        expect(results.class).to eq(Array)
+        expect(results.count).to eq( Gem.win_platform? ? 7 : 4)
+
+        results.each do |result|
+          Settings.supported_docs.include? File.extname(result[:file])
+          File.exist?( result[:file] ).should be(true)
+          result[:paragraphs].each{ |paragraph| expect( paragraph.include?(term) ).to be true }
+        end
+      end
+
+      it "Search phrase in directory number #{number}" do
+        dir_path  = Settings.fixtures_path
+        term      = 'Often needs to retrieve documents on its'
+        results   = SearchInFile.search_in_directory( dir_path, term )
+
+        expect(results.class).to eq(Array)
+        expect(results.count).to eq( Gem.win_platform? ? 7 : 4)
 
-        results.each do |result| 
-          ['.doc', '.docx', '.pdf', '.txt'].include? File.extname(result[:file])
-          File.exist?( result[:file] ).should == true 
-          
-          result[:paragraphs].each{ |paragraph| paragraph.include? term }
+        results.each do |result|
+          Settings.supported_docs.include? File.extname(result[:file])
+          File.exist?( result[:file] ).should be(true)
+          result[:paragraphs].each{ |paragraph| expect( paragraph.include?(term) ).to be true }
         end
       end
+
+      Settings.supported_docs.each do |f_type|
+
+        it "Search phrase in #{f_type} file number #{number}" do
+          f_path    = Settings.fixtures_path + "/basic#{f_type}"
+          term      = 'Often needs to retrieve documents on its'
+          results   = SearchInFile.search_in_file( f_path, term )
+
+          expect(results.class).to eq(Array)
+          results.each do |result| 
+            Settings.supported_docs.include? File.extname(result[:file])
+            expect( File.exist?( result[:file] ) ).to be true
+            result[:paragraphs].each{ |paragraph| paragraph.include? term }
+          end
+        end
+
+      end
+
     end
 
   end

data/spec/specs/search_files_spec.rb

@@ -4,40 +4,38 @@ describe "Folder Search System" do
 
   describe "Search in folder tests" do
 
-    10.times do |number|
+    Settings.test_count.times do |number|
+
       it "Search TXT files test number #{number}" do
-        tmp_path = File.expand_path('../..', __FILE__) + '/fixtures'
-        files = SearchInFile.find_by_type_in( tmp_path, '.txt' )
+        tmp_path = Settings.fixtures_path
+        files    = SearchInFile.find_by_type_in( tmp_path, '.txt' )
         files.each{ |f| File.extname(f) == '.txt' }
         files.count.should == 2
       end
-    end
 
-    10.times do |number|
+
       it "Search PDF files test number #{number}" do
-        tmp_path = File.expand_path('../..', __FILE__) + '/fixtures'
+        tmp_path = Settings.fixtures_path
         files = SearchInFile.find_by_type_in( tmp_path, '.pdf' )
         files.each{ |f| File.extname(f) == '.pdf' }
-        files.count.should == 3
+        files.count.should == 4
       end
-    end
-
-    10.times do |number|
+   
       it "Search DOC files test number #{number}" do
-        tmp_path = File.expand_path('../..', __FILE__) + '/fixtures'
-        files = SearchInFile.find_by_type_in( tmp_path, '.doc' )
+        tmp_path = Settings.fixtures_path
+        files    = SearchInFile.find_by_type_in( tmp_path, '.doc' )
         files.each{ |f| File.extname(f) == '.doc' }
         files.count.should == 3
-      end
-    end
+      end 
+
 
-    10.times do |number|
       it "Search DOCX files test number #{number}" do
-        tmp_path = File.expand_path('../..', __FILE__) + '/fixtures'
-        files = SearchInFile.find_by_type_in( tmp_path, '.docx' )
+        tmp_path = Settings.fixtures_path
+        files    = SearchInFile.find_by_type_in( tmp_path, '.docx' )
         files.each{ |f| File.extname(f) == '.docx' }
         files.count.should == 2
       end
+
     end       
 
   end

data/spec/specs/search_phrase_spec.rb

@@ -4,36 +4,35 @@ describe "File Search System" do
 
   describe "Document Parsers Testing" do
 
-    10.times do |number|
+    Settings.test_count.times do |number|
+
       it "Read TXT files test number #{number}" do
-        tmp_path  = File.expand_path('../..', __FILE__) + '/fixtures/basic.txt'
+        tmp_path  = Settings.fixtures_path + '/basic.txt'
         data_read = TxtParser.new.read_file(tmp_path)
         data_read.include? 'Often needs to retrieve documents on its'
       end
-    end
 
-    10.times do |number|
+
       it "Read DOC files test number #{number}" do
-        tmp_path  = File.expand_path('../..', __FILE__) + '/fixtures/basic.doc'
+        tmp_path  = Settings.fixtures_path + '/basic.doc'
         data_read = DocParser.new.read_file(tmp_path)
         data_read.include? 'Often needs to retrieve documents on its'
-      end
-    end
+      end if Gem.win_platform?
+
 
-    10.times do |number|
       it "Read DOCX files test number #{number}" do
-        tmp_path  = File.expand_path('../..', __FILE__) + '/fixtures/basic.docx'
+        tmp_path  = Settings.fixtures_path + '/basic.docx'
         data_read = DocxParser.new.read_file(tmp_path)
         data_read.include? 'Often needs to retrieve documents on its'
       end
-    end  
 
-    10.times do |number|
+
       it "Read PDF files test number #{number}" do
-        tmp_path  = File.expand_path('../..', __FILE__) + '/fixtures/basic.pdf'
+        tmp_path  = Settings.fixtures_path + '/basic.pdf'
         data_read = PdfParser.new.read_file(tmp_path)
         data_read.include? 'Often needs to retrieve documents on its'
       end
+
     end       
 
   end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: search_in_file
 version: !ruby/object:Gem::Version
-  version: 1.0.0
+  version: 1.0.1
   prerelease: 
 platform: ruby
 authors:
@@ -91,7 +91,8 @@ dependencies:
     - - ! '>='
       - !ruby/object:Gem::Version
         version: '0'
-description: Search files with phrase by directory path
+description: Search word or phrase in .pdf,.txt,.doc and .docx files by directory
+  path
 email:
 - b.malets@gmail.com
 executables: []
@@ -108,6 +109,7 @@ files:
 - lib/file_parsers/pdf_parser.rb
 - lib/file_parsers/txt_parser.rb
 - lib/search_in_file.rb
+- lib/search_in_file/settings.rb
 - lib/search_in_file/version.rb
 - search_in_file.gemspec
 - spec/fixtures/basic.doc