rtesseract 3.0.0 → 3.0.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 00e35fcd674159d47c7252be586ba214c42f374d1e671cab2ae8a47672237d92
-  data.tar.gz: e6601b67346092513af4399c49bb161b99afccb84f277395fe8f27417d2f61da
+  metadata.gz: c40e571fce623118c523c005a8d8404d99390ae90fa73d43a50ee873b103d431
+  data.tar.gz: a4d7325c79141f3bb9625def8b28b1ef808f6dc7e6ab59abefb2178a73b59277
 SHA512:
-  metadata.gz: 92713fe105ad0b96ca28fc61f530c4dcfb8a6905aeb9d7206dfb951afab7f482e10b1ffc397f5d2e8cf86f9bd0a69be65e8da791ba0dd32655c29341bf0a67af
-  data.tar.gz: 4601e51dc96489292cfd0b54653f149bfcde9a830b3c3071ed2bc55778ab393405c267172ea06aa1e2ad3f64c615b845ba0aaf9a5317288555ee50e913bc196f
+  metadata.gz: 39198fd34d327c75172433a3c811ced60b6cf4be19bb7071a922ea66a9336f3023f0d1edaa70e0f9cf695c95e9b9eb1928690449213f6740b0ad4f5779be3d61
+  data.tar.gz: 97684d198ce69b722e03c6f2bde257eac62f772d23e5fd2695eafe47a053c72731f0335c95e779ded290cc50d70c4fc21d3841ca27691c80856c02655630c6a4

data/.gitignore

@@ -10,3 +10,4 @@
 
 # rspec failure tracking
 .rspec_status
+*.gem

data/.travis.yml

@@ -14,3 +14,4 @@ before_install:
 rvm:
   - 2.4.5
   - 2.5.3
+  - 2.6.0

data/README.md

@@ -57,7 +57,7 @@ It's very simple to use rtesseract.
 
 ```ruby
   image = RTesseract.new("my_image.jpg")
-  image.to_tsv  # Getting open file of pdf
+  image.to_tsv  # Getting open file of tsv
 ```
 
 This will preserve the image colors, pictures and structure in the generated pdf.
@@ -100,23 +100,24 @@ This will preserve the image colors, pictures and structure in the generated pdf
     RTesseract.new('test_words.png').to_box
   ```
 
-  # => [
-  #   {:word => 'If',  :x_start=>52, :y_start=>13, :x_end=>63, :y_end=>27},
-  #   {:word => 'you', :x_start=>69, :y_start=>17, :x_end=>100, :y_end=>31},
-  #   {:word => 'are', :x_start=>108, :y_start=>17, :x_end=>136, :y_end=>27},
-  #   {:word => 'a', :x_start=>143, :y_start=>17, :x_end=>151, :y_end=>27},
-  #   {:word => 'friend,', :x_start=>158, :y_start=>13, :x_end=>214, :y_end=>29},
-  #   {:word => 'you', :x_start=>51, :y_start=>39, :x_end=>82, :y_end=>53},
-  #   {:word => 'speak', :x_start=>90, :y_start=>35, :x_end=>140, :y_end=>53},
-  #   {:word => 'the', :x_start=>146, :y_start=>35, :x_end=>174, :y_end=>49},
-  #   {:word => 'password,', :x_start=>182, :y_start=>35, :x_end=>267, :y_end=>53},
-  #   {:word => 'and', :x_start=>51, :y_start=>57, :x_end=>81, :y_end=>71},
-  #   {:word => 'the', :x_start=>89, :y_start=>57, :x_end=>117, :y_end=>71},
-  #   {:word => 'doors', :x_start=>124, :y_start=>57, :x_end=>172, :y_end=>71},
-  #   {:word => 'will', :x_start=>180, :y_start=>57, :x_end=>208, :y_end=>71},
-  #   {:word => 'open.', :x_start=>216, :y_start=>61, :x_end=>263, :y_end=>75}
-  # ]
-
+  ```ruby
+    => [
+      {:word => 'If',  :x_start=>52, :y_start=>13, :x_end=>63, :y_end=>27},
+      {:word => 'you', :x_start=>69, :y_start=>17, :x_end=>100, :y_end=>31},
+      {:word => 'are', :x_start=>108, :y_start=>17, :x_end=>136, :y_end=>27},
+      {:word => 'a', :x_start=>143, :y_start=>17, :x_end=>151, :y_end=>27},
+      {:word => 'friend,', :x_start=>158, :y_start=>13, :x_end=>214, :y_end=>29},
+      {:word => 'you', :x_start=>51, :y_start=>39, :x_end=>82, :y_end=>53},
+      {:word => 'speak', :x_start=>90, :y_start=>35, :x_end=>140, :y_end=>53},
+      {:word => 'the', :x_start=>146, :y_start=>35, :x_end=>174, :y_end=>49},
+      {:word => 'password,', :x_start=>182, :y_start=>35, :x_end=>267, :y_end=>53},
+      {:word => 'and', :x_start=>51, :y_start=>57, :x_end=>81, :y_end=>71},
+      {:word => 'the', :x_start=>89, :y_start=>57, :x_end=>117, :y_end=>71},
+      {:word => 'doors', :x_start=>124, :y_start=>57, :x_end=>172, :y_end=>71},
+      {:word => 'will', :x_start=>180, :y_start=>57, :x_end=>208, :y_end=>71},
+      {:word => 'open.', :x_start=>216, :y_start=>61, :x_end=>263, :y_end=>75}
+    ]
+  ```
 
 ## Development
 

data/lib/rtesseract.rb

@@ -1,6 +1,7 @@
 require "rtesseract/check"
 require "rtesseract/configuration"
 require "rtesseract/command"
+require "rtesseract/base"
 require "rtesseract/text"
 require "rtesseract/pdf"
 require "rtesseract/box"
@@ -9,8 +10,6 @@ require "rtesseract/tsv"
 class RTesseract
   class Error < StandardError; end
 
-  check_version!
-
   attr_reader :config, :source
 
   def initialize(src = '', options = {})

data/lib/rtesseract/base.rb

@@ -0,0 +1,12 @@
+require 'tmpdir'
+require 'securerandom'
+
+class RTesseract
+  module Base
+    def temp_file(ext = '')
+      @rand_file ||= "rtesseract_#{SecureRandom.uuid}"
+
+      Pathname.new(Dir.tmpdir).join("#{@rand_file}#{ext}").to_s
+    end
+  end
+end

data/lib/rtesseract/box.rb

@@ -1,19 +1,15 @@
 require 'nokogiri'
-require 'tmpdir'
 
 class RTesseract
   module Box
-    def self.temp_dir
-      @file_path = Pathname.new(Dir.tmpdir)
-    end
+    extend RTesseract::Base
 
     def self.run(source, options)
-      name = "rtesseract_#{SecureRandom.uuid}"
       options.tessedit_create_hocr = 1
 
-      RTesseract::Command.new(source, temp_dir.join(name).to_s, options).run
+      RTesseract::Command.new(source, temp_file, options).run
 
-      parse(temp_dir.join("#{name}.hocr").read)
+      parse(File.read(temp_file('.hocr')))
     end
 
     def self.parse(content)

data/lib/rtesseract/command.rb

@@ -1,5 +1,3 @@
-require 'tmpdir'
-
 class RTesseract
   class Command
     FIXED = [:command, :psm, :oem, :lang, :tessdata_dir, :user_words, :user_patterns, :config_file]
@@ -10,32 +8,44 @@ class RTesseract
       @source = source
       @output = output
       @options = options
-    end
-
-    def configs
-      @options.to_h.map { |key, value| ['-c', "#{key}=#{value}"] unless FIXED.include?(key) }.compact
+      @full_command = [ options.command, @source, @output]
     end
 
     def full_command
-      command = [options.command, @source, @output]
+      add_option('--psm', options.psm)
+      add_option('--oem', options.oem)
+      add_option('-l', options.lang)
+      add_option('--tessdata_dir', options.tessdata_dir)
+      add_option('--user_words', options.user_words)
+      add_option('--user_patterns', options.user_patterns)
 
-      command << ['--psm', options.psm.to_s] if options.psm
-      command << ['--oem', options.oem.to_s] if options.oem
-      command << ['-l', options.lang] if options.lang
+      other_configs
 
-      command << ['--tessdata_dir', options.tessdata_dir] if options.tessdata_dir
-      command << ['--user_words', options.user_words] if options.user_words
-      command << ['--user_patterns', options.user_patterns] if options.user_patterns
+      add_option(options.config_file)
 
-      command << configs
+      @full_command
+    end
 
-      command << options.config_file.to_s if options.config_file
+    def add_option(*args)
+      return unless args.last
 
-      command.flatten
+      @full_command << args.map(&:to_s)
+    end
+
+    def other_configs
+      @options.to_h.map do |key, value|
+        next if FIXED.include?(key)
+
+        add_option('-c', "#{key}=#{value}")
+      end
     end
 
     def run
-      Open3.capture2e(*full_command)
+      output, status = Open3.capture2e(*full_command.flatten)
+
+      return output if status.success?
+
+      raise RTesseract::Error.new(output)
     end
   end
 end

data/lib/rtesseract/pdf.rb

@@ -1,18 +1,13 @@
-require 'tmpdir'
-
 class RTesseract
   module Pdf
-    def self.temp_dir
-      @file_path = Pathname.new(Dir.tmpdir)
-    end
+    extend Base
 
     def self.run(source, options)
-      name = "rtesseract_#{SecureRandom.uuid}"
       options.tessedit_create_pdf = 1
 
-      RTesseract::Command.new(source, temp_dir.join(name).to_s, options).run
+      RTesseract::Command.new(source, temp_file, options).run
 
-      File.open(temp_dir.join("#{name}.pdf").to_s, 'r')
+      File.open(temp_file('.pdf'), 'r')
     end
   end
 end

data/lib/rtesseract/text.rb

@@ -3,7 +3,7 @@ require 'open3'
 class RTesseract
   module Text
     def self.run(source, options)
-      RTesseract::Command.new(source, 'stdout', options).run.first
+      RTesseract::Command.new(source, 'stdout', options).run
     end
   end
 end

data/lib/rtesseract/tsv.rb

@@ -1,18 +1,13 @@
-require 'tmpdir'
-
 class RTesseract
   module Tsv
-    def self.temp_dir
-      @file_path = Pathname.new(Dir.tmpdir)
-    end
+    extend Base
 
     def self.run(source, options)
-      name = "rtesseract_#{SecureRandom.uuid}"
       options.tessedit_create_tsv = 1
 
-      RTesseract::Command.new(source, temp_dir.join(name).to_s, options).run
+      RTesseract::Command.new(source, temp_file, options).run
 
-      File.open(temp_dir.join("#{name}.tsv").to_s, 'r')
+      File.open(temp_file('.tsv'), 'r')
     end
   end
 end

data/lib/rtesseract/version.rb

@@ -1,3 +1,3 @@
 class RTesseract
-  VERSION = '3.0.0'.freeze
+  VERSION = '3.0.1'.freeze
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rtesseract
 version: !ruby/object:Gem::Version
-  version: 3.0.0
+  version: 3.0.1
 platform: ruby
 authors:
 - Danilo Jeremias da Silva
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2019-01-01 00:00:00.000000000 Z
+date: 2019-01-04 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -115,6 +115,7 @@ files:
 - bin/console
 - bin/setup
 - lib/rtesseract.rb
+- lib/rtesseract/base.rb
 - lib/rtesseract/box.rb
 - lib/rtesseract/check.rb
 - lib/rtesseract/command.rb