paperless 0.1.0 → 0.2.0

data/bin/paperless

@@ -131,6 +131,9 @@ desc 'Create a new note from a file'
 arg_name 'file_name'
 command :create do |c|
 
+  c.desc 'Dump the OCR text for the document to the terminal.'
+  c.switch :dump, :negatable => false, :default_value => false
+
   c.desc 'Open a prompt to rename the file before its processed through the rules.'
   c.switch :prompt, :negatable => false, :default_value => false
 
@@ -147,14 +150,15 @@ command :create do |c|
 
     args.each do |file|
 
-      if File.exists?(File.expand_path file) 
+      if File.exists?(File.expand_path file) && !File.directory?(File.expand_path file)
         file = File.expand_path file
       else
-        raise "File does not exist (#{file})"
+        raise "File does not exist or is a directory (#{file})"
       end
 
-      if options[:prompt] && !global_options[:simulate]
-        # Cannot rename the file in simulate mode since we dont want to change the file name
+      old_filename = nil
+
+      if options[:prompt]
         file_ext = File.extname(file)
         filename = File.basename(file, file_ext)
         user_input = `#{COCOADIALOG} standard-inputbox --title "Paperless Prompt" --informative-text "Rename your file before its processed with rules..." --text "#{filename}" --no-newline --string-output`.split("\n")
@@ -162,6 +166,10 @@ command :create do |c|
         if user_input[0].match(/ok/i) && user_input[1] != File.basename(file, file_ext)
           new_filename = File.join(File.dirname(file), user_input[1] + file_ext)
           puts "Renaming file based on input to #{new_filename}"
+          if global_options[:simulate]
+            # save the file name to rename it back later
+            old_filename = file
+          end
           File.rename(file, new_filename) 
           file = new_filename
         end
@@ -182,7 +190,7 @@ command :create do |c|
       file_ext = File.extname(file).gsub(/\./,'')
       if file_ext == Paperless::PDF_EXT && options[:ocr]
         puts "OCRing file..."
-        engine.ocr
+        engine.ocr(options[:dump])
       end
 
       if options[:proc_rules]
@@ -193,6 +201,11 @@ command :create do |c|
       if global_options[:simulate]
         puts "Simulating changes..."
         engine.print
+
+        if global_options[:simulate] && options[:prompt] && !old_filename.nil?
+          puts "Renaming file back to #{old_filename}"
+          File.rename(file, old_filename) 
+        end
       else
         puts "Saving #{file} to #{engine.service}"
         engine.create({:delete => options[:delete]})

data/lib/paperless/date_search.rb

@@ -2,6 +2,7 @@ require 'date'
 
 module DateSearch
 
+	SEP_NOSPACE = '\.\/\-\,'
 	SEP = '\. \/\-\,'
 	DAY = '(\d{1,2})'
 	MONTH = '([a-zA-Z]{3,15})'
@@ -49,7 +50,25 @@ module DateSearch
 
 	def date_search(text,date_locale)
 		date = nil
-		if match = text.match(/#{MONTH}[#{SEP}]{0,3}#{DAY}[#{SEP}]{1,3}#{YEAR}#{END_DATE}/i)
+		if match = text.match(/#{DAY}[#{SEP_NOSPACE}]+#{DAY}[#{SEP_NOSPACE}]+#{YEAR}/)
+			# US:   12-29-2011
+			# Euro: 29-12-2011
+			year  = valid_year(match[3])
+			day   = date_locale == 'us' ? valid_day(match[2])   : valid_day(match[1])
+			month = date_locale == 'us' ? valid_month(match[1]) : valid_month(match[2])
+			
+			if month && day && year
+				puts "Basing the date off the discovered string (3): #{match[0]}"
+				begin
+					date = DateTime.new(year,month,day)
+				rescue
+					puts "WARNING: Unable to create date object. #{$!}"
+					date = nil
+				end
+			else
+				puts "WARNING: The discovered date string does not validate: #{match[0]}"						
+			end
+		elsif match = text.match(/#{MONTH}[#{SEP}]{0,3}#{DAY}[#{SEP}]{1,3}#{YEAR}/i)
 			# December 29, 2011
 			if valid_day(match[2]) && valid_year(match[3])
 				puts "Basing the date off the discovered string (1): #{match[0]}"
@@ -60,7 +79,7 @@ module DateSearch
 					date = nil
 				end
 			end
-		elsif match = text.match(/#{DAY}[#{SEP}]{0,3}#{MONTH}[#{SEP}]{0,3}#{YEAR}#{END_DATE}/i)
+		elsif match = text.match(/#{DAY}[#{SEP}]{0,3}#{MONTH}[#{SEP}]{0,3}#{YEAR}/i)
 			# 29 December 2011
 			if valid_day(match[1]) && valid_year(match[3])
 				puts "Basing the date off the discovered string (2): #{match[0]}"
@@ -71,23 +90,16 @@ module DateSearch
 					date = nil
 				end
 			end
-		elsif match = text.match(/#{DAY}[#{SEP}]+#{DAY}[#{SEP}]+#{YEAR}#{END_DATE}/)
-			# US:   12-29-2011
-			# Euro: 29-12-2011
-			year  = valid_year(match[3])
-			day   = date_locale == 'us' ? valid_day(match[2])   : valid_day(match[1])
-			month = date_locale == 'us' ? valid_month(match[1]) : valid_month(match[2])
-			
-			if month && day && year
-				puts "Basing the date off the discovered string (3): #{match[0]}"
+		elsif match = text.match(/#{MONTH}[#{SEP}]{0,3}#{YEAR}/i)
+			# December 2011
+			if valid_year(match[2])
+				puts "Basing the date off the discovered string (2): #{match[0]}"
 				begin
-					date = DateTime.new(year,month,day)
+					date = DateTime.parse(repair_ocr_string(match[0]))
 				rescue
 					puts "WARNING: Unable to create date object. #{$!}"
 					date = nil
 				end
-			else
-				puts "WARNING: The discovered date string does not validate: #{match[0]}"						
 			end
 		end
 		date

data/lib/paperless/engine.rb

@@ -8,6 +8,7 @@ module Paperless
   PDF_EXT   = 'pdf'
   DATE_VAR  = '<date>'
   MATCH_VAR = '<match>'
+  FILENAME_VAR = '<filename>'
   FILEDATE  = 'filedate'
   TODAY     = 'today'
 
@@ -15,6 +16,7 @@ module Paperless
 
     PDFPEN_ENGINE         = 'pdfpen'
     PDFPENPRO_ENGINE      = 'pdfpenpro'
+    PDFPENPRO6_ENGINE     = 'pdfpenpro6'
     ACROBAT_ENGINE        = 'acrobat'
     DEVONTHINKPRO_ENGINE  = 'devonthinkpro'
     DEVONTHINKPRO_SERVICE = 'devonthinkpro'
@@ -25,7 +27,7 @@ module Paperless
 
 		def initialize(options)
       @destination         = nil
-      @service             = nil
+      @service             = options[:default_service]
       @title               = nil
       @date                = DateTime.now
       @tags                = Array.new()
@@ -119,7 +121,7 @@ module Paperless
       # First check if there are actually any date rules
       @rules.each do |rule|
         if rule.condition == Paperless::DATE_VAR
-          @date = date_search(text,@date_locale)
+          @date = date_search(text,@date_locale) || date_search(@file,@date_locale)
         end
       end
 
@@ -141,6 +143,8 @@ module Paperless
 			    reader.pages.each do |page|
 			    	break if @date = date_search(page.text,@date_locale)
 			    end
+          # Check for the date in the file name if not found in the content
+          @date = date_search(@file,@date_locale) if @date.nil?
 			    break
 	    	end
 			end
@@ -151,9 +155,21 @@ module Paperless
 	    end
 		end
 
-		def ocr
+		def ocr(dump = false)
+      reader = PDF::Reader.new(@file)
+      if reader.pages.length > 0
+        text = reader.pages[0].text
+        if !text.nil? && text != ''
+          puts text if dump
+          puts "This doc already seems to be OCR'd. Not processing through #{@ocr_engine}"
+          return
+        end
+      end
+
 			puts "Running OCR on file with #{@ocr_engine}"
       ocr_engine = case @ocr_engine
+        when /^#{PDFPENPRO6_ENGINE}$/i    then PaperlessOCR::PDFpenPro6.new
+        when /^#{PDFPEN6_ENGINE}$/i       then PaperlessOCR::PDFpen6.new
         when /^#{PDFPENPRO_ENGINE}$/i     then PaperlessOCR::PDFpenPro.new
         when /^#{PDFPEN_ENGINE}$/i        then PaperlessOCR::PDFpen.new
         when /^#{ACROBAT_ENGINE}$/i       then PaperlessOCR::Acrobat.new
@@ -163,6 +179,16 @@ module Paperless
       
       if ocr_engine
         ocr_engine.ocr({:file => @file})
+
+        if dump
+          puts "Dumping Page Content..."
+          # Print the contents of the doc
+          reader = PDF::Reader.new(@file)
+          reader.pages.each do |page|
+            puts page.text
+          end
+        end
+
       else
         puts "WARNING: No valid OCR engine was defined."
       end
@@ -180,15 +206,16 @@ module Paperless
       if service
         self.print
         
-        destination = @destination.nil? ? @default_destination : @destination
-        # :created => @date
+        destination = @destination.nil? ? @default_destination                      : @destination
+        title       = @title.nil?       ? File.basename(@file, File.extname(@file)) : @title
+
         service.create({ 
           :delete => options[:delete], 
           :destination => destination, 
           :text_ext => @text_ext, 
           :file => @file, 
           :date => @date, 
-          :title => @title, 
+          :title => title, 
           :tags => @tags
         })
       else 
@@ -198,7 +225,7 @@ module Paperless
 
 		def print
       service = @service.nil? ? @default_service : @service
-      title = @title.nil? ? File.basename(@file) : @title
+      title = @title.nil? ? File.basename(@file, File.extname(@file)) : @title
 
       destination = @destination.nil? ? @default_destination : @destination
       if destination == PaperlessService::Finder::NO_MOVE && service == PaperlessService::FINDER.downcase

data/lib/paperless/ocr_engines/pdfpen.rb

@@ -23,6 +23,7 @@ module PaperlessOCR
           sleep 1
         end
         doc.close(:saving => :yes)
+        sleep 3
       rescue 
         puts "WARNING: There was an error OCRing the document with #{@engine}: #{$!}"
       end

data/lib/paperless/ocr_engines/pdfpen6.rb

@@ -0,0 +1,33 @@
+require 'appscript'
+include Appscript
+
+module PaperlessOCR
+
+  PDFPEN6 = 'PDFpen 6.app'
+
+	class PDFpen6
+		def initialize
+      @engine = PaperlessOCR::PDFPEN6
+      @app = app(@engine)
+      @app.activate
+		end
+
+    def ocr(options)
+      begin
+        doc = @app.open MacTypes::Alias.path(options[:file])
+        doc.ocr
+
+        app("System Events").processes['PDFpen 6'].visible.set(false)
+
+        while doc.performing_ocr.get
+          sleep 1
+        end
+        doc.close(:saving => :yes)
+        sleep 3
+      rescue 
+        puts "WARNING: There was an error OCRing the document with #{@engine}: #{$!}"
+      end
+    end
+
+  end
+end

data/lib/paperless/ocr_engines/pdfpenpro.rb

@@ -23,6 +23,7 @@ module PaperlessOCR
           sleep 1
         end
         doc.close(:saving => :yes)
+        sleep 3
       rescue 
         puts "WARNING: There was an error OCRing the document with #{@engine}: #{$!}"
       end

data/lib/paperless/ocr_engines/pdfpenpro6.rb

@@ -0,0 +1,33 @@
+require 'appscript'
+include Appscript
+
+module PaperlessOCR
+
+  PDFPENPRO6 = 'PDFpenPro 6.app'
+
+	class PDFpenPro6
+		def initialize
+      @engine = PaperlessOCR::PDFPENPRO6
+      @app = app(@engine)
+      @app.activate
+		end
+
+    def ocr(options)
+      begin
+        doc = @app.open MacTypes::Alias.path(options[:file])
+        doc.ocr
+
+        app("System Events").processes['PDFpenPro 6'].visible.set(false)
+
+        while doc.performing_ocr.get
+          sleep 1
+        end
+        doc.close(:saving => :yes)
+        sleep 3
+      rescue 
+        puts "WARNING: There was an error OCRing the document with #{@engine}: #{$!}"
+      end
+    end
+
+  end
+end

data/lib/paperless/rule.rb

@@ -14,6 +14,7 @@ module Paperless
 			@description         = options['description']
 			@tags                = options['tags'].nil? ? Array.new : options['tags'].split
 			@date_stamp          = DateTime.now
+			@filename 			 = ''
 			@date_default_format = '%Y-%m-%d'
 			@matched             = false
 		end
@@ -28,6 +29,9 @@ module Paperless
 		def match(file,text)
 			return @matched if @matched
 
+			file_ext = File.extname(file)
+	        @filename = File.basename(file, file_ext)
+
 			if @condition == Paperless::DATE_VAR
 				@date    = date
 				@matched = true
@@ -57,6 +61,7 @@ module Paperless
 
 		def sub_var(attribute, value)
 			unless attribute.nil?
+				attribute.gsub!(/#{Paperless::FILENAME_VAR}/, @filename) 
 				attribute.gsub!(/#{Paperless::MATCH_VAR}/, value) 
 				attribute.gsub!(/#{Paperless::DATE_VAR}/, @date_stamp.strftime(@date_default_format))
 

data/lib/paperless/services/evernote.rb

@@ -24,17 +24,28 @@ module PaperlessService
       text_ext    = options[:text_ext]
 
       create_options = { :created => date }
-      file_ext = File.extname(from_file.gsub(/\./,''))
+      file_ext = File.extname(from_file)
+      file_dir = File.dirname(from_file)
+      file_name = File.basename(from_file)
 
-      if text_ext.index file_ext
+      if file_name != title
+        new_filename = File.join(file_dir, title + file_ext)
+        File.rename(from_file, new_filename)
+        from_file = new_filename
+      end
+
+      if text_ext.index file_ext.gsub!(/\./,'')
+        puts "Adding text note into Evernote"
         create_options[:with_text] = File.open(from_file, "rb") {|io| io.read}
       else
         if file_ext.match(/md$/i)
           # If this is a mardown file insert it into Evernote as html
+          puts "Converting Markdown to HTML"
           text = File.open(from_file, "rb") {|io| io.read}
           create_options[:with_html] = Markdown.new(text).to_html
         else
           # Create a note from a file and let Evernote choose how to attach the file
+          puts "Adding note into Evernote"
           create_options[:from_file] = MacTypes::FileURL.path(from_file)
         end
       end

data/lib/paperless/services/finder.rb

@@ -18,28 +18,34 @@ module PaperlessService
       destination = options[:destination]
       date        = options[:date]
       from_file   = options[:file]
-      title       = options[:title]
+      title       = options[:title] || File.basename(from_file, File.extname(from_file))
       tags        = options[:tags].collect!{|x| x="'#{x}'"} # Add quotes around each tag in case there is a space
 
       if destination == NO_MOVE || destination == File.dirname(from_file)
         new_filename = File.join(File.dirname(from_file), title + File.extname(from_file))
+        puts "New filename (1): #{new_filename}"
       else
         FileUtils.mkdir_p destination unless File.exists?(destination)
         new_filename = File.join(destination, title + File.extname(from_file))
+        puts "New filename (2): #{new_filename}"
       end
 
-      FileUtils.cp from_file, new_filename, :force => true
+      puts "Copying File..."
+      FileUtils.cp from_file, new_filename, :verbose => true
 
       time = Time.new(date.year, date.month, date.day)
+      puts "Modifying the time of the file to be #{time.to_s}"
       FileUtils.touch new_filename, {:mtime => time}
 
       if tags.length > 0
         # Add open meta tags to file
+        puts "Tagging file"
         system("#{OPENMETA} -p '#{new_filename}' -a #{tags.join(' ')}")
       end
 
       if options[:delete] && from_file != new_filename
-        FileUtils.rm from_file, :force => true
+        puts "Removing original file"
+        FileUtils.rm from_file, :force => true, :verbose => true
       end
     end
 

data/lib/paperless/version.rb

@@ -1,3 +1,3 @@
 module Paperless
-  VERSION = '0.1.0'
+  VERSION = '0.2.0'
 end

data/lib/paperless.rb

@@ -6,6 +6,8 @@ require 'paperless/services/evernote.rb'
 require 'paperless/services/devonthinkpro.rb'
 require 'paperless/services/finder.rb'
 require 'paperless/ocr_engines/acrobat.rb'
+require 'paperless/ocr_engines/pdfpen6.rb'
+require 'paperless/ocr_engines/pdfpenpro6.rb'
 require 'paperless/ocr_engines/pdfpen.rb'
 require 'paperless/ocr_engines/pdfpenpro.rb'
 require 'paperless/ocr_engines/devonthinkpro.rb'

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: paperless
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.2.0
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-03-07 00:00:00.000000000 Z
+date: 2013-03-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake
@@ -190,6 +190,8 @@ files:
 - lib/paperless/services/finder.rb
 - lib/paperless/services/devonthinkpro.rb
 - lib/paperless/ocr_engines/acrobat.rb
+- lib/paperless/ocr_engines/pdfpen6.rb
+- lib/paperless/ocr_engines/pdfpenpro6.rb
 - lib/paperless/ocr_engines/pdfpen.rb
 - lib/paperless/ocr_engines/pdfpenpro.rb
 - lib/paperless/ocr_engines/devonthinkpro.rb
@@ -222,7 +224,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
       segments:
       - 0
-      hash: 763194951971081562
+      hash: -3000972400223417895
 requirements: []
 rubyforge_project: 
 rubygems_version: 1.8.24