paperless 0.1.0

data/lib/paperless/engine.rb

@@ -0,0 +1,220 @@
+require 'date'
+require 'pdf/reader'
+require 'paperless/date_search'
+include DateSearch
+
+module Paperless
+
+  PDF_EXT   = 'pdf'
+  DATE_VAR  = '<date>'
+  MATCH_VAR = '<match>'
+  FILEDATE  = 'filedate'
+  TODAY     = 'today'
+
+	class Engine
+
+    PDFPEN_ENGINE         = 'pdfpen'
+    PDFPENPRO_ENGINE      = 'pdfpenpro'
+    ACROBAT_ENGINE        = 'acrobat'
+    DEVONTHINKPRO_ENGINE  = 'devonthinkpro'
+    DEVONTHINKPRO_SERVICE = 'devonthinkpro'
+    FINDER_SERVICE        = 'finder'
+    EVERNOTE_SERVICE      = 'evernote'
+
+    attr_reader :service
+
+		def initialize(options)
+      @destination         = nil
+      @service             = nil
+      @title               = nil
+      @date                = DateTime.now
+      @tags                = Array.new()
+      
+      @file                = options[:file]
+      @text_ext            = options[:text_ext]
+      @pdf_ext             = [Paperless::PDF_EXT]
+      @default_service     = options[:default_service]
+      @date_format         = options[:date_format]
+      @date_locale         = options[:date_locale]
+      @date_default        = options[:date_default]
+      @default_destination = options[:default_destination]
+      @rules               = Array.new()
+
+			options[:rules].each do |rule|
+				@rules.push Paperless::Rule.new(rule)
+			end
+
+      @ocr_engine = options[:ocr_engine]||false
+		end
+
+		def process_rules
+      markdown_ext = ['md','mmd']
+			text_ext = @text_ext + markdown_ext
+			file_ext = File.extname(@file).gsub(/\./,'')
+
+			if file_ext == Paperless::PDF_EXT
+        
+        self.process_pdf
+
+      elsif text_ext.index file_ext
+
+        self.process_text
+      
+      else
+      	puts "Unknown file type. No rules were processed."
+      end
+		end
+
+		def add_tags(tags)
+			if tags.length > 0
+				@tags = (@tags + tags).collect {|x| x = x.downcase }
+				@tags.uniq!
+			end
+		end
+
+		def set_destination(destination)
+			@destination = destination if destination && @destination.nil?
+		end
+
+		def set_title(title)
+			@title = title if title && @title.nil?
+		end
+
+		def set_service(service)
+			@service = service if service && @service.nil?
+		end
+
+    def set_date_default()
+      puts "Using default date..."
+      # Set the default date to the date of the file or else to now
+      if @date_default == Paperless::FILEDATE
+        t = File.stat(@file).mtime
+        @date = Date.new(t.year,t.month,t.day) 
+      else
+        @date = DateTime.now
+      end
+    end
+
+    def process_rules_engine(text)
+      self.set_date_default if @date.nil?
+      # Process each page and pass it through the rules engine
+      @rules.each do |rule|
+        rule.set_date(@date,@date_format)
+        if !rule.matched && rule.match(@file, text)
+          self.add_tags(rule.tags)
+          self.set_destination(rule.destination)
+          self.set_title(rule.title)
+          self.set_service(rule.service)
+        end
+      end
+    end
+
+    def process_text
+      puts "Processing Text file..."
+
+      text = File.open(@file, "rb") {|io| io.read}
+
+      # Verify that we need to search for date or just set to today
+      # Need to prcess file for date in case the rules need to use it.
+      # 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)
+        end
+      end
+
+      # Process each page and pass it through the rules engine
+      process_rules_engine(text)
+    end
+
+
+		def process_pdf
+			puts "Processing PDF pages..."
+
+		  reader = PDF::Reader.new(@file)
+
+		  # Verify that we need to search for date or just set to today
+		  # Need to prcess file for date in case the rules need to use it.
+		  # First check if there are actually any date rules
+      @rules.each do |rule|
+			  if rule.condition == Paperless::DATE_VAR
+			    reader.pages.each do |page|
+			    	break if @date = date_search(page.text,@date_locale)
+			    end
+			    break
+	    	end
+			end
+
+			# Process each page and pass it through the rules engine
+	    reader.pages.each do |page|
+        process_rules_engine(page.text)
+	    end
+		end
+
+		def ocr
+			puts "Running OCR on file with #{@ocr_engine}"
+      ocr_engine = case @ocr_engine
+        when /^#{PDFPENPRO_ENGINE}$/i     then PaperlessOCR::PDFpenPro.new
+        when /^#{PDFPEN_ENGINE}$/i        then PaperlessOCR::PDFpen.new
+        when /^#{ACROBAT_ENGINE}$/i       then PaperlessOCR::Acrobat.new
+        when /^#{DEVONTHINKPRO_ENGINE}$/i then PaperlessOCR::DevonThinkPro.new
+        else false
+      end
+      
+      if ocr_engine
+        ocr_engine.ocr({:file => @file})
+      else
+        puts "WARNING: No valid OCR engine was defined."
+      end
+		end
+
+		def create(options)
+      # May need to externalize this so other methods can access it.
+      service = case @service.nil? ? @default_service : @service
+        when /^#{EVERNOTE_SERVICE}$/i then PaperlessService::Evernote.new
+        when /^#{FINDER_SERVICE}$/i then PaperlessService::Finder.new
+        when /^#{DEVONTHINKPRO_SERVICE}$/i then PaperlessService::DevonThinkPro.new
+        else false
+      end
+
+      if service
+        self.print
+        
+        destination = @destination.nil? ? @default_destination : @destination
+        # :created => @date
+        service.create({ 
+          :delete => options[:delete], 
+          :destination => destination, 
+          :text_ext => @text_ext, 
+          :file => @file, 
+          :date => @date, 
+          :title => @title, 
+          :tags => @tags
+        })
+      else 
+        puts "WARNING: No valid Service was defined."
+      end
+		end
+
+		def print
+      service = @service.nil? ? @default_service : @service
+      title = @title.nil? ? File.basename(@file) : @title
+
+      destination = @destination.nil? ? @default_destination : @destination
+      if destination == PaperlessService::Finder::NO_MOVE && service == PaperlessService::FINDER.downcase
+        destination = File.dirname(@file)
+      end
+
+      puts "* ---------------------------------------------"
+			puts "* File: #{@file}"
+      puts "* Service: #{service}"
+			puts "* Destination: #{destination}"
+			puts "* Title: #{title}"
+			puts "* Date: #{@date.strftime('%Y-%m-%d')}"
+			puts "* Tags: #{@tags.join(', ')}"
+      puts "* ---------------------------------------------"
+		end
+
+	end
+
+end

data/lib/paperless/ocr_engines/acrobat.rb

@@ -0,0 +1,51 @@
+require 'appscript'
+include Appscript
+
+module PaperlessOCR
+
+  ACROBAT = 'Adobe Acrobat Pro'
+
+  class PDFpen
+    def initialize
+      @engine = PaperlessOCR::ACROBAT
+      @app = app(@engine)
+      @app.activate
+    end
+
+    def ocr(options)
+
+      begin
+
+      rescue 
+        puts "WARNING: There was an error OCRing the document with #{@engine}: #{$!}"
+      end
+    end
+
+  end
+end
+
+__END__
+
+The Following does not work with Acrobat X
+
+tell application "Adobe Acrobat Professional"
+      activate
+      open theFile
+      tell application "System Events"
+         tell process "Acrobat"
+            tell menu bar 1
+               tell menu "Document"
+                  tell menu item "OCR Text Recognition"
+                     tell menu 1
+                        click menu item "Recognize Text Using OCR..."
+                     end tell
+                  end tell
+               end tell
+ 
+            end tell
+            keystroke return
+         end tell
+      end tell
+      save the front document
+      close the front document
+   end tell

data/lib/paperless/ocr_engines/devonthinkpro.rb

@@ -0,0 +1,26 @@
+require 'appscript'
+include Appscript
+
+module PaperlessOCR
+
+  DEVONTHINKPRO = 'DEVONthink Pro'
+
+	class DevonThinkPro
+    
+		def initialize
+      @engine = PaperlessOCR::DEVONTHINKPRO
+      @app    = app(@engine)
+      @app.activate
+		end
+
+    def ocr(options)
+      begin
+        app("System Events").processes[@engine].visible.set(false)
+        @app.ocr(:file => options[:file])
+      rescue 
+        puts "WARNING: There was an error OCRing the document with #{@engine}: #{$!}"
+      end
+    end
+
+  end
+end

data/lib/paperless/ocr_engines/pdfpen.rb

@@ -0,0 +1,32 @@
+require 'appscript'
+include Appscript
+
+module PaperlessOCR
+
+  PDFPEN = 'PDFpen'
+
+	class PDFpen
+		def initialize
+      @engine = PaperlessOCR::PDFPEN
+      @app = app(@engine)
+      @app.activate
+		end
+
+    def ocr(options)
+      begin
+        doc = @app.open MacTypes::Alias.path(options[:file])
+        doc.ocr
+
+        app("System Events").processes[@engine].visible.set(false)
+
+        while doc.performing_ocr.get
+          sleep 1
+        end
+        doc.close(:saving => :yes)
+      rescue 
+        puts "WARNING: There was an error OCRing the document with #{@engine}: #{$!}"
+      end
+    end
+
+  end
+end

data/lib/paperless/ocr_engines/pdfpenpro.rb

@@ -0,0 +1,32 @@
+require 'appscript'
+include Appscript
+
+module PaperlessOCR
+
+  PDFPENPRO = 'PDFpenPro'
+
+	class PDFpenPro
+		def initialize
+      @engine = PaperlessOCR::PDFPENPRO
+      @app = app(@engine)
+      @app.activate
+		end
+
+    def ocr(options)
+      begin
+        doc = @app.open MacTypes::Alias.path(options[:file])
+        doc.ocr
+
+        app("System Events").processes[@engine].visible.set(false)
+
+        while doc.performing_ocr.get
+          sleep 1
+        end
+        doc.close(:saving => :yes)
+      rescue 
+        puts "WARNING: There was an error OCRing the document with #{@engine}: #{$!}"
+      end
+    end
+
+  end
+end

data/lib/paperless/rule.rb

@@ -0,0 +1,82 @@
+module Paperless
+
+	class Rule
+
+		def initialize(options)
+			# @todo: Add logic to validate rule has proper settings
+
+			@scope               = options['scope'].split
+			@condition           = options['condition']
+			@destination         = options['destination']
+			@service             = options['service']
+			@title               = options['title']
+			@date                = options['date']
+			@description         = options['description']
+			@tags                = options['tags'].nil? ? Array.new : options['tags'].split
+			@date_stamp          = DateTime.now
+			@date_default_format = '%Y-%m-%d'
+			@matched             = false
+		end
+
+		attr_reader :matched, :tags, :destination, :condition, :title, :date, :service
+
+		def set_date(date,format)
+			@date_stamp = date
+			@date_default_format = format
+		end
+
+		def match(file,text)
+			return @matched if @matched
+
+			if @condition == Paperless::DATE_VAR
+				@date    = date
+				@matched = true
+			else
+				@scope.each do |type|
+					case type
+						when 'content' then
+							self.match_text(text)
+						when 'filename' then
+							self.match_text(file)
+						else
+							puts "WARNING: Unknown scope defined - #{type}"
+					end
+				end
+			end
+			return @matched
+		end
+
+		def match_text(text)
+			match = text.match(/#{@condition}/i) {|md| md[0]}
+			unless match.nil?
+				puts "Rule Matched: #{@description}"
+				self.sub_vars(match.gsub(/\s*$/,'')) # chomp! was not working... had to do this
+				@matched = true
+			end
+		end
+
+		def sub_var(attribute, value)
+			unless attribute.nil?
+				attribute.gsub!(/#{Paperless::MATCH_VAR}/, value) 
+				attribute.gsub!(/#{Paperless::DATE_VAR}/, @date_stamp.strftime(@date_default_format))
+
+				# Custom date formats set at the rule level
+				custom_date_regex = "#{Paperless::DATE_VAR.chop}=([a-zA-Z\%\-]+)>"
+				if match = attribute.match(/#{custom_date_regex}/)
+					attribute.gsub!(/#{custom_date_regex}/, @date_stamp.strftime(match[1])) 
+				end
+			end
+			attribute
+		end
+
+		def sub_vars(value)
+			# Need to do substituation for supported variables
+			@destination = self.sub_var(@destination,value)
+			@title = self.sub_var(@title,value)
+			@date = self.sub_var(@date,value)
+			@tags.collect! {|tag| tag = self.sub_var(tag,value) }			
+		end
+
+	end
+
+end

data/lib/paperless/services/devonthinkpro.rb

@@ -0,0 +1,54 @@
+require 'fileutils'
+include FileUtils
+require 'appscript'
+include Appscript
+
+module PaperlessService
+
+  DEVONTHINKPRO = 'DEVONthink Pro'
+
+  class DevonThinkPro
+
+    def initialize
+      @app = app(PaperlessService::DEVONTHINKPRO)
+      @app.activate
+    end
+
+    def create(options)
+      destination = options[:destination]
+      date        = options[:date]
+      from_file   = options[:file]
+      title       = options[:title] || File.basename(from_file)
+      tags        = options[:tags].collect!{|x| x="'#{x}'"} # Add quotes around each tag in case there is a space
+
+      time = Time.new(date.year, date.month, date.day)
+      FileUtils.touch from_file, :mtime => time
+
+      if tags.length > 0
+        # Add open meta tags to file
+        system("#{OPENMETA} -p '#{from_file}' -a #{tags.join(' ')}")
+      end
+
+      # Extract the database and folder name from the destination
+      matches     = destination.match(/(.+)::(.+)/)
+      dt_database = matches[1]
+      dt_folder   = matches[2]
+
+      unless dt_folder && dt_database
+        raise "Unable to determine database and folder for DevonThinkPro: #{destination}"
+      end
+
+      db_conn = @app.open_database(dt_database)
+
+      app("System Events").processes[PaperlessService::DEVONTHINKPRO].visible.set(false)
+
+      group = @app.create_location(dt_folder, {:in => db_conn})
+      @app.import(from_file, {:to => group, :name => title} )
+
+      if options[:delete]
+        FileUtils.rm from_file, :force => true
+      end
+    end
+
+  end
+end

data/lib/paperless/services/evernote.rb

@@ -0,0 +1,55 @@
+require 'fileutils'
+include FileUtils
+require 'markdown'
+require 'appscript'
+include Appscript
+
+module PaperlessService
+
+  EVERNOTE = 'Evernote'
+
+	class Evernote
+
+		def initialize
+      @app = app(PaperlessService::EVERNOTE)
+      @app.activate
+		end
+
+    def create(options)
+      destination = options[:destination]
+      date        = options[:date]
+      from_file   = options[:file]
+      title       = options[:title] || File.basename(from_file)
+      tags        = options[:tags]
+      text_ext    = options[:text_ext]
+
+      create_options = { :created => date }
+      file_ext = File.extname(from_file.gsub(/\./,''))
+
+      if text_ext.index file_ext
+        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
+          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
+          create_options[:from_file] = MacTypes::FileURL.path(from_file)
+        end
+      end
+
+      create_options[:tags] = tags if tags.length > 0
+      create_options[:notebook] = destination if destination
+
+      @app.create_note(create_options)
+
+      if options[:delete]
+        FileUtils.rm from_file, :force => true
+      end
+
+      @app.synchronize
+    end
+
+	end
+end

data/lib/paperless/services/finder.rb

@@ -0,0 +1,47 @@
+require 'fileutils'
+include FileUtils
+
+module PaperlessService
+
+  FINDER = 'Finder'
+
+	class Finder
+
+    NO_MOVE = '<nomove>'
+
+		def initialize
+      @app = app(PaperlessService::FINDER)
+      @app.activate
+		end
+
+    def create(options)
+      destination = options[:destination]
+      date        = options[:date]
+      from_file   = options[:file]
+      title       = options[:title]
+      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))
+      else
+        FileUtils.mkdir_p destination unless File.exists?(destination)
+        new_filename = File.join(destination, title + File.extname(from_file))
+      end
+
+      FileUtils.cp from_file, new_filename, :force => true
+
+      time = Time.new(date.year, date.month, date.day)
+      FileUtils.touch new_filename, {:mtime => time}
+
+      if tags.length > 0
+        # Add open meta tags to file
+        system("#{OPENMETA} -p '#{new_filename}' -a #{tags.join(' ')}")
+      end
+
+      if options[:delete] && from_file != new_filename
+        FileUtils.rm from_file, :force => true
+      end
+    end
+
+	end
+end

data/lib/paperless/version.rb

@@ -0,0 +1,3 @@
+module Paperless
+  VERSION = '0.1.0'
+end

data/lib/paperless.rb

@@ -0,0 +1,14 @@
+require 'paperless/version.rb'
+require 'paperless/engine.rb'
+require 'paperless/rule.rb'
+require 'paperless/date_search.rb'
+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/pdfpen.rb'
+require 'paperless/ocr_engines/pdfpenpro.rb'
+require 'paperless/ocr_engines/devonthinkpro.rb'
+
+# Add requires for other files you add to your project here, so
+# you just need to require this one file in your bin file

data/paperless.rdoc

@@ -0,0 +1,5 @@
+= paperless
+
+Generate this with
+    paperless rdoc
+After you have described your command line interface