pdfmd 1.5.0 → 1.6.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 71b69c4e8462e3e8e9b8fe27df5fc9b6a472292b
-  data.tar.gz: 55cb0726b99937d3cbaad55bb803d8c652315f4c
+  metadata.gz: 910525eeb9a1498f7b6345e987367eca6bc1a16e
+  data.tar.gz: 4fe25f14ef7c6366e2cd2188c68b5a5a7e56bba5
 SHA512:
-  metadata.gz: 01a4b983e7eaaade73851e51d8193ad5e3c6911db00f87852b816c0848b2d0042c3acc651419e1d77226129582691d544fef85cce4395eb665fa51274eee7d64
-  data.tar.gz: 87b8640a0b5c3a1286a5fcd214afdc01c2c1f72a07a996da93a50603caa13a20699e6d907dcd18c03344266d686e7dc934114b93289e2e200da006b8506d5e4a
+  metadata.gz: f2016aae25674bb4be3366ffda1954a1b799b0662f3f050eaf74c68e00a9f0195593b9113850dbfe1c8155fa9a7901a86d2899abea0e66f80284dca3c53e4a20
+  data.tar.gz: c5aa754e21ac36ad4166111099b721e972b3a4524e3fa33917c278d6bb268cbe40707bcc17f12070a7cb9c14fe4fe780610e742a5504410ba8de6152ffb1e296

data/CHANGELOG.md

@@ -1,3 +1,13 @@
+# Version 1.6.0
+- Added command 'config'
+- Added dependency 'yaml'
+- Added logging to command 'rename'
+- Bugfix for comand 'rename' with settings in Hiera.
+- Added Batchmode for command 'edit'
+- Bugfix: value '00' not longer accepted for a month in 'createdate'
+- Added logging to command 'edit'
+- Bugfix: Removing the ampersand (&) from authornames as well now.
+
 # Version 1.5.0
 - Added option 'dryrun' to command 'sort'.
 - Added option 'logfilepath' to command 'sort'

data/lib/pdfmd/config.rb

@@ -0,0 +1,40 @@
+#
+# == File: config.rb
+#
+# Show the current default configuration settings
+#
+
+# Options
+opt_show = ENV.fetch('PDFMD_SHOW')
+
+require_relative '../string_extend.rb'
+require 'yaml'
+require 'pp'
+
+# TODO: the output can be probably made more pretty without
+# adding another requirement, can't it?
+
+#
+# If now options are set,
+# show the current settings
+if opt_show.blank? or opt_show == 'true'
+  opt_show = true
+end
+
+case opt_show
+  # Show the current settings
+when true
+
+  # As long as only Hiera is supported as external storage
+  # for configuration (unless I need otherwise), read the
+  # hiera configuration
+  puts 'Current default configuration:'
+  puts ''
+  hieraConfig = eval `hiera pdfmd::config`
+  hieraConfig.sort.each do |key,value|
+    puts 'Command : ' + key
+    puts value.to_yaml
+    puts "---\n\n"
+  end
+
+end

data/lib/pdfmd/edit.rb

@@ -1,12 +1,67 @@
 #
 # Thor command 'edit' for changing the common
 # ExifTags within the PDF file
+# TODO: Put rename into Hiera
+# TODO: Back backup file/path into hiera and options
 #
-filename  = ENV.fetch('PDFMD_FILENAME')
-optTag    = ENV['PDFMD_TAG'] || nil
-optRename = ENV['PDFMD_RENAME'] == 'true' ? true : false
-pdfmd     = ENV['PDFMD']
 
+require_relative '../string_extend'
+
+filename    = ENV.fetch('PDFMD_FILENAME')
+optTag      = ENV['PDFMD_TAG'] || nil
+optRename   = ENV['PDFMD_RENAME'] == 'true' ? true : false
+pdfmd       = ENV['PDFMD']
+opt_log     = ENV['PDFMD_LOG']
+opt_logfile = ENV['PDFMD_LOGFILE']
+hieraDefaults = queryHiera('pdfmd::config')
+
+# Define logging state
+if ( hieraDefaults['edit'].nil? or
+     hieraDefaults['edit']['log'].nil? or
+     not hieraDefaults['edit']['log'] == true ) and
+  (opt_log == 'false' or opt_log.blank?)
+
+  logenable = false
+  
+else
+
+  logenable = true
+
+end
+
+# If logging is enabled, set parameters and create the obkject
+if logenable
+
+  if opt_logfile.nil?  and
+    ( hieraDefaults['edit'].nil? or
+    hieraDefaults['edit']['logfile'].nil? or
+    hieraDefaults['edit']['logfile'].blank? ) 
+
+    logfile = Dir.pwd.chomp('/') + '/' + File.basename(ENV['PDFMD'], '.*') + '.log'
+
+  elsif not opt_logfile.nil? and not opt_logfile.blank?
+
+    if File.directory? opt_logfile
+      abort('Logfile path is a directory. Abort.')
+      exit 1
+    end
+
+    logfile = opt_logfile
+
+  elsif (opt_logfile.nil? or opt_logfile.blank?) and
+    not hieraDefaults['edit']['logfile'].blank?
+
+    logfile = hieraDefaults['edit']['logfile']
+
+  else
+
+    logfile = Dir.pwd.chomp('/') + '/' + File.basename(ENV['PDFMD'], '.*') + '.log'
+
+  end
+
+  $logger = Logger.new(logfile)
+
+end
 
 metadata = readMetadata(filename)
 
@@ -15,19 +70,42 @@ if optTag == 'all'
 else
   tags = optTag.split(',')
 end
+
 tags.each do |currentTag|
 
-  # Change the tag to something we can use here
-  puts "Current value: '#{metadata[currentTag.downcase]}'"
-  answer   = readUserInput("Enter new value for #{currentTag} :")
-  if currentTag.downcase == 'createdate'
-    while not answer = identifyDate(answer)
-      puts 'Invalid date format'
-      answer = readUserInput("Enter new value for #{currentTag} :")
+  # If the tags contain an '=', set the value for the tag
+  # automatically. Otherwise enter interactive mode.
+  if currentTag.match(/\=/)
+
+    tag, value = currentTag.split('=')
+
+    # Include Date identifier
+    if tag.downcase == 'createdate'
+      value = identifyDate(value)
     end
-  end
-  puts "Changing value for #{currentTag}: '#{metadata[currentTag]}' => #{answer}"
-  `exiftool -#{currentTag.downcase}='#{answer}' -overwrite_original '#{filename}'`
+
+    logenable ? $logger.info("#{filename}: Setting value for tag '#{tag.downcase}': '#{value}'") : ''
+    `exiftool -#{tag.downcase}='#{value}' -overwrite_original '#{filename}'`
+
+  else
+
+    # Change the tag to something we can use here
+    puts "Current value: '#{metadata[currentTag.downcase]}'"
+    answer     = readUserInput("Enter new value for #{currentTag} :")
+    answerCopy = answer
+    if currentTag.downcase == 'createdate'
+      while not answer = identifyDate(answer)
+        logenable ? $logger.warn("${filename}: Invalid date provided: '#{answerCopy}'.") : ''
+        logenable ? $logger.info("${filename}: Asking for new user provided date.") : ''
+        puts 'Invalid date format.'
+        answer = readUserInput("Enter new value for #{currentTag} :")
+      end
+    end
+    puts "Changing value for #{currentTag}: '#{metadata[currentTag]}' => #{answer}"
+    logenable ? $logger.info("#{filename}: Setting value for tag '#{currentTag.downcase}': '#{answer}'") : ''
+    `exiftool -#{currentTag.downcase}='#{answer}' -overwrite_original '#{filename}'`
+  
+  end # If interactive/batch mode
 end
 
 #
@@ -35,6 +113,7 @@ end
 # This is not pretty, but seems to be the only way to do this in THOR
 #
 if optRename
+  logenable ? $logger.info("#{filename}: Trigger file renaming.") : ''
   `#{pdfmd} rename '#{filename}'`
 end
 

data/lib/pdfmd/methods.rb

@@ -107,7 +107,7 @@ end
 def identifyDate(datestring)
   identifiedDate = ''
   year    = '[1-2][90][0-9][0-9]'
-  month   = '0[0-9]|10|11|12'
+  month   = '0[1-9]|10|11|12'
   day     = '[1-9]|0[1-9]|1[0-9]|2[0-9]|3[0-1]'
   hour    = '[0-1][0-9]|2[0-3]|[1-9]'
   minute  = '[0-5][0-9]'

data/lib/pdfmd/rename.rb

@@ -1,8 +1,7 @@
 #
 # Thor command 'rename'
 #
-# TODO: Add option to create outputdir if not existing
-# TODO: Define option to create outputdir via Hiera
+# TODO: align keyword abbreviations for all languages
 #
 require_relative '../string_extend'
 
@@ -12,8 +11,55 @@ outputdir           = ENV.fetch('PDFMD_OUTPUTDIR') == 'false' ? false : ENV.fetc
 dryrun              = ENV.fetch('PDFMD_DRYRUN') == 'false' ? false : true
 opt_numberKeywords  = ENV.fetch('PDFMD_NUMBERKEYWORDS')
 opt_copy            = ENV.fetch('PDFMD_COPY')
+opt_log             = ENV.fetch('PDFMD_LOG')
+opt_logfile         = ENV.fetch('PDFMD_LOGFILE')
 hieraDefaults       = queryHiera('pdfmd::config')
 
+if (opt_log.blank? and not hieraDefaults['rename'].nil? and not hieraDefaults['rename']['log'].nil? and hieraDefaults['rename']['log'] == true) or
+  opt_log == 'false' or
+  opt_log.blank? 
+
+  logenable = false
+
+else
+
+  logenable = true
+
+end
+
+if logenable
+
+  if opt_logfile.blank? and
+    ( hieraDefaults['rename']['logfilepath'].nil? or
+     hieraDefaults['rename']['logfilepath'].blank? or
+     hieraDefaults['rename'].nil? )
+
+    logfile = Dir.pwd.chomp('/') + '/' + File.basename(ENV['PDFMD'], '.*') + '.log'
+
+  elsif not opt_logfile.blank?
+
+    if File.directory? opt_logfile
+      abort('Logfilepath is a directory. Abort.')
+      exit 1
+    end
+
+    logfile = opt_logfile
+
+  elsif opt_logfile.blank? and
+    not hieraDefaults['rename']['logfilepath'].blank?
+
+    logfile = hieraDefaults['rename']['logfilepath']
+
+  else
+
+    logfile = Dir.pwd.chomp('/') + '/' + File.basename(ENV['PDFMD'], '.*') + '.log'
+
+  end
+
+  $logger = Logger.new(logfile)
+
+end
+
 metadata = readMetadata(filename).each do |key,value|
 
   # Check if the metadata is complete
@@ -83,7 +129,7 @@ end
 
 
 date   = metadata['createdate'].gsub(/\ \d{2}\:\d{2}\:\d{2}.*$/,'').gsub(/\:/,'')
-author = metadata['author'].gsub(/\./,'_').gsub(/\-/,'').gsub(/\s/,'_').gsub(/\,/,'_').gsub(/\_\_/,'_')
+author = metadata['author'].gsub(/\./,'_').gsub(/\&/,'').gsub(/\-/,'').gsub(/\s/,'_').gsub(/\,/,'_').gsub(/\_\_/,'_')
 I18n.enforce_available_locales = false
 author = I18n.transliterate(author) # Normalising
 
@@ -93,9 +139,12 @@ case metadata['title']
 when /(Tilbudt|Angebot|Offer)/i
   doktype = 'til'
   keywords_preface = setKeywordsPreface(metadata,doktype.gsub(/\-/,''))
-when /Orderbekrefelse/i
+when /Orderbekreftelse/i
   doktype = 'odb'
   keywords_preface = setKeywordsPreface(metadata,doktype.gsub(/\-/,''))
+when /Auftragsbestätigung/i
+  doktype = 'abg'
+  keywords_preface = setKeywordsPreface(metadata,doktype.gsub(/\-/,''))
 when /faktura/i
   doktype = 'fak'
   keywords_preface = setKeywordsPreface(metadata,doktype.gsub(/\-/,''))
@@ -108,6 +157,9 @@ when /rechnung/i
 when /order/i
   doktype = 'ord'
   keywords_preface = setKeywordsPreface(metadata,doktype.gsub(/\-/,''))
+when /bestilling/i
+  doktype = 'bes'
+  keywords_preface = setKeywordsPreface(metadata,doktype.gsub(/\-/,''))
 when /(kontrakt|avtale|vertrag|contract)/i
   doktype = 'avt'
   keywords_preface = setKeywordsPreface(metadata,doktype.gsub(/\-/,''))
@@ -211,10 +263,14 @@ end
 
 if not dryrun and filename != newFilename.downcase
 
+  logenable ? $logger.info(filename + ' => ' + outputdir + '/' + newFilename.downcase): ''
+
   # Copy of me the file to the new name
   command = opt_copy ? 'cp' : 'mv'
   `#{command} -v '#{filename}' '#{outputdir}/#{newFilename.downcase}'`
 
 else
-  puts filename + "\n   => " + outputdir + '/' + newFilename.downcase
+
+  logenable ? $logger.info('Dryrun: ' + filename + ' => ' + outputdir + '/' + newFilename.downcase): ''
+
 end

data/lib/pdfmd.rb

@@ -1,6 +1,4 @@
 #!/usr/bin/env ruby
-# == Version 1.3
-#
 # == File: pdfmd.rb
 #
 # Show and edit Metadata of PDF files and rename the files accordingly.
@@ -40,8 +38,7 @@
 #
 # TODO: Include password protected PDF documents as well
 # TODO: Fix broken PDF files automatically
-# TODO: Enable logging in more functions than only "sort"
-# TODO: command 'hiera' to show the current settings
+# TODO: Enable logging in 'edit'
 # TODO: Read this: http://lostechies.com/derickbailey/2011/04/29/writing-a-thor-application/
 # TODO: ... and this: http://blog.paracode.com/2012/05/17/building-your-tools-with-thor/
 # gs \
@@ -62,7 +59,7 @@ require "i18n"
 require 'pathname'
 require 'logger'
 
-VERSION = '1.5.0'
+VERSION = '1.6.0'
 
 # Include general usage methods
 require_relative('pdfmd/methods.rb')
@@ -79,6 +76,7 @@ class DOC < Thor
   desc 'show', 'Show metadata of a file'
   method_option :all, :type => :boolean, :aliases => '-a', :desc => 'Show all metatags', :default => false, :required => false
   method_option :tag, :type => :string, :aliases => '-t', :desc => 'Show specific tag(s), comma separated', :required => false
+  #method_option :format, :type => :string, :aliases => '-f', :desc => 'Define output format', :required => false
   long_desc <<-LONGDESC
   == General
 
@@ -128,6 +126,21 @@ class DOC < Thor
 
   end
 
+  #
+  # Show current settings
+  #
+  desc 'config', 'Show config defaults'
+  long_desc <<-LONGDESC
+
+  LONGDESC
+  method_option :show, :type => :boolean, :aliases => '-s', :required => false
+  def config
+
+    ENV['PDFMD_SHOW'] = options[:show].to_s
+    require_relative('./pdfmd/config.rb')
+
+  end
+
   #
   # Change a MetaTag Attribute
   #
@@ -141,7 +154,7 @@ class DOC < Thor
   specified or 'all'.
 
   The command will invoke an interactive user input and request the values
-  for the metatag.
+  for the metatag if no value is provided.
 
   Additionally the file can be renamed at the end according to the new meta
     tags. See `$ #{__FILE__} help rename` for details.
@@ -151,6 +164,8 @@ class DOC < Thor
   --tag, -t
   \x5 Names or list of names of Metatag fields to set, separated by commata.
 
+  If a value is provided, the current Value will be replaced by the new value.
+
   --rename, -r
   \x5 Rename file after updating the meta tag information according to the fields.
 
@@ -164,9 +179,11 @@ class DOC < Thor
   # Edit tag 'Author' and set new value interactive.
   \x5>CLI edit -t author example.pdf
 
-  # Edit mulitple Tags and set a new value.
+  # Edit multiple Tags and set a new value interactive.
   \x5>CLI edit -t tag1,tag2,tag3 <filename>
 
+  # Edit multiple Tags and set a new value in batch mode.
+  \x5 CLI edit -t tag1='value1',tag2='value2' <filename>
 
   == Multiple Tags
 
@@ -178,9 +195,14 @@ class DOC < Thor
   \x5>CLI edit -t author,title,subject example.pdf`
 
   # Set tags 'Author', 'Title', 'Subject', 'CreateDate', 'Keywords' in
-  example.pdf interactive.
+  example.pdf interactive:
   \x5>CLI edit -t all example.pdf
 
+  # Set tags 'Author', 'CreateDate' in example.pdf in batch mode (non-interactive:
+
+  CLI edit -t author='Me',createdate='1970:00:00 01:01:01' example.pdf
+  CLI edit -t author='Me',Createdate=19700000 example.pdf
+
   == Tag: CreateDate
 
   In order to enter a value for the 'CreateDate' field, some internal matching is going on in order to make it easier and faster to enter dates and times.
@@ -217,12 +239,17 @@ class DOC < Thor
   LONGDESC
   method_option :tag, :type => :string, :aliases => '-t', :desc => 'Name of the Tag(s) to Edit', :default => false, :required => true
   method_option :rename, :type => :boolean, :aliases => '-r', :desc => 'Rename file after changing meta-tags', :default => false, :required => false
+  method_option :log, :aliases => '-l', :type => :boolean, :desc => 'Enable logging'
+  method_option :logfile, :aliases => '-p', :type => :string, :desc => 'Define path to logfile'
   def edit(filename)
 
     ENV['PDFMD_FILENAME'] = filename
     ENV['PDFMD_TAG']      = options[:tag]
     ENV['PDFMD_RENAME']   = options[:rename].to_s
     ENV['PDFMD']          = __FILE__
+    ENV['PDFMD_LOG']      = options[:log].to_s
+    ENV['PDFMD_LOGFILE']  = options[:logfile]
+
     require_relative('./pdfmd/edit.rb')
 
   end
@@ -322,6 +349,10 @@ class DOC < Thor
   [*logfilepath|p*]
   \x5 Set an alternate path for the logfile. If not path is chosen, the logfile is being created in the current working directory as `pdfmd.log`.
 
+  [*logfilepath|p*]
+  \x5 Set an alternate path for the logfile. If not path is chosen, the logfile
+  is being created in the current working directory as `pdfmd.log`.
+
   [*interactive|i*]
   \x5 Disable/Enable interactive sorting. This will ask for confirmation for each sorting action.
 
@@ -367,6 +398,10 @@ class DOC < Thor
   [*logfilepath*]
   \x5 Specifes the default path for the logfile. If no path is set and logging is enable, the logfile will be created in the current working directory.
 
+  [*logfilepath*]
+  \5x Specifes the default path for the logfile. If no path is set and logging is enable,
+  the logfile will be created in the current working directory.
+
   Default is the current working directory with the filename `pdfmd.log`
 
   [*interactive*]
@@ -447,6 +482,16 @@ class DOC < Thor
 
   The directory must exist at runtime.
 
+  --log, -l
+  \x5 Enable logging.
+
+  Values: true|false
+
+  --logfile, -p
+  \x5 Define logfile path
+
+  Default: current working-dir/pdfmd.log
+
   == Example
 
   # Rename the file according to the metatags
@@ -536,6 +581,8 @@ class DOC < Thor
   method_option :keywords, :type => :numeric, :aliases => '-k', :desc => 'Number of keywords to include (Default: 3)', :required => false
   method_option :outputdir, :aliases => '-o', :type => :string, :desc => 'Speficy output directory', :default => false, :required => :false
   method_option :copy, :aliases => '-c', :type => :boolean, :desc => 'Copy instead of moving the file when renaming'
+  method_option :log, :aliases => '-l', :type => :boolean, :desc => 'Enable logging'
+  method_option :logfile, :aliases => '-p', :type => :string, :desc => 'Define path to logfile'
   def rename(filename)
 
     ENV['PDFMD_FILENAME']       = filename
@@ -544,6 +591,9 @@ class DOC < Thor
     ENV['PDFMD_OUTPUTDIR']      = options[:outputdir].to_s
     ENV['PDFMD_NUMBERKEYWORDS'] = options[:keywords].to_s
     ENV['PDFMD_COPY']           = options[:copy].to_s
+    ENV['PDFMD_LOG']            = options[:log].to_s
+    ENV['PDFMD_LOGFILE']        = options[:logfile].to_s
+    ENV['PDFMD']                = __FILE__
     require_relative('./pdfmd/rename.rb')
 
   end

data/test/test_rename.rb

@@ -1,7 +1,7 @@
 # Standard renaming
 # Test 001
 initTmpDir
-commandparameter = " rename -o #{TMPDIR}"
+commandparameter = " rename -o #{TMPDIR} -c false "
 `#{PDFMD} #{commandparameter} #{TARGETPDF}`.chomp
 files = readFilesInDir(TMPDIR)
 if files.size == 1 and
@@ -45,7 +45,7 @@ $testResults.store('003', {:result => result, :command => commandparameter })
 # Test 004
 # Testing all keywords (-a)
 initTmpDir
-commandparameter = " rename -a -o #{TMPDIR}"
+commandparameter = " rename -a -o #{TMPDIR} -c false"
 `#{PDFMD} #{commandparameter} #{TARGETPDF}`.chomp
 files = readFilesInDir(TMPDIR)
 if files.size == 1 and
@@ -60,7 +60,7 @@ $testResults.store('004', {:result => result, :command => commandparameter })
 # Testing number of keywords
 # this might be buggy
 initTmpDir
-commandparameter = " rename -k 1 -o #{TMPDIR}"
+commandparameter = " rename -k 1 -o #{TMPDIR} -c false"
 `#{PDFMD} #{commandparameter} #{TARGETPDF}`.chomp
 files = readFilesInDir(TMPDIR)
 if files.size == 1 and
@@ -70,3 +70,19 @@ else
   result = 'failed'
 end
 $testResults.store('005', {:result => result, :command => commandparameter })
+
+# Test 006
+# Testing logging enabled
+initTmpDir
+commandparameter = " rename -l -p #{TMPDIR}/pdfmd.log -c false"
+`#{PDFMD} #{commandparameter} #{TARGETPDF}`.chomp
+files = readFilesInDir(TMPDIR)
+if files.size == 1 and
+  File.basename(files.keys[0]) == '19700101-example_author-dok-some_keywords-kdn1111111-test_subject.pdf' and
+  `tail -n 1 #{TMPDIR}/pdfmd.log | wc -l`.to_i == 1
+  result = 'OK'
+else
+  result = 'failed'
+end
+$testResults.store('006', {:result => result, :command => commandparameter })
+

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: pdfmd
 version: !ruby/object:Gem::Version
-  version: 1.5.0
+  version: 1.6.0
 platform: ruby
 authors:
 - Daniel Roos
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-03-31 00:00:00.000000000 Z
+date: 2015-04-10 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: thor
@@ -108,6 +108,7 @@ files:
 - bin/pdfmd
 - lib/pdfmd.rb
 - lib/pdfmd/check.rb
+- lib/pdfmd/config.rb
 - lib/pdfmd/edit.rb
 - lib/pdfmd/explain.author.md
 - lib/pdfmd/explain.createdate.md