docsplit 0.3.4 → 0.4.0

data/docsplit.gemspec

@@ -1,7 +1,7 @@
 Gem::Specification.new do |s|
   s.name      = 'docsplit'
-  s.version   = '0.3.4'         # Keep version in sync with docsplit.rb
-  s.date      = '2010-8-20'
+  s.version   = '0.4.0'         # Keep version in sync with docsplit.rb
+  s.date      = '2010-8-23'
 
   s.homepage    = "http://documentcloud.github.com/docsplit/"
   s.summary     = "Break Apart Documents into Images, Text, Pages and PDFs"

data/lib/docsplit.rb

@@ -1,7 +1,7 @@
 # The Docsplit module delegates to the Java PDF extractors.
 module Docsplit
 
-  VERSION       = '0.3.4' # Keep in sync with gemspec.
+  VERSION       = '0.4.0' # Keep in sync with gemspec.
 
   ROOT          = File.expand_path(File.dirname(__FILE__) + '/..')
 
@@ -54,7 +54,8 @@ module Docsplit
   def self.extract_pdf(docs, opts={})
     [docs].flatten.each do |doc|
       basename = File.basename(doc, File.extname(doc))
-      run "-jar #{ROOT}/vendor/jodconverter/jodconverter-cli-2.2.2.jar #{doc} #{opts[:output] || '.'}/#{basename}.pdf", [], {}
+      options = "-jar #{ROOT}/vendor/jodconverter/jodconverter-core-3.0-beta-3.jar -r #{ROOT}/vendor/conf/document-formats.js"
+      run "#{options} \"#{doc}\" \"#{opts[:output] || '.'}/#{basename}.pdf\"", [], {}
     end
   end
 

data/lib/docsplit/image_extractor.rb

@@ -46,6 +46,7 @@ module Docsplit
           raise ExtractionFailed, result if $? != 0
         end
       end
+    ensure
       FileUtils.remove_entry_secure tempdir if File.exists?(tempdir)
     end
 

data/lib/docsplit/text_extractor.rb

@@ -71,6 +71,7 @@ module Docsplit
         run "MAGICK_TMPDIR=#{tempdir} OMP_NUM_THREADS=2 gm convert #{MEMORY_ARGS} #{OCR_FLAGS} #{pdf} #{tiff} 2>&1"
         run "tesseract #{tiff} #{base_path} -l eng 2>&1"
       end
+    ensure
       FileUtils.remove_entry_secure tempdir if File.exists?(tempdir)
     end
 

data/vendor/conf/document-formats.js

@@ -0,0 +1,236 @@
+//
+// JODConverter Document Formats Configuration
+//
+[
+  {
+    "name": "Portable Document Format",
+    "extension": "pdf",
+    "mediaType": "application/pdf",
+    "storePropertiesByFamily": {
+      "DRAWING": {"FilterName": "draw_pdf_Export"},
+      "SPREADSHEET": {"FilterName": "calc_pdf_Export"},
+      "PRESENTATION": {"FilterName": "impress_pdf_Export"},
+      "TEXT": {"FilterName": "writer_pdf_Export"}
+    }
+  },
+  {
+    "name": "Macromedia Flash",
+    "extension": "swf",
+    "mediaType": "application/x-shockwave-flash",
+    "storePropertiesByFamily": {
+      "DRAWING": {"FilterName": "draw_flash_Export"},
+      "PRESENTATION": {"FilterName": "impress_flash_Export"}
+    }
+  },
+  {
+    "name": "HTML",
+    "extension": "html",
+    "mediaType": "text/html",
+    "inputFamily": "TEXT",
+    "storePropertiesByFamily": {
+      "SPREADSHEET": {"FilterName": "HTML (StarCalc)"},
+      "PRESENTATION": {"FilterName": "impress_html_Export"},
+      "TEXT": {"FilterName": "HTML (StarWriter)"}
+    }
+  },
+  {
+    "name": "OpenDocument Text",
+    "extension": "odt",
+    "mediaType": "application/vnd.oasis.opendocument.text",
+    "inputFamily": "TEXT",
+    "storePropertiesByFamily": {"TEXT": {"FilterName": "writer8"}}
+  },
+  {
+    "name": "OpenOffice.org 1.0 Text Document",
+    "extension": "sxw",
+    "mediaType": "application/vnd.sun.xml.writer",
+    "inputFamily": "TEXT",
+    "storePropertiesByFamily": {"TEXT": {"FilterName": "StarOffice XML (Writer)"}}
+  },
+  {
+    "name": "Microsoft Word",
+    "extension": "doc",
+    "mediaType": "application/msword",
+    "inputFamily": "TEXT",
+    "storePropertiesByFamily": {"TEXT": {"FilterName": "MS Word 97"}}
+  },
+  {
+    "name": "Microsoft Word 2007 XML",
+    "extension": "docx",
+    "mediaType": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
+    "inputFamily": "TEXT"
+  },
+  {
+    "name": "Rich Text Format",
+    "extension": "rtf",
+    "mediaType": "text/rtf",
+    "inputFamily": "TEXT",
+    "storePropertiesByFamily": {"TEXT": {"FilterName": "Rich Text Format"}}
+  },
+  {
+    "name": "WordPerfect",
+    "extension": "wpd",
+    "mediaType": "application/wordperfect",
+    "inputFamily": "TEXT"
+  },
+  {
+    "name": "Plain Text",
+    "extension": "txt",
+    "mediaType": "text/plain",
+    "inputFamily": "TEXT",
+    "loadProperties": {
+      "FilterName": "Text (encoded)",
+      "FilterOptions": "utf8"
+    },
+    "storePropertiesByFamily": {"TEXT": {
+      "FilterName": "Text (encoded)",
+      "FilterOptions": "utf8"
+    }}
+  },
+  {
+    "name": "MediaWiki wikitext",
+    "extension": "wiki",
+    "mediaType": "text/x-wiki",
+    "storePropertiesByFamily": {"TEXT": {"FilterName": "MediaWiki"}}
+  },
+  {
+    "name": "OpenDocument Spreadsheet",
+    "extension": "ods",
+    "mediaType": "application/vnd.oasis.opendocument.spreadsheet",
+    "inputFamily": "SPREADSHEET",
+    "storePropertiesByFamily": {"SPREADSHEET": {"FilterName": "calc8"}}
+  },
+  {
+    "name": "OpenOffice.org 1.0 Spreadsheet",
+    "extension": "sxc",
+    "mediaType": "application/vnd.sun.xml.calc",
+    "inputFamily": "SPREADSHEET",
+    "storePropertiesByFamily": {"SPREADSHEET": {"FilterName": "StarOffice XML (Calc)"}}
+  },
+  {
+    "name": "Microsoft Excel",
+    "extension": "xls",
+    "mediaType": "application/vnd.ms-excel",
+    "inputFamily": "SPREADSHEET",
+    "storePropertiesByFamily": {"SPREADSHEET": {"FilterName": "MS Excel 97"}}
+  },
+  {
+    "name": "Microsoft Excel 2007 XML",
+    "extension": "xlsx",
+    "mediaType": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+    "inputFamily": "SPREADSHEET"
+  },
+  {
+    "name": "Comma Separated Values",
+    "extension": "csv",
+    "mediaType": "text/csv",
+    "inputFamily": "SPREADSHEET",
+    "loadProperties": {
+      "FilterName": "Text - txt - csv (StarCalc)",
+      "FilterOptions": "44,34,0"
+    },
+    "storePropertiesByFamily": {"SPREADSHEET": {
+      "FilterName": "Text - txt - csv (StarCalc)",
+      "FilterOptions": "44,34,0"
+    }}
+  },
+  {
+    "name": "Tab Separated Values",
+    "extension": "tsv",
+    "mediaType": "text/tab-separated-values",
+    "inputFamily": "SPREADSHEET",
+    "loadProperties": {
+      "FilterName": "Text - txt - csv (StarCalc)",
+      "FilterOptions": "9,34,0"
+    },
+    "storePropertiesByFamily": {"SPREADSHEET": {
+      "FilterName": "Text - txt - csv (StarCalc)",
+      "FilterOptions": "9,34,0"
+    }}
+  },
+  {
+    "name": "OpenDocument Presentation",
+    "extension": "odp",
+    "mediaType": "application/vnd.oasis.opendocument.presentation",
+    "inputFamily": "PRESENTATION",
+    "storePropertiesByFamily": {"PRESENTATION": {"FilterName": "impress8"}}
+  },
+  {
+    "name": "OpenOffice.org 1.0 Presentation",
+    "extension": "sxi",
+    "mediaType": "application/vnd.sun.xml.impress",
+    "inputFamily": "PRESENTATION",
+    "storePropertiesByFamily": {"PRESENTATION": {"FilterName": "StarOffice XML (Impress)"}}
+  },
+  {
+    "name": "Microsoft PowerPoint",
+    "extension": "ppt",
+    "mediaType": "application/vnd.ms-powerpoint",
+    "inputFamily": "PRESENTATION",
+    "storePropertiesByFamily": {"PRESENTATION": {"FilterName": "MS PowerPoint 97"}}
+  },
+  {
+    "name": "Microsoft PowerPoint 2007 XML",
+    "extension": "pptx",
+    "mediaType": "application/vnd.openxmlformats-officedocument.presentationml.presentation",
+    "inputFamily": "PRESENTATION"
+  },
+  {
+    "name": "OpenDocument Drawing",
+    "extension": "odg",
+    "mediaType": "application/vnd.oasis.opendocument.graphics",
+    "inputFamily": "DRAWING",
+    "storePropertiesByFamily": {"DRAWING": {"FilterName": "draw8"}}
+  },
+  {
+    "name": "Scalable Vector Graphics",
+    "extension": "svg",
+    "mediaType": "image/svg+xml",
+    "storePropertiesByFamily": {"DRAWING": {"FilterName": "draw_svg_Export"}}
+  },
+  {
+    "name": "Portable Network Graphic",
+    "extension": "png",
+    "mediaType": "image/png",
+    "storePropertiesByFamily": {
+      "DRAWING": {"FilterName": "draw_png_Export"},
+      "PRESENTATION": {"FilterName": "impress_png_Export"}
+    }
+  },
+  {
+    "name": "Graphics Interchange Format",
+    "extension": "gif",
+    "mediaType": "image/gif",
+    "storePropertiesByFamily": {
+      "DRAWING": {"FilterName": "draw_gif_Export"},
+      "PRESENTATION": {"FilterName": "impress_gif_Export"}
+    }
+  },
+  {
+    "name": "Joint Photographic Experts Group",
+    "extension": "jpg",
+    "mediaType": "image/jpeg",
+    "storePropertiesByFamily": {
+      "DRAWING": {"FilterName": "draw_jpg_Export"},
+      "PRESENTATION": {"FilterName": "impress_jpg_Export"}
+    }
+  },
+  {
+    "name": "Windows Bitmap",
+    "extension": "bmp",
+    "mediaType": "image/bmp",
+    "storePropertiesByFamily": {
+      "DRAWING": {"FilterName": "draw_bmp_Export"},
+      "PRESENTATION": {"FilterName": "impress_bmp_Export"}
+    }
+  },
+  {
+    "name": "Tagged Image File Format",
+    "extension": "tif",
+    "mediaType": "image/tiff",
+    "storePropertiesByFamily": {
+      "DRAWING": {"FilterName": "draw_tif_Export"},
+      "PRESENTATION": {"FilterName": "impress_tif_Export"}
+    }
+  }
+]

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: docsplit
 version: !ruby/object:Gem::Version 
-  hash: 27
+  hash: 15
   prerelease: false
   segments: 
   - 0
-  - 3
   - 4
-  version: 0.3.4
+  - 0
+  version: 0.4.0
 platform: ruby
 authors: 
 - Jeremy Ashkenas
@@ -16,7 +16,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-08-20 00:00:00 -04:00
+date: 2010-08-23 00:00:00 -04:00
 default_executable: 
 dependencies: []
 
@@ -37,16 +37,15 @@ files:
 - lib/docsplit/transparent_pdfs.rb
 - lib/docsplit.rb
 - bin/docsplit
-- vendor/jodconverter/commons-cli-1.2.jar
+- vendor/conf/document-formats.js
+- vendor/jodconverter/commons-cli-1.1.jar
 - vendor/jodconverter/commons-io-1.4.jar
-- vendor/jodconverter/jodconverter-2.2.2.jar
-- vendor/jodconverter/jodconverter-cli-2.2.2.jar
-- vendor/jodconverter/juh-3.0.1.jar
-- vendor/jodconverter/jurt-3.0.1.jar
-- vendor/jodconverter/ridl-3.0.1.jar
-- vendor/jodconverter/slf4j-api-1.5.6.jar
-- vendor/jodconverter/slf4j-jdk14-1.5.6.jar
-- vendor/jodconverter/unoil-3.0.1.jar
+- vendor/jodconverter/jodconverter-core-3.0-beta-3.jar
+- vendor/jodconverter/json-20080701.jar
+- vendor/jodconverter/juh-3.1.0.jar
+- vendor/jodconverter/jurt-3.1.0.jar
+- vendor/jodconverter/ridl-3.1.0.jar
+- vendor/jodconverter/unoil-3.1.0.jar
 - vendor/logging.properties
 - docsplit.gemspec
 - LICENSE