doc2text 0.3.2 → 0.3.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: e795071f6df878e0427e2728848d170ecb1277d0
-  data.tar.gz: 4c02f808c1d6c47bf7edf8f1d81fc534bdf2ab93
+  metadata.gz: f2530e3597409c35637493d9d8e9842b2f70f005
+  data.tar.gz: 304eb91cfa9b109bfa853cbc7d9736a14b7817bd
 SHA512:
-  metadata.gz: d7a9314e425c425dc228e4f742874f3393ec77ab604e22ead973cb1430e5a01bd7395a5e8f6b6f2c71114aaad2736b5d43abcad30bb797a17a0ce842f6474ecd
-  data.tar.gz: c82f62f3fabd0ccdf42ec6fb23d06cb3f6884676184e6a44e670e4a9f803b7fa2bc71ebd107ff067315e1ac52214c79c0477d9fd0a0388a0010f5c70fd915678
+  metadata.gz: 9ff147cd7f87f233a6fbbc9270e682426ffc3b5df514a5b7ca2b16b0405a4048901cb53972d5264c853eabce7532b9bb386f97847e5b072189119ec616b2961f
+  data.tar.gz: f992a33120ae4273966a54a15e02768abf4baa16473808775e938fed21163882e46c46333dfa7ea9a63cf5bd9e4ef02de4bbb8f50b7d9ab86dbe5a4a8ee6d3cc

data/lib/doc2text.rb

@@ -2,10 +2,11 @@ require 'nokogiri'
 #require 'nokogiri/xml'
 require 'fileutils'
 
+require 'doc2text/resolution'
 require 'doc2text/odt'
 require 'doc2text/odt_xml_node'
 require 'doc2text/odt_xml_namespaces'
 require 'doc2text/markdown_odt_parser'
 require 'doc2text/errors'
 
-require 'doc2text/content'
+require 'doc2text/styles_parser'

data/lib/doc2text/markdown_odt_parser.rb

@@ -2,13 +2,14 @@ require 'logger'
 
 module Doc2Text
   module Markdown
-    class OdtParser
-      def initialize(output)
+    class OdtParser < Nokogiri::XML::SAX::Document
+      def initialize(output, styles_xml_root = nil)
+        @styles_xml_root = styles_xml_root
         @output = output
         @automatic_styles = {}
       end
 
-      def new_node(prefix, name, attrs)
+      def start_element_namespace(name ,attrs = [], prefix = nil, uri = nil, ns = [])
         unless @xml_root
           @xml_root = @current_node = Odt::XmlNodes::Node.create_node prefix, name, nil, attrs, self
         else
@@ -18,22 +19,19 @@ module Doc2Text
         end
       end
 
-      def close_node(prefix, name)
-        # if Odt::XmlNodes::Node.create_node(prefix, name, nil, [], self).eql? @current_node
-          if @current_node.parent and @current_node.parent.office_text?
-            @output << @current_node.expand
-            @current_node.delete
-          end
-          @current_node = @current_node.parent
-        # else
-        #   # TODO remove this redundant(tree build algorithm) checks
-        #   raise Doc2Text::XmlError, "!Close node child #{prefix} #{name} IS NOT correct, CURRENT_ELEM #{@current_node}"
-        # end
+      def end_element_namespace(name, prefix = nil, uri = nil)
+        if @current_node.parent and @current_node.parent.office_text?
+          @output << @current_node.expand
+          @current_node.delete
+        end
+        @current_node = @current_node.parent
       end
 
-      def text(string)
-        plain_text = Odt::XmlNodes::PlainText.new(string)
-        @current_node.children << plain_text
+      def characters(string)
+        unless string.strip.empty?
+          plain_text = Odt::XmlNodes::PlainText.new(string)
+          @current_node.children << plain_text
+        end
       end
 
       def close
@@ -50,18 +48,29 @@ module Doc2Text
       # Select nodes xpath style
       # - supports selecting from the root node
       def xpath(string)
-        if /^(\/[\w:\-]+)+$/ =~ string
-          path = string.scan /[\w:\-]+/
-          seek_nodes = [@xml_root]
-          path.each_with_index do |xml_name, index|
-            seek_nodes.select! { |node| node.xml_name == xml_name }
-            seek_nodes = seek_nodes.map(&:children).flatten unless index == path.length - 1
-            break if seek_nodes.empty?
+        patterns = string.split '|'
+        raise Doc2Text::XmlError, 'it does not support this xpath syntax' if patterns.length == 0
+        result = []
+        patterns.each do |pattern|
+          if /^(\/[\w:\-]+)+$/ =~ pattern
+            path = pattern.scan /[\w:\-]+/
+            result += xpath_search_nodes(path, @xml_root)
+            result += xpath_search_nodes(path, @styles_xml_root) if @styles_xml_root
+          else
+            raise Doc2Text::XmlError, 'it does not support this xpath syntax'
           end
-          seek_nodes
-        else
-          raise Doc2Text::XmlError, 'it does not support this xpath syntax'
         end
+        result
+      end
+
+      def xpath_search_nodes(path, xml_root)
+        seek_nodes = [xml_root]
+        path.each_with_index do |xml_name, index|
+          seek_nodes.select! { |node| node.xml_name == xml_name }
+          seek_nodes = seek_nodes.map(&:children).flatten unless index == path.length - 1
+          break if seek_nodes.empty?
+        end
+        seek_nodes
       end
 
       def logger

data/lib/doc2text/odt.rb

@@ -9,8 +9,9 @@ module Doc2Text
         odt = new input
         begin
           odt.unpack
+          styles_xml_root = odt.parse_styles
           output = File.open output_filename, 'w'
-          markdown = Markdown::OdtParser.new output
+          markdown = Markdown::OdtParser.new output, styles_xml_root
           begin
             odt.parse markdown
           ensure
@@ -21,9 +22,15 @@ module Doc2Text
         end
       end
 
+      def parse_styles
+        styles_parser = Doc2Text::Odt::StylesParser.new
+        xml = Nokogiri::XML::SAX::Parser.new(styles_parser)
+        xml.parse open 'styles.xml'
+        styles_parser.xml_root
+      end
+
       def parse(markdown)
-        content = ::Doc2Text::Odt::Content::Document.new markdown
-        parser = Nokogiri::XML::SAX::Parser.new(content) # { |config| config.strict}
+        parser = Nokogiri::XML::SAX::Parser.new(markdown) # { |config| config.strict}
         parser.parse open 'content.xml'
       end
 

data/lib/doc2text/odt_xml_namespaces.rb

@@ -88,7 +88,7 @@ module Doc2Text
       module Text
         def initialize(parent = nil, attrs = [], prefix = nil, name = nil, markdown_odt_parser = nil)
           super parent, attrs, prefix, name
-          @markdown_odt_parser = markdown_odt_parser
+          @xml_parser = markdown_odt_parser
           style_index = attrs.index { |attr| attr.prefix == 'text' && attr.localname == 'style-name' }
           @enclosing_style = []
           if style_index and fetch_style?
@@ -118,7 +118,7 @@ module Doc2Text
         end
 
         def find_style(style_name)
-          styles = @markdown_odt_parser.xpath '/office:document-content/office:automatic-styles/style:style'
+          styles = @xml_parser.xpath '/office:document-content/office:automatic-styles/style:style'
           styles.find { |style| style.attrs.index { |attr| attr.prefix == 'style' && attr.localname == 'family' } &&
               style.attrs.index { |attr| attr.prefix == 'style' && attr.localname == 'name' && attr.value == style_name } }
         end
@@ -169,6 +169,12 @@ module Doc2Text
           def close
             @enclosing_style.reverse.join
           end
+
+          def expand
+            expanded = "#{open}#{@children.map(&:expand).join}#{close}"
+            delete
+            expanded.clone
+          end
         end
 
         # http://docs.oasis-open.org/office/v1.2/os/OpenDocument-v1.2-os-part1.html#__RefHeading__1415154_253892949

data/lib/doc2text/odt_xml_node.rb

@@ -21,7 +21,7 @@ module Doc2Text
         end
 
         def initialize(parent = nil, attrs = [], prefix = nil, name = nil, markdown_odt_parser = nil)
-          @parent, @attrs, @prefix, @name, @markdown_odt_parser = parent, attrs, prefix, name, markdown_odt_parser
+          @parent, @attrs, @prefix, @name, @xml_parser = parent, attrs, prefix, name, markdown_odt_parser
           @children = []
           @has_text = false
         end

data/lib/doc2text/resolution.rb

@@ -0,0 +1,17 @@
+module Doc2Text
+  class Resolution
+    def self.parse_and_save(source, output)
+      case File.extname source
+        when '.doc', '.docx'
+          mid_name = File.join(File.dirname(output),
+                               File.basename(source, File.extname(source)) + '.odt')
+          system "soffice --headless --convert-to odt #{source}  --outdir #{File.dirname output}"
+          source = mid_name
+          Doc2Text::Odt::Document.parse_and_save source, output
+          File.delete(mid_name)
+        else
+          Doc2Text::Odt::Document.parse_and_save source, output
+      end
+    end
+  end
+end

data/lib/doc2text/styles_parser.rb

@@ -0,0 +1,28 @@
+module Doc2Text
+  module Odt
+    class StylesParser < Nokogiri::XML::SAX::Document
+      attr_reader :xml_root
+
+      def start_element_namespace(name ,attrs = [], prefix = nil, uri = nil, ns = [])
+        unless @xml_root
+          @xml_root = @current_node = Odt::XmlNodes::Node.create_node prefix, name, nil, attrs, self
+        else
+          new_node = Odt::XmlNodes::Node.create_node prefix, name, @current_node, attrs, self
+          @current_node.children << new_node
+          @current_node = new_node
+        end
+      end
+
+      def end_element_namespace(name, prefix = nil, uri = nil)
+        @current_node = @current_node.parent
+      end
+
+      def characters(_)
+      end
+
+      def xpath(_)
+        []
+      end
+    end
+  end
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: doc2text
 version: !ruby/object:Gem::Version
-  version: 0.3.2
+  version: 0.3.3
 platform: ruby
 authors:
 - Valentin Aitken
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-10-23 00:00:00.000000000 Z
+date: 2014-10-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: nokogiri
@@ -57,12 +57,13 @@ extensions: []
 extra_rdoc_files: []
 files:
 - lib/doc2text.rb
-- lib/doc2text/content.rb
 - lib/doc2text/errors.rb
 - lib/doc2text/markdown_odt_parser.rb
 - lib/doc2text/odt.rb
 - lib/doc2text/odt_xml_namespaces.rb
 - lib/doc2text/odt_xml_node.rb
+- lib/doc2text/resolution.rb
+- lib/doc2text/styles_parser.rb
 homepage: http://doc2text.com
 licenses:
 - GPL
@@ -88,3 +89,4 @@ signing_key:
 specification_version: 4
 summary: Translates odt to markdown
 test_files: []
+has_rdoc: 

data/lib/doc2text/content.rb

@@ -1,25 +0,0 @@
-module Doc2Text
-  module Odt
-    module Content
-      class Document < ::Nokogiri::XML::SAX::Document
-        def initialize(markdown_odt_parser)
-          @markdown_odt_parser = markdown_odt_parser
-        end
-
-        def start_element_namespace(name ,attrs = [], prefix = nil, uri = nil, ns = [])
-          @markdown_odt_parser.new_node prefix, name, attrs
-        end
-
-        def end_element_namespace(name, prefix = nil, uri = nil)
-          @markdown_odt_parser.close_node prefix, name
-        end
-
-        def characters(string)
-          unless string.strip.empty?
-            @markdown_odt_parser.text string
-          end
-        end
-      end
-    end
-  end
-end