asciidoctor-bibtex 0.4.1 → 0.5.0

data/lib/asciidoctor-bibtex/StringUtils.rb

@@ -0,0 +1,57 @@
+# frozen_string_literal: true
+
+#
+# StringUtils.rb
+#
+# Some utilities for strings.
+#
+# Copyright (c) Peter Lane, 2012-13.
+# Copyright (c) Zhang Yang, 2019.
+#
+# Released under Open Works License, 0.9.2
+#
+
+module AsciidoctorBibtex
+  module StringUtils
+    # Converts html output produced by citeproc to asciidoc markup
+    def self.html_to_asciidoc(s)
+      s = s.gsub(%r{</?i>}, '_')
+      s = s.gsub(%r{</?b>}, '*')
+      s = s.gsub(%r{</?span.*?>}, '')
+      s = s.gsub(/\{|\}/, '')
+      s
+    end
+
+    # Provides a check that a string is in integer
+    def self.is_i?(s)
+      !!(s =~ /^[-+]?[0-9]+$/)
+    end
+
+    # Merge consecutive number so that "1,2,3,5" becomes "1-3,5"
+    def self.combine_consecutive_numbers(str)
+      nums = str.split(',').collect(&:strip)
+      res = ''
+      # Loop through ranges
+      start_range = 0
+      while start_range < nums.length
+        end_range = start_range
+        while (end_range < nums.length - 1) &&
+              is_i?(nums[end_range]) &&
+              is_i?(nums[end_range + 1]) &&
+              (nums[end_range + 1].to_i == nums[end_range].to_i + 1)
+          end_range += 1
+        end
+        if end_range - start_range >= 2
+          res += "#{nums[start_range]}-#{nums[end_range]}, "
+        else
+          start_range.upto(end_range) do |i|
+            res += "#{nums[i]}, "
+          end
+        end
+        start_range = end_range + 1
+      end
+      # finish by removing last comma
+      res.gsub(/, $/, '')
+    end
+  end
+end

data/lib/asciidoctor-bibtex/{styles.rb → StyleUtils.rb}

@@ -3,25 +3,25 @@
 # Simple checks on available styles through CSL
 #
 
-module AsciidoctorBibtex
-
-  module Styles
+require 'citeproc'
+require 'csl/styles'
 
-    def Styles.available
+module AsciidoctorBibtex
+  module StyleUtils
+    def self.available
       CSL::Style.ls
     end
 
-    def Styles.default_style
+    def self.default_style
       'apa'
     end
 
-    def Styles.valid? style
+    def self.valid?(style)
       Styles.available.include? style
     end
 
-    def Styles.is_numeric? style
+    def self.is_numeric?(style)
       CSL::Style.load(style).citation_format == :numeric
     end
   end
 end
-

data/lib/asciidoctor-bibtex/extensions.rb

@@ -1,64 +1,174 @@
-# Some extension and helper methods. 
 #
-# Copyright (c) Peter Lane, 2012-13.
-# Released under Open Works License, 0.9.2
+# Treeprocessor extension for asciidoctor
+#
 
-module AsciidoctorBibtexArrayExtensions
+require 'asciidoctor'
+require 'asciidoctor/extensions'
+require 'asciidoctor/reader'
+require 'asciidoctor/parser'
 
-  # Retrieve the third item of an array
-  # Note: no checks for validity
-  def third
-    self[2]
-  end
+require_relative 'PathUtils'
+require_relative 'Processor'
 
-  # Join items in array using commas and 'and' on last item
-  def comma_and_join
-    if size < 2
-      return self.join("")
-    end
-    result = ""
-    self.each_with_index do |item, index|
-      if index.zero?
-        result << item
-      elsif index == size-1
-        result << " and #{item}"
-      else
-        result << ", #{item}"
+module AsciidoctorBibtex
+  module Asciidoctor
+    # Placeholder paragraph for the bibliography paragraph. Choose a uuid so
+    # that it is a special word unlikeky to conflict with normal texts.
+    BibliographyBlockMacroPlaceholder = %(a5d42deb-3cfc-4293-b96a-fcb47316ce56)
+
+    # BibliographyBlockMacro
+    #
+    # The `bibliography::[] block macro` processor.
+    #
+    # This macro processor does only half the work. It just parse the macro
+    # and set bibtex file and style if found in the macro. The macro is then
+    # expanded to a special paragraph, which is then replaced with generated
+    # references by the treeprocessor.
+    #
+    class BibliographyBlockMacro < ::Asciidoctor::Extensions::BlockMacroProcessor
+      use_dsl
+      named :bibliography
+      name_positional_attributes :style, :locale
+
+      def process(parent, target, attrs)
+        # NOTE: bibtex-file and bibtex-style set by this macro shall be
+        # overridable by document attributes and commandline arguments. So we
+        # respect the convention here.
+        if target && (!parent.document.attr? 'bibtex-file')
+          parent.document.set_attribute 'bibtex-file', target
+        end
+        if attrs.key?(:style) && (!parent.document.attr? 'bibtex-style')
+          parent.document.set_attribute 'bibtex-style', attrs[:style]
+        end
+        if attrs.key?(:locale) && (!parent.document.attr? 'bibtex-locale')
+          parent.document.set_attribute 'bibtex-locale', attrs[:locale]
+        end
+        create_paragraph parent, BibliographyBlockMacroPlaceholder, {}
       end
     end
 
-    return result
-  end
-end
+    # CitationProcessor
+    #
+    # A tree processor to replace all citations and bibliography.
+    #
+    # This processor scans the document, generates a list of citations, replaces
+    # each citation with citation text and the reference block macro placeholder
+    # with the final bibliography list. It relys on the block macro processor to
+    # generate the place holder.
+    #
+    # NOTE: According to the asiidoctor extension policy, the tree processor can
+    # only produce texts with inline macros.
+    #
+    class CitationProcessor < ::Asciidoctor::Extensions::Treeprocessor
+      def process(document)
+        bibtex_file = (document.attr 'bibtex-file').to_s
+        bibtex_style = ((document.attr 'bibtex-style') || 'ieee').to_s
+        bibtex_locale = ((document.attr 'bibtex-locale') || 'en-US').to_s
+        bibtex_order = ((document.attr 'bibtex-order') || 'appearance').to_sym
+        bibtex_format = ((document.attr 'bibtex-format') || 'asciidoc').to_sym
+        bibtex_throw = ((document.attr 'bibtex-throw') || 'false').to_s.downcase
 
-# monkey patch the extension methods to Array
-class Array
-  include AsciidoctorBibtexArrayExtensions
-end
+        # Fild bibtex file automatically if not supplied.
+        if bibtex_file.empty?
+          bibtex_file = AsciidoctorBibtex::PathUtils.fild_bibfile '.'
+        end
+        if bibtex_file.empty?
+          bibtex_file = AsciidoctorBibtex::PathUtils.find_bibfile "#{ENV['HOME']}/Documents"
+        end
+        if bibtex_file.empty?
+          puts 'Error: bibtex-file is not set and automatic search failed'
+          exit
+        end
 
-# Converts html output produced by citeproc to asciidoc markup
-module StringHtmlToAsciiDoc
-  def html_to_asciidoc
-    r = self.gsub(/<\/?i>/, '_')
-    r = r.gsub(/<\/?b>/, '*')
-    r = r.gsub(/<\/?span.*?>/, '')
-    r = r.gsub(/\{|\}/, '')
-    r
-  end
-end
+        # Extract all AST nodes that can contain citations.
+        prose_blocks = document.find_by do |b|
+          (b.content_model == :simple) ||
+            (b.context == :list_item) ||
+            (b.context == :table_cell)
+        end
+        return nil if prose_blocks.nil?
 
-# Provides a check that a string is in integer
-# Taken from:
-# http://stackoverflow.com/questions/1235863/test-if-a-string-is-basically-an-integer-in-quotes-using-ruby
-module IntegerCheck
-  def is_i?
-    !!(self =~ /^[-+]?[0-9]+$/)
+        processor = Processor.new bibtex_file, true, bibtex_style, bibtex_locale,
+                                  bibtex_order == :appearance, bibtex_format,
+                                  bibtex_throw == 'true'
+
+        # First pass: extract all citations.
+        prose_blocks.each do |block|
+          if block.context == :list_item || block.context == :table_cell
+            line = block.text
+            processor.process_citation_macros line
+          else
+            block.lines.each do |line|
+              processor.process_citation_macros line
+            end
+          end
+        end
+        # Make processor finalize macro processing as required.
+        processor.finalize_macro_processing
+
+        # Second pass: replace citations with citation texts.
+        prose_blocks.each do |block|
+          if block.context == :list_item || block.context == :table_cell
+            line = block.text
+            line = processor.replace_citation_macros(line)
+            block.text = line
+          else
+            block.lines.each_with_index do |line, index|
+              line = processor.replace_citation_macros(line)
+              block.lines[index] = line
+            end
+          end
+        end
+
+        references_asciidoc = []
+        if (bibtex_format == :latex) || (bibtex_format == :bibtex)
+          references_asciidoc << %(+++\\bibliography{#{bibtex_file}}{}+++)
+          references_asciidoc << %(+++\\bibliographystyle{#{bibtex_style}}+++)
+        elsif bibtex_format == :biblatex
+          references_asciidoc << %(+++\\printbibliography+++)
+        else
+          references_asciidoc = processor.build_bibliography_list
+        end
+
+        # Third pass: replace the bibliography paragraph with the bibliography
+        # list.
+        biblio_blocks = document.find_by do |b|
+          # for fast search (since most searches shall fail)
+          (b.content_model == :simple) && (b.lines.size == 1) \
+            && (b.lines[0] == BibliographyBlockMacroPlaceholder)
+        end
+        biblio_blocks.each do |block|
+          block_index = block.parent.blocks.index do |b|
+            b == block
+          end
+          reference_blocks = parse_asciidoc block.parent, references_asciidoc
+          reference_blocks.reverse.each do |b|
+            block.parent.blocks.insert block_index, b
+          end
+          block.parent.blocks.delete_at block_index + reference_blocks.size
+        end
+
+        nil
+      end
+
+      # This is an adapted version of Asciidoctor::Extension::parse_content,
+      # where resultant blocks are returned as a list instead of attached to
+      # the parent.
+      def parse_asciidoc(parent, content, attributes = {})
+        result = []
+        reader = ::Asciidoctor::Reader.new content
+        while reader.has_more_lines?
+          block = ::Asciidoctor::Parser.next_block reader, parent, attributes
+          result << block if block
+        end
+        result
+      end
+    end
   end
 end
 
-# monkey patch the extension methods into String
-class String
-  include StringHtmlToAsciiDoc
-  include IntegerCheck
+# Register the extensions to asciidoctor
+Asciidoctor::Extensions.register do
+  block_macro AsciidoctorBibtex::Asciidoctor::BibliographyBlockMacro
+  treeprocessor AsciidoctorBibtex::Asciidoctor::CitationProcessor
 end
-

data/lib/asciidoctor-bibtex/version.rb

@@ -1,3 +1,3 @@
 module AsciidoctorBibtex
-  VERSION = '0.4.1'
+  VERSION = '0.5.0'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: asciidoctor-bibtex
 version: !ruby/object:Gem::Version
-  version: 0.4.1
+  version: 0.5.0
 platform: ruby
 authors:
 - Zhang YANG
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-05-20 00:00:00.000000000 Z
+date: 2019-08-11 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: asciidoctor
@@ -16,7 +16,7 @@ dependencies:
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 1.5.0
+        version: 2.0.0
     - - "<"
       - !ruby/object:Gem::Version
         version: 3.0.0
@@ -26,7 +26,7 @@ dependencies:
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 1.5.0
+        version: 2.0.0
     - - "<"
       - !ruby/object:Gem::Version
         version: 3.0.0
@@ -86,6 +86,48 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '0.2'
+- !ruby/object:Gem::Dependency
+  name: minitest
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '5'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '5'
+- !ruby/object:Gem::Dependency
+  name: bundler
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '2'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '2'
+- !ruby/object:Gem::Dependency
+  name: rake
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '12'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '12'
 description: |
   asciidoctor-bibtex adds bibtex support for asciidoc documents by introducing
   two new macros: `cite:[KEY]` and `bibliography::[]`. Citations are parsed and
@@ -100,19 +142,14 @@ files:
 - LICENSE.txt
 - README.md
 - lib/asciidoctor-bibtex.rb
-- lib/asciidoctor-bibtex/asciidoctor.rb
-- lib/asciidoctor-bibtex/bibextension.rb
-- lib/asciidoctor-bibtex/citation.rb
-- lib/asciidoctor-bibtex/citationdata.rb
-- lib/asciidoctor-bibtex/citations.rb
-- lib/asciidoctor-bibtex/citationutils.rb
+- lib/asciidoctor-bibtex/CitationMacro.rb
+- lib/asciidoctor-bibtex/CitationUtils.rb
+- lib/asciidoctor-bibtex/PathUtils.rb
+- lib/asciidoctor-bibtex/Processor.rb
+- lib/asciidoctor-bibtex/StringUtils.rb
+- lib/asciidoctor-bibtex/StyleUtils.rb
 - lib/asciidoctor-bibtex/extensions.rb
-- lib/asciidoctor-bibtex/filehandlers.rb
-- lib/asciidoctor-bibtex/processor.rb
-- lib/asciidoctor-bibtex/processorutils.rb
-- lib/asciidoctor-bibtex/styles.rb
 - lib/asciidoctor-bibtex/version.rb
-- lib/utils.rb
 homepage: https://github.com/asciidoctor/asciidoctor-bibtex
 licenses:
 - OWL

data/lib/asciidoctor-bibtex/asciidoctor.rb

@@ -1,7 +0,0 @@
-require 'asciidoctor/extensions'
-require_relative 'bibextension'
-
-Asciidoctor::Extensions.register do
-  block_macro AsciidoctorBibtex::Asciidoctor::BibliographyBlockMacro
-  treeprocessor AsciidoctorBibtex::Asciidoctor::CitationProcessor
-end

data/lib/asciidoctor-bibtex/bibextension.rb

@@ -1,177 +0,0 @@
-#
-# Treeprocessor extension for asciidoctor
-#
-
-require 'asciidoctor'
-require 'asciidoctor/extensions'
-require 'asciidoctor/reader'
-require 'asciidoctor/parser'
-require 'bibtex/filters'
-require 'latex/decode/base'
-require 'latex/decode/maths'
-require 'latex/decode/accents'
-require 'latex/decode/diacritics'
-require 'latex/decode/punctuation'
-require 'latex/decode/symbols'
-require 'latex/decode/greek'
-require_relative 'styles'
-require_relative 'filehandlers'
-
-module AsciidoctorBibtex
-  module Asciidoctor
-
-    # This filter extends the original latex filter in bibtex-ruby to handle
-    # unknown latex macros more gracefully. We could have used latex-decode
-    # gem together with our custom replacement rules, but latex-decode eats up
-    # all braces after it finishes all decoding. So we hack over the
-    # LaTeX.decode function and insert our rules before `strip_braces`.
-    class LatexFilter < ::BibTeX::Filter
-      def apply(value)
-        text = value.to_s
-        LaTeX::Decode::Base.normalize(text)
-        LaTeX::Decode::Maths.decode!(text)
-        LaTeX::Decode::Accents.decode!(text)
-        LaTeX::Decode::Diacritics.decode!(text)
-        LaTeX::Decode::Punctuation.decode!(text)
-        LaTeX::Decode::Symbols.decode!(text)
-        LaTeX::Decode::Greek.decode!(text)
-        text = text.gsub(/\\url\{(.+?)\}/, " \\1 ").gsub(/\\\w+(?=\s+\w)/, "").gsub(/\\\w+(?:\[.+?\])?\s*\{(.+?)\}/, "\\1")
-        LaTeX::Decode::Base.strip_braces(text)
-        LaTeX.normalize_C(text)
-      end
-    end
-
-    BibliographyBlockMacroPlaceholder = %(BIBLIOGRAPHY BLOCK MACRO PLACEHOLDER)
-
-    # This macro processor does only half the work. It just parse the macro
-    # and set bibtex file and style if found in the macro. The macro is then
-    # expanded to a special paragraph, which is then replaced with generated
-    # references by the treeprocessor.
-    class BibliographyBlockMacro < ::Asciidoctor::Extensions::BlockMacroProcessor
-      use_dsl
-      named :bibliography
-      name_positional_attributes :style
-
-      def process parent, target, attrs
-        # NOTE: bibtex-file and bibtex-style set by this macro shall be
-        # overridable by document attributes and commandline arguments. So we
-        # respect the convention here.
-        if target and not parent.document.attr? 'bibtex-file'
-          parent.document.set_attribute 'bibtex-file', target
-        end
-        if attrs.key? :style and not parent.document.attr? 'bibtex-style'
-          parent.document.set_attribute 'bibtex-style', attrs[:style]
-        end
-        if attrs.key? :locale and not parent.document.attr? 'bibtex-locale'
-          parent.document.set_attribute 'bibtex-locale', attrs[:locale]
-        end
-        create_paragraph parent, BibliographyBlockMacroPlaceholder, {}
-      end
-    end
-
-    # This processor scans the document, generates a list of citations,
-    # replace each citation with correct text and the reference block macro
-    # placeholder with the final reference list. It relys on the block macro
-    # processor to generate the place holder.
-    class CitationProcessor < ::Asciidoctor::Extensions::Treeprocessor
-
-      def process document
-        bibtex_file = (document.attr 'bibtex-file').to_s
-        bibtex_style = ((document.attr 'bibtex-style') || 'ieee').to_s
-        bibtex_locale = ((document.attr 'bibtex-locale') || 'en-US').to_s
-        bibtex_order = ((document.attr 'bibtex-order') || 'appearance').to_sym
-        bibtex_format = ((document.attr 'bibtex-format') || 'asciidoc').to_sym
-
-        if bibtex_file.empty?
-          bibtex_file = AsciidoctorBibtex::FileHandlers.find_bibliography "."
-        end
-        if bibtex_file.empty?
-          bibtex_file = AsciidoctorBibtex::FileHandlers.find_bibliography "#{ENV['HOME']}/Documents"
-        end
-        if bibtex_file.empty?
-          puts "Error: bibtex-file is not set and automatic search failed"
-          exit
-        end
-
-        bibtex = BibTeX.open bibtex_file, :filter => [LatexFilter]
-        processor = Processor.new bibtex, true, bibtex_style, bibtex_locale, bibtex_order == :appearance, bibtex_format
-
-        prose_blocks = document.find_by {|b| b.content_model == :simple or b.context == :list_item}
-        prose_blocks.each do |block|
-          if block.context == :list_item
-            line = block.instance_variable_get :@text
-            processor.citations.add_from_line line
-          else
-            block.lines.each do |line|
-              processor.citations.add_from_line line
-            end
-          end
-        end
-
-        prose_blocks.each do |block|
-          if block.context == :list_item
-            line = block.instance_variable_get :@text
-            processor.citations.retrieve_citations(line).each do |citation|
-              line = line.gsub(citation.original, processor.complete_citation(citation))
-            end
-            block.instance_variable_set :@text, line
-          else
-            block.lines.each do |line|
-              tmp = line.clone
-              processor.citations.retrieve_citations(line).each do |citation|
-                tmp = tmp.gsub(citation.original, processor.complete_citation(citation))
-              end
-              line.replace tmp
-            end
-          end
-        end
-
-        references_asciidoc = []
-        if bibtex_format == :latex or bibtex_format == :bibtex
-          references_asciidoc << %(+++\\bibliography{#{bibtex_file}}{}+++)
-          references_asciidoc << %(+++\\bibliographystyle{#{bibtex_style}}+++)
-        elsif bibtex_format == :biblatex
-          references_asciidoc << %(+++\\printbibliography+++)
-        else
-          processor.cites.each do |ref|
-            references_asciidoc << processor.get_reference(ref)
-            references_asciidoc << ''
-          end
-        end
-
-        biblio_blocks = document.find_by do |b|
-          # for fast search (since most searches shall fail)
-          b.content_model == :simple and b.lines.size == 1 \
-            and b.lines[0] == BibliographyBlockMacroPlaceholder
-        end
-        biblio_blocks.each do |block|
-          block_index = block.parent.blocks.index do |b|
-            b == block
-          end
-          reference_blocks = parse_asciidoc block.parent, references_asciidoc
-          reference_blocks.reverse.each do |b|
-            block.parent.blocks.insert block_index, b
-          end
-          block.parent.blocks.delete_at block_index + reference_blocks.size
-        end
-
-        nil
-      end
-
-      # This is an adapted version of Asciidoctor::Extension::parse_content,
-      # where resultant blocks are returned as a list instead of attached to
-      # the parent.
-      def parse_asciidoc parent, content, attributes = {}
-        result = []
-        reader = ::Asciidoctor::Reader.new content
-        while reader.has_more_lines?
-          block = ::Asciidoctor::Parser.next_block reader, parent, attributes
-          result << block if block
-        end
-        result
-      end
-
-    end
-
-  end
-end