asciidoctor-bibliography 0.0.1.dev

data/deprecated/asciidoctor-bibliography/processor.rb

@@ -0,0 +1,208 @@
+#
+# Manage the current set of citations, the document settings,
+# and main operations.
+#
+
+module AsciidoctorBibliography
+
+  # Class used through utility method to hold data about citations for
+  # current document, and run the different steps to add the citations
+  # and bibliography
+  class Processor
+    include ProcessorUtils
+
+    attr_reader :biblio, :links, :style, :citations
+
+    def initialize biblio, links, style, numeric_in_appearance_order = false, output = :asciidoc, bibfile = ""
+      @biblio = biblio
+      @links = links
+      @numeric_in_appearance_order = numeric_in_appearance_order
+      @style = style
+      @citations = Citations.new
+      @filenames = Set.new
+      @output = output
+      @bibfile = bibfile
+
+      if output != :latex
+        @citeproc = CiteProc::Processor.new style: @style, format: :html
+        @citeproc.import @biblio.to_citeproc
+      end
+    end
+
+    # Return the complete citation text for given cite_data
+    def complete_citation cite_data
+
+      if @output == :latex
+        result = '+++'
+        cite_data.cites.each do |cite|
+          # NOTE: xelatex does not support "\citenp", so we output all
+          # references as "cite" here.
+          # result << "\\" << cite_data.type
+          result << "\\" << 'cite'
+          if cite.pages != ''
+            result << "[p. " << cite.pages << "]"
+          end
+          result << "{" << "#{cite.ref}" << "},"
+        end
+        if result[-1] == ','
+          result = result[0..-2]
+        end
+        result << "+++"
+        return result
+      else
+        result = ''
+        ob, cb = '(', ')'
+
+        cite_data.cites.each_with_index do |cite, index|
+          # before all items apart from the first, insert appropriate separator
+          result << "#{separator} " unless index.zero?
+
+          # @links requires adding hyperlink to reference
+          result << "<<#{cite.ref}," if @links and (cite_data.type != 'fullcite')
+
+          # if found, insert reference information
+          unless biblio[cite.ref].nil?
+            item = biblio[cite.ref].clone
+            cite_text, ob, cb = make_citation item, cite.ref, cite_data, cite
+          else
+            puts "Unknown reference: #{cite.ref}"
+            cite_text = "#{cite.ref}"
+          end
+
+          result << cite_text.html_to_asciidoc
+          # @links requires finish hyperlink
+          result << ">>" if @links and (cite_data.type != 'fullcite')
+        end
+
+        unless @links
+          # combine numeric ranges
+          if Styles.is_numeric? @style
+            result = combine_consecutive_numbers result
+          end
+        end
+
+        return include_pretext result, cite_data, ob, cb
+      end
+    end
+
+    # Retrieve text for reference in given style
+    # - ref is reference for item to give reference for
+    def get_reference ref
+      result = ""
+      result << ". " if Styles.is_numeric? @style
+
+      begin
+        cptext = @citeproc.render :bibliography, id: ref
+      rescue Exception => e
+        puts "Failed to render #{ref}: #{e}"
+      end
+      result << "[[#{ref}]]" if @links
+      if cptext.nil?
+        return result+ref
+      else
+        result << cptext.first
+      end
+
+      return result.html_to_asciidoc
+    end
+
+    def separator
+      if Styles.is_numeric? @style
+        ','
+      else
+        ';'
+      end
+    end
+
+    # Format pages with pp/p as appropriate
+    def with_pp pages
+      return '' if pages.empty?
+
+      if @style.include? "chicago"
+        pages
+      elsif pages.include? '-'
+        "pp.&#160;#{pages}"
+      else
+        "p.&#160;#{pages}"
+      end
+    end
+
+    # Return page string for given cite
+    def page_str cite
+      result = ''
+      unless cite.pages.empty?
+        result << "," unless Styles.is_numeric? @style
+        result << " #{with_pp(cite.pages)}"
+      end
+
+      return result
+    end
+
+    def include_pretext result, cite_data, ob, cb
+      pretext = cite_data.pretext
+      pretext += ' ' unless pretext.empty? # add space after any content
+
+      if Styles.is_numeric? @style
+        "#{pretext}#{ob}#{result}#{cb}"
+      elsif cite_data.type == "cite"
+        "#{ob}#{pretext}#{result}#{cb}"
+      else
+        "#{pretext}#{result}"
+      end
+    end
+
+    # Numeric citations are handled by computing the position of the reference
+    # in the list of used citations.
+    # Other citations are formatted by citeproc.
+    def make_citation item, ref, cite_data, cite
+      if cite_data.type == "fullcite"
+        cite_text = @citeproc.render(:bibliography, id: ref).join
+
+        fc = ''
+        lc = ''
+      elsif Styles.is_numeric? @style
+        cite_text = if @numeric_in_appearance_order
+                      "#{@citations.cites_used.index(cite.ref) + 1}"
+                    else
+                      "#{sorted_cites.index(cite.ref) + 1}"
+                    end
+        fc = '['
+        lc = ']'
+      else
+        cite_text = @citeproc.process id: ref, mode: :citation
+
+        fc = cite_text[0,1]
+        lc = cite_text[-1,1]
+        cite_text = cite_text[1..-2]
+      end
+
+      if cite_data.type == "fullcite"
+        cite_text = cite_text[0...-1] + page_str(cite) + cite_text[-1]
+      elsif Styles.is_numeric? @style
+        cite_text << "#{page_str(cite)}"
+      elsif cite_data.type == "citenp"
+        cite_text.gsub!(item.year, "#{fc}#{item.year}#{page_str(cite)}#{lc}")
+        cite_text.gsub!(", #{fc}", " #{fc}")
+      else
+        cite_text << page_str(cite)
+      end
+
+      cite_text.gsub!(",", "&#44;") if @links # replace comma
+
+      return cite_text, fc, lc
+    end
+
+    def sorted_cites
+      @citations.sorted_cites @biblio
+    end
+
+    def cites
+      if Styles.is_numeric?(@style) and @numeric_in_appearance_order
+        @citations.cites_used
+      else
+        sorted_cites
+      end
+    end
+
+  end
+end

data/deprecated/asciidoctor-bibliography/processorutils.rb

@@ -0,0 +1,34 @@
+
+module AsciidoctorBibliography
+  module ProcessorUtils
+    # Used with numeric styles to combine consecutive numbers into ranges
+    # e.g. 1,2,3 -> 1-3, or 1,2,3,6,7,8,9,12 -> 1-3,6-9,12
+    # leave references with page numbers alone
+    def combine_consecutive_numbers str
+      nums = str.split(",").collect(&:strip)
+      res = ""
+      # Loop through ranges
+      start_range = 0
+      while start_range < nums.length do
+        end_range = start_range
+        while (end_range < nums.length-1 and
+               nums[end_range].is_i? and
+               nums[end_range+1].is_i? and
+               nums[end_range+1].to_i == nums[end_range].to_i + 1) do
+          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/deprecated/asciidoctor-bibliography/styles.rb

@@ -0,0 +1,27 @@
+#
+# styles.rb
+# Simple checks on available styles through CSL
+#
+
+module AsciidoctorBibliography
+
+  module Styles
+
+    def Styles.available
+      CSL::Style.ls
+    end
+
+    def Styles.default_style
+      'apa'
+    end
+
+    def Styles.valid? style
+      Styles.available.include? style
+    end
+
+    def Styles.is_numeric? style
+      CSL::Style.load(style).citation_format == :numeric
+    end
+  end
+end
+

data/lib/asciidoctor-bibliography.rb

@@ -0,0 +1,3 @@
+require 'byebug'
+
+require_relative 'asciidoctor-bibliography/asciidoctor'

data/lib/asciidoctor-bibliography/asciidoctor.rb

@@ -0,0 +1,17 @@
+require 'asciidoctor/extensions'
+
+require_relative 'asciidoctor/bibliographer_preprocessor'
+require_relative 'bibliographer'
+
+Asciidoctor::Extensions.register do
+  preprocessor AsciidoctorBibliography::Asciidoctor::BibliographerPreprocessor
+end
+
+module Asciidoctor
+  class Document
+    # All our document-level permanence passes through this attribute accessor.
+    def bibliographer
+      @bibliographer ||= AsciidoctorBibliography::Bibliographer.new
+    end
+  end
+end

data/lib/asciidoctor-bibliography/asciidoctor/bibliographer_preprocessor.rb

@@ -0,0 +1,68 @@
+require 'asciidoctor'
+
+require_relative '../helpers'
+require_relative '../formatters/csl'
+require_relative '../formatters/tex'
+require_relative '../database'
+require_relative '../citation'
+require_relative '../index'
+
+module AsciidoctorBibliography
+  module Asciidoctor
+    class BibliographerPreprocessor < ::Asciidoctor::Extensions::Preprocessor
+      def process document, reader
+
+        # We peek at the document attributes we need, without perturbing the parsing flow.
+        # NOTE: we're in a preprocessor and they haven't been parsed yet; doing it manually.
+        document_attributes =
+          ::Asciidoctor::Parser
+            .parse(reader, ::Asciidoctor::Document.new, header_only: true)
+            .attributes
+        # We extract only the ones we recognize.
+        document.bibliographer.options = Hash[Helpers.slice(document_attributes, 'bibliography-citation-style', 'bibliography-order', 'bibliography-reference-style', 'bibliography-database').map {|k, v| [k.sub(/^bibliography-/, ''), v] }]
+
+        # We're handling single database/formatters; generalization will be straightforward when needed.
+        document.bibliographer.database = Database.new(document.bibliographer.options['database'])
+        document.bibliographer.index_formatter = Formatters::CSL.new(document.bibliographer.options['reference-style'])
+        document.bibliographer.index_formatter.import document.bibliographer.database
+        document.bibliographer.citation_formatter = Formatters::TeX.new(document.bibliographer.options['citation-style'])
+        document.bibliographer.citation_formatter.import document.bibliographer.database
+
+        # Find, store and replace citations with uuids.
+        processed_lines = reader.read_lines.map do |line|
+          line.gsub(Citation::REGEXP) do
+            citation = Citation.new(*Regexp.last_match.captures)
+            document.bibliographer.add_citation(citation)
+            citation.uuid
+          end
+        end
+        reader = ::Asciidoctor::Reader.new processed_lines
+
+        # NOTE: retrieval and formatting are separated to allow sorting and numeric styles.
+
+        # Find and replace uuids with formatted citations.
+        processed_lines = reader.lines.join("\n") # for quicker matching
+        document.bibliographer.citations.each do |citation|
+          processed_lines.sub!(citation.uuid) do
+            citation.render document.bibliographer
+          end
+        end
+        processed_lines = processed_lines.lines.map(&:chomp)
+
+        reader = ::Asciidoctor::Reader.new processed_lines
+
+        # Find and format indices.
+        processed_lines = reader.read_lines.map do |line|
+          if line =~ Index::REGEXP
+            index = Index.new(*Regexp.last_match.captures)
+            index.render document.bibliographer
+          else
+            line
+          end
+        end
+        processed_lines.flatten!
+        reader = ::Asciidoctor::Reader.new processed_lines
+      end
+    end
+  end
+end

data/lib/asciidoctor-bibliography/bibliographer.rb

@@ -0,0 +1,31 @@
+module AsciidoctorBibliography
+  class Bibliographer
+    attr_accessor :citations
+    attr_accessor :indices
+    attr_accessor :database
+    attr_accessor :index_formatter
+    attr_accessor :citation_formatter
+    attr_reader :occurring_keys
+    attr_accessor :options
+
+    # NOTE: while database and formatter are singular, they're meant for future generalization.
+
+    def initialize
+      @options = {}
+      @citations = []
+      @indices = []
+      @database = nil
+      @index_formatter = nil
+      @citation_formatter = nil
+      @occurring_keys = []
+    end
+
+    def add_citation(citation)
+      citations << citation
+      @occurring_keys.concat(citation.keys).uniq!
+      citations.last.cites.each do |cite|
+        cite.occurrence_index = @occurring_keys.index(cite.key)
+      end
+    end
+  end
+end

data/lib/asciidoctor-bibliography/citation.rb

@@ -0,0 +1,73 @@
+require 'securerandom'
+require 'asciidoctor/attribute_list'
+
+module AsciidoctorBibliography
+  class Citation
+    TEX_MACROS_NAMES = Formatters::TeX::MACROS.keys.map { |s| Regexp.escape s }.concat(['fullcite']).join('|')
+    REGEXP = /\\?(#{TEX_MACROS_NAMES}):(?:(\S*?)?\[(|.*?[^\\])\])(?:\+(\S*?)?\[(|.*?[^\\])\])*/
+
+    # No need for a fully fledged class right now.
+    Cite = Struct.new(:key, :occurrence_index, :target, :positional_attributes, :named_attributes)
+
+    attr_reader :macro, :cites
+
+    def initialize(macro, *targets_and_attributes_list)
+      @uuid = SecureRandom.uuid
+      @macro = macro
+      @cites = []
+      targets_and_attributes_list.compact.each_slice(2).each do |target, attributes|
+        positional_attributes, named_attributes = # true, false
+          ::Asciidoctor::AttributeList.new(attributes).parse
+            .group_by { |hash_key, _| hash_key.is_a? Integer }
+            .values.map { |a| Hash[a] }
+        positional_attributes = positional_attributes.values
+        @cites << Cite.new(
+          positional_attributes.first,
+          nil,
+          target,
+          positional_attributes,
+          named_attributes
+        )
+      end
+    end
+
+    def render(bibliographer)
+      if macro == 'fullcite'
+        # NOTE: we reinstantiate to avoid tracking used keys.
+        formatter = Formatters::CSL.new(bibliographer.options['reference-style'])
+        formatter.import bibliographer.database
+        # TODO: as is, cites other than the first are simply ignored.
+        '{empty}' + Helpers.html_to_asciidoc(formatter.render(:bibliography, id: cites.first.key).join)
+      elsif Formatters::TeX::MACROS.keys.include? macro
+        bibliographer.citation_formatter.render(self)
+      end
+    end
+
+    def uuid
+      ":#{@uuid}:"
+    end
+
+    def keys
+      @cites.map { |h| h[:key] }
+    end
+
+    def xref(key, label)
+      "xref:#{self.render_id(key)}[#{label.gsub(']','\]')}]"
+    end
+
+    def render_id(key)
+      ['bibliography', key].compact.join('-')
+    end
+
+    private
+
+    def render_label(formatter, key)
+      formatter.render(:citation, id: key)
+    end
+
+    def render_xref(formatter, key)
+      "xref:#{render_id(key)}[#{render_label(formatter, key)}]"
+    end
+  end
+end
+