# TODO: 
#   - DOI fetching?
#   - Protect math expressions with braces
#   - 


require "bibtex"

module BibtexMunge
  @@split_fields = [                          # specifying how to handle {PART 1: PART 2}
                                              # in fields where it is a frequent issue:
    [:title, :title, :subtitle],              #  * if the title contains a separator, 
                                              #    the second part becomes a subtitle
    [:booktitle, :booktitle, :booksubtitle],  #  * if the booktitle contains a separator,
                                              #    the second part becomes a booksubtitle
    [:publisher, :address, :publisher]        #  * if the publisher contains a separator,
                                              #    the first part becomes an address
  ]
  @@name_fields = [:author, :editor]          # specifying which fields shouldn't have
                                              # trailing periods removed (because
                                              # they might be initials)

  @@prune_fields = [:isbn, :issn, :abstract]  # common nuisance fields to delete
 
  # Run a block on every field in every entry
  def each_field(options = {})
    except   = options[:except]
    except ||= []
    if block_given?
      each_entry do |e|
        e.fields.each do |k,v|
          yield v unless except.include? k
        end
      end
    end
  end

  def normalize_initials!
    each_entry do |e|
      @@name_fields.each do |k|
        if e[k]
          e[k].each do |name|
            if name.first
              name.first = name.normalize_initials
            end
          end
        end
      end
    end
    self
  end

  def depunctuate!
    each_field :except => @@name_fields do |f|
      f.chomp!(".")
    end
    self
  end
 
  def extend_braces!
    each_field do |f|
      f.gsub! /\ {([^ ])}([^ ]+)/, ' {\1\2}'  # The initial space is to avoid
                                              # fucking up braces after an accent.
                                              # this is *probably* redundant now that we're
                                              # unicodifying everything, but better safe than
                                              # sorry
    end
    self
  end

  def remove_double_braces!
    each_field do |f|
      f.gsub! /{{(.*)}}/, '{\1}'    # Eliminate overt double braces

      f.gsub! /^{(.*)}$/, '\1'      # Eliminate whole-entry braces (which would have been
                                    # double braces in the unparsed BibTeX file, since
                                    #       publisher = {{Blackwell}}
                                    # parses to 
                                    #       @bib[:key].publisher = "{Blackwell}"
    end
    self
  end

  def split_field!(e, oldkey, newsuperkey, newsubkey)
    if e[oldkey]
      elements = e[oldkey].split(/([.:?]) /,2)    # Split after .:?, retaining the separator
      if elements.length == 3                       
        e[newsubkey] = elements[2]                     
        if elements[1] == "?"                       # Keep the separator as part of the first
          e[newsuperkey] = elements[0..1].join      # field if it is a ?, to handle titles of
                                                    # the form {Question? An Answer}
        else
          e[newsuperkey] = elements[0]
        end
      end
    end
  end

  def split_fields!
    each_entry do |e|
      @@split_fields.each do |oldkey, newsuperkey, newsubkey|
        split_field!(e, oldkey, newsuperkey, newsubkey)
      end
    end
    self
  end

  def prune_fields!
    each_entry do |e|
      @@prune_fields.each do |k|
        e.delete(k)
      end
    end
    self
  end

  def normalize_page_ranges!
    each_entry do |e|
      if e.pages
        e.pages.gsub! /^(\d+)\s*(–|—|-|--|---)\s*(\d+)$/, '\1 -- \3'
      end
    end
    self
  end

  def fix_everything!
    remove_double_braces!
    depunctuate!
    extend_braces!
    split_fields!
    prune_fields!
    normalize_initials!
    normalize_page_ranges!
  end
end