cskit 1.0.1 → 1.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: b63a117f2cc1e418414654ab99412b9a70355c05
-  data.tar.gz: 59293273440f24aea1a23c80c6b96d845d601a12
+  metadata.gz: fc2a9550e46d7df48af95443aa4bd70b8a6d9f51
+  data.tar.gz: 2c74a6f0feefde08a17af8ba7e7113e9bb97e5fa
 SHA512:
-  metadata.gz: cd54044d09b723ccfd504dd87d9a44283df32ddd56905a40df1b0e36314368b04388c343e2d9190b175f2f786ecf0e89e2e824488fb72ff11bf808f2238d9b3f
-  data.tar.gz: 204c2eec966c773dfabb68a3e70a2778ef7811691238dbc26a01ce66f0bbf55651b943e6eebf4c1fd39ddda6f883027ad1889d980526a2fed6570e2ce3616ee0
+  metadata.gz: fa92477f03276a97d9e77964b5eb2b9ea295870d810c97037166e31a091d2fbd808227325949febebdd4efef5894344d127bb3639fbb6fd237f806fec49cba73
+  data.tar.gz: eaad5970540813c5a6f489fa60668b15476bed1ae281d092d46898bb8a29e22766d687c400d5b2b67d0b55bbd27d4886d6b199105a98cd455319df83b8fb486e

data/Gemfile

@@ -1,8 +1,11 @@
-source "http://rubygems.org"
+source 'http://rubygems.org'
 
 gemspec
 
-group :development do
-  gem "pry"
-  gem "pry-nav"
+group :test do
+  gem 'rspec'
+end
+
+group :development, :test do
+  gem 'pry-byebug'
 end

data/History.txt

@@ -6,3 +6,8 @@
 
 * Fix preface regex to detect pages starting at vii instead of vi for Science and Health.
 
+== 1.1.0
+
+* Parser rewrite.
+* Add parser tests.
+* General cleanup.

data/cskit.gemspec

@@ -6,20 +6,17 @@ require 'cskit/version'
 Gem::Specification.new do |s|
   s.name     = "cskit"
   s.version  = ::CSKit::VERSION
-  s.authors  = ["Cameron Dutro"]
-  s.email    = ["camertron@gmail.com"]
-  s.homepage = "http://github.com/camertron"
+  s.authors  = ['Cameron Dutro']
+  s.email    = ['camertron@gmail.com']
+  s.homepage = 'http://github.com/camertron'
 
-  s.description = s.summary = "Christian Science citation library for Ruby."
+  s.description = s.summary = 'Christian Science citation library for Ruby.'
 
   s.platform = Gem::Platform::RUBY
   s.has_rdoc = true
 
   s.add_dependency 'json'
-  s.add_dependency 'treetop'
-
-  s.add_development_dependency 'rake'
 
   s.require_path = 'lib'
-  s.files = Dir["{lib,spec,resources}/**/*", "Gemfile", "History.txt", "LICENSE", "README.md", "Rakefile", "cskit.gemspec"]
+  s.files = Dir['{lib,spec,resources}/**/*', 'Gemfile', 'History.txt', 'LICENSE', 'README.md', 'Rakefile', 'cskit.gemspec']
 end

data/lib/cskit.rb

@@ -1,73 +1,22 @@
 # encoding: UTF-8
 
-require 'cskit/formatters'
-require 'cskit/lesson'
-require 'cskit/parsers'
-require 'cskit/readers'
-require 'cskit/volume'
-require 'cskit/resources/volumes'
-
-require 'treetop'
-require 'json'
-
 module CSKit
 
-  autoload :Annotator,       "cskit/annotator"
-  autoload :Annotation,      "cskit/annotator"
-  autoload :AnnotatedString, "cskit/annotated_string"
-
-  class << self
-
-    def register_volume(config)
-      register(:volume, config[:id], config[:volume].new(config))
-    end
-
-    def register_annotator(config)
-      register(:annotator, config[:id], config[:annotator].new(config))
-    end
-
-    def get_volume(volume_id)
-      get(:volume, volume_id)
-    end
-
-    def get_annotator(annotator_id)
-      get(:annotator, annotator_id)
-    end
-
-    def volume_available?(volume_id)
-      available?(:volume, volume_id)
-    end
-
-    def annotator_available?(annotator_id)
-      available?(:annotator, annotator_id)
-    end
-
-    private
-
-    def register(family, key, obj)
-      (registry[family.to_sym] ||= {})[key.to_sym] = obj
-    end
-
-    def get(family, key)
-      registry[family.to_sym][key.to_sym] || get_for_type(family, key)
-    end
-
-    def get_for_type(family, type)
-      found = registry[family.to_sym].find do |key, obj|
-        obj.config[:type] == type
-      end
-
-      found.last if found
-    end
-
-    def available?(family, key)
-      !!registry[family.to_sym][key.to_sym] rescue false
-    end
+  autoload :Annotator,       'cskit/annotator'
+  autoload :Annotation,      'cskit/annotator'
+  autoload :AnnotatedString, 'cskit/annotated_string'
+  autoload :Formatters,      'cskit/formatters'
+  autoload :Lesson,          'cskit/lesson'
+  autoload :Parsers,         'cskit/parsers'
+  autoload :Readers,         'cskit/readers'
+  autoload :Registry,        'cskit/registry'
+  autoload :RegistryFactory, 'cskit/registry'
+  autoload :Volume,          'cskit/volume'
+  autoload :Volumes,         'cskit/resources/volumes'
 
-    def registry
-      @registry ||= {}
-    end
+  extend Registry
 
-  end
+  family :volume, :volumes
+  family :annotator, :annotators
 
-end
+end

data/lib/cskit/annotated_string.rb

@@ -70,4 +70,4 @@ module CSKit
     end
 
   end
-end
+end

data/lib/cskit/annotator.rb

@@ -2,15 +2,13 @@
 
 module CSKit
   class Annotator
-
     attr_reader :config
 
     def initialize(config)
       @config = config
     end
-
   end
 
   Annotation = Struct.new(:start, :finish, :data)
 
-end
+end

data/lib/cskit/formatters.rb

@@ -3,9 +3,8 @@
 module CSKit
   module Formatters
 
-    autoload :Bible,         "cskit/formatters/bible"
-    autoload :ScienceHealth, "cskit/formatters/science_health"
-
+    autoload :Bible,         'cskit/formatters/bible'
+    autoload :ScienceHealth, 'cskit/formatters/science_health'
 
     class Formatter
       attr_reader :options
@@ -16,4 +15,4 @@ module CSKit
     end
 
   end
-end
+end

data/lib/cskit/formatters/bible.rb

@@ -3,10 +3,9 @@
 module CSKit
   module Formatters
     module Bible
-
-      autoload :BiblePlainTextFormatter, "cskit/formatters/bible/bible_plain_text_formatter"
-      autoload :BibleHtmlFormatter,      "cskit/formatters/bible/bible_html_formatter"
-
+      autoload :BibleHtmlFormatter,      'cskit/formatters/bible/bible_html_formatter'
+      autoload :BibleJsonFormatter,      'cskit/formatters/bible/bible_json_formatter'
+      autoload :BiblePlainTextFormatter, 'cskit/formatters/bible/bible_plain_text_formatter'
     end
   end
-end
+end

data/lib/cskit/formatters/bible/bible_html_formatter.rb

@@ -5,7 +5,7 @@ module CSKit
     module Bible
       class BibleHtmlFormatter < BiblePlainTextFormatter
 
-        protected
+        private
 
         def format_verse_texts(texts, verse)
           texts.each_with_index.map do |text, index|
@@ -38,4 +38,4 @@ module CSKit
       end
     end
   end
-end
+end

data/lib/cskit/formatters/bible/bible_json_formatter.rb

@@ -0,0 +1,17 @@
+# encoding: UTF-8
+
+module CSKit
+  module Formatters
+    module Bible
+      class BibleJsonFormatter < CSKit::Formatters::Formatter
+
+        def format_readings(readings)
+          readings.map do |reading|
+            reading.to_hash
+          end
+        end
+
+      end
+    end
+  end
+end

data/lib/cskit/formatters/bible/bible_plain_text_formatter.rb

@@ -21,7 +21,7 @@ module CSKit
           texts.join(separator)
         end
 
-        protected
+        private
 
         def join_readings(readings)
           join(
@@ -78,7 +78,7 @@ module CSKit
         end
 
         def format_starter(text, pos)
-          pos ? "..." + text : text
+          pos ? '...' + text : text
         end
 
         def terminator_position(text, terminator)
@@ -114,4 +114,4 @@ module CSKit
       end
     end
   end
-end
+end

data/lib/cskit/formatters/science_health.rb

@@ -3,10 +3,8 @@
 module CSKit
   module Formatters
     module ScienceHealth
-
-      autoload :ScienceHealthPlainTextFormatter, "cskit/formatters/science_health/science_health_plain_text_formatter"
-      autoload :ScienceHealthHtmlFormatter,      "cskit/formatters/science_health/science_health_html_formatter"
-
+      autoload :ScienceHealthPlainTextFormatter, 'cskit/formatters/science_health/science_health_plain_text_formatter'
+      autoload :ScienceHealthHtmlFormatter,      'cskit/formatters/science_health/science_health_html_formatter'
     end
   end
-end
+end

data/lib/cskit/formatters/science_health/science_health_html_formatter.rb

@@ -5,10 +5,9 @@ module CSKit
     module ScienceHealth
       class ScienceHealthHtmlFormatter < ScienceHealthPlainTextFormatter
 
-        # _word_
         ITALICS_REGEX = /_([a-zA-Z0-9\-\.,\"\'\?]+)_/
 
-        protected
+        private
 
         def format_lines(lines, citation_line)
           detect_italics(super)
@@ -23,10 +22,10 @@ module CSKit
         end
 
         def separator
-          options[:separator] || "<br />"
+          options[:separator] || '<br />'
         end
 
       end
     end
   end
-end
+end

data/lib/cskit/formatters/science_health/science_health_plain_text_formatter.rb

@@ -8,7 +8,8 @@ module CSKit
         # semicolon, question mark, or period
         SENTENCE_TERMINATOR_REGEX = /[;\?\.]/
 
-        # either a period + space, quotes, or start of line followed by the first capital letter or number.
+        # either a period + space, quotes, or start of line followed by the
+        # first capital letter or number.
         SENTENCE_START_REGEX = /(\.\s+|\.\"\s+|\.\'\s+|\?\s+|\!\s+|^)[A-Z0-9\"\']/
 
         def format_readings(readings)
@@ -23,7 +24,7 @@ module CSKit
           texts.join(separator)
         end
 
-        protected
+        private
 
         def format_lines(lines, citation_line)
           lines.each_with_index.map do |line, line_index|
@@ -37,7 +38,7 @@ module CSKit
                 matches = line_text.match(SENTENCE_START_REGEX)
                 if matches && matches.length == 2
                   offset = matches.offset(1)
-                  spaces = " " * offset.first  # indent to match physical position in S&H book
+                  spaces = ' ' * offset.first  # indent to match physical position in S&H book
                   spaces + line_text[matches.offset(1).last..-1]
                 else
                   line_text
@@ -70,4 +71,4 @@ module CSKit
       end
     end
   end
-end
+end

data/lib/cskit/lesson.rb

@@ -2,9 +2,7 @@
 
 module CSKit
   module Lesson
-
-    autoload :Lesson,  "cskit/lesson/lesson"
-    autoload :Section, "cskit/lesson/section"
-
+    autoload :Lesson,  'cskit/lesson/lesson'
+    autoload :Section, 'cskit/lesson/section'
   end
-end
+end

data/lib/cskit/lesson/lesson.rb

@@ -17,8 +17,8 @@ module CSKit
         def from_json(data)
           sections = JSON.parse(data).map do |section_data|
             Section.new(
-              section_data["section"],
-              section_data["readings"].inject({}) do |ret, (k, v)|
+              section_data['section'],
+              section_data['readings'].inject({}) do |ret, (k, v)|
                 ret[k.to_sym] = v; ret
               end
             )
@@ -69,4 +69,4 @@ module CSKit
 
     end
   end
-end
+end

data/lib/cskit/lesson/section.rb

@@ -23,4 +23,4 @@ module CSKit
 
     end
   end
-end
+end

data/lib/cskit/parsers.rb

@@ -2,7 +2,10 @@
 
 module CSKit
   module Parsers
-    autoload :ScienceHealthParser, "cskit/parsers/science_health/science_health"
-    autoload :BibleParser,         "cskit/parsers/bible/bible"
+    autoload :Bible,         'cskit/parsers/bible'
+    autoload :Parser,        'cskit/parsers/parser'
+    autoload :ScienceHealth, 'cskit/parsers/science_health'
+    autoload :Token,         'cskit/parsers/token'
+    autoload :Tokenizer,     'cskit/parsers/tokenizer'
   end
-end
+end

data/lib/cskit/parsers/bible.rb

@@ -0,0 +1,10 @@
+# encoding: UTF-8
+
+module CSKit
+  module Parsers
+    module Bible
+      autoload :BibleParser,    'cskit/parsers/bible/bible_parser'
+      autoload :BibleTokenizer, 'cskit/parsers/bible/bible_tokenizer'
+    end
+  end
+end

data/lib/cskit/parsers/bible/bible_parser.rb

@@ -0,0 +1,192 @@
+# encoding: UTF-8
+
+module CSKit
+  module Parsers
+    module Bible
+
+      DEFAULT_CARDINALITY = 1
+
+      Citation = Struct.new(:book, :chapter_list) do
+        def to_s
+          "#{book} #{chapter_list.map(&:to_s).join('; ')}"
+        end
+
+        def to_hash
+          {
+            book: book,
+            chapters: chapter_list.map(&:to_hash)
+          }
+        end
+      end
+
+      Chapter = Struct.new(:chapter_number, :verse_list) do
+        def to_s
+          "#{chapter_number}:#{verse_list.map(&:to_s).join(', ')}"
+        end
+
+        def to_hash
+          {
+            chapter_number: chapter_number,
+            verses: verse_list.map(&:to_hash)
+          }
+        end
+      end
+
+      Verse = Struct.new(:start, :finish, :starter, :terminator) do
+        def to_s
+          str = if start == finish
+            start.to_s
+          else
+            "#{start}-#{finish}"
+          end
+
+          str << " #{starter}" if starter
+          str << " (to #{terminator})" if terminator
+          str
+        end
+
+        def to_hash
+          {
+            start: start,
+            finish: finish,
+            starter: starter ? starter.to_hash : nil,
+            terminator: terminator ? terminator.to_hash : nil
+          }
+        end
+      end
+
+      Positional = Struct.new(:cardinality, :fragment) do
+        def to_s
+          card_s = case cardinality
+            when 1 then '1st'
+            when 2 then '2nd'
+            when 3 then '3rd'
+          end
+
+          if cardinality
+            "#{card_s} #{fragment}"
+          else
+            fragment
+          end
+        end
+
+        def to_hash
+          {
+            cardinality: cardinality,
+            fragment: fragment
+          }
+        end
+      end
+
+
+      class BibleParser < CSKit::Parsers::Parser
+        def entry_point
+          Citation.new(book, chapter_list)
+        end
+
+        private
+
+        def get_token_stream
+          BibleTokenizer.new(citation_text).each_token.lazy
+        end
+
+        def book
+          current.value.tap { next_token(:text) }
+        end
+
+        def chapter_list
+          [].tap do |list|
+            loop do
+              list << chapter
+
+              case current.type
+                when :semicolon
+                  next_token(:semicolon)
+                else
+                  break
+              end
+
+              break if eos?
+            end
+          end
+        end
+
+        def chapter
+          chap_num = chapter_number
+          next_token(:colon)
+          vlist = verse_list
+
+          Chapter.new(chap_num, vlist)
+        end
+
+        def chapter_number
+          current.value.tap { next_token(:number) }.to_i
+        end
+
+        def verse_list
+          [].tap do |list|
+            loop do
+              list << verse
+
+              case current.type
+                when :comma
+                  next_token(:comma)
+                else
+                  break
+              end
+
+              break if eos?
+            end
+          end
+        end
+
+        def verse
+          start = current.value.tap { next_token(:number) }.to_i
+          finish = start
+          starter = nil
+
+          if current.type == :dash
+            next_token(:dash)
+            finish = current.value.tap { next_token(:number) }.to_i
+          end
+
+          starter = verse_starter
+          terminator = verse_terminator
+
+          Verse.new(start, finish, starter, terminator)
+        end
+
+        def verse_starter
+          case current.type
+            when :text, :cardinality
+              positional
+          end
+        end
+
+        def verse_terminator
+          if current.type == :left_paren
+            next_token(:left_paren)
+            next_token(:to)
+            positional.tap { next_token(:right_paren) }
+          end
+        end
+
+        def positional
+          card = cardinality
+          fragment = current.value
+          next_token(:text, :semicolon, :colon, :comma)
+          Positional.new(card, fragment)
+        end
+
+        def cardinality
+          if current.type == :cardinality
+            current.value.tap { next_token(:cardinality) }.to_i
+          else
+            DEFAULT_CARDINALITY
+          end
+        end
+      end
+
+    end
+  end
+end