datafile 0.2.5 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: cf81810d2af02c7ccd64d4e103dc912737d1d585
-  data.tar.gz: 7d1b6dac27a44fce550bd203ed76cb6a6df1e32d
+  metadata.gz: e02f51529c40c3573f5bf51670b8e704e518306e
+  data.tar.gz: 309d29e81297fcf2a63b95a1c26f72003aa842b3
 SHA512:
-  metadata.gz: e43a593bc7dfebf016aff5c382f3032664988c672a7950d887dc4f901998754e1a384aa41d86e2d88b0cdc63caed1f957d59f6aafefa92659ab15f91028e8a61
-  data.tar.gz: bd225c4c8daa8261151e0568a77b3930fc9335089d22b0bd1d1325cbcfcadf8bb535d617768c933f58165cabd69524b0462e61db0a14282667eed7442c348ebb
+  metadata.gz: 77bf2afdca97946e52e99bb6f7b2ba96e714b5890a3be11df6f4a3a3b19e43281e30ee78002125928039535b6813c4ec362e1bd8ecf184fa0935e6487d7fa0e4
+  data.tar.gz: 41d975fe70d41f75e46031dd7ce006629375b9703dd9449199e6ee3fd3e6461b71db33618ac57720347a5c9fd99889f42caddf3c354bf72ea16aa05cad1a8180

data/Manifest.txt

@@ -1,30 +1,23 @@
-HISTORY.md
-Manifest.txt
-README.md
-Rakefile
-data/football.txt
-lib/datafile.rb
-lib/datafile/builder.rb
-lib/datafile/builder2.rb
-lib/datafile/datafile.rb
-lib/datafile/datasets/dataset.rb
-lib/datafile/version.rb
-lib/datafile/workers/dataset.rb
-lib/datafile/workers/file/dataset.rb
-lib/datafile/workers/file/registry.rb
-lib/datafile/workers/file/worker.rb
-lib/datafile/workers/zip/beer.rb
-lib/datafile/workers/zip/dataset.rb
-lib/datafile/workers/zip/football.rb
-lib/datafile/workers/zip/worker.rb
-lib/datafile/workers/zip/world.rb
-test/datafile/eurocup.rb
-test/datafile/stadiums.rb
-test/datafile/world.rb
-test/datafile2/at.rb
-test/helper.rb
-test/test_builder.rb
-test/test_builder2.rb
-test/test_file_dataset_registry.rb
-test/test_file_worker.rb
-test/test_football_dataset.rb
+CHANGELOG.md
+Manifest.txt
+README.md
+Rakefile
+data/football.txt
+lib/datafile.rb
+lib/datafile/builder.rb
+lib/datafile/datafile.rb
+lib/datafile/dataset.rb
+lib/datafile/version.rb
+lib/datafile/workers/file/dataset.rb
+lib/datafile/workers/file/registry.rb
+lib/datafile/workers/file/worker.rb
+lib/datafile/workers/zip/dataset.rb
+lib/datafile/workers/zip/worker.rb
+test/datafile/eurocup.rb
+test/datafile/stadiums.rb
+test/datafile/world.rb
+test/helper.rb
+test/test_builder.rb
+test/test_file_dataset_registry.rb
+test/test_file_worker.rb
+test/test_football_dataset.rb

data/Rakefile

@@ -15,17 +15,16 @@ Hoe.spec 'datafile' do
 
   # switch extension to .markdown for gihub formatting
   self.readme_file  = 'README.md'
-  self.history_file = 'HISTORY.md'
+  self.history_file = 'CHANGELOG.md'
 
   self.extra_deps = [
     ['logutils'],
-    ['textutils'],
     ['fetcher'],
   ]
 
   self.licenses = ['Public Domain']
 
   self.spec_extras = {
-   :required_ruby_version => '>= 1.9.2'
+    required_ruby_version: '>= 2.2.2'
   }
 end

data/data/football.txt

@@ -24,22 +24,31 @@ openfootball/confed-cup
 # -- Clubs
 
 openfootball/clubs
-
-openfootball/eng-england
-openfootball/at-austria
-openfootball/de-deutschland
-openfootball/es-espana
-openfootball/it-italy
-openfootball/fr-france
-openfootball/ch-confoederatio-helvetica
+openfootball/world
+
+openfootball/england
+openfootball/austria
+openfootball/deutschland
+openfootball/espana
+openfootball/italy
+openfootball/france
 openfootball/europe-champions-league
 
-openfootball/major-league-soccer
-openfootball/mx-mexico
+openfootball/mexico
 openfootball/north-america-champions-league
 
-openfootball/br-brazil
+openfootball/brazil
 openfootball/copa-sudamericana
 openfootball/copa-libertadores
 
 openfootball/club-world-cup
+
+
+
+############
+# in csv format
+
+footballcsv/england
+footballcsv/deutschland
+
+## todo/fix: add some more ...

data/lib/datafile.rb

@@ -1,39 +1,49 @@
 # encoding: utf-8
 
 require 'logutils'
+require 'fetcher'    ## add (HTTP/S) download support (of blobs/binaries/zips)
 
 
 # our own code
-
 require 'datafile/version'  # let it always go first
 
-require 'datafile/datasets/dataset'
 
-require 'datafile/workers/dataset'
+module Datafile
+
+  class Error < StandardError
+  end
+
+  ####
+  # todo/check:
+  # rename to DatasetNotFound or similar??
+  # use "common" error class - why? why not?
+  class DatasetNotFoundError < Error
+    attr_reader :message
+
+    def initialize( message )
+      @message = message
+    end
+
+    def to_s
+      "datset not found => #{@message}"
+    end
+  end
+end   # module Datafile
+
+
+require 'datafile/dataset'
+require 'datafile/builder'
+
 
-require 'datafile/workers/file/dataset'
 require 'datafile/workers/file/registry'
+require 'datafile/workers/file/dataset'
 require 'datafile/workers/file/worker'
 
 require 'datafile/workers/zip/dataset'
-require 'datafile/workers/zip/beer'
-require 'datafile/workers/zip/football'
-require 'datafile/workers/zip/world'
 require 'datafile/workers/zip/worker'
 
 require 'datafile/datafile'
-require 'datafile/builder'
-require 'datafile/builder2'
-
-
-module Datafile
-  def self.data_path
-    "#{root}/data"
-  end
-end # module Datafile
-
 
 
 # say hello
-puts Datafile.banner    if defined?($RUBYLIBS_DEBUG) && $RUBYLIBS_DEBUG
-
+puts Datafile.banner    if defined?($RUBYCOCO_DEBUG) && $RUBYCOCO_DEBUG

data/lib/datafile/builder.rb

@@ -5,7 +5,7 @@ module Datafile
 class Builder     ## "simple" builder (one file, one datafile)
 
   def self.load_file( path )
-    code = File.read_utf8( path )
+    code = File.open( path, 'r:utf-8' ).read
     self.load( code )
   end
 
@@ -18,36 +18,14 @@ class Builder     ## "simple" builder (one file, one datafile)
 
   include LogUtils::Logging
 
-  ## add models shortcuts/constants if exist
-  ##   todo/check: is there a better way ?
-  ##    e.g. just include in inline section/block - for example?
-  ##    check if already included? and check on load to include?
-  ##     models might not yet be required
-
-  if defined?( WorldDb ) && defined?( WorldDb::Models )
-    include WorldDb::Models
-  else
-    puts "*** sorry; can't include WorldDb::Models; not yet defined (required)"
-  end
-  ## todo/check: also include FootballDb::Models and BeerDb::Models ???
-
-
   def initialize
     @datafile = Datafile.new
   end
 
   attr_reader :datafile
 
-  ## "special" datasets
-
-  def inline( &block )
-    logger.info( "[builder] add inline script-block" )
-    @datafile.inlines << Inline.new( block )
-  end
-
 
   ## "classic" standard datasets
-
   def beer( name, opts={} )
     logger.info( "[builder] add beer-dataset '#{name}'" )
     @datafile.datasets << BeerDataset.new( name, opts )
@@ -65,4 +43,3 @@ class Builder     ## "simple" builder (one file, one datafile)
 
 end # class Builder
 end # module Datafile
-

data/lib/datafile/datafile.rb

@@ -2,52 +2,12 @@
 
 module Datafile
 
-class Script
-  include LogUtils::Logging
-
-  def initialize( proc )
-    @proc = proc
-  end
-
-  def call
-    logger.info( "[script] calling calc block" )
-    @proc.call
-  end
-  
-  def dump
-    puts "  script: #{@proc.inspect}"
-  end
-end  ## class Script
-
-
-### todo/check: use Script for Inline too?? - why, why not???
-###  - use setup/pre/before and post/after or something??
-##  - note: for now always is pre/before
-
-class Inline
-  include LogUtils::Logging
-
-  def initialize( proc )
-    @proc = proc
-  end
-
-  def call
-    logger.info( "[inline] calling script block" )
-    @proc.call
-  end
-
-  def dump
-    puts "  script: #{@proc.inspect}"
-  end
-end  ## class Inline
-
-
 
 class Datafile
 
   ## convenience method - use like Datafile.load_file()
   def self.load_file( path='./Datafile' )
-    code = File.read_utf8( path )
+    code = File.open( path, 'r:utf-8' ).read
     self.load( code )
   end
 
@@ -68,25 +28,34 @@ class Datafile
 
   include LogUtils::Logging
 
+
   def initialize( opts={} )
     @opts     = opts
     @datasets = []
-    @scripts  = []   ## calculation scripts (calc blocks)
-    @inlines  = []   ## inline (setup) scripts (run before reading datasets)
-
-    ## (target)name - return nil if noname (set/defined/assigned)
-    @name  = opts[:name] || nil
-    ## deps (dependencies) - note: always returns an array (empty array if no deps)
-    @deps  = opts[:deps] || []     
-
-    if opts[:file]
-      @worker  = FileWorker.new( self )
-    else
-      ## default to zip worker for now
-      @worker   = ZipWorker.new( self )
-    end
+
+    @worker = if opts[:file]
+                FileWorker.new( self )
+              else   ## default to zip worker for now
+                ZipWorker.new( self )
+              end
+  end
+
+  attr_reader :datasets
+
+
+  def worker=( value )  # lets you change worker - find a better way - how, why, why not??
+    @worker =  if value.is_a?( Class )   ## let's you pass in FileWorker or ZipWorker etc.
+                 value.new( self )
+               elsif value.to_sym == :file
+                 FileWorker.new( self )
+               elsif value.to_sym == :zip
+                 ZipWorker.new( self )
+               else
+                 value
+               end
   end
 
+
   def guess_file_or_zip_worker   ## change/rename to configure_file_or_zip_worker - why? why not??
     ## if opts file or zip exists do NOT change (assume set manually)
     return  if @opts[:file] || @opts[:zip]
@@ -99,20 +68,10 @@ class Datafile
   end
 
 
-  attr_reader   :datasets
-  attr_reader   :scripts    ## calc(ulation) scripts (calc blocks)
-  attr_reader   :inlines    ## inline script blocks  -- use before?? run before datasets
-  attr_reader   :name
-  attr_reader   :deps       ## dep(endencies)
-
-  attr_accessor :worker      # lets you change worker - find a better way - how, why, why not??
-
-
   def run
     logger.info( "[datafile] begin - run" )
     download     # step 1 - download zips for datasets
     read         # step 2 - read in datasets from zips  - note: includes running inlines
-    calc         # step 3 - run calc(ulations) scripts
     logger.info( "[datafile] end - run" )
   end
 
@@ -120,7 +79,7 @@ class Datafile
   def download
     logger.info( "[datafile] dowload" )
     @worker.download
-    ## check: use @worker.download( @datasets) - why, why not??  link worker w/ datafile - why, why not??  
+    ## check: use @worker.download( @datasets) - why, why not??  link worker w/ datafile - why, why not??
   end
 
   def read
@@ -128,89 +87,11 @@ class Datafile
     @worker.read
   end
 
-  def calc
-    logger.info( "[datafile] calc" )
-    @worker.calc
-  end
-
   def dump
     ## for debugging dump datasets (note: will/might also check if zip exits)
     logger.info( "[datafile] dump datasets (for debugging)" )
     @worker.dump
   end
 
-
-=begin
-  def download_world   ## only dl world datasets (skip all others)
-    logger.info( "[datafile] dowload world datasets" )
-    @datasets.each do |dataset|
-      if dataset.kind_of? WorldDataset
-        dataset.download()
-      else
-        # skip all others
-      end
-    end
-  end
-
-  def download_beer   ## only dl beer datasets (skip all others)
-    logger.info( "[datafile] dowload beer datasets" )
-    @datasets.each do |dataset|
-      if dataset.kind_of? BeerDataset
-        dataset.download()
-      else
-        # skip all others
-      end
-    end
-  end
-
-  def download_football   ## only dl football datasets (skip all others)
-    logger.info( "[datafile] dowload football datasets" )
-    @datasets.each do |dataset|
-      if dataset.kind_of? FootballDataset
-        dataset.download()
-      else
-        # skip all others
-      end
-    end
-  end
-=end
-
-
-=begin
-  def read_world
-    logger.info( "[datafile] read world datasets" )
-    @datasets.each do |dataset|
-      if dataset.kind_of?( WorldDataset )
-        dataset.read()
-      else
-        # skip all others
-      end
-    end
-  end
-
-  def read_beer
-    logger.info( "[datafile] read beer datasets" )
-    @datasets.each do |dataset|
-      if dataset.kind_of?( BeerDataset )
-        dataset.read()
-      else
-        # skip all others
-      end
-    end
-  end
-
-  def read_football
-    logger.info( "[datafile] read football datasets" )
-    @datasets.each do |dataset|
-      if dataset.kind_of?( FootballDataset )
-        dataset.read()
-      else
-        # skip all others
-      end
-    end
-  end
-=end
-
 end # class Datafile
 end # module Datafile
-