datafile 0.3.0 → 0.3.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: e02f51529c40c3573f5bf51670b8e704e518306e
-  data.tar.gz: 309d29e81297fcf2a63b95a1c26f72003aa842b3
+  metadata.gz: a70f59e6054e3fae9f97e113bd1ec672b02b2cf8
+  data.tar.gz: fef94a40bf6e7d190ec9a2605de4c2469307710c
 SHA512:
-  metadata.gz: 77bf2afdca97946e52e99bb6f7b2ba96e714b5890a3be11df6f4a3a3b19e43281e30ee78002125928039535b6813c4ec362e1bd8ecf184fa0935e6487d7fa0e4
-  data.tar.gz: 41d975fe70d41f75e46031dd7ce006629375b9703dd9449199e6ee3fd3e6461b71db33618ac57720347a5c9fd99889f42caddf3c354bf72ea16aa05cad1a8180
+  metadata.gz: b1c96fcf41d2aa5664aff47255976638827b2b34f959f41f941d1272b567622027c3d0bb8658af71f17e5a27ce96f58b95a47b93fcfd52985b4d2f4bce0b3607
+  data.tar.gz: b4ed10d3b5aace8c699bcdf5a8039c9cfaeae25f88c7f80e8ef622badc84c43d0f59ad57b9c9c71be91a21864955ccc8846d3f58ed5548880e72a9ae328d17b5

data/lib/datafile/datafile.rb

@@ -19,9 +19,6 @@ class Datafile
     # Note: return datafile (of course, NOT the builder)
     #  if you want a builder use Datafile::Builder ;-)
     datafile = builder.datafile
-    ## check for auto-configure (just guessing)
-    ##   zip or file worker
-    datafile.guess_file_or_zip_worker
     datafile
   end
 
@@ -56,18 +53,6 @@ class Datafile
   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]
-
-    ## for now only change if single (just 1) dataset and it's present
-    if @datasets.size == 1 && @datasets[0].file?
-      puts "  bingo!! assume (in-situ) datafile; use file workers"
-      @worker = FileWorker.new( self )
-    end
-  end
-
-
   def run
     logger.info( "[datafile] begin - run" )
     download     # step 1 - download zips for datasets

data/lib/datafile/dataset.rb

@@ -32,35 +32,9 @@ class Dataset
   ## convenience helpers for known opts
   def setup()   @opts[:setup];           end    ## note: return nil if not found/set
   def format()  @opts[:format] || 'txt'; end    ## note: assume default is txt (other formats incl. csv) for now - why? wh not?
-
-
-  def file?                    # note: use file? (not exit? might use zip? later to check if zip exists? -why? why not?)
-    ## hack/convenience shortcut:
-    ##   1) check if dataset exists as local (in situ) file dataset
-    ##    e.g. working folder  == name of dataset
-    ##
-    ##   todo/fix: 2) also check
-    ##                   via (file)registry - why, why not??
-
-    ## split name in org/user + project (e.g. openfootball/at-austria)
-    parts = @name.split( '/' )
-
-    basename = parts[1]
-    ##  e.g.
-    ##  ./ (working folder) => at-austria
-    ##  openfootball/at-austria
-    if File.basename( Dir.getwd ) == basename
-      puts "  bingo!! working folder >#{basename}< matches dataset"
-      true     ## return true
-    else
-      false    ## return false
-    end
-  end  ## file?
-
 end # class Dataset
 
 
-
 class WorldDataset < Dataset
   def initialize( name_easy, opts={} )
 
@@ -129,6 +103,11 @@ class BeerDataset < Dataset
     super( name, opts )
   end
 end  # class BeerDataset
+end # module Datafile
 
 
-end # module Datafile
+## note:  add "known" dataset classes to global namespace - why? why not?
+##  add "global" shortcuts for now
+WorldDataset    = Datafile::WorldDataset
+FootballDataset = Datafile::FootballDataset
+BeerDataset     = Datafile::BeerDataset

data/lib/datafile/version.rb

@@ -4,7 +4,7 @@ module Datafile
 
   MAJOR = 0    ## todo: namespace inside version or something - why? why not??
   MINOR = 3
-  PATCH = 0
+  PATCH = 1
   VERSION = [MAJOR,MINOR,PATCH].join('.')
 
   def self.version

data/lib/datafile/workers/file/dataset.rb

@@ -18,18 +18,7 @@ class FileDataset
   end
 
   def repo_dir     ### check: use (rename to) include dir (or local_repo_dir) - why, why not ???
-    ## note: for easy testing allow "in situ" datasets
-    ##   e.g.  ./ (e.g. mu-mauritius)  is openfootball/mu-mauritius
-    ## split name in org/user + project (e.g. openfootball/at-austria)
-    parts = @dataset.name.split( '/' )
-
-    basename = parts[1]
-    if File.basename( Dir.getwd ) == basename
-      puts "  bingo!! working folder >#{basename}< matches dataset"
-      return Dir.getwd     ## assume working directory/folder is repo dir
-    end
-
-    registry.lookup( @dataset.name )
+    self.class.registry.lookup( @dataset.name )
   end
 
   def dump
@@ -48,9 +37,5 @@ class FileDataset
       logger.info( "TODO/FIX: read dataset (file) '#{@dataset.name}', '#{@dataset.setup}'; sorry" )
     end
   end
-
-
-private
-  def registry()  self.class.registry;  end    ## convenience method to access "static" shared class variable
 end # class FileDataset
 end # module Datafile

data/test/helper.rb

@@ -6,8 +6,7 @@ require 'minitest/autorun'
 
 ### require 'worlddb'
 require 'sportdb/models'   # note: will include worlddb
-
+require 'sportdb/readers'    ## pulls in SportDb::ZipPackage or SportDb::DirPackage
 
 ## our own code
 require 'datafile'
-

data/test/test_builder.rb

@@ -10,7 +10,7 @@ require 'helper'
 class TestBuilder < MiniTest::Test
 
   def test_builder
-    code =<<EOS
+    code =<<TXT
 ## comments
 
   world  'world.db', setup: 'countries'
@@ -24,15 +24,15 @@ class TestBuilder < MiniTest::Test
 
 ## more comments
 
-EOS
+TXT
 
     builder = Datafile::Builder.load( code )
 
     datafile = builder.datafile
-    datafile.run
+    ## datafile.run
 
     datafile.download
-    datafile.read
+    ## datafile.read
     datafile.dump
 
     ## change worker (defaults to ZipWorker)

data/test/test_football_dataset.rb

@@ -10,26 +10,33 @@ require 'helper'
 class TestFootballDataset < MiniTest::Test
 
   def test_names
-    pp Datafile::FootballDataset.known_datasets
+    pp FootballDataset.known_datasets
 
-    at   = Datafile::FootballDataset.new( 'austria' )
-    at2  = Datafile::FootballDataset.new( 'openfootball/austria' )
-    xxx  = Datafile::FootballDataset.new( 'xxx' )
-    xxx2 = Datafile::FootballDataset.new( 'openfootball/xxx' )
+    at   = FootballDataset.new( 'austria' )
+    at2  = FootballDataset.new( 'openfootball/austria' )
+    xxx  = FootballDataset.new( 'xxx' )
+    xxx2 = FootballDataset.new( 'openfootball/xxx' )
 
     assert_equal 'openfootball/austria', at.name
     assert_equal 'openfootball/austria', at2.name
 
     assert_equal 'openfootball/xxx', xxx.name
     assert_equal 'openfootball/xxx', xxx2.name
+
+    ## check csv format
+    at   = FootballDataset.new( 'austria', format: 'csv' )
+    at2  = FootballDataset.new( 'footballcsv/austria' )
+
+    assert_equal 'footballcsv/austria', at.name
+    assert_equal 'footballcsv/austria', at2.name
   end
 
   ## move to test datasets - why, why not ??
   def test_world
-    at   = Datafile::WorldDataset.new( 'austria.db' )
-    at2  = Datafile::WorldDataset.new( 'openmundi/austria.db' )
-    xxx  = Datafile::WorldDataset.new( 'xxx' )
-    xxx2 = Datafile::WorldDataset.new( 'openmundi/xxx' )
+    at   = WorldDataset.new( 'austria.db' )
+    at2  = WorldDataset.new( 'openmundi/austria.db' )
+    xxx  = WorldDataset.new( 'xxx' )
+    xxx2 = WorldDataset.new( 'openmundi/xxx' )
 
     assert_equal 'openmundi/austria.db', at.name
     assert_equal 'openmundi/austria.db', at2.name
@@ -39,10 +46,10 @@ class TestFootballDataset < MiniTest::Test
   end
 
   def test_beer
-    at   = Datafile::BeerDataset.new( 'austria' )
-    at2  = Datafile::BeerDataset.new( 'openbeer/austria' )
-    xxx  = Datafile::BeerDataset.new( 'xxx' )
-    xxx2 = Datafile::BeerDataset.new( 'openbeer/xxx' )
+    at   = BeerDataset.new( 'austria' )
+    at2  = BeerDataset.new( 'openbeer/austria' )
+    xxx  = BeerDataset.new( 'xxx' )
+    xxx2 = BeerDataset.new( 'openbeer/xxx' )
 
     assert_equal 'openbeer/austria', at.name
     assert_equal 'openbeer/austria', at2.name

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: datafile
 version: !ruby/object:Gem::Version
-  version: 0.3.0
+  version: 0.3.1
 platform: ruby
 authors:
 - Gerald Bauer
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-11-21 00:00:00.000000000 Z
+date: 2019-11-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: logutils