sportdb 0.9.7 → 1.0.0

data/lib/sportdb/data/fixtures.rb

@@ -0,0 +1,205 @@
+# encoding: utf-8
+
+def find_football_db_path_from_gemfile_gitref!
+  puts "[debug] find_football_db_path..."
+  puts "load path:"
+  pp $LOAD_PATH
+  
+  candidates = []
+  $LOAD_PATH.each do |path|
+    if path =~ /\/(football\.db-[a-z0-9]+)|(football\.db)\//
+      candidates << path.dup
+    end
+  end
+  
+  puts "found candidates:"
+  pp candidates
+  
+  ## cut-off everything after /football.db
+  # e.g. lib/ruby/gems/1.9.1/bundler/gems/football.db-38279c414449/lib becomes
+  #      lib/ruby/gems/1.9.1/bundler/gems/football.db-38279c414449
+  
+  cand = candidates[0]
+  
+  puts "cand before: #{cand}"
+  
+  ## nb: *? is non-greedy many operator
+  
+  ## todo: why not just cut off trailing /lib - is it good enough??
+  # it's easier
+  
+  regex = /(\/football\.db.*?)(\/.*)/
+  cand = cand.sub( regex ) do |_|
+    puts "cutting off >>#{$2}<<"
+    $1
+  end
+  
+  puts "cand after: #{cand}"
+  
+  ## todo:exit with error if not found!!!
+  
+  cand
+end
+
+
+
+module SportDB
+  
+  module Fixtures   # use shortcut/alias Fx?
+  
+  ## todo/ check: remove _FIXTURES and use module namespace instead?
+  
+  
+  SEASONS = [
+    'seasons'
+  ]
+  
+  LEAGUES = [
+    'leagues',
+    'leagues_club'
+  ]
+  
+  ## todo/check: put teams into its own module/namespace (remove _TEAMS?)
+  EUROPE_TEAMS = [
+    'europe/teams'
+  ]
+
+  ## NB: copa america requires jpn as team invitee (from asia)
+  AMERICA_TEAMS = [
+    'world/teams',
+    'america/teams'
+  ]
+  
+  WORLD_TEAMS = [
+    'world/teams',
+    'america/teams',
+    'europe/teams'
+  ]
+
+  AT_TEAMS = [ 'at/teams', 'at/teams2' ]
+  DE_TEAMS = [ 'de/teams' ]
+  EN_TEAMS = [ 'en/teams' ]
+  ES_TEAMS = [ 'es/teams' ]
+  FR_TEAMS = [ 'fr/teams' ]
+  IT_TEAMS = [ 'it/teams' ]
+  RO_TEAMS = [ 'ro/teams' ]
+
+  CLUB_EUROPE_TEAMS =
+    AT_TEAMS +
+    DE_TEAMS +
+    EN_TEAMS +
+    ES_TEAMS +
+    FR_TEAMS +
+    IT_TEAMS + 
+    RO_TEAMS +
+   [
+    'club/europe/teams'
+   ]
+  
+  
+  AR_TEAMS = [ 'ar/teams' ]
+  BR_TEAMS = [ 'br/teams' ]
+  MX_TEAMS = [ 'mx/teams' ]
+  
+  CLUB_AMERICA_TEAMS = 
+    AR_TEAMS +
+    BR_TEAMS +
+    MX_TEAMS +
+  [
+    'club/america/teams_c',
+    'club/america/teams_n',
+    'club/america/teams_s',
+  ]
+  
+  AR_FIXTURES = []
+
+  ### todo: ? get event_key automatically from event_reader ?? why? why not??
+  BR_FIXTURES = [
+    ['br.2013', 'br/2013/cb' ]
+  ]
+
+  MX_FIXTURES = [
+    ['mx.apertura.2012.2', 'mx/2012/apertura' ],
+    ['mx.clausura.2013.1', 'mx/2013/clausura' ]
+  ]
+
+
+  AT_FIXTURES = [
+    ['at.2011/12',     'at/2011_12/bl' ],
+    ['at.cup.2011/12', 'at/2011_12/cup' ],
+    ['at.2012/13',     'at/2012_13/bl', 'at/2012_13/bl2'],
+    ['at.cup.2012/13', 'at/2012_13/cup']
+  ]
+  
+  DE_FIXTURES = [
+    ['de.2012/13',     'de/2012_13/bl' ]
+  ]
+  
+  EN_FIXTURES = [
+    ['en.2012/13',     'en/2012_13/pl' ]
+  ]
+  
+  RO_FIXTURES = [
+    ['ro.2012/13',     'ro/2012_13/l1' ]
+  ]
+  
+  
+  EUROPE_FIXTURES = [
+    ['euro.2008', 'europe/2008/euro'],
+    ['euro.2012', 'europe/2012/euro'],
+  ]
+    
+  AMERICA_FIXTURES = [
+   ['america.copa.2011', 'america/2011/copa'],
+   ['america.gold.2011', 'america/2011/gold'],
+   ['america.gold.2013', 'america/2013/gold'],
+   ['america.copa.2015', 'america/2015/copa'],
+  ]
+  
+  WORLD_FIXTURES = [
+   ['world.conf.2009', 'world/2009/conf'],
+   ['world.2010',      'world/2010/cup'],
+   ['world.quali.america.2014', 'world/2014/quali_america'],
+   ['world.quali.europe.c.2014', 'world/2014/quali_europe_c'],
+   ['world.quali.europe.2014', 'world/2014/quali_europe', 'world/2014/quali_europe_c', 'world/2014/quali_europe_i'],
+   ['world.2014',      'world/2014/cup'],
+  ]
+  
+  CLUB_EUROPE_FIXTURES = [
+    ['cl.2011/12',     'club/europe/2011_12/cl'],
+    ['el.2011/12',     'club/europe/2011_12/el'],
+    ['cl.2012/13',     'club/europe/2012_13/cl'],
+    ['el.2012/13',     'club/europe/2012_13/el'],
+  ]
+  
+  CLUB_AMERICA_FIXTURES = [
+   ['america.cl.2011/12',     'club/america/2011_12/cl'],
+   ['copa.libertadores.2012', 'club/america/2012/libertadores' ],
+   ['copa.sud.2012',          'club/america/2012/sud'],
+   ['america.cl.2012/13',     'club/america/2012_13/cl'],
+   ['copa.libertadores.2013', 'club/america/2013/libertadores' ]
+  ]
+  
+  def self.all
+    SEASONS + 
+    LEAGUES +
+    CLUB_AMERICA_TEAMS +
+    AR_FIXTURES +
+    BR_FIXTURES +
+    MX_FIXTURES +
+    CLUB_AMERICA_FIXTURES +
+    CLUB_EUROPE_TEAMS +
+    AT_FIXTURES +
+    DE_FIXTURES +
+    EN_FIXTURES +
+    RO_FIXTURES +
+    CLUB_EUROPE_FIXTURES +
+    WORLD_TEAMS +
+    AMERICA_FIXTURES +
+    EUROPE_FIXTURES +
+    WORLD_FIXTURES
+  end  # method all
+
+  end # module Fixtures
+  
+end # module SportDB

data/lib/sportdb/{keys.rb → data/keys.rb}

@@ -1,8 +1,15 @@
+# encoding: utf-8
+
+### todo/fix: move to sportdb-data gem/plugin/addon ??
+
+
+### fix: rename to ::Key (singular) - why? why not??
 
 module SportDB::Keys
 
     module EventKeys
   # use constants for known keys; lets us define aliases (if things change)
+
   
   AT_2011_12     = 'at.2011/12'
   AT_2012_13     = 'at.2012/13'
@@ -15,8 +22,8 @@ module SportDB::Keys
 
   WORLD_2010     = 'world.2010'
   
-  WORLD_QUALI_EURO_2012_13    = 'world.quali.euro.2012/13'
-  WORLD_QUALI_AMERICA_2012_13 = 'world.quali.america.2012/13'
+  WORLD_QUALI_EUROPE_2014  = 'world.quali.europe.2014'
+  WORLD_QUALI_AMERICA_2014 = 'world.quali.america.2014'
 
   ############################
   ## NB: see db/leagues.rb for keys in use

data/lib/sportdb/data/models.rb

@@ -0,0 +1,43 @@
+# encoding: utf-8
+
+module SportDB::Models
+
+  ## add convenience finders to some model classes
+  
+  class Event
+
+  include SportDB::Keys::EventKeys
+
+  def self.find_at_2012_13!
+    self.find_by_key!( AT_2012_13 )
+  end
+  
+  def self.find_at_cup_2012_13!
+    self.find_by_key!( AT_CUP_2012_13 )
+  end
+  
+  def self.find_cl_2012_13!
+    self.find_by_key!( CL_2012_13 )
+  end
+  
+  def self.find_euro_2012!
+    self.find_by_key!( EURO_2012 )
+  end
+  
+  def self.find_world_2010!
+    self.find_by_key!( WORLD_2010 )
+  end
+  
+  def self.find_world_quali_europe_2014!
+    self.find_by_key!( WORLD_QUALI_EUROPE_2014 )
+  end
+  
+  def self.find_world_quali_america_2014!
+    self.find_by_key!( WORLD_QUALI_AMERICA_2014 )
+  end
+    
+  end # class Event
+
+end # module SportDB::Models
+
+

data/lib/sportdb/deleter.rb

@@ -0,0 +1,27 @@
+
+module SportDB
+
+  class Deleter
+    ######
+    # NB: make models available in sportdb module by default with namespace
+    #  e.g. lets you use Team instead of Models::Team 
+    include SportDB::Models
+
+    def run
+      # for now delete all tables
+      
+      Team.delete_all
+      Game.delete_all
+      Event.delete_all
+      EventTeam.delete_all
+      Group.delete_all
+      GroupTeam.delete_all
+      Round.delete_all
+      Badge.delete_all
+      League.delete_all
+      Season.delete_all
+    end
+    
+  end # class Deleter
+  
+end # module SportDB

data/lib/sportdb/lang.rb

@@ -0,0 +1,265 @@
+# encoding: utf-8
+
+module SportDB
+
+class Lang
+
+  def initialize( logger=nil )
+
+    @lang = 'en'   # make default lang english/en
+
+    # load word lists
+
+    @fixtures_en = YAML.load( File.read_utf8( "#{SportDB.config_path}/fixtures.en.yml" ))
+    @fixtures_de = YAML.load( File.read_utf8( "#{SportDB.config_path}/fixtures.de.yml" ))
+    @fixtures_es = YAML.load( File.read_utf8( "#{SportDB.config_path}/fixtures.es.yml" ))
+
+    @fixtures = {
+      'en' => @fixtures_en,
+      'de' => @fixtures_de,
+      'es' => @fixtures_es,
+    }
+
+
+    @words_en = fixtures_hash_to_words_ary( @fixtures_en )
+    @words_de = fixtures_hash_to_words_ary( @fixtures_de )
+    @words_es = fixtures_hash_to_words_ary( @fixtures_es )
+    
+    puts "en - #{@words_en.size} words: #{@words_en}"
+    puts "de - #{@words_de.size} words: #{@words_de}"
+    puts "es - #{@words_es.size} words: #{@words_es}"
+    
+  end
+
+  attr_reader :words_en
+  attr_reader :words_de
+  attr_reader :words_es
+  attr_reader :lang
+
+  def lang=(value)
+    puts "setting lang to #{value}"
+    
+    if @lang != value
+      # reset cached values on language change
+      puts "reseting cached lang values (lang changed from #{@lang} to #{value})"
+
+      @group = nil
+      @round = nil
+      @knockout_round = nil
+      @leg1 = nil
+      @leg2 = nil
+      
+      @regex_group = nil
+      @regex_round = nil
+      @regex_knockout_round = nil
+      @regex_leg1 = nil
+      @regex_leg2 = nil
+    end
+    
+    @lang = value
+    
+  end
+
+
+  def group
+    @group ||= group_getter
+  end
+  
+  def round
+    @round ||= round_getter
+  end
+ 
+  def knockout_round
+    @knock_round ||= knockout_round_getter
+  end
+  
+  def leg1
+    @leg1 ||= leg1_getter
+  end
+  
+  def leg2
+    @leg2 ||= leg2_getter
+  end
+
+
+  def regex_group
+    @regex_group ||= regex_group_getter
+  end
+   
+  def regex_round
+    @regex_round ||= regex_round_getter
+  end
+  
+  def regex_knockout_round
+    @regex_knockout_round ||= regex_knockout_round_getter
+  end
+  
+  def regex_leg1
+    @regex_leg1 ||= regex_leg1_getter
+  end
+  
+  def regex_leg2
+    @regex_leg2 ||= regex_leg2_getter
+  end
+  
+  
+  
+private
+  def group_getter
+    h = @fixtures[ lang ]
+    values = h['group']
+    values
+  end
+  
+  def round_getter
+    # e.g. Spieltag|Runde|Achtelfinale|Viertelfinale|Halbfinale|Finale
+    
+    h = @fixtures[ lang ]
+    values = h['round']
+    values << "|" << h['matchday']  ## todo/check: fold round n matchday into one key? why? why not??
+    
+    ### add knockout rounds values too
+    values << "|" << h['round32']
+    values << "|" << h['round16']
+    values << "|" << h['quarterfinals']
+    values << "|" << h['semifinals']
+    values << "|" << h['thirdplace']
+    values << "|" << h['final']
+    values
+  end
+
+  def leg1_getter
+    h = @fixtures[ lang ]
+    values = h['leg1']
+    values
+  end
+
+  def leg2_getter
+    h = @fixtures[ lang ]
+    values = h['leg2']
+    values
+  end
+
+  def knockout_round_getter
+    h = @fixtures[ lang ]
+    values = h['round32']
+    values << "|" << h['round16']
+    values << "|" << h['quarterfinals']
+    values << "|" << h['semifinals']
+    values << "|" << h['thirdplace']
+    values << "|" << h['final']
+    values
+  end
+  
+  def regex_group_getter
+    ## todo: escape for regex?
+    /#{group}/
+  end
+  
+  def regex_round_getter
+    ## todo: escape for regex?
+    ## todo: sort by length - biggest words go first? does regex match biggest word automatically?? - check
+    /#{round}/
+  end
+
+  def regex_knockout_round_getter
+    ## todo: escape for regex?
+    /#{knockout_round}/
+  end
+  
+  def regex_leg1_getter
+    ## todo: escape for regex?
+    /#{leg1}/
+  end
+
+  def regex_leg2_getter
+    ## todo: escape for regex?
+    /#{leg2}/
+  end
+
+
+  def fixtures_hash_to_words_ary( hash )
+    ary = []
+    hash.each do |key_wild, values_wild|
+      key     = key_wild.to_s.strip
+      values  = values_wild.to_s.strip
+       
+      puts "processing key >>#{key}<< with words >>#{values}<<"
+       
+      ary += values.split('|')
+    end
+    ary
+  end
+  
+end # class Lang
+
+
+class LangChecker
+
+  def initialize( logger=nil )
+    ## add logger here
+  end
+
+  def analyze( name, include_path )
+    # return lang code e.g. en, de, es
+    
+    path = "#{include_path}/#{name}.txt"
+
+    puts "*** parsing data '#{name}' (#{path})..."
+
+    text = File.read_utf8( path )
+    
+    ### todo/fix:
+    # remove comment lines and end of line comments from text
+    
+    en = count_words_in_text( SportDB.lang.words_en, text )
+    de = count_words_in_text( SportDB.lang.words_de, text )
+    es = count_words_in_text( SportDB.lang.words_es, text )
+      
+    lang_counts = [
+      [ 'en', en ],
+      [ 'de', de ],
+      [ 'es', es ]
+    ]
+    
+    # sort by word count (reverse sort e.g. highest count goes first)
+    lang_counts = lang_counts.sort {|l,r| r[1] <=> l[1] }
+    
+    # dump stats
+    
+    puts "****************************************"
+    lang_counts.each_with_index do |item,index|
+      ## e.g. 1. en: 20 words
+      ##      2. de: 2 words
+      puts "#{index+1}. #{item[0]}: #{item[1]}"
+    end
+    
+    ## return lang code w/ highest count
+    lang_counts[0][0]
+  end
+
+private
+  def count_word_in_text( word, text )
+    count = 0
+    pos = text.index( word )
+    while pos.nil? == false
+      count += 1
+      puts "bingo - found >>#{word}<< on pos #{pos}, count: #{count}"
+      ### todo: check if pos+word.length/size needs +1 or similar
+      pos = text.index( word, pos+word.length)
+    end
+    count
+  end
+  
+  def count_words_in_text( words, text )
+    count = 0
+    words.each do |word|
+      count += count_word_in_text( word, text )
+    end
+    count
+  end
+
+end # class LangChecker
+
+
+end # module SportDB

data/lib/sportdb/models/event.rb

@@ -73,40 +73,6 @@ class Event < ActiveRecord::Base
     
     known_teams
   end # method known_teams_table
-  
-  ###########################
-  ## convenience helpers
-
-  include SportDB::Keys::EventKeys
-
-  def self.find_at_2012_13!
-    self.find_by_key!( AT_2012_13 )
-  end
-  
-  def self.find_at_cup_2012_13!
-    self.find_by_key!( AT_CUP_2012_13 )
-  end
-  
-  def self.find_cl_2012_13!
-    self.find_by_key!( CL_2012_13 )
-  end
-  
-  def self.find_euro_2012!
-    self.find_by_key!( EURO_2012 )
-  end
-  
-  def self.find_world_2010!
-    self.find_by_key!( WORLD_2010 )
-  end
-  
-  def self.find_world_quali_euro_2012_13!
-    self.find_by_key!( WORLD_QUALI_EURO_2012_13 )
-  end
-  
-  def self.find_world_quali_america_2012_13!
-    self.find_by_key!( WORLD_QUALI_AMERICA_2012_13 )
-  end
-
 
 end # class Event