sportdb-formats 0.2.1 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: '08b90c09533628f176c7c08778634c557898b5ca'
-  data.tar.gz: 434a9655f2d11614016322bb7829b975866aa9ca
+  metadata.gz: 4ebc60eb9651185cf5aa0ba1c7707fc8678f7803
+  data.tar.gz: 67d24f30f4a8586f54ea73135fd58239420e1e22
 SHA512:
-  metadata.gz: 6dc601d34ae8d5bb66b181cb09228d40792c2af2a360b82203f711fdfae309353bf1d3f01b79f5c0d4d41e0a64b0d69709193b553065950364a7d4e7bf5a72d3
-  data.tar.gz: a583d31ff5fd7d6201c7f85499b448e0f73111a71d679511ca13b82d798ba3a12a40a57be42a821760a3dbafeafe9ab16a8c48b7d46c7f582df27c4f43de175c
+  metadata.gz: 6c2fdcdf34f4ea95ee765d1b1cfedd9aef84787836fb7fb2c21c10d4000e3b8417109f5ec9ce4d5ae552a0a63d5e91a9c976847fd33b175cccad5aa4d5ea4762
+  data.tar.gz: 4ef8c4ba7538c738e2c44b6f16aafbdc11d640485ecf99a603a764c069b038e290ea320918ec16ecd3405f321e6b4bba5f8b350c6835c5f3a70808954da93350

data/Manifest.txt

@@ -9,13 +9,22 @@ lib/sportdb/formats/outline_reader.rb
 lib/sportdb/formats/package.rb
 lib/sportdb/formats/scores.rb
 lib/sportdb/formats/season_utils.rb
+lib/sportdb/formats/structs/club.rb
+lib/sportdb/formats/structs/match.rb
+lib/sportdb/formats/structs/matchlist.rb
+lib/sportdb/formats/structs/season.rb
+lib/sportdb/formats/structs/standings.rb
+lib/sportdb/formats/structs/team_usage.rb
 lib/sportdb/formats/version.rb
 test/helper.rb
+test/test_club_helpers.rb
+test/test_clubs.rb
 test/test_csv_reader.rb
 test/test_datafile.rb
 test/test_datafile_match.rb
 test/test_goals.rb
+test/test_match.rb
 test/test_outline_reader.rb
 test/test_package.rb
 test/test_scores.rb
-test/test_season_utils.rb
+test/test_season.rb

data/Rakefile

@@ -20,7 +20,7 @@ Hoe.spec 'sportdb-formats' do
   self.licenses = ['Public Domain']
 
   self.extra_deps = [
-    ['alphabets',     '>= 0.1.0'],
+    ['alphabets',     '>= 0.1.3'],
     ['date-formats',  '>= 0.2.4'],
     ['csvreader',     '>= 1.2.4'],
     ['sportdb-langs', '>= 0.0.1'],

data/lib/sportdb/formats/season_utils.rb

@@ -2,110 +2,21 @@
 
 
 module SeasonHelper ## use Helpers why? why not?
-  def prev( season )
-    ## todo: add 1964-1965 format too!!!
-    if season =~ /^(\d{4})-(\d{2})$/    ## season format is  1964-65
-      fst = $1.to_i - 1
-      snd = (100 + $2.to_i - 1) % 100    ## note: add 100 to turn 00 => 99
-      "%4d-%02d" % [fst,snd]
-    elsif season =~ /^(\d{4})$/
-      fst = $1.to_i - 1
-      "%4d" % [fst]
-    else
-      puts "*** !!!! wrong season format >>#{season}<<; exit; sorry"
-      exit 1
-    end
-  end  # method prev
 
+  ##############################################
+  ### deprecated!!! use new Season class!!!
+  ##   this code will get removed!!!!
+  ###################################################
 
-  def key( basename )
-    if basename =~ /^(\d{4})[\-\/](\d{4})$/      ## e.g. 2011-2012 or 2011/2012 => 2011/12
-      "%4d/%02d" % [$1.to_i, $2.to_i % 100]
-    elsif basename =~ /^(\d{4})[\-\/](\d{2})$/   ## e.g. 2011-12 or 2011/12  => 2011/12
-      "#{$1}/#{$2}"
-    elsif basename =~ /^(\d{4})$/
-      $1
-    else
-      puts "*** !!!! wrong season format >>#{basename}<<; exit; sorry"
-      exit 1
-    end
-  end  # method key
+  def prev( str )   SportDb::Import::Season.new( str ).prev;  end
+  def key( str )    SportDb::Import::Season.new( str ).key;   end
+  def directory( str, format: nil )  SportDb::Import::Season.new( str ).directory( format: format );  end
 
-
-  def directory( season, format: nil )
-    ##  todo: find better names for formats - why? why not?:
-    ##    long | archive | decade(?)   =>   1980s/1988-89, 2010s/2017-18, ...
-    ##    short | std(?)   =>   1988-89, 2017-18, ...
-
-    ## convert season name to "standard" season name for directory
-
-    ## todo/fix: move to parse / validate season (for (re)use)!!!! - why? why not?
-    if season =~ /^(\d{4})[\-\/](\d{4})$/   ## e.g. 2011-2012 or 2011/2012 => 2011-12
-      years = [$1.to_i, $2.to_i]
-    elsif season =~ /^(\d{4})[\-\/](\d{2})$/   ## e.g. 2011-12 or 2011/12  => 2011-12
-      years = [$1.to_i, $1.to_i+1]
-      ## note: check that season end year is (always) season start year + 1
-      if ($1.to_i+1) % 100 != $2.to_i
-        puts "*** !!!! wrong season format >>#{season}<<; season end year MUST (always) equal season start year + 1; exit; sorry"
-        exit 1
-      end
-    elsif season =~ /^(\d{4})$/
-      years = [$1.to_i]
-    else
-      puts "*** !!!! wrong season format >>#{season}<<; exit; sorry"
-      exit 1
-    end
-
-
-    if ['l', 'long', 'archive' ].include?( format.to_s )   ## note: allow passing in of symbol to e.g. 'long' or :long
-      if years.size == 2
-        "%3d0s/%4d-%02d" % [years[0] / 10, years[0], years[1] % 100]   ## e.g. 2000s/2001-02
-      else  ## assume size 1 (single year season)
-        "%3d0s/%4d" % [years[0] / 10, years[0]]
-      end
-    else  ## default 'short' format / fallback
-      if years.size == 2
-        "%4d-%02d" % [years[0], years[1] % 100]   ## e.g. 2001-02
-      else  ## assume size 1 (single year season)
-        "%4d" % years[0]
-      end
-    end
-  end # method directory
-
-
-
-  def start_year( season ) ## get start year
-    ## convert season name to "standard" season name for directory
-
-    ## todo/check:  just return year from first for chars - keep it simple - why? why not?
-    if season =~ /^(\d{4})[\-\/](\d{4})$/   ## e.g. 2011-2010 or 2011/2011 => 2011-10
-      $1
-    elsif season =~ /^(\d{4})[\-\/](\d{2})$/
-      $1
-    elsif season =~ /^(\d{4})$/
-      $1
-    else
-      puts "*** !!!! wrong season format >>#{season}<<; exit; sorry"
-      exit 1
-    end
-  end
-
-  def end_year( season ) ## get end year
-    ## convert season name to "standard" season name for directory
-    if season =~ /^(\d{4})[\-\/](\d{4})$/   ## e.g. 2011-2010 or 2011/2011 => 2011-10
-      $2
-    elsif season =~ /^(\d{4})[\-\/](\d{2})$/
-      ## note: assume second year is always +1
-      ##    todo/fix: add assert/check - why? why not?
-      ## eg. 1999-00 => 2000 or 1899-00 => 1900
-      ($1.to_i+1).to_s
-    elsif season =~ /^(\d{4})$/
-      $1
-    else
-      puts "*** !!!! wrong season format >>#{season}<<; exit; sorry"
-      exit 1
-    end
-  end
+  ## note: new start_year now returns an integer number (no longer a string)!!!
+  def start_year( str )  SportDb::Import::Season.new( str ).start_year;  end
+  ## note: new end_year now returns an integer number (no longer a string)!!!
+  ##       if now end_year (year? == true) than returns nil (no longer the start_year "as fallback")!!!
+  def end_year( str )   SportDb::Import::Season.new( str ).end_year;  end
 end  # module SeasonHelper
 
 

data/lib/sportdb/formats/structs/club.rb

@@ -0,0 +1,221 @@
+# encoding: utf-8
+
+module SportDb
+  module Import
+
+##
+#  note: use our own (internal) club struct for now - why? why not?
+#    - check that shape/structure/fields/attributes match
+#      the Team struct in sportdb-text (in SportDb::Struct::Team)  !!!!
+
+
+## more attribs - todo/fix - also add "upstream" to struct & model!!!!!
+#   district, geos, year_end, country, etc.
+
+
+
+
+class Club
+
+  def self.create( **kwargs )
+    new.update( kwargs )
+  end
+
+  def update( **kwargs )
+    @name        = kwargs[:name]       if kwargs.has_key?( :name )
+    @alt_names   = kwargs[:alt_names]  if kwargs.has_key?( :alt_names )
+    @city        = kwargs[:city]       if kwargs.has_key?( :city )
+    ## todo/fix:  use city struct - why? why not?
+    ## todo/fix: add country too  or report unused keywords / attributes - why? why not?
+
+    self   ## note - MUST return self for chaining
+  end
+
+
+  ##  todo: use just names for alt_names - why? why not?
+  attr_accessor :key, :name, :alt_names,
+                :code,    ## code == abbreviation e.g. ARS etc.
+                :year, :year_end,   ## todo/fix: change year_end to end_year (like in season)!!!
+                :ground
+
+
+  alias_method :title, :name  ## add alias/compat - why? why not
+
+  def names
+    ## todo/check: add alt_names_auto too? - why? why not?
+    [@name] + @alt_names
+  end   ## all names
+
+
+  ## special import only attribs
+  attr_accessor :alt_names_auto    ## auto-generated alt names
+  attr_accessor :wikipedia   # wikipedia page name (for english (en))
+
+
+  def historic?()  @year_end ? true : false; end
+  alias_method  :past?, :historic?
+
+
+  attr_accessor :a, :b
+  def a?()  @a == nil; end  ## is a (1st) team / club (i)?            if a is NOT set
+  def b?()  @a != nil; end  ## is b (2nd/reserve/jr) team / club (ii) if a is     set
+
+  ## note: delegate/forward all geo attributes for team b for now (to team a) - keep - why? why not?
+  attr_writer  :city, :district, :country, :geos
+  def city()      @a == nil ?  @city     : @a.city;     end
+  def district()  @a == nil ?  @district : @a.district; end
+  def country()   @a == nil ?  @country  : @a.country;  end
+  def geos()      @a == nil ?  @geos     : @a.geos;     end
+
+
+  def wikipedia?()  @wikipedia; end
+  def wikipedia_url
+    if @wikipedia
+      ##  note: replace spaces with underscore (-)
+      ##  e.g. Club Brugge KV => Club_Brugge_KV
+      ##  todo/check/fix:
+      ##    check if "plain" dash (-) needs to get replaced with typographic dash??
+      "https://en.wikipedia.org/wiki/#{@wikipedia.gsub(' ','_')}"
+    else
+      nil
+    end
+  end
+
+
+  def initialize
+    @alt_names      = []
+    @alt_names_auto = []
+  end
+
+
+  ## helper methods for import only
+  ## check for duplicates
+  def duplicates?
+    names = [name] + alt_names + alt_names_auto
+    names = names.map { |name| normalize( sanitize(name) ) }
+
+    names.size != names.uniq.size
+  end
+
+  def duplicates
+    names = [name] + alt_names + alt_names_auto
+
+    ## calculate (count) frequency and select if greater than one
+    names.reduce( Hash.new ) do |h,name|
+       norm = normalize( sanitize(name) )
+       h[norm] ||= []
+       h[norm] << name; h
+    end.select { |norm,names| names.size > 1 }
+  end
+
+  def add_variants( name_or_names )
+    names = name_or_names.is_a?(Array) ? name_or_names : [name_or_names]
+    names.each do |name|
+      name = sanitize( name )
+      self.alt_names_auto += variants( name )
+    end
+  end
+
+
+   ###################################
+   # "global" helper - move to ___ ? why? why not?
+
+   ## note: allow placeholder years to e.g. (-___) or (-????)
+   ##    for marking missing (to be filled in) years
+   YEAR_REGEX = /\([0-9, ?_-]+?\)/    # note: non-greedy (minimum/first) match
+   def self.strip_year( name )
+     ## check for year(s) e.g. (1887-1911), (-2013),
+     ##                        (1946-2001, 2013-) etc.
+     name.gsub( YEAR_REGEX, '' ).strip
+   end
+
+   def self.has_year?( name ) name =~ YEAR_REGEX; end
+
+   LANG_REGEX = /\[[a-z]{1,2}\]/   ## note also allow [a] or [d] or [e] - why? why not?
+   def self.strip_lang( name )
+     name.gsub( LANG_REGEX, '' ).strip
+   end
+
+   def self.has_lang?( name ) name =~ LANG_REGEX; end
+
+  def self.sanitize( name )
+    ## check for year(s) e.g. (1887-1911), (-2013),
+    ##                        (1946-2001,2013-) etc.
+    name = strip_year( name )
+    ## check lang codes e.g. [en], [fr], etc.
+    name = strip_lang( name )
+    name
+  end
+
+
+  ## note: also add (),’,−  etc. e.g.
+  ##   Estudiantes (LP) => Estudiantes LP
+  ##   Saint Patrick’s Athletic FC => Saint Patricks Athletic FC
+  ##   Myllykosken Pallo −47 => Myllykosken Pallo 47
+
+  NORM_REGEX =  %r{
+                    [.'’º/()−-]
+                  }x   # note: in [] dash (-) if last doesn't need to get escaped
+  ## note: remove all dots (.), dash (-), ', º, /, etc.
+  #   .  U+002E (46) - FULL STOP
+  #   '  U+0027 (39) - APOSTROPHE
+  #   ’  U+2019 (8217) - RIGHT SINGLE QUOTATION MARK
+  #   º  U+00BA (186) - MASCULINE ORDINAL INDICATOR
+  #   /  U+002F (47) - SOLIDUS
+  #   (  U+0028 (40) - LEFT PARENTHESIS
+  #   )  U+0029 (41) - RIGHT PARENTHESIS
+  #   −  U+2212 (8722) - MINUS SIGN
+  #   -  U+002D (45) - HYPHEN-MINUS
+
+  ##         for norm(alizing) names
+  def self.strip_norm( name )
+    name.gsub( NORM_REGEX, '' )
+  end
+
+  def self.normalize( name )
+    # note: do NOT call sanitize here (keep normalize "atomic" for reuse)
+    name = strip_norm( name )
+    name = name.gsub( ' ', '' )  # note: also remove all spaces!!!
+
+    ## todo/fix: use our own downcase - why? why not?
+    name = downcase_i18n( name )     ## do NOT care about upper and lowercase for now
+    name
+  end
+
+
+  def self.strip_wiki( name )     # todo/check: rename to strip_wikipedia_en - why? why not?
+    ## note: strip disambiguationn qualifier from wikipedia page name if present
+    ##        note: only remove year and foot... for now
+    ## e.g. FC Wacker Innsbruck (2002) => FC Wacker Innsbruck
+    ##      Willem II (football club)  => Willem II
+    ##
+    ## e.g. do NOT strip others !! e.g.
+    ##   América Futebol Clube (MG)
+    ##  only add more "special" cases on demand (that, is) if we find more
+    name = name.gsub( /\([12][^\)]+?\)/, '' ).strip  ## starting with a digit 1 or 2 (assuming year)
+    name = name.gsub( /\(foot[^\)]+?\)/, '' ).strip  ## starting with foot (assuming football ...)
+    name
+  end
+
+
+private
+  ## private "shortcut" convenience helpers
+  def sanitize( name )    self.class.sanitize( name ); end
+  def normalize( name )   self.class.normalize( name ); end
+
+  def variants( name )  Variant.find( name ); end
+end # class Club
+
+
+
+############
+#  convenience
+#   Club and Team are for now alias
+#     in the future make
+#        Club > Team
+#        NationalTeam > Team  - why? why not?
+Team = Club
+
+
+end   # module Import
+end   # module SportDb

data/lib/sportdb/formats/structs/match.rb

@@ -0,0 +1,131 @@
+# encoding: utf-8
+
+##
+# note: add all "former" structs to the SportDb::Import module / namespace
+
+module SportDb
+  module Import
+
+
+class Match
+
+  def self.create( **kwargs )    ## keep using create why? why not?
+    new.update( kwargs )
+  end
+
+  def update( **kwargs )
+    ## note: check with has_key?  because value might be nil!!!
+    @date     = kwargs[:date]     if kwargs.has_key? :date
+    @team1    = kwargs[:team1]    if kwargs.has_key? :team1
+    @team2    = kwargs[:team2]    if kwargs.has_key? :team2
+    @conf1    = kwargs[:conf1]    if kwargs.has_key? :conf1
+    @conf2    = kwargs[:conf2]    if kwargs.has_key? :conf2
+    @country1 = kwargs[:country1]  if kwargs.has_key? :country1
+    @country2 = kwargs[:country2]  if kwargs.has_key? :country2
+
+    ## note: round is a string!!!  e.g. '1', '2' for matchday or 'Final', 'Semi-final', etc.
+    ##   todo: use to_s - why? why not?
+    @round    = kwargs[:round]    if kwargs.has_key? :round
+    @stage    = kwargs[:stage]    if kwargs.has_key? :stage
+    @leg      = kwargs[:leg]      if kwargs.has_key? :leg
+    @group    = kwargs[:group]    if kwargs.has_key? :group
+    @comments = kwargs[:comments] if kwargs.has_key? :comments
+
+
+    @score1     = kwargs[:score1]      if kwargs.has_key? :score1
+    @score1i    = kwargs[:score1i]     if kwargs.has_key? :score1i
+    @score1et   = kwargs[:score1et]    if kwargs.has_key? :score1et
+    @score1p    = kwargs[:score1p]     if kwargs.has_key? :score1p
+    @score1agg  = kwargs[:score1agg]   if kwargs.has_key? :score1agg
+
+    @score2     = kwargs[:score2]      if kwargs.has_key? :score2
+    @score2i    = kwargs[:score2i]     if kwargs.has_key? :score2i
+    @score2et   = kwargs[:score2et]    if kwargs.has_key? :score2et
+    @score2p    = kwargs[:score2p]     if kwargs.has_key? :score2p
+    @score2agg  = kwargs[:score2agg]   if kwargs.has_key? :score2agg
+
+    ## note: (always) (auto-)convert scores to integers
+    @score1     = @score1.to_i      if @score1
+    @score1i    = @score1i.to_i     if @score1i
+    @score1et   = @score1et.to_i    if @score1et
+    @score1p    = @score1p.to_i     if @score1p
+    @score1agg  = @score1agg.to_i   if @score1agg
+
+    @score2     = @score2.to_i      if @score2
+    @score2i    = @score2i.to_i     if @score2i
+    @score2et   = @score2et.to_i    if @score2et
+    @score2p    = @score2p.to_i     if @score2p
+    @score2agg  = @score2agg.to_i   if @score2agg
+
+
+    ## todo/fix:
+    ##  gr-greece/2014-15/G1.csv:
+    ##     G1,10/05/15,Niki Volos,OFI,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,
+    ##
+
+    ##  for now score1 and score2 must be present
+    if @score1.nil? || @score2.nil?
+      puts "*** missing scores for match:"
+      pp kwargs
+      ## exit 1
+    end
+
+    ## todo/fix: auto-calculate winner
+    # return 1,2,0   1 => team1, 2 => team2, 0 => draw/tie
+    ### calculate winner - use 1,2,0
+    if @score1 && @score2
+       if @score1 > @score2
+          @winner = 1
+       elsif @score2 > @score1
+          @winner = 2
+       elsif @score1 == @score2
+          @winner = 0
+       else
+       end
+    else
+      @winner = nil   # unknown / undefined
+    end
+
+    self   ## note - MUST return self for chaining
+  end
+
+
+  attr_reader :date,
+              :team1,     :team2,      ## todo/fix: use team1_name, team2_name or similar - for compat with db activerecord version? why? why not?
+              :score1,    :score2,     ## full time
+              :score1i,   :score2i,    ## half time (first (i) part)
+              :score1et,  :score2et,   ## extra time
+              :score1p,   :score2p,    ## penalty
+              :score1agg, :score2agg,  ## full time (all legs) aggregated
+              :winner,    # return 1,2,0   1 => team1, 2 => team2, 0 => draw/tie
+              :round,     ## todo/fix:  use round_num or similar - for compat with db activerecord version? why? why not?
+              :leg,      ## e.g. '1','2','3','replay', etc.   - use leg for marking **replay** too - keep/make leg numeric?! - why? why not?
+              :stage,
+              :group,
+              :conf1,    :conf2,      ## special case for mls e.g. conference1, conference2 (e.g. west, east, central)
+              :country1, :country2,    ## special case for champions league etc. - uses FIFA country code
+              :comments
+
+
+
+  def initialize( **kwargs )
+    update( kwargs )  unless kwargs.empty?
+  end
+
+
+  def over?()      true; end  ## for now all matches are over - in the future check date!!!
+  def complete?()  true; end  ## for now all scores are complete - in the future check scores; might be missing - not yet entered
+
+
+  def score_str    # pretty print (full time) scores; convenience method
+    "#{@score1}-#{@score2}"
+  end
+
+  def scorei_str    # pretty print (half time) scores; convenience method
+    "#{@score1i}-#{@score2i}"
+  end
+
+end  # class Match
+end # module Import
+
+end # module SportDb