leagues 0.1.0

data/config/timezones_asia.csv

@@ -0,0 +1,9 @@
+key,   zone
+
+jp,   Asia/Tokyo
+kr,   Asia/Seoul
+cn,   Asia/Shanghai
+
+kz,   Asia/Almaty      # see en.wikipedia.org/wiki/Time_in_Kazakhstan
+
+

data/config/timezones_europe.csv

@@ -0,0 +1,98 @@
+key,   zone
+
+eng,    Europe/London
+sco,    Europe/London     ## check if entry or Edingburgh or such exits
+wal,    Europe/London
+nir,    Europe/London
+ie,     Europe/Dublin
+
+
+es,     Europe/Madrid
+pt,     Europe/Lisbon
+gi,     Europe/Gibraltar
+ad,	    Europe/Andorra
+
+fr,     Europe/Paris
+mc,	    Europe/Monaco
+
+it,     Europe/Rome
+sm,     Europe/San_Marino
+va,     Europe/Vatican
+
+
+
+gr,      Europe/Athens
+tr,     Europe/Istanbul
+cy,     Asia/Nicosia
+mt,     Europe/Malta
+
+
+ro,      Europe/Bucharest
+bg,      Europe/Sofia
+
+ua,     Europe/Kyiv
+by,     Europe/Minsk
+ru,     Europe/Moscow
+
+
+de,     Europe/Berlin
+at,     Europe/Vienna
+ch,     Europe/Zurich
+li,     Europe/Vaduz
+
+hu,     Europe/Budapest
+cz,    Europe/Prague
+sk,    Europe/Bratislava     # link to Prague !!!
+si,    Europe/Ljubljana
+pl,   Europe/Warsaw
+
+
+hr,   Europe/Zagreb    # link to Belgrade
+ba,   Europe/Sarajevo   # link to Belgrade
+rs,   Europe/Belgrade     # see en.wikipedia.org/wiki/Time_in_Serbia
+al,   Europe/Tirane
+me,	  Europe/Podgorica
+mk,	   Europe/Skopje
+md,    Europe/Chisinau
+
+kos,  Europe/Belgrade       #   en.wikipedia.org/wiki/Time_in_Kosovo
+xk,   Europe/Belgrade    # note - add both codes for now eg. xk & kos
+
+
+
+be,  Europe/Brussels
+lu,  Europe/Luxembourg      # link to Brussels !!!
+nl,  Europe/Amsterdam       # link to Brussels !!!
+
+
+dk,   Europe/Copenhagen    ## link to Berlin
+se,   Europe/Stockholm     ## link to Berlin
+no,   Europe/Oslo           # link to Berlin
+fi,   Europe/Helsinki
+
+lv,   Europe/Riga
+ee,   Europe/Tallinn
+lt,    Europe/Vilnius
+
+is,    Iceland
+fo,   Atlantic/Faroe
+
+
+
+am,  Asia/Yerevan      # en.wikipedia.org/wiki/Armenia_Time
+ge,  Asia/Tbilisi      # en.wikipedia.org/wiki/Georgia_Time
+az,  Asia/Baku
+
+
+
+uefa,          Europe/Paris
+uefa.cl,       Europe/Paris
+uefa.champs,   Europe/Paris  ## for champs default for now to cet (central european time) - why? why not?
+uefa.el,       Europe/Paris
+uefa.europa,   Europe/Paris
+uefa.conf,     Europe/Paris
+uefa.con,      Europe/Paris
+uefa.nl,       Europe/Paris
+uefa.nations,  Europe/Paris
+euro,          Europe/Paris
+

data/config/timezones_middle_east.csv

@@ -0,0 +1,5 @@
+key, zone
+
+il,  	Asia/Jerusalem
+
+

data/config/timezones_pacific.csv

@@ -0,0 +1,5 @@
+key,   zone
+
+
+au,  Australia/Sydney
+nz,  Pacific/Auckland

data/config/timezones_world.csv

@@ -0,0 +1,45 @@
+key,   zone
+
+
+## todo/check - allow/use country ref
+##            e.g.  world+1998,  fr
+##                  world+2006,  de
+##        and lookup timezone fr instead - why? why not?
+
+
+world+2022,    Asia/Qatar            # Qatar
+world+2018,    Europe/Moscow         # Russia
+world+2014,    America/Sao_Paulo     # Brazil
+world+2010,    Africa/Johannesburg   # South Africa
+world+2006,    Europe/Berlin         # Germany
+world+2002,    Asia/Tokyo            # Japan & South Korea
+world+1998,    Europe/Paris          # France
+
+
+friendlies,    Europe/London      ##  use (generic) utc - why? why not?
+
+
+
+## add club world cups
+
+world.clubs+2000,  America/Sao_Paulo    # Brazil 
+world.clubs+2005,  Asia/Tokyo      # Japan 
+world.clubs+2006,  Asia/Tokyo      # Japan
+world.clubs+2007,  Asia/Tokyo      # Japan
+world.clubs+2008,  Asia/Tokyo      # Japan
+world.clubs+2009,  Asia/Dubai      # UAE
+world.clubs+2010,  Asia/Dubai      # UAE
+world.clubs+2011,  Asia/Tokyo    # Japan
+world.clubs+2012,  Asia/Tokyo    # Japan
+world.clubs+2013,  Africa/Casablanca   # Morocco
+world.clubs+2014,  Africa/Casablanca   # Morocco
+world.clubs+2015,  Asia/Tokyo   # Japan
+world.clubs+2016,  Asia/Tokyo   # Japan
+world.clubs+2017,  Asia/Dubai   # UAE
+world.clubs+2018,  Asia/Dubai   # UAE
+world.clubs+2019,  Asia/Qatar   # Qatar
+world.clubs+2020,  Asia/Qatar   # Qatar
+world.clubs+2021,  Asia/Dubai   # UAE
+world.clubs+2022,  Africa/Casablanca    # Morocco
+world.clubs+2023,  Asia/Riyadh       # Saudi Arabia
+world.clubs+2025,  America/New_York    # United States

data/lib/leagues/league_codes.rb

@@ -0,0 +1,169 @@
+#####
+#
+#  quick & dirty league code lookup (and mapping)
+
+
+module SportDb
+class LeagueCodes 
+    
+####
+## (public) api
+def self.valid?( code )   
+    ## check if code is valid
+    builtin.valid?( code )
+end
+
+def self.find_by( code:, season: )
+    ## return league code record/item or nil
+    builtin.find_by( code: code, season: season )
+end
+
+
+#####
+## (static) helpers
+def self.builtin
+   ## get builtin league code index (build on demand)
+   @leagues ||= begin
+        leagues = SportDb::LeagueCodes.new
+        ['leagues',
+         'leagues_more',
+        ].each do |name|
+           recs = read_csv( "#{SportDb::Module::Leagues.root}/config/#{name}.csv" )
+           leagues.add( recs )
+        end
+
+        ['codes_alt',
+        ].each do |name|
+           recs = read_csv( "#{SportDb::Module::Leagues.root}/config/#{name}.csv" )
+           leagues.add_alt( recs )
+        end
+        leagues
+   end
+   @leagues
+end
+
+def self.norm( code )      ## use norm_(league)code - why? why not?
+  ## norm league code
+  ##   downcase
+  ##   and remove all non-letters/digits e.g. at.1 => at1, at 1 => at1 etc.
+  ##                                            ö.1 => ö1
+  ##   note - allow unicode letters!!! 
+  ##    note - assume downcase works for unicode too e.g. Ö=>ö
+  ##           for now no need to use our own downcase - why? why not?
+
+  code.downcase.gsub( /[^\p{Ll}0-9]/, '' )
+end
+
+
+
+
+def initialize
+    ## keep to separate (hash) table for now - why? why not?
+    @leagues = {}
+    @codes   = {}
+end    
+
+
+def add( recs )
+  recs.each do |rec|
+    key = LeagueCodes.norm( rec['code'] )
+    @leagues[ key ] ||= []
+
+    ## note: auto-change seasons to season object or nil
+    @leagues[ key ] << {  'code'         => rec['code'],
+                          'name'         => rec['name'],
+                          'basename'     => rec['basename'],
+                          'start_season' => rec['start_season'].empty? ? nil : Season.parse( rec['start_season'] ),
+                          'end_season'   => rec['end_season'].empty?   ? nil : Season.parse( rec['end_season'] ),
+                       }
+  end
+end
+
+### step two - add alt(ernative) codes
+def add_alt( recs )
+  recs.each do |rec|
+    key = LeagueCodes.norm( rec['alt'] )
+    @codes[ key ] ||= []
+
+  ### double check code reference 
+  ##    MUST be present for now!!
+     ref_key = LeagueCodes.norm( rec['code'] )
+     unless @leagues.has_key?( ref_key )
+       raise ArgumentError, "league code >#{rec['code']}< for alt code >#{rec['alt']}< not found; sorry" 
+     end  
+
+    ## note: auto-change seasons to season object or nil
+    @codes[ key ] << {  'code'         => rec['code'],
+                        'alt'          => rec['alt'],
+                        'start_season' => rec['start_season'].empty? ? nil : Season.parse( rec['start_season'] ),
+                        'end_season'   => rec['end_season'].empty?   ? nil : Season.parse( rec['end_season'] ),
+                     }
+ end
+end
+
+
+def valid?( code )
+  ## check if code is valid
+  ##   1) canonical codes  (check first - why? why not?)
+  ##   2) alt codes
+  raise ArgumentError, "league code as string|symbol expected"  unless code.is_a?(String) || code.is_a?(Symbol)
+
+  key = LeagueCodes.norm( code )
+  found = @leagues.has_key?( key )
+  found = @codes.has_key?( key )   if found == false
+  found
+end
+
+
+def find_by( code:, season: )
+  raise ArgumentError, "league code as string|symbol expected"  unless code.is_a?(String) || code.is_a?(Symbol)
+
+  ## return league code record/item or nil
+  ## check for alt code first
+  season = Season( season )
+  key    = LeagueCodes.norm( code )
+  rec    = nil
+
+  if !@leagues.has_key?( key )    ## try alt codes
+     key = _find_alt_code( key, season )
+  end
+
+  if key
+    recs = @leagues[ key ] 
+    if recs
+      rec =  _find_by_season( recs, season )
+    end
+  end
+
+  rec   ## return nil if no code record/item found
+end
+
+
+def _find_alt_code( key, season )
+     ## check alt keys
+     ref_key = nil
+     recs = @codes[key]
+     if recs
+        rec = _find_by_season( recs, season )
+        ## norm code
+        ref_key = LeagueCodes.norm( rec['code'] )  if rec  
+     end
+
+     ref_key    ## return nil if no mapping found
+end
+
+
+def _find_by_season( recs, season )
+  recs.each do |rec|
+      start_season = rec['start_season']
+      end_season   = rec['end_season']
+      return rec  if (start_season.nil? || start_season <= season) &&
+                     (end_season.nil? || end_season >= season)
+  end
+  nil
+end
+
+
+end # class LeagueCodes
+end # module SportDb    
+    

data/lib/leagues/leagueset.rb

@@ -0,0 +1,185 @@
+
+
+##
+# use/find a better name
+#    League Set, League Sheet,
+#    Leagueset/LeagueSet, LeagueSheet/Leaguesheet  
+#    or Leagues (only)??
+#    or League Book, League Setup, ??
+#    or Workset, Worksheet, Workbook, ...
+#
+#   move league config over here from sportdb-writers too!!!!!
+#
+#
+#   find a better way to handle league codes
+#            always map to canoncial codes - why? why not?
+
+
+
+module SportDb
+class Leagueset
+
+def self.parse_args( args )
+    ### split args in datasets with leagues and seasons
+    datasets = []
+    args.each do |arg|
+       if arg =~ %r{^[0-9/-]+$}   ##  season
+           if datasets.empty?
+             puts "!! ERROR - league required before season arg; sorry"
+             exit 1
+           end
+
+           season = Season.parse( arg )  ## check season
+           datasets[-1][1] << season
+       else ## assume league key
+           key = arg.downcase
+           datasets << [key, []]
+       end
+    end
+    new(datasets)
+end
+
+
+def self.parse( txt )
+    ### split args in datasets with leagues and seasons
+    datasets = []
+    recs = parse_csv( txt )
+    recs.each do |rec|
+        key = rec['league'].downcase
+        datasets << [key, []]
+
+        seasons_str = rec['seasons']
+        seasons = seasons_str.split( /[ ]+/ )
+
+        seasons.each do |season_str|
+            ## note - add support for ranges e.g. 2001/02..2010/11
+            if season_str.index( '..' )
+                    fst,snd = season_str.split( '..' )
+                    # pp [fst,snd]
+                    fst = Season.parse( fst )
+                    snd = Season.parse( snd )
+                    if fst < snd && fst.year? == snd.year?
+                        datasets[-1][1] += (fst..snd).to_a
+                    else
+                       raise ArgumentError, "parse error - invalid season range >#{str}<, 1) two seasons required, 2) first < second, 3) same (year/academic) type"
+                    end
+            else
+               season = Season.parse( season_str )  ## check season
+               datasets[-1][1] << season
+            end
+        end
+    end
+    new(datasets)
+end
+
+def self.read( path ) parse( read_text( path )); end
+
+
+
+def initialize( recs )
+    @recs = recs
+end
+
+def size() @recs.size; end
+
+def each( &blk )
+    @recs.each do |league_key, seasons|
+       blk.call( league_key, seasons )
+    end
+end
+
+
+### use a function for (re)use
+###   note - may add seasons in place!! (if seasons is empty)
+##
+##   todo/check - change source_path to (simply) path - why? why not?
+def validate!( source_path: ['.'] )
+    each do |league_key, seasons|
+  
+      unless LeagueCodes.valid?( league_key ) 
+        puts "!! ERROR - (leagueset) no league (config) found for code >#{league_key}<; sorry"
+        exit 1
+      end
+  
+  
+      if seasons.empty?
+        ## simple heuristic to find current season
+        [ Season( '2024/25'), Season( '2025') ].each do |season|
+           league_info = LeagueCodes.find_by( code: league_key, season: season )
+           filename = "#{season.to_path}/#{league_info['code']}.csv"
+           path = find_file( filename, path: source_path )
+           if path
+              seasons << season
+              break
+           end
+        end
+  
+        if seasons.empty?
+          puts "!! ERROR - (leagueset) no latest auto-season via source found for #{league_key}; sorry"
+          exit 1
+        end
+      end
+  
+      ## check source path too upfront - why? why not?
+      seasons.each do |season|
+           ## check league code config too - why? why not?
+           league_info = LeagueCodes.find_by( code: league_key, season: season )
+           if league_info.nil?
+             puts "!! ERROR - (leagueset) no league config found for code #{league_key}  AND season #{season}; sorry"
+             exit 1
+           end
+   
+           filename = "#{season.to_path}/#{league_info['code']}.csv"
+           path = find_file( filename, path: source_path )
+  
+           if path.nil?
+             puts "!! ERROR - (leagueset) no source found for #{filename}; sorry"
+             exit 1
+           end
+      end
+   end # each record
+end
+  
+
+
+## todo/check: find a better name for helper?
+##   find_all_datasets, filter_datatsets - add alias(es???
+##  queries (lik ARGV) e.g. ['at'] or ['eng', 'de'] etc. list of strings
+##
+##  todo/fix - check if used anywhere???
+##            - check if works with new alt codes too (or needs update)???
+
+def filter( queries=[] )
+  ## find all matching leagues (that is, league keys)
+  if queries.empty?  ## no filter - get all league keys
+    self
+  else
+    recs = @recs.find_all do |league_key, seasons|
+                               found = false
+                              ## note: normalize league key 
+                              ##        (remove dot and downcase)
+                              norm_key = league_key.gsub( '.', '' )
+                              queries.each do |query|
+                                 q = query.gsub( '.', '' ).downcase
+                                 if norm_key.start_with?( q )
+                                   found = true
+                                   break
+                                 end
+                              end
+                              found
+                           end
+    ## return new typed leagueset
+    self.class.new( recs )
+  end
+end
+
+
+def pretty_print( printer )
+    printer.text( "<Leagueset: " )
+    printer.text( @recs )
+    printer.text( ">")
+end
+
+end  # module Leagueset
+end  # module SportDb
+