sportdb-config 0.0.1

data/lib/sportdb/config/league.rb

@@ -0,0 +1,118 @@
+# encoding: utf-8
+
+module SportDb
+  module Import
+
+
+class LeagueConfig   ## use LeagueInfo or LeagueMap or LeagueHash or similar
+
+def initialize
+
+  ## just use leagues without latest for latest - why? why not?
+  @leagues_latest = {
+   'es' => { '1' => 'liga',  # spanish liga 1
+             '2' => 'liga2',  # spanish liga 2
+           },
+   'it' => { '1' => 'seriea',  # italian serie a
+             '2' => 'serieb',  # italian serie b
+           },
+   'de' => { '1' => 'bundesliga',  # german bundesliga
+             '2' => 'bundesliga2', # german 2. bundesliga
+           },
+  'nl'  =>  { '1' => 'eredivisie' },   # dutch eredivisie
+  'be'  =>  { '1' => 'proleague' },    # belgian pro league
+  'pt'  =>  { '1' => 'liga' },         # portugese Primeira Liga
+  'tr'  =>  { '1' => 'superlig' },     # turkish Süper Lig
+
+#  note: eng now read from txt
+#     'eng'  => { '1' => 'premierleague',  # english premier league
+#                 '2' => 'championship',  # english championship league
+#                 '3' => 'league1',  # english league 1
+#               },
+  }
+
+  ## change history to past or changes/changelog something - why? why not?
+  @leagues_history = {
+
+#  note: eng now read from txt
+#    'eng' => {
+#               ## until (including) 2003-04 season
+#               '2003-04' => { '1' => 'premierleague', # english premier league
+#                              '2' => 'division1',     # english division 1
+#                            },
+#               ## until (including) 1991-92} season
+#               '1991-92' => { '1' => 'division1', # english division 1
+#                              '2' => 'division2', # english division 2
+#                            }
+#             }
+  }
+
+  pp @leagues_latest
+  pp @leagues_history
+
+  %w(eng sco fr gr).each do |country|
+    hash = LeagueReader.from_file( "#{Boot.data_dir}/leagues/#{country}.txt" )
+    pp hash
+
+    hash.each do |season,league_hash|
+      if season == '*'  ## assume latest / default season
+        @leagues_latest[ country ] = league_hash
+      else
+        @leagues_history[ country ] ||= {}
+        @leagues_history[ country ][ season ]  = league_hash
+      end
+    end
+  end
+
+  pp @leagues_latest
+  pp @leagues_history
+end
+
+
+
+def basename( league, country:, season: )
+  ## todo/check: rename league: to key: - why? why not?
+
+  if country.include?( '-' )  ## assume package name e.g. eng-england etc.
+    ## cut off country code from package name
+    cc = country.split( '-' )[0]   # use first part
+  else
+    cc = country
+  end
+
+  if season
+    puts "  checking season >#{season}<"
+    ## check history if season is provided / supplied / known
+    history = @leagues_history[ cc ]
+    if history
+       season_start_year = SeasonUtils.start_year( season ).to_i
+       ##
+       ##  todo: sorty season keys - why? why not?  -- assume reverse chronological order for now
+       history.keys.reverse.each do |key|
+          history_season_start_year = SeasonUtils.start_year( key ).to_i
+          puts "  #{season_start_year} <= #{history_season_start_year} - #{season_start_year <= history_season_start_year}"
+          if season_start_year <= history_season_start_year
+            result = history[ key ][ league ]
+            if result
+              return "#{league}-#{result}"
+            else
+              return nil
+            end
+          end
+       end
+    end
+  end
+
+  latest  = @leagues_latest[ cc ]
+  if latest
+     result = latest[ league ]
+     return "#{league}-#{result}"   if result
+  end
+
+  nil
+end # method basename
+end # class LeagueConfig
+
+
+end ## module Import
+end ## module SportDb

data/lib/sportdb/config/league_reader.rb

@@ -0,0 +1,65 @@
+# encoding: utf-8
+
+
+module SportDb
+  module Import
+
+class LeagueReader
+
+def self.from_file( path )   ## use - rename to read_file - why? why not?
+  txt = File.open( path, 'r:utf-8' ).read
+  read( txt )
+end
+
+
+
+SEASON_REGEX = /\[
+                (?<season>
+                   \d{4}
+                   (-\d{2,4})?
+                )
+               \]/x
+
+def self.read( txt )
+  hash = {}
+  season = '*'   ## use '*' for default/latest season
+
+  txt.each_line do |line|
+    line = line.strip
+
+    next if line.empty?
+    next if line.start_with?( '#' )   ## skip comments too
+
+    ## strip inline comments too
+    line = line.sub( /#.*/, '' ).strip
+
+    pp line
+
+
+    if (m=line.match( SEASON_REGEX ))
+        season = m[:season]
+    else
+      key_line, values_line = line.split( '=>' )
+      values  = values_line.split( ',' )
+
+      ## remove leading and trailing spaces
+      key_line = key_line.strip
+      values = values.map { |value| value.strip }
+      pp values
+
+      league_key      = key_line
+      league_basename = values[0]
+
+      hash[season] ||= {}
+      hash[season][league_key] = league_basename
+    end
+  end
+  hash
+end # method read
+
+end ## class LeagueReader
+
+
+
+end ## module Import
+end ## module SportDb

data/lib/sportdb/config/structs/team.rb

@@ -0,0 +1,45 @@
+# encoding: utf-8
+
+module SportDb
+  module Struct
+
+
+
+class Team
+  attr_accessor  :name,
+                 :alt_names,
+                 :city,
+                 :country
+
+
+   ## add alias/compat - why? why not
+   def title() @name; end
+   def names()
+     ary = [@name]
+     ary += @alt_names  if @alt_names
+     ary
+   end   ## all names
+
+
+
+  def initialize
+    ## do nothing for now (use from_csv to setup data)
+  end
+
+  def self.create( **kwargs )
+    self.new.update( kwargs )
+  end
+
+  def update( **kwargs )
+    @name        = kwargs[:name]
+    @alt_names   = kwargs[:alt_names]
+    @city        = kwargs[:city]    ## use city struct - why? why not?
+    ## todo: add country too
+
+    self   ## note - MUST return self for chaining
+  end
+end # class Team
+
+
+end # module Struct
+end # module SportDb

data/lib/sportdb/config/team_reader.rb

@@ -0,0 +1,108 @@
+# encoding: utf-8
+
+
+module SportDb
+module Import
+
+
+class TeamReader
+
+
+def self.from_file( path )   ## use - rename to read_file - why? why not?
+  txt = File.open( path, 'r:utf-8' ).read
+  read( txt )
+end
+
+
+def self.read( txt )   ## rename to parse - why? why not? and use read for file read?
+  recs = []
+  txt.each_line do |line|
+    line = line.strip
+
+    next if line.empty?
+    next if line.start_with?( '#' )   ## skip comments too
+
+    ## strip inline comments too
+    ##  e.g Eupen        => KAS Eupen,    ## [de]
+    ##   => Eupen        => KAS Eupen,
+    line = line.sub( /#.*/, '' ).strip
+
+    pp line
+    names_line, team_line = line.split( '=>' )
+
+    names = names_line.split( /[|,]/ )   # team names - allow comma(,) or pipe(|)
+    team  = team_line.split( ',' )   # (canoncial) team name, team_city
+
+    ## remove leading and trailing spaces
+    names = names.map { |name| name.strip }
+    team  = team.map { |team| team.strip }
+    pp names
+    pp team
+
+    canonical_name = team[0]
+    city           = team[1]
+
+    ## squish (white)spaces
+    #   e.g. New York FC      (2011-)  => New York FC (2011-)
+    #   e.g. León     › Guanajuato     => León › Guanajuato
+    canonical_name = canonical_name.gsub( /[ \t]+/, ' ' )
+    city           = city.gsub( /[ \t]+/, ' ' )   if city
+
+    ## note:
+    ##   check/todo!!!!!!!!!!!!!!!!!-
+    ##  strip year if to present e.g. (2011-)
+    ##
+    ##  do NOT strip for defunct / historic clubs e.g.
+    ##    (1899-1910)
+    ## or (-1914) or (-2011) etc.
+
+
+
+    ###
+    ##  todo: move year out of canonical team name - why? why not?
+    ##
+
+    ## check if canonical name include (2011-) or similar in name
+    ##   if yes, remove (2011-) and add to (alt) names
+    ##   e.g. New York FC (2011) => New York FC
+    if canonical_name =~ /\(.+?\)/   ## note: use non-greedy (?) match
+      name = canonical_name.gsub( /\(.+?\)/, '' ).strip
+      names << name
+    end
+
+    ### todo/fix:
+    ##  auto-add alt name with dots stripped - why? why not?
+    ##    e.g.  D.C. United    => DC United
+    ##    e.g.  Liverpool F.C. => Liverpool FC
+    ##    e.g.  St. Albin       => St Albin etc.
+    ##    e.g.  1. FC Köln     => 1 FC Köln  -- make special case for 1. - why? why not?
+
+    ##
+    ## todo/fix:  unify mapping entries
+    ##   always lowercase !!!!  (case insensitive)
+    ##   always strip (2011-) !!!
+    ##   always strip dots (e.g. St., F.C, etc.)
+
+
+    names = names.uniq  ## remove duplicates - todo/fix: warn about duplicates - why? why not?
+
+    ## note: remove from alt_names if canonical name (mapping 1:1)
+    ##                            or if empty
+    alt_names = names.select {|name| (name.empty? || name == canonical_name) ? false : true }
+
+    ## todo: add country (code) too!!!
+    rec = SportDb::Struct::Team.create(
+                                    name:      canonical_name,
+                                    city:      city,     ## note: team_city is optional for now (might be nil!!!)
+                                    alt_names: alt_names
+                                  )
+    ## pp rec
+    recs << rec
+  end
+  recs
+end  # method read
+end  # class TeamReader
+
+
+end ## module Import
+end ## module SportDb

data/lib/sportdb/config/utils/league_utils.rb

@@ -0,0 +1,24 @@
+# encoding: utf-8
+
+
+module LeagueHelper
+  def basename( league, country:, season: nil )
+     ## e.g. eng-england, 2011-12, 1  returns  1-premierleague
+     ##
+     ##  allow country code or (repo) package name
+     ##    e.g. eng-england or eng
+     ##         de-deutschland or de etc.
+
+     leagues = SportDb::Import.config.leagues
+
+     result = leagues.basename( league, country: country, season: season )
+
+     ##
+     # note: if no mapping / nothing found return league e.g. 1, 2, 3, 3a, 3b, cup(?), etc.
+     result ? result : league
+  end
+end
+
+module LeagueUtils
+  extend LeagueHelper
+end

data/lib/sportdb/config/utils/season_utils.rb

@@ -0,0 +1,116 @@
+# encoding: utf-8
+
+
+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 = $1.to_i - 1
+      "%4d" % [fst]
+    else
+      puts "*** !!!! wrong season format >>#{season}<<; exit; sorry"
+      exit 1
+    end
+  end  # method prev
+
+
+  def key( basename )
+    if basename =~ /^(\d{4})[\-\/](\d{4})$/      ## 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}/#{$2}"
+    elsif basename =~ /^(\d{4})$/
+      $1
+    else
+      puts "*** !!!! wrong season format >>#{basename}<<; exit; sorry"
+      exit 1
+    end
+  end  # method key
+
+
+  def directory( season, format: nil )
+    ##  todo: find better names for formats - why? why not?:
+    ##    long | archive | decade(?)   =>   1980s/1988-89, 2010s/2017-18, ...
+    ##    short | std(?)   =>   1988-89, 2017-18, ...
+
+    ## convert season name to "standard" season name for directory
+
+    ## todo/fix: move to parse / validate season (for (re)use)!!!! - why? why not?
+    if season =~ /^(\d{4})[\-\/](\d{4})$/   ## 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
+end  # module SeasonHelper
+
+
+module SeasonUtils
+  extend SeasonHelper
+  ##  lets you use SeasonHelper as "globals" eg.
+  ##     SeasonUtils.prev( season ) etc.
+end # SeasonUtils