sportdb-formats 1.0.0 → 1.0.1

data/lib/sportdb/formats/team/club_reader_props.rb

@@ -0,0 +1,75 @@
+# encoding: utf-8
+
+
+module SportDb
+module Import
+
+
+class ClubPropsReader
+
+  def catalog() Import.catalog; end
+
+
+  def self.read( path )   ## use - rename to read_file or from_file etc. - why? why not?
+    txt = File.open( path, 'r:utf-8' ) {|f| f.read }
+    parse( txt )
+  end
+
+  def self.parse( txt )
+    new( txt ).parse
+  end
+
+
+  def initialize( txt )
+    @txt = txt
+  end
+
+  def parse
+    recs = parse_csv( @txt )
+    recs.each do |rec|
+      name = rec[:name]
+      if name.nil?
+        puts "** !!! ERROR !!! Name column required / missing / NOT found in row:"
+        pp rec
+        exit 1
+      end
+
+      ## find / match club by (canocial) name
+      m = catalog.clubs.match( name )
+      if m && m.size > 1
+        puts "** !!! WARN !!! ambigious (multiple) club matches (#{m.size}) for name >#{name}< in props row:"
+        pp rec
+        pp m
+
+        ## todo/fix:  try filter by canonical name if more than one match
+        m = m.select { |club| club.name == name }
+        m = nil    if m.empty?     ## note: reset to nil if no more matches
+      end
+
+      if m.nil?
+        puts "** !!! ERROR !!! no club match for (canonical) name >#{name}< in props row:"
+        pp rec
+        exit 1
+      elsif m.size > 1
+        puts "** !!! ERROR !!! ambigious (multiple) club matches (#{m.size}) for (canonical) name >#{name}< in props row:"
+        pp rec
+        pp m
+        exit 1
+      else   ## assume size == 1, bingo!!!
+        club_rec = m[0]
+        ## todo/fix:  warn if name differes from (canonical) name
+        ## todo/fix:  also add props to in-memory structs/records!!!
+        ## todo/fix:   only updated "on-demand" from in-memory struct/records!!!!
+
+        ##  update attributes
+        club_rec.key  = rec[:key]      if rec[:key]
+        club_rec.code = rec[:code]     if rec[:code]
+        ## todo/fix: add (some) more props e.g. address, web, etc.
+      end
+    end
+  end # method parse
+
+end # class ClubPropsReader
+
+end ## module Import
+end ## module SportDb

data/lib/sportdb/formats/team/national_team_index.rb

@@ -0,0 +1,114 @@
+
+module SportDb
+  module Import
+
+class NationalTeamIndex
+
+  attr_reader :teams     ## all (national) team records
+
+  def initialize( recs )
+    @teams         = []
+    @teams_by_code = {}
+    @teams_by_name = {}
+
+    add( recs )
+  end
+
+  include NameHelper
+  ## incl. strip_year( name )
+  ##       has_year?( name)
+  ##       strip_lang( name )
+  ##       normalize( name )
+
+
+  def add( recs )
+    ###########################################
+    ## auto-fill national teams
+    ## pp recs
+    recs.each do |rec|
+      @teams << rec
+
+      ## add fifa code lookup
+      if @teams_by_code[ rec.code.downcase ]
+        puts "** !! ERROR !! national team code (code) >#{rec.code}< already exits!!"
+        exit 1
+      else
+        @teams_by_code[ rec.code.downcase ] = rec
+      end
+
+
+      ##  add all names (canonical name + alt names
+      names = [rec.name] + rec.alt_names
+      more_names = []
+      ## check "hand-typed" names for year (auto-add)
+      ## check for year(s) e.g. (1887-1911), (-2013),
+      ##                        (1946-2001,2013-) etc.
+      names.each do |name|
+        if has_year?( name )
+          more_names <<  strip_year( name )
+        end
+      end
+
+      names += more_names
+      ## check for duplicates - simple check for now - fix/improve
+      ## todo/fix: (auto)remove duplicates - why? why not?
+      count      = names.size
+      count_uniq = names.uniq.size
+      if count != count_uniq
+        puts "** !!! ERROR !!! - #{count-count_uniq} duplicate name(s) in national teams:"
+        pp names
+        pp rec
+        exit 1
+      end
+
+      names.each_with_index do |name,i|
+        ## check lang codes e.g. [en], [fr], etc.
+        ##  todo/check/fix:  move strip_lang up in the chain - check for duplicates (e.g. only lang code marker different etc.) - why? why not?
+        name = strip_lang( name )
+        norm = normalize( name )
+        old_rec = @teams_by_name[ norm ]
+        if old_rec
+          ## check if tame name already is included or is new team rec
+            msg = "** !!! ERROR !!! - national team name conflict/duplicate - >#{name}< will overwrite >#{old_rec.name}< with >#{rec.name}<"
+            puts msg
+            exit 1
+        else
+          @teams_by_name[ norm ] = rec
+        end
+      end
+    end  ## each record
+  end # method initialize
+
+  ##  fix/todo: add  find_by (alias for find_by_name/find_by_code)
+  def find_by_code( code )
+    code = code.to_s.downcase   ## allow symbols (and always downcase e.g. AUT to aut etc.)
+    @teams_by_code[ code ]
+  end
+
+  def find_by_name( name )
+    name = normalize( name.to_s )  ## allow symbols too (e.g. use to.s first)
+    @teams_by_name[ name ]
+  end
+
+  def find( q )
+    ## check longest match first (assume name is longer than code)
+    ## try lookup / find by (normalized) name first
+    team = find_by_name( q )
+    team = find_by_code( q )  if team.nil?
+    team
+  end
+
+  def find!( q )
+    team = find( q )
+    if team.nil?
+      puts "** !!! ERROR - no match for national team >#{q}< found"
+      exit 1
+    end
+    team
+  end
+end   # class NationalTeamIndex
+
+
+end   # module Import
+end   # module SportDb
+

data/lib/sportdb/formats/team/team_index.rb

@@ -0,0 +1,43 @@
+
+module SportDb
+  module Import
+
+
+class TeamIndex
+  ## note: "virtual" index lets you search clubs and/or national_teams (don't care)
+
+  def catalog() Import.catalog; end
+
+  ## todo/check: rename to/use map_by! for array version - why? why not?
+  def find_by!( name:, league:, mods: nil )
+    if name.is_a?( Array )
+      recs = []
+      name.each do |q|
+        recs << __find_by!( name: q, league: league, mods: mods )
+      end
+      recs
+    else  ## assume single name
+      __find_by!( name: name, league: league, mods: mods )
+    end
+  end
+
+  def __find_by!( name:, league:, mods: nil )
+    if mods && mods[ league.key ] && mods[ league.key ][ name ]
+      mods[ league.key ][ name ]
+    else
+      if league.clubs?
+        if league.intl?    ## todo/fix: add intl? to ActiveRecord league!!!
+          catalog.clubs.find!( name )
+        else  ## assume clubs in domestic/national league tournament
+          catalog.clubs.find_by!( name: name, country: league.country )
+        end
+      else   ## assume national teams (not clubs)
+        catalog.national_teams.find!( name )
+      end
+    end
+  end # method __find_by!
+
+end  # class TeamIndex
+
+end   # module Import
+end   # module SportDb

data/lib/sportdb/formats/team/wiki_reader.rb

@@ -0,0 +1,108 @@
+# encoding: utf-8
+
+
+module SportDb
+module Import
+
+
+class WikiReader    ## todo/check: rename to WikiClubReader - why? why not?
+
+  class WikiClub  # nested class
+    attr_reader :name, :country
+    def initialize( name, country )
+      @name, @country = name, country
+    end
+  end  # (nested) class WikiClub
+
+
+  def catalog() Import.catalog; end
+
+
+def self.read( path )   ## use - rename to read_file or from_file etc. - why? why not?
+  txt = File.open( path, 'r:utf-8' ) { |f| f.read }
+  parse( txt )
+end
+
+def self.parse( txt )
+  new( txt ).parse
+end
+
+def initialize( txt )
+  @txt = txt
+end
+
+def parse
+  recs = []
+  last_country = nil  ## note: supports only one level of headings for now (and that is a country)
+
+  @txt.each_line do |line|
+    line = line.strip
+
+    next if line.empty?
+    next if line.start_with?( '#' )   ## skip comments too
+
+    ## strip inline (until end-of-line) comments too
+    ##  e.g Eupen        => KAS Eupen,    ## [de]
+    ##   => Eupen        => KAS Eupen,
+    line = line.sub( /#.*/, '' ).strip
+    pp line
+
+
+    next if line =~ /^={1,}$/          ## skip "decorative" only heading e.g. ========
+
+     ## note: like in wikimedia markup (and markdown) all optional trailing ==== too
+     ##  todo/check:  allow ===  Text  =-=-=-=-=-=   too - why? why not?
+    if line =~ /^(={1,})       ## leading ======
+                 ([^=]+?)      ##  text   (note: for now no "inline" = allowed)
+                 =*            ## (optional) trailing ====
+                 $/x
+       heading_marker = $1
+       heading_level  = $1.length   ## count number of = for heading level
+       heading        = $2.strip
+
+       puts "heading #{heading_level} >#{heading}<"
+
+       if heading_level > 1
+         puts "** !!! ERROR [wiki reader] !!! -  - headings level too deep - only top / one level supported for now; sorry"
+         exit 1
+       end
+
+      ## assume country in heading; allow all "formats" supported by parse e.g.
+      ##   Österreich • Austria (at)
+      ##   Österreich • Austria
+      ##   Austria
+      ##   Deutschland (de) • Germany
+      country = catalog.countries.parse( heading )
+      ## check country code - MUST exist for now!!!!
+      if country.nil?
+        puts "!!! error [wiki reader] - unknown country >#{heading}< - sorry - add country to config to fix"
+        exit 1
+      end
+
+      last_country = country
+      pp last_country
+    else
+      ## strip and  squish (white)spaces
+      #   e.g. New York FC      (2011-)  => New York FC (2011-)
+      value = line.strip.gsub( /[ \t]+/, ' ' )
+
+      ## normalize (allow underscore (-) - replace with space)
+      ##  e.g. Cercle_Brugge_K.S.V. =>  Cercle Brugge K.S.V.
+      value = value.gsub( '_', ' ' )
+
+      if last_country.nil?
+        puts "** !!! ERROR [wiki reader] !!! - country heading missing for club name; sorry - add country heading to fix"
+        exit 1
+      end
+
+      rec = WikiClub.new( value, last_country )
+      recs << rec
+    end
+  end  # each_line
+  recs
+end  # method read
+
+end  # class WikiReader
+
+end ## module Import
+end ## module SportDb

data/lib/sportdb/formats/version.rb

@@ -6,7 +6,7 @@ module Formats
 
   MAJOR = 1    ## todo: namespace inside version or something - why? why not??
   MINOR = 0
-  PATCH = 0
+  PATCH = 1
   VERSION = [MAJOR,MINOR,PATCH].join('.')
 
   def self.version

data/test/helper.rb

@@ -10,6 +10,78 @@ require 'minitest/autorun'
 require 'sportdb/formats'
 
 
+
+module SportDb
+  module Import
+
+class TestCatalog
+  def build_country_index
+    recs = CountryReader.read( "#{Test.data_dir}/world/countries.txt" )
+    index = CountryIndex.new( recs )
+    index
+  end
+
+  def build_league_index
+    recs = SportDb::Import::LeagueReader.parse( <<TXT )
+  = England =
+  1       English Premier League
+            | ENG PL | England Premier League | Premier League
+  2       English Championship
+            | ENG CS | England Championship | Championship
+  3       English League One
+            | England League One | League One
+  4       English League Two
+  5       English National League
+
+  cup      EFL Cup
+            | League Cup | Football League Cup
+            | ENG LC | England Liga Cup
+
+  = Scotland =
+  1       Scottish Premiership
+  2       Scottish Championship
+  3       Scottish League One
+  4       Scottish League Two
+TXT
+
+    leagues = SportDb::Import::LeagueIndex.new
+    leagues.add( recs )
+    leagues
+  end
+
+  def build_club_index
+    recs = ClubReader.parse( <<TXT )
+= England
+
+Chelsea FC
+Arsenal FC
+Tottenham Hotspur
+West Ham United
+Crystal Palace
+Manchester United
+Manchester City
+TXT
+
+    index = ClubIndex.new
+    index.add( recs )
+    index
+  end
+
+
+  def countries() @countries ||= build_country_index; end
+  def leagues()   @leagues   ||= build_league_index; end
+  def clubs()     @clubs     ||= build_club_index; end
+end
+
+configure do |config|
+  config.catalog = TestCatalog.new
+end
+
+end  # module Import
+end  # module SportDb
+
+
+
 ################
 ## helper
 

data/test/test_club_index.rb

@@ -0,0 +1,183 @@
+# encoding: utf-8
+
+###
+#  to run use
+#     ruby -I ./lib -I ./test test/test_club_index.rb
+
+
+require 'helper'
+
+class TestClubIndex < MiniTest::Test
+
+  def test_match
+
+    recs = SportDb::Import::ClubReader.parse( <<TXT )
+=================================
+=  Austria
+
+== Wien ==
+
+FK Austria Wien,   Wien (Favoriten)
+  | Austria Vienna | Austria Wien
+SK Rapid Wien,     Wien (Hütteldorf)
+  | Rapid Vienna | Rapid Wien
+
+
+====================================
+=  England
+
+Arsenal FC, 1886, @ Emirates Stadium, London   ## Greater London
+  | Arsenal | FC Arsenal
+Chelsea FC, 1905, @ Stamford Bridge, London   ## Greater London
+  | Chelsea | FC Chelsea
+
+
+=====================================
+=  Russia
+
+Arsenal Tula,  Tula
+  | Arsenal | FC Arsenal Tula
+
+
+===========================
+=  Argentina
+
+== Buenos Aires ==
+
+Arsenal de Sarandí,  Sarandí › Buenos Aires    #  Sarandí es una ciudad de la Zona Sur del Gran Buenos Aires
+  | Arsenal | Arsenal Sarandi
+  | Arsenal FC | Arsenal Fútbol Club
+TXT
+
+   clubs = SportDb::Import::ClubIndex.new
+   clubs.add( recs )
+
+    pp clubs.errors
+
+    clubs.dump_duplicates
+
+    m = clubs.match( 'Rapid Wien' )
+    assert_equal 'SK Rapid Wien', m[0].name
+    assert_equal 'Austria',       m[0].country.name
+    assert_equal 'Wien',          m[0].city
+
+    m = clubs.match( 'rapid wien' )
+    assert_equal 'SK Rapid Wien', m[0].name
+    assert_equal 'Austria',       m[0].country.name
+    assert_equal 'Wien',          m[0].city
+
+    ## note: all dots (.) get always removed
+    m = clubs.match( '...r.a.p.i.d w.i.e.n...' )
+    assert_equal 'SK Rapid Wien', m[0].name
+    assert_equal 'Austria',       m[0].country.name
+    assert_equal 'Wien',          m[0].city
+
+    ## note: all spaces and dashes (-) get always removed
+    m = clubs.match( '--- r a p i d  w i e n ---' )
+    assert_equal 'SK Rapid Wien', m[0].name
+    assert_equal 'Austria',       m[0].country.name
+    assert_equal 'Wien',          m[0].city
+
+    m = clubs.match( 'RAPID WIEN' )
+    assert_equal 'SK Rapid Wien', m[0].name
+    assert_equal 'Austria',       m[0].country.name
+    assert_equal 'Wien',          m[0].city
+
+
+    c = clubs[ 'SK Rapid Wien' ]   ## check canoncial name match (only)
+    assert_equal 'SK Rapid Wien', c.name
+    assert_equal 'Austria',       c.country.name
+    assert_equal 'Wien',          c.city
+
+
+    m = clubs.match( 'Arsenal' )
+    assert_equal 3, m.size
+
+    m = clubs.match( 'ARSENAL' )
+    assert_equal 3, m.size
+
+    m = clubs.match_by( name: 'Arsenal', country: 'eng' )
+    assert_equal 1, m.size
+    assert_equal 'Arsenal FC', m[0].name
+    assert_equal 'England',    m[0].country.name
+    assert_equal 'London',     m[0].city
+
+    club = clubs.find_by!( name: 'Arsenal', country: 'eng' )
+    assert_equal 'Arsenal FC', club.name
+    assert_equal 'England',    club.country.name
+    assert_equal 'London',     club.city
+
+
+    m = clubs.match_by( name: 'Arsenal', country: 'ar' )
+    assert_equal 1, m.size
+    assert_equal 'Arsenal de Sarandí', m[0].name
+    assert_equal 'Argentina',          m[0].country.name
+    assert_equal 'Sarandí',            m[0].city
+
+    club = clubs.find_by!( name: 'Arsenal', country: 'ar' )
+    assert_equal 'Arsenal de Sarandí', club.name
+    assert_equal 'Argentina',          club.country.name
+    assert_equal 'Sarandí',            club.city
+
+
+    m = clubs.match_by( name: 'Arsenal', country: 'ru' )
+    assert_equal 1, m.size
+    assert_equal 'Arsenal Tula', m[0].name
+    assert_equal 'Russia',       m[0].country.name
+    assert_equal 'Tula',         m[0].city
+
+
+    m = clubs.match( 'Arsenal FC' )
+    assert_equal 2, m.size
+
+    m = clubs.match( 'Arsenal F.C.' )
+    assert_equal 2, m.size
+
+    m = clubs.match( '...A.r.s.e.n.a.l... F.C...' )
+    assert_equal 2, m.size
+  end
+
+
+  def test_wikipedia    # test wikipedia names and links/urls
+
+    recs = SportDb::Import::ClubReader.parse( <<TXT )
+==================================
+=  Belgium
+
+== Brussels ==
+
+RSC Anderlecht, 1908,   Brussels      ## use (just) Anderlecht or Brussel-Anderlecht ??
+  | Anderlecht | R.S.C. Anderlecht | Royal Sporting Club Anderlecht
+
+== West-Vlaanderen › Vlaanderen ==
+
+Club Brugge, 1891,    Brugge › West-Vlaanderen › Vlaanderen
+  | Club Brugge KV | Club Brugge Koninklijke Voetbalvereniging
+TXT
+
+
+    clubs = SportDb::Import::ClubIndex.new
+    clubs.add( recs )
+
+    recs = SportDb::Import::WikiReader.parse( <<TXT )
+==================================
+=  Belgium
+
+R.S.C. Anderlecht
+Club Brugge KV
+TXT
+    clubs.add_wiki( recs )
+
+
+    m = clubs.match( 'Club Brugge KV' )
+    assert_equal 1, m.size
+    assert_equal 'Club Brugge KV', m[0].wikipedia
+    assert_equal 'https://en.wikipedia.org/wiki/Club_Brugge_KV', m[0].wikipedia_url
+
+    m = clubs.match( 'RSC Anderlecht' )
+    assert_equal 1, m.size
+    assert_equal 'R.S.C. Anderlecht', m[0].wikipedia
+    assert_equal 'https://en.wikipedia.org/wiki/R.S.C._Anderlecht', m[0].wikipedia_url
+  end
+
+end # class TestClubIndex