sportdb-config 0.3.4 → 0.3.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 8e732ec27a048790023945b81ee4bfb641b511df
-  data.tar.gz: c54dc64a1b019accb8337e1ef80d252338c8416c
+  metadata.gz: b96366c57a674d8504918759864f3b2d77242ab1
+  data.tar.gz: 867a992ef3dd1b31b5e360e0b2b80b123a2abff3
 SHA512:
-  metadata.gz: d73d9aafbd977d57be80e230de8c9bb8d0090fe816ee235df9854e8d4f6877e1df863b74e1a22ff80ed360534141c46e7e1b23f1c219ebd418d3cf7d9c87d7f9
-  data.tar.gz: 730cf120ff994ac024b462e164faf2aba875e3f3e29ee72d73e4594706cbe088253050f85b9557e78881061c923cea30b12ee278a9de5b38b2fbffe0152eb830
+  metadata.gz: 0c5197829655457ac0b38a8d995e611804f3e385cf25b78f0a4c4e83c68fececfed3d16c4b51201c6f4f0d8f29f9189e629dc20a30a602f4f534c46bc6eaf2ea
+  data.tar.gz: 676a5a68e7827b9c34c7d9b07ed9d5ab7894bff4514841f98fbf989c29049f17d44968b3032bbf3a00bc273f7917a572460f632b6aea4e6b404385cdd10bde89

data/Manifest.txt

@@ -2,25 +2,27 @@ HISTORY.md
 Manifest.txt
 README.md
 Rakefile
-config/geos/eng.txt
 config/leagues/eng.txt
 config/leagues/fr.txt
 config/leagues/gr.txt
 config/leagues/sco.txt
+config/world/countries.txt
+config/world/eng.txt
 lib/sportdb/config.rb
+lib/sportdb/config/club_reader.rb
+lib/sportdb/config/clubs.rb
 lib/sportdb/config/config.rb
 lib/sportdb/config/countries.rb
 lib/sportdb/config/league.rb
 lib/sportdb/config/league_reader.rb
 lib/sportdb/config/league_utils.rb
 lib/sportdb/config/season_utils.rb
-lib/sportdb/config/team_reader.rb
 lib/sportdb/config/version.rb
 test/helper.rb
+test/test_club_reader.rb
 test/test_clubs.rb
 test/test_config.rb
 test/test_countries.rb
 test/test_league_reader.rb
 test/test_league_utils.rb
 test/test_season_utils.rb
-test/test_team_reader.rb

data/Rakefile

@@ -19,6 +19,10 @@ Hoe.spec 'sportdb-config' do
 
   self.licenses = ['Public Domain']
 
+  self.extra_deps = [
+    ['csvreader',   '>= 1.2.4'],
+  ]
+
   self.spec_extras = {
    :required_ruby_version => '>= 2.2.2'
   }

data/config/world/countries.txt

@@ -0,0 +1,307 @@
+##########################
+# FIFA country codes
+#
+#  see   en.wikipedia.org/wiki/List_of_FIFA_country_codes
+#   and  github.com/openmundi/world.db/blob/master/1-codes/fifa.txt
+#   and  www.rsssf.com/miscellaneous/fifa-codes.html
+
+##
+# note: all dependencies n territories marked w/
+#            (UK), (FR), (NZ), (AU), (US), (NL), (DK), (CN) etc.
+
+
+###
+# todo: group by world.db "continents" - south america, africa, etc.
+
+
+
+#################
+## 209  (now in 2019 211(!) which two got added ??) regular FIFA members
+
+
+Key,  FIFA,  Name
+
+af,  AFG, Afghanistan
+al,  ALB, Albania
+dz,  ALG, Algeria
+as,  ASA, American Samoa (US)
+ad,  AND, Andorra
+ao,  ANG, Angola
+ai,  AIA, Anguilla (UK)
+ag,  ATG, Antigua and Barbuda
+ar,  ARG, Argentina
+am,  ARM, Armenia
+aw,  ARU, Aruba (NL)
+au,  AUS, Australia
+at,  AUT, Austria
+az,  AZE, Azerbaijan
+
+bs,  BAH, Bahamas
+bh,  BHR, Bahrain
+bd,  BAN, Bangladesh
+bb,  BRB, Barbados
+by,  BLR, Belarus
+be,  BEL, Belgium
+bz,  BLZ, Belize
+bj,  BEN, Benin
+bm,  BER, Bermuda (UK)
+bt,  BHU, Bhutan
+bo,  BOL, Bolivia
+ba,  BIH, Bosnia and Herzegovina      ## or use Bosnia-Herzegovina ?
+bw,  BOT, Botswana
+br,  BRA, Brazil
+bn,  BRU, Brunei
+bg,  BUL, Bulgaria
+bf,  BFA, Burkina Faso
+bi,  BDI, Burundi
+
+kh,  CAM, Cambodia
+cm,  CMR, Cameroon
+ca,  CAN, Canada
+cv,  CPV, Cabo Verde
+ky,  CAY, Cayman Islands (UK)
+cf,  CTA, Central African Republic
+td,  CHA, Chad
+cl,  CHI, Chile
+cn,  CHN, China
+co,  COL, Colombia
+km,  COM, Comoros
+cg,  CGO, Congo
+cd,  COD, Congo DR
+ck,  COK, Cook Islands (NZ)
+cr,  CRC, Costa Rica
+hr,  CRO, Croatia
+cu,  CUB, Cuba
+cw,  CUW, Curaçao (NL)
+cy,  CYP, Cyprus
+cz,  CZE, Czech Republic
+ci,  CIV, Côte d'Ivoire      # Ivory Coast
+
+dk,  DEN, Denmark
+dj,  DJI, Djibouti
+dm,  DMA, Dominica
+do,  DOM, Dominican Republic
+
+ec,  ECU, Ecuador
+eg,  EGY, Egypt
+eng, ENG, England (UK)
+gq,  EQG, Equatorial Guinea
+er,  ERI, Eritrea
+ee,  EST, Estonia
+et,  ETH, Ethiopia
+es,  ESP, Spain
+
+fo,  FRO, Faroe Islands (DK)
+fj,  FIJ, Fiji
+fi,  FIN, Finland
+fr,  FRA, France
+
+ga,  GAB, Gabon
+gm,  GAM, Gambia
+ge,  GEO, Georgia
+de,  GER, Germany
+gh,  GHA, Ghana
+gr,  GRE, Greece
+gd,  GRN, Grenada
+gu,  GUM, Guam (US)
+gt,  GUA, Guatemala
+gn,  GUI, Guinea
+gw,  GNB, Guinea-Bissau
+gy,  GUY, Guyana
+
+ht,  HAI, Haiti
+hn,  HON, Honduras
+hk,  HKG, Hong Kong (CN)
+hu,  HUN, Hungary
+
+is,  ISL, Iceland
+in,  IND, India
+id,  IDN, Indonesia
+ir,  IRN, Iran
+iq,  IRQ, Iraq
+il,  ISR, Israel
+it,  ITA, Italy
+ie,  IRL, Ireland
+
+jm,  JAM, Jamaica
+jp,  JPN, Japan
+jo,  JOR, Jordan
+
+kz,  KAZ, Kazakhstan
+ke,  KEN, Kenya
+kw,  KUW, Kuwait
+kg,  KGZ, Kyrgyzstan
+sa,  KSA, Saudi Arabia
+kr,  KOR, South Korea
+
+la,  LAO, Laos
+lv,  LVA, Latvia
+lb,  LIB, Lebanon
+ls,  LES, Lesotho
+lr,  LBR, Liberia
+ly,  LBY, Libya
+li,  LIE, Liechtenstein
+lt,  LTU, Lithuania
+lu,  LUX, Luxembourg
+lc,  LCA, Saint Lucia
+
+mo,  MAC, Macao (CN)
+mk,  MKD, Macedonia
+mg,  MAD, Madagascar
+mw,  MWI, Malawi
+my,  MAS, Malaysia
+mv,  MDV, Maldives
+ml,  MLI, Mali
+mt,  MLT, Malta
+mr,  MTN, Mauritania
+mu,  MRI, Mauritius
+mx,  MEX, Mexico
+md,  MDA, Moldova
+mn,  MNG, Mongolia
+me,  MNE, Montenegro
+ms,  MSR, Montserrat (UK)
+ma,  MAR, Morocco
+mz,  MOZ, Mozambique
+mm,  MYA, Myanmar
+
+na,  NAM, Namibia
+np,  NEP, Nepal
+nl,  NED, Netherlands
+nc,  NCL, New Caledonia (FR)
+nz,  NZL, New Zealand
+ni,  NCA, Nicaragua
+ne,  NIG, Niger
+ng,  NGA, Nigeria
+nir, NIR, Northern Ireland (UK)
+no,  NOR, Norway
+
+om,  OMA, Oman
+
+pk,  PAK, Pakistan
+ps,  PLE, Palestine
+pa,  PAN, Panama
+pg,  PNG, Papua New Guinea
+py,  PAR, Paraguay
+pe,  PER, Peru
+ph,  PHI, Philippines
+pl,  POL, Poland
+pt,  POR, Portugal
+pr,  PUR, Puerto Rico (US)
+kp,  PRK, North Korea
+
+qa,  QAT, Qatar
+
+ro,  ROU, Romania
+ru,  RUS, Russia
+rw,  RWA, Rwanda
+za,  RSA, South Africa
+
+kn,  SKN, Saint Kitts and Nevis
+ws,  SAM, Samoa
+sm,  SMR, San Marino
+st,  STP, São Tomé and Príncipe
+sco, SCO, Scotland (UK)
+sn,  SEN, Senegal
+rs,  SRB, Serbia
+sc,  SEY, Seychelles
+sl,  SLE, Sierra Leone
+sg,  SIN, Singapore
+sk,  SVK, Slovakia
+si,  SVN, Slovenia
+sb,  SOL, Solomon Islands
+so,  SOM, Somalia
+ss,  SSD, South Sudan
+lk,  SRI, Sri Lanka
+sd,  SDN, Sudan
+sr,  SUR, Suriname
+sz,  SWZ, Swaziland
+se,  SWE, Sweden
+ch,  SUI, Switzerland
+sy,  SYR, Syria
+sv,  SLV, El Salvador
+
+pf,  TAH, Tahiti (FR)   # todo/check: use French Polynesia (FR) as name - why? why not?
+tj,  TJK, Tajikistan
+tz,  TAN, Tanzania
+th,  THA, Thailand
+tl,  TLS, Timor-Leste
+tg,  TOG, Togo
+to,  TGA, Tonga
+tt,  TRI, Trinidad and Tobago
+tn,  TUN, Tunisia
+tr,  TUR, Turkey
+tm,  TKM, Turkmenistan
+tc,  TCA, Turks and Caicos Islands (UK)
+tw,  TPE, Taiwan              # Chinese Tapei
+
+ug,  UGA, Uganda
+ua,  UKR, Ukraine
+ae,  UAE, United Arab Emirates
+us,  USA, United States
+uy,  URU, Uruguay
+uz,  UZB, Uzbekistan
+
+vu,  VAN, Vanuatu
+ve,  VEN, Venezuela
+vn,  VIE, Vietnam
+vi,  VIR, US Virgin Islands (US)    # United States Virgin Islands; Virgin Islands (U.S.)
+vc,  VIN, Saint Vincent and the Grenadines
+vg,  VGB, British Virgin Islands (UK)
+
+wal, WAL, Wales (UK)
+
+ye,  YEM, Yemen
+
+zm,  ZAM, Zambia
+zw,  ZIM, Zimbabwe
+
+
+###################
+## Non-FIFA member codes
+
+bq,  BOE, Bonaire        # CONCACAF
+gf,  GYF, French Guiana (FR)  # CONCACAF
+gp,  GPE, Guadeloupe (FR)     # CONCACAF
+mq,  MTQ, Martinique (FR)     # CONCACAF
+sx,  SXM, Sint Maarten (NL)   # CONCACAF
+
+gi,  GIB, Gibraltar (UK)   # UEFA
+xk,  KOS, Kosovo         # UEFA     -- note: Kosovo is not listed as an ISO standard country. The unofficial 2 and 3-digit codes are used by the European Commission and others until Kosovo is assigned an ISO code
+
+re,  REU, Réunion (FR)        # CAF
+zan, ZAN, Zanzibar (TZ)       # CAF     -- note:  is a semi-autonomous region of Tanzania
+
+tv,  TUV, Tuvalu         # OFC
+
+
+####################
+## Irregular codes
+
+mp,  NMI, Northern Mariana Islands (US)  # AFC
+
+mf,  SMT, Saint Martin (FR)    # CONCACAF
+
+fm,  FSM, Micronesia      # OFC
+ki,  KIR, Kiribati        # OFC
+nu,  NIU, Niue (NZ)       # OFC
+pw,  PLW, Palau           # OFC
+
+
+#################
+# not members
+#
+# in europe:
+# - Great Britain - only all its member e.g. England/Wales/Scotland/Northern Irland
+
+va,  VAT,  Vatican City    ## no "official" fifa code - listed in "irregular" codes
+mc,  MCO,  Monaco          ## no "official" fifa code - listed in "irregular" codes
+
+
+#
+# in oceania:
+# - Kiribati
+# - Marshall Islands
+# - Micronesia
+# - Nauru
+# - Palau
+# - Tuvalu

data/lib/sportdb/config/{team_reader.rb → club_reader.rb}

@@ -5,31 +5,7 @@ module SportDb
 module Import
 
 
-class TeamReader
-
-##
-#  note: use our own (internal) team struct for now - why? why not?
-#    - check that shape/structure/fields/attributes match
-#      the Team struct in sportdb-text (in SportDb::Struct::Team)  !!!!
-class Team
-  ##  todo: use just names for alt_names - why? why not?
-  attr_accessor :name, :alt_names, :year, :ground, :city
-
-  ## more attribs - todo/fix - also add "upstream" to struct & model!!!!!
-  attr_accessor :district, :geos, :year_end, :country
-
-  def historic?()  @year_end ? true : false; end
-  alias_method  :past?, :historic?
-
-
-  def initialize
-    @alt_names = []
-  end
-
-
-
-end # class Team
-
+class ClubReader
 
 
 def self.read( path )   ## use - rename to read_file or from_file etc. - why? why not?
@@ -89,12 +65,13 @@ def self.parse( txt )
          ## quick hack:   if level is 1 assume country for now
          ##                 and extract country code e.g.
          ##                    Austria (at) => at
+         ##  todo/fix:  allow code only e.g. at or aut without enclosing () too - why? why not?
          if heading_level == 1
-             if heading =~ /\(([a-z]{2,3})\)/
+             if heading =~ /\(([a-z]{2,3})\)/i    ## note allow (at) or (AUT) too
                country_code = $1
 
                ## check country code - MUST exist for now!!!!
-               country = SportDb::Import.config.countries[ country_code.to_sym ]
+               country = SportDb::Import.config.countries[ country_code ]
                if country.nil?
                  puts "!!! error [team reader] - unknown country with code >#{country_code}< - sorry - add country to config to fix"
                  exit 1
@@ -150,7 +127,7 @@ def self.parse( txt )
     else
       values = line.split( ',' )
 
-      rec = Team.new
+      rec = Club.new
       value = values.shift    ## get first item
       ## strip and  squish (white)spaces
       #   e.g. New York FC      (2011-)  => New York FC (2011-)
@@ -231,7 +208,7 @@ def self.parse( txt )
 
       ## 1) add country if present
       if headings.size > 0 && headings[0]
-        country = SportDb::Import.config.countries[ headings[0].to_sym ]
+        country = SportDb::Import.config.countries[ headings[0] ]
         rec.country = country
       else
         ## make it an error - why? why not?
@@ -292,7 +269,7 @@ def self.split_geo( str )
   geos
 end
 
-end  # class TeamReader
+end  # class ClubReader
 
 
 end ## module Import

data/lib/sportdb/config/clubs.rb

@@ -0,0 +1,126 @@
+# 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)  !!!!
+class Club
+  ##  todo: use just names for alt_names - why? why not?
+  attr_accessor :name, :alt_names, :year, :ground, :city
+
+  ## more attribs - todo/fix - also add "upstream" to struct & model!!!!!
+  attr_accessor :district, :geos, :year_end, :country
+
+  def historic?()  @year_end ? true : false; end
+  alias_method  :past?, :historic?
+
+  def initialize
+    @alt_names = []
+  end
+end # class Club
+
+
+
+class ClubIndex
+
+  def initialize
+    @clubs          = {}   ## clubs (indexed) by canonical name
+    @clubs_by_name  = {}
+    @errors         = []
+  end
+
+  attr_reader :errors
+  def errors?() @errors.empty? == false; end
+
+
+  def add( rec_or_recs )   ## add club record / alt_names
+    recs = rec_or_recs.is_a?( Array ) ? rec_or_recs : [rec_or_recs]      ## wrap (single) rec in array
+
+    recs.each do |rec|
+      ## puts "adding:"
+      ## pp rec
+      ### step 1) add canonical name
+      old_rec = @clubs[ rec.name ]
+      if old_rec
+        puts "** !!! ERROR !!! - (canonical) name conflict - duplicate - >#{rec.name}< will overwrite >#{old_rec.name}<:"
+        pp old_rec
+        pp rec
+        exit 1
+      else
+        @clubs[ rec.name ] = rec
+      end
+
+      ## step 2) add all names (canonical name + alt names)
+      names = [rec.name] + rec.alt_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):"
+        pp rec
+        exit 1
+      end
+
+      names.each_with_index do |name,i|
+        alt_recs = @clubs_by_name[ name ]
+        if alt_recs
+          msg = "** !!! WARN !!! - name conflict/duplicate - >#{name}< will overwrite >#{alt_recs[0].name}, #{alt_recs[0].country.name}< with >#{rec.name}, #{rec.country.name}<"
+          puts msg
+          @errors << msg
+          alt_recs << rec
+        else
+          @clubs_by_name[ name ] = [rec]
+        end
+      end
+    end
+  end # method add
+
+
+  def []( name )    ## lookup by canoncial name only
+    @clubs[ name ]
+  end
+
+  def match( name )
+    ## todo/check: return empty array if no match!!! and NOT nil (add || []) - why? why not?
+    @clubs_by_name[ name ]
+  end
+
+
+  def match_by( name:, country: )
+    ## note: match must for now always  include name
+    m = match( name )
+    if m    ## filter by country
+      ## note: country assumes / allows the country key or fifa code for now
+      country_rec = SportDb::Import.config.countries[ country ]
+      if country_rec.nil?
+        puts "** !!! ERROR !!! - unknown country >#{country}< - no match found, sorry - add to world/countries.txt in config"
+        exit 1
+      end
+
+      m = m.select { |club| club.country.key == country_rec.key }
+      m = nil   if m.empty?     ## note: reset to nil if no more matches
+    end
+    m
+  end
+
+
+
+  def dump_duplicates # debug helper - report duplicate club name records
+     @clubs_by_name.each do |name, clubs|
+       if clubs.size > 1
+         puts "#{clubs.size} duplicates for >#{name}<:"
+         pp clubs
+       end
+     end
+  end
+
+end # class ClubIndex
+
+
+end   # module Import
+end   # module SportDb

data/lib/sportdb/config/config.rb

@@ -6,41 +6,36 @@ module SportDb
 
 class Configuration
 
-
-  def team_mappings    ## todo/fix: rename to clubs_mappings - why? why not?
-    read_teams()        if @team_mappings.nil?
-    @team_mappings
+  ## todo/check:  rename to country_mappings/index - why? why not?
+  ##    or countries_by_code or countries_by_key
+  def countries
+    @countries ||= build_country_index
+    @countries
   end
 
-  def teams             ## todo/fix: rename to clubs - why? why not?
-    read_teams()        if @teams.nil?
-    @teams
+  def build_country_index    ## todo/check: rename to setup_country_index or read_country_index - why? why not?
+    recs = read_csv( "#{SportDb::Boot.data_dir}/world/countries.txt" )
+    CountryIndex.new( recs )
   end
 
-  def leagues
-    read_leagues()      if @leagues.nil?
-    @leagues
-  end
 
 
+  def clubs
+    @clubs  ||= build_club_index
+    @clubs
+  end
+
   ####
   #  todo/fix:  find a better way to configure club / team datasets
   attr_accessor   :clubs_dir
   def clubs_dir()   @clubs_dir ||= './clubs'; end
 
 
-  CLUBS_REGEX1 = %r{  (?:^|/)            # beginning (^) or beginning of path (/)
-                       (?<country>[a-z]{1,3})\.clubs\.txt$
-                   }x
-
-  CLUBS_REGEX2 = %r{  (?:^|/)            # beginning (^) or beginning of path (/)
-                       (?<country>[a-z]{2,3})-[^/]+?
-                         /clubs\.txt$
+  CLUBS_REGEX = %r{  (?:^|/)               # beginning (^) or beginning of path (/)
+                       (?:[a-z]{1,3}\.)?   # optional country code/key e.g. eng.clubs.txt
+                       clubs\.txt$
                    }x
 
-  CLUBS_REGEX = Regexp.union( CLUBS_REGEX1, CLUBS_REGEX2 )
-
-
   def find_clubs_datafiles( path )
      datafiles = []   ## note: [country, path] pairs for now
 
@@ -48,106 +43,15 @@ class Configuration
      candidates = Dir.glob( "#{path}/**/*.txt" )
      pp candidates
      candidates.each do |candidate|
-       m = CLUBS_REGEX.match( candidate )
-       if m
-         datafiles << [m[:country], candidate]
-       end
+       datafiles << candidate    if CLUBS_REGEX.match( candidate )
      end
 
      pp datafiles
      datafiles
   end
 
-=begin
-  CLUBS_DATAFILES = {   ## path by country to clubs.txt
-      de:  'europe/de-deutschland',
-      fr:  'europe/fr-france',
-      mc:  'europe/mc-monaco',
-      es:  'europe/es-espana',
-      it:  'europe/it-italy',
-      pt:  'europe/pt-portugal',
-      nl:  'europe/nl-netherlands',
-      be:  'europe/be-belgium',
-      tr:  'europe/tr-turkey',
-      gr:  'europe/gr-greece',
-      eng: 'europe/eng-england',
-      wal: 'europe/wal-wales',
-      sco: 'europe/sco-scotland',
-      nir: 'europe/nir-northern-ireland',
-      at:  'europe/at-austria',
-      al:  'europe/al-albania',
-      bg:  'europe/bg-bulgaria',
-      by:  'europe/by-belarus',
-      ch:  'europe/ch-confoederatio-helvetica',
-      cy:  'europe/cy-cyprus',
-      cz:  'europe/cz-czech-republic',
-      dk:  'europe/dk-denmark',
-      fi:  'europe/fi-finland',
-      ge:  'europe/ge-georgia',
-      hr:  'europe/hr-croatia',
-      hu:  'europe/hu-hungary',
-      ie:  'europe/ie-ireland',
-      is:  'europe/is-iceland',
-      lu:  'europe/lu-luxembourg',
-      md:  'europe/md-moldova',
-      mt:  'europe/mt-malta',
-      no:  'europe/no-norway',
-      pl:  'europe/pl-poland',
-      ro:  'europe/ro-romania',
-      rs:  'europe/rs-serbia',
-      ru:  'europe/ru-russia',
-      se:  'europe/se-sweden',
-      si:  'europe/si-slovenia',
-      sk:  'europe/sk-slovakia',
-      ua:  'europe/ua-ukraine',
-      mx:  'north-america/mx-mexico',
-      us:  'north-america/us-united-states',
-      ca:  'north-america/ca-canada',
-      ar:  'south-america/ar-argentina',
-      br:  'south-america/br-brazil',
-      jp:  'asia/jp-japan',
-      cn:  'asia/cn-china' }
-=end
-
-
-
-  class ClubsHash   ## todo/check: use (rename to) ClubsMapping(s) - why? why not?
-    def initialize
-      @h      = {}
-      @errors = []
-    end
-
-    attr_reader :errors
-    def errors?() @errors.empty? == false; end
-
-    def add( rec )   ## add club record / alt_names
-      rec.alt_names.each do |alt_name|
-        alt_recs = @h[ alt_name ]
-        if alt_recs
-          msg = "** !!! WARN !!! - alt name conflict/duplicate - >#{alt_name}< will overwrite >#{alt_recs[0].name}, #{alt_recs[0].country.name}< with >#{rec.name}, #{rec.country.name}<"
-          puts msg
-          @errors << msg
-          alt_recs << rec
-        else
-          @h[ alt_name ] = [rec]
-        end
-      end
-    end
 
-    def fetch( alt_name ) @h[alt_name]; end
-    def []( alt_name )   ## quick fix: for now returns first name; use fetch for recs array for now!!!
-       ## keep "compatible" with old code for now
-       alt_recs = @h[ alt_name ]
-       if alt_recs
-         alt_recs[0].name   ## use canonical name of first match for now
-       else
-         nil   # nothing found
-       end
-    end
-  end # class ClubHash
-
-
-  def read_teams   ## todo/fix: rename to read_clubs - why? why not?
+  def build_club_index
     ## unify team names; team (builtin/known/shared) name mappings
     ## cleanup team names - use local ("native") name with umlaut etc.
     recs = []
@@ -159,61 +63,32 @@ class Configuration
     ##              check that name and alt_names for a club are all unique (not duplicates)
     datafiles = find_clubs_datafiles( clubs_dir )
     datafiles.each do |datafile|
-       country  = datafile[0]    ## country code e.g. eng, at, de, etc.
-       path     = datafile[1]
-       recs += TeamReader.read( path )
+       recs += ClubReader.read( datafile )
     end
 
-    ############################
-    ## add team mappings
-    ##   alt names to canonical pretty (recommended unique) name
-    @team_mappings = ClubsHash.new
-    recs.each { |rec| @team_mappings.add( rec ) }
-
-###
-## reverse hash for lookup/list of "official / registered(?)"
-##    pretty recommended canonical unique (long form)
-##    team names
-
-    @teams = {}
-    recs.each do |rec|
-      old_rec = @teams[ rec.name ]
-      if old_rec
-        puts "** !!! ERROR !!! - (canonical) name conflict - duplicate - >#{rec.name}< will overwrite >#{old_rec.name}<:"
-        pp old_rec
-        pp rec
-        exit 1
-      else
-        @teams[ rec.name ] = rec
-
-        ## note: check if there's an alternative name registered
-        ##           for the canonical name
-        ##    todo/fix: find a better scheme for (alt) name mapping - why? why not?
-        ##   allow alt names same as canonical names if not is the same club/record - why? why not?
-        alt_recs =  @team_mappings.fetch( rec.name )
-        if alt_recs
-          puts "** !!! ERROR !!! - (canonical) name conflict - >#{rec.name}< registered already as alternative name mapping:"
-          pp alt_recs
-
-          ## note: exit now only if it's the same club (that has alt name that is the canonical name)
-          exit 1    if alt_recs.include?(rec)
-        end
-      end
-    end
 
-    if @team_mappings.errors?
+    clubs  = ClubIndex.new
+    clubs.add( recs )
+
+    if clubs.errors?
       puts ""
       puts "!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"
-      puts " #{@team_mappings.errors.size} errors:"
-      pp @team_mappings.errors
+      puts " #{clubs.errors.size} errors:"
+      pp clubs.errors
       ## exit 1
     end
 
-    self  ## return self for chaining
-end
+    clubs
+end # method build_club_index
+
 
 
 
+def leagues
+  read_leagues()      if @leagues.nil?
+  @leagues
+end
+
 def read_leagues
     #####
     # add / read-in leagues config

data/lib/sportdb/config/countries.rb

@@ -3,154 +3,75 @@
 module SportDb
   module Import
 
-class Configuration
-
-  ## built-in countries for (quick starter) auto-add
-  COUNTRIES = {    ## rename to AUTO or BUILTIN_COUNTRIES or QUICK_COUNTRIES - why? why not?
-    ## british isles
-    eng: ['England',          'ENG'],     ## title/name, code
-    sco: ['Scotland',         'SCO'],
-    ie:  ['Ireland',          'IRL'],
-    wal: ['Wales',            'WAL'],
-    nir: ['Northern Ireland', 'NIR'],
-
-    ## central europe
-    at:  ['Austria',      'AUT'],
-    de:  ['Germany',      'GER'],   ## use fifa code or iso? - using fifa for now
-    ch:  ['Switzerland',  'SUI'],
-    li:  ['Liechtenstein', 'LIE'],
-    pl:  ['Poland',       'POL'],
-    cz:  ['Czech Republic', 'CZE'],
-    sk:  ['Slovakia', 'SVK'],
-    hu:  ['Hungary',       'HUN'],
-    si:  ['Slovenia', 'SVN'],
-
-    ## western europe
-    be:  ['Belgium',     'BEL'],
-    nl:  ['Netherlands', 'NED'],
-    lu:  ['Luxembourg',  'LUX'],
-    fr:  ['France',      'FRA'],
-    mc:  ['Monaco',      'MCO'],
-    es:  ['Spain',       'ESP'],
-    ad:  ['Andorra',     'AND'],
-
-    ## northern europe / skandinavia
-    dk:  ['Denmark',      'DEN'],
-    fi:  ['Finland',      'FIN'],
-    no:  ['Norway',       'NOR'],
-    se:  ['Sweden',       'SWE'],
-
-    ## eastern europe
-    ro:  ['Romania',      'ROU'],
-    ru:  ['Russia',       'RUS'],
-    al:  ['Albania',      'ALB'],
-    am:  ['Armenia',      'ARM'],
-    az:  ['Azerbaijan',   'AZE'],
-    ba:  ['Bosnia and Herzegovina', 'BIH'],
-    bg:  ['Bulgaria', 'BUL'],
-    by:  ['Belarus', 'BLR'],
-
-    ## more european
-    cy: ['Cyprus', 'CYP'],
-    is: ['Iceland', 'ISL'],
-    fo: ['Faroe Islands', 'FRO'],
-    ge: ['Georgia', 'GEO'],
-    gi: ['Gibraltar', 'GIB'],
-    gr: ['Greece', 'GRE'],
-    hr: ['Croatia', 'CRO'],
-    it: ['Italy', 'ITA'],
-    sm: ['San Marino', 'SMR'],
-    ee: ['Estonia', 'EST'],
-    lt: ['Lithuania', 'LTU'],
-    lv: ['Latvija', 'LVA'],
-    md: ['Moldova', 'MDA'],
-    me: ['Montenegro', 'MNE'],
-    mk: ['Macedonia', 'MKD'],
-    mt: ['Malta', 'MLT'],
-    pt: ['Portugal', 'POR'],
-    rs: ['Serbia', 'SRB'],
-    tr: ['Turkey', 'TUR'],
-    ua: ['Ukraine', 'UKR'],
-
-
-    ## north america
-    us:  ['United States', 'USA'],
-    mx:  ['Mexico',        'MEX'],
-    ca:  ['Canada',        'CAN'],
-
-    ## caribbean
-    ht:  ['Haiti',               'HAI'],
-    pr:  ['Puerto Rico',         'PUR'],
-    tt:  ['Trinidad and Tobago', 'TRI'],
-
-    ## central america
-    cr:  ['Costa Rica',  'CRC'],
-    gt:  ['Guatemala',   'GUA'],
-    hn:  ['Honduras',    'HON'],
-    ni:  ['Nicaragua',   'NCA'],
-    pa:  ['Panama',      'PAN'],
-    sv:  ['El Salvador', 'SLV'],
-
-
-    ## south america
-    ar:  ['Argentina',   'ARG'],
-    br:  ['Brazil',      'BRA'],
-    bo:  ['Bolivia',     'BOL'],
-    cl:  ['Chile',       'CHI'],
-    co:  ['Colombia',    'COL'],
-    ec:  ['Ecuador',     'ECU'],
-    gy:  ['Guyana',      'GUY'],
-    pe:  ['Peru',        'PER'],
-    py:  ['Paraguay',    'PAR'],
-    uy:  ['Uruguay',     'URU'],
-    ve:  ['Venezuela',   'VEN'],
-
-    ## middle east
-    il:  ['Israel', 'ISR'],
+## built-in countries for (quick starter) auto-add
+
+##
+#  note: use our own (internal) country struct for now - why? why not?
+#    - check that shape/structure/fields/attributes match
+#      the Country struct in sportdb-text (in SportDb::Struct::Country)
+##       and the ActiveRecord model !!!!
+class Country
+  ## note: is read-only/immutable for now - why? why not?
+  ##          add cities (array/list) - why? why not?
+  attr_reader :key, :name, :fifa
+  def initialize( key, name, fifa )
+    @key, @name, @fifa = key, name, fifa
+  end
+end  # class Country
 
-    ## asia
-    cn:  ['China',       'CHN'],
-    jp:  ['Japan',       'JPN'],
-    kz:  ['Kazakhstan',  'KAZ'],
-    kr:  ['South Korea', 'KOR'],
 
-    ## africa
-    eg:  ['Egypt',   'EGY'],
-    ma:  ['Morocco', 'MAR'],
+class CountryIndex
 
-    ## pacific
-    au:  ['Australia',   'AUS'],
-    nz:  ['New Zealand', 'NZL'],
-  }
+  def initialize( recs )
+    @countries         = []
+    @countries_by_code = {}
 
+    add( recs )
+  end
 
-  ##
-  #  note: use our own (internal) country struct for now - why? why not?
-  #    - check that shape/structure/fields/attributes match
-  #      the Country struct in sportdb-text (in SportDb::Struct::Country)
-  ##       and the ActiveRecord model !!!!
-  class Country
-    ## note: is read-only/immutable for now - why? why not?
-    ##          add cities (array/list) - why? why not?
-    attr_reader :key, :name, :code
-    def initialize( key, name, code )
-      @key, @name, @code = key, name, code
+  def add( recs )
+    ###########################################
+    ## auto-fill countries
+    ## pp recs
+    recs.each do |rec|
+      ## rec e.g. { key:'af', fifa:'AFG', name:'Afghanistan'}
+
+      key  = rec[:key]
+      ## note: remove territory of marker e.g. (UK), (US), etc. from name
+      ##    e.g. England (UK)     => England
+      ##         Puerto Rico (US) => Puerto Rico
+      name = rec[:name].sub( /\([A-Z]{2}\)/, '' ).strip
+      fifa = rec[:fifa]
+
+      country = Country.new( key, name, fifa )
+      @countries << country
+
+      ## add codes lookups - key, fifa, ...
+      if @countries_by_code[ country.key ]
+        puts "** !! ERROR !! country code (key) >#{country.key}< already exits!!"
+        exit 1
+      else
+        @countries_by_code[ country.key ] = country
+      end
+
+      ## add fifa code (only) if different from key
+      if country.key != country.fifa.downcase
+        if @countries_by_code[ country.fifa.downcase ]
+          puts "** !! ERROR !! country code (fifa) >#{country.fifa.downcase}< already exits!!"
+          exit 1
+        else
+          @countries_by_code[ country.fifa.downcase ] = country
+        end
+      end
     end
-  end  # class Country
+  end # method initialize
 
-   ## todo/check:  rename to country_mappings - why? why not?
-   ##    or countries_by_code or countries_by_key
-  def countries
-    ## note: convertcountry data to "proper" struct
-    @countries ||= COUNTRIES.each.reduce({}) do |h,(key,value)|
-      ## note: convert key (e.g. :eng to string e.g. 'eng')
-      h[key] = Country.new( key.to_s, value[0], value[1] )
-      h
-    end
-    @countries
+  def []( key )
+    key = key.to_s.downcase   ## allow symbols (and always downcase e.g. AUT to aut etc.)
+    @countries_by_code[ key ]
   end
+end # class CountryIndex
+
 
-end   # class Configuration
 end   # module Import
 end   # module SportDb
-

data/lib/sportdb/config/version.rb

@@ -8,7 +8,7 @@ module Boot      ## note: use a different module than Config to avoid confusion
 
   MAJOR = 0    ## todo: namespace inside version or something - why? why not??
   MINOR = 3
-  PATCH = 4
+  PATCH = 5
   VERSION = [MAJOR,MINOR,PATCH].join('.')
 
   def self.version
@@ -28,9 +28,5 @@ module Boot      ## note: use a different module than Config to avoid confusion
   end
 
 
-#  def self.test_data_dir
-#    "#{root}/test/data"
-#  end
-
 end # module Boot
 end # module SportDb

data/lib/sportdb/config.rb

@@ -6,6 +6,12 @@ require 'date'
 require 'fileutils'
 
 
+## 3rd party gems
+require 'csvreader'
+
+def read_csv( path )
+  CsvHash.read( path, :header_converters => :symbol )
+end
 
 ###
 # our own code
@@ -15,9 +21,10 @@ require 'sportdb/config/season_utils'
 require 'sportdb/config/league_utils'
 require 'sportdb/config/league'
 require 'sportdb/config/league_reader'
-require 'sportdb/config/team_reader'
 
 require 'sportdb/config/countries'
+require 'sportdb/config/clubs'
+require 'sportdb/config/club_reader'
 require 'sportdb/config/config'
 
 

data/test/{test_team_reader.rb → test_club_reader.rb}

@@ -2,15 +2,15 @@
 
 ###
 #  to run use
-#     ruby -I ./lib -I ./test test/test_team_reader.rb
+#     ruby -I ./lib -I ./test test/test_club_reader.rb
 
 
 require 'helper'
 
-class TestTeamReader < MiniTest::Test
+class TestClubReader < MiniTest::Test
 
   def test_parse_at
-    recs = SportDb::Import::TeamReader.parse( <<TXT )
+    recs = SportDb::Import::ClubReader.parse( <<TXT )
 ==================================
 =  Austria (at)
 
@@ -29,7 +29,7 @@ TXT
   end
 
   def test_parse_us
-    recs = SportDb::Import::TeamReader.parse( <<TXT )
+    recs = SportDb::Import::ClubReader.parse( <<TXT )
 ==================================================
 = United States (us)
 
@@ -59,7 +59,7 @@ TXT
 
 
   def test_parse_years
-    recs = SportDb::Import::TeamReader.parse( <<TXT )
+    recs = SportDb::Import::ClubReader.parse( <<TXT )
 = United States (us)
 FC Dallas (1996-),         Frisco        › Texas
 Miami Fusion (1998-2001),  Fort Lauderdale › Florida
@@ -84,7 +84,7 @@ TXT
   end
 
   def test_parse_geos
-    recs = SportDb::Import::TeamReader.parse( <<TXT )
+    recs = SportDb::Import::ClubReader.parse( <<TXT )
 =  England (eng)
 == Greater London
 
@@ -123,7 +123,7 @@ TXT
 
 
   def test_parse_headings
-    recs = SportDb::Import::TeamReader.parse( <<TXT )
+    recs = SportDb::Import::ClubReader.parse( <<TXT )
 ==============
 ====
 ===========
@@ -147,4 +147,4 @@ TXT
     assert_equal 0, recs.size
   end
 
-end # class TestTeamReader
+end # class TestClubReader

data/test/test_clubs.rb

@@ -9,28 +9,45 @@ require 'helper'
 
 class TestClubs < MiniTest::Test
 
+  def test_clubs
+    ## todo/check: move config.clubs_dir to test/helper for global setting - why? why not?
+    SportDb::Import.config.clubs_dir = '../../../openfootball/clubs'
 
-  def match1( txt ) SportDb::Import::Configuration::CLUBS_REGEX1.match( txt ); end
-  def match2( txt ) SportDb::Import::Configuration::CLUBS_REGEX2.match( txt ); end
+    pp SportDb::Import.config.clubs.errors
 
-  def test_find_clubs
-    m = match1( 'de.clubs.txt' )
-    assert_equal 'de', m[:country]
+    SportDb::Import.config.clubs.dump_duplicates
 
-    m = match1( 'deutschland/de.clubs.txt' )
-    assert_equal 'de', m[:country]
+    m = SportDb::Import.config.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 = match2( 'europe/de-deutschland/clubs.txt' )
-    assert_equal 'de', m[:country]
+    c = SportDb::Import.config.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 = match2( 'de-deutschland/clubs.txt' )
-    assert_equal 'de', m[:country]
 
+    m = SportDb::Import.config.clubs.match( 'Arsenal' )
+    assert_equal 3, m.size
 
-    assert match1( 'clubs.txt' ) == nil
-    assert match2( 'clubs.txt' ) == nil
-    assert match1( 'deutschland/clubs.txt' ) == nil
-    assert match2( 'deutschland/clubs.txt' ) == nil
+    m = SportDb::Import.config.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
+
+    m = SportDb::Import.config.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
+
+    m = SportDb::Import.config.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
   end
 
 end # class TestClubs

data/test/test_config.rb

@@ -9,12 +9,15 @@ require 'helper'
 
 class TestConfig < MiniTest::Test
 
-  def test_clubs
-    SportDb::Import.config.clubs_dir = '../../../openfootball/clubs'
-
-    SportDb::Import.config.teams
-
-    assert true ## assume ok if we get here
-  end  # method test_clubs
+  def match( txt ) SportDb::Import::Configuration::CLUBS_REGEX.match( txt ); end
+
+  def test_find_clubs
+      assert match( 'de.clubs.txt' )
+      assert match( 'deutschland/de.clubs.txt' )
+      assert match( 'europe/de-deutschland/clubs.txt' )
+      assert match( 'de-deutschland/clubs.txt' )
+      assert match( 'clubs.txt' )
+      assert match( 'deutschland/clubs.txt' )
+  end
 
 end # class TestConfig

data/test/test_countries.rb

@@ -9,18 +9,34 @@ require 'helper'
 
 class TestCountries < MiniTest::Test
 
+  def test_read_countries
+    recs = read_csv( "#{SportDb::Boot.data_dir}/world/countries.txt" )
+    ## pp recs
+
+    assert_equal  [{ key:'af', fifa:'AFG', name:'Afghanistan'},
+                   { key:'al', fifa:'ALB', name:'Albania'}], recs[0..1]
+  end
+
+
   def test_countries
-    pp SportDb::Import.config.countries
+    ## pp SportDb::Import.config.countries
 
     eng = SportDb::Import.config.countries[:eng]
     assert_equal 'eng',      eng.key
     assert_equal 'England',  eng.name
-    assert_equal 'ENG',      eng.code
+    assert_equal 'ENG',      eng.fifa
 
     at  = SportDb::Import.config.countries[:at]
     assert_equal 'at',       at.key
     assert_equal 'Austria',  at.name
-    assert_equal 'AUT',      at.code
+    assert_equal 'AUT',      at.fifa
+
+    aut  = SportDb::Import.config.countries[:aut]
+    assert_equal 'at',       aut.key
+    assert_equal 'Austria',  aut.name
+    assert_equal 'AUT',      aut.fifa
+
+    assert aut == at
   end
 
 end # class TestCountries

metadata

@@ -1,15 +1,29 @@
 --- !ruby/object:Gem::Specification
 name: sportdb-config
 version: !ruby/object:Gem::Version
-  version: 0.3.4
+  version: 0.3.5
 platform: ruby
 authors:
 - Gerald Bauer
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-07-21 00:00:00.000000000 Z
+date: 2019-07-28 00:00:00.000000000 Z
 dependencies:
+- !ruby/object:Gem::Dependency
+  name: csvreader
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: 1.2.4
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: 1.2.4
 - !ruby/object:Gem::Dependency
   name: rdoc
   requirement: !ruby/object:Gem::Requirement
@@ -46,38 +60,41 @@ extra_rdoc_files:
 - HISTORY.md
 - Manifest.txt
 - README.md
-- config/geos/eng.txt
 - config/leagues/eng.txt
 - config/leagues/fr.txt
 - config/leagues/gr.txt
 - config/leagues/sco.txt
+- config/world/countries.txt
+- config/world/eng.txt
 files:
 - HISTORY.md
 - Manifest.txt
 - README.md
 - Rakefile
-- config/geos/eng.txt
 - config/leagues/eng.txt
 - config/leagues/fr.txt
 - config/leagues/gr.txt
 - config/leagues/sco.txt
+- config/world/countries.txt
+- config/world/eng.txt
 - lib/sportdb/config.rb
+- lib/sportdb/config/club_reader.rb
+- lib/sportdb/config/clubs.rb
 - lib/sportdb/config/config.rb
 - lib/sportdb/config/countries.rb
 - lib/sportdb/config/league.rb
 - lib/sportdb/config/league_reader.rb
 - lib/sportdb/config/league_utils.rb
 - lib/sportdb/config/season_utils.rb
-- lib/sportdb/config/team_reader.rb
 - lib/sportdb/config/version.rb
 - test/helper.rb
+- test/test_club_reader.rb
 - test/test_clubs.rb
 - test/test_config.rb
 - test/test_countries.rb
 - test/test_league_reader.rb
 - test/test_league_utils.rb
 - test/test_season_utils.rb
-- test/test_team_reader.rb
 homepage: https://github.com/sportdb/sport.db
 licenses:
 - Public Domain