sportdb-formats 1.0.3 → 1.0.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 4a27c364d2003ece8da886c892d34d595ffcffcc
-  data.tar.gz: 81530dfec5decf5d9476a3f2be90bcbeb0195824
+  metadata.gz: 5ccd241256a964b430d9c9aa515901641ed5a0e2
+  data.tar.gz: 1d997241f388000d7b2c19e26a92c9ba39c8e172
 SHA512:
-  metadata.gz: 4a2aabb65968939d55f2000c64e4ac5df4cbedfe6b3786fd12ff86d26bd5e6af12c32ea129b59a7ddf30736afa6f093e0e4f2e97f2c6f0b23bffd6901a83ca91
-  data.tar.gz: 63f4ca42ebb537173334a931a02331b8cbdc27d9e953052c8126d9e683e54a3103eb496103fea400283c6b0b08fa7c6e738466f215e26f9567a3b822a80c172d
+  metadata.gz: 1d7d1e6b71009ec598f01f59ead597269ea3c4b48bc3e2f5be0a7257879114f2b3110937dfb23c153962cd7dd0a61e23d0050a8eb40c6652acf34a8a1720071c
+  data.tar.gz: 392a8eaf8c0d1f14b021a45ca39ce491f8edceac0cf9ba33ffde920e638a2cf1fd2adf836e596dc2100c233f5ca1fe8aadf47526fb1837de855acaf84974b19a

data/lib/sportdb/formats/country/country_index.rb

@@ -30,11 +30,11 @@ class CountryIndex
     ## auto-fill countries
     ## pp recs
     recs.each do |rec|
-      ## rec e.g. { key:'af', fifa:'AFG', name:'Afghanistan'}
+      ## rec e.g. { key:'af', code:'AFG', name:'Afghanistan'}
 
       @countries << rec
 
-      ## add codes lookups - key, fifa, ...
+      ## add codes lookups - key, code, ...
       if @countries_by_code[ rec.key ]
         puts "** !! ERROR !! country code (key) >#{rec.key}< already exits!!"
         exit 1
@@ -42,13 +42,13 @@ class CountryIndex
         @countries_by_code[ rec.key ] = rec
       end
 
-      ## add fifa code (only) if different from key
-      if rec.key != rec.fifa.downcase
-        if @countries_by_code[ rec.fifa.downcase ]
-          puts "** !! ERROR !! country code (fifa) >#{rec.fifa.downcase}< already exits!!"
+      ## add  code (only) if different from key
+      if rec.key != rec.code.downcase
+        if @countries_by_code[ rec.code.downcase ]
+          puts "** !! ERROR !! country code  >#{rec.code.downcase}< already exits!!"
           exit 1
         else
-          @countries_by_code[ rec.fifa.downcase ] = rec
+          @countries_by_code[ rec.code.downcase ] = rec
         end
       end
 

data/lib/sportdb/formats/country/country_reader.rb

@@ -43,11 +43,31 @@ def parse
         #   e.g. East Germany        (-1989)  => East Germany (-1989)
         values = values.map { |value| value.strip.gsub( /[ \t]+/, ' ' ) }
         last_country.alt_names += values
+      elsif line =~ /^-[ ]*(\d{4})
+                        [ ]+
+                       (.+)$
+                    /x     ## check for historic lines e.g. -1989
+         year   = $1.to_i
+         parts  = $2.split( /=>|⇒/ )
+         values = parts[0].split( ',' )
+         values = values.map { |value| value.strip.gsub( /[ \t]+/, ' ' ) }
+
+         name = values[0]
+         code = values[1]
+
+         last_country = country = Country.new( name: "#{name} (-#{year})",
+                                               code: code )
+         country.alt_names << name    ## note: do NOT forget - add name without year to alt_names!!!
+
+         countries << country
+         ## todo/fix: add reference to country today (in parts[1] !!!!)
       else
         ## assume "regular" line
         ##  check if starts with id  (todo/check: use a more "strict"/better regex capture pattern!!!)
         ##   note: allow country codes upto 4 (!!) e.g. Northern Cyprus
-        if line =~ /^([a-z]{2,4})[ ]+(.+)$/
+        if line =~ /^([a-z]{2,4})
+                        [ ]+
+                       (.+)$/x
           key    = $1
           values = $2.split( ',' )
           ## strip and squish (white)spaces
@@ -60,14 +80,14 @@ def parse
           geos = split_geo( values[0] )
           name = geos[0]    ## note: ignore all other geos for now
 
-          ##   note: allow fifa country codes upto 4 (!!) e.g. Northern Cyprus
-          fifa = if values[1] && values[1] =~ /^[A-Z]{3,4}$/   ## note: also check format
+          ##   note: allow country codes up to 4 (!!) e.g. Northern Cyprus
+          code = if values[1] && values[1] =~ /^[A-Z]{3,4}$/   ## note: also check format
                    values[1]
                  else
                    if values[1]
-                     puts "** !!! ERROR !!! wrong fifa code format >#{values[1]}<; expected three (or four)-letter all up-case"
+                     puts "** !!! ERROR !!! wrong code format >#{values[1]}<; expected three (or four)-letter all up-case"
                    else
-                     puts "** !!! ERROR !!! missing fifa code for (canonical) country name"
+                     puts "** !!! ERROR !!! missing code for (canonical) country name"
                    end
                    exit 1
                  end
@@ -80,7 +100,7 @@ def parse
 
           last_country = country = Country.new( key: key,
                                                 name: name,
-                                                fifa: fifa,
+                                                code: code,
                                                 tags: tags )
           countries << country
         else

data/lib/sportdb/formats/league/league_reader.rb

@@ -113,9 +113,9 @@ def parse
             ##             add a list of (auto-)excluded country codes with conflicts? why? why not?
             ##                 cl - a) Chile  b) Champions League
             alt_names_auto << "#{country.key.upcase}"   if league_key == '1'   ## add shortcut for top level 1 (just country key)
-            if country.key.upcase != country.fifa
-              alt_names_auto << "#{country.fifa} #{league_key.upcase.gsub('.', ' ')}"
-              alt_names_auto << "#{country.fifa}"    if league_key == '1'   ## add shortcut for top level 1 (just country key)
+            if country.key.upcase != country.code
+              alt_names_auto << "#{country.code} #{league_key.upcase.gsub('.', ' ')}"
+              alt_names_auto << "#{country.code}"    if league_key == '1'   ## add shortcut for top level 1 (just country key)
             end
             alt_names_auto << "#{country.name} #{league_key}"  if league_key =~ /^[0-9]+$/   ## if all numeric e.g. add Austria 1 etc.
           else   ## assume int'l (no country) e.g. champions league, etc.

data/lib/sportdb/formats/package.rb

@@ -4,31 +4,50 @@ module SportDb
 
     ## todo/fix:   make all regexes case-insensitive with /i option - why? why not?
     ##                e.g. .TXT and .txt
+    ##   yes!! use /i option!!!!!
 
     CONF_RE = %r{  (?: ^|/ )               # beginning (^) or beginning of path (/)
         \.conf\.txt$
     }x
 
+    ## leagues.txt or leagues_en.txt
+    ##   remove support for en.leagues.txt - why? why not?
     LEAGUES_RE = %r{  (?: ^|/ )               # beginning (^) or beginning of path (/)
        (?: [a-z]{1,4}\. )?   # optional country code/key e.g. eng.clubs.wiki.txt
-        leagues\.txt$
+        leagues
+          (?:_[a-z0-9_-]+)?
+        \.txt$
     }x
 
+    ## clubs.txt or clubs_en.txt
+    ##   remove support for en.clubs.txt - why? why not?
     CLUBS_RE = %r{  (?: ^|/ )               # beginning (^) or beginning of path (/)
        (?: [a-z]{1,4}\. )?   # optional country code/key e.g. eng.clubs.txt
-        clubs\.txt$
+        clubs
+          (?:_[a-z0-9_-]+)?
+        \.txt$
     }x
 
     CLUBS_WIKI_RE = %r{  (?:^|/)               # beginning (^) or beginning of path (/)
         (?:[a-z]{1,4}\.)?   # optional country code/key e.g. eng.clubs.wiki.txt
-       clubs\.wiki\.txt$
+       clubs
+         (?:_[a-z0-9_-]+)?
+       \.wiki\.txt$
     }x
 
     CLUB_PROPS_RE = %r{  (?: ^|/ )               # beginning (^) or beginning of path (/)
       (?: [a-z]{1,4}\. )?   # optional country code/key e.g. eng.clubs.props.txt
-        clubs\.props\.txt$
+        clubs
+          (?:_[a-z0-9_-]+)?
+        \.props\.txt$
     }x
 
+    ##  teams.txt or teams_history.txt
+    TEAMS_RE = %r{  (?: ^|/ )               # beginning (^) or beginning of path (/)
+                      teams
+                        (?:_[a-z0-9_-]+)?
+                      \.txt$
+    }x
 
     ### season folder:
     ##            e.g. /2019-20   or
@@ -36,7 +55,7 @@ module SportDb
     ##                 /2016--france
     SEASON_RE = %r{ (?:
                        \d{4}-\d{2}
-                     | \d{4}(--[^/]+)?
+                     | \d{4}(--[a-z0-9_-]+)?
                     )
                   }x
     SEASON = SEASON_RE.source    ## "inline" helper for embedding in other regexes - keep? why? why not?
@@ -74,6 +93,8 @@ module SportDb
    end
 
 
+   def self.find_teams( path, pattern: TEAMS_RE )  find( path, pattern ); end
+   def self.match_teams( path )  TEAMS_RE.match( path ); end
 
    def self.find_clubs( path, pattern: CLUBS_RE )            find( path, pattern ); end
    def self.find_clubs_wiki( path, pattern: CLUBS_WIKI_RE )  find( path, pattern ); end
@@ -98,6 +119,9 @@ module SportDb
    ## add match_match and match_match_csv  - why? why not?
 
    class << self
+     alias_method :match_teams?, :match_teams
+     alias_method :teams?,       :match_teams
+
      alias_method :match_clubs?, :match_clubs
      alias_method :clubs?,       :match_clubs
 

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

@@ -11,11 +11,14 @@ class Country
 
   ## note: is read-only/immutable for now - why? why not?
   ##          add cities (array/list) - why? why not?
-  attr_reader   :key, :name, :fifa, :tags
+  attr_reader   :key, :name, :code, :tags
   attr_accessor :alt_names
 
-  def initialize( key:, name:, fifa:, tags: [] )
-    @key, @name, @fifa = key, name, fifa
+  def initialize( key: nil, name:, code:, tags: [] )
+    ## note: auto-generate key "on-the-fly" if missing for now - why? why not?
+    ## note: quick hack - auto-generate key, that is, remove all non-ascii chars and downcase
+    @key = key || name.downcase.gsub( /[^a-z]/, '' )
+    @name, @code = name, code
     @alt_names      = []
     @tags           = tags
   end

data/lib/sportdb/formats/version.rb

@@ -7,7 +7,7 @@ module Formats
 
   MAJOR = 1    ## todo: namespace inside version or something - why? why not??
   MINOR = 0
-  PATCH = 3
+  PATCH = 4
   VERSION = [MAJOR,MINOR,PATCH].join('.')
 
   def self.version

data/test/test_country_index.rb

@@ -16,12 +16,12 @@ class TestCountryIndex < MiniTest::Test
     eng = countries[:eng]
     assert_equal 'eng',      eng.key
     assert_equal 'England',  eng.name
-    assert_equal 'ENG',      eng.fifa
+    assert_equal 'ENG',      eng.code
 
     at  = countries[:at]
     assert_equal 'at',                at.key
     assert_equal 'Austria',           at.name
-    assert_equal 'AUT',               at.fifa
+    assert_equal 'AUT',               at.code
     assert_equal ['Österreich [de]'], at.alt_names
 
     assert at == countries['AT']
@@ -48,13 +48,13 @@ class TestCountryIndex < MiniTest::Test
     assert at == countries.parse( 'Österreich • Austria' )
     assert at == countries.parse( 'Austria' )
     assert at == countries.parse( 'at' )    ## (iso alpha2) country code
-    assert at == countries.parse( 'AUT' )   ## fifa code
+    assert at == countries.parse( 'AUT' )   ## (fifa) country code
 
 
     de  = countries[:de]
     assert_equal 'de',                 de.key
     assert_equal 'Germany',            de.name
-    assert_equal 'GER',                de.fifa
+    assert_equal 'GER',                de.code
     assert_equal ['Deutschland [de]'], de.alt_names
 
     assert de == countries.parse( 'Deutschland (de) • Germany' )

data/test/test_country_reader.rb

@@ -16,12 +16,12 @@ class TestCountryReader < MiniTest::Test
     assert_equal 232, recs.size
 
     assert_equal 'Albania',        recs[0].name
-    assert_equal 'ALB',            recs[0].fifa
+    assert_equal 'ALB',            recs[0].code
     assert_equal 'al',             recs[0].key
     assert_equal ['fifa', 'uefa'], recs[0].tags
 
     assert_equal 'Andorra',        recs[1].name
-    assert_equal 'AND',            recs[1].fifa
+    assert_equal 'AND',            recs[1].code
     assert_equal 'ad',             recs[1].key
     assert_equal ['fifa', 'uefa'], recs[1].tags
   end
@@ -44,16 +44,46 @@ TXT
 
     assert_equal 4, recs.size
     assert_equal 'Afghanistan',    recs[0].name
-    assert_equal 'AFG',            recs[0].fifa
+    assert_equal 'AFG',            recs[0].code
     assert_equal 'af',             recs[0].key
     assert_equal [],               recs[0].alt_names
     assert_equal ['fifa', 'afc'],  recs[0].tags
 
     assert_equal 'American Samoa', recs[3].name
-    assert_equal 'ASA',            recs[3].fifa
+    assert_equal 'ASA',            recs[3].code
     assert_equal 'as',             recs[3].key
     assert_equal ['Am. Samoa'],    recs[3].alt_names
     assert_equal [],               recs[3].tags
   end
 
+  def test_parse_historic
+    recs = SportDb::Import::CountryReader.parse( <<TXT )
+###########################################
+# Former national teams
+#   with former FIFA country codes etc.
+-1992   Czechoslovakia, TCH      ⇒  Czech Republic
+
+-1991   Soviet Union,   URS      ⇒  Russia
+
+-1989   West Germany,   FRG      =>  Germany
+-1989   East Germany,   GDR      =>  Germany
+TXT
+
+    pp recs
+
+    assert_equal 4, recs.size
+    assert_equal 'Czechoslovakia (-1992)', recs[0].name
+    assert_equal 'TCH',                    recs[0].code
+    assert_equal 'czechoslovakia',         recs[0].key
+    assert_equal ['Czechoslovakia'],       recs[0].alt_names
+    assert_equal [],                       recs[0].tags
+
+    assert_equal 'East Germany (-1989)',   recs[3].name
+    assert_equal 'GDR',                    recs[3].code
+    assert_equal 'eastgermany',            recs[3].key
+    assert_equal ['East Germany'],         recs[3].alt_names
+    assert_equal [],                       recs[3].tags
+  end
+
+
 end # class TestCountryReader

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: sportdb-formats
 version: !ruby/object:Gem::Version
-  version: 1.0.3
+  version: 1.0.4
 platform: ruby
 authors:
 - Gerald Bauer
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-05-16 00:00:00.000000000 Z
+date: 2020-05-18 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: alphabets