sportdb-config 0.3.2 → 0.3.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 955375833b909f386106aedfc6726b15454819dc
-  data.tar.gz: 82508fd16aa4c60a4e9654aaaa604ed71d3494d7
+  metadata.gz: 2a4445ef87f3b56b3ef17a685cdd0e1b2b576220
+  data.tar.gz: 508640a8a0a0e2519892d15d1ba7065e71738eea
 SHA512:
-  metadata.gz: faa819f096e961388cb7fa5ef0627d2fb754ccba38d4c598023c5179cb687b8fb298a033e9af0ec4b502e78d2b64c1bed370cda41411d7cac2703300860b140f
-  data.tar.gz: b03b985e5ec0ac0caf36adef463c30cccdb16b245365889f16ad8d67fcf2fa8ee446c1ce7ca6296678382cad547beb25a0adc2df402c63d456cae9527d4fd2fa
+  metadata.gz: 44cc3d79fcb8a16c34fe27c8ecdfb158a99ef308bd9a17901fd85c0849f1bddef3882c3878161738d1cb23f2a068d8e706bce5f28fb5b21bc2df94b427316447
+  data.tar.gz: d0e784ae220704714ebbd3187246bbd44b87c6dd55a01a6d79812d959fa2ad2772b788e4c5f910581fb3db7dfb82a72339d1cd0b18593579c932b46b71701f08

data/Manifest.txt

@@ -9,6 +9,7 @@ config/leagues/gr.txt
 config/leagues/sco.txt
 lib/sportdb/config.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
@@ -17,6 +18,7 @@ lib/sportdb/config/team_reader.rb
 lib/sportdb/config/version.rb
 test/helper.rb
 test/test_config.rb
+test/test_countries.rb
 test/test_league_reader.rb
 test/test_league_utils.rb
 test/test_season_utils.rb

data/lib/sportdb/config.rb

@@ -16,6 +16,8 @@ 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/config'
 
 

data/lib/sportdb/config/config.rb

@@ -10,7 +10,6 @@ class Configuration
     @errors = []     ## make parsing errors "global" for now
   end
 
-
   def team_mappings
     read_teams()        if @team_mappings.nil?
     @team_mappings

data/lib/sportdb/config/countries.rb

@@ -0,0 +1,200 @@
+# encoding: utf-8
+
+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'],
+
+    ## central europe
+    at:  ['Austria',      'AUT'],
+    de:  ['Germany',      'GER'],   ## use fifa code or iso? - using fifa for now
+    ch:  ['Switzerland',  'SWZ'],
+    pl:  ['Poland',       'POL'],
+
+    ## western europe
+    fr:  ['France',      'FRA'],
+    es:  ['Spain',       'ESP'],
+    be:  ['Belgium',     '' ],
+    nl:  ['Netherlands', '' ],
+
+    ## northern europe / skandinavia
+    dk:  ['Denmark',      'DNK'],
+    fi:  ['Finland',      'FIN'],
+    no:  ['Norway',       'NOR'],
+    se:  ['Sweden',       'SWE'],
+
+    ## eastern europe
+    ro:  ['Romania',      'ROU'],
+    ru:  ['Russia',       'RUS'],
+
+    ## north america
+    us:  ['United States', 'USA'],
+    mx:  ['Mexico',        'MEX'],
+
+    ## caribbean
+    ht:  ['Haiti',               ''],
+    pr:  ['Puerto Rico',         ''],
+    tt:  ['Trinidad and Tobago', ''],
+
+    ## central america
+    cr:  ['Costa Rica',  ''],
+    gt:  ['Guatemala',   ''],
+    hn:  ['Honduras',    ''],
+    ni:  ['Nicaragua',   ''],
+    pa:  ['Panama',      ''],
+    sv:  ['El Salvador', ''],
+
+    ## south america
+    ar:  ['Argentina',   'ARG'],
+    br:  ['Brazil',      'BRA'],
+    bo:  ['Bolivia',     ''],
+    cl:  ['Chile',       ''],
+    co:  ['Colombia',    ''],
+    ec:  ['Ecuador',     ''],
+    gy:  ['Guyana',      ''],
+    pe:  ['Peru',        ''],
+    py:  ['Paraguay',    ''],
+    uy:  ['Uruguay',     ''],
+    ve:  ['Venezuela',   ''],
+
+    ## asia
+    cn:  ['China',       'CHN'],
+    jp:  ['Japan',       'JPN'],
+    kz:  ['Kazakhstan',  '' ],
+    kr:  ['South Korea', '' ],
+
+    ## africa
+    eg:  ['Egypt',   ''],
+    ma:  ['Morocco', ''],
+  }
+
+## add more countries - see/check datasets !!
+=begin
+[["eg", "../../../openfootball/clubs/africa/egypt/eg.clubs.txt"],
+ ["ma", "../../../openfootball/clubs/africa/morocco/ma.clubs.txt"],
+ ["cn", "../../../openfootball/clubs/asia/china/cn.clubs.txt"],
+ ["jp", "../../../openfootball/clubs/asia/japan/jp.clubs.txt"],
+ ["kz", "../../../openfootball/clubs/asia/kazakhstan/kz.clubs.txt"],
+ ["kr", "../../../openfootball/clubs/asia/south-korea/kr.clubs.txt"],
+ ["ht", "../../../openfootball/clubs/caribbean/haiti/ht.clubs.txt"],
+ ["pr", "../../../openfootball/clubs/caribbean/puerto-rico/pr.clubs.txt"],
+ ["tt",
+  "../../../openfootball/clubs/caribbean/trinidad-n-tobago/tt.clubs.txt"],
+ ["cr", "../../../openfootball/clubs/central-america/costa-rica/cr.clubs.txt"],
+ ["gt", "../../../openfootball/clubs/central-america/guatemala/gt.clubs.txt"],
+ ["hn", "../../../openfootball/clubs/central-america/hn-honduras/clubs.txt"],
+ ["ni", "../../../openfootball/clubs/central-america/ni-nicaragua/clubs.txt"],
+ ["pa", "../../../openfootball/clubs/central-america/pa-panama/clubs.txt"],
+ ["sv",
+  "../../../openfootball/clubs/central-america/sv-el-salvador/clubs.txt"],
+ ["ad", "../../../openfootball/clubs/europe/ad-andorra/clubs.txt"],
+ ["al", "../../../openfootball/clubs/europe/al-albania/clubs.txt"],
+ ["am", "../../../openfootball/clubs/europe/am-armenia/clubs.txt"],
+ ["at", "../../../openfootball/clubs/europe/at-austria/clubs.txt"],
+ ["az", "../../../openfootball/clubs/europe/az-azerbaijan/clubs.txt"],
+ ["ba",
+  "../../../openfootball/clubs/europe/ba-bosnia-n-herzegovina/clubs.txt"],
+ ["be", "../../../openfootball/clubs/europe/be-belgium/clubs.txt"],
+ ["bg", "../../../openfootball/clubs/europe/bg-bulgaria/clubs.txt"],
+ ["by", "../../../openfootball/clubs/europe/by-belarus/clubs.txt"],
+ ["ch",
+  "../../../openfootball/clubs/europe/ch-confoederatio-helvetica/clubs.txt"],
+ ["cy", "../../../openfootball/clubs/europe/cy-cyprus/clubs.txt"],
+ ["cz", "../../../openfootball/clubs/europe/cz-czech-republic/clubs.txt"],
+ ["de", "../../../openfootball/clubs/europe/de-deutschland/clubs.txt"],
+ ["dk", "../../../openfootball/clubs/europe/dk-denmark/clubs.txt"],
+ ["ee", "../../../openfootball/clubs/europe/ee-estonia/clubs.txt"],
+ ["eng", "../../../openfootball/clubs/europe/eng-england/clubs.txt"],
+ ["es", "../../../openfootball/clubs/europe/es-espana/clubs.txt"],
+ ["fi", "../../../openfootball/clubs/europe/fi-finland/clubs.txt"],
+ ["fo", "../../../openfootball/clubs/europe/fo-faroe-islands/clubs.txt"],
+ ["fr", "../../../openfootball/clubs/europe/fr-france/clubs.txt"],
+ ["ge", "../../../openfootball/clubs/europe/ge-georgia/clubs.txt"],
+ ["gi", "../../../openfootball/clubs/europe/gi-gibraltar/clubs.txt"],
+ ["gr", "../../../openfootball/clubs/europe/gr-greece/clubs.txt"],
+ ["hr", "../../../openfootball/clubs/europe/hr-croatia/clubs.txt"],
+ ["hu", "../../../openfootball/clubs/europe/hu-hungary/clubs.txt"],
+ ["ie", "../../../openfootball/clubs/europe/ie-ireland/clubs.txt"],
+ ["is", "../../../openfootball/clubs/europe/is-iceland/clubs.txt"],
+ ["it", "../../../openfootball/clubs/europe/it-italy/clubs.txt"],
+ ["li", "../../../openfootball/clubs/europe/li-liechtenstein/clubs.txt"],
+ ["lt", "../../../openfootball/clubs/europe/lt-lithuania/clubs.txt"],
+ ["lu", "../../../openfootball/clubs/europe/lu-luxembourg/clubs.txt"],
+ ["lv", "../../../openfootball/clubs/europe/lv-latvija/clubs.txt"],
+ ["mc", "../../../openfootball/clubs/europe/mc-monaco/clubs.txt"],
+ ["md", "../../../openfootball/clubs/europe/md-moldova/clubs.txt"],
+ ["me", "../../../openfootball/clubs/europe/me-montenegro/clubs.txt"],
+ ["mk", "../../../openfootball/clubs/europe/mk-macedonia/clubs.txt"],
+ ["mt", "../../../openfootball/clubs/europe/mt-malta/clubs.txt"],
+ ["nir", "../../../openfootball/clubs/europe/nir-northern-ireland/clubs.txt"],
+ ["nl", "../../../openfootball/clubs/europe/nl-netherlands/clubs.txt"],
+ ["no", "../../../openfootball/clubs/europe/no-norway/clubs.txt"],
+ ["pl", "../../../openfootball/clubs/europe/pl-poland/clubs.txt"],
+ ["pt", "../../../openfootball/clubs/europe/pt-portugal/clubs.txt"],
+ ["ro", "../../../openfootball/clubs/europe/ro-romania/clubs.txt"],
+ ["rs", "../../../openfootball/clubs/europe/rs-serbia/clubs.txt"],
+ ["ru", "../../../openfootball/clubs/europe/ru-russia/clubs.txt"],
+ ["sco", "../../../openfootball/clubs/europe/sco-scotland/clubs.txt"],
+ ["se", "../../../openfootball/clubs/europe/se-sweden/clubs.txt"],
+ ["si", "../../../openfootball/clubs/europe/si-slovenia/clubs.txt"],
+ ["sk", "../../../openfootball/clubs/europe/sk-slovakia/clubs.txt"],
+ ["sm", "../../../openfootball/clubs/europe/sm-san-marino/clubs.txt"],
+ ["tr", "../../../openfootball/clubs/europe/tr-turkey/clubs.txt"],
+ ["ua", "../../../openfootball/clubs/europe/ua-ukraine/clubs.txt"],
+ ["wal", "../../../openfootball/clubs/europe/wal-wales/clubs.txt"],
+ ["il", "../../../openfootball/clubs/middle-east/il-israel/clubs.txt"],
+ ["ca", "../../../openfootball/clubs/north-america/ca-canada/clubs.txt"],
+ ["mx", "../../../openfootball/clubs/north-america/mx-mexico/clubs.txt"],
+ ["us",
+  "../../../openfootball/clubs/north-america/us-united-states/clubs.txt"],
+ ["au", "../../../openfootball/clubs/pacific/australia/au.clubs.txt"],
+ ["nz", "../../../openfootball/clubs/pacific/new-zealand/nz.clubs.txt"],
+ ["ar", "../../../openfootball/clubs/south-america/argentina/ar.clubs.txt"],
+ ["bo", "../../../openfootball/clubs/south-america/bolivia/bo.clubs.txt"],
+ ["br", "../../../openfootball/clubs/south-america/brazil/br.clubs.txt"],
+ ["cl", "../../../openfootball/clubs/south-america/chile/cl.clubs.txt"],
+ ["co", "../../../openfootball/clubs/south-america/co-colombia/clubs.txt"],
+ ["ec", "../../../openfootball/clubs/south-america/ec-ecuador/clubs.txt"],
+ ["gy", "../../../openfootball/clubs/south-america/gy-guyana/clubs.txt"],
+ ["pe", "../../../openfootball/clubs/south-america/pe-peru/clubs.txt"],
+ ["py", "../../../openfootball/clubs/south-america/py-paraguay/clubs.txt"],
+ ["uy", "../../../openfootball/clubs/south-america/uy-uruguay/clubs.txt"],
+ ["ve", "../../../openfootball/clubs/south-america/ve-venezuela/clubs.txt"]]
+=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
+    end
+  end  # class Country
+
+   ## 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
+  end
+
+end   # class Configuration
+end   # module Import
+end   # module SportDb

data/lib/sportdb/config/team_reader.rb

@@ -16,15 +16,14 @@ class Team
   attr_accessor :name, :alt_names, :year, :ground, :city
 
   ## more attribs - todo/fix - also add "upstream" to struct & model!!!!!
-  attr_accessor :district, :geos, :year_end
+  attr_accessor :district, :geos, :year_end, :country
 
-  def historic?()  @year_end.nil? == false; end
+  def historic?()  @year_end ? true : false; end
   alias_method  :past?, :historic?
 
 
   def initialize
     @alt_names = []
-    @year_end  = nil
   end
 
 
@@ -42,6 +41,7 @@ end
 def self.parse( txt )
   recs = []
   last_rec  = nil
+  headings = []   ## headings stack
 
   txt.each_line do |line|
     line = line.strip
@@ -69,7 +69,64 @@ def self.parse( txt )
        heading        = $2.strip
 
        puts "heading #{heading_level} >#{heading}<"
-       ## skip heading for now
+
+       ## 1) first pop headings if present
+       while headings.size+1 > heading_level
+         headings.pop
+       end
+
+       ## 2) add missing (hierarchy) level if
+       while headings.size+1 < heading_level
+         ##  todo/fix: issue warning about "skipping" hierarchy level
+         puts "!!! warn [team reader] - skipping hierarchy level in headings "
+         headings.push( nil )
+       end
+
+       if heading =~ /^\?+$/    ## note: use ? or ?? or ?? to reset level to nil
+         ## keep level empty
+       else
+
+         ## quick hack:   if level is 1 assume country for now
+         ##                 and extract country code e.g.
+         ##                    Austria (at) => at
+         if heading_level == 1
+             if heading =~ /\(([a-z]{2,3})\)/
+               country_code = $1
+
+               ## check country code - MUST exist for now!!!!
+               country = SportDb::Import.config.countries[ country_code.to_sym ]
+               if country.nil?
+                 puts "!!! error [team reader] - unknown country with code >#{country_code}< - sorry - add country to config to fix"
+                 exit 1
+               end
+
+               headings.push( country_code )
+             else
+               puts "!!! error - heading level 1 - missing country code - >#{heading}<"
+               exit 1
+             end
+         else
+          ## quick hack:
+          ##   remove known fill/dummy words incl:
+          ##     Provincia San Juan  =>  San Juan   (see argentina, for example)
+          ##
+          ##   use geo tree long term with alternative names - why? why not?
+           words = ['Provincia']
+           words.each { |word| heading = heading.gsub( word, '' ) }
+           heading = heading.strip
+
+           headings.push( heading )
+         end
+
+         ## assert that hierarchy level is ok
+         if headings.size != heading_level
+           puts "!!! error - headings hierarchy/stack out of order - #{heading.size}<=>#{heading_level}"
+           exit 1
+         end
+       end
+
+       pp headings
+
     elsif line.start_with?( '|' )
       ## assume continuation with line of alternative names
       ##  note: skip leading pipe
@@ -152,10 +209,8 @@ def self.parse( txt )
         else
           ## assume city / geo tree
           ## split into geo tree
-          geos = value.split( /[<>‹›]/ )   ## note: allow > < or › ‹
-          geos = geos.map { |geo| geo.strip }   ## remove all whitespaces
-
-          city     = geos[0]
+          geos = split_geo( value )
+          city = geos[0]
           ## check for "embedded" district e.g. London (Fulham) or Hamburg (St. Pauli) etc.
           if city =~ /\((.+?)\)/   ## note: use non-greedy (?) match
             rec.district  = $1.strip
@@ -168,6 +223,37 @@ def self.parse( txt )
              rec.geos = geos[1..-1]
           end
         end
+      end  ## while values
+
+
+      ###############
+      ## use headings text for geo tree
+
+      ## 1) add country if present
+      if headings.size > 0 && headings[0]
+        country = SportDb::Import.config.countries[ headings[0].to_sym ]
+        rec.country = country
+      else
+        ## make it an error - why? why not?
+        puts "!!! warn - country missing in headings hierarchy"
+      end
+
+      ## 2) check geo tree with headings hierarchy
+      if headings.size > 1 && headings[1]
+         geos = split_geo( headings[1] )
+         if rec.geos
+           if rec.geos[0] != geos[0]
+             puts "!!! error - geo tree - headings mismatch >#{rec.geos[0]}< <=> >#{geos[0]}<"
+             exit 1
+           end
+           if rec.geos[1] && rec.geos[1] != geos[1]   ## check optional 2nd level too
+             puts "!!! error - geo tree - headings mismatch >#{rec.geos[1]}< <=> >#{geos[1]}<"
+             exit 1
+           end
+         else
+           ## add missing region (state/province) from headings hierarchy
+           rec.geos = geos
+         end
       end
 
       last_rec = rec
@@ -191,6 +277,20 @@ def self.parse( txt )
   end  # each_line
   recs
 end  # method read
+
+###  helpers
+def self.split_geo( str )
+  ## assume city / geo tree
+  ##  strip and squish (white)spaces
+  #   e.g. León     › Guanajuato     => León › Guanajuato
+  str = str.strip.gsub( /[ \t]+/, ' ' )
+
+  ## split into geo tree
+  geos = str.split( /[<>‹›]/ )   ## note: allow > < or › ‹
+  geos = geos.map { |geo| geo.strip }   ## remove all whitespaces
+  geos
+end
+
 end  # class TeamReader
 
 

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 = 2
+  PATCH = 3
   VERSION = [MAJOR,MINOR,PATCH].join('.')
 
   def self.version

data/test/test_countries.rb

@@ -0,0 +1,26 @@
+# encoding: utf-8
+
+###
+#  to run use
+#     ruby -I ./lib -I ./test test/test_countries.rb
+
+
+require 'helper'
+
+class TestCountries < MiniTest::Test
+
+  def test_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
+
+    at  = SportDb::Import.config.countries[:at]
+    assert_equal 'at',       at.key
+    assert_equal 'Austria',  at.name
+    assert_equal 'AUT',      at.code
+  end
+
+end # class TestCountries

data/test/test_team_reader.rb

@@ -60,6 +60,7 @@ TXT
 
   def test_parse_years
     recs = SportDb::Import::TeamReader.parse( <<TXT )
+= United States (us)
 FC Dallas (1996-),         Frisco        › Texas
 Miami Fusion (1998-2001),  Fort Lauderdale › Florida
 CD Chivas USA (-2014),     Carson          › California
@@ -84,12 +85,17 @@ TXT
 
   def test_parse_geos
     recs = SportDb::Import::TeamReader.parse( <<TXT )
+=  England (eng)
+== Greater London
 
 Fulham FC, 1879,  @ Craven Cottage,   London (Fulham)   › Greater London
   | Fulham | FC Fulham
 Charlton Athletic FC,  @ The Valley,  London (Charlton) › Greater London
   | Charlton | Charlton Athletic
 
+=  Deutschland (de)
+== Hamburg
+
 St. Pauli,   Hamburg (St. Pauli)
 TXT
 
@@ -99,13 +105,17 @@ TXT
     assert_equal 'London',           recs[0].city
     assert_equal 'Fulham',           recs[0].district
     assert_equal ['Greater London'], recs[0].geos
+    assert_equal 'eng',              recs[0].country
 
     assert_equal 'London',           recs[1].city
     assert_equal 'Charlton',         recs[1].district
     assert_equal ['Greater London'], recs[1].geos
+    assert_equal 'eng',              recs[1].country
 
     assert_equal 'Hamburg',          recs[2].city
     assert_equal 'St. Pauli',        recs[2].district
+    assert_equal ['Hamburg'],        recs[2].geos
+    assert_equal 'de',               recs[2].country
   end
 
 
@@ -114,8 +124,8 @@ TXT
 ==============
 ====
 ===========
-= Heading 1
-= Heading 1 ==================
+= Heading 1   - Austria (at)
+= Heading 1   - Austria (at) ==================
 == Heading 2
 == Heading 2 =========
 === Heading 3
@@ -123,6 +133,10 @@ TXT
 === Heading 3             # with end-of-line comment
 === Heading 3             ## with end-of-line comment
 === Heading 3 =========   # with end-of-line comment
+== Heading 2
+==== Heading 4
+= ?????
+== ???
 TXT
 
     pp recs

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: sportdb-config
 version: !ruby/object:Gem::Version
-  version: 0.3.2
+  version: 0.3.3
 platform: ruby
 authors:
 - Gerald Bauer
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-07-10 00:00:00.000000000 Z
+date: 2019-07-16 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rdoc
@@ -63,6 +63,7 @@ files:
 - config/leagues/sco.txt
 - lib/sportdb/config.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
@@ -71,6 +72,7 @@ files:
 - lib/sportdb/config/version.rb
 - test/helper.rb
 - test/test_config.rb
+- test/test_countries.rb
 - test/test_league_reader.rb
 - test/test_league_utils.rb
 - test/test_season_utils.rb