sportdb-formats 1.2.0 → 2.0.0

data/lib/sportdb/formats/name_helper.rb

@@ -0,0 +1,87 @@
+
+module SportDb
+  module NameHelper
+
+
+  ## note: allow placeholder years to e.g. (-___) or (-????)
+  ##    for marking missing (to be filled in) years
+  ##  e.g. (1887-1911), (-2013),
+  ##      (1946-2001, 2013-) etc.
+  ##  todo/check: make more strict  e.g. only accept 4-digit years? - why? why not?
+  YEAR_RE =  %r{\(
+                  [0-9, ?_-]+?     # note: non-greedy (minimum/first) match
+              \)}x
+
+  def strip_year( name )
+    ## check for year(s) e.g. (1887-1911), (-2013),
+    ##                        (1946-2001, 2013-) etc.
+    ##  todo/check: only sub once (not global) - why? why not?
+    name.gsub( YEAR_RE, '' ).strip
+  end
+
+  def has_year?( name ) name =~ YEAR_RE; end
+
+
+  LANG_RE =  %r{\[
+                [a-z]{1,2}    # note also allow single-letter [a] or [d] or [e] - why? why not?
+                \]}x
+  def strip_lang( name )
+     name.gsub( LANG_RE, '' ).strip
+  end
+
+  def has_lang?( name ) name =~ LANG_RE; end
+
+
+  def sanitize( name )
+    ## check for year(s) e.g. (1887-1911), (-2013),
+    ##                        (1946-2001,2013-) etc.
+    name = strip_year( name )
+    ## check lang codes e.g. [en], [fr], etc.
+    name = strip_lang( name )
+    name
+  end
+
+
+  ## note: also add (),’,−  etc. e.g.
+  ##   Estudiantes (LP) => Estudiantes LP
+  ##   Saint Patrick’s Athletic FC => Saint Patricks Athletic FC
+  ##   Myllykosken Pallo −47 => Myllykosken Pallo 47
+  ##
+  ##  add & too!!
+  ##   e.g. Brighton & Hove Albion => Brighton Hove Albion  -- and others in England
+
+  NORM_RE =  %r{
+                    [.'’º/()&_−-]
+                  }x   # note: in [] dash (-) if last doesn't need to get escaped
+  ## note: remove all dots (.), dash (-), ', º, /, etc.
+  #   .  U+002E (46) - FULL STOP
+  #   '  U+0027 (39) - APOSTROPHE
+  #   ’  U+2019 (8217) - RIGHT SINGLE QUOTATION MARK
+  #   º  U+00BA (186) - MASCULINE ORDINAL INDICATOR
+  #   /  U+002F (47) - SOLIDUS
+  #   (  U+0028 (40) - LEFT PARENTHESIS
+  #   )  U+0029 (41) - RIGHT PARENTHESIS
+  #   −  U+2212 (8722) - MINUS SIGN
+  #   -  U+002D (45) - HYPHEN-MINUS
+
+  ##         for norm(alizing) names
+  def strip_norm( name )
+    name.gsub( NORM_RE, '' )
+  end
+
+  def normalize( name )
+    # note: do NOT call sanitize here (keep normalize "atomic" for reuse)
+    name = strip_norm( name )
+    name = name.gsub( ' ', '' )  # note: also remove all spaces!!!
+
+    ## todo/check: use our own downcase - why? why not?
+    name = downcase_i18n( name )     ## do NOT care about upper and lowercase for now
+    name
+  end
+
+
+  def variants( name )  Variant.find( name ); end
+
+  end  # module NameHelper
+end   # module SportDb
+

data/lib/sportdb/formats/search/sport.rb

@@ -6,7 +6,7 @@
 
 class SportSearch
 
-class Search    ## base search service - use/keep - why? why not? 
+class Search    ## base search service - use/keep - why? why not?
   def initialize( service ) @service = service; end
 end  # class Search
 
@@ -15,48 +15,62 @@ class PlayerSearch < Search
   ###################
   ## core required delegates  - use delegate generator - why? why not?
   def match_by( name:, country: nil, year: nil )
-    @service.match_by( name:    name, 
+    @service.match_by( name:    name,
                        country: country,
-                       year:    year ) 
+                       year:    year )
   end
 
   ###############
   ### more deriv support functions / helpers
   def match( name ) match_by( name: name ); end
-  ## add more here - why? why not? 
+  ## add more here - why? why not?
 end   # class PlayerSearch
 
 
 class LeagueSearch < Search
+
   ###################
   ## core required delegates  - use delegate generator - why? why not?
-  def match_by( name:, country: nil ) 
-    @service.match_by( name: name, 
-                       country: country ) 
+  def match_by( name: nil, code: nil, country: nil )
+     ## todo/fix upstream - remove "generic" match_by() - why? why not?
+     ###
+     if code && name.nil?
+      @service.match_by_code( code, country: country )
+    elsif name && code.nil?
+      @service.match_by_name( name, country: country )
+    else
+      raise ArgumentError, "LeagueSearch#match_by - one (and only one arg) required - code: or name:"
+    end
+    ## @service.match_by( name: name,
+    ##                   country: country )
   end
 
+  ## all-in-one query (name or code)
+  def match( q, country: nil )
+     @service.match_by_name_or_code( q, country: country )
+  end
+
+
   ###############
   ### more deriv support functions / helpers
-  def match( name )   match_by( name: name ); end 
-
-  def find!( name )
-    league = find( name )
+  def find!( q )
+    league = find( q )
     if league.nil?
-      puts "** !!! ERROR - no league match found for >#{name}<, add to leagues table; sorry"
+      puts "** !!! ERROR - no league match found for >#{q}<, add to leagues table; sorry"
       exit 1
     end
     league
   end
 
-  def find( name )
+  def find( q )
     league = nil
-    recs = match( name )
+    recs = match( q )
     # pp m
 
     if recs.empty?
       ## fall through/do nothing
     elsif recs.size > 1
-      puts "** !!! ERROR - ambigious league name; too many leagues (#{recs.size}) found:"
+      puts "** !!! ERROR - ambigious league query; too many leagues (#{recs.size}) found:"
       pp recs
       exit 1
     else
@@ -68,19 +82,20 @@ class LeagueSearch < Search
 end # class LeagueSearch
 
 
+
 class GroundSearch  < Search
   ###################
   ## core required delegates  - use delegate generator - why? why not?
   def match_by( name:, country: nil, city: nil )
-    @service.match_by( name:    name, 
+    @service.match_by( name:    name,
                        country: country,
-                       city:    city ) 
+                       city:    city )
   end
 
   ###############
   ### more deriv support functions / helpers
   def match( name ) match_by( name: name ); end
-  ## add more here - why? why not? 
+  ## add more here - why? why not?
 end  # class GroundSearch
 
 
@@ -106,31 +121,31 @@ class ClubSearch  < Search
                        league:  nil,
                        mods:    nil )
     ## for now assume "global" mods  - checks only for name
-    ##     
+    ##
     if mods && mods[ name ]
       club = mods[ name ]
       return [club]   # note: wrap (single record) in array
-    end   
+    end
 
     ## note: add "auto-magic" country calculation via league record
     ##            if league is a national league for football clubs
     if league
         raise ArgumentError, "match_by - league AND country NOT supported; sorry"  if country
-### find countries via league        
+### find countries via league
 ###     support league.intl? too - why? why not?
 ###       or only nationa league
         raise ArgumentError, "match_by - league - only national club leagues supported (not int'l or national teams for now); sorry"   unless league.national? && league.clubs?
-       
+
         ### calc countries
         ### uses "global" func in sports-catalogs for now
-        ##   move code here - why? why not? 
+        ##   move code here - why? why not?
         country = find_countries_for_league( league )
-        @service.match_by( name:    name, 
+        @service.match_by( name:    name,
                            country: country )
     else
-        @service.match_by( name:    name, 
+        @service.match_by( name:    name,
                            country: country )
-    end 
+    end
   end
 
 
@@ -145,10 +160,10 @@ class ClubSearch  < Search
   ###############
   ### more deriv support functions / helpers
   def match( name ) match_by( name: name ); end
- 
+
   ##########
-  #  "legacy" finders - return zero or one club 
-  ##    (if more than one match, exit/raise error/exception) 
+  #  "legacy" finders - return zero or one club
+  ##    (if more than one match, exit/raise error/exception)
   def find( name )   find_by( name: name ); end
   def find!( name )  find_by!( name: name ); end
 
@@ -156,7 +171,7 @@ class ClubSearch  < Search
   ##   if there is more than one match than find aborts / fails
   def find_by!( name:, country: nil,
                        league:  nil )    ## todo/fix: add international or league flag?
-    club = find_by( name:    name, 
+    club = find_by( name:    name,
                     country: country,
                     league:  league )
 
@@ -174,8 +189,8 @@ class ClubSearch  < Search
     ## note: allow passing in of country key too (auto-counvert)
     ##       and country struct too
     ##     - country assumes / allows the country key or fifa code for now
-    recs = match_by( name:    name, 
-                     country: country, 
+    recs = match_by( name:    name,
+                     country: country,
                      league:  league )
 
     club = nil
@@ -243,12 +258,12 @@ end # class EventSearch
 
 ####
 ## virtual table for season lookup
-##   note - use EventSeaon  to avoid name conflict with (global) Season class 
-##      find a better name SeasonInfo or SeasonFinder or SeasonStore 
+##   note - use EventSeaon  to avoid name conflict with (global) Season class
+##      find a better name SeasonInfo or SeasonFinder or SeasonStore
 ##                       or SeasonQ or ??
 class EventSeasonSearch
-    def initialize( events: ) 
-        @events = events 
+    def initialize( events: )
+        @events = events
     end
 
   ###############
@@ -256,7 +271,7 @@ class EventSeasonSearch
   ##
   def find_by( date:, league: )
     date = Date.strptime( date, '%Y-%m-%d' )   if date.is_a?( String )
-  
+
     infos = @events.seasons( league )
 
     infos.each do |info|
@@ -265,17 +280,17 @@ class EventSeasonSearch
     nil
   end
 end # class EventSeasonSearch
-  
+
 
 ######
 ### add virtual team search ( clubs + national teams)
 ##   note: no record base!!!!!
 class TeamSearch
     ## note: "virtual" index lets you search clubs and/or national_teams (don't care)
-  
-  def initialize( clubs:, national_teams: ) 
+
+  def initialize( clubs:, national_teams: )
     @clubs          = clubs
-    @national_teams = national_teams 
+    @national_teams = national_teams
   end
 
     ## todo/check: rename to/use map_by! for array version - why? why not?
@@ -290,19 +305,19 @@ class TeamSearch
         _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!!!
-            @club.find!( name )
+            @clubs.find!( name )
           else  ## assume clubs in domestic/national league tournament
-             ## note - search by league countries (may incl. more than one country 
+             ## note - search by league countries (may incl. more than one country
              ##             e.g. us incl. ca, fr incl. mc, ch incl. li, etc.
-            @club.find_by!( name: name, league: league )
+            @clubs.find_by!( name: name, league: league )
           end
         else   ## assume national teams (not clubs)
           @national_teams.find!( name )
@@ -310,8 +325,8 @@ class TeamSearch
       end
     end # method _find_by!
   end  # class TeamSearch
-  
-  
+
+
 
    def initialize( leagues:,
                    national_teams:,
@@ -320,11 +335,11 @@ class TeamSearch
                    events:,
                    players:
                    )
-       @leagues        = LeagueSearch.new( leagues ) 
+       @leagues        = LeagueSearch.new( leagues )
        @national_teams = NationalTeamSearch.new( national_teams )
        @clubs          = ClubSearch.new( clubs )
-       @events         = EventSearch.new( events ) 
-       
+       @events         = EventSearch.new( events )
+
        @grounds        = GroundSearch.new( grounds )
 
        @players        = PlayerSearch.new( players )
@@ -332,8 +347,8 @@ class TeamSearch
        ## virtual deriv ("composite") search services
        @teams          = TeamSearch.new( clubs:          @clubs,
                                          national_teams: @national_teams )
-       @event_seasons  = EventSeasonSearch.new( events: @events )                                
-       
+       @event_seasons  = EventSeasonSearch.new( events: @events )
+
    end
 
     def countries
@@ -351,7 +366,7 @@ class TeamSearch
  def grounds()         @grounds; end
 
  def players()        @players; end
- 
+
  def teams()          @teams; end         ## note - virtual table
  def seasons()        @event_seasons; end ## note - virtual table
 end  # class SportSearch

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

@@ -0,0 +1,116 @@
+###
+#  note - extend all structs for with search api
+#
+#  todo - add more helpers!!!!
+#
+#
+#  todo - fix - move all non-core search functionality/machinery
+#                  over here - why? why not?
+
+
+
+module Sports
+
+class Country
+    def self._search #### use service/api or such - why? why not?
+        SportDb::Import.world.countries
+    end
+   def self.find_by( code: nil, name: nil )
+        _search.find_by( code: code, name: name )
+   end
+
+   def self.find( q )   ## find by code (first) or name (second)
+       _search.find( q )
+   end
+
+   def self.parse_heading( line )
+      ## fix - move parse code here from search - why? why not?
+      _search.parse( line )
+   end
+
+   ## add alternate names/aliases
+   class << self
+    alias_method :[],      :find    ### keep shortcut - why? why not?
+    alias_method :heading, :parse_heading
+   end
+
+
+# open question - what name to use build or  parse_line or ?
+#                              or   parse_recs for CountryReader?
+#          remove CountryReader helper methods - why? why not?
+#   use parse_heading/heading for now !!!
+#
+#   def self.parse( line )  or build( line ) ??
+#      SportDb::Import.world.countries.parse( line )
+#   end
+#
+# !!!! note - conflict with
+#     def self.read( path )  CountryReader.read( path ); end
+#     def self.parse( txt )  CountryReader.parse( txt ); end
+#
+end # class Country
+
+
+###
+## todo/fix - add find_by( code: ), find_by( name: )
+##                   split - why? why not?
+
+
+class League
+    def self._search #### use service/api or such - why? why not?
+        SportDb::Import.catalog.leagues
+    end
+    def self.match_by( name: nil, code: nil,
+                        country: nil )
+       _search.match_by( name: name, code: code,
+                         country: country )
+    end
+    def self.match( q ) _search.match( q ); end
+
+    def self.find!( q ) _search.find!( q ); end
+    def self.find( q )  _search_find( q ); end
+end # class League
+
+
+class NationalTeam
+    def self._search #### use service/api or such - why? why not?
+        SportDb::Import.catalog.national_teams
+    end
+
+    def self.find( q )   _search.find( q ); end
+    def self.find!( q )  _search_find!( q ); end
+end # class NationalTeam
+
+
+class Club
+    def self._search #### use service/api or such - why? why not?
+        SportDb::Import.catalog.clubs
+    end
+
+    def self.match_by( name:, country: nil,
+                         league:  nil,
+                         mods:    nil )
+        _search.match_by( name: name, country: country,
+                          league: league, mods: mods )
+    end
+    def self.match( name ) match_by( name: name ); end
+
+    def self.find( name )   _search.find_by( name: name ); end
+    def self.find!( name )  _search.find_by!( name: name ); end
+
+    def self.find_by!( name:, country: nil,
+                              league:  nil )
+       _search.find_by!( name: name, country: country,
+                                     league: league )
+    end
+    def self.find_by( name:, country: nil,
+                             league:  nil )
+       _search.find_by( name: name, country: country,
+                                    league: league )
+    end
+
+   def self.build_mods( mods )
+       _search_build_mods( mods )
+    end
+  end # class Club
+end   # module Sports

data/lib/sportdb/formats/search/world.rb

@@ -6,7 +6,7 @@
 # -                 .find_by_name
 
 
-class WorldSearch    
+class WorldSearch
 
 class CitySearch
   def initialize( service ) @service = service; end
@@ -14,29 +14,45 @@ class CitySearch
   ###################
   ## core required delegates  - use delegate generator - why? why not?
   def match_by( name: )
-    @service.match_by( name: name ) 
+    @service.match_by( name: name )
   end
 end  # class CitySearch
 
 
 class CountrySearch
   def initialize( service ) @service = service; end
-  
+
     ###################
     ## core required delegates  - use delegate generator - why? why not?
-    def find_by_code( code ) @service.find_by_code( code ); end
-    def find_by_name( name ) @service.find_by_name( name ); end
+    def find_by_code( code )
+      puts "!! DEPRECATED - use CountrySearch#find_by( code: )"
+      @service.find_by_code( code )
+    end
+    def find_by_name( name )
+      puts "!! DEPRECATED - use CountrySearch#find_by( name: )"
+      @service.find_by_name( name )
+    end
 
+    def find_by( code: nil, name: nil )
+      ## todo/fix upstream - change to find_by( code:, name:, ) too
+      if code && name.nil?
+        @service.find_by_code( code )
+      elsif name && code.nil?
+        @service.find_by_name( name )
+      else
+        raise ArgumentError, "CountrySearch#find_by - one (and only one arg) required - code: or name:"
+      end
+    end
 
-    ###############
-    ### more deriv support functions / helpers
     def find( q )
-       country = find_by_code( q )
-       country = find_by_name( q )  if country.nil?     ## try lookup / find by (normalized) name
-       country
+      @service.find_by_name_or_code( q )
     end
     alias_method :[], :find    ### keep shortcut - why? why not?
-         
+
+
+    ###############
+    ### more deriv support functions / helpers
+
  ###
  ##   split/parse country line
  ##
@@ -56,29 +72,31 @@ class CountrySearch
    country = nil
    values.each do |value|
       value = value.strip
-      ## check for trailing country code e.g. (at), (eng), etc.
-      if value =~ /[ ]+\((?<code>[a-z]{1,4})\)$/  ## e.g. Austria (at)
+      ## check for trailing country code e.g. (at), (eng), etc
+      ##   allow code 1 to 5 for now - northern cyprus(fifa) with 5 letters?.
+      ##     add/allow  gb-eng, gb-wal (official iso2!!), in the future too - why? why not?
+      if value =~ /[ ]+\((?<code>[A-Za-z]{1,5})\)$/  ## e.g. Austria (at)
         code =  $~[:code]
         name = value[0...(value.size-code.size-2)].strip  ## note: add -2 for brackets
-        candidates = [ find_by_code( code ), find_by_name( name ) ]
+        candidates = [ find_by( code: code ), find_by( name: name ) ]
         if candidates[0].nil?
-          puts "** !!! ERROR !!! country - unknown code >#{code}< in line: #{line}"
+          puts "** !!! ERROR Country.parse_heading - unknown code >#{code}< in line: #{line}"
           pp line
           exit 1
         end
         if candidates[1].nil?
-          puts "** !!! ERROR !!! country - unknown name >#{code}< in line: #{line}"
+          puts "** !!! ERROR Country.parse_heading - unknown name >#{code}< in line: #{line}"
           pp line
           exit 1
         end
         if candidates[0] != candidates[1]
-          puts "** !!! ERROR !!! country - name and code do NOT match the same country:"
+          puts "** !!! ERROR Country.parse_heading - name and code do NOT match the same country:"
           pp line
           pp candidates
           exit 1
         end
         if country && country != candidates[0]
-          puts "** !!! ERROR !!! country - names do NOT match the same country:"
+          puts "** !!! ERROR Country.parse_heading - names do NOT match the same country:"
           pp line
           pp country
           pp candidates
@@ -89,12 +107,12 @@ class CountrySearch
         ## just assume value is name or code
         candidate = find( value )
         if candidate.nil?
-          puts "** !!! ERROR !!! country - unknown name or code >#{value}< in line: #{line}"
+          puts "** !!! ERROR Country.parse_heading - unknown name or code >#{value}< in line: #{line}"
           pp line
           exit 1
         end
         if country && country != candidate
-          puts "** !!! ERROR !!! country - names do NOT match the same country:"
+          puts "** !!! ERROR Country.parse_heading - names do NOT match the same country:"
           pp line
           pp country
           pp candidate
@@ -126,9 +144,9 @@ end  # class WorldSearch
 
 
 class DummyCountrySearch
-    def find_by_code( code ) 
+    def find_by_code( code )
         puts "[WARN] no world search configured; cannot find country by code"
-        nil 
+        nil
     end
     def find_by_name( name )
         puts "[WARN] no world search configured; cannot find country by name"

data/lib/sportdb/formats/version.rb

@@ -2,8 +2,8 @@ module SportDb
 module Module
 module Formats
 
-  MAJOR = 1    ## todo: namespace inside version or something - why? why not??
-  MINOR = 2
+  MAJOR = 2    ## todo: namespace inside version or something - why? why not??
+  MINOR = 0
   PATCH = 0
   VERSION = [MAJOR,MINOR,PATCH].join('.')