sportdb 0.9.0 → 0.9.1

data/lib/sportdb/models/event.rb

@@ -44,39 +44,52 @@ class Event < ActiveRecord::Base
   ### fix/todo: share helper for all text readers/parsers- where to put it?  
   ###
   
-  def title_esc_regex( title )
-      ## todo: how to mark value as regex?
-      ##  for now escape regex special chars e.g. . to \.
-      title_for_regex = title.gsub( '.', '\.' ) # e.g. Benfica Lis.
-      title_for_regex = title_for_regex.gsub( '(', '\(' ) # e.g. Club Atlético Colón (Santa Fe)
-      title_for_regex = title_for_regex.gsub( ')', '\)' )
+  def title_esc_regex( title_unescaped )
       
-      ## fix: todo: match accented char with or without accents
+      ##  escape regex special chars e.g. . to \. and ( to \( etc.
+      # e.g. Benfica Lis.
+      # e.g. Club Atlético Colón (Santa Fe)
+
+      ## NB: cannot use Regexp.escape! will escape space '' to '\ '
+      ## title = Regexp.escape( title_unescaped )
+      title = title_unescaped.gsub( '.', '\.' )
+      title = title.gsub( '(', '\(' )
+      title = title.gsub( ')', '\)' )
+
+      ##  match accented char with or without accents
       ##  add (ü|ue) etc.
       ## also make - optional change to (-| ) e.g. Blau-Weiss == Blau Weiss
-      ##  reuse for all readers!
-      title_for_regex = title_for_regex.gsub( '-', '(-| )' )
-      title_for_regex = title_for_regex.gsub( 'ß', '(ß|ss)' )
-      title_for_regex = title_for_regex.gsub( 'æ', '(æ|ae)' )
-      title_for_regex = title_for_regex.gsub( 'á', '(á|a)' )  ## e.g. Bogotá
-      title_for_regex = title_for_regex.gsub( 'ã', '(ã|a)' )   ## e.g  São Paulo
-      title_for_regex = title_for_regex.gsub( 'ä', '(ä|ae)' )  ## add a ?
-      title_for_regex = title_for_regex.gsub( 'ö', '(ö|oe)' )  ## add o ?
-      title_for_regex = title_for_regex.gsub( 'ó', '(ó|o)' )  ## e.g. Colón
-      title_for_regex = title_for_regex.gsub( 'ü', '(ü|ue)' )  ## add u ?
-      title_for_regex = title_for_regex.gsub( 'é', '(é|e)' )  ## e.g. Vélez
-      title_for_regex = title_for_regex.gsub( 'ê', '(ê|e)' )  ## e.g. Grêmio
-      title_for_regex = title_for_regex.gsub( 'ñ', '(ñ|n)' )  ## e.g. Porteño
-      title_for_regex = title_for_regex.gsub( 'ú', '(ú|u)' )  ## e.g. Fútbol
-
-      ## todo: add some more; use array for config?
+
+      ## todo: add some more
       ## see http://en.wikipedia.org/wiki/List_of_XML_and_HTML_character_entity_references  for more
+      ##
+      ##  reuse for all readers!
+      
+      alternatives = [
+        ['-', '(-| )'],
+        ['ß', '(ß|ss)'],
+        ['æ', '(æ|ae)'],
+        ['á', '(á|a)'],  ## e.g. Bogotá
+        ['ã', '(ã|a)'],  ## e.g  São Paulo
+        ['ä', '(ä|ae)'],  ## add a ?
+        ['Ö', '(Ö|Oe)'], ## e.g. Österreich
+        ['ö', '(ö|oe)'],  ## add o ?
+        ['ó', '(ó|o)'],  ## e.g. Colón
+        ['ü', '(ü|ue)'],  ## add u ?
+        ['é', '(é|e)'],  ## e.g. Vélez
+        ['ê', '(ê|e)'],  ## e.g. Grêmio
+        ['ñ', '(ñ|n)'],  ## e.g. Porteño
+        ['ú', '(ú|u)']  ## e.g. Fútbol
+      ]
+      
+      alternatives.each do |alt|
+        title = title.gsub( alt[0], alt[1] )
+      end
 
-      title_for_regex
+      title
   end
 
 
-
   def known_teams_table
     
     ## build known teams table w/ synonyms e.g.

data/lib/sportdb/reader.rb

@@ -40,14 +40,13 @@ class Reader
 
     puts "*** parsing data '#{name}' (#{path})..."
 
-    code = File.read( path )
+    ## nb: assume/enfore utf-8 encoding (with or without BOM - byte order mark)
+    ## - see sportdb/utils.rb
+    code = File.read_utf8( path )
     
     load_fixtures_worker( event_key, code )
 
-    ### fix: 
-    ## for loaded from fs use filestat? for version - why? why not?
-
-    Prop.create!( key: "db.#{fixture_name_to_prop_key(name)}.version", value: "txt.1" )
+    Prop.create!( key: "db.#{fixture_name_to_prop_key(name)}.version", value: "file.txt.#{File.mtime(path).strftime('%Y.%m.%d')}" )
   end
 
   def load_fixtures_builtin( event_key, name ) # load from gem (built-in)
@@ -55,7 +54,7 @@ class Reader
 
     puts "*** parsing data '#{name}' (#{path})..."
 
-    code = File.read( path )
+    code = File.read_utf8( path )
     
     load_fixtures_worker( event_key, code )
     
@@ -84,6 +83,12 @@ private
     ## assume active activerecord connection
     ##
     
+    ## reset cached values
+    @patch_rounds  = {}
+    @knockout_flag = false
+    @round         = nil
+    
+    
     @event = Event.find_by_key!( event_key )
     
     puts "Event #{@event.key} >#{@event.title}<"
@@ -99,8 +104,15 @@ private
     line =~ /Spieltag|Runde|Achtelfinale|Viertelfinale|Halbfinale|Finale/
   end
   
+  def is_group?( line )
+    # NB: check after is_round? (round may contain group reference!)
+    line =~ /Gruppe|Group/
+  end
+  
+  ### todo: rename to is_knockout_round?
+  ##
   def find_knockout_flag( line )
-    if line =~ /Achtelfinale|Viertelfinale|Halbfinale|Finale|K\.O\.|Knockout/
+    if line =~ /Achtelfinale|Viertelfinale|Halbfinale|Spiel um Platz 3|Finale|K\.O\.|Knockout/
       puts "   setting knockout flag to true"
       true
     else
@@ -108,6 +120,50 @@ private
     end
   end
   
+  def find_group_title_and_pos!( line )
+    ## group pos - for now support single digit e.g 1,2,3 or letter e.g. A,B,C
+    ## nb:  (?:)  = is for non-capturing group(ing)
+    regex = /(?:Group|Gruppe)\s+((?:\d{1}|[A-Z]{1}))\b/
+    
+    match = regex.match( line )
+    unless match.nil?
+      if match[1] == 'A'
+        pos = 1
+      elsif match[1] == 'B'
+        pos = 2
+      elsif match[1] == 'C'
+        pos = 3
+      elsif match[1] == 'D'
+        pos = 4
+      elsif match[1] == 'E'
+        pos = 5
+      elsif match[1] == 'F'
+        pos = 6
+      elsif match[1] == 'G'
+        pos = 7
+      elsif match[1] == 'H'
+        pos = 8
+      elsif match[1] == 'I'
+        pos = 9
+      elsif match[1] == 'J'
+        pos = 10
+      else
+        pos = match[1].to_i
+      end
+
+      title = match[0]
+
+      puts "   title: >#{title}<"
+      puts "   pos: >#{pos}<"
+      
+      line.sub!( regex, '[GROUP|TITLE+POS]' )
+
+      return [title,pos]
+    else
+      return [nil,nil]
+    end
+  end
+  
   def find_round_pos!( line )
     ## fix/todo:
     ##  if no round found assume last_pos+1 ??? why? why not?
@@ -118,7 +174,7 @@ private
       value = $1.to_i
       puts "   pos: >#{value}<"
       
-      line.sub!( regex, '[POS]' )
+      line.sub!( regex, '[ROUND|POS]' )
 
       return value
     else
@@ -241,6 +297,20 @@ private
       return nil
     end
   end
+  
+  def find_teams!( line )
+    counter = 1
+    teams = []
+    
+    team = find_team_worker!( line, counter )
+    while team.present?
+      teams << team
+      counter += 1
+      team = find_team_worker!( line, counter )
+    end
+    
+    teams
+  end
 
   def find_team1!( line )
     find_team_worker!( line, 1 )
@@ -257,7 +327,7 @@ private
       ## (thus add it, allows match for Benfica Lis.  for example - note . at the end)
 
       ## check add $ e.g. (\b| |\t|$) does this work? - check w/ Benfica Lis.$
-      regex = /\b#{value}(\b| |\t)/   # wrap with world boundry (e.g. match only whole words e.g. not wac in wacker) 
+      regex = /\b#{value}(\b| |\t|$)/   # wrap with world boundry (e.g. match only whole words e.g. not wac in wacker) 
       if line =~ regex
         puts "     match for team >#{key}< >#{value}<"
         # make sure @@oo{key}oo@@ doesn't match itself with other key e.g. wacker, wac, etc.
@@ -276,135 +346,195 @@ private
     end # each known_teams    
   end # method translate_teams!
   
+  
+  def parse_group( line )
+    puts "parsing group line: >#{line}<"
+    
+    match_teams!( line )
+    team_keys = find_teams!( line )
+      
+    title, pos = find_group_title_and_pos!( line )
 
-  def parse_fixtures( data )
+    puts "  line: >#{line}<"
+
+    ## world2 = Group.create!( event: world, pos: 2, title: 'Gruppe 2' )
+    ## world2.add_teams_from_ary!( team_keys_world2 )
+
+    group_attribs = {
+      title: title
+    }
+        
+    @group = Group.find_by_event_id_and_pos( @event.id, pos )
+    if @group.present?
+      puts "*** update group #{@group.id}:"
+    else
+      puts "*** create group:"
+      @group = Group.new
+      group_attribs = group_attribs.merge( {
+        event_id: @event.id,
+        pos:   pos
+      })
+    end
       
-    data.each_line do |line|
+    puts  group_attribs.to_json
+   
+    @group.update_attributes!( group_attribs )
+
+    @group.teams.clear  # remove old teams
+    ## add new teams
+    team_keys.each do |team_key|
+      team = Team.find_by_key!( team_key )
+      puts "  adding team #{team.title} (#{team.code})"
+      @group.teams << team
+    end
+  end
   
-      if line =~ /^\s*#/
-        # skip komments and do NOT copy to result (keep comments secret!)
-        logger.debug 'skipping comment line'
-        next
-      end
+  def parse_round( line )
+    puts "parsing round line: >#{line}<"
+    pos = find_round_pos!( line )
         
-      if line =~ /^\s*$/ 
-        # kommentar oder leerzeile überspringen 
-        logger.debug 'skipping blank line'
-        next
-      end
+    @knockout_flag = find_knockout_flag( line )
 
-      # remove leading and trailing whitespace
-      line = line.strip
+    group_title, group_pos = find_group_title_and_pos!( line )
 
-      if is_round?( line )
-        puts "parsing round line: >#{line}<"
-        pos = find_round_pos!( line )
-        
-        @knockout_flag = find_knockout_flag( line )
-        puts "  line: >#{line}<"
+    if group_pos.present?
+      @group = Group.find_by_event_id_and_pos!( @event.id, group_pos )
+    else
+      @group = nil   # reset group to no group
+    end
+
+    puts "  line: >#{line}<"
         
-        ## NB: dummy/placeholder start_at, end_at date
-        ##  replace/patch after adding all games for round
+    ## NB: dummy/placeholder start_at, end_at date
+    ##  replace/patch after adding all games for round
         
-        round_attribs = {
-          title: "#{pos}. Runde"
-        }
+    round_attribs = {
+      title: "#{pos}. Runde"
+    }
+
         
-        @round = Round.find_by_event_id_and_pos( @event.id, pos )
-        if @round.present?
-          puts "*** update round #{@round.id}:"
-        else
-          puts "*** create round:"
-          @round = Round.new
+    @round = Round.find_by_event_id_and_pos( @event.id, pos )
+    if @round.present?
+      puts "*** update round #{@round.id}:"
+    else
+      puts "*** create round:"
+      @round = Round.new
           
-          round_attribs = round_attribs.merge( {
-            event_id: @event.id,
-            pos:   pos,
-            start_at: Time.utc('1999-12-12'),
-            end_at:   Time.utc('1999-12-12')
-                                 })
-        end
+      round_attribs = round_attribs.merge( {
+        event_id: @event.id,
+        pos:   pos,
+        start_at: Time.utc('1912-12-12'),
+        end_at:   Time.utc('1912-12-12')
+      })
+    end
         
-        puts round_attribs.to_json
+    puts round_attribs.to_json
    
-        @round.update_attributes!( round_attribs )
+    @round.update_attributes!( round_attribs )
 
-        ### store list of round is for patching start_at/end_at at the end
-        @patch_rounds ||= {}
-        @patch_rounds[ @round.id ] = @round.id
-        
-        
-      else
-        puts "parsing game (fixture) line: >#{line}<"
+    ### store list of round is for patching start_at/end_at at the end
+    @patch_rounds[ @round.id ] = @round.id
+  end
+
+  def parse_game( line )
+    puts "parsing game (fixture) line: >#{line}<"
 
-        pos = find_game_pos!( line )
+    pos = find_game_pos!( line )
 
-        match_teams!( line )
-        team1_key = find_team1!( line )
-        team2_key = find_team2!( line )
+    match_teams!( line )
+    team1_key = find_team1!( line )
+    team2_key = find_team2!( line )
 
-        date  = find_date!( line )
-        scores = find_scores!( line )
+    date  = find_date!( line )
+    scores = find_scores!( line )
         
-        puts "  line: >#{line}<"
+    puts "  line: >#{line}<"
 
 
-        ### todo: cache team lookups in hash?
+    ### todo: cache team lookups in hash?
 
-        team1 = Team.find_by_key!( team1_key )
-        team2 = Team.find_by_key!( team2_key )
+    team1 = Team.find_by_key!( team1_key )
+    team2 = Team.find_by_key!( team2_key )
 
-        ### check if games exists
-        ##  with this teams in this round if yes only update
-        game = Game.find_by_round_id_and_team1_id_and_team2_id(
+    ### check if games exists
+    ##  with this teams in this round if yes only update
+    game = Game.find_by_round_id_and_team1_id_and_team2_id(
                          @round.id, team1.id, team2.id
-        )
-
-        game_attribs = {
-          score1:    scores[0],
-          score2:    scores[1],
-          score3:    scores[2],
-          score4:    scores[3],
-          score5:    scores[4],
-          score6:    scores[5],
-          play_at:   date,
-          knockout:  @knockout_flag
-        }
-        
-        game_attribs[ :pos ] = pos   if pos.present?
-
-        if game.present?
-          puts "*** update game #{game.id}:"
-        else
-          puts "*** create game:"
-          game = Game.new
-
-          more_game_attribs = {
-            round_id:  @round.id,
-            team1_id: team1.id,
-            team2_id: team2.id
-          }
+    )
+
+    game_attribs = {
+      score1:    scores[0],
+      score2:    scores[1],
+      score3:    scores[2],
+      score4:    scores[3],
+      score5:    scores[4],
+      score6:    scores[5],
+      play_at:   date,
+      knockout:  @knockout_flag,
+      group_id:  @group.present? ? @group.id : nil
+    }
+
+    game_attribs[ :pos ]      = pos        if pos.present?
+
+    if game.present?
+      puts "*** update game #{game.id}:"
+    else
+      puts "*** create game:"
+      game = Game.new
+
+      more_game_attribs = {
+        round_id:  @round.id,
+        team1_id: team1.id,
+        team2_id: team2.id
+      }
           
-          ## NB: use round.games.count for pos
-          ##  lets us add games out of order if later needed
-          more_game_attribs[ :pos ] = @round.games.count+1  if pos.nil? 
+      ## NB: use round.games.count for pos
+      ##  lets us add games out of order if later needed
+      more_game_attribs[ :pos ] = @round.games.count+1  if pos.nil? 
 
-          game_attribs = game_attribs.merge( more_game_attribs )
-        end
+      game_attribs = game_attribs.merge( more_game_attribs )
+    end
+
+    puts game_attribs.to_json
+
+    game.update_attributes!( game_attribs )
+  end
+
+
+  def parse_fixtures( data )
+      
+    data.each_line do |line|
+  
+      if line =~ /^\s*#/
+        # skip komments and do NOT copy to result (keep comments secret!)
+        logger.debug 'skipping comment line'
+        next
+      end
+        
+      if line =~ /^\s*$/ 
+        # kommentar oder leerzeile überspringen 
+        logger.debug 'skipping blank line'
+        next
+      end
 
-        puts game_attribs.to_json
+      # remove leading and trailing whitespace
+      line = line.strip
 
-        game.update_attributes!( game_attribs )
+      if is_round?( line )
+        parse_round( line )
+      elsif is_group?( line ) ## NB: group goes after round (round may contain group marker too)
+        parse_group( line )
+      else
+        parse_game( line )
       end
-    end # oldlines.each
+    end # lines.each
     
-    @patch_rounds ||= {}
     @patch_rounds.each do |k,v|
       puts "*** patch start_at/end_at date for round #{k}:"
       round = Round.find( k )
       games = round.games.order( 'play_at asc' ).all
       
-      ## skip rouns w/ no games
+      ## skip rounds w/ no games
       
       ## todo/fix: what's the best way for checking assoc w/ 0 recs?
       next if games.size == 0