sportdb-models 1.10.0

data/lib/sportdb/stats.rb

@@ -0,0 +1,27 @@
+
+module SportDb
+  
+  class Stats
+    include Models
+
+    def tables
+      puts "  #{League.count} leagues  /  #{Season.count} seasons"
+      puts "  #{Event.count} events (league+season recs)  /  #{Round.count} rounds  /  #{Group.count} groups"
+      puts "  #{Team.count} teams"
+      puts "  #{Game.count} games"
+      puts "  #{Badge.count} badges"
+
+      puts "  #{Roster.count} rosters (person+team+event recs)"
+      puts "  #{Goal.count} goals (person+game recs)"
+
+      puts "  #{Assoc.count}  assocs|orgs"
+      puts "  #{Ground.count}  grounds|stadiums"
+
+## note: moved to racing.db  -- remove/delete!!
+##      puts "  #{Track.count} tracks / #{Race.count} races (track+event recs) / #{Run.count} runs"
+##      puts "  #{Record.count} records (race|run+person recs)"
+    end
+
+  end  # class Stats
+
+end  # module SportDb

data/lib/sportdb/utils.rb

@@ -0,0 +1,89 @@
+# encoding: utf-8
+
+### note: some utils moved to worldbdb/utils for reuse
+
+
+####
+## move to folder matcher(s)/finder(s)
+#  -- rename to FixtureFinder or FixtureFinders
+#  or just GeneralFinder
+#   TeamFinder etc.  ???
+
+
+module SportDb
+  module FixtureHelpers
+
+
+  def cut_off_end_of_line_comment!( line )
+    #  cut off (that is, remove) optional end of line comment starting w/ #
+
+    line.sub!( /#.*$/ ) do |_|
+      logger.debug "   cutting off end of line comment - >>#{$&}<<"
+      ''
+    end
+
+    # NB: line = line.sub  will NOT work - thus, lets use line.sub!
+  end
+
+
+  def find_nationality!( line )
+    # extract optional nationality - three-letter country code from line e.g.  Lionel Messi (ARG)
+    # and return it
+    # NB: side effect - removes num from line string
+
+    regex = /\(([A-Z]{3})\)/     # e.g. (ARG)
+    if line =~ regex
+      logger.debug "   nationality: >#{$1}<"
+
+      line.sub!( regex, '[NATIONALITY]' )
+      return $1.to_s
+    else
+      return nil
+    end
+  end
+
+
+  def find_leading_num!( line )
+    # extract optional leading num from line   e.g.   9 Lionel Messi
+    # and return it
+    # NB: side effect - removes num from line string
+
+    # e.g.  9  Lionel Messi   - must start line
+    ## note: use lookahead (?=) for trailing spaces - do NOT cosume
+    regex = /^[ \t]*(\d{1,3})(?=[ \t]+)/
+    if line =~ regex
+      logger.debug "   num: >#{$1}<"
+
+      line.sub!( regex, '[NUM]' )
+      return $1.to_i
+    else
+      return nil
+    end
+  end
+
+  def find_leading_pos!( line )
+    # extract optional game pos from line
+    # and return it
+    # NB: side effect - removes pos from line string
+
+    # e.g.  (1)   - must start line 
+    regex = /^[ \t]*\((\d{1,3})\)[ \t]+/
+    if line =~ regex
+      logger.debug "   pos: >#{$1}<"
+
+      line.sub!( regex, '[POS] ' ) # NB: add trailing space
+      return $1.to_i
+    else
+      return nil
+    end
+  end
+
+  def find_game_pos!( line )
+    ## fix: add depreciation warning - remove - use find_leading_pos!
+    find_leading_pos!( line )
+  end
+
+
+
+  end # module FixtureHelpers
+end # module SportDb

data/lib/sportdb/utils_date.rb

@@ -0,0 +1,26 @@
+# encoding: utf-8
+
+module SportDb
+  module FixtureHelpers
+
+  def is_postponed?( line )
+    # check if line include postponed marker e.g. =>
+    line =~ /=>/
+  end
+
+
+  def find_date!( line, opts={} )
+    ## NB: lets us pass in start_at/end_at date (for event)
+    #   for auto-complete year
+
+    # extract date from line
+    # and return it
+    # NB: side effect - removes date from line string
+    
+    finder = DateFinder.new
+    finder.find!( line, opts )
+  end
+
+  end # module FixtureHelpers
+end # module SportDb
+

data/lib/sportdb/utils_goals.rb

@@ -0,0 +1,20 @@
+# encoding: utf-8
+
+module SportDb
+  module FixtureHelpers
+
+  def is_goals?( line )
+    # check if is goals line
+    #  e.g. looks like
+    #   Neymar 29', 71' (pen.) Oscar 90+1';  Marcelo 11' (o.g.)
+    #  check for
+    #    <space>90'  or
+    #    <space>90+1'
+
+    line =~ /[ ](\d{1,3}\+)?\d{1,3}'/
+  end
+
+
+  end # module FixtureHelpers
+end # module SportDb
+

data/lib/sportdb/utils_group.rb

@@ -0,0 +1,63 @@
+# encoding: utf-8
+
+module SportDb
+  module FixtureHelpers
+
+  def is_group_def?( line )
+    # NB: check after is_round? (round may contain group reference!)
+    ## must include bar (|) marker (make required)
+    line =~ /\|/ && is_group?( line )
+  end
+
+  def is_group?( line )
+    # NB: check after is_round? (round may contain group reference!)
+    ## note: =~ return nil if not match found, and 0,1, etc for match
+    (line =~ SportDb.lang.regex_group) != nil
+  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 or HEX
+    ## nb:  (?:)  = is for non-capturing group(ing)
+
+    ## fix:
+    ##   get Group|Gruppe|Grupo from lang!!!! do NOT hardcode in place
+
+    ## todo:
+    ##   check if Group A:  or [Group A]  works e.g. : or ] get matched by \b ???
+    regex = /(?:Group|Gruppe|Grupo)\s+((?:\d{1}|[A-Z]{1,3}))\b/
+
+    match = regex.match( line )
+    
+    return [nil,nil] if match.nil?
+
+    pos = case match[1]
+          when 'A' then 1
+          when 'B' then 2
+          when 'C' then 3
+          when 'D' then 4
+          when 'E' then 5
+          when 'F' then 6
+          when 'G' then 7
+          when 'H' then 8
+          when 'I' then 9
+          when 'J' then 10
+          when 'K' then 11
+          when 'L' then 12
+          when 'HEX' then 666    # HEX for Hexagonal - todo/check: map to something else ??
+          else  match[1].to_i
+          end
+
+    title = match[0]
+
+    logger.debug "   title: >#{title}<"
+    logger.debug "   pos: >#{pos}<"
+      
+    line.sub!( regex, '[GROUP.TITLE+POS]' )
+
+    return [title,pos]
+  end
+
+
+
+  end # module FixtureHelpers
+end # module SportDb

data/lib/sportdb/utils_map.rb

@@ -0,0 +1,44 @@
+# encoding: utf-8
+
+module SportDb
+  module FixtureHelpers
+
+
+  def find_ground!( line )
+    TextUtils.find_key_for!( 'ground', line )
+  end
+
+  ## todo/fix: pass in known_grounds as a parameter? why? why not?
+  ### todo/fix:
+  ##    remove =nil in para - make param required w/o fallback
+
+  def map_ground!( line, known_grounds=nil )
+    if known_grounds.nil?
+      puts "depreciated API call map_ground! (pass in mapping table as 2nd param)"
+      known_grounds = @known_grounds
+    end
+
+    TextUtils.map_titles_for!( 'ground', line, known_grounds )
+  end
+
+
+
+  def find_person!( line )
+    TextUtils.find_key_for!( 'person', line )
+  end
+
+  ### todo/fix:
+  ##    remove =nil in para - make param required w/o fallback
+  def map_person!( line, known_persons=nil )
+    if known_persons.nil?
+      puts "depreciated API call map_person! (pass in mapping table as 2nd param)"
+      known_persons = @known_persons
+    end
+    
+    TextUtils.map_titles_for!( 'person', line, known_persons )
+  end
+
+
+  end # module FixtureHelpers
+end # module SportDb
+

data/lib/sportdb/utils_round.rb

@@ -0,0 +1,165 @@
+# encoding: utf-8
+
+module SportDb
+  module FixtureHelpers
+
+  def is_round_def?( line )
+    ## must include bar (|) marker (make required)
+    line =~ /\|/ && is_round?( line ) 
+  end
+
+  def is_round?( line )
+    ## note: =~ return nil if not match found, and 0,1, etc for match
+    (line =~ SportDb.lang.regex_round) != nil
+  end
+
+  def is_knockout_round?( line )
+
+    ## todo: check for adding ignore case for regex (e.g. 1st leg/1st Leg)
+
+    if line =~ SportDb.lang.regex_leg1
+      logger.debug "  two leg knockout; skip knockout flag on first leg"
+      false
+    elsif line =~ SportDb.lang.regex_knockout_round
+      logger.debug "   setting knockout flag to true"
+      true
+    elsif line =~ /K\.O\.|K\.o\.|Knockout/
+        ## NB: add two language independent markers, that is, K.O. and Knockout
+      logger.debug "   setting knockout flag to true (lang independent marker)"
+      true
+    else
+      false
+    end
+  end
+
+
+  ##
+  ## fix/todo: check that [ROUND.TITLE2] and friends do NOT use pipes (|)
+  ##  change all pipes (|) to dot (.)
+  ## - pipes get used for def markers!!!
+
+  def find_round_header_title2!( line )
+    ## todo/fix:
+    ##  cleanup method
+    ##   use  buf.index( '//' ) to split string (see found_round_def)
+    ##     why? simpler why not?
+    ##  - do we currently allow groups if title2 present? add example if it works?
+
+
+    # assume everything after // is title2 - strip off leading n trailing whitespaces
+    regex = /\/{2,}\s*(.+)\s*$/
+    if line =~ regex
+      logger.debug "   title2: >#{$1}<"
+      
+      line.sub!( regex, '[ROUND.TITLE2]' )
+      return $1
+    else
+      return nil    # no round title2 found (title2 is optional)
+    end
+  end
+
+
+  def find_round_header_title!( line )
+    # assume everything left is the round title
+    #  extract all other items first (round title2, round pos, group title n pos, etc.)
+
+    ## todo/fix:
+    ##  cleanup method
+    ##   use  buf.index( '//' ) to split string (see found_round_def)
+    ##     why? simpler why not?
+    ##  - do we currently allow groups if title2 present? add example if it works?
+
+    buf = line.dup
+    logger.debug "  find_round_header_title! line-before: >>#{buf}<<"
+
+    buf.gsub!( /\[[^\]]+\]/, '' )   # e.g. remove [ROUND.POS], [ROUND.TITLE2], [GROUP.TITLE+POS] etc.
+    buf.sub!( /\s+[\/\-]{1,}\s+$/, '' )    # remove optional trailing / or / chars (left over from group)
+    buf.strip!    # remove leading and trailing whitespace
+
+    logger.debug "  find_round_title! line-after: >>#{buf}<<"
+
+    ### bingo - assume what's left is the round title
+
+    logger.debug "   title: >>#{buf}<<"
+    line.sub!( buf, '[ROUND.TITLE]' )
+
+    buf
+  end
+
+
+  def find_round_def_title!( line )
+    # assume everything before pipe (\) is the round title
+    #  strip [ROUND.POS],  todo:?? [ROUND.TITLE2]
+
+    # todo/fix: add title2 w/  // or /  why? why not?
+    #  -- strip / or / chars
+
+    buf = line.dup
+    logger.debug "  find_round_def_title! line-before: >>#{buf}<<"
+
+    ## cut-off everything after (including) pipe (|)
+    buf = buf[ 0...buf.index('|') ]
+
+    # e.g. remove [ROUND.POS], [ROUND.TITLE2], [GROUP.TITLE+POS] etc.
+    buf.gsub!( /\[[^\]]+\]/, '' )    ## fix: use helper for (re)use e.g. remove_match_placeholder/marker or similar?
+    # remove leading and trailing whitespace
+    buf.strip!
+
+    logger.debug "  find_round_def_title! line-after: >>#{buf}<<"
+
+    logger.debug "   title: >>#{buf}<<"
+    line.sub!( buf, '[ROUND.TITLE]' )
+
+    buf
+  end
+
+
+  def find_round_pos!( line )
+    # pass #1) extract optional round pos from line
+    # e.g.  (1)   - must start line 
+    regex_pos = /^[ \t]*\((\d{1,3})\)[ \t]+/
+
+    # pass #2) find free standing number  e.g. Matchday 3 or Round 5 or 3. Spieltag etc.
+    # note: /\b(\d{1,3})\b/
+    #   will match -12
+    #  thus, use space required - will NOT match  -2 e.g. Group-2 Play-off
+    #  note:  allow  1. Runde  n
+    #                1^ Giornata
+    regex_num = /(?:^|\s)(\d{1,3})(?:[.\^\s]|$)/
+
+    if line =~ regex_pos
+      logger.debug "   pos: >#{$1}<"
+      
+      line.sub!( regex_pos, '[ROUND.POS] ' )  ## NB: add back trailing space that got swallowed w/ regex -> [ \t]+
+      return $1.to_i
+    elsif line =~ regex_num
+      ## assume number in title is pos (e.g. Jornada 3, 3 Runde etc.)
+      ## NB: do NOT remove pos from string (will get removed by round title)
+
+      num = $1.to_i  # note: clone capture; keep a copy (another regex follows; will redefine $1)
+
+      #### fix:
+      #  use/make keywords required
+      #  e.g. Round of 16  -> should NOT match 16!
+      #    Spiel um Platz 3  (or 5) etc -> should NOT match 3!
+      #  Round 16 - ok
+      #  thus, check for required keywords
+
+      ## quick hack for round of 16
+      # todo: mask match e.g. Round of xxx ... and try again - might include something
+      #  reuse pattern for Group XX Replays for example
+      if line =~ /^\s*Round of \d{1,3}\b/
+         return nil
+      end
+
+      logger.debug "   pos: >#{num}<"
+      return num
+    else
+      ## fix: add logger.warn no round pos found in line
+      return nil
+    end
+  end # method find_round_pos!
+
+
+  end # module FixtureHelpers
+end # module SportDb

data/lib/sportdb/utils_scores.rb

@@ -0,0 +1,17 @@
+# encoding: utf-8
+
+module SportDb
+  module FixtureHelpers
+
+
+  def find_scores!( line, opts={} )
+    # note: always call after find_dates !!!
+    #  scores match date-like patterns!!  e.g. 10-11  or 10:00 etc.
+    #   -- note: score might have two digits too
+
+    finder = ScoresFinder.new
+    finder.find!( line, opts )
+  end
+
+  end # module FixtureHelpers
+end # module SportDb