sportdb-formats 1.1.5 → 1.1.6

data/lib/sportdb/formats/event/event_reader.rb

@@ -1,183 +1,183 @@
-
-module SportDb
-module Import
-
-
-class EventInfo
-    ##  "high level" info (summary) about event  (like a "wikipedia infobox")
-    ##    use for checking dataset imports; lets you check e.g.
-    ##    - dates within range
-    ##    - number of teams e.g. 20
-    ##    - matches played e.g. 380
-    ##    - goals scored e.g. 937
-    ##    etc.
-
-    attr_reader :league,
-                :season,
-                :teams,
-                :matches,
-                :goals,
-                :start_date,
-                :end_date
-
-    def initialize( league:, season:,
-                    start_date: nil, end_date: nil,
-                    teams:   nil,
-                    matches: nil,
-                    goals:   nil )
-
-      @league      = league
-      @season      = season
-
-      @start_date  = start_date
-      @end_date    = end_date
-
-      @teams       = teams    ## todo/check: rename/use teams_count ??
-      @matches     = matches  ## todo/check: rename/use match_count ??
-      @goals       = goals
-    end
-
-    def include?( date )
-       ## todo/fix: add options e.g.
-       ##  - add delta/off_by_one or such?
-       ##  - add strict (for) only return true if date range (really) defined (no generic auto-rules)
-
-      ### note: for now allow off by one error (via timezone/local time errors)
-      ##    todo/fix: issue warning if off by one!!!!
-      if @start_date && @end_date
-        date >= (@start_date-1) &&
-        date <= (@end_date+1)
-      else
-        if @season.year?
-           # assume generic rule
-           ## same year e.g. Jan 1 - Dec 31; always true for now
-           date.year == @season.start_year
-        else
-           # assume generic rule
-           ##  July 1 - June 30 (Y+1)
-           ##  - todo/check -start for some countries/leagues in June 1 or August 1 ????
-           date >= Date.new( @season.start_year, 7, 1 ) &&
-           date <= Date.new( @season.end_year, 6, 30 )
-        end
-      end
-    end  # method include?
-    alias_method :between?, :include?
-  end # class EventInfo
-
-
-  class EventInfoReader
-    def catalog() Import.catalog; end
-
-
-    def self.read( path )
-      txt = File.open( path, 'r:utf-8') {|f| f.read }
-      new( txt ).parse
-    end
-
-    def self.parse( txt )
-      new( txt ).parse
-    end
-
-    def initialize( txt )
-      @txt = txt
-    end
-
-    def parse
-      recs = []
-
-      parse_csv( @txt ).each do |row|
-        league_col = row['League']
-        season_col = row['Season'] || row['Year']
-        dates_col  = row['Dates']
-
-        season = Import::Season.parse( season_col )
-        league = catalog.leagues.find!( league_col )
-
-
-        dates = []
-        if dates_col.nil? || dates_col.empty?
-           ## do nothing; no dates - keep dates array empty
-        else
-          ## squish spaces
-          dates_col = dates_col.gsub( /[ ]{2,}/, ' ' )  ## squish/fold spaces
-
-          puts "#{league.name} (#{league.key}) | #{season.key} | #{dates_col}"
-
-          ### todo/check:  check what parts "Aug 15" return ???
-          ###                       short form for "Aug 15 -" - works?
-
-          ## todo/fix!!! - check EventInfo.include?
-          ##    now allow dates with only start_date too!! (WITHOUT end_date)
-          parts = dates_col.split( /[ ]*[–-][ ]*/ )
-          if parts.size == 1
-            pp parts
-            dates << DateFormats.parse( parts[0], start: Date.new( season.start_year, 1, 1 ), lang: 'en' )
-            pp dates
-          elsif parts.size == 2
-            pp parts
-            dates << DateFormats.parse( parts[0], start: Date.new( season.start_year, 1, 1 ), lang: 'en' )
-            dates << DateFormats.parse( parts[1], start: Date.new( season.end_year ? season.end_year : season.start_year, 1, 1 ), lang: 'en' )
-            pp dates
-
-            ## assert/check if period is less than 365 days for now
-            diff = dates[1].to_date.jd - dates[0].to_date.jd
-            puts "#{diff}d"
-            if diff > 365
-              puts "!! ERROR - date range / period assertion failed; expected diff < 365 days"
-              exit 1
-            end
-          else
-            puts "!! ERRROR - expected data range / period - one or two dates; got #{parts.size}:"
-            pp dates_col
-            pp parts
-            exit 1
-          end
-        end
-
-
-        teams_col    = row['Clubs'] || row['Teams']
-        goals_col    = row['Goals']
-
-        ## note: remove (and allow) all non-digits e.g. 370 goals, 20 clubs, etc.
-        teams_col    = teams_col.gsub( /[^0-9]/, '' )    if teams_col
-        goals_col    = goals_col.gsub( /[^0-9]/, '' )    if goals_col
-
-        teams  =   (teams_col.nil? || teams_col.empty?) ? nil : teams_col.to_i
-        goals  =   (goals_col.nil? || goals_col.empty?) ? nil : goals_col.to_i
-
-        matches_col  = row['Matches']
-        ## note: support additions in matches (played) e.g.
-        #   132 + 63 Play-off-Spiele
-        matches_col   = matches_col.gsub( /[^0-9+]/, '' )  if matches_col
-
-        matches =  if matches_col.nil? || matches_col.empty?
-                     nil
-                   else
-                     if matches_col.index( '+' )  ### check for calculations
-                        ## note: for now only supports additions
-                        matches_col.split( '+' ).reduce( 0 ) do |sum,str|
-                          sum + str.to_i
-                        end
-                     else            ## assume single (integer) number
-                        matches_col.to_i
-                     end
-                  end
-
-        rec = EventInfo.new( league:     league,
-                             season:     season,
-                             start_date: dates[0],
-                             end_date:   dates[1],
-                             teams:      teams,
-                             matches:    matches,
-                             goals:      goals
-                           )
-        recs << rec
-      end  # each row
-      recs
-    end  # method parse
-  end  # class EventInfoReader
-
-
-end ## module Import
-end ## module SportDb
-
+
+module SportDb
+module Import
+
+
+class EventInfo
+    ##  "high level" info (summary) about event  (like a "wikipedia infobox")
+    ##    use for checking dataset imports; lets you check e.g.
+    ##    - dates within range
+    ##    - number of teams e.g. 20
+    ##    - matches played e.g. 380
+    ##    - goals scored e.g. 937
+    ##    etc.
+
+    attr_reader :league,
+                :season,
+                :teams,
+                :matches,
+                :goals,
+                :start_date,
+                :end_date
+
+    def initialize( league:, season:,
+                    start_date: nil, end_date: nil,
+                    teams:   nil,
+                    matches: nil,
+                    goals:   nil )
+
+      @league      = league
+      @season      = season
+
+      @start_date  = start_date
+      @end_date    = end_date
+
+      @teams       = teams    ## todo/check: rename/use teams_count ??
+      @matches     = matches  ## todo/check: rename/use match_count ??
+      @goals       = goals
+    end
+
+    def include?( date )
+       ## todo/fix: add options e.g.
+       ##  - add delta/off_by_one or such?
+       ##  - add strict (for) only return true if date range (really) defined (no generic auto-rules)
+
+      ### note: for now allow off by one error (via timezone/local time errors)
+      ##    todo/fix: issue warning if off by one!!!!
+      if @start_date && @end_date
+        date >= (@start_date-1) &&
+        date <= (@end_date+1)
+      else
+        if @season.year?
+           # assume generic rule
+           ## same year e.g. Jan 1 - Dec 31; always true for now
+           date.year == @season.start_year
+        else
+           # assume generic rule
+           ##  July 1 - June 30 (Y+1)
+           ##  - todo/check -start for some countries/leagues in June 1 or August 1 ????
+           date >= Date.new( @season.start_year, 7, 1 ) &&
+           date <= Date.new( @season.end_year, 6, 30 )
+        end
+      end
+    end  # method include?
+    alias_method :between?, :include?
+  end # class EventInfo
+
+
+  class EventInfoReader
+    def catalog() Import.catalog; end
+
+
+    def self.read( path )
+      txt = File.open( path, 'r:utf-8') {|f| f.read }
+      new( txt ).parse
+    end
+
+    def self.parse( txt )
+      new( txt ).parse
+    end
+
+    def initialize( txt )
+      @txt = txt
+    end
+
+    def parse
+      recs = []
+
+      parse_csv( @txt ).each do |row|
+        league_col = row['League']
+        season_col = row['Season'] || row['Year']
+        dates_col  = row['Dates']
+
+        season = Import::Season.parse( season_col )
+        league = catalog.leagues.find!( league_col )
+
+
+        dates = []
+        if dates_col.nil? || dates_col.empty?
+           ## do nothing; no dates - keep dates array empty
+        else
+          ## squish spaces
+          dates_col = dates_col.gsub( /[ ]{2,}/, ' ' )  ## squish/fold spaces
+
+          puts "#{league.name} (#{league.key}) | #{season.key} | #{dates_col}"
+
+          ### todo/check:  check what parts "Aug 15" return ???
+          ###                       short form for "Aug 15 -" - works?
+
+          ## todo/fix!!! - check EventInfo.include?
+          ##    now allow dates with only start_date too!! (WITHOUT end_date)
+          parts = dates_col.split( /[ ]*[–-][ ]*/ )
+          if parts.size == 1
+            pp parts
+            dates << DateFormats.parse( parts[0], start: Date.new( season.start_year, 1, 1 ), lang: 'en' )
+            pp dates
+          elsif parts.size == 2
+            pp parts
+            dates << DateFormats.parse( parts[0], start: Date.new( season.start_year, 1, 1 ), lang: 'en' )
+            dates << DateFormats.parse( parts[1], start: Date.new( season.end_year ? season.end_year : season.start_year, 1, 1 ), lang: 'en' )
+            pp dates
+
+            ## assert/check if period is less than 365 days for now
+            diff = dates[1].to_date.jd - dates[0].to_date.jd
+            puts "#{diff}d"
+            if diff > 365
+              puts "!! ERROR - date range / period assertion failed; expected diff < 365 days"
+              exit 1
+            end
+          else
+            puts "!! ERRROR - expected data range / period - one or two dates; got #{parts.size}:"
+            pp dates_col
+            pp parts
+            exit 1
+          end
+        end
+
+
+        teams_col    = row['Clubs'] || row['Teams']
+        goals_col    = row['Goals']
+
+        ## note: remove (and allow) all non-digits e.g. 370 goals, 20 clubs, etc.
+        teams_col    = teams_col.gsub( /[^0-9]/, '' )    if teams_col
+        goals_col    = goals_col.gsub( /[^0-9]/, '' )    if goals_col
+
+        teams  =   (teams_col.nil? || teams_col.empty?) ? nil : teams_col.to_i
+        goals  =   (goals_col.nil? || goals_col.empty?) ? nil : goals_col.to_i
+
+        matches_col  = row['Matches']
+        ## note: support additions in matches (played) e.g.
+        #   132 + 63 Play-off-Spiele
+        matches_col   = matches_col.gsub( /[^0-9+]/, '' )  if matches_col
+
+        matches =  if matches_col.nil? || matches_col.empty?
+                     nil
+                   else
+                     if matches_col.index( '+' )  ### check for calculations
+                        ## note: for now only supports additions
+                        matches_col.split( '+' ).reduce( 0 ) do |sum,str|
+                          sum + str.to_i
+                        end
+                     else            ## assume single (integer) number
+                        matches_col.to_i
+                     end
+                  end
+
+        rec = EventInfo.new( league:     league,
+                             season:     season,
+                             start_date: dates[0],
+                             end_date:   dates[1],
+                             teams:      teams,
+                             matches:    matches,
+                             goals:      goals
+                           )
+        recs << rec
+      end  # each row
+      recs
+    end  # method parse
+  end  # class EventInfoReader
+
+
+end ## module Import
+end ## module SportDb
+

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

@@ -151,6 +151,7 @@ class LeagueOutlineReader   ## todo/check - rename to LeaguePageReader / LeagueP
     'Playoffs - Relegation',
     'Playoffs - Challenger',
     'Finals',
+    'Match 6th Place',  # e.g. Super League Greece 2012/13
 
     'Apertura',
     'Apertura - Liguilla',

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

@@ -1,168 +1,168 @@
-# encoding: utf-8
-
-
-module SportDb
-module Import
-
-
-class LeagueReader
-
-  def catalog() Import.catalog; end
-
-
-def self.read( path )   ## use - rename to read_file or from_file etc. - why? why not?
-  txt = File.open( path, 'r:utf-8' ) { |f| f.read }
-  parse( txt )
-end
-
-def self.parse( txt )
-  new( txt ).parse
-end
-
-
-
-include Logging
-
-def initialize( txt )
-  @txt = txt
-end
-
-def parse
-  recs = []
-  last_rec = nil
-
-  country  = nil    # last country
-  intl     = false  # is international (league/tournament/cup/competition)
-  clubs    = true   # or clubs|national teams
-
-  OutlineReader.parse( @txt ).each do |node|
-    if [:h1,:h2,:h3,:h4,:h5,:h6].include?( node[0] )
-      heading_level  = node[0][1].to_i
-      heading        = node[1]
-
-      logger.debug "heading #{heading_level} >#{heading}<"
-
-      if heading_level != 1
-        puts "** !!! ERROR !!! unsupported heading level; expected heading 1 for now only; sorry"
-        pp line
-        exit 1
-      else
-        logger.debug "heading (#{heading_level}) >#{heading}<"
-        last_heading = heading
-        ## map to country or international / int'l or national teams
-        if heading =~ /national team/i   ## national team tournament
-          country = nil
-          intl    = true
-          clubs   = false
-        elsif heading =~ /international|int'l/i  ## int'l club tournament
-          country = nil
-          intl    = true
-          clubs   = true
-        else
-          ## assume country in heading; allow all "formats" supported by parse e.g.
-          ##   Österreich • Austria (at)
-          ##   Österreich • Austria
-          ##   Austria
-          ##   Deutschland (de) • Germany
-          country = catalog.countries.parse( heading )
-          intl    = false
-          clubs   = true
-
-          ## check country code - MUST exist for now!!!!
-          if country.nil?
-            puts "!!! error [league reader] - unknown country >#{heading}< - sorry - add country to config to fix"
-            exit 1
-          end
-        end
-      end
-    elsif node[0] == :p   ## paragraph with (text) lines
-      lines = node[1]
-      lines.each do |line|
-
-      if line.start_with?( '|' )
-          ## assume continuation with line of alternative names
-          ##  note: skip leading pipe
-          values = line[1..-1].split( '|' )   # team names - allow/use pipe(|)
-          ## 1) strip (commerical) sponsor markers/tags e.g. $$ Liga $$BBV$$ MX
-          ## 2) strip and  squish (white)spaces
-          #   e.g. New York FC      (2011-)  => New York FC (2011-)
-          values = values.map { |value| value.gsub( '$', '' )
-                                             .gsub( /[ \t]+/, ' ' )
-                                             .strip  }
-          logger.debug "alt_names: #{values.join( '|' )}"
-
-          last_rec.alt_names += values
-      else
-        ## assume "regular" line
-        ##  check if starts with id  (todo/check: use a more "strict"/better regex capture pattern!!!)
-        if line =~ /^([a-z0-9][a-z0-9.]*)[ ]+(.+)$/
-          league_key  = $1
-          ## 1) strip (commercial) sponsor markers/tags e.g $$
-          ## 2) strip and squish (white)spaces
-          league_name = $2.gsub( '$', '' )
-                          .gsub( /[ \t]+/, ' ' )
-                          .strip
-
-          logger.debug "key: >#{league_key}<, name: >#{league_name}<"
-
-
-          alt_names_auto = []
-          if country
-            alt_names_auto << "#{country.key.upcase} #{league_key.upcase.gsub('.', ' ')}"
-            ## todo/check: add "hack" for cl (chile) and exclude?
-            ##             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.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.
-
-            ## auto-add with country prepended
-            ##   e.g. England Premier League, Austria Bundesliga etc.
-            ##  todo/check: also add variants with country alt name if present!!!
-            ##  todo/check: exclude cups or such from country + league name auto-add - why? why not?
-            alt_names_auto << "#{country.name} #{league_name}"
-          else   ## assume int'l (no country) e.g. champions league, etc.
-            ## only auto-add key (e.g. CL, EL, etc.)
-            alt_names_auto << league_key.upcase.gsub('.', ' ')   ## note: no country code (prefix/leading) used
-          end
-
-          ## pp alt_names_auto
-
-          ## prepend country key/code if country present
-          ##   todo/fix: only auto-prepend country if key/code start with a number (level) or incl. cup
-          ##    why? lets you "overwrite" key if desired - use it - why? why not?
-          if country
-            league_key = "#{country.key}.#{league_key}"
-          end
-
-          rec = League.new( key:            league_key,
-                            name:           league_name,
-                            alt_names_auto: alt_names_auto,
-                            country:        country,
-                            intl:           intl,
-                            clubs:          clubs)
-          recs << rec
-          last_rec = rec
-        else
-          puts "** !!! ERROR !!! missing key for (canonical) league name"
-          exit 1
-        end
-      end
-      end  # each line
-    else
-      puts "** !!! ERROR !!! [league reader] - unknown line type:"
-      pp node
-      exit 1
-    end
-    ## pp line
-  end
-  recs
-end # method parse
-
-end # class LeagueReader
-
-end ## module Import
-end ## module SportDb
+# encoding: utf-8
+
+
+module SportDb
+module Import
+
+
+class LeagueReader
+
+  def catalog() Import.catalog; end
+
+
+def self.read( path )   ## use - rename to read_file or from_file etc. - why? why not?
+  txt = File.open( path, 'r:utf-8' ) { |f| f.read }
+  parse( txt )
+end
+
+def self.parse( txt )
+  new( txt ).parse
+end
+
+
+
+include Logging
+
+def initialize( txt )
+  @txt = txt
+end
+
+def parse
+  recs = []
+  last_rec = nil
+
+  country  = nil    # last country
+  intl     = false  # is international (league/tournament/cup/competition)
+  clubs    = true   # or clubs|national teams
+
+  OutlineReader.parse( @txt ).each do |node|
+    if [:h1,:h2,:h3,:h4,:h5,:h6].include?( node[0] )
+      heading_level  = node[0][1].to_i
+      heading        = node[1]
+
+      logger.debug "heading #{heading_level} >#{heading}<"
+
+      if heading_level != 1
+        puts "** !!! ERROR !!! unsupported heading level; expected heading 1 for now only; sorry"
+        pp line
+        exit 1
+      else
+        logger.debug "heading (#{heading_level}) >#{heading}<"
+        last_heading = heading
+        ## map to country or international / int'l or national teams
+        if heading =~ /national team/i   ## national team tournament
+          country = nil
+          intl    = true
+          clubs   = false
+        elsif heading =~ /international|int'l/i  ## int'l club tournament
+          country = nil
+          intl    = true
+          clubs   = true
+        else
+          ## assume country in heading; allow all "formats" supported by parse e.g.
+          ##   Österreich • Austria (at)
+          ##   Österreich • Austria
+          ##   Austria
+          ##   Deutschland (de) • Germany
+          country = catalog.countries.parse( heading )
+          intl    = false
+          clubs   = true
+
+          ## check country code - MUST exist for now!!!!
+          if country.nil?
+            puts "!!! error [league reader] - unknown country >#{heading}< - sorry - add country to config to fix"
+            exit 1
+          end
+        end
+      end
+    elsif node[0] == :p   ## paragraph with (text) lines
+      lines = node[1]
+      lines.each do |line|
+
+      if line.start_with?( '|' )
+          ## assume continuation with line of alternative names
+          ##  note: skip leading pipe
+          values = line[1..-1].split( '|' )   # team names - allow/use pipe(|)
+          ## 1) strip (commerical) sponsor markers/tags e.g. $$ Liga $$BBV$$ MX
+          ## 2) strip and  squish (white)spaces
+          #   e.g. New York FC      (2011-)  => New York FC (2011-)
+          values = values.map { |value| value.gsub( '$', '' )
+                                             .gsub( /[ \t]+/, ' ' )
+                                             .strip  }
+          logger.debug "alt_names: #{values.join( '|' )}"
+
+          last_rec.alt_names += values
+      else
+        ## assume "regular" line
+        ##  check if starts with id  (todo/check: use a more "strict"/better regex capture pattern!!!)
+        if line =~ /^([a-z0-9][a-z0-9.]*)[ ]+(.+)$/
+          league_key  = $1
+          ## 1) strip (commercial) sponsor markers/tags e.g $$
+          ## 2) strip and squish (white)spaces
+          league_name = $2.gsub( '$', '' )
+                          .gsub( /[ \t]+/, ' ' )
+                          .strip
+
+          logger.debug "key: >#{league_key}<, name: >#{league_name}<"
+
+
+          alt_names_auto = []
+          if country
+            alt_names_auto << "#{country.key.upcase} #{league_key.upcase.gsub('.', ' ')}"
+            ## todo/check: add "hack" for cl (chile) and exclude?
+            ##             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.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.
+
+            ## auto-add with country prepended
+            ##   e.g. England Premier League, Austria Bundesliga etc.
+            ##  todo/check: also add variants with country alt name if present!!!
+            ##  todo/check: exclude cups or such from country + league name auto-add - why? why not?
+            alt_names_auto << "#{country.name} #{league_name}"
+          else   ## assume int'l (no country) e.g. champions league, etc.
+            ## only auto-add key (e.g. CL, EL, etc.)
+            alt_names_auto << league_key.upcase.gsub('.', ' ')   ## note: no country code (prefix/leading) used
+          end
+
+          ## pp alt_names_auto
+
+          ## prepend country key/code if country present
+          ##   todo/fix: only auto-prepend country if key/code start with a number (level) or incl. cup
+          ##    why? lets you "overwrite" key if desired - use it - why? why not?
+          if country
+            league_key = "#{country.key}.#{league_key}"
+          end
+
+          rec = League.new( key:            league_key,
+                            name:           league_name,
+                            alt_names_auto: alt_names_auto,
+                            country:        country,
+                            intl:           intl,
+                            clubs:          clubs)
+          recs << rec
+          last_rec = rec
+        else
+          puts "** !!! ERROR !!! missing key for (canonical) league name"
+          exit 1
+        end
+      end
+      end  # each line
+    else
+      puts "** !!! ERROR !!! [league reader] - unknown line type:"
+      pp node
+      exit 1
+    end
+    ## pp line
+  end
+  recs
+end # method parse
+
+end # class LeagueReader
+
+end ## module Import
+end ## module SportDb