sportdb-formats 1.1.6 → 1.2.1

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

@@ -1,183 +1,184 @@
-
-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 = 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/goals.rb

@@ -1,4 +1,3 @@
-# encoding: utf-8
 
 module SportDb
 
@@ -15,6 +14,15 @@ class GoalsPlayerStruct
   def initialize
     @minutes = []
   end
+
+  def pretty_print( printer ) 
+    buf = String.new
+    buf << "<GoalsPlayerStruct: #{@name} "
+    buf << @minutes.pretty_print_inspect
+    buf << ">"
+
+    printer.text( buf ) 
+  end
 end
 
 
@@ -27,6 +35,18 @@ class GoalsMinuteStruct
     @penalty = false
     @owngoal = false
   end
+
+  def pretty_print( printer ) 
+    buf = String.new
+    buf << "<GoalsMinuteStruct: #{@minute}"
+    buf << "+#{@offset}"    if @offset && @offset > 0
+    buf << "'"
+    buf << " (o.g.)"  if @owngoal
+    buf << " (pen.)"  if @penalty
+    buf << ">"
+
+    printer.text( buf ) 
+  end
 end
 
 
@@ -59,6 +79,22 @@ class GoalStruct
   def state
     [@name, @team, @minute, @offset, @penalty, @owngoal, @score1, @score2]
   end
+
+
+  def pretty_print( printer ) 
+    buf = String.new
+    buf << "<GoalStruct: #{@score1}-#{@score2} #{@name} #{@minute}"
+    buf << "+#{@offset}"    if @offset && @offset > 0
+    buf << "'"
+    buf << " (o.g.)"  if @owngoal
+    buf << " (pen.)"  if @penalty
+    buf << " for #{@team}"     ### team 1 or 2 - use home/away
+    buf << ">"
+
+    printer.text( buf ) 
+  end
+
+
 end
 
 
@@ -173,6 +209,10 @@ class GoalsParser
   include LogUtils::Logging
 
 
+  ### todo/fix:
+  ##    let's use stringscanner for parsing line - why? why not?
+
+
   # note: use ^ for start of string only!!!
   # - for now slurp everything up to digits (inlc. spaces - use strip to remove)
   # todo/check: use/rename to NAME_UNTIL_REGEX ??? ( add lookahead for spaces?)
@@ -184,18 +224,21 @@ class GoalsParser
   # todo/check: change to MINUTE_REGEX ??
   # add MINUTE_SKIP_REGEX or MINUTE_SEP_REGEX /^[ ,]+/
   # todo/fix:  split out  penalty and owngoal flag in PATTERN constant for reuse
+  #   note - offset  90+10 possible!!!!
+  #    note - allow  p/pen./pen  or  o.g. or og
   MINUTES_REGEX = /^      # note: use ^ for start of string only!!!
                     (?<minute>[0-9]{1,3})
                     (?:\+
-                      (?<offset>[1-9]{1})
+                      (?<offset>[0-9]{1,2})
                     )?
                     '
                     (?:[ ]*
                       \(
-                      (?<type>P|pen\.|o\.g\.)
+                      (?<type>p|pen\.?|
+                              og|o\.g\.)
                       \)
                     )?
-                  /x
+                  /ix
 
 
 
@@ -220,22 +263,21 @@ class GoalsParser
       player = GoalsPlayerStruct.new
       player.name = name
 
-      minute_hash = get_minute_hash!( line )
-      while minute_hash
+      minute_hash = nil 
+      while minute_hash=get_minute_hash!( line )   ## note: returns nil if no (regex) match
         logger.debug "  found minutes >#{minute_hash.inspect}< - remaining >#{line}<"
 
         minute = GoalsMinuteStruct.new
         minute.minute = minute_hash[:minute].to_i
         minute.offset = minute_hash[:offset].to_i  if minute_hash[:offset]
         if minute_hash[:type]
-          minute.owngoal = true  if minute_hash[:type] =~ /o\.g\./
-          minute.penalty = true  if minute_hash[:type] =~ /P|pen\./
+          minute.owngoal = true  if minute_hash[:type] =~ /og|o\.g\./i
+          minute.penalty = true  if minute_hash[:type] =~ /p|pen\.?/i
         end
         player.minutes << minute
 
         # remove commas and spaces (note: use ^ for start of string only!!!)
         line.sub!( /^[ ,]+/, '' )
-        minute_hash = get_minute_hash!( line )
       end
 
       players << player
@@ -261,6 +303,8 @@ private
     m = MINUTES_REGEX.match( line ) # note: use ^ for start of string only!!!
     if m
       h = {}
+      ## todo/fix - hash conversion no longer need in ruby 3+!!
+      ##               double check - and remove (simplify) !!!!
       # - note: do NOT forget to turn name into symbol for lookup in new hash (name.to_sym)
       m.names.each { |n| h[n.to_sym] = m[n] } # or use match_data.names.zip( match_data.captures ) - more cryptic but "elegant"??