sportdb-formats 1.1.6 → 1.2.1

data/lib/sportdb/formats/match/match_parser.rb

@@ -1,4 +1,3 @@
-# encoding: utf-8
 
 module SportDb
 
@@ -21,13 +20,44 @@ class MatchParser   ## simple match parser for team match schedules
     # for convenience split string into lines
     ##    note: removes/strips empty lines
     ## todo/check: change to text instead of array of lines - why? why not?
-    @lines        = lines.is_a?( String ) ? read_lines( lines ) : lines
-
+ 
+    ## note - wrap in enumerator/iterator a.k.a lines reader
+    @lines = LinesReader.new( lines.is_a?( String ) ?
+                                 read_lines( lines ) : 
+                                 lines 
+                            )
+ 
     @mapper_teams = TeamMapper.new( teams )
     @start        = start
   end
 
 
+  
+
+  ## note:  colon (:) MUST be followed by one (or more) spaces
+  ##      make sure mon feb 12 18:10 will not match
+  ##        allow 1. FC Köln etc.
+  ##               Mainz 05:
+  ##           limit to 30 chars max
+  ##          only allow  chars incl. intl buut (NOT ()[]/;)
+  ##
+  ##   Group A:
+  ##   Group B:   - remove colon
+  ##    or lookup first
+
+  ATTRIB_REGEX = /^  
+                   [ ]*?     # slurp leading spaces
+                (?<key>[^:|\]\[()\/; -]
+                       [^:|\]\[()\/;]{0,30}
+                 )
+                   [ ]*?     # slurp trailing spaces
+                   :[ ]+
+                (?<value>.+)
+                    [ ]*?   # slurp trailing spaces
+                   $
+                /ix
+
+
   def parse
     @last_date    = nil
     @last_round   = nil
@@ -39,11 +69,10 @@ class MatchParser   ## simple match parser for team match schedules
 
     @warns        = []    ## track list of warnings (unmatched lines)  too - why? why not?
 
-
+ ## todo/fix - use @lines.rewind first  here - why? why not?
     @lines.each do |line|
-      if is_goals?( line )
-        logger.debug "skipping matched goals line: >#{line}<"
-      elsif is_round_def?( line )
+       
+      if is_round_def?( line )
         ## todo/fix:  add round definition (w begin n end date)
         ## todo: do not patch rounds with definition (already assume begin/end date is good)
         ##  -- how to deal with matches that get rescheduled/postponed?
@@ -56,6 +85,32 @@ class MatchParser   ## simple match parser for team match schedules
       elsif is_group?( line )
         ##  -- lets you set group  e.g. Group A etc.
         parse_group_header( line )
+
+      elsif m=ATTRIB_REGEX.match( line )
+         ## note: check attrib regex AFTER group def e.g.:
+         ##         Group A: 
+         ##         Group B:  etc.
+         ##     todo/fix - change Group A: to Group A etc.
+         ##                       Group B: to Group B
+
+         ## check if line ends with dot
+         ##  if not slurp up lines to the next do!!!
+         logger.debug "skipping key/value line - >#{line}<"
+         while !line.end_with?( '.' ) || line.nil? do
+             line = @lines.next
+             logger.debug "skipping key/value line (cont.) - >#{line}<"
+         end       
+       elsif is_goals?( line )
+          ## note - goals must be AFTER attributes!!!
+          logger.debug "matched goals line: >#{line}<"
+          logger.debug "  try parse:"
+  
+          goals = GoalsFinder.new.find!( line )
+          pp goals 
+          ## quick & dirty - auto add goals to last match
+          match = @matches[-1]
+          match.goals = goals
+       
       elsif try_parse_game( line )
         # do nothing here
       elsif try_parse_date_header( line )
@@ -307,6 +362,27 @@ class MatchParser   ## simple match parser for team match schedules
   end
 
 
+  ### todo/check - include (optional) leading space in regex - why? why not?
+  NUM_RE = /^[ ]*\(
+                    (?<num>[0-9]{1,3})
+                  \)
+           /x
+
+  def find_num!( line )
+      ## check for leading match number e.g.
+      ##    (1) Fri Jun/14 21:00         Germany   5-1 (3-0)  Scotland 
+      m = line.match( NUM_RE )
+      if m
+        num = m[:num].to_i(10)  ## allows 01/02/07 etc. -- why? why not?
+        match_str = m[0]
+        line.sub!( match_str, '[NUM]' )
+        num
+      else
+        nil
+      end
+  end
+
+
   def try_parse_game( line )
     # note: clone line; for possible test do NOT modify in place for now
     # note: returns true if parsed, false if no match
@@ -317,11 +393,27 @@ class MatchParser   ## simple match parser for team match schedules
   def parse_game( line )
     logger.debug "parsing game (fixture) line: >#{line}<"
 
-    ## split by geo (@) - remove for now
+    ## split by geo (@)
     ##   split into parts e.g. break using @ !!!
     values = line.split( '@' )
-    line = values[0]
 
+    ## for now pass along ground, city (timezone) as string as is
+    ##         parse (map) later - why? why not??
+    ### check for ground/stadium and cities
+    ground =  if values.size == 1
+                 nil ## no stadium
+              elsif values.size == 2   # bingo!!!
+                 ## process stadium, city (timezone) etc.
+                 ## for now keep it simple - pass along "unparsed" all-in-one
+                 values[1].gsub( /[ \t]+/, ' ').strip   ## squish
+              else
+                 puts "!! ERROR - too many @-markers found in line:"
+                 puts line
+                 exit 1
+              end
+
+
+    line   = values[0]
 
     @mapper_teams.map_teams!( line )   ### todo/fix: limit mapping to two(2) teams - why? why not?  might avoid matching @ Barcelona ??
     teams = @mapper_teams.find_teams!( line )
@@ -334,13 +426,23 @@ class MatchParser   ## simple match parser for team match schedules
       return false
     end
 
+
+    ## try optional match number e.g. 
+    ##        (1) Fri Jun/14 21:00         Germany   5-1 (3-0)  Scotland 
+    num = find_num!( line )
+    ## pos = find_game_pos!( line )
+
     ## find (optional) match status e.g. [abandoned] or [replay] or [awarded]
     ##                                   or [cancelled] or [postponed] etc.
     status = find_status!( line )   ## todo/check: allow match status also in geo part (e.g. after @) - why? why not?
 
-    ## pos = find_game_pos!( line )
 
-      date      = find_date!( line, start: @start )
+      date      = find_date!( line, start: @start )  ## date or datetime (but NOT time!)
+
+    ## todo/fix:
+    ##   add support for find_time!   e.g. 21.00 (or 21:00 ?)
+
+
 
     ###
     # check if date found?
@@ -387,13 +489,28 @@ class MatchParser   ## simple match parser for team match schedules
 
     ## todo/check: pass along round and group refs or just string (canonical names) - why? why not?
 
-    @matches << Import::Match.new( date:    date,
+
+    ## split date in date & time if DateTime
+    time_str = nil
+    date_str = nil
+    if date.is_a?( DateTime )
+        date_str = date.strftime('%Y-%m-%d')
+        time_str = date.strftime('%H:%M')
+    elsif date.is_a?( Date )
+        date_str = date.strftime('%Y-%m-%d')
+    else  # assume date is nil
+    end        
+
+    @matches << Import::Match.new( num:     num,
+                                   date:    date_str,
+                                   time:    time_str,
                                    team1:   team1,  ## note: for now always use mapping value e.g. rec (NOT string e.g. team1.name)
                                    team2:   team2,  ## note: for now always use mapping value e.g. rec (NOT string e.g. team2.name)
                                    score:   score,
                                    round:   round       ? round.name       : nil,   ## note: for now always use string (assume unique canonical name for event)
                                    group:   @last_group ? @last_group.name : nil,   ## note: for now always use string (assume unique canonical name for event)
-                                   status:  status )
+                                   status:  status,
+                                   ground:  ground )
     ### todo: cache team lookups in hash?
 
 =begin