sportdb-parser 0.7.1 → 0.7.2

data/lib/sportdb/parser/racc_parser.rb

@@ -5,61 +5,70 @@
 class RaccMatchParser
 
 
-def initialize( txt,  debug: false )
-    ## puts "==> txt:"
-    ## puts txt
- 
-    @tree   = [] 
+
+  def initialize( txt,  debug: false )
+    @tree   = []
     @errors = []
 
-    ### todo:
-    ##  -  pass along debug flag
     lexer = SportDb::Lexer.new( txt, debug: debug )
     ##  note - use tokenize_with_errors and add/collect tokenize errors
     @tokens, @errors = lexer.tokenize_with_errors
     ## pp @tokens
-    
-    ## quick hack - convert to racc format single char literal tokens e.g. '@' etc.
-    @tokens = @tokens.map do |tok|
-                 if tok.size == 1
-                   [tok[0].to_s, tok[0].to_s]
-                 else 
-                   tok
-                 end
-               end
+
   end
 
 
   def debug( value ) @debug = value; end   ### fix: use setter-style e.g. debug=(value) !!!
   def debug?()  @debug == true; end
 
+  ###
+  ###  fix-fix-fix   rename   to _trace
+  ##                   check lexer - add [debug] Parser -  or such!!!
+  ##
   ## debug - trace / print message
   def trace( msg )
-     puts "  [parse] " + msg    if debug?
+    puts " fix-fix-fix use _trace() in parser!!"
+    puts "  [parse] " + msg
   end
 
-
+  def _trace( *args )
+    ## if debug?
+      print "[DEBUG] Parser -- "
+      args.each { |arg| puts args }
+    ## end
+  end
 
 
   def next_token
     tok = @tokens.shift
-    trace( "next_token => #{tok.pretty_inspect}" )
+   _trace( "next_token => #{tok.pretty_inspect}" )
+
+    ##  convert to racc format single char literal tokens e.g. '@' etc.
+    ##                  note - literal token MUST be string (NOT symbol)
+    ##    note - racc expects array with to items
+    ##               -  item[0] is the token id
+    ##               -  item[1] is the token value
+
+    ## note - returns nil for end-of-file !!!
+    tok = [tok.type, tok]   if tok
+
     tok
   end
-  
+
+
 #  on_error do |error_token_id, error_value, value_stack|
 #      puts "Parse error on token: #{error_token_id}, value: #{error_value}"
-#  end  
+#  end
 
-  def parse_with_errors    
-     trace( "start parse:" )
+  def parse_with_errors
+     _trace( "start parse:" )
      do_parse
      [@tree, @errors]
   end
 
   def parse  ## convenience shortcut (ignores errors)
     tree, _ = parse_with_errors
-    tree 
+    tree
   end
 
 
@@ -69,20 +78,15 @@ def initialize( txt,  debug: false )
 
   def on_error(error_token_id, error_value, value_stack)
     ## auto-add error_token (as string)
-    error_token = Racc_token_to_s_table[error_token_id] 
+    error_token = Racc_token_to_s_table[error_token_id]
     args = [error_token, error_token_id, error_value, value_stack]
+
     puts
     puts "!! on parse error:"
     puts "args=#{args.pretty_inspect}"
 
-    @errors << "parse error on token: #{error_token} (#{error_token_id}) with value: #{error_value}, stack: #{value_stack.pretty_inspect}" 
-    ## exit 1  ##   exit for now  -  get and print more info about context etc.!!
+    @errors << "parse error on token: #{error_token} >#{error_value.pretty_inspect}<, stack: #{value_stack.pretty_inspect}"
   end
 
 
-=begin
-on_error do |error_token_id, error_value, value_stack|
-    puts "Parse error on token: #{error_token_id}, value: #{error_value}"
-end
-=end
 end   # class RaccMatchParser

data/lib/sportdb/parser/racc_tree.rb

@@ -5,7 +5,7 @@
 class RaccMatchParser
 
 =begin
-RefereeLine = Struct.new( :name, :country ) do 
+RefereeLine = Struct.new( :name, :country ) do
   def pretty_print( printer )
     printer.text( "<RefereeLine " )
     printer.text( self.name )
@@ -16,7 +16,7 @@ end
 =end
 
 ## support multiple referees (incl. assistant refs etc.)
-RefereeLine = Struct.new( :referees ) do 
+RefereeLine = Struct.new( :referees ) do
   def pretty_print( printer )
     printer.text( "<RefereeLine " )
     printer.text( self.referees.pretty_inspect )
@@ -28,16 +28,16 @@ Referee = Struct.new( :name, :country ) do
   def to_s
     buf = String.new
     buf <<  self.name
-    buf << " (#{self.country})"    if self.country 
+    buf << " (#{self.country})"    if self.country
     buf
   end
 
   def pretty_print( printer )
     printer.text( to_s )
-  end  
+  end
 end
 
-AttendanceLine = Struct.new( :att ) do 
+AttendanceLine = Struct.new( :att ) do
   def pretty_print( printer )
     printer.text( "<AttendanceLine #{self.att}>" )
   end
@@ -45,7 +45,7 @@ end
 
 
 
-PenaltiesLine = Struct.new( :penalties ) do   
+PenaltiesLine = Struct.new( :penalties ) do
   def pretty_print( printer )
     printer.text( "<PenaltiesLine " )
     printer.text( self.penalties.pretty_inspect )
@@ -59,21 +59,21 @@ Penalty = Struct.new( :name, :score, :note ) do
     buf = String.new
     buf << "#{self.score[0]}-#{self.score[1]} "   if self.score
     buf <<  self.name
-    buf << " (#{self.note})"    if self.note 
+    buf << " (#{self.note})"    if self.note
     buf
   end
 
 
   def pretty_print( printer )
     printer.text( to_s )
-  end  
+  end
 end
 
 
 
 
 ##  find a better name for player (use bookings?) - note - red/yellow card for trainer possible
-CardsLine = Struct.new( :type, :bookings ) do   
+CardsLine = Struct.new( :type, :bookings ) do
   def pretty_print( printer )
     printer.text( "<CardsLine " )
     printer.text( self.type )
@@ -92,7 +92,7 @@ Booking = Struct.new( :name, :minute ) do
 
   def pretty_print( printer )
     printer.text( to_s )
-  end  
+  end
 end
 
 
@@ -111,11 +111,11 @@ end
 Lineup     = Struct.new( :name, :captain, :cards, :sub ) do
   def pretty_print( printer )
     buf = String.new
-    buf <<  self.name 
+    buf <<  self.name
     buf << " [c]"   if captain
     buf << " cards=" + self.cards.pretty_inspect    if cards
     buf << " sub=" + self.sub.pretty_inspect      if sub
-    printer.text( buf ) 
+    printer.text( buf )
   end
 end
 
@@ -130,18 +130,18 @@ Card       = Struct.new( :name, :minute ) do
 
   def pretty_print( printer )
     printer.text( to_s )
-  end  
+  end
 end
 
 
 Sub        = Struct.new( :minute, :sub )  do
   def pretty_print( printer )
-    buf = String.new 
+    buf = String.new
     buf << "("    ## note - possibly recursive (thus, let minute go first/print first/upfront)
-    buf << "#{self.minute.to_s} "   if self.minute  
-    buf << "#{self.sub.pretty_inspect}" 
+    buf << "#{self.minute.to_s} "   if self.minute
+    buf << "#{self.sub.pretty_inspect}"
     buf << ")"
-    printer.text( buf ) 
+    printer.text( buf )
   end
 end
 
@@ -201,13 +201,6 @@ class BlankLine
   end
 end
 
-## todo/check - find a better name for hruler - divider?  or ??? - why? why not?
-class HRuler   ## h(orizontal) ruler (for breaks; new scopes) 
-  def pretty_print( printer )
-    printer.text( "<HRuler>" )
-  end
-end
-
 
 
 NoteLine = Struct.new( :text ) do
@@ -223,32 +216,6 @@ NotaBene = Struct.new( :text ) do
 end
 
 
-##  todo/check - rename TableHeading to TableHeader - why? why not?
-TableHeading = Struct.new( :text ) do
-  def pretty_print( printer )
-    printer.text( "<TableHeading #{self.text}>" )
-  end
-end
-
-##  todo/check - rename TableDivider to TableRule/TableRuler/TableLine - why? why not?
-TableDivider = Struct.new( :text ) do
-  def pretty_print( printer )
-    printer.text( "<TableDivider #{self.text}>" )
-  end
-end
-
-TableLine = Struct.new( :text ) do
-  def pretty_print( printer )
-    printer.text( "<TableLine #{self.text}>" )
-  end
-end
-
-TableNote = Struct.new( :text ) do
-  def pretty_print( printer )
-    printer.text( "<TableNote #{self.text}>" )
-  end
-end
-
 
 
 ## todo/check - use a generic Heading instead of Heading1/2/3 - why? why not?
@@ -278,7 +245,7 @@ MatchLineBye   = Struct.new( :team, :note ) do
     printer.text( "#{self.team} bye")
     printer.text( " note=#{self.note.pretty_inspect}" )  if self.note
     printer.text( ">" )
-  end  
+  end
 end
 
 MatchLineWalkover   = Struct.new( :team1, :team2, :note ) do
@@ -287,10 +254,10 @@ MatchLineWalkover   = Struct.new( :team1, :team2, :note ) do
     printer.text( "#{self.team1} w/o #{self.team2}")
     printer.text( " note=#{self.note.pretty_inspect}" )  if self.note
     printer.text( ">" )
-  end  
+  end
 end
 
-MatchLineLegs   = Struct.new( :team1, :team2, 
+MatchLineLegs   = Struct.new( :team1, :team2,
                               :score )  do   ## change to geos - why? why not?
   def pretty_print( printer )
     printer.text( "<MatchLineLegs " )
@@ -300,26 +267,26 @@ MatchLineLegs   = Struct.new( :team1, :team2,
     members.zip(values) do |name, value|
       next if [:team1, :team2].include?( name )
       next if value.nil?
-      
+
       printer.text( "#{name}=#{value.pretty_inspect}" )
-    end    
+    end
 
     printer.text( ">" )
-  end  
+  end
 end
 
 
 #
 #  note: use two status attributes for now
 #         1) inline_status and 2) (note_)status
-#              for now  e.g.  A abd. B     vs   A  v B [abadoned] 
+#              for now  e.g.  A abd. B     vs   A  v B [abadoned]
 #                             A 3-0 awd B  vs   A  3-0 B [awarded]
 #   note - BOTH might be present at the same time
 
 
 MatchLine   = Struct.new( :header,  :tty,   ## tty = TELETYPE MODE for teams and score!!
                           :num, :date, :time, :time_local, :year,
-                          :team1, :team2, 
+                          :team1, :team2,
                           :score,
                           :status,  :status_inline, :status_note,
                           :geo,
@@ -335,12 +302,12 @@ MatchLine   = Struct.new( :header,  :tty,   ## tty = TELETYPE MODE for teams and
     members.zip(values) do |name, value|
       next if [:team1, :team2].include?( name )
       next if value.nil?
-      
+
       printer.text( "#{name}=#{value.pretty_inspect}" )
-    end    
+    end
 
     printer.text( ">" )
-  end  
+  end
 
 end
 
@@ -351,12 +318,12 @@ GoalLine    = Struct.new( :goals1, :goals2 ) do
     printer.text( "goals1=" + self.goals1.pretty_inspect + "," )
     printer.breakable
     printer.text( "goals2=" + self.goals2.pretty_inspect + ">" )
-  end  
+  end
 
 ##  def clone
-##         _clone = GoalLine.new( goals1: goals1.clone, 
+##         _clone = GoalLine.new( goals1: goals1.clone,
 ##                           goals2: goals2.clone )
-##                  
+##
 ##         puts "[debug] clone #{self.pretty_inspect} => #{_clone.pretty_inspect}"
 ##
 ##         _clone
@@ -386,7 +353,7 @@ Goal        = Struct.new( :player, :minutes, :count ) do
 
   def pretty_print( printer )
     printer.text( to_s )
-  end  
+  end
 
 ##  def clone
 ##      puts "[debug] clone #{self.pretty_inspect}"
@@ -402,7 +369,7 @@ GoalLineAlt = Struct.new( :goals ) do
   def pretty_print( printer )
     printer.text( "<GoalLineAlt " )
     printer.text( "goals=" + self.goals.pretty_inspect + ">" )
-  end  
+  end
 end
 
 
@@ -423,7 +390,7 @@ GoalAlt   = Struct.new( :score, :player, :minute, :goal_type ) do
 
   def pretty_print( printer )
     printer.text( to_s )
-  end  
+  end
 end
 
 
@@ -432,7 +399,7 @@ GoalLineCompat = Struct.new( :goals ) do
   def pretty_print( printer )
     printer.text( "<GoalLineCompat " )
     printer.text( "goals=" + self.goals.pretty_inspect + ">" )
-  end  
+  end
 end
 
 ##
@@ -443,7 +410,7 @@ end
 GoalCompat    = Struct.new( :score, :player, :minute, :goal_type ) do
   def to_s
     buf = String.new
-    buf << "#{self.minute}"    
+    buf << "#{self.minute}"
     buf << " #{self.player}"
     buf << " #{self.goal_type}"                  if self.goal_type
     buf << " #{self.score[0]}-#{self.score[1]}"  if self.score
@@ -452,7 +419,7 @@ GoalCompat    = Struct.new( :score, :player, :minute, :goal_type ) do
 
   def pretty_print( printer )
     printer.text( to_s )
-  end  
+  end
 end
 
 
@@ -477,31 +444,31 @@ GoalMinute      = Struct.new( :m, :offset, :secs,
       buf = String.new
       buf << "#{self.m}"
       buf << "'"
-      buf << "+#{self.offset}"  if self.offset 
-      buf << " (og)"   if self.og
-      buf << " (pen)"  if self.pen
-      buf << " (f)"  if self.freekick
-      buf << " (h)"  if self.header
+      buf << "+#{self.offset}"  if self.offset
+      buf << "(og)"   if self.og
+      buf << "(p)"  if self.pen
+      buf << "(f)"  if self.freekick
+      buf << "(h)"  if self.header
       buf << " (#{self.secs} secs)"   if self.secs
       buf
     end
- 
-    def pretty_print( printer ) 
-       printer.text( to_s ) 
-    end  
+
+    def pretty_print( printer )
+       printer.text( to_s )
+    end
 
     ### quick hack:
     ### split struct into  Minute+GoalType structs
     def to_minute
-        Minute.new( m:      self.m, 
+        Minute.new( m:      self.m,
                     offset: self.offset,
                     secs:   self.secs )
     end
 
     def to_goal_type
-        if self.og || self.pen || self.header || self.freekick   
-          GoalType.new( og:       self.og, 
-                        pen:      self.pen, 
+        if self.og || self.pen || self.header || self.freekick
+          GoalType.new( og:       self.og,
+                        pen:      self.pen,
                         header:   self.header,
                         freekick: self.freekick )
         else
@@ -511,14 +478,14 @@ GoalMinute      = Struct.new( :m, :offset, :secs,
 
 ##    def clone
 ##        puts "[debug] clone #{self.pretty_inspect}"
-##        GoalMinute.new( m: m.clone, 
-##                        offset: offset.clone, 
+##        GoalMinute.new( m: m.clone,
+##                        offset: offset.clone,
 ##                        secs: secs.clone,
-##                        og: og.clone, 
-##                        pen: pen.clone, 
-##                        header: header.clone, 
+##                        og: og.clone,
+##                        pen: pen.clone,
+##                        header: header.clone,
 ##                        freekick: freekick.clone )
-##                 
+##
 ##    end
 end
 
@@ -532,10 +499,10 @@ GoalType  = Struct.new( :og, :pen, :header, :freekick )  do
       buf << "(h)"    if self.header
       buf
     end
- 
-    def pretty_print( printer ) 
-       printer.text( to_s ) 
-    end  
+
+    def pretty_print( printer )
+       printer.text( to_s )
+    end
 end
 
 
@@ -544,14 +511,14 @@ Minute      = Struct.new( :m, :offset, :secs )  do
       buf = String.new
       buf << "#{self.m}"
       buf << "'"
-      buf << "+#{self.offset}"      if self.offset 
+      buf << "+#{self.offset}"      if self.offset
       buf << "/#{self.secs} secs"   if self.secs
       buf
     end
- 
-    def pretty_print( printer ) 
-       printer.text( to_s ) 
-    end  
+
+    def pretty_print( printer )
+       printer.text( to_s )
+    end
 end
 
 end  # class RaccMatchParser

data/lib/sportdb/parser/token-date--helpers.rb

@@ -0,0 +1,130 @@
+module SportDb
+class Lexer
+
+
+##  "internal" date helpers
+def self._build_date( m )
+            date = {}
+         ## map month names
+         ## note - allow any/upcase JULY/JUL etc. thus ALWAYS downcase for lookup
+            date[:y]  = m[:year].to_i(10)  if m[:year]
+            ## check - use y too for two-digit year or keep separate - why? why not?
+            date[:yy] = m[:yy].to_i(10)    if m[:yy]    ## two digit year (e.g. 25 or 78 etc.)
+            date[:m] = m[:month].to_i(10)  if m[:month]
+            date[:m] = MONTH_MAP[ m[:month_name].downcase ]   if m[:month_name]
+            date[:d]  = m[:day].to_i(10)   if m[:day]
+            date[:wday] = DAY_MAP[ m[:day_name].downcase ]   if m[:day_name]
+
+            date
+end
+
+def self._build_date_legs( m )
+           legs = {}
+        ## map month names
+         ## note - allow any/upcase JULY/JUL etc. thus ALWAYS downcase for lookup
+            date = {}
+            date[:m] = MONTH_MAP[ m[:month_name1].downcase ]
+            date[:d]  = m[:day1].to_i(10)
+            legs[:date1] = date
+
+            date = {}
+            date[:m] = MONTH_MAP[ m[:month_name2].downcase ]   if m[:month_name2]
+            date[:d]  = m[:day2].to_i(10)
+            legs[:date2] = date
+
+            legs
+end
+
+
+def self._build_duration( m )
+            ## todo/check/fix - if end: works for kwargs!!!!!
+            duration = { start: {}, end: {}}
+
+            duration[:start][:y] = m[:year1].to_i(10)  if m[:year1]
+            duration[:start][:m] = MONTH_MAP[ m[:month_name1].downcase ]   if m[:month_name1]
+            duration[:start][:d]  = m[:day1].to_i(10)   if m[:day1]
+            duration[:start][:wday] = DAY_MAP[ m[:day_name1].downcase ]   if m[:day_name1]
+
+            duration[:end][:y] = m[:year2].to_i(10)  if m[:year2]
+            duration[:end][:m] = MONTH_MAP[ m[:month_name2].downcase ]   if m[:month_name2]
+            duration[:end][:d]  = m[:day2].to_i(10)   if m[:day2]
+            duration[:end][:wday] = DAY_MAP[ m[:day_name2].downcase ]   if m[:day_name2]
+
+            duration
+end
+
+
+
+
+def _build_date( m )      self.class._build_date( m ); end
+def _build_date_legs( m ) self.class._build_date_legs( m ); end
+def _build_duration( m )  self.class._build_duration( m ); end
+
+
+
+
+#############
+## "top-level" add a date parser helper
+
+##  note:  parse_date  - returns Date object
+##        _parse_date (with underscore) - return  hash of "parsed" regex match data!!
+
+def self.parse_date( str, start: nil )
+    if m = _parse_date( str )
+       year  = m[:y]
+       yy    = m[:yy]
+
+       ####
+       ## support two digit shortcut for year
+       if yy && year.nil?
+          ###
+          ## for now assume 00,01 to 30 is 2000,2001 to 2030
+          ##   and          31 to 99   is  1931 to 1999
+          year =   yy <= 30 ?  2000+yy : 1900+yy
+       end
+
+       month = m[:m]
+       day   = m[:d]
+       wday  = m[:wday]
+
+
+      if year.nil?     ## try to calculate year
+        raise ArgumentError, "year required in date >#{str}< or pass along start date"   if start.nil?
+
+        year =  if  month > start.month ||
+                   (month == start.month && day >= start.day)
+                  # assume same year as start_at event (e.g. 2013 for 2013/14 season)
+                  start.year
+                else
+                  # assume year+1 as start_at event (e.g. 2014 for 2013/14 season)
+                  start.year+1
+                end
+      end
+      Date.new( year,month,day )
+    else
+      raise ArgumentError, "unexpected date format; cannot parse >#{str}<"
+    end
+end
+
+
+
+def self._parse_date( str )
+    ## note - strip - leading/trailing spaces automatic - why? why not?
+    m = DATE_RE.match( str.strip )
+
+    if m && m.pre_match == '' && m.post_match == ''
+      ## return hash table with captured components
+      date = _build_date( m )
+      date
+    elsif  m
+        ## note - match BUT not anchored to start and end-of-string!!!
+        ##  report, error somehow??
+      nil
+    else
+      nil  ## no match - return nil
+    end
+end
+
+
+end  # class Lexer
+end  # module SportDb