fbtxt-parser 0.9.0

data/lib/fbtxt/parser/token-score_fuller.rb

@@ -0,0 +1,434 @@
+module Fbtxt
+class Lexer
+
+###
+# alternate (more literate) score style
+#      found in books & magazines
+#         (with no half-time score)
+#
+#         4-4 (aet, 6-6 on aggregate, win 3-5 on pens)
+#            -or-
+#         4-4 (aet, 6-6 an aggregate, 3-5 on pens)
+#            -or-
+#         4-4 (aet, 6-6agg, 3-5p)
+#
+#         2-2 (3-3 on aggregate, win on away goals)
+#            -or-
+#         2-2 (3-3 agg, away goals)
+#          etc.
+
+
+## regex score helpers
+##    note - MUST double escape \d e.g. \\d!!!   if not "simple" string (e.g. '' but %Q<>)
+
+def self._mk_score_fuller_agg( win: )    ## with optional win - true|false
+   %Q<
+                 (?:
+                    ############
+                    ## opt 1)  with win
+                    (?:
+                       #{ win ? '(?: win [ ] )?' : '' }
+                        (?<agg1>\\d{1,2}) - (?<agg2>\\d{1,2})
+                          [ ] on [ ] agg (?: regate )?
+                    )
+                    |
+                    #####
+                    ## opt 2)  "classic" (post)
+                    (?:
+                       (?<agg1>\\d{1,2}) - (?<agg2>\\d{1,2})
+                          [ ]*
+                        #{AGG_EN}
+                    )
+                    |
+                    #####
+                    ## opt 3) agg up-front (pre)
+                    (?:
+                         agg [ ]
+                       (?<agg1>\\d{1,2}) - (?<agg2>\\d{1,2})
+                    )
+                 )
+    >
+end
+
+def self._mk_score_fuller_p( win: )    ## with optional win - true|false
+   %Q<
+                 (?:
+                    ############
+                    ## opt 1)  with win
+                    (?:
+                        #{ win ? '(?: win [ ] )?' : '' }
+                        (?<p1>\\d{1,2}) - (?<p2>\\d{1,2})
+                          [ ] on [ ] pens
+                    )
+                    |
+                    #####
+                    ## opt 2)  "classic" (post)
+                    (?:
+                       (?<p1>\\d{1,2}) - (?<p2>\\d{1,2})
+                          [ ]*
+                        #{P_EN}
+                    )
+                    |
+                    #####
+                    ## opt 3) up-front (pre)
+                    (?:
+                         (?: pen|p) [ ]
+                       (?<p1>\\d{1,2}) - (?<p2>\\d{1,2})
+                    )
+                 )
+    >
+end
+
+
+SCORE_FULLER_AGG     =  _mk_score_fuller_agg( win: false )
+SCORE_FULLER_AGG_WIN =  _mk_score_fuller_agg( win: true )
+
+SCORE_FULLER_P     =  _mk_score_fuller_p( win: false )
+SCORE_FULLER_P_WIN =  _mk_score_fuller_p( win: true )
+
+SCORE_FULLER_AWAY_WIN  = %Q<
+                 (?:
+                  (?<away>
+                    ############
+                    ## opt 1)  with win
+                    (?:
+                        (?: win [ ] )?
+                        (?: (?<away1>\\d{1,2}) - (?<away2>\\d{1,2}) [ ] )?
+                         on [ ] away [ ] goals?     # goal or goals
+                    )
+                    |
+                    #####
+                    ## opt 2)  "classic" (post)
+                    (?:
+                       (?: (?<away1>\\d{1,2}) - (?<away2>\\d{1,2}) [ ] )?
+                          [ ]* away
+                    )
+                    |
+                    #####
+                    ## opt 3) up-front (pre)
+                    (?:
+                          away
+                       (?:  [ ]
+                            (?<away1>\\d{1,2}) - (?<away2>\\d{1,2})
+                       )?
+                    )
+                 ))
+            >
+
+
+
+SCORE_FULLER_HT_OPT   =   %Q<
+                              (?:   HT [ ]
+                                  (?: (?<ht1>\\d{1,2}) - (?<ht2>\\d{1,2}))
+                                  [ ]*,[ ]*
+                              )?  ## note - make optional
+                            >
+
+SCORE_FULLER_FT_OPT =  %Q<
+                              (?:   FT [ ]
+                                  (?: (?<ft1>\\d{1,2}) - (?<ft2>\\d{1,2}))
+                                  [ ]*,[ ]*
+                              )?  ## note - make optional
+                            >
+
+
+#############
+#     4-4 (HT 2-1)
+#           or
+#    Team A  4-1  Team B  (HT 2-1)
+
+SCORE_FULLER__HT = %Q<
+             \\(  HT [ ]
+                  (?<ht1>\\d{1,2}) - (?<ht2>\\d{1,2})
+             \\)
+>
+
+SCORE_FULLER__HT_FT__RE  =  %r{
+        (?<score_fuller>
+           \b
+            (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})
+            [ ]+
+             #{SCORE_FULLER__HT}
+        )}ix
+
+SCORE_FULLER_MORE__HT_FT__RE = %r{
+        (?<score_fuller_more>
+             #{SCORE_FULLER__HT}
+        )}ix
+
+
+
+
+############
+#        4-4 (aet)
+#        4-4 (a.e.t.)
+#        or
+#   add golden goal/sudden death and silver goal e.g.
+#        5-4 (aet/gg)            - note: adds golden (goal) flag
+#        5-4 (a.e.t./g.g.)       - note: adds silver (goal) flag
+#
+#     Team A  4-4  Team B  (aet)
+#     Team A  4-4  Team B  (a.e.t.)
+#
+#        or
+#        4-4 (FT 3-2, AET)
+#        4-4 (HT 2-1, FT 3-2, AET)
+
+
+SCORE_FULLER__ET = %Q<
+             \\(
+                #{SCORE_FULLER_HT_OPT}
+                #{SCORE_FULLER_FT_OPT}
+                (?:
+                  (?<aetgg> #{AETGG_EN})
+                   |
+                  (?<aetsg> #{AETSG_EN})
+                   |
+                  (?<aet> #{ET_EN})
+                 )
+             \\)
+>
+
+SCORE_FULLER__ET__RE  =  %r{
+        (?<score_fuller>
+           \b
+            (?<et1>\d{1,2}) - (?<et2>\d{1,2})
+            [ ]+
+             #{SCORE_FULLER__ET}
+        )}ix
+
+SCORE_FULLER_MORE__ET__RE = %r{
+        (?<score_fuller_more>
+             #{SCORE_FULLER__ET}
+        )}ix
+
+
+#############
+#         4-4 (aet, win 3-5 on pens)
+#         4-4 (aet, 3-5 on pens)
+#         4-4 (aet, 3-5 pen)
+#         4-4 (a.e.t., 3-5 pen.)
+#            or
+#         Team A  4-4  Team B  (aet, win 3-5 on pens)
+#         Team A  4-4  Team B  (aet, 3-5 on pens)
+#         Team A  4-4  Team B  (aet, 3-5 pen)
+#         Team A  4-4  Team B  (a.e.t., 3-5 pen.)
+
+SCORE_FULLER__ET_P = %Q<
+             \\(
+                #{SCORE_FULLER_HT_OPT}
+                #{SCORE_FULLER_FT_OPT}
+                (?<aet> #{ET_EN})
+                 [ ]*,[ ]*
+                 #{SCORE_FULLER_P_WIN}
+             \\)
+>
+
+SCORE_FULLER__ET_P__RE  =  %r{
+        (?<score_fuller>
+           \b
+            (?<et1>\d{1,2}) - (?<et2>\d{1,2})
+            [ ]+
+             #{SCORE_FULLER__ET_P}
+        )}ix
+
+SCORE_FULLER_MORE__ET_P__RE = %r{
+        (?<score_fuller_more>
+             #{SCORE_FULLER__ET_P}
+        )}ix
+
+
+##########################
+#         4-4 (win 3-5 on pens)
+#         4-4 (3-5 pen)
+#         4-4 (3-5p)
+#           or
+#       Team A  4-4  Team B (win 3-5 on pens)
+#       Team A  4-4  Team B (3-5 pen)
+#       Team A  4-4  Team B (3-5p)
+
+SCORE_FULLER__FT_P  =  %Q<
+             \\(
+                  #{SCORE_FULLER_HT_OPT}
+                  #{SCORE_FULLER_P_WIN}
+             \\)
+>
+
+SCORE_FULLER__FT_P__RE  =  %r{
+        (?<score_fuller>
+           \b
+            (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})
+            [ ]+
+             \(
+                 #{SCORE_FULLER_P_WIN}
+             \)
+        )}ix
+
+SCORE_FULLER_MORE__FT_P__RE = %r{
+        (?<score_fuller_more>
+             #{SCORE_FULLER__FT_P}
+        )}ix
+
+
+#####################
+#   3-2 (win 4-5 on aggregate)
+#   3-2 (4-5 on aggregate)
+#   3-2 (4-5 on agg)
+#   3-2 (4-5 agg)
+#   3-2 (4-5 agg.)
+#     or
+#   3-2 (agg 4-5)
+
+SCORE_FULLER__FT_AGG  =  %Q<
+             \\(
+                 #{SCORE_FULLER_HT_OPT}
+                 #{SCORE_FULLER_AGG_WIN}
+             \\)
+>
+
+SCORE_FULLER__FT_AGG__RE  =  %r{
+        (?<score_fuller>
+           \b
+            (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})
+            [ ]+
+             #{SCORE_FULLER__FT_AGG}
+        )}ix
+
+SCORE_FULLER_MORE__FT_AGG__RE = %r{
+        (?<score_fuller_more>
+             #{SCORE_FULLER__FT_AGG}
+        )}ix
+
+####
+#  ft + agg + away
+#     2-1 (3-3 on aggregate, win on away goals)
+#     2-1 (3-3 on aggregate, win 2-1 on away goals)
+
+SCORE_FULLER__FT_AGG_AWAY = %Q<
+             \\(
+                #{SCORE_FULLER_HT_OPT}
+                #{SCORE_FULLER_AGG}
+                   [ ]*,[ ]*
+                 #{SCORE_FULLER_AWAY_WIN}
+             \\)
+>
+
+SCORE_FULLER__FT_AGG_AWAY__RE  =  %r{
+        (?<score_fuller>
+           \b
+            (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})
+            [ ]+
+             #{SCORE_FULLER__FT_AGG_AWAY}
+        )}ix
+
+SCORE_FULLER_MORE__FT_AGG_AWAY__RE = %r{
+        (?<score_fuller_more>
+             #{SCORE_FULLER__FT_AGG_AWAY}
+        )}ix
+
+
+#####################
+#   2-1 (aet, 3-3 on aggregate, win 5-2 on pens)
+#   2-1 (aet, 3-3 agg, 5-2 pen.)
+
+SCORE_FULLER__ET_AGG_P  =  %Q<
+             \\(
+                #{SCORE_FULLER_HT_OPT}
+                #{SCORE_FULLER_FT_OPT}
+                (?<aet> #{ET_EN})
+                    [ ]*,[ ]*
+                    #{SCORE_FULLER_AGG}
+                    [ ]*,[ ]*
+                    #{SCORE_FULLER_P_WIN}
+             \\)
+>
+
+SCORE_FULLER__ET_AGG_P__RE  =  %r{
+        (?<score_fuller>
+           \b
+            (?<et1>\d{1,2}) - (?<et2>\d{1,2})
+            [ ]+
+             #{SCORE_FULLER__ET_AGG_P}
+        )}ix
+
+SCORE_FULLER_MORE__ET_AGG_P__RE = %r{
+        (?<score_fuller_more>
+             #{SCORE_FULLER__ET_AGG_P}
+        )}ix
+
+
+#############################################
+# map tables
+#  note: order matters - first come-first matched/served
+
+SCORE_FULLER_RE = Regexp.union(
+  SCORE_FULLER__HT_FT__RE,       ## e.g.  3-2 (HT 2-1)
+  SCORE_FULLER__ET_P__RE,        ## e.g.  2-2 (aet, win 5-3 on pens)
+  SCORE_FULLER__ET__RE,          ## e.g.  2-3 (aet)
+  SCORE_FULLER__FT_P__RE,        ## e.g.  2-2 (win 5-3 on pens)
+  SCORE_FULLER__FT_AGG__RE,      ## e.g.  2-3 (win 5-4 on aggregate)
+  SCORE_FULLER__FT_AGG_AWAY__RE, ## e.g.  2-1 (3-3 on aggreate, win 2-1 on away goals)
+  SCORE_FULLER__ET_AGG_P__RE,    ## e.g.  2-1 (aet, 3-3 on aggregate, win 5-2 on pens)
+  )
+
+
+
+##
+##  add support for "stand-alone"  (HT) and (FT) - keep why? why not?
+SCORE_FULLER_MORE__HT__RE = %r{
+        (?<score_fuller_more>
+            \( (?<ht> ht ) \)
+        )}ix
+
+SCORE_FULLER_MORE__FT__RE = %r{
+        (?<score_fuller_more>
+             \( (?<ft> ft ) \)
+        )}ix
+
+
+###
+#  add special for fuller_more
+#     (aet 4-3)   -  core score is ft, and fuller more incl. et!!!
+
+SCORE_FULLER_MORE__FT_ET__RE = %r{
+        (?<score_fuller_more>
+              \(#{ET_EN}
+                   [ ]
+               (?<et1>\d{1,2}) - (?<et2>\d{1,2})
+              \)
+        )}ix
+
+
+
+##
+#  note - simply   (1-1)  !!!!!
+##   note - special attention needed for placemenent in processing error!!!
+##     make sure it is the last (or on of the last) match(es)
+
+SCORE_FULLER_MORE__HT_FT__CLASSIC_RE = %r{
+        (?<score_fuller_more>
+             \(
+                  (?<ht1>\d{1,2}) - (?<ht2>\d{1,2})
+             \)
+        )}ix
+
+
+SCORE_FULLER_MORE_RE = Regexp.union(
+  SCORE_FULLER_MORE__FT__RE,          ## e.g. (ft)
+  SCORE_FULLER_MORE__HT__RE,          ## e.g. (ht)
+  SCORE_FULLER_MORE__HT_FT__RE,       ## e.g. (HT 2-1)
+  SCORE_FULLER_MORE__ET_P__RE,        ## e.g. (aet, win 5-3 on pens)
+  SCORE_FULLER_MORE__ET__RE,          ## e.g. (aet)
+  SCORE_FULLER_MORE__FT_ET__RE,       ## e.g. (aet 3-2) - (SPECIAL) incl. after extra-time score!!
+  SCORE_FULLER_MORE__FT_P__RE,        ## e.g. (win 5-3 on pens)
+  SCORE_FULLER_MORE__FT_AGG__RE,      ## e.g. (win 5-4 on aggregate)
+  SCORE_FULLER_MORE__FT_AGG_AWAY__RE, ## e.g. (3-3 on aggreate, win 2-1 on away goals)
+  SCORE_FULLER_MORE__ET_AGG_P__RE,    ## e.g. (aet, 3-3 on aggregate, win 5-2 on pens)
+
+  SCORE_FULLER_MORE__HT_FT__CLASSIC_RE,   ## e.g. (2-1)  half-time !!!!
+)
+
+
+
+end  #  class Lexer
+end  # module Fbtxt

data/lib/fbtxt/parser/token-score_legs.rb

@@ -0,0 +1,59 @@
+module Fbtxt
+class Lexer
+
+
+##
+##  note - for now only two legs (1st,2nd) supported
+##             maybe more in the future (if there is a real-world sample/use)
+
+##
+##  win on away goals
+##  aet
+##
+
+SCORE_LEGS_RE  =  %r{
+        (?<score_legs>
+           \b
+            (?<leg1_ft1>\d{1,2}) - (?<leg1_ft2>\d{1,2})
+               (?: [ ]+ |  [ ]*,[ ]*)   # separate by spaces OR comma
+            (?:
+                ## opt 1 - after extra-time (et) score
+                    (?<leg2_et1>\d{1,2}) - (?<leg2_et2>\d{1,2})
+                       [ ]? #{ET_EN}   ## a.e.t./aet
+                        ### note - might end in dot (.) not alpha
+                        ###  thus, wordboundary NOT working
+                       #{SCORE_LOOKAHEAD}
+                  |
+                ## opt 2 - full-time (ft)
+                (?<leg2_ft1>\d{1,2}) - (?<leg2_ft2>\d{1,2})
+                    \b
+            )
+            (?:   ## check optional aggregate e.g. (agg 4-4)
+                [ ]+
+                 \(
+                     agg [ ]
+                      (?<agg1>\d{1,2}) - (?<agg2>\d{1,2})
+
+                     ### add win options
+                     (?:
+                         ## opt 1 - on away goals
+                        (?<away> [ ]*,[ ]*
+                                 (?:win [ ])? on [ ] away [ ] goals?
+                         )
+                           |
+                         ## opt 2 - on penalties
+                        (?:
+                           [ ]*,[ ]*
+                           (?:win [ ])?
+                            (?<leg2_p1>\d{1,2}) - (?<leg2_p2>\d{1,2})
+                            [ ] on [ ] pens
+                        )
+                     )?
+                 \)
+            )?
+        )}ix
+
+
+
+end  #  class Lexer
+end  # module Fbtxt

data/lib/fbtxt/parser/token-status.rb

@@ -0,0 +1,192 @@
+module Fbtxt
+class Lexer
+
+##  (match) status
+##    note: english usage - cancelled (in UK), canceled (in US)
+##
+##  add more variants - why? why not?
+
+
+POSTPONED = %Q{ (?<postponed> postponed  | pst\\.? | po?stp\\.?  | ppd\\.? ) }
+CANCELED  = %Q{ (?<canceled>  cancell?ed | canc\\.? ) }    ##  add can/can. - why? why not?
+WALKOVER  = %Q{ (?<walkover>  walkover   | w/o  | wo ) }   ## add o/w too - why? why not?
+AWARDED   = %Q{ (?<awarded>   awarded    | awd\\.? ) }
+SUSPENDED = %Q{ (?<suspended> suspended  | susp\\.? ) }
+ABANDONED = %Q{ (?<abandoned> abandoned  | aban\\.?  | abd\\.? ) }
+ANNULLED  = %Q{ (?<annulled>  annulled ) }
+VOIDED    = %Q{ (?<voided>    voided     | void ) }  ### note - alternative (name) to annulled
+
+REPLAY    = %Q{ (?<replay>    replay     | repl\\.? ) }
+
+
+##
+##  note - status_note incl. complete  text incl. <status> (not normalized)
+##            <status> gets normalized e.g. ppt => postponed etc.
+
+STATUS_RE = %r{
+            \[
+      (?:
+#############################################
+### opt 1 - allow long forms with note/comment for some stati
+##                    e.g. [postponed due to tropical storm "Hanna"]
+##                         [suspended at 84' by storm; result stood]
+#########################
+           (?: (?<status_note>
+                  (?<status>
+               ####################
+               ## pre-match (not played)
+                    #{POSTPONED}
+                           |
+                    #{CANCELED}
+                           |
+                    #{WALKOVER}
+                           |
+               ######################
+               ## pre/post match
+                     #{AWARDED}
+                            |
+               ########################
+               ## post match - (partially) played
+                    #{SUSPENDED}
+                            |
+                    #{ABANDONED}
+                            |
+                    #{ANNULLED}
+                            |
+                    #{VOIDED} ### note - alternative to annulled
+              )     ## end-of-<status>
+                  [ :;,-]+     ## leading spaces (or separators)
+                  [^\]]+?      ## note - add non-greedy match
+              ) ## end-of-<status-note>
+              [ ]*  ## eat-up optional trailing spaces
+            )
+            |
+########################################
+## opt 2 - short form only (no note/comments) e.g. [postponed], [Canceled], etc.
+####################################
+            (?<status>
+         ####################
+         ## pre-match (not played)
+               #{POSTPONED}
+                 |
+               #{CANCELED}
+                 |
+               #{WALKOVER}
+                 |
+         ######################
+         ## pre/post match
+               #{AWARDED}
+                 |
+         ########################
+         ## post match - (partially) played
+               #{SUSPENDED}
+                 |
+               #{ABANDONED}
+                 |
+               #{ANNULLED}
+                 |
+               #{VOIDED}   ### note - alternative to annulled
+                 |
+               #{REPLAY}       ### todo/fix - keep replay - why? why not?
+                                  ###   prefer replay in round e.g.
+                                  ##       ▪ Round 17, Replay
+                                  ##       ▪ Semi-finals, Replays
+            )
+      )
+    \]
+}ix
+
+
+def self._build_status( m )
+        status = {}
+        ## note - norm status text - why? why not?
+        status[:status] = if    m[:postponed] then 'postponed'
+                          elsif m[:canceled]  then 'canceled'
+                          elsif m[:walkover]  then 'walkover'
+                          elsif m[:awarded]   then 'awarded'
+                          elsif m[:suspended] then 'suspended'
+                          elsif m[:abandoned] then 'abandoned'
+                          elsif m[:annulled] ||
+                                m[:voided]    then 'annulled'
+                          elsif m[:replay]    then 'replay'
+                          else  ## fallback on "generic" status (shouldn't happen)
+                            m[:status]
+                          end
+
+        ## includes note? e.g.  awarded; originally 2-0
+        status[:status_note] = m[:status_note]   if m[:status_note]
+
+        status
+end
+def _build_status( m ) self.class._build_status( m ); end
+
+
+end  #  class Lexer
+end  # module Fbtxt
+
+
+__END__
+
+
+
+
+####################
+## pre-match (not played)
+postponed|postp\.|ppd\.
+             ## e.g. [postponed due to problems with the screen of the stadium]
+             ##      [postponed by storm]
+             ##      [postponed due to tropical storm "Hanna"]
+             ##      [postponed from Sep 10-12 due to death Queen Elizabeth II]
+
+cancell?ed|canc.\
+
+walkover|w/o|wo
+## A victory awarded to one team because the opponent was unable
+##  or unwilling to compete (e.g., failing to show up or being disqualified).
+##  -or-
+##  A walkover or "win over" reveals when a team has won a game
+##    without it being played.
+##  -or-
+##  see <https://en.wikipedia.org/wiki/Walkover>
+
+
+
+######################
+## pre/post match
+awarded|awd\.
+
+            ## e.g. [awarded match to Leones Negros by undue alignment; original result 1-2]
+             ##     [awarded 3-0 to Cafetaleros by undue alignment; originally ended 2-0]
+             ##     [awarded 3-0; originally 0-2, América used ineligible player (Federico Viñas)]
+
+## A result that is decided by a governing body
+##  (like FIFA or a domestic league) rather than by the play on the pitch.
+## Usually follows a Forfeit or Walkover.
+## If a team refuses to play, abandons a match, or fields an ineligible player,
+##  the opponent is typically awarded a 3-0 victory.
+
+########################
+## post match - (partially) played
+suspended|susp\.
+
+            ## e.g. [suspended at 0-0 in 12' due to storm]
+             ##      [suspended at 84' by storm; result stood]
+
+## The match is temporarily halted but intended to be resumed or restarted later.
+
+abandoned|aban.\|abd\.
+
+             ## e.g. [abandoned at 1-1 in 65' due to cardiac arrest
+             ##          Luton player Tom Lockyer]
+             ##      [abandoned at 0-0 in 6' due to waterlogged pitch]
+             ##     [abandoned at 5-0 in 80' due to attack
+             ##          on assistant referee by Cerro; result stood]
+             ##    [abandoned at 1-0 in 31']
+             ##    [abandoned at 0-1' in 85 due to crowd trouble]
+
+## The match started but was stopped by the referee before the final whistle
+##  (e.g., due to a waterlogged pitch or player injury) and did not resume
+
+annulled  OR  voided|void
+## The match result is struck from the record entirely,
+##  usually due to a team's withdrawal from the league or a severe rule violation.