sportdb-parser 0.7.1 → 0.7.2

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

@@ -0,0 +1,189 @@
+module SportDb
+class Lexer
+
+
+def self._build_score( m )
+             ##  note - score is "generic"
+            ##      might be full-time (ft) or
+            ##         after extra-time (aet) or such
+            ##         or even undecided/unknown
+            ##    thus, use score1/score2 and NOT ft1/ft2
+            ##      thus, use (simply an) array e.g. [1,2]
+            ##           and NOT hash (table) e.g. { ft: [1,2] } !!!
+
+            score  = [m[:score1].to_i(10),
+                      m[:score2].to_i(10)]
+
+           score
+end
+
+def self._build_score_awd( m )    # score awarded (awd/awd.)
+            ### note - use "generic" score for now
+            ##         to match  A 3-0 B [awarded] etc.
+            score = [m[:score1].to_i(10),
+                      m[:score2].to_i(10)]
+            ## add score[:awarded] = true ???
+            ##  note - for now uses its own token e.g SCORE_AWD
+            score
+end
+
+def self._build_score_abd( m )      # score abandonded (abd/abd.)
+            ### note - use "generic" score for now
+            score   = [m[:score1].to_i(10),
+                       m[:score2].to_i(10)]
+            ## add score[:abd] = true ???
+            ##  note - for now uses its own token e.g SCORE_ABD
+            score
+end
+
+
+def self._build_score_full( m )
+              score = {}
+              score[:p] = [m[:p1].to_i(10),
+                           m[:p2].to_i(10)]  if m[:p1] && m[:p2]
+              score[:et] = [m[:et1].to_i(10),
+                            m[:et2].to_i(10)]  if m[:et1] && m[:et2]
+              score[:ft] = [m[:ft1].to_i(10),
+                            m[:ft2].to_i(10)]  if m[:ft1] && m[:ft2]
+              score[:ht] = [m[:ht1].to_i(10),
+                            m[:ht2].to_i(10)]  if m[:ht1] && m[:ht2]
+
+              ## add golden/silver flags
+              score[:golden] = true   if m[:aetgg]  ## golden goal (gg)/sudden death (sd)
+              score[:silver] = true   if m[:aetsg]  ## silver goal (sg)
+
+              score
+end
+
+def self._build_score_fuller( m )
+              score = {}
+              score[:p] = [m[:p1].to_i(10),
+                           m[:p2].to_i(10)]  if m[:p1] && m[:p2]
+              score[:et] = [m[:et1].to_i(10),
+                            m[:et2].to_i(10)]  if m[:et1] && m[:et2]
+              score[:ft] = [m[:ft1].to_i(10),
+                            m[:ft2].to_i(10)]  if m[:ft1] && m[:ft2]
+              score[:ht] = [m[:ht1].to_i(10),
+                            m[:ht2].to_i(10)]  if m[:ht1] && m[:ht2]
+              score[:agg] = [m[:agg1].to_i(10),
+                             m[:agg2].to_i(10)]  if m[:agg1] && m[:agg2]
+
+              if m[:away1] && m[:away2]
+                 score[:away] = [m[:away1].to_i(10),
+                                 m[:away2].to_i(10)]
+              elsif m[:away]    ## fallback if no away score; check away flag
+                 score[:away] = true
+              end
+
+              ## add golden/silver flags
+              score[:golden] = true   if m[:aetgg]  ## golden goal (gg)/sudden death (sd)
+              score[:silver] = true   if m[:aetsg]  ## silver goal (sg)
+
+              score
+end
+
+
+def self._build_score_fuller_more( m )
+               ##    SCORE + SCORE_FULLER_MORE
+               ## note -  after extra-time (aet) or full-time (ft)
+               ##           score may be present in SCORE!!!
+              score = {}
+              score[:p] = [m[:p1].to_i(10),
+                           m[:p2].to_i(10)]  if m[:p1] && m[:p2]
+              score[:et] = [m[:et1].to_i(10),
+                            m[:et2].to_i(10)]  if m[:et1] && m[:et2]
+              score[:ft] = [m[:ft1].to_i(10),
+                            m[:ft2].to_i(10)]  if m[:ft1] && m[:ft2]
+              score[:ht] = [m[:ht1].to_i(10),
+                            m[:ht2].to_i(10)]  if m[:ht1] && m[:ht2]
+              score[:agg] = [m[:agg1].to_i(10),
+                             m[:agg2].to_i(10)]  if m[:agg1] && m[:agg2]
+
+              if m[:away1] && m[:away2]
+                 score[:away] = [m[:away1].to_i(10),
+                                 m[:away2].to_i(10)]
+              elsif m[:away]    ## fallback if no away score; check away flag
+                 score[:away] = true
+              end
+
+              ## add golden/silver flags
+              score[:golden] = true   if m[:aetgg]  ## golden goal (gg)/sudden death (sd)
+              score[:silver] = true   if m[:aetsg]  ## silver goal (sg)
+
+              ## add flag in score for et/ft/ht
+              ##    used for "dangling" (generic) score
+              score[:score] = 'et'   if m[:aet] || m[:aetgg] || m[:aetsg]
+              score[:score] = 'ft'   if m[:ft]
+              score[:score] = 'ht'   if m[:ht]
+
+              score
+end
+
+
+def self._build_score_legs( m )
+              legs = {}
+
+              ############
+              ### build leg1 (score)
+              score = {}
+              score[:ft] = [m[:leg1_ft1].to_i(10),
+                            m[:leg1_ft2].to_i(10)]
+              legs['leg1'] = score
+
+              ##################
+              ### build leg2 (score)
+              score = {}
+              score[:ft] = [m[:leg2_ft1].to_i(10),
+                            m[:leg2_ft2].to_i(10)]  if m[:leg2_ft1] && m[:leg2_ft2]
+              score[:et] = [m[:leg2_et1].to_i(10),
+                            m[:leg2_et2].to_i(10)]  if m[:leg2_et1] && m[:leg2_et2]
+              score[:p]  = [m[:leg2_p1].to_i(10),
+                            m[:leg2_p2].to_i(10)]  if m[:leg2_p1] && m[:leg2_p2]
+              legs['leg2'] = score
+
+              ## check for (opt) aggregate - keep on "top-level"
+              legs[:agg] = [m[:agg1].to_i(10),
+                            m[:agg2].to_i(10)]  if m[:agg1] && m[:agg2]
+              legs[:away] = true  if m[:away]
+
+              legs
+end
+
+
+def _build_score( m )             self.class._build_score( m ); end
+def _build_score_awd( m )         self.class._build_score_awd( m ); end
+def _build_score_abd( m )         self.class._build_score_abd( m ); end
+def _build_score_full( m )        self.class._build_score_full( m ); end
+def _build_score_fuller( m )      self.class._build_score_fuller( m ); end
+def _build_score_fuller_more( m ) self.class._build_score_fuller_more( m ); end
+def _build_score_legs( m )        self.class._build_score_legs( m ); end
+
+
+
+
+###
+## add parser helpers
+
+def self._parse_score_full( str )
+    ## note - strip - leading/trailing spaces automatic - why? why not?
+
+    m = Regexp.union(
+              SCORE_FULL_1ST_RE,
+              SCORE_FULL_RE ).match( str.strip )
+
+    if m && m.pre_match == '' && m.post_match == ''
+       pp m
+       _build_score_full( m )
+    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

data/lib/sportdb/parser/token-score.rb

@@ -2,318 +2,10 @@ module SportDb
 class Lexer
 
 
-   
-
-    ## todo/check: use ‹› (unicode chars) to mark optional parts in regex constant name - why? why not?
-
-    #####
-    #  english helpers (penalty, extra time, ...)
-    ##   note - p must go last (shortest match)
-    #     pso = penalty shootout
-    ###  - note - remove PSO for now (may add later back) - why? why not? 
-    #
-    #  todo/fix/clean-up - keep it simple -  remove optional trailing dot (.)
-    #                       from pen., p., agg. etc. - why? why not?
-    #                        always use (simply) pen, p, agg 
-    #                      (also) remove  a.e.t. / a.e.t option - why? why not?
-    #
-    ##  UPDATE mar/2026:  addd pens too - keep - why? why not?
-    ##                     (4-3 pens)
-    ##  (4-3 Pens)  -- keep mixed Pens/Pen. too - why? why not?
-    ##  (4-3 Pen.)
-    P_EN  =  '(?-i: PEN | P |' +
-                   '[Pp]ens | [Pp]en\.? | p\.? )'     # e.g. p., p, pen, pen., etc.
-
-
-    ## fix - change ET_EN  to AET_EN!!! - why? why not?
-    ##   check - allow Aet too - why? why not?
-    ##              or A.e.t ??
-    ET_EN =  '(?-i: AET | ' +
-                   'aet | a\.e\.t\.? )'     # note: make last . optional (e.g a.e.t) allowed too
-    # AET_EN = ET_EN
-
-    ####
-    ## after (golden goal/sudden death) extra time   - add more options/styles - why? why not?
-    AETGG_EN  = '(?-i: AET/GG | AGGET | ASDET | ' +
-                      'aet/gg | a\.e\.t\.?/g\.g\.? | agget | asdet )'
-    ## after (silver goal) extra time
-    AETSG_EN  = '(?-i: AET/SG | ASGET | ' +
-                      'aet/sg | a\.e\.t\.?/s\.g\.? | asget  )'
-    
-    ##  agg/agg.  or AGG
-    AGG_EN = '(?-i: AGG | agg\.? )'   ## aggregate e..g  4-4 agg etc.
-  
-
-    
-    ## regex score helpers
-    ##    note - MUST double escape \d e.g. \\d!!!   if not "simple" string (e.g. '' but %Q<>)
-
-    ##
-    ##  fix - change SCORE_P to SCORE_FULL_P
-    ##               SCORE_ET to SCORE_FULL_ET
-    ##
-    ##   (re)use SCORE_P, SCORE_ET for score only part!!!
-
-    SCORE_P   = %Q<  (?<p1>\\d{1,2}) - (?<p2>\\d{1,2})
-                          [ ]? #{P_EN}
-                  >
-    SCORE_ET  = %Q<  (?<et1>\\d{1,2}) - (?<et2>\\d{1,2})
-                          [ ]? #{ET_EN}
-                  >
-    SCORE_LOOKAHEAD = '(?= [ ,\]] | $)'
-
-    
-    ####
-    ## after extra-time with golden goal/sudden death & silver goal rule
-    ##        note - golden goal & silver goal EXCLUDE penalties!!!
-    ##
-    ##  4-3 a.e.t/g.g.
-    ##  4-3 aet/gg
-    ##  4-3agget   -or-   4-3 asdet
-    ##  2-1 aet/sg
-    ##   -or-
-    ##   4-3 aet/gg (3-3, 2-1)
-    SCORE__ET_GG_SG__RE = %r{
-        (?<score_full>
-           \b
-           (?<et1>\d{1,2}) - (?<et2>\d{1,2})
-                          [ ]? (?:
-                                   (?<aetgg> #{AETGG_EN})
-                                      |
-                                   (?<aetsg> #{AETSG_EN})
-                                )
-           ### note:
-           ## add optional full-time, half-time score
-             (?:
-                 [ ]+
-                 \(
-                    [ ]*
-                   (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})
-                      [ ]*
-                    (?:
-                       , [ ]*
-                       (?: (?<ht1>\d{1,2}) - (?<ht2>\d{1,2})
-                         [ ]*
-                      )?
-                   )? # note: make half time (HT) score optional for now
-                 \)
-             )?                     
-            #{SCORE_LOOKAHEAD}
-    )}ix
-
-
-    ##  note: allow SPECIAL cases WITHOUT full time scores (just a.e.t or pen. + a.e.t.)
-    ##      3-4 pen. 2-2 a.e.t.
-    ##      3-4 pen.   2-2 a.e.t.
-    ##               2-2 a.e.t.
-    SCORE__P_ET__RE = %r{
-        (?<score_full>
-           \b
-            (?: #{SCORE_P} [ ]+ 
-             )?             ## note: make penalty (P) score optional for now
-            #{SCORE_ET}
-            #{SCORE_LOOKAHEAD}
-        )}ix
-                ## todo/check:  remove loakahead assertion here - why require space?
-                ## note: \b works only after non-alphanum e.g. )
-
-
-    ##  note: allow SPECIAL cases WITHOUT full time scores 
-    ##         AND with pen in last position!
-    ##           2-2 a.e.t., 3-4 pen. 
-    ##           2-2 a.e.t.  3-4 pen.  ## or without comma separator - why? why not?   
-    SCORE__ET_P__RE = %r{
-        (?<score_full>
-           \b
-            #{SCORE_ET}  
-               (?: [ ]*,[ ]* | [ ]+ )
-            #{SCORE_P}  
-            #{SCORE_LOOKAHEAD}
-        )}ix
-                ## todo/check:  remove loakahead assertion here - why require space?
-                ## note: \b works only after non-alphanum e.g. )
-
-    ### special case (i)  - full time with penalties
-    ##          2-2, 3-4 pen.
-    SCORE__FT_P__RE = %r{
-        (?<score_full>
-           \b
-            (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})  
-                [ ]*,[ ]*    ## note - comma required!!! 
-            #{SCORE_P}  
-            #{SCORE_LOOKAHEAD}
-        )}ix
-
-    ### special case (ii)  - full time & half-time with penalties
-    ##          2-2 (1-1), 3-4 pen.
-    SCORE__FT_HT_P__RE = %r{
-        (?<score_full>
-           \b
-            (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})
-                [ ]*
-                 \(
-                     (?<ht1>\d{1,2}) - (?<ht2>\d{1,2})
-                 \)
-                [ ]*,[ ]*    ## note - comma required!!! 
-            #{SCORE_P}  
-            #{SCORE_LOOKAHEAD}
-        )}ix
-
-
-
-
-    ##  note: allow SPECIAL with penalty only
-    ##      3-4 pen.  or 3-4p etc.
-    SCORE__P__RE = %r{
-        (?<score_full>
-           \b
-             #{SCORE_P}  
-             #{SCORE_LOOKAHEAD}
-         )}ix
-                ## todo/check:  remove loakahead assertion here - why require space?
-                ## note: \b works only after non-alphanum e.g. )
-
-   ####
-   ## support short all-in-one e.g.
-   ##  e.g.      3-4 pen. 2-2 a.e.t. ( 1-1, 1-1 ) becomes
-   ##   3-4 pen. (2-2, 1-1, 1-1)            
-         
-   SCORE__P_ET_FT_HT_V2__RE = %r{
-          (?<score_full>
-               \b
-                #{SCORE_P} [ ]+       
-                   \(
-                   [ ]*
-               (?<et1>\d{1,2}) - (?<et2>\d{1,2})
-                   [ ]*, [ ]*
-               (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})
-                   [ ]*, [ ]*
-               (?<ht1>\d{1,2}) - (?<ht2>\d{1,2})
-                   [ ]*
-                \)
-               #{SCORE_LOOKAHEAD}
-            )}ix       ## todo/check:  remove loakahead assertion here - why require space?
-                               ## note: \b works only after non-alphanum e.g. )
-
-
-    # e.g. 2-2 a.e.t. (1-1, 1-0), 5-1 pen. 
-    SCORE__ET_FT_HT_P__RE = %r{
-          (?<score_full>
-               \b
-               #{SCORE_ET} [ ]+
-                   \(
-                   [ ]*
-              (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})
-                   [ ]*
-                (?:
-                     , [ ]*
-                    (?: (?<ht1>\d{1,2}) - (?<ht2>\d{1,2})
-                        [ ]*
-                    )?
-                )?              # note: make half time (HT) score optional for now
-              \)
-               (?: [ ]*,[ ]* | [ ]+)
-               #{SCORE_P}
-               #{SCORE_LOOKAHEAD}
-            )}ix       ## todo/check:  remove loakahead assertion here - why require space?
-                               ## note: \b works only after non-alphanum e.g. )
-
-    
-
-    ## e.g. 3-4 pen. 2-2 a.e.t. (1-1, 1-1)  or
-    ##      3-4p 2-2aet (1-1, )     or
-    ##      3-4 pen.  2-2 a.e.t. (1-1)       or
-    ##               2-2 a.e.t. (1-1, 1-1)  or
-    ##               2-2 a.e.t. (1-1, )     or
-    ##               2-2 a.e.t. (1-1)
-
-    SCORE__P_ET_FT_HT__RE = %r{
-          (?<score_full>
-               \b
-               (?:
-                  #{SCORE_P} [ ]+
-                )?            ## note - make penalty (P) score optional for now
-               #{SCORE_ET} [ ]+
-                   \(
-                   [ ]*
-              (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})
-                   [ ]*
-                (?:
-                     , [ ]*
-                    (?: (?<ht1>\d{1,2}) - (?<ht2>\d{1,2})
-                        [ ]*
-                    )?
-                )?              # note: make half time (HT) score optional for now
-              \)
-             #{SCORE_LOOKAHEAD}
-            )}ix       ## todo/check:  remove loakahead assertion here - why require space?
-                               ## note: \b works only after non-alphanum e.g. )
-
-    ###
-    ##   special case for case WITHOUT extra time!!
-    ##     same as above (but WITHOUT extra time and pen required)
-    SCORE__P_FT_HT__RE = %r{
-             (?<score_full>
-                \b
-               #{SCORE_P} [ ]+
-        \(
-        [ ]*
-      (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})
-        [ ]*
-     (?:
-          , [ ]*
-         (?: (?<ht1>\d{1,2}) - (?<ht2>\d{1,2})
-             [ ]*
-         )?
-     )?              # note: make half time (HT) score optional for now
-   \)
-    #{SCORE_LOOKAHEAD}
-    )}ix    ## todo/check:  remove loakahead assertion here - why require space?
-            ## note: \b works only after non-alphanum e.g. )
-
-
-    ##########
-    ## e.g. 2-1 (1-1)
-    SCORE__FT_HT__RE = %r{
-            (?<score_full>
-              \b
-              (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})
-                   [ ]+ \( [ ]*
-                (?<ht1>\d{1,2}) - (?<ht2>\d{1,2})
-                   [ ]* \)
-             #{SCORE_LOOKAHEAD}
-             )}ix    ## todo/check:  remove loakahead assertion here - why require space?
-                    ## note: \b works only after non-alphanum e.g. )
-
-
-
-             
-#############################################
-# map tables
-#  note: order matters; first come-first matched/served
-
-SCORE_FULL_RE = Regexp.union(
-  SCORE__ET_GG_SG__RE,       # e.g. 3-1 aet/gg  
-  SCORE__P_ET_FT_HT_V2__RE,  # e.g. 5-1 pen. (2-2, 1-1, 1-0)  
-  SCORE__ET_FT_HT_P__RE,    # e.g. 2-2 a.e.t. (1-1, 1-0), 5-1 pen. 
-  SCORE__P_ET_FT_HT__RE,    # e.g. 5-1 pen. 2-2 a.e.t. (1-1, 1-0)
-  SCORE__P_FT_HT__RE,     # e.g. 5-1 pen. (1-1)
-  SCORE__ET_P__RE,        # e.g. 2-2 a.e.t., 5-1 pen.
-  SCORE__FT_P__RE,        # e.g. 2-2, 5-1 pen.
-  SCORE__FT_HT_P__RE,     # e.g. 2-2 (1-1), 5-1 pen.
-  SCORE__P_ET__RE,        # e.g.  5-1 pen. 2-2 a.e.t.  or  2-2 a.e.t. (w/o pen)
-  SCORE__P__RE,           # e.g. 5-1 pen.
-  SCORE__FT_HT__RE,        # e.g. 1-1 (1-0)
-  ##  note - keep basic score as its own token!!!!
-  ##   that is, SCORE & SCORE_MORE
-  ### SCORE__FT__RE,           # e.g. 1-1  -- note - must go last!!!
-)
-
-
 ###
 ##
 ##  add support for score awarded (inline style)
-##    3-0 awd  3-0 awd. 3-0awd 
+##    3-0 awd  3-0 awd. 3-0awd
 ##    0-1 awd  or 0-1 AWD etc.
 
 ##
@@ -327,7 +19,7 @@ SCORE_AWD_RE  = %r{
                    (?-i: awd\.? | AWD )
                ## POSITIVE lookahead - requires space
                (?= [ ])
-             )}ix  
+             )}ix
 
 ###
 ##
@@ -341,7 +33,7 @@ SCORE_ABD_RE  = %r{
                   (?-i: abd\.? | ABD )
                ## POSITIVE lookahead - requires space
                (?= [ ])
-             )}ix  
+             )}ix
 
 #####
 ##      2-1
@@ -349,10 +41,10 @@ SCORE_ABD_RE  = %r{
 ###  note - was SCORE__FT__RE
 ###           changed to "generic" SCORE_RE
 ###                and
-##             (?<ft1>\d{1,2}) - (?<ft2>\d{1,2}) 
+##             (?<ft1>\d{1,2}) - (?<ft2>\d{1,2})
 ##      changed
-##             (?<score1>\d{1,2}) - (?<score2>\d{1,2}) 
-##                to 
+##             (?<score1>\d{1,2}) - (?<score2>\d{1,2})
+##                to
 ##             pattern match not necessarily the full-time (ft) scoreline!!!
 ##    - pattern also used for goal seq(uence) e.g. 1-0 Kane, 1-1 Johnson
 SCORE_RE  = %r{
@@ -360,85 +52,9 @@ SCORE_RE  = %r{
               \b
                (?<score1>\d{1,2}) - (?<score2>\d{1,2})
               \b
-             )}ix  
-
-
-######
-#  add support for "split" score
-#    note - for now (2) 1  is REQUIRED
-
-SCORE_TEAM_RE = %r{
-            (?<score_team>
-                 \(
-                    (?<score_i> \d{1,2}) 
-                 \)
-                 [ ]*   ## note - space optional- why? why not?
-                    (?<score_ii> \d{1,2})
-                \b 
-            )
-        }ix   
-
-#     "penalty"-style  (4) is assumed penalty score
-#    note - for now 1 (4) is REQUIRED
-
-SCORE_TEAM_PEN_RE = %r{
-            (?<score_team_pen>
-                 \b
-                    (?<score_i> \d{1,2})
-                 \b
-                   [ ]*  ## note - space optional- why? why not?  
-                 \(
-                    (?<score_pen> \d{1,2}) 
-                 \)
-            )
-        }ix   
-
-########
-## note - score_team_num (<100) e.g. 0, 1, .., 10, 11, .. 99
-##      use a different name - why? why not? 
-##   note - must be surrouned by space
-SCORE_TEAM_NUM_RE = %r{
-    ## positive lookbehind
-     (?<= [ ])
-
-      (?<score_team_num> \d{1,2} )
-
-     ## positive lookahead
-     (?= [ ]|\z)
-}x
-
-
-
-def self._build_score_team( m )
-            score = {}
-            ##  note - score team is "generic"
-            ##      might be full-time (ft) or
-            ##         after extra-time (aet) or such
-            ##         or even undecided/unknown
-            ##    thus, use score_i/score_ii 
-            score[:score] = [m[:score_i].to_i(10),
-                             m[:score_ii].to_i(10)]  
-            score
-end 
-def _build_score_team( m ) self.class._build_score_team( m ); end 
-
-
-def self._build_score_team_pen( m )
-            score = {}
-            score[:score] = [m[:score_i].to_i(10),
-                             m[:score_pen].to_i(10)]  
-            score
-end 
-def _build_score_team_pen( m ) self.class._build_score_team_pen( m ); end 
-
+             )}ix
 
-def self._build_score_team_num( m )
-            score = {}
-            score[:score] = m[:score_team_num].to_i(10)
-            score
-end 
-def _build_score_team_num( m ) self.class._build_score_team_num( m ); end 
 
 
-end  #  class Lexer
-end  # module SportDb
+end # class Lexer
+end # module SportDb