sportdb-parser 0.6.7 → 0.6.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 366948a35eecdf2ea0b1e83e48a4ae53b399f5f91436c066e8b3328fb3d9dda4
-  data.tar.gz: 8fba19e829e8d81873f4a263c5d0b2a1db27095fddebb90233a0d075c2ec7ea4
+  metadata.gz: 6d2e6a1f243cdd9e255d659ffd616e875d3baf978edada8e663eb54e9bff35df
+  data.tar.gz: 76ddc2f54d6ac9e117963e15ae5fa3cadfbaebce8d942021efcd849aaed7811d
 SHA512:
-  metadata.gz: d4cc2627f892ab128df8cbccab77117dd97b85ac9b9900f0434f4ada02fdad1ba7a19cf6055aeb7a88fe1dce477190bfa8b028276837e30546bce0656ba4f901
-  data.tar.gz: 671f21e8bdb895223504e2b8776905d169c7ba97086cf843ebcc6e361a074eaad4c2fa62c6296328e265bae511eb4c4574bbdfd3d76ce92649212547898f726a
+  metadata.gz: 750b13d813a7fddf80b991c4fdb244856b4bd7d07410caa75cacabcd8cc724631b87eb15b3f61de0e84c378475a1dbc42529933997389102378527b623c39f90
+  data.tar.gz: a88f83b33728a761cbb44c134456aeb3b49f003b422a42b30642c43a95514f8a2825929b56d7d54d788f4c197f680e7830d4dffcb7728ab284ea6e23bae64877

data/CHANGELOG.md

@@ -1,4 +1,4 @@
-### 0.6.7
+### 0.6.9
 ### 0.0.1 / 2024-07-12
 
 * Everything is new. First release.

data/config/rounds_en.txt

@@ -17,6 +17,10 @@ Major Semi-Final
 Minor Semi-Final
 
 
+1/8 Finals
+
+
+
 ## keep weirdo matchday ??
 Matchday 0    ## in 2003-04/az.1.txt
 

data/lib/sportdb/parser/lang.rb

@@ -15,6 +15,7 @@ module Lang
 ## Group 1-99
 ## Group HEX  # used in concaf world cup quali
 ## Group 1A or A1, B1  - used anywhere
+##   yes - A1, A2, B1, C1, etc. used in UEFA Nations League for example!!
 ##
 ##  use "key" of group - why? why not?
 
@@ -197,7 +198,17 @@ end
 
 
 def self.is_round?( text )
-    ROUND_RE.match?( text ) || more_round_names.include?( text )
+    ### note - use check for case-insensitive 
+    ##   was:
+    ##       more_round_names.include?( text )
+    ##   change to:
+    ##       more_round_names.any?{ |str| str.casecmp( text )==0 }
+    ##
+    ##  todo/fix:
+    ##    maybe in the future use our own unaccent and downcase - wyh? why not?
+    ##      note - for now ROUND_RE is also case-insensitive!!
+
+    ROUND_RE.match?( text ) || more_round_names.any?{ |str| str.casecmp( text )==0 }
 end
 
 ##

data/lib/sportdb/parser/lexer.rb

@@ -339,11 +339,17 @@ def _tokenize_line( line )
           @re = PROP_CARDS_RE  
           tokens << [:PROP_YELLOWCARDS, m[:key]]
         elsif ['ref', 'referee'].include?( key.downcase )
-          @re = PROP_RE     ## (re)use prop setup for now - why? why not?
+          @re = PROP_REFEREE_RE     
           tokens << [:PROP_REFEREE, m[:key]]
+        elsif ['att', 'attn', 'attendance'].include?( key.downcase )
+          @re = PROP_ATTENDANCE_RE
+          tokens << [:PROP_ATTENDANCE, m[:key]]         
         elsif ['goals'].include?( key.downcase )
           @re = PROP_GOAL_RE
           tokens << [:PROP_GOALS, m[:key]]
+        elsif ['penalties', 'penalty shootout'].include?( key.downcase )
+          @re = PROP_PENALTIES_RE
+          tokens << [:PROP_PENALTIES, m[:key]]
         else   ## assume (team) line-up
           @re = PROP_RE           ## use LINEUP_RE ???
           tokens << [:PROP, m[:key]]
@@ -511,6 +517,97 @@ def _tokenize_line( line )
              puts "!!! TOKENIZE ERROR (PROP_RE) - no match found"
              nil 
          end
+      elsif @re == PROP_ATTENDANCE_RE
+         if m[:space] || m[:spaces]
+              nil    ## skip space(s)
+         elsif m[:enclosed_name]
+              ## reserverd for use for sold out or such (in the future) - why? why not?
+             [:ENCLOSED_NAME, m[:name]]
+         elsif m[:num]
+             [:PROP_NUM, [m[:num], { value: m[:value].to_i(10) } ]]
+=begin             
+         elsif m[:sym]
+            sym = m[:sym]
+            case sym
+            when ',' then [:',']
+            when ';' then [:';']
+            # when '[' then [:'[']
+            # when ']' then [:']']
+            else
+              nil  ## ignore others (e.g. brackets [])
+            end
+=end
+         else
+            ## report error
+            puts "!!! TOKENIZE ERROR (PROP_ATTENDANCE_RE) - no match found"
+            nil 
+         end
+      elsif @re == PROP_REFEREE_RE
+         if m[:space] || m[:spaces]
+              nil    ## skip space(s)
+         elsif m[:prop_key]   ## check for inline prop keys
+              key = m[:key]   
+              ##  supported for now coach/trainer (add manager?)
+              if ['att', 'attn', 'attendance' ].include?( key.downcase )
+                [:ATTENDANCE, m[:key]]   ## use COACH_KEY or such - why? why not?
+              else
+                ## report error - for unknown (inline) prop key in lineup
+                nil
+              end
+         elsif m[:prop_name]    ## note - change prop_name to player
+             [:PROP_NAME, m[:name]]    ### use PLAYER for token - why? why not?
+         elsif m[:num]
+             [:PROP_NUM, [m[:num], { value: m[:value].to_i(10) } ]]
+         elsif m[:enclosed_name]
+              ## use HOLD,SAVE,POST or such keys - why? why not?
+             [:ENCLOSED_NAME, m[:name]]
+         elsif m[:sym]
+            sym = m[:sym]
+            case sym
+            when ',' then [:',']
+            when ';' then [:';']
+ #           when '[' then [:'[']
+ #           when ']' then [:']']
+            else
+              nil  ## ignore others (e.g. brackets [])
+            end
+         else
+            ## report error
+            puts "!!! TOKENIZE ERROR (PROP_REFEREE_RE) - no match found"
+            nil 
+         end       
+      elsif @re == PROP_PENALTIES_RE
+        if m[:space] || m[:spaces]
+              nil    ## skip space(s)
+         elsif m[:prop_name]    ## note - change prop_name to player
+             [:PROP_NAME, m[:name]]    ### use PLAYER for token - why? why not?
+         elsif m[:enclosed_name]
+              ## use HOLD,SAVE,POST or such keys - why? why not?
+             [:ENCLOSED_NAME, m[:name]]
+         elsif m[:score]
+              score = {}
+              ## must always have ft for now e.g. 1-1 or such
+              ###  change to (generic) score from ft -
+              ##     might be score a.e.t. or such - why? why not?
+              score[:ft] = [m[:ft1].to_i(10),
+                            m[:ft2].to_i(10)]  
+              ## note - for debugging keep (pass along) "literal" score
+              [:SCORE, [m[:score], score]]
+         elsif m[:sym]
+            sym = m[:sym]
+            case sym
+            when ',' then [:',']
+            when ';' then [:';']
+            when '[' then [:'[']
+            when ']' then [:']']
+            else
+              nil  ## ignore others (e.g. brackets [])
+            end
+         else
+            ## report error
+            puts "!!! TOKENIZE ERROR (PROP_PENALTIES_RE) - no match found"
+            nil 
+         end
       elsif @re == GOAL_RE || @re == PROP_GOAL_RE
          if m[:space] || m[:spaces]
               nil    ## skip space(s)
@@ -678,7 +775,12 @@ def _tokenize_line( line )
           end
         elsif m[:any]
            ## todo/check log error
-           puts "!!! TOKENIZE ERROR (any) - no match found >#{m[:any]}<"
+           msg = "parse error (tokenize) - skipping any match>#{m[:any]}< @#{offsets[0]},#{offsets[1]} in line >#{line}<"
+           puts "!! WARN - #{msg}"
+
+           errors << msg
+           log( "!! WARN - #{msg}" )
+     
            nil   
         else
           ## report error
@@ -715,7 +817,9 @@ def _tokenize_line( line )
    ##
    ## if in prop mode continue if   last token is [,-]
    ##        otherwise change back to "standard" mode
-   if @re == PROP_RE || @re == PROP_CARDS_RE || @re == PROP_GOAL_RE
+   if @re == PROP_RE            || @re == PROP_CARDS_RE ||
+      @re == PROP_GOAL_RE       || @re == PROP_PENALTIES_RE ||
+      @re == PROP_ATTENDANCE_RE || @re == PROP_REFEREE_RE
      if [:',', :'-', :';'].include?( tokens[-1][0] )
         ## continue/stay in PROP_RE mode
         ##  todo/check - auto-add PROP_CONT token or such