sportdb-parser 0.6.8 → 0.6.10

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b8d65c02b835d52ddb9f7fd61bd44dd6ff4392bd7ddcef751dcc2ec7a0934065
-  data.tar.gz: 92b797acbb7c1652ea50910207f86ae53fa5097bec9acaadeae96e161b005aa2
+  metadata.gz: d0b047ceeb4dc8b4bf446fd47770aa89bd381f35a6375e630dd823e0e8a9f960
+  data.tar.gz: a035121961c238510506d0f58862220cf8f7c767845585f3e0d26b4e28367471
 SHA512:
-  metadata.gz: dee3a2ff615fe8789593db7f627d00a1f3931d18bc150af50e75c4d68b9ad74c87dc359ebfe4268a5f974de7c681b4ddfdb4cd87e66b69b86c4fd6633c1b98a5
-  data.tar.gz: 54762780eb57ba090caa44c4efd53e2fcc8149c87a2aa97b7d052efc58f51da5d4f2f49cfe4ce7b1a45db437a3c9d82856cf7bb887382f75942acda6f495be4c
+  metadata.gz: 19a354aa23e88a776573e3d481f6754c670b7195e40b3f7063fdeee1d8c6155bb2dfe67129e07e8dd665221205e586df0c0bed6735d227659773ff8bfbd126c2
+  data.tar.gz: d89aaedee183b9a8ccc2de58cd7b35e0aa6216463d2873534008bdb02c7e8b8fc9d6cbf555dfd6da67e8c9323d9fb113098ef23ddfa6eb8331b48b74e43477d9

data/CHANGELOG.md

@@ -1,4 +1,4 @@
-### 0.6.8
+### 0.6.10
 ### 0.0.1 / 2024-07-12
 
 * Everything is new. First release.

data/config/rounds_en.txt

@@ -18,6 +18,7 @@ Minor Semi-Final
 
 
 1/8 Finals
+Qualifying Round
 
 
 

data/lib/sportdb/parser/lexer.rb

@@ -339,8 +339,11 @@ 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]]
@@ -410,6 +413,7 @@ def _tokenize_line( line )
   end
 
 
+  old_pos = -1   ## allows to backtrack to old pos (used in geo)
 
   while m = @re.match( line, pos )
     # if debug?
@@ -428,12 +432,14 @@ def _tokenize_line( line )
       log( msg )
     end
 
+
     ##
     ## todo/fix - also check if possible
     ##   if no match but not yet end off string!!!!
     ##    report skipped text run too!!!
 
-    pos = offsets[1]
+    old_pos = pos
+    pos     = offsets[1]
 
 #    pp offsets   if debug?
 
@@ -441,7 +447,46 @@ def _tokenize_line( line )
     ## note: racc requires pairs e.g. [:TOKEN, VAL]
     ##         for VAL use "text" or ["text", { opts }]  array
 
-  t = if @re == PROP_CARDS_RE 
+  t = if @re == GEO_RE
+         ### note - possibly end inline geo on [ (and others?? in the future
+         if m[:space] || m[:spaces]
+            nil    ## skip space(s)
+         elsif m[:text]
+            [:GEO, m[:text]]   ## keep pos - why? why not?
+         elsif m[:timezone]
+            [:TIMEZONE, m[:timezone]]
+         elsif m[:sym]
+            sym = m[:sym]
+            ## return symbols "inline" as is - why? why not?
+            ## (?<sym>[;,@|\[\]-])
+   
+            case sym
+            when ',' then [:',']
+            when '[' then
+                 ## get out-off geo mode and backtrack (w/ next)
+                 puts "  LEAVE GEO_RE MODE, BACK TO TOP_LEVEL/RE"  if debug?
+                 @re = RE
+                 pos = old_pos
+                 next   ## backtrack (resume new loop step)                 
+            else
+              puts "!!! TOKENIZE ERROR (sym) - ignore sym >#{sym}<"
+              nil  ## ignore others (e.g. brackets [])
+            end
+          elsif m[:any]
+             ## todo/check log error
+             msg = "parse error (tokenize geo) - skipping any match>#{m[:any]}< @#{offsets[0]},#{offsets[1]} in line >#{line}<"
+             puts "!! WARN - #{msg}"
+  
+             errors << msg
+             log( "!! WARN - #{msg}" )
+       
+             nil   
+          else
+            ## report error/raise expection
+             puts "!!! TOKENIZE ERROR - no match found"
+             nil 
+          end
+      elsif @re == PROP_CARDS_RE 
         if m[:space] || m[:spaces]
               nil    ## skip space(s)
          elsif m[:prop_name]
@@ -514,6 +559,65 @@ 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)
@@ -639,8 +743,6 @@ def _tokenize_line( line )
             date[:wday] = DAY_MAP[ m[:day_name].downcase ]   if m[:day_name]
             ## note - for debugging keep (pass along) "literal" date
             [:DATE, [m[:date], date]]
-        elsif m[:timezone]
-          [:TIMEZONE, m[:timezone]]
         elsif m[:duration]
             ## todo/check/fix - if end: works for kwargs!!!!!
             duration = { start: {}, end: {}}
@@ -696,10 +798,13 @@ def _tokenize_line( line )
           ## (?<sym>[;,@|\[\]-])
  
           case sym
+          when '@'    ##  enter geo mode
+            puts "  ENTER GEO_RE MODE"  if debug?
+            @re = GEO_RE
+            [:'@']
           when ',' then [:',']
           when ';' then [:';']
           when '/' then [:'/']
-          when '@' then [:'@']
           when '|' then [:'|']
           when '[' then [:'[']
           when ']' then [:']']
@@ -713,7 +818,12 @@ 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
@@ -747,11 +857,17 @@ def _tokenize_line( line )
      @re = RE 
    end
  
+   if @re == GEO_RE   ### ALWAYS switch back to top level mode
+     puts "  LEAVE GEO_RE MODE, BACK TO TOP_LEVEL/RE"  if debug?
+     @re = RE 
+   end
+
    ##
    ## 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  || @re == PROP_PENALTIES_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