sportdb-parser 0.6.12 → 0.6.14

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 45ae1f806a3734a32c34fd96a32784568443875cd82c9ab1e8ab867bd7358977
-  data.tar.gz: 3bc7ed4c867722b5d950ff759afdf02191405949fedffa4609d81c74daf882be
+  metadata.gz: 51b568bfcb315049faf125718d09615346959cdd7301934a5131cb0fac1b6f9b
+  data.tar.gz: c8bd6486e70d28d4b121a3fb1ebec206b46880571a2404de64418fd76fe8b039
 SHA512:
-  metadata.gz: a6d43efebff7d7558d084fea36d83bc6ec9fdb4aed83aa23d8226f79b10018ddbaf6b607539ac6d664b1038d7772062fcf4cf6f22791444e1c87353889e77d91
-  data.tar.gz: ae2151c60bb90bc4dd61d3fadd3fed6e44b865739ebfb7b28bd8c4a9e3b373cfb1827a2202ebc740ec8a69103d1073c6fe1caa5241a3580495aef1e503d9f55c
+  metadata.gz: 1236c350c6b2cef3a7b9caff9b943ef09d69a02e5e6f2fe3ed55c5677d4a20ed17bba6423c77d035a4ea4c772bdaf598a81ee0738f25185ad2bbf5b63ab4e901
+  data.tar.gz: a16da2520f22b6392330bfbec90e2c8c2ca0f784bb31556ddb0140e25d53d7a293d6ca615b4a2a08d426e0069b0f6dcab8ed06ba35e688a03befa6429d9c732d

data/CHANGELOG.md

@@ -1,4 +1,4 @@
-### 0.6.12
+### 0.6.14
 ### 0.0.1 / 2024-07-12
 
 * Everything is new. First release.

data/config/rounds_es.txt

@@ -17,6 +17,9 @@ Zona B
 Final de Grupos
 Repechaje
 
+Final Absoluta
+
+
 Copa Libertadores
 Copa Sudamericana
 

data/config/zones_en.txt

@@ -3,7 +3,9 @@
 
 
 Western Region
+West Region
 Eastern Region
+East Region
 
 Western Conference
 Eastern Conference  

data/lib/sportdb/parser/lang.rb

@@ -17,12 +17,17 @@ module Lang
 ## Group 1A or A1, B1  - used anywhere
 ##   yes - A1, A2, B1, C1, etc. used in UEFA Nations League for example!!
 ##
+##     exlcude
 ##  use "key" of group - why? why not?
+##
+##  note - will include group stage too
+##           make sure is_round gets called before is_group for now!!!
 
 GROUP_RE = %r{^
                 Group [ ]
-                   (?<key>[a-z0-9]+)
+                   (?<key> [a-z0-9]+ )
               $}ix
+
 def self.is_group?( text )
    ## use regex for match
    GROUP_RE.match?( text )
@@ -39,8 +44,8 @@ ROUND_RE = %r{^
    ##
    ### note - allow Group ("stand-alone") as "generic" round for now
    ##      BUT do NOT allow Group 1, Group 2, Group A, Group B, etc.
-     (?: Group [ ] [A-Z0-9]+ [ ] Play-?offs?  |
-         Group (?: [ ] phase|stage)?  |
+     (?: Group [ ] [a-z0-9]+ [ ] Play-?offs?  |
+         Group (?: [ ] (?: phase|stage))?  |
          League (?: [ ] phase)?
      )
         |
@@ -54,10 +59,13 @@ ROUND_RE = %r{^
            )
            [ ] [1-9][0-9]*
           (?:    ## note - add optional   Matchday 1 of 2 or such
-               [ ] of [1-9][0-9]*
+               [ ] of [ ] [1-9][0-9]*
           )?
       )
        |
+      (?:  Round [ ]  One
+       )
+       |
    ##  starting with qual(ification)
    ##   Qual. Round 1 / Qual. Round 2 / Qual. Round 3
    ##  or
@@ -116,16 +124,28 @@ ROUND_RE = %r{^
         |
    # round32
         (?: Round[ ]of[ ]32 |
-            Last[ ]32 )
+            Last[ ]32 |
+            16th[ ]finals |
+            1/16[ ]finals  )
           |
    # round16
         (?: Round[ ]of[ ]16 |
-            Last[ ]16 )
+            Last[ ]16 |
+            8th[ ]finals |
+            1/8[ ]finals  )
            |
+   # round8 aka quarterfinals
+   #   note - allow quarter-finals/quarter finals/quarterfinals 
+         (?:  Round[ ]of[ ]8 |
+              Last[ ]8  |   
+              1/4[ ]finals  | 
+              Quarter[ -]?finals? |
+              Quarters  )
+          |
    # fifthplace
          (?:
              (?: (Fifth|5th)[ -]place
-                  (?: [ ] (?: match|play[ -]?off|final ))?
+                  (?: [ ] (?: match|final|play[ -]?off ))?
               ) |
              (?: Match[ ]for[ ](?: fifth|5th )[ -]place )
          )
@@ -133,33 +153,21 @@ ROUND_RE = %r{^
    # thirdplace
           (?:
               (?: (Third|3rd)[ -]place
-                     (?: [ ] (?: match|play[ -]?off|final ))?
+                     (?: [ ] (?: match|final|play[ -]?off ))?
                ) |
               (?: Match[ ]for[ ](?: third|3rd )[ -]place )
            )
            |
-   # quarterfinals
-         (?:
-              ## note - allow quarter-finals/quarter finals/quarterfinals
-              Quarter[ -]?finals? |
-              Quarters |
-              Last[ ]8 | 
-              8th[ ]finals |
-              1/8[ ]finals      ## check 1/8 finals is same as querter-finals?
-          )
-          |
-   # semifinals
+   # round4 aka semifinals
         (?:
+             Round[ ]of[ ]4 |
+             Last[ ]4 | 
              Semi[ -]?finals? |
-             Semis |
-             Last[ ]4 |
-             1/4[ ]finals    ## check 1/4 finals is same as semi-finals?
-        )
+             Semis )
         |
-   # final
+   # round2 aka final
          Finals?
-         |
-         |
+        |
     ## add replays
     ##  e.g. Final Replay
     ##       Quarter-finals replays
@@ -271,6 +279,7 @@ LEG_RE = %r{^
  #  leg 1 of 2 / leg 2 of 2   
  #  note - leg limited to ALWAY 1/2 of 2 for now - why? why not?
  #             for more use match 1/2/3 etc. 
+ ##   allow leg of three (e.g. leg 1 of 3) - why? why not?
      (?:  leg [ ] [12]     
           (?: [ ] of [ ] 2)?  )
      |

data/lib/sportdb/parser/lexer.rb

@@ -182,13 +182,16 @@ def tokenize_with_errors
         ## pass 1
         ##   replace all texts with keyword matches
         ##     (e.g. group, round, leg, etc.)
+        ##
+        ##   note - let is_round? get first (before is_group?)
+        ##            will match group stage  as round (NOT group)
         tokens = tokens.map do |t|        
                     if t[0] == :TEXT
                        text = t[1]
-                       t = if is_group?( text )
+                       t =  if is_round?( text ) || is_leg?( text ) || is_zone?( text )
+                               [:ROUND, text]   
+                            elsif is_group?( text )
                                [:GROUP, text]
-                             elsif is_round?( text ) || is_leg?( text ) || is_zone?( text )
-                               [:ROUND, text]
                              else
                                t  ## pass through as-is (1:1)
                              end
@@ -359,6 +362,17 @@ def _tokenize_line( line )
         pos = offsets[1]    ## update pos
     end
 
+    m = ROUND_OUTLINE_RE.match( line )
+    if m
+      puts "   ROUND_OUTLINE"  if debug?
+
+      tokens << [:ROUND_OUTLINE, m[:round_outline]]
+
+      ## note - eats-up line for now (change later to only eat-up marker e.g. »|>>)
+      offsets = [m.begin(0), m.end(0)]
+      pos = offsets[1]    ## update pos
+    end
+
     m = PLAYER_WITH_SCORE_RE.match( line )
     if m
       ##  switch context to GOAL_RE (goalline(s)
@@ -462,6 +476,8 @@ def _tokenize_line( line )
    
             case sym
             when ',' then [:',']
+            when '›' then [:',']  ## note - treat geo sep › (unicode) like comma for now!!!
+            when '>' then [:',']  ## note - treat geo sep > (ascii) like comma for now!!!
             when '[' then
                  ## get out-off geo mode and backtrack (w/ next)
                  puts "  LEAVE GEO_RE MODE, BACK TO TOP_LEVEL/RE"  if debug?