fifadat-convert 0.0.1 → 0.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 98a8e2d8a9d493086e5b0758df77f1c372437961fadb6fdbf30a715b50316af7
-  data.tar.gz: 1efd005037d164e2ffdf71f677616d0427a558ee076654128d8e6d2047f27498
+  metadata.gz: 72f197ccce455bdc5de862831d320aabaa06f5ed2665e8496ec324573135398c
+  data.tar.gz: 6e2d54b61c281bcbc8d812e8d7a7bc562811a4deb6663d6e3cf6a73c68cc0b60
 SHA512:
-  metadata.gz: 1d01a876f48d6a96ddecc640d90daeb1b507ed88e6f513510bfdbba5a0154c05c3502fe5354b6129908e6f3547853897ab09661683c1d7c99d3440c42dec6d26
-  data.tar.gz: f98f022b43b2b0483fe66446a95fa64f9306a9d2be623759e8a2b93a0664c23f584e3900b340e0b4bb54bbeca7b42d0edb03762ce91903c25763a2906b9b7ced
+  metadata.gz: 1c5aca3857ba0409c4070845fcb03dd812cbabb011689108437776ffe24e5d3934dc9145b55fb5895db47f26b17f2f6189f67ef4828a41a9fde799e01c2d0a2e
+  data.tar.gz: 340cbd14fd0b74db8af3144426b6ca857b1acf2d79911495e33e82589dd28c48e6ba9c659bffae4799c93df0debd271d70dd0e9dd9bd97e97af6bfd5ad144c9b

data/CHANGELOG.md

@@ -1,4 +1,5 @@
 
+### 0.1.0
 ### 0.0.1 / 2026-09-24
 
 * Everything is new. First release.

data/Manifest.txt

@@ -4,24 +4,28 @@ README.md
 Rakefile
 bin/fifadat
 lib/fifadat/convert.rb
-lib/fifadat/convert/build_match.rb
-lib/fifadat/convert/build_report.rb
 lib/fifadat/convert/convert-reports.rb
 lib/fifadat/convert/convert.rb
 lib/fifadat/convert/errata.rb
-lib/fifadat/convert/goals-calc_score.rb
-lib/fifadat/convert/goals.rb
 lib/fifadat/convert/helper-date.rb
 lib/fifadat/convert/helper-minute.rb
 lib/fifadat/convert/helper-score.rb
 lib/fifadat/convert/norm.rb
-lib/fifadat/convert/officials.rb
-lib/fifadat/convert/penalties.rb
-lib/fifadat/convert/players.rb
-lib/fifadat/convert/stadiums.rb
-lib/fifadat/convert/stages.rb
-lib/fifadat/convert/substitutions.rb
-lib/fifadat/convert/teams.rb
+lib/fifadat/convert/reader.rb
 lib/fifadat/convert/tool.rb
 lib/fifadat/convert/version.rb
+lib/fifadat/models/goal.rb
+lib/fifadat/models/match.rb
+lib/fifadat/models/match_report.rb
+lib/fifadat/models/official.rb
+lib/fifadat/models/penalty.rb
+lib/fifadat/models/player.rb
+lib/fifadat/models/stadium.rb
+lib/fifadat/models/stage.rb
+lib/fifadat/models/substitution.rb
+lib/fifadat/models/team.rb
+lib/fifadat/tables/players.rb
+lib/fifadat/tables/stadiums.rb
+lib/fifadat/tables/stages.rb
+lib/fifadat/tables/teams.rb
 lib/fifadata-convert.rb

data/lib/fifadat/convert/convert-reports.rb

@@ -3,84 +3,54 @@
 ##
 ##   read & build match report
 
-## use mk/build_report_basename - why? why not?
-##     fix-fix-fix - change to _match_basename !!!
-##        use by (match)report and (match)timeline !!!
-def _report_basename( m )
-      ### get match (live) details
-      team1_code = m['Home']['Abbreviation']
-      team2_code = m['Away']['Abbreviation']
-      localDateTime  = parse_date_utc( m['LocalDate'] )
-
-      ##  e.g.  2020-06-24_RAP-AUS__54545454
-      basename = String.new
-      basename += localDateTime.strftime('%Y-%m-%d')    ##   i) date
-      basename += "_#{team1_code}-#{team2_code}"        ##  ii) team1 v team2
-
-      basename
-end
 
+def convert_reports( slug:, season:,
+                         indir: '.',
+                         outdir: './tmp' )
 
+   season = Season(season)
 
-def _read_timeline( m, timeline_dir: )
-      return nil     if m['Home'].nil? || m['Away'].nil?
+   data =  read_json_v2( "#{indir}/#{slug}/#{season.to_path}_matches.json" )
+   matches = data['Results']  ## only use results (match) array
+   puts "==> convert reports - #{matches.size} match(es) in #{slug} #{season}"
 
-      basename = _report_basename( m )
-      basename += "__#{m['IdMatch']}"
 
-      path = "#{timeline_dir}/#{basename}.json"
+   ## read in stages
+   ##   incl.  SequenceOrder, StageLevel (optional)
+   stages = Stages.read( "#{indir}/#{slug}/misc/#{season.to_path}_stages.json" )
+   stages.add_matches( matches )
 
-      ## note - skip if no match report available!!!
-      if File.file?( path )
-         timeline = read_json_v2( path )
-         timeline
-      else
-         nil
-      end
-end
+   teams = Teams.new
+   teams.add_matches( matches )
 
-def _read_report( m, report_dir: )
-      return nil     if m['Home'].nil? || m['Away'].nil?
+   stadiums = Stadiums.new
+   stadiums.add_matches( matches )
 
-      ##  e.g.  2020-06-24_RAP-AUS__54545454
-      #   note - add match id for "raw" match reports!!!
-      basename = _report_basename( m )
-      basename += "__#{m['IdMatch']}"
-
-      ## #{indir}/#{slug}/matches/#{season.to_path}
-      path = "#{report_dir}/#{basename}.json"
-
-      ## note - skip if no match report available!!!
-      if File.file?( path )
-         live = read_json_v2( path )
-         live
-      else
-         nil
-      end
-end
 
 
 
+   report_dir = "#{indir}/#{slug}/matches/#{season.to_path}"
+   timeline_dir = "#{indir}/#{slug}/timelines/#{season.to_path}"
 
 
 
-def convert_reports( slug:, season:,
-                         indir: '.',
-                         outdir: './tmp' )
-
-   season = Season(season)
+   matches.each_with_index do |m, i|
 
-   data =  read_json_v2( "#{indir}/#{slug}/#{season.to_path}_matches.json" )
-   matches = data['Results']  ## only use results (match) array
-   puts "  #{matches.size} match(es) in season #{season}"
+       print "."
+       print i+1    if (i+1) %10 == 0
 
 
-   report_dir = "#{indir}/#{slug}/matches/#{season.to_path}"
-   timeline_dir = "#{indir}/#{slug}/timelines/#{season.to_path}"
+       ## note skip if SCHEDULED == 1
+       ##   0 =>   FINISHED/complete (OK)
+       ##   1 =>   SCHEDULED/not yet played
+       ##   2 =>   LIVE
+       ##   7 =>   POSTPONED
 
+       next if m['MatchStatus'] == 1 ||
+               m['MatchStatus'] == 2 ||
+               m['MatchStatus'] == 7
 
 
-   matches.each_with_index do |m, i|
       live     = _read_report( m, report_dir: report_dir )
 
       ## no report found; continue
@@ -89,26 +59,38 @@ def convert_reports( slug:, season:,
       ## check for optional timeline
       timeline = _read_timeline( m, timeline_dir: timeline_dir )
 
-      rec = {
+
+      rec    = Match.build( live,   teams:    teams,
+                                    stadiums: stadiums,
+                                    stages:   stages )
+
+
+       ## try  update of score via goals from (match) report
+        score_more =  _build_report_score( live, timeline )
+        if score_more
+           rec.score = {}.merge( rec.score||{}, score_more )
+        end
+
+
+      report = MatchReport.build( live, timeline )
+
+      data = {
                meta: {
                   name: desc(live['SeasonName']),
                   slug: slug,
                   season: season.to_s,
                   generated: Time.now.to_s,
                }
-            }.merge( _build_match( live ),
-                     _build_report( live, timeline ))
+            }
 
-      ## try  update of score via goals from report
-      score_more =  _build_report_score( live, timeline )
-      if score_more
-           rec[:score] = rec[:score].merge( score_more )
-      end
+      data = data.merge( rec.as_json, report.as_json )
 
 
       ## build basename e.g  2026-07-15_ARG-ENG
       basename = _report_basename( m )
       outpath = "#{outdir}/#{season.to_path}/#{slug}/#{basename}.json"
-      write_json( outpath, rec )
+      write_json( outpath, data )
    end
+
+   print "DONE\n"
 end

data/lib/fifadat/convert/convert.rb

@@ -5,44 +5,49 @@ def convert( slug:, season:,
                 indir: '.',
                 outdir: './tmp' )
 
-    season = Season(season)
-
-    ## fix - change data to top or ??? - why? why not?
-    data = {}
-
-    ## add slug & seasons (add name to be done!!)
-    data[:meta] = { slug:      slug,
-                    season:    season.to_s,
-                    generated: Time.now.to_s,
-                  }
+   season = Season(season)
 
 
+   ### fix-fix-fix  change to read_json_results or ??? - why? why not?
    matches =  read_json_v2( "#{indir}/#{slug}/#{season.to_path}_matches.json" )
    matches = matches['Results']  ## only use results (match) array
 
    ## pp matches
-   puts "  #{matches.size} match(es) in season #{season}"
-
-
-  ## read in stages
-  ##   incl.  SequenceOrder, StageLevel (optional)
-  stages = Stages.read( "#{indir}/#{slug}/misc/#{season.to_path}_stages.json" )
+   puts "==> convert - #{matches.size} match(es) in #{slug} #{season}"
 
 
-  stages.add_matches( matches )
-  data[:stages] = stages.as_json
-
+   ## read in stages
+   ##   incl.  SequenceOrder, StageLevel (optional)
+   stages = Stages.read( "#{indir}/#{slug}/misc/#{season.to_path}_stages.json" )
+   stages.add_matches( matches )
 
    teams = Teams.new
    teams.add_matches( matches )
-   data[:teams] = teams.as_json
-
 
    stadiums = Stadiums.new
    stadiums.add_matches( matches )
+
+
+
+    data = {}
+
+    ## add slug & seasons (add name to be done!!)
+    data[:meta] = { slug:      slug,
+                    season:    season.to_s,
+                    generated: Time.now.to_s,
+                    version:   FifadatConvert::VERSION,
+                    teams:    teams.size,
+                    matches:  matches.size,
+                    stages:   stages.size,
+                    stadiums:  stadiums.size,
+                  }
+
+   data[:stages]   = stages.as_json
+   data[:teams]    = teams.as_json
    data[:stadiums] = stadiums.as_json
 
 
+
    ## for match-by-match live reports
    report_dir   = "#{indir}/#{slug}/matches/#{season.to_path}"
    timeline_dir = "#{indir}/#{slug}/timelines/#{season.to_path}"
@@ -56,112 +61,78 @@ def convert( slug:, season:,
 
 
         ## add/fill-up match basic
-       rec = _build_match( m )
-
+       rec = Match.build( m,  teams:    teams,
+                              stadiums: stadiums,
+                              stages:   stages )
 
-   ### get match (live) details
    ###
-   ##   check if match report exits
-   ##    optional for now!!
+   ##   note - skip check for match events / goals etc.
+   ##      if not yet played!!!
+   #         1-scheduled and
+   #         2-live
+   #         7-postponed (too)
+   #    return empty score!!!
+      if !(m['MatchStatus'] == 1 ||
+           m['MatchStatus'] == 2 ||
+           m['MatchStatus'] == 7)
+
 
-      live = _read_report( m, report_dir: report_dir )
 
+      ### get match (live) details
+      ###
+       ##   check if match report & timeline exits
+      ##    optional for now!!
+
+      live     = _read_report( m, report_dir: report_dir )
+      timeline = _read_timeline( m, timeline_dir: timeline_dir )
 
       ## if live.nil?
       ##   puts "warn no match report for #{_report_basename(m)}"
       ## end
 
 
-   ###
-   ##   note - skip check for match events / goals etc.
-   ##      if not yet played!!!
-
-   if live && !(rec[:status] == 'TIMED' ||
-                rec[:status] == 'SCHED' || rec[:status] == 'SCHEDULED' ||
-                rec[:status] == 'LIVE')
-
-
-      ## check for timeline
-      timeline = _read_timeline( m, timeline_dir: timeline_dir )
-      ## reuse generated output from report
-        report = _build_report( live, timeline )
-
         ## try  update of score via goals from (match) report
         score_more =  _build_report_score( live, timeline )
         if score_more
-           rec[:score] = rec[:score].merge( score_more )
+           rec.score = {}.merge( rec.score||{}, score_more )
         end
 
 
+        ## reuse generated output from report
+        report =  MatchReport.build( live, timeline )
 
-        goals1 = report[:goals1]
-        goals2 = report[:goals2]
-
+       ############
+       ## add goals
 
-       if goals1.empty? && goals2.empty?
+       if report.goals1.empty? && report.goals2.empty?
           ## skip if no goals
        else
-         rec[:goals1] = goals1
-         rec[:goals2] = goals2
-
-
-
+         rec.goals1 = report.goals1
+         rec.goals2 = report.goals2
        end
 
 
        #########
-       ##  add penalties
-         penalties = report[:penalties]
-          rec[:penalties] = penalties   if penalties && !penalties.empty?
-
-
-      sentoff1 =  (report[:red1]||[]) + (report[:yellowred1]||[])
-      sentoff2 =  (report[:red2]||[]) + (report[:yellowred2]||[])
-
-      ## sort by minute
-      sentoff1 = sentoff1.sort { |l,r|  l[:minute] <=> r[:minute] }
-      sentoff2 = sentoff2.sort { |l,r|  l[:minute] <=> r[:minute] }
+       ## add penalties
+       rec.penalties = report.penalties   if report.penalties &&
+                                            !report.penalties.empty?
 
-     rec[:sentoff1] = sentoff1    unless sentoff1.empty?
-     rec[:sentoff2] = sentoff2    unless sentoff2.empty?
- end
-
-
-
-=begin
-   "Officials":
-     [{"IdCountry": "BRA",
-       "OfficialId": "361561",
-       "NameShort": [{"Locale": "en-GB", "Description": "Wilton SAMPAIO"}],
-       "Name": [{"Locale": "en-GB", "Description": "Wilton SAMPAIO"}],
-       "OfficialType": 1,
-       "TypeLocalized": [{"Locale": "en-GB", "Description": "Referee"}]},
-      {"IdCountry": "PAR",
-       "OfficialId": "416159",
-       "NameShort":
-        [{"Locale": "en-GB", "Description": "Juan Gabriel BENITEZ"}],
-       "Name": [{"Locale": "en-GB", "Description": "Juan Gabriel Benítez"}],
-       "OfficialType": 4,
-       "TypeLocalized":
-        [{"Locale": "en-GB", "Description": "Fourth official"}]}],
-=end
-
-###
-##  add referees
-    officials = build_officials( m['Officials'], id: false )
-
-    if officials.empty?
-       ## puts "!! WARN no refs / officials found"
-    else
-         rec[:referees] = officials
-    end
 
+      sentoff1 = report.sentoff1
+      sentoff2 = report.sentoff2
+      rec.sentoff1 = sentoff1    unless sentoff1.empty?
+      rec.sentoff2 = sentoff2    unless sentoff2.empty?
 
+      ###
+      ##  add referees
+      rec.officials = report.officials   if report.officials && !report.officials.empty?
+      end
 
 
       recs << rec
    end
-   data[:matches] = recs
+   data[:matches] = recs.as_json
+
 
 
     outpath =  "#{outdir}/#{season.to_path}/#{slug}.json"

data/lib/fifadat/convert/helper-date.rb

@@ -11,3 +11,15 @@ def _fmt_date_local(  date_utc, date_local )
   ## use   20:30 UTC+1  or 20:30 UTC-3
   "#{date_local.strftime( '%Y-%m-%d %H:%M' )} UTC%+d" % diff_in_hours
 end
+
+def _fmt_time_local(  date_utc, date_local )
+
+  ## note:  returns Rational (e.g. 3/1 or 1/4 etc.) use to_f/to_i to convert
+  diff_in_hours = ((date_local - date_utc) * 24).to_f
+  diff_in_days  =  date_local.jd - date_utc.jd
+  ## pp [diff_in_hours, diff_in_days]
+
+
+  ## use   20:30 UTC+1  or 20:30 UTC-3
+  "#{date_local.strftime( '%H:%M' )} UTC%+d" % diff_in_hours
+end

data/lib/fifadat/convert/helper-score.rb

@@ -1,4 +1,20 @@
 
+
+def _parse_score( m )
+   h = { }
+
+   ###
+   #  note: for match status
+   #       1-scheduled and
+   #       2-live
+   #       7-postponed
+   #    return empty score!!!
+
+   return h    if m['MatchStatus'] == 1 ||
+                  m['MatchStatus'] == 2 ||
+                  m['MatchStatus'] == 7
+
+
 #
 # add quick fix for club world cup?!
 #      elsif  resultType == 1  ||  ## assume 1 - regular (90 mins+stoppage/injury time)
@@ -6,14 +22,13 @@
 #        m['IdMatch'] == '400019191'  ##  fix for pachuca vs salzburg !!!
 #   has resultType == 0!!!
 
-
-def _parse_score( m )
-   h = { }
-
    m = errata_autofix_score( m )
 
 
   resultType  = m['ResultType']
+   ## add score[:type] - why? why not?
+   h[:type] =  RESULT_TYPE[resultType] || "???-#{resultType}"
+
 
   score =  if m['HomeTeam'] && m['AwayTeam']  ## assume inside (match) report
              [m['HomeTeam']['Score'], m['AwayTeam']['Score']].compact
@@ -72,68 +87,96 @@ end
 
 
 
-def _fmt_score( m )
+def _build_report_score( live, timeline=nil )
 
-   m = errata_autofix_score( m )
+    _scores = _build_score_from_goals( live )
 
-  ## m = (full) match hash incl.  IdMatch, etc.
-  ##  returns string e.g.  4-4  or 4-3 a.e.t etc
 
-  resultType  = m['ResultType']
+    score = nil
 
-  # resultType
-  #            0 =>  no result / not played yet
-  #            1 => regular (90 mins)
-  #            2 => aet (120 mins), win on pens
-  #            3 => aet (120 mins)
-  #            8 =>  same as 3?  -aet with golden goal/silver goal in 1998 FRA-PAR
-
-  score = if resultType == 2   ## aet, win on pens
-             "#{m['HomeTeamScore']}-#{m['AwayTeamScore']}" +
-             " a.e.t., #{m['HomeTeamPenaltyScore']}-#{m['AwayTeamPenaltyScore']} pen."
-           elsif resultType == 3 || resultType == 8  ## aet
-             "#{m['HomeTeamScore']}-#{m['AwayTeamScore']} a.e.t."
-           elsif  resultType == 1  ||  ## assume 1 - regular (90 mins+stoppage/injury time)
-                  resultType == 4
-              "#{m['HomeTeamScore']}-#{m['AwayTeamScore']}"
-           elsif  resultType == 0
-               ##  double check if score present
-               ##   e.g. pachuca vs salzburg in cwc 2025??
-               raise ArgumentError,
-                  " resultType == 0 but score present idMatch #{m['IdMatch']} #{m['HomeTeamScore']}-#{m['AwayTeamScore']}"  if m['HomeTeamScore'] &&
-                                                                                                                               m['AwayTeamScore']
-              ""
-           else
-              raise ArgumentError, "unknown/unexpected result type #{resultType}"
-           end
-
-  score
-end
+     case live['ResultType']
+     when 1,4
+         ## 1 => 'REGULAR',
+         ## 4 => 'REGULAR/AGG',  ##  aggregate (1st/2nd leg) - regular
+        ## add ht
+        ##  assert ft match
+        score      = {  ht: _scores[:ht][:score],
+                         ft: _scores[:ft][:score]
+                      }
+     when 3,5,8
+       score       = {   ht: _scores[:ht][:score],
+                         ft: _scores[:ft][:score],
+                         et: _scores[:et][:score],
+                      }
+     when 2
+        ##  todo/fix - pull in timeline check too to check
+        ##     if any events recorded for extra time!!
+        score      = {   ht: _scores[:ht][:score],
+                         ft: _scores[:ft][:score],
+                      }
 
+        ## note - only incl. extra-time (et) if a goal scored in extra time
+        score[:et] = _scores[:et][:score]   unless _scores[:et][:minutes].empty?
+     end
+
+     score
+end
 
 
 
-__END__
-
-  score = if resultType == 2   ## aet, win on pens
-             { et: [m['HomeTeamScore'],m['AwayTeamScore']],
-               p:  [m['HomeTeamPenaltyScore'],m['AwayTeamPenaltyScore']] }
-           elsif resultType == 3 || resultType == 8  ## aet
-             { et: [m['HomeTeamScore'], m['AwayTeamScore']] }
-           elsif  resultType == 1  ||  ## assume 1 - regular (90 mins+stoppage/injury time)
-                  resultType == 4  ||
-                  m['IdMatch'] == '400019191'  ##  fix for pachuca vs salzburg !!!
-             { ft: [m['HomeTeamScore'],m['AwayTeamScore']] }
-           elsif  resultType == 12
-               ## fix/fix/fix - check for score too
-                nil
-           elsif  resultType == 0
-              ##  pachuca vs salzburg in cwc 2025??
-               ##  double check if score present
-               raise ArgumentError,
-                  " resultType == 0 but score present idMatch #{m['IdMatch']} #{m['HomeTeamScore']}-#{m['AwayTeamScore']}"  if m['HomeTeamScore'] &&
-                                                                                                                               m['AwayTeamScore']
-              nil
-           else
-              raise ArgumentError, "unknown/unexpected result type #{resultType}"
-           end
+def _build_score_from_goals( m )
+   ##
+   ## 5 periods - 3 (1ST_HALF), 5 (2ND_HALF),
+   ##             7 (EXTRA_TIME_1ST_HALF), 9 (EXTRA_TIME_2ND_HALF),
+   ##             11 (PENALTY_SHOOTOUT) possibly
+   ##
+   goals = {
+      count: 0,
+      score: [0,0],
+      ht:   { score: [0,0], minutes: [] },
+      ft:   { score: [0,0], minutes: [] },
+      et:   { score: [0,0], minutes: [] },
+      p:    { score: [0,0], minutes: [] },
+   }
+
+   ## note - quick fix
+   ##            add  period 2  for USA v Belgium (PKO) - worldcup
+
+   ## note - i is 0|1  -  array index for team
+   [m['HomeTeam']['Goals'],
+    m['AwayTeam']['Goals']].each_with_index do |recs,i|
+      recs.each do |h|
+
+        h = errata_autofix_goal( h )
+
+        period = h['Period']
+        ## include penalty shootout (11) - why? why not?
+        assert(  [3,5,7,9,11].include?(period),
+                  "goal in period 3/5/7/9/11 expected; got #{h.pretty_inspect} in match #{m.pretty_inspect}" )
+
+        minute = h['Minute']
+        key =  case period
+               when 3    then  :ht
+               when 5    then  :ft
+               when 7, 9 then  :et
+               when 11   then  :p
+               else
+                raise ArgumentError,
+                  "goal in period 3/5/7/9/11 expected; got #{h.pretty_inspect}"
+               end
+
+        goals[:count] += 1
+        goals[:score][i] += 1   unless key == :p
+        goals[key][:score][i] += 1
+      end
+   end
+
+   ## use/calc cummulative score  (BUT not for penalties)
+   goals[:ft][:score][0] += goals[:ht][:score][0]
+   goals[:ft][:score][1] += goals[:ht][:score][1]
+
+   goals[:et][:score][0] += goals[:ft][:score][0]
+   goals[:et][:score][1] += goals[:ft][:score][1]
+
+   goals
+end